Mittal, K.L. - Contact Angle, Wettability and Adhesion Volume 6-VSP (2009)

The topic of wettability (measured in terms of contact
angle) is of tremendous interest from both fundamental

and applied points of view, Wettability plays an essential
role in many industrial processes, so an understanding
of factors dictating wettability and how to modulate it
is of paramount importance. In the last years there has
been an explosive interest in superhydrophobic surfaces
(i.e., surfaces with water contact angle of 150° or higher)
because of their relevance/importance in many areas
ranging from self-cleaning windows to nanofluidics. Also
recently there has been heightened activity in the field of
electrowetting.
Contact Angle, Wettability and Adhesion

This book is divided into four parts: Part 1: Fundamental
Aspects; Part 2: Wettability Control/Modification; Part Contact Angle,
3: Superhydrophobic Surfaces; and Part 4: Surface Free
Energy and Relevance of Wettability in Adhesion. The
topics covered include: a guide to the equilibrium contact
angles maze: fundamental aspects of wetting of rough
Wettability and
and chemically heterogeneous surfaces: work of adhesion
for rock-oil-brine systems; Is the world basic?; wettability
control/modification using various approaches; superhy-
Adhesion
drophobic surfaces and ways to impart superhydrophobici-
ty; adsorption on superhydrophobic surfaces; solid surface
energy determination; surface modification of different
Volume 6
materials; relevance of wettability and adhesion aspects
in a variety of reinforced composites.
Edited by
In essence, this volume reflects the cumulative wisdom
of many active and renowned researchers and provides
a commentary on contemporary research in the fascina-
K.L. Mittal
ting world of contact angles and wettability. This volume
and its predecessors (5 volumes), containing bountiful
information, will be of much value to anyone interested/ Volume 6
involved in controlling wetting phenomena and their ap-
plications.
K.L. Mittal
(Ed.)
9 789004 169326
VSP
brill.nl A.S.
mittal2.indd 1 10-06-2009 16:07:31

Contact Angle, Wettability and Adhesion
Volume 6
This page intentionally left blank
Contact Angle, Wettability
and Adhesion
Volume 6
Edited by
K. L. Mittal
LEIDEN • BOSTON
2009
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2009 by Koninklijke Brill NV Leiden The Netherlands
CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works

Version Date: 20120525
International Standard Book Number-13: 978-9-00-418102-1 (eBook - PDF)
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts
have been made to publish reliable data and information, but the author and publisher cannot assume
responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize
to copyright holders if permission to publish in this form has not been obtained. If any copyright material
has not been acknowledged please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, trans-
mitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter
invented, including photocopying, microfilming, and recording, or in any information storage or retrieval
system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the
CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
Contents
Preface ix
Part 1: Fundamental Aspects
A Guide to the Equilibrium Contact Angles Maze
A. Marmur 3
Kinetics of Triple Line Motion during Evaporation
M. E. R. Shanahan and K. Sefiane 19
What Can We Learn from the Vibration of Drops Deposited on Rough
Surfaces? Wetting Transitions Occurring on Rough Surfaces
E. Bormashenko, G. Whyman and R. Pogreb 33
Length Scale Effects in Wetting of Chemically Heterogeneous Surfaces
N. Anantharaju, M. V. Panchagnula and S. Vedantam 53
Effects of Water Adsorption on Silicon Oxide Nano-asperity Adhesion in
Ambient Conditions
D. B. Asay, A. L. Barnette and S. H. Kim 65
Work of Wetting Associated with the Spreading of Sessile Drops
C. W. Extrand 81
Effect of Line Tension on Work of Adhesion for Rock–Oil–Brine
Systems
D. Saini and D. N. Rao 95
Is the World Basic? Lessons from Surface Science
K. L. Mittal and F. M. Etzler 111
Part 2: Wettability Control/Modification
Reversible Transition from Hydrophobicity to Hydrophilicity of Photon
Responsive Surfaces: From Photochromic Molecules to Nanocrystals
A. Athanassiou 127
Femtosecond Laser-Induced Surface Structures on Platinum and Their
Effects on Surface Wettability and Fibroblast Cell Proliferation
E. Fadeeva, S. Schlie, J. Koch, B. N. Chichkov, A. Y. Vorobyev and C. Guo 163
vi Contents
A Novel Design of Water- and Oil-Repellent Surface Modifier Having

Double-Fluoroalkyl Groups
T. Kawase, S. Ohshita, C. Yoshimasu and T. Oida 173
Toward Superlyophobic Surfaces
W. Ming, B. Leng, R. Hoefnagels, D. Wu, G. de With and Z. Shao 191
The Porosity and Wettability Properties of Hydrogen Ion Treated
Poly(tetrafluoroethylene)
H. S. Salapare III, G. Q. Blantocas, V. R. Noguera and H. J. Ramos 207
Part 3: Superhydrophobic Surfaces
Superhydrophobicity: Localized Parameters and Gradient Surfaces
G. McHale, S. J. Elliott, M. I. Newton and N. J. Shirtcliffe 219
Design of Superhydrophobic Paper/Cellulose Surfaces via Plasma
Enhanced Etching and Deposition
B. Balu, J. S. Kim, V. Breedveld and D. W. Hess 235
Superhydrophobic Aluminum Surfaces: Preparation Routes, Properties
and Artificial Weathering Impact
M. Thieme, C. Blank, A. Pereira de Oliveira, H. Worch, R. Frenzel,
S. Höhne, F. Simon, H. G. Pryce Lewis and A. J. White 251
Aqueous and Non-aqueous Liquids on Superhydrophobic Surfaces:
Recent Developments
M. Ferrari 269
Part 4: Surface Free Energy and Relevance of Wettability
in Adhesion
Comparison of Apparent Surface Free Energy of Some Solids
Determined by Different Approaches
E. Chibowski and K. Terpilowski 283
Surface Free Energy of Viscoelastic Thermal Compressed Wood
M. Petrič, A. Kutnar, B. Kričej, M. Pavlič, F. A. Kamke and M. Šernek 301
Modification of Sugar Maple Wood Board Surface by Plasma
Treatments at Low Pressure
V. Blanchard, B. Riedl, P. Blanchet and P. Evans 311
The Effect of a Plasma Pre-treatment on the Quality of Flock Coatings
on Polymer Substrates
T. Bahners, G. Hoffmann, J. Nagel, E. Schollmeyer and A. Voigt 325
Adhesion Properties of Wood–Plastic Composite Surfaces:
Atomic Force Microscopy as a Complimentary Analysis Tool
G. S. Oporto, D. J. Gardner and D. J. Neivandt 341
Contents vii
Wettability Behavior and Adhesion Properties of a Nano-epoxy Matrix

with Organic Fibers
W. H. Zhong, Y. Fu, S. Jana, A. Salehi-Khojin, A. Zhamu and
M. T. Wingert 359
Enhancing the Fiber–Matrix Adhesion in Woven Jute Fabric Reinforced
Polyester Resin-Based Composites
A. A. Kafi and B. L. Fox 377
Scratch and Hydrophobic Properties of Al Alloy Surface by
Combination of Transparent Inorganic and Silane-Based Coatings
A. R. Phani, P. De Marco and S. Santucci 391
Preface
This volume embodies the proceedings of the Sixth International Symposium

on Contact Angle, Wettability and Adhesion held under the auspices of MST Con-
ferences at the University of Maine, Orono, Maine during July 14–16, 2008. The
premier symposium with the same title was held in 1992 in honor of Prof. Robert J.
Good as a part of the American Chemical Society meeting in San Francisco, Cali-
fornia and the second, third, fourth and fifth events in this series were held in 2000
in Newark, New Jersey; 2002 in Providence, Rhode Island; 2004 in Philadelphia,
Pennsylvania; and 2006 in Toronto, Canada, respectively, and all were held under
the aegis of MST Conferences. The proceedings of these earlier symposia have been
properly documented as five hard-bound books [1–5].
The topic of wettability (measured in terms of contact angle) is of tremendous
interest from both fundamental and applied points of view. Wettablility plays an
essential role in many industrial processes, concomitantly an understanding of fac-
tors dictating wettability and how to modulate it is of paramount importance. In
the last years there has been an explosive interest in superhydrophobic surfaces
(i.e., surfaces with water contact angle of 150◦ or higher) because of their rele-
vance/importance in many areas ranging from self-cleaning windows to nanoflu-
idics to MEMS/NEMS to biomedical technology. The high tempo of research and
interest in this burgeoning field can be gauged from recently published book cover-
ing many and varied ramifications of superhydrophobic surfaces [6]. Also recently
there has been heightened activity in the field of electrowetting.
Apropos, I would like to underscore that since the seminal paper published by
Young (Thomas Young, Phil. Trans. Royal Soc. 95, 65–87 (1805)) there has been
voluminous literature published relative to contact angles and wettability. However,
in spite of a large body of literature (thousands of papers) still some vexing ques-
tions remain, for example: What is the most appropriate method to measure contact
angle? What kind of contact angle should be used in determining the surface free
energy of solid surfaces? What is the most appropriate or justified model/approach
to determine solid surface free energy? If one sifts the existing literature, one will
find a great deal of confusion and discordance and very many divergent views in
the arena of contact angles and wettability. Let us hope that this bewildered state
Contact Angle, Wettability and Adhesion, Vol. 6
© Koninklijke Brill NV, Leiden, 2009
x Preface
of affairs will become clear in the future and there will be consensus (ecumenical
agreement) regarding the questions posed above among surface scientists world-
wide. Parenthetically, I would like to add a personal note. I was very fortunate to
learn surface chemistry in 1968 from Prof. Arthur W. Adamson, a veritable “giant”
in surface chemistry, and he told us one day in the class that there were two kinds
of contributions: either you add something worthwhile to the literature or take out
what confounds the situation (in plain vernacular this means rid the literature of
nonsense).
Now coming to this volume, it contains a total of 25 papers covering many and
varied aspects of contact angle, wettability and adhesion. As done in previous vol-
umes, it must be recorded that all manuscripts were rigorously peer-reviewed and
revised (some twice or thrice) and properly edited before inclusion in this book.
Concomitantly, this volume represents an archival publication of the highest stan-
dard. It should not be considered a proceedings volume in the usual and ordinary
sense, as many so-called proceedings volumes are neither peer-reviewed nor ade-
quately edited. By the way, the technical program for the symposium consisted of
59 papers but the remaining papers are not included in this book for a variety of
reasons (including some which did not pass muster).
This book (designated as Volume 6) is divided into four parts: Part 1: Fundamen-
tal Aspects; Part 2: Wettablility Control/Modification; Part 3: Superhydrophobic
Surfaces; and Part 4: Surface Free Energy and Relevance of Wettability in Adhe-
sion. The topics covered include: a guide to the equilibrium contact angles maze;
fundamental aspects of wetting of rough and chemically heterogeneous surfaces;
work of adhesion for rock-oil-brine systems; Is the world basic?; wettability con-
trol/modification using various approaches (both wet and dry); superhydrophobic
surfaces and ways to impart superhydrophobicity; adsorption on superhydrophobic
materials; relevance of wettability and adhesion aspects in a variety of reinforced
composites.
In essence, this volume reflects the cumulative wisdom of many active and
renowned researchers and provides a commentary on contemporary research in the
fascinating world of contact angles and wettability. This volume and its predeces-
sors (5 volumes) containing bountiful information (a total of ∼3300 pages) should
be of much value and interest to anyone interested/involved in controlling wetting
phenomena and their applications. This set of 6 volumes provides a unique reposi-
tory of information in a single source.
Acknowledgements
Now comes the pleasant task of thanking all those who helped in varied ways. First
of all, I (on behalf of MST Conferences) would like to express my sincere gratitude
to Prof. Douglas J. Gardner of the University of Maine for sponsoring this sym-
posium as well as for his generous support. Prof. Gardner and his colleagues were
extremely helpful during the course of this symposium and we appreciated it very
Preface xi
much. Then, it is a pleasure to express my sincere thanks to my colleague and dear

friend, Dr. Robert H. Lacombe, for taking care of the requisite details entailed in
organizing this symposium. Thanks are extended to all the contributors to this book
for their interest, enthusiasm, patience and cooperation without which this book
would not have been in the hands of the readers. The unsung heroes (reviewers) are
profusely thanked for their time and efforts in providing many valuable comments
which contributed significantly towards improving the quality of manuscripts. Fi-
nally, my appreciation goes to the staff of VSP/Brill (publisher) for incarnating this
book.
References
1. K. L. Mittal (Ed.), Contact Angle, Wettability and Adhesion. VSP, Utrecht (1993).
2. K. L. Mittal (Ed.), Contact Angle, Wettability and Adhesion, Vol. 2. VSP, Utrecht (2002).
3. K. L. Mittal (Ed.), Contact Angle, Wettability and Adhesion, Vol. 3. VSP, Utrecht (2003).
4. K. L. Mittal (Ed.), Contact Angle, Wettability and Adhesion, Vol. 4. VSP/Brill, Leiden (2006).
5. K. L. Mittal (Ed.), Contact Angle, Wettability and Adhesion, Vol. 5. VSP/Brill, Leiden (2008).
6. A. Carré and K. L. Mittal (Eds), Superhydrophobic Surfaces. VSP/Brill, Leiden (2009).
K. L. Mittal
P.O. Box 1280
Hopewell Jct., NY 12533, USA
Part 1
Fundamental Aspects
A Guide to the Equilibrium Contact Angles Maze
Abraham Marmur ∗
Department of Chemical Engineering, Technion — Israel Institute of Technology,
32000 Haifa, Israel
Abstract
Understanding, measuring, and interpreting equilibrium contact angles appear to be simple, but may actually
be quite confusing. This paper is an attempt at a guide to the perplexed. First, a comprehensive, clearly
defined terminology is suggested. Then, the theory of equilibrium contact angles on smooth, rough, or
chemically heterogeneous surfaces is briefly discussed. Finally, the practical implications of the theory to
contact angle measurement and interpretation are indicated and explained.
Keywords
Contact angle, hysteresis, surface tension, interfacial tension, roughness, chemical heterogeneity
1. Introduction
Systems that include solids and liquids are ubiquitous, e.g. soil and water, food
powder and water, eyeballs and tears, artificial surfaces and blood, paper and ink,
metal and adhesive, bearing balls and oil, wall and paint, and many more. The
process of bringing a liquid to contact a solid surface, which is referred to as wet-
ting, has attracted scientific attention over more than two centuries (e.g. [1–15]).
Wetting usually occurs within an environment that may consist of a gas or another
immiscible liquid, each of which may be referred to as a fluid. A wetting system
is characterized by a contact angle (CA), which is defined as the angle between
the tangent to the liquid–fluid interface and the tangent to the solid surface at the
contact line between the three phases (Fig. 1). The CA is usually measured on
the liquid side (when the system involves two liquids, it is usually measured on the
denser liquid side). A low contact angle means that the solid is well wetted by the
liquid (hygrophilic solid surface), while a high contact angle indicates a preference
for solid–fluid contact (hygrophobic solid surface).1 When the process of interest
is static, equilibrium CAs are discussed; when it involves very slow motion, quasi-
* Fax: 972-4-829-3088; e-mail: marmur@technion.ac.il

1 Hygrophilic and hygrophobic literally mean liquid-loving and liquid-fearing (hygro in Greek means liq-
uid) [42]. In the specific case of water in air, the terms hydrophilic and hydrophobic are used.

4 A. Marmur
equilibrium may exist, which closely approximates the equilibrium state; when the
process requires high speeds, such as in tape coating, dynamic CAs are considered.
CAs may be used to characterize specific wetting systems as such, or for as-
sessing the surface tension of a solid surface as an independent component of a
wetting system (in this paper, for simplicity, no distinction is made between the
surface tension of a solid and its specific surface energy; also, only rigid solid sur-
faces are discussed). The main point underlying the latter usage is that there is no
direct method to measure the surface tension of solid surfaces, since they are not
deformable as are liquid surfaces. The measurement of CAs and their interpretation
appear to be straightforward and simple; however, this impression is very deceptive
[15]. To start with, there are a few types of CAs to consider, some of which may
be easier to measure but more difficult to interpret. Therefore, in order to avoid
questionable measurement and misleading interpretations, the theory of CAs needs
to be well understood. The theory for a general system in three dimensions is quite
complicated and not entirely understood as yet. Fortunately, however, even the most
complicated theoretical considerations that do exist lead to simple conclusions that
can be practically applied [15].
There are many possible configurations of solid–liquid–fluid systems, such as a
liquid drop on a solid surface (sessile drop), a fluid bubble underneath a solid sur-
face (captive bubble), a solid plate partially dipped into a liquid, a liquid inside a
porous medium, or a particle floating on a liquid–fluid interface. For CA measure-
ment, the sessile drop method is frequently used, since this is the most convenient
method; however, other contact angle measurement methods are used as well, such
as the captive bubble, and the Wilhelmy plate [8]. This paper focuses on sessile
drops; nonetheless, all theories and conclusions remain valid for or can be easily
adapted to all the other systems as well.
The objective of this paper is to assist in understanding the measurement and
interpretation of equilibrium CAs. The next section describes the variety of CAs
that need to be considered. It is followed by a brief summary of the theory as we
currently understand it, and is concluded with practical indications for CA mea-
surement and assessment of the surface tension of solid surfaces. The discussion
is deliberately brief, in order to emphasize the main points. It mainly reflects the
viewpoint of the author, which has evolved from work with colleagues over the last
two decades or so. The reference list is not intended to be comprehensive — the
cited papers include many more references to the work of researchers who laid the
foundations on which this viewpoint has been developed.
2. The Contact Angle Lexicon

The general, operational definition of the CA was given in the Introduction. How-
ever, there are sub-definitions of CAs that need to be used under various circum-
stances. This section describes them in a concise form, in order to serve as a quick
reference and enable easy comparison.
A Guide to the Equilibrium Contact Angles Maze 5
(a) (b)
(c)
Figure 1. Contact angles: (a) various geometric contact angles for the same volume of the drop; (b) the
actual contact angle; (c) the apparent contact angle.
The geometric CA, θ , (Fig. 1a) is the one calculated using geometric consider-
ations only, regardless of whether the system is at equilibrium [15]. It serves as an
independent variable for calculating all possible Gibbs energy states of the system.
For example, for an axisymmetric spherical drop (in the absence of gravity), it is
related to the base radius of the drop, rb , and its volume, V , by
(2 − 3 cos θ + cos3 θ )/ sin3 θ = (3V )/(πrb3 ). (1)
The ideal CA, θi , is the equilibrium CA that a liquid makes with an ideal solid
surface. The latter is defined as smooth, rigid, chemically homogeneous, insoluble,
and non-reactive surface. The ideal CA is also sometimes referred to as the intrinsic
or inherent CA.
The Young CA, θY , is defined by the equation conceived by Young [1]
σsf − σsl
cos θY = , (2)
σlf
where σlf , σsl and σsf are the liquid–fluid, solid–liquid, and solid–fluid interfacial
tensions, respectively. For most macroscopic systems, the Young CA represents the
ideal CA; however, as argued in the following, for systems of surfaces with very
small radii of curvature, they may be different due to line tension. The Young CA
depends only on the physico-chemical nature of the three phases, and is independent
of both the system geometric shape and gravity. The latter may affect the shape of
the liquid–fluid interface, but not θY . It is important to notice that σsf represents the
6 A. Marmur
surface tension of the solid surface in its equilibrium state with the fluid and liquid
environments, not that of the same solid surface when perfectly clean under high
vacuum. Due to possible adsorption on the solid surface, the two surface tensions
may be different; the difference is termed “film pressure” [7] and may be difficult
to measure since CA measurements under high vacuum are usually impractical. For
assessment of the surface tension of a solid, σs , it is more convenient to work with
systems for which the fluid is air. In these cases, σsf and σlf are replaced by σs
and σl , the surface tension of the liquid. As explained below, σs may be calculated
from equation (2) if the value of the Young CA is known (assuming σsl can be
calculated). However, θY may not be directly measurable. The question of how to
obtain θY from measurable parameters is a central issue in CA theory.
The actual CA, θac , is the local equilibrium CA at a point on a solid surface
that may be rough or chemically heterogeneous (Fig. 1b). For most macroscopic
systems, the actual CA is equal to the Young CA; however, as discussed below,
when line tension is meaningful they may be different.
The apparent CA, θap , is an equilibrium CA measured macroscopically on a solid
surface that may be rough or chemically heterogeneous. The detailed topography
of a rough solid surface cannot be viewed with regular optical means; therefore,
this CA is defined as the angle between the tangent to the liquid–fluid interface and
the apparent solid surface, as macroscopically observed (Fig. 1c). θap is actually
an equilibrium value of the geometric CA, θ . It is important to note that, as shown
schematically in Fig. 1c, the apparent and actual CAs may be very different. As ex-
plained below, there exists a range of discrete apparent CAs for any non-ideal solid
surface.
The advancing CA, θad , is the highest possible apparent CA that can be achieved
for a given wetting system (usually it is realized by increasing the volume of the
drop). The advancing CA is an equilibrium (though metastable) CA, since θap is,
by definition, an equilibrium CA.
The receding CA, θr , is the lowest possible apparent CA that can be achieved
for a given wetting system (usually it is realized by decreasing the volume of the
drop). Similarly to the advancing CA, the receding CA is an equilibrium (though
metastable) CA.
The most stable CA, θms , is the apparent CA that is associated with the lowest
possible Gibbs energy of the wetting system. The value of this CA is expressed,
under appropriate conditions, by the Wenzel, Cassie, or Cassie–Baxter equations
[2–4], as discussed below.
The dynamic CA is the one measured when there is relative motion between the
liquid and the solid at appreciable velocities. Its value depends on the contact line
velocity. In order to avoid ambiguity, this term should not be used for CAs measured
by pushing the liquid very slowly in order to obtain advancing or receding values.
For such measurements, the term quasi-static seems to be the most appropriate.
3. Equilibrium Contact Angle Theory

3.1. Contact Angles on Ideal Solid Surfaces
Equilibrium of a wetting system at constant pressure and temperature is achieved
(as that of any system) when the Gibbs energy is minimal [16]. The Gibbs en-
ergy, G, of a wetting system with an ideal solid surface is given by
G = σlf Alf + σsl Asl + σsf Asf = σlf Alf + (σsl − σsf )Asl + σsf Atotal , (3)
where Alf , Asl and Asf are the liquid–fluid, solid–liquid, and solid–fluid interfa-
cial areas, respectively, and Atotal is the (constant) total area of the solid surface
(Atotal = Asl + Asf = constant). In the case of an ideal solid surface, the Gibbs en-
ergy of a wetting system, as a function of the geometric CA, has a single minimum
(Fig. 2). The geometric CA leading to this minimum is defined as the ideal CA.
If the interfacial tensions are constant over the whole corresponding interfaces, the
ideal CA equals the Young CA [1].
However, as already noted by Gibbs [16], the values of the interfacial tensions
at a close proximity to the contact line may not be the same as their values further
away from it. This is so, because any of the three phases may affect the interaction
between the other two in the immediate vicinity of the contact line. Most significant
could be the effect of the solid (due to its high density) on the surface tension of the
liquid very close to the contact line [11, 17]. Gibbs [16] proposed that this three-
phase mutual interaction could be accounted for by the concept of “line tension”, τ ,
which was later quantified for a solid–liquid–gas system in the following form [18–
20]:
τ
cos θi = cos θY − . (4)
rb σl
Figure 2. The Gibbs energy for a drop on an ideal solid surface.

8 A. Marmur
Another way to account for this effect is to indeed consider σl to be a function

of the distance from the solid rather than a constant property [17]. However, the
line tension approach seems more convenient, and in any case the two approaches
should be, in principle, equivalent.
The order of magnitude and sign of line tension have been debated for many
years [21–25]. Values claimed in the literature are spread over a few orders of mag-
nitude, between 10−11 N and 10−6 N [21–25]. However, it has become clear that
line tension may be roughly of the order of magnitude of 10−9 –10−10 N, and may
affect the CA of drops only when they are extremely small, i.e. when the radius
of curvature is in the nano range [26–29]. Therefore, for all wetting systems with
larger radii of curvature, the line tension effect on the ideal CA is negligible. Thus,
the ideal CA is practically identical with the Young CA, and this equality is assumed
in the following.
3.2. Contact Angles on Rough or Chemically Heterogeneous Surfaces
Real surfaces are usually rough or chemically heterogeneous, or both. On rough

surfaces, the local angle of inclination of the solid surface with respect to some
reference plane varies from point to point. Similarly, on chemically heterogeneous
surfaces, the chemical composition is different at various locations. Therefore, the
first fundamental question is: What is the value of the actual CA (Fig. 1b)? The
answer was theoretically given in a very general way for three-dimensional systems:
at equilibrium θac = θY locally, if line tension is negligible [30–32]. Thus, for rough
and chemically heterogeneous surfaces in general, the actual CA varies along the
contact line, according to the local chemical composition and inclination of the solid
surface. Consequently, the contact line is “wavy”, and the shape of the liquid–air
interface has to adjust itself accordingly (Fig. 3). However, the actual CA is usually
not a measurable quantity, since it may be very difficult, or even impossible, to make
local CA measurements, especially on rough surfaces. Thus, the next question is:
What is the experimentally measurable quantity, and how is it related to the Young
CA?
Figure 3. A three-dimensional view of a drop, showing the variation in the actual contact angle along
the contact line.
It is well known that the apparent contact angle is the measurable quantity that
can be observed macroscopically (Fig. 1c). The relationship between θap and θY
depends on the nature of the solid surface. For an ideal solid surface, the apparent
CA is, obviously, identical to the actual CA as well as to the ideal CA. In contrast,
for real surfaces, which may be rough or chemically heterogeneous, the apparent
CA may vary from one point on the solid surface to the other. If the surface is
chemically heterogeneous but perfectly smooth, it may be possible to measure the
various local values of the apparent CA (even though it may be difficult). In this
case, the local apparent CAs equal the local Young CAs. Nonetheless, if the surface
is also rough, it may be impossible to assess the Young CA from local values of the
apparent CA. Moreover, as discussed below, a range of apparent CAs exist. Thus, it
is essential to understand how to meaningfully measure CAs and what information
can be derived from each apparent CA.
This understanding can be derived from the Gibbs energy curve for rough or
chemically heterogeneous surfaces. Because of the complexity of three-dimensional
situations, it is useful to first consider two-dimensional systems. These could be, for
example, either long plates dipped into a liquid, or axisymmetric drops on surfaces
with axisymmetric patterns of roughness or chemical heterogeneity [5]. The Gibbs
energy for such systems is readily calculated for all possible states of the system,
which are defined by the geometric CA (Fig. 4). The most prominent feature of the
Gibbs energy curve for a wetting system with a real solid surface is the existence
of multiple minima [5], in contrast to the case of an ideal solid surface, which is
characterized by a single minimum. Each such minimum defines an equilibrium
state. The minimum that is associated with the lowest Gibbs energy value defines
the most stable equilibrium state. The geometric CA associated with this state is
Figure 4. The Gibbs energy for a drop on a rough or chemically heterogeneous solid surface.
10 A. Marmur
the most stable CA (Fig. 4). The other minima in the Gibbs energy curve represent
metastable equilibrium states, to be discussed below. In-between every pair of min-
ima points, there is, obviously, a maximum point. The difference in Gibbs energy
between a minimum and the following maximum defines the energy barrier to be
overcome when moving from one metastable equilibrium state to the next (Fig. 4).
As Fig. 4 also shows, the minima in the Gibbs energy exist only over a finite
range of apparent CAs. This is “the CA hysteresis range”. The highest CA in this
range is the (theoretical) advancing CA. The lowest CA is the (theoretical) reced-
ing CA. As can also be seen in Fig. 4, the energy barrier near the advancing or
receding CA is rather low (approaching zero), and increases as the CA is closer to
the most stable one. Since “natural” vibrations in the laboratory enable the drop
to overcome some magnitude of a potential barrier, the advancing CA in prac-
tice is lower than the theoretically expected value. Similarly, the receding CA in
practice is higher than the theoretically expected value. It is also worthwhile to
emphasize a terminology issue that is frequently confusing. While the transition
between metastable states is a dynamic process, the apparent CAs themselves are
equilibrium CAs (though metastable). Thus, the term “dynamic CA measurement”,
when used with reference to advancing and receding CA measurements, may be
misleading.
The advancing and receding CAs can be relatively well defined from an experi-
mental point of view. However, a theory that relates these CAs with the Young CA
or with the surface tension of the solid in any other way has not yet been developed
(some initial steps, though, have been made (see [14, 33–35])). Therefore, the hys-
teresis range can be currently used in only two ways: (a) as an empirical measure
of the extent of heterogeneity of the solid surface; and (b) as a means to estimate
the most stable CA. Two suggestions for the latter option have been given in the
literature. One is [36]
cos θms = (cos θad + cos θr )/2 (5)
and the other is [37]
θms = (θad + θr )/2. (6)
These suggestions, however, are not rooted in any fundamental theory.
The most stable CA is experimentally less accessible than the advancing or
receding CA; however, its theory is well defined. Therefore, efforts to define ex-
perimental procedures for its measurement have been initiated [36–38], mainly by
applying vibrations to overcome the Gibbs energy barriers. Mathematically, there
is no way to identify the lowest minimum among multiple minima of a function but
to check all of them one by one. Therefore, there is no analytical way to calculate
the most stable equilibrium state of a wetting system without searching through the
Gibbs energy curve for all metastable states. However, some years ago it was proven
[39, 40] that the equations developed intuitively by Wenzel, Cassie and Baxter [2–
4] are good approximations for the most stable CA if the drop size is sufficiently
large compared with the scale of roughness or heterogeneity. Although it was not
explicitly explained in the original papers, the idea underlying these approxima-
tions is to assume that the drop spreads on an ideal, equivalent solid surface; the
properties of this equivalent surface are the average of the actual properties over the
actual surface. Since the equivalent solid surface is, by definition, ideal, there will
be only one minimum in the Gibbs energy curve for this surface. The apparent CA
associated with this single minimum serves as an approximation for the most stable
CA in the real case.
For a rough but chemically homogeneous solid surface, the surface area is larger
than its projected area by a factor r, which is called the roughness ratio. Now, each
term contributing to the Gibbs energy of a wetting system consists of a product of
the corresponding interfacial tension by the interfacial area; therefore, the increase
in the solid–liquid and solid–fluid interfacial areas by a factor of r can be mathe-
matically viewed as an increase in the interfacial tensions by the same factor. This
immediately leads to the Wenzel equation [2]
cos θW = r cos θY , (7)
where the subscript W indicates the apparent CA calculated according to this
equation. A rigorous mathematical study for a general three-dimensional system
concluded that the Wenzel CA becomes a better and better approximation to the
most stable CA as the size of the drop increases relatively to the typical scale of
the roughness [39]. Initial experiments [38] also seem to confirm the theoretical
validity of these equations for sufficiently large drops.
For a smooth, but chemically heterogeneous surface, the same approach of aver-
aging the interfacial tensions immediately leads to the Cassie equation [4] for the
apparent CA, θC :

cos θC = xi cos θYi . (8)
i
In this equation, xi is the area fraction characterized by a given chemistry i. This
is a generalized form of the original Cassie equation that referred to only two
chemistries. The summation can be replaced by an integral when the chemical het-
erogeneity is continuously varying over the surface. Similarly to the rough surface
case, two-dimensional and three-dimensional numerical simulations [40] showed
that the Cassie CA becomes a very good approximation for the most stable CA as
the drop becomes larger in comparison with the scale of heterogeneity.
A rough surface may, of course, also be chemically heterogeneous. A special
case of roughness and chemical heterogeneity occurs when air is trapped under the
drop within the roughness grooves [3, 5, 41, 42]. This phenomenon is of great in-
terest and importance, especially with regard to super-hygrophobic2 surfaces (e.g.
2 The term “super-hygrophobic surfaces” was suggested as a general term for surfaces that are not wetted
by any liquid [42].
12 A. Marmur
[41–50]). The approximate equation for the most stable CA in this case is the
Cassie–Baxter equation [3], which is actually a combination of the Wenzel and
Cassie equations. This equation is of fundamental importance for designing super-
hygrophobic surfaces, but somewhat of less interest for the assessment of σs .
Recently, based on experiments with chemically heterogeneous and rough sur-
faces, a doubt was cast over the correctness of the Wenzel and Cassie equations
[51]. However, it was argued in response that the experiments were performed with
drop sizes that were of the same order of magnitude as the scale of heterogeneity
[52]. Therefore, these experiments actually emphasize the importance of applying
the Wenzel or Cassie equation only when the drop is sufficiently large rather than
undermining the correctness of these equations.
The Gibbs energy curve in Fig. 4 is typical of two-dimensional systems. The
situation with three-dimensional systems, in general, may be much more com-
plex, since the drop may be asymmetric and there may not be a unique value of
the geometric CA. Thus, the Gibbs energy may not be presented as a function of
one variable only. Also, from an experimental point of view, observing the drop
from various angles may result in different values of the apparent CA. Fortunately,
however, the analysis of the most stable CA on three-dimensional rough [39] and
heterogeneous [40] surfaces enables a major simplification of the problem. It was
found that when the drop is sufficiently large compared with the scale of roughness
or heterogeneity it becomes almost axisymmetric (Fig. 5). This observation enables
an unequivocal definition of the geometric or apparent CA in three-dimensional
systems [39], as follows. When the drop becomes larger and larger, the deviations
from axisymmetry (at the most stable state) become more and more limited to the
close vicinity of the contact line (Fig. 5). When the drop is sufficiently large, most
of it appears to be identical with an equivalent, perfectly axisymmetric drop on
an ideal surface. The apparent CA that this equivalent drop makes with the solid
surface is actually the experimentally measurable apparent CA; theoretically it is
expressed by the Wenzel or Cassie equation [2–4]. Thus, because of this approach
towards axisymmetry with increasing drop size, all the conclusions drawn above for
two-dimensional drops hold also for three-dimensional ones, if they are sufficiently
large.
Figure 5. Three-dimensional views of drops of increasing size (from left to right) relative to the scale
of chemical heterogeneity. The drop becomes more axisymmetric as its size increases.
4. Contact Angle Measurement and Interpretation
As mentioned in the Introduction, a most important application of CA measurement

is the assessment of the surface tension of a solid surface. At present, the only
way to assess σs is through the Young equation, which, in principle, enables this
calculation if θY and σsl are known (σl is independently known). Since σsl itself
cannot be experimentally measured, it has to be calculated from σs and σl using
semi-empirical correlations designed for this purpose ([53–59] and papers cited
therein). Although the question of the proper correlation is still an open problem,
the present discussion is dedicated only to the elucidation of θY from measured
apparent CAs. The above theoretical conclusions are now implemented to achieve
this end.
Following the placement of a drop on a real solid surface, it acquires an apparent
CA somewhere within the hysteresis range. This “as is” apparent CA depends on
the balance between the energy barriers and available drop energy. The latter is de-
termined by the dynamic process that the drop undergoes during touching the solid
surface and the oscillations that follow [60, 61]. Thus, the state of the drop when it
lands on a surface is rather randomly determined. Consequently, it is extremely dif-
ficult, if not impossible, to gain useful information from an apparent CA measured
“as is” (sometimes referred to as the “static CA”).
The only uniquely defined apparent CAs are the advancing, receding, and most
stable ones. The former two are easier to measure than the latter. As is well known,
they can be experimentally determined by increasing or decreasing the drop volume
until the maximum (advancing) or minimum (receding) CA is reached, respectively.
However, there is always a difference between the theoretical and practical values of
these CAs, and there is yet no substantiated theory available for their interpretation.
In contrast, the most stable CA is the one that can be theoretically interpreted in
terms of Young CAs, using the Wenzel or Cassie equation, but its experimental
determination (e.g. using vibrations) is still under development [38]. Sometimes
these equations are being used for the advancing or receding CA, but this has no
justification whatsoever. Regardless of which of these apparent CAs is measured,
there are some common requirements that have to be carefully fulfilled in order to
perform meaningful measurement and interpretation.
The most important requirement that is extracted from the equilibrium CA theory
is that the ratio of the drop size to the scale of roughness or chemical heterogeneity
must be sufficiently large. This requirement is essential for the theoretical inter-
pretation of the most stable CA (in order to be able to use the Wenzel or Cassie
equation); it is also crucial for ensuring axisymmetry of the drop, without which
the apparent CA measurement may be meaningless in many situations. In addition,
it is advantageous also for advancing CA measurements, since this angle is less
sensitive to the drop volume if the latter is sufficiently large [62]. It is not known
yet how large should this ratio be, but it appears that a ratio of two to three orders of
magnitude is satisfactory [38–40]. For example, since a typical practical roughness
14 A. Marmur
scale is of the order of magnitude of a few micrometers, a drop of a few millimeters

in size may be sufficiently large.
The next requirement is closely related to the former: the drop must be axisym-
metric in order for the CA measurement and interpretation to be meaningful. This
requirement necessitates an experimental measurement of axisymmetry. The eas-
iest way to do it is by taking pictures of the drop from above [38]. The contact
angles can still be measured from pictures taken from the side, using a second cam-
era, or calculated from the maximum drop diameter captured from above, using
the Young–Laplace equation with input values for the drop volume, liquid surface
tension, and its density [38].
As explained above, the only substantiated theoretical correlation presently avail-
able links the Young CA with the most stable CA, using the Wenzel or Cassie
equation. Thus, one needs the most stable CA in order to calculate the Young CA.
This can be achieved by two approaches, each having its drawbacks. The first ap-
proach suggests calculating the most stable CA from the advancing and receding
CAs, using various averages as shown in equations (5) and (6). This approach is
experimentally convenient; however, its main drawback is the lack of theoretical
substantiation. The second approach advocates a direct measurement of the most
stable CA by applying, for example, vibrations. The main disadvantage of this ap-
proach is the uncertainty in the experimental definition of the most stable CA.
Once the most stable CA is calculated for a rough surface, the Wenzel equation
can be used to calculate θY . Of course, the roughness ratio must also be known. For
a chemically heterogeneous surface, the most stable CA of a sufficiently large drop
(the Cassie CA) serves as an average θY for the surface. Currently, there is no way
to determine the local surface tension values over a chemically heterogeneous solid
surface from the average θY . Lastly, though a theory for CA hysteresis is still in its
infancy [33–35], measurement of the CA hysteresis range may be informative as an
empirical measure for surface roughness or chemical heterogeneity.
5. Concluding Remarks
As stated in the Introduction, the process of wetting and the measurement of CAs
appear to be straightforward, but actually are very complex. The theory that was
started in 1805 and has been developed mostly over the last five decades still poses
many challenges for future research. The following paragraphs list both practical
hints that may help to correctly deal with CA data, as well as some challenges for
future research.
5.1. Practical Conclusions
The following points should be kept in mind regarding contact angle measurement
and interpretation:
• The measured drop should be sufficiently large (compared with the scale of
roughness or chemical heterogeneity) and axisymmetric.
• The Young CA (which, for macroscopic drops, practically equals the ideal CA)
is the one necessary for calculating the surface tension of a solid surface.
• Three apparent CAs are well defined and measurable for sufficiently large, ax-
isymmetric drops: the most stable, the advancing, and the receding.
• The Young contact angle for a rough surface may be calculated only from the
most stable contact angle, using the Wenzel equation.
• The average Young contact angle for a chemically heterogeneous surface is,
from the Cassie equation, the most stable contact angle.
• The most stable CA can, in principle, be directly measured by applying external
energy to overcome the energy barriers between the metastable states of the
drop, e.g. by using vibrations.
• The most stable contact angle can be roughly estimated by a proper average be-
tween the advancing and receding contact angles; however, this approximation
is not substantiated yet by theory.
5.2. Future Challenges
The above discussion attempted to briefly summarize the current state of the theory
that is essential for equilibrium CA measurement and interpretation. Obviously, ma-
jor challenges still remain for future work. One of them is the theory of CA hystere-
sis. An approach that has been often used to deal with this problem is by referring to
the phenomenon of contact line pinning (e.g. [10, 11, 63, 64]) and analyzing the dy-
namics of the stick–slip process. However, one of the open questions, for example,
is whether contact line pinning is actual or apparent. Two-dimensional calculations,
for example, demonstrated that pinning may be an apparent phenomenon that ap-
pears to exist only when observed macroscopically, whereas microscopically the
contact line does move, but only over very short distances [14, 65]. Another ques-
tion is whether advancing or receding CAs may be measured during slow motion of
a drop, for example when it slides down an inclined plane. It was shown that for a
sliding two-dimensional drop the front or rear CAs do not necessarily coincide with
the advancing or receding equilibrium CAs [66]. The reason is that the drop volume
and shape impose a geometrical relationship between the front and rear CAs, which
does not necessarily coincide with thermodynamic equilibrium requirements. It re-
mains to be seen if this conclusion is valid in three dimensions, and under what
conditions.
Another main challenge is the equilibrium theory of drops that are not much
bigger than the scale of roughness or heterogeneity (e.g. [65, 67, 68]). In these
situations, the Gibbs energy depends on a few spatial variables, and identification
of the most stable minimum is indeed a difficult task that may be done only by
three-dimensional numerical calculations for each individual case. In addition, if
the drops are also small enough to be in the nano-size range, then line tension
effects pose an additional challenge that needs to be addressed.
16 A. Marmur
References
1. T. Young, An essay on the cohesion of fluids, Phil. Trans. Roy. Soc. (London) 95, 65–87 (1805).
2. R. N. Wenzel, Resistance of solid surfaces to wetting by water, Ind. Eng. Chem. 28, 988–994
(1936).
3. A. B. D. Cassie and S. Baxter, Wettability of porous surfaces, Trans. Faraday Soc. 40, 546–551
(1944).
4. A. B. D. Cassie, Contact angles, Disc. Faraday Soc. 3, 11–16 (1948).
5. R. E. Johnson Jr. and R. H. Dettre, Wettability and contact angle, in: Surface and Colloid Science,
Vol. 2, E. Matijevic (Ed.), pp. 85–153. Wiley-Interscience, New York (1969).
6. C. Huh and S. G. Mason, Effects of surface roughness on wetting (theoretical), J. Colloid Interface
Sci. 60, 11–38 (1977).
7. R. J. Good, Contact angles and the surface free energy of solids, in: Surface and Colloid Science,
Vol. 11, R. J. Good and R. R. Stromberg (Eds), pp. 1–29. Plenum Press, New York (1979).
8. A. W. Neumann and R. J. Good, Techniques of measuring contact angles, in: Surface and Colloid
Science, Vol. 11, R. J. Good and R. R. Stromberg (Eds), pp. 31–91. Plenum Press, New York
(1979).
9. A. Marmur, Equilibrium and spreading of liquids on solid surfaces, Adv. Colloid Interface Sci. 19,
75–102 (1983).
10. J. F. Joanny and P. G. de Gennes, A model for contact angle hysteresis, J. Chem. Phys. 81, 552–562
(1984).
11. P. G. de Gennes, Wetting: statics and dynamics, Rev. Mod. Phys. 7, 827–863 (1985).
12. M. E. R. Shanahan, A simple analysis of local wetting hysteresis on a Wilhelmy plate, Surface
Interface Anal. 17, 489–495 (1991).
13. J. M. di Meglio, Contact angle hysteresis and interacting surface defects, Europhys. Lett. 17, 607–
612 (1992).
14. A. Marmur, Thermodynamic aspects of contact angle hysteresis, Adv. Colloid Interface Sci. 50,
121–141 (1994).
15. A. Marmur, Soft contact: measurement and interpretation of contact angles, Soft Matter 2, 12–17
(2006).
16. J. W. Gibbs, The Scientific Papers of J. Willard Gibbs, Vol. 1, p. 288. Dover Publications, New
York (1961).
17. A. Marmur, Contact angle and thin film equilibrium, J. Colloid Interface Sci. 148, 541–550 (1992).
18. V. S. Veselovski and V. N. Pertchov, Zh. Fiz. Khim. 8, 245–259 (1936) (in Russian).
19. B. A. Pethica, Contact angles, Rep. Prog. Appl. Chem. 46, 14–17 (1961).
20. B. A. Pethica, The contact angle equilibrium, J. Colloid Interface Sci. 62, 567–569 (1977).
21. A. Scheludko, B. V. Toshev and D. T. Bojadjiev, Attachment of particles to a liquid surface (cap-
illary theory of flotation), J. Chem. Soc. Faraday Trans. I 72, 2815–2828 (1976).
22. R. J. Good and M. N. Koo, The effect of drop size on contact angle, J. Colloid Interface Sci. 71,
283–292 (1979).
23. J. Drelich and J. D. Miller, The effect of solid surface heterogeneity and roughness on the contact
angle/drop (bubble) size relationship, J. Colloid Interface Sci. 164, 252–259 (1994).
24. D. Duncan, D. Li, J. Gaydos and A. W. Neumann, Correlation of line tension and solid–liquid
interfacial tension from the measurement of drop size dependence of contact angles, J. Colloid
Interface Sci. 169, 256–261 (1995).
25. J. Drelich, The significance and magnitude of the line tension in three-phase (solid–liquid–fluid)
systems, Colloids Surfaces A 116, 43–54 (1996).
26. A. Marmur, Line tension and the intrinsic contact angle, J. Colloid Interface Sci. 186, 462–466
(1997).
27. T. Pompe, A. Fery and S. Herminghaus, Measurement of contact line tension by analysis of the
three-phase boundary with nanometer resolution, J. Adhesion Sci. Technol. 13, 1155–1164 (1999).
28. K. W. Stöckelhuber, B. Radoev and H. J. Schulze, Some new observations on line tension of
microscopic droplets, Colloids Surfaces A 156, 323–333 (1999).
29. A. Marmur and B. Krasovitski, Line tension on curved surfaces: liquid drops on solid micro- and
nano-spheres, Langmuir 18, 8919–8923 (2002).
30. L. Boruvka and A. W. Neumann, Generalization of the classical theory of capillarity, J. Chem.
Phys. 66, 5464–5476 (1977).
31. G. Wolansky and A. Marmur, The actual contact angle on a heterogeneous rough surface in three
dimensions, Langmuir 14, 5292–5297 (1998).
32. P. S. Swain and R. Lipowsky, Contact angles on heterogeneous surfaces: a new look at Cassie’s
and Wenzel’s laws, Langmuir 14, 6772–6780 (1998).
33. B. He, J. Lee and N. A. Patankar, Contact angle hysteresis on rough hydrophobic surfaces, Col-
loids Surfaces A 248, 101–104 (2004).
34. W. Li and A. Amirfazli, A thermodynamic approach for determining the contact angle hysteresis
for superhydrophobic surfaces, J. Colloid Interface Sci. 292, 195–201 (2005).
35. S. Vedantam and M. V. Panchagnula, Constitutive modeling of contact angle hysteresis, J. Colloid
Interface Sci. 321, 393–400 (2008).
36. E. L. Decker and S. Garoff, Using vibrational noise to probe energy barriers producing contact
angle hysteresis, Langmuir 12, 2100–2110 (1996).
37. C. Andrieu, C. Sykes and F. Brochard, Average spreading parameter on heterogeneous surfaces,
Langmuir 10, 2077–2080 (1994).
38. T. S. Meiron, A. Marmur and I. S. Saguy, Contact angle measurement on rough surfaces, J. Colloid
Interface Sci. 274, 637–644 (2004).
39. G. Wolansky and A. Marmur, Apparent contact angles on rough surfaces: the Wenzel equation
revisited, Colloids Surfaces A 156, 381–388 (1999).
40. S. Brandon, N. Haimovich, E. Yeger and A. Marmur, Partial wetting of chemically patterned
surfaces: the effect of drop size, J. Colloid Interface Sci. 263, 237–243 (2003).
41. A. Marmur, Wetting on hydrophobic rough surfaces: to be heterogeneous or not to be?, Langmuir
19, 8343–8348 (2003).
42. A. Marmur, From hygrophilic to superhygrophobic: theoretical conditions for making high-
contact-angle surfaces from low-contact-angle materials, Langmuir 24, 7573–7579 (2008).
43. A. Marmur, The Lotus effect: super-hydrophobicity and metastability, Langmuir 20, 3517–3519
(2004).
44. N. A. Patankar, Mimicking the Lotus effect: influence of double roughness structures and slender
pillars, Langmuir 20, 8209–8213 (2004).
45. C. W. Extrand, Designing for optimum liquid repellency, Langmuir 22, 1711–1714 (2006).
46. Y. Ma, X. Cao, X. Feng, Y. Ma and H. Zou, Fabrication of super-hydrophobic film from PMMA
with intrinsic water contact angle below 90◦ , Polymer 48, 7455–7460 (2007).
47. J. Zhang, X. Gao and L. Jiang, Application of superhydrophobic edge effects in solving the liquid
outflow phenomena, Langmuir 23, 3230–3235 (2007).
48. W. Li and A. Amirfazli, Microtextured superhydrophobic surfaces: a thermodynamic analysis,
Adv. Colloid Interface Sci. 132, 51–68 (2007).
49. M. Nosonovsky and B. Bhushan, Hierarchical roughness optimization for biomimetic superhy-
drophobic surfaces, Ultramicroscopy 107, 969–979 (2007).
18 A. Marmur
50. D. Quéré, Wetting and roughness, Annu. Rev. Mater. Res. 38, 71–99 (2008).
51. L. Gao and T. J. McCarthy, How Wenzel and Cassie were wrong, Langmuir 23, 3762–3765 (2007).
52. A. Marmur, When Wenzel and Cassie are right: reconciling local and global considerations, Lang-
muir 25, 1277–1281 (2009).
53. L. A. Girifalco and R. J. Good, A theory for the estimation of surface and interfacial energies. I.
Derivation and application to interfacial tension, J. Phys. Chem. 61, 904–909 (1957).
54. F. M. Fowkes, Additivity of intermolecular forces at interfaces. I. Determination of the contribu-
tion to surface and interfacial tensions of dispersion forces in various liquids, J. Phys. Chem. 67,
2538–2541 (1963).
55. D. K. Owens and R. C. Wendt, Estimation of the surface free energy of polymers, J. Appl. Polym.
Sci. 13, 1741–1747 (1969).
56. C. J. van Oss, R. J. Good and R. J. Busscher, Estimation of the polar surface tension parameters
of glycerol and formamide, for use in contact angle measurements on polar solids, J. Dispersion
Sci. Technol. 11, 75–81 (1990).
57. C. J. van Oss, Acid–base interfacial interactions in aqueous media, Colloids Surfaces A 78, 1–49
(1993).
58. C. Della Volpe and S. J. Siboni, Some reflections on acid–base solid surface free energy theories,
J. Colloid Interface Sci. 195, 121–136 (1997).
59. S. Shalel-Levanon and A. Marmur, Validity and accuracy in evaluating surface tension of solids
by additive approaches, J. Colloid Interface Sci. 262, 489–499 and 268, 272 (2003).
60. I. V. Roisman, B. Prunet-Foch, C. Tropea and M. Vignes-Adler, Multiple drop impact onto a dry
solid substrate, J. Colloid Interface Sci. 256, 396–410 (2002).
61. A. Rozhkov, B. Prunet-Foch and M. Vignes-Adler, Impact of water drops on small targets, Phys.
Fluids 14, 3485–3501 (2002).
62. A. Marmur, Contact angle hysteresis on heterogeneous smooth surfaces: theoretical comparison
of the captive bubble and drop methods, Colloids Surfaces A 136, 209–215 (1998).
63. P. Lenz and R. Lipowsky, Morphological transitions of wetting layers on structured surfaces, Phys.
Rev. Lett. 80, 1920–1923 (1998).
64. H. Kusumaatmaja and J. M. Yeomans, Modeling contact angle hysteresis on chemically patterned
and superhydrophobic surfaces, Langmuir 23, 6019–6032 (2007).
65. S. Brandon, A. Wachs and A. Marmur, Simulated contact angle hysteresis of a three-dimensional
drop on a chemically heterogeneous surface: a numerical example, J. Colloid Interface Sci. 191,
110–116 (1997).
66. B. Krasovitski and A. Marmur, Drops down the hill: theoretical study of limiting contact angles
and the hysteresis range on a tilted plate, Langmuir 21, 3881–3885 (2005).
67. J. Y. Chung, J. P. Youngblood and C. M. Stafford, Anisotropic wetting on tunable micro-wrinkled
surfaces, Soft Matter 3, 1163–1169 (2007).
68. H. Kusumaatmaja, R. J. Vrancken, C. W. M. Bastiaansen and J. M. Yeomans, Anisotropic drop
morphologies on corrugated surfaces, Langmuir 24, 7299–7308 (2008).
Kinetics of Triple Line Motion during Evaporation
M. E. R. Shanahan a,∗ and K. Sefiane b

a
Université Bordeaux 1, Laboratoire de Mécanique Physique, CNRS UMR 5469,
351 Cours de la Libération, 33405 Talence Cedex, France
b
University of Edinburgh, School of Engineering and Electronics, Kings Buildings, Mayfield Road,
Edinburgh, EH9 3JL, UK. E-mail: k.sefiane@ed.ac.uk
Abstract
When a sessile drop evaporates in an unsaturated environment, it may change its geometry during mass loss
in a variety of ways, depending largely on the surface state of the solid in contact. Under some circumstances,
“pinning” of the wetting triple line (TL) to the solid surface may occur, leading to decrease of contact angle.
Subsequent “de-pinning” leads to relatively rapid TL recession and accompanying contact angle increase,
only to be followed by pinning again. Thus a “stick–slip” cycle is set up. Here we consider experimental
results of ethanol drops on Teflon® , and both apply and develop further the ideas presented some years ago
in a simple theoretical study of the possible mechanisms involved in stick–slip behaviour.
Keywords
Contact angle, drop evaporation, “stick–slip”, triple line, wetting hysteresis
1. Introduction
In recent years, the apparently trivial phenomenon of the evaporation of small drops
of liquids has proven not to be nearly so easy to comprehend satisfactorily as might
have been initially expected [1–12]. Considerations of heat and mass transfer are
relevant whilst wetting and capillary phenomena must be taken into account. Possi-
bly one of the most fascinating aspects discovered recently has been the existence
of flow phenomena within a drop associated with evaporation and manifested by
the drying of such suspensions as coffee, leaving stain rings of grounds near the
initial position of the wetting front [3, 4]. Advective flow to replace evaporated liq-
uid brings additional solid to deposit in the triple line region. Although such annular
stains have presumably been observed by human beings over centuries, their origin
has apparently only recently been investigated! (at least, successfully).
A perfunctory assessment of wetting combined with evaporation may lead one
to suppose that an axisymmetric sessile drop of liquid disappears due to a constant
* To whom correspondence should be addressed. E-mail: m.shanahan@lmp.u-bordeaux1.fr

20 M. E. R. Shanahan and K. Sefiane
evaporation flux over its exposed liquid/vapour surface, thus leading to the drop
shrinking with consistently diminishing contact radius, R, and height, h, whilst
maintaining a constant equilibrium value of contact angle, θo , at the wetting triple
line (TL), as given by the Young equation.
However, apart from an increased evaporation rate found near the triple line, at
least for small contact angles [3, 4, 6, 9], phenomena related to wetting hysteresis,
or deviation from the expected Young’s equilibrium contact angle, come into play
[1, 2]. The TL may effectively remain “stuck”, or “pinned” to the solid surface,
at least for some time. Contact radius, R, and therefore liquid/solid contact area,
πR 2 , remain constant during this stage of the evaporation, and, as a consequence,
the contact angle, θ , must decrease with time. In the case of a liquid containing a
suspended solid, this pinning may be permanent, virtually up until final drop dis-
appearance [3, 4], but for a pure liquid, it is more often temporary. In the latter
case, once the contact angle has decreased sufficiently, the TL may recede rela-
tively rapidly reducing liquid/solid contact area and increasing drop height, such
that the low contact angle may increase again, at least to some extent, tending to-
wards its assumed equilibrium value. The TL then again becomes pinned, and after
sufficient reduction of θ , following further evaporation, it may then jump again to a
smaller value of drop contact radius. The cycle is then repeated and we obtain the
phenomenon of “stick–slip” wetting hysteresis.
A simple theoretical argument to explain such observations, and based on excess
free energy as a criterion for TL movement, was put forward a few years ago [13].
Although simple, it appears to contain the basic ingredients to explain stick–slip
hysteresis. In this paper, we present experimental results which are coherent with
the previously proposed theory, and extend various aspects.
2. Simplified Considerations of Stick–Slip Evaporation

We shall commence with a brief summary of the main points of reference 13 in
order to present the proposed basics of stick–slip TL movement during drop evapo-
ration. Assuming gravitational flattening to be negligible, the drop is represented by
a spherical cap of contact angle θ and contact radius R (see Fig. 1). Drop volume,
V , and liquid/vapour surface area, A, are given by:
πR 3 (1 − cos θ )2 (2 + cos θ )
V= , (1)
3 sin3 θ
2πR 2
A= . (2)
(1 + cos θ )
The associated Gibbs free energy, G, is given to within an additive constant by:

2
G = γ A + πR (γSL − γSV ) = γ πR
2 2
− cos θo , (3)
(1 + cos θ )
where use of Young’s equation has been made to replace (γSL − γSV ) by −γ cos θo ,
θo representing the equilibrium contact angle, γ , γSL and γSV , being, respectively,
Kinetics of Triple Line Motion during Evaporation 21
(a)
(b)
Figure 1. Sketch representing evaporation leading to stick–slip motion of triple line (TL) of a sessile
drop. (a) Initial evaporation leads to decreased height, h, and contact angle, θ , at constant contact
radius, R, due to TL pinning. (b) When the TL de-pins, it recedes leading to increased h and θ , the
latter tending towards its equilibrium value. The process then repeats itself with further evaporation,
giving successive periods of “stick” and “slip”.
liquid surface free energy, and solid/liquid and solid/vapour interfacial free ener-
gies.
For a given volume, V , we have an equilibrium contact radius, Ro , corresponding
to contact angle θo , but if the drop is out of thermodynamic equilibrium with, say,
contact radius R = Ro + δR (and therefore contact angle θo − δθ ), its excess free
energy, δG, is given by:

dG (δR)2 d2 G
δG = G(R) − G(Ro ) = δR + + O[(δR)3 ]. (4)
dR R=Ro 2 dR 2 R=Ro
Since [dG/dR]R=Ro = 0, as Ro is the equilibrium value, we may evaluate the ex-

cess free energy per unit length of TL, δ G̃ = δG/(2πR) to O[(δR)2 ] from equations
(1)–(4):
γ sin2 θo (2 + cos θo )(δR)2

δ G̃ = . (5)
2R
Alternatively, we may use a Taylor expansion in powers of δθ , rather than δR as in

equation (4), to obtain:
γ R(δθ )2
δ G̃ = . (6)
2(2 + cos θo )
During evaporation, δ G̃ increases for a pinned TL (this is clearer from equation (6)
than from equation (5), since δθ physically increases during pinned evaporation),
until it just exceeds the potential energy barrier, U , physical or chemical, for de-
pinning. At this moment, the TL may de-pin and recede, leading to a smaller value
of R, assumed to be a new equilibrium value of Ro corresponding to the recently
diminished drop volume. The new Ro will be smaller than the previous one, but
since the height of the drop, h, will increase during the (assumed rapid) TL motion,
the original θo may be regained, at least in this simplified model. δ G̃ would again
become zero. As discussed below, whether the original value of θo would be ex-
pected to be attained after the initial de-pinning is doubtful. It should be noted that
a further simplification assumed here is that TL motion occurs over a sufficiently
short period for concomitant mass loss by evaporation to be considered negligible.
Also, explicit time dependence is neglected in this preliminary model.
3. Experimental
The experimental equipment used to carry out the investigation was a First Ten
Ångstroms FTA200 dynamic contact angle and surface tension analyzer used for
drop shape analysis. This setup allows the deposition of droplets on substrates at
controlled rate and volume, as well as performing subsequent analysis of data ob-
tained. The FTA package can be used to evaluate the contact angle, θ , volume, V ,
height, h, and diameter, 2R, of axisymmetric sessile drops. An automatic injection
pump is used to deposit a calibrated drop on the substrate. The optical technique
uses a projection method to determine the drop profile (contact angle, base, height
and volume) vs time. As the deposited droplet evaporates, its profile can be analysed
using the FTA200 drop shape analysis software. The drop profile is analysed based
on a 2D view assuming an axisymmetric shape. Optical measurement is calibrated
and permits precision of the drop angle of better than 0.5 degrees, and base diame-
ter, height and drop volume to less than 2%, on the assumption of true axisymmetry.
For the present experiments, the substrate was a thin (thickness 1 mm) plate
of poly(tetrafluoroethylene), Teflon® , heated by a resistive heater from below. The
heater was connected to a temperature Proportional, Integral, Derivative (PID) con-
troller. Plate temperature was controlled to within ±0.5◦ C. Surface temperature was
measured using an embedded thermocouple, which was centered within the plate.
It should be noted that all temperatures reported in this study correspond to those
measured by the thermocouple. Surface analysis using a ZYGO profilometer re-
vealed that the substrate was randomly rough (Rrms = 148 nm). Ethanol droplets of
initial volume of 10 µl were deposited on the substrate at a controlled rate using a

step motor syringe.
4. Results and Discussion

Figure 2(a) shows typical examples of the evolution of contact radius, R, with time,
t, for ethanol drops of volume ca. 10 µl on Teflon® , at various temperatures from
20◦ C to 60◦ C. The stick–slip behaviour can be clearly seen, after the initial rela-
tively long stick period. Behaviour is similar at all temperatures, allowing for more
rapid kinetics at higher temperatures. Figure 2(b) shows a specific case correspond-
ing to behaviour at 60◦ C. Both contact radius, R, and contact angle, θ , are given.
In this case, the first period of evaporation lasting ca. 10 seconds at constant R,
but decreasing θ , corresponds to initial TL pinning. This is the primary “stick” pe-
riod during which the contact angle decreases by ca. 13◦ . It seems different from
the subsequent cycles and we shall return to this point below. The primary pinning
is followed by a “slip” phase of decreasing R during which θ initially stagnates
but then increases, albeit slightly (by ca. 2◦ ). Henceforth, the observed cycles are
smaller, at least as far as θ is concerned, and somewhat erratic, but typically θ may
be seen to vary by ca. 1–2◦ over a given cycle, whilst R either pins or decreases
monotonically, depending on the part of the cycle and in agreement with the sim-
ple model [13] presented succinctly above. Over a period of several cycles, θ tends
to increase its average value. However, this is probably not significant since the
Teflon® used as the substrate was quite rough, with a random surface pattern, and
so it is not too surprising that the stick–slip cycles are somewhat erratic. Never-
theless, the essential features of the stick–slip model are present. Similar overall
behaviour was found at the other test temperatures.
4.1. Time Dependence of Stick–Slip
The model so far has no explicit time dependence, but we may attempt to incorpo-
rate this. Let us first consider the pinning phase. Under conditions of constant R,
overall volume evaporation rate from the drop is given by:
dV dV dθ πR 3 dθ
= · = , (7)
dt dθ dt (1 + cos θ )2 dt
from equation (1). In order to estimate the pinning period, t1 , we integrate equation
(7) between θ = θo and θ = θo − δθ , where θo − δθ is the threshold contact angle
at which the TL starts to recede. (We assume that pinning commences at the equi-
librium value of contact angle.) Given the slight variability of θ during pinning, the
time t1 , is given approximately by:
−πR 3 δθ
t1 = , (8)
(1 + cos θo )2 dV /dt
where the negative sign allows for the fact that dV /dt represents mass loss. Un-
der the given conditions, approximate values of the various parameters are R =
(a)
(b)
Figure 2. Experimental evidence of stick–slip behaviour of ethanol on Teflon® : (a) contact radius,
R, vs time, t, at various substrate temperatures from 20◦ C to 60◦ C, (b) contact angle, θ , and contact
radius, R, vs time, t, at 60◦ C.
ca. 2.5 mm, θo = ca. 25◦ and δθ = ca. 2◦ . A typical value of dV /dt has been calcu-
lated from evolving drop shape and is of the order of −0.3 µl/s. These values lead to
an estimate for t1 of the order of 2 seconds, which is consistent with observations.
Given the variability of the results, a more precise estimate has no merit.
We also consider the typical time of slip, t2 . It is probable that the hydrodynamic
theory of wetting will apply [14], given that re-condensation [15, 16] is unlikely
under the experimental conditions considered here. The receding nature of the TL
during the slip period is really a case of de-wetting. A well-established treatment
of the phenomenon using the lubrication approximation leads, in the case of de-
wetting, to the equation [14]:
dR γ θ (t)(θo2 − θ 2 (t))
∼− , (9)
dt 6ηL
where θ (t) is time-dependent contact angle, η liquid viscosity and L the logarithm
of the ratio of a macroscopic distance (ca. drop radius) and a microscopic cut-off
length to avoid divergence of the flow-field near the contact line (L is estimated
typically to be of the order of 10–12, and approximately constant because of its
logarithmic nature). Since θ (t) varies only little during slip, we may reasonably
rewrite equation (9) as:
dR γ θ 2 (θo − θ (t))
∼− o , (10)
dt 3ηL
which is readily integrated with the constant volume condition, V ∼ πR 3 θ/4, to
give:
θ (t) ∼ θo − δθ exp(−t/τ ), (11)
where the time constant, τ , is given by:

ηL 4V 1/3
τ= . (12)
γ πθo10
With approximate values of η = ca. 10−3 Pa s, L = ca. 10, γ = ca. 25 mJ/m2 ,
V = ca. 10−8 m3 , and θo = ca. 30◦ , we find τ = ca. 10−2 seconds, and therefore,
assuming the asymptotic approach to θo to be virtually finished after 3 time con-
stants, the overall slip time, t2 , is typically of the order of 1/30 second, which is
very short.
Since slip is a fairly rapid process, one may wonder whether inertial forces play
a significant role. Let us consider the initiation of slip. The initial capillary force,
fc , per unit length of TL is given by:

fc = γ cos(θo − δθ ) − cos θo ≈ γ θo δθ. (13)
From equations (10) and (11), we have:
d2 R γ θo2 δθ
∼ . (14)
dt 2 3ηLτ
We take a segment of drop (like a slice of cake) subtending an angle corresponding
to unit length of TL, whose mass, m, is given by πθ R 3 /4ρ/(2πR) = θ R 2 ρ/8,
where ρ is liquid density. For axisymmetric TL motion, the corresponding inertial

force, fin , is then given by:
d2 R kγ θo3 ρR 2 δθ
fin = m ≈ , (15)
dt 2 24ηLτ
where k is a numerical constant, k < 1, and corresponds to a correction for the fact
that during TL motion, it is not the entire segment which moves: there is gradient
from virtually zero motion at the centre of the drop to full motion at R. Finally, the
ratio fin /fc can be expressed as:
fin kρθo2 R 2
≈ . (16)
fc 24ηLτ
Constant k is expected to be of order 0.5. With values given above and ρ =
ca. 800 kg/m3 , we then find a value of fin /fc of ca. 0.3, or less. Thus the inertial
force will have only little impact on the capillary force, and the order of magnitude
of τ of 10−2 seconds remains unchanged.
Examination of the data suggests that in some cases, the slip time is effectively
very short, thus not precisely measurable. However, in others, the overall slip time
is much longer, suggesting a more gradual process, perhaps involving metastable
energy barriers, or “micro-sticking”. Again, the roughness of the Teflon® surface
being random, variation in behaviour may be expected, especially as what we are
observing is an average behaviour around the entire drop contact line. A possible
explanation is suggested. If slip conditions are attained at a given small length of
the TL, δL, whereas neighbouring zones on either side are not yet at their slip
threshold, due to the random nature of the solid surface, the local, lateral, elastic
response of the TL [14] will hinder the slip of δL. Effectively, the neighbouring
zones will supply a braking force working against the capillary imbalance tending
to move δL and thus slow down slip. The overall time of slip will effectively involve
the sum of all the contributions from the various TL sections, which do not attain
threshold conditions simultaneously. We note also that under these conditions, some
evaporation will occur and perhaps modify TL movement.
4.2. Free Energy Barriers
The above theory invokes an excess free energy, δ G̃, which is taken with reference
to a supposed equilibrium state with contact angle θo . In the simplified model, it is
assumed that evaporation with pinning leads to a monotonic increase in δ G̃ associ-
ated with decrease in θ , given by equations (5) or (6), until de-pinning occurs at a
threshold value of θ . This is then followed by a rapidly receding motion of the TL
until θo is again attained, with concomitant reduction of δ G̃.
In Fig. 3 is presented the evolution of normalised excess free energy, δ G̃/γ , with
normalised time, t/tf (tf is the overall drop lifetime), for various test temperatures,
as evaluated from equation (6) and using the assumed value of θo (corresponding
to the initial contact angle, at t = 0) and the actual drop volume at t (obtained from
Figure 3. Normalised excess free energy, δ G̃/γ , vs normalised time, t/tf , where tf is total drop life-
time, at various temperatures.
θ (t) and R(t)). The zigzag shape of the curves clearly points to the existence of po-
tential energy barriers to TL de-pinning. Peaks of δ G̃/γ correspond to threshold θ ,
attained just preceding the onset of slip. The qualitative features expected from the
description of Section 2 are present, yet Fig. 3 shows a primary maximum of δ G̃/γ
just preceding the first de-pinning, and henceforth continuously smaller maxima for
subsequent slip thresholds. In addition, δ G̃/γ , does not return to zero.
In order to explain this observation, we postulate that there are two values of
θo , depending on whether the solid surface has previously been in contact with
ethanol or not. The initial value of θo , denoted θo,1 , then corresponds to that of the
liquid when the external solid surface is dry (corresponding to γSV ). However, for
subsequent cycles, the newly exposed solid surface has previously been in contact
with ethanol, and therefore some liquid may remain, either due to adsorption or,
more likely, slight physical retention within the surface roughness. Thus a second
value of θo , denoted θo,2 , is more appropriate for later behaviour. The value of θo,2
has been estimated from θ at the onset of slip at the second cycle. An example is
shown in Fig. 4.
The values of δ G̃ vs t/tf , at 60◦ C, following use of the two values of θo , viz.
θo,1 and θo,2 , can be seen in Fig. 4. Also shown are the corresponding θ . Note that
positive or negative deviations of θ from the expected equilibrium lead to positive
excess free energy changes, because of the dependence on (δθ )2 (equation (6)).
A primary maximum of δ G̃ is present, followed by secondary maxima. The pri-
mary maximum is markedly greater than all of the secondary peaks: the relative
Figure 4. Contact angle, θ , and excess free energy, δ G̃, vs time, t, where the initial value of θo,1 has
been adopted for the first cycle, and the maximal value of the second cycle, θo,2 , for the second and
subsequent cycles. (θo,1 and θo,2 are indicated in the figure.) The inset, on a different scale, allows the
totality of the first energy peak to be seen.
magnitudes can be seen in the inset of Fig. 4. After the primary peak, although the
first secondary peak is considerably smaller than the primary peak, it is still greater
than most of the following peaks. The return to a zero value of δ G̃ at the appropriate
point in each cycle is quite convincing. We conclude that the simple theory based
on free energy reasonably explains most of the observed features, if we allow for a
difference between an initial and then a secondary value for θo .
Considering Fig. 4, it can be seen that δ G̃ is typically of the order of 10−7 N for
the first cycle based on θo,1 , and 10−8 N for subsequent cycles based on θo,2 . The
units of δ G̃ are the same as those of line tension and clearly there is a similarity
between the two quantities. They may even possibly be construed as manifestations
of the same effect. Theoretical predictions of values of line tension are usually in
the range of 10−12 N to 10−10 N [17, 18]. However, certain experimental studies
suggest values nearer to 10−6 N [17]. Our values of δ G̃ lie between experimental
and predicted values of line tension. Is this coincidental, or is line tension intimately
connected with the wetting hysteresis free energy barrier? We open the debate.
4.3. Slip Distance
In reference [13], it was postulated that δR scales with R 1/2 : a result of equating
U to (the present) expression (5) and rearrangement. It is of interest to investigate
the validity of this postulate. Figure 5 represents δR vs R 1/2 data obtained from
Figure 5. Slip distance, δR, vs square root of drop contact radius, R 1/2 , from the data of Fig. 2(b).
Fig. 2(b). The data are restricted to early stick slip cycles, as indicated in Fig. 2(b),
since later behaviour is difficult to analyse, the randomness of the surface reduc-
ing the clarity of the jumps. There is an acceptably linear relation, but the line
clearly does not pass through the origin. This may be related to the fact that the
simple model assumes a single value, U , for free energy barrier, whereas in fact
there is variability, as suggested above to explain the initially high value of δ G̃.
(Although the use of two values of θo could possibly be involved, an explanation is
not presently available.)
Figure 2(b) clearly shows that towards the end of drop life, contact angle steadily
decreases to zero. Instead of the drop getting smaller and smaller, yet following the
stick–slip cycles with ever decreasing δR (δR ∼ R 1/2 : see above), and oscillating θ ,
we attain a phase in which θ tends to zero. This may be explained by an insufficient
capillary excess free energy, δ G̃, to overcome the hysteretic barrier, U , as suggested
in reference [13]. Consider the free energy, Gc , of equation (3) corresponding to
a critical value of drop radius, Rc . Gc may be considered to be the excess free
energy compared to that of the solid when the drop has totally vanished (assuming
no adsorption). As a consequence, this is the maximum of free energy that can be
supplied in order to surmount the hysteretic barrier, assumed to be constant around
the drop periphery. Thus:

2
Gc = γ πRc 2
− cos θo < 2πRc U. (17)
(1 + cos θ )
Henceforth, the drop will have insufficient energy in order to engage in slip be-
haviour. It will remain pinned with evaporation leading to contact angle, θ , and
drop height, h, tending to zero at final disappearance. With a dependence on Rc2 for
available free energy, and a dependence on Rc for the overall barrier energy, it can
be seen that the threshold value occurs at some small limiting value of contact ra-
dius. Our results suggest final TL pinning at R = Rc = ca. 1 mm, and thus a rough
estimate of U is given by:

1 cos θo
U ∼ γ Rc − . (18)
(1 + cos θo ) 2
For our data, this suggests a potential energy barrier of the order of 10−6 N, which
is comparable to some figures quoted for experimentally obtained values of line
tension [17] (also see above). Are line tension and our free energy barrier directly
related?
5. Conclusion
The wetting hysteresis, during evaporation, of sessile drops of ethanol on a ran-
domly rough Teflon® surface has been studied. Stick–slip behaviour has been
clearly identified. A period during which the wetting triple line is pinned, main-
taining constant drop contact radius, but involving the decrease of drop height and
contact angle, is followed by a phase of receding wetting front accompanied by
increasing contact angle and height. The cycle then repeats itself.
A comparison with a simple model using excess free energy as a criterion for
stick–slip behaviour has been made. The basic features of the model and exper-
iment correlate well. Time dependence has been added to the model and we ex-
plain satisfactorily the order of magnitude of the pinning time. However, there is
more variability of the slip time. This has been tentatively attributed to the non-
simultaneous attainment of the slip threshold around the drop periphery: the slip
period is then “spread out”.
Evaluation of excess free energy barriers for slip to occur suggests a difference
between primary and subsequent cycles, possibly related to the retention of a slight
amount of liquid on the exposed solid after wetting line recession.
Slip distances diminish approximately with the square root of contact radius, but
the agreement between model and observations is feeble. However, analysis of final
drop disappearance behaviour suggests a hysteretic energy barrier of ca. 10−6 N,
comparable with reported values of line tension.
References
1. C. Bourgès and M. E. R. Shanahan, Compt. Rendus Acad. Sci. (Paris) 316 (II), 311 (1993).
2. C. Bourgès and M. E. R. Shanahan, Langmuir 11, 2820 (1995).
3. R. D. Deegan, O. Bakajin, T. F. Dupont, G. Huber, S. R. Nagel and T. A. Witten, Nature 389, 827
(1997).
4. R. D. Deegan, O. Bakajin, T. F. Dupont, G. Huber, S. R. Nagel and T. A. Witten, Phys. Rev. E 62,
756 (2000).
5. H. Y. Erbil and M. Dogan, Langmuir 16, 9267 (2000).

6. H. Hu and R. G. Larson, J. Phys. Chem. B 106, 1334 (2002).
7. M. E. R. Shanahan, Langmuir 18, 7763 (2002).
8. Y. O. Popov and T. A. Witten, Phys. Rev. E 68, 036306 (2003).
9. K. Sefiane, L. Tadrist and M. Douglas, Int. J. Heat Mass Transfer 46, 4527 (2003).
10. C. Poulard, G. Guéna and A. M. Cazabat, J. Phys.: Condens. Matter 17, S4213 (2005).
11. H. Hu and R. G. Larson, J. Phys. Chem. B 110, 7090 (2006).
12. F. Girard, M. Antoni, S. Faure and A. Steinchen, Langmuir 22, 11085 (2006).
14. P. G. de Gennes, Rev. Mod. Phys. 57, 827 (1985).
15. P. C. Wayner, Langmuir 9, 294 (1993).
17. J. Drelich, Colloids Surfaces A 116, 43 (1996).
18. J. Y. Wang, S. Betelu and B. M. Law, Phys. Rev. E 63, 031601 (2001).
What Can We Learn from the Vibration of Drops Deposited
on Rough Surfaces? Wetting Transitions Occurring on
Rough Surfaces
Edward Bormashenko ∗ , Gene Whyman and Roman Pogreb

Ariel University Center of Samaria, Applied Physics Faculty, Ariel 40700, Israel
Abstract
Wetting transitions were studied with vertically and horizontally-vibrated drops on various artificial and
natural rough substrates. Alternative pathways of wetting transitions were observed. The model of wetting
transition is presented. Multiple minima of the Gibbs free energy of a drop deposited on a rough surface
explain the alternative pathways of wetting transitions. It is demonstrated that the wetting transition occurs
when the constant force resulting from vibration, Laplace and hydrostatic pressures acts on the triple line. It
is shown that the final wetting state is mainly the Cassie impregnating wetting state with water penetrating
the pores in the outer vicinity of the droplet or the Wenzel state with water inside the pores under the droplet
whereas the substrate ahead the drop is dry. For horizontally vibrated drops, the resonance character of
wetting transition is noteworthy: the threshold amplitude necessary for transition is decreased dramatically
in the vicinity of the eigenfrequencies of surface waves.
Keywords
Wetting transition, rough surfaces, vibrated drops, triple line, resonance
1. Introduction
Wetting of rough surfaces is a ubiquitous phenomenon in various natural and tech-
nological processes including the wetting of soil, printing, painting, textile impreg-
nation, etc. Wetting phenomena on rough interfaces are diverse: from the famous
“lotus-like” water repellency to superhydrophilicity [1–5]. The spreading of liquid
on a rough surface is controlled by the interplay of hydrodynamic and adhesion
effects, which are governed by long-range forces [1, 2]; thus it is clear that a pro-
found understanding of the dynamic wetting of rough surfaces is a complicated and
challenging task. The static situation is more transparent and easy for analysis, and
a clear understanding has been achieved in its analysis during the two past decades
[3–6]. The main static characteristic of the wetting properties of rough surfaces is
* To whom correspondence should be addressed. E-mail: edward@ariel.ac.il

34 E. Bormashenko et al.
(A) (B) (C)
Figure 1. Wetting regimes on rough surfaces.
the so-called apparent contact angle (APCA) [3–6]. It is well known that the appar-
ent contact angles are different from those predicted by the Young equation [6–12].
Two main hypotheses were proposed in order to explain this discrepancy, namely
the Wenzel and Cassie wetting models [6–10]. According to the traditional Cassie
model, air can remain trapped under the drop, forming “air pockets” (Fig. 1A).
Thus, the hydrophobicity of the surface is enhanced because the drop sits partially
on air. On the other hand, according to the Wenzel model the roughness increases
the area of a solid surface, which also geometrically modifies the wetting properties
of this surface (see Fig. 1B). In spite of the fact that the Wenzel and Cassie models
were developed more than 50 years ago, a rigorous derivation of the Wenzel and
Cassie equations based on the virtual work principle was carried out only recently
by Bico et al. [13], and also by the present authors in the recent work [14] based on
minimization of the droplet free energy.
It is a widespread delusion that the Cassie air trapping state and the Wenzel
wetting one exhaust all observable wetting situations. Actually, there is one more
option, namely the Cassie impregnating wetting regime when a liquid film pen-
etrates a texture ahead the drop (pores are filled with liquid and solid “islands”
ahead the drop are dry, as depicted in Fig. 1C, see also Ref. [1]). We demonstrate
here that the Cassie impregnating wetting regime occurs actually in various experi-
mental situations.
Another important characteristic of the solid/liquid interface is contact angle hys-
teresis [15–24]. As a rule, the experimentally established apparent contact angle θ
is confined within a certain range θrec < θ < θadv , where θadv and θrec are the so-
called advancing and receding contact angles. The difference between θadv and θrec
is called the contact angle hysteresis (CAH) [16–24]. Contact angle hysteresis is
caused by pinning forces on the outline of the droplet, called the triple line. This
pinning may be due to physical or chemical heterogeneities of the substrate, or it
may be due to long-range (van der Waals) forces when the substrate is atomically
flat. It has to be emphasized that the phenomenon of contact angle hysteresis is not
completely understood, and very contradictory theoretical approaches have been
proposed recently for explaining and predicting CAH [25]. The phenomenon of
CAH is not understood to a full extent even for atomically flat surfaces; it is quite
clear that on the rough surfaces the situation is more complicated.
Wetting Transitions Occurring on Rough Surfaces 35
On the same rough surface both the Wenzel and Cassie wetting regimes may
co-exist. It was also demonstrated that under certain external stimuli, such as pres-
sure, vibration, impact or electric field, the Cassie air trapping wetting state could
be converted into the Wenzel one [4, 5, 26–35]. It is now well accepted that the
Cassie air trapping wetting regime corresponds to a higher energetic state, and
the Cassie–Wenzel transition is irreversible [4–6]. However, the kinetics of the
transition remains unclear. Ishino and Okumura suggested that the Cassie–Wenzel
transition occurred via a nucleation mechanism starting from the drop center [36].
On the other hand, a recent investigation performed by the authors with polymer
honeycomb surfaces demonstrates that the Cassie–Wenzel transition is more likely
due to the displacement of the triple line under an external stimulus [29, 30]. Ini-
tially, the idea that CAH is governed by effects occurring in the nearest vicinity
of the triple line belongs to Joanny and de Gennes [37]. In the present study per-
formed with a variety of organic and inorganic rough surfaces under both vertical
and horizontal vibrations we show that wetting transition is indeed due to the dis-
placement of the triple line. We also demonstrate that the wetting transition under
vibration proceeds from the Cassie air trapping state to the Cassie impregnating or
the Wenzel wetting states.
Moreover, the investigation performed with vibrated drops supplied one more
quantitative parameter (in addition to the apparent contact angle and contact angle
hysteresis) featuring the rough solid/liquid interface, namely the threshold force
acting on unit length of a triple line that is sufficient for a wetting transition. This
force, which quantitatively describes the pinning of the triple line, is responsible for
a variety of phenomena, including the contact angle hysteresis.
In the first section we report and discuss the results obtained with vertically vi-
brated drops at the frequency which is far from the eigenfrequency of the drop, in
the second part we discuss specific resonance effects which take place under hor-
izontal vibrations in the vicinity of eigenfrequencies of surface waves. In the both
cases the wetting transition is governed by the displacement of the triple line.
2. Vertical Vibrations
2.1. Experimental
Five kinds of rough surfaces depicted in Fig. 2A–E were studied. The surfaces are
called A, B, C, D and E respectively. Surface A was a polystyrene (PS) honey-
comb surface prepared by the fast dip-coating method described in our previous
papers [38, 39]. When the substrate is dip-coated, it is pulled vertically with a con-
stant speed from a polymer solution. The liquid film runs out from the polymer
solution, adheres to the substrate surface and solidifies during evaporation of the
solvent under controlled humidity. A Scanning Electron Microscopy (SEM) image
of surface A is depicted in Fig. 2A.
Surface B was produced in a two-stage process. In the first stage, a poly-
etherimide (PEI) honeycomb template was prepared using the dip-coating method
(A)
(B)
Figure 2. (A) SEM image of surface A. The PS honeycomb structure (the scale bar is 10 µm). (B) SEM
image of surface B. The PEI relief coated with nickel (the scale bar is 5 µm). (C) SEM image of
surface C. LDPE embossed with a crimped textile surface (the scale bar is 500 µm). (D) SEM image of
surface D. Polyester fabric textile (the scale bar is 500 µm). (E) SEM image of surface E. The surface
of a pigeon feather (the scale bar is 10 µm). Large barbs, like one in the center, form a large-scale
roughness while small barbules form a small-scale roughness.
described in detail in our papers [38, 39]. Polypropylene (PP) substrates with
a thickness of 30 µm were coated with a solution containing 5% PEI, 84.5%
dichloromethane (CH2 Cl2 ), and 10.5% chloroform (CHCl3 ) (by weight). Dip-
coating was performed at a pull speed of 22 cm/min, at room temperature and
relative humidity of 40–50%. Dichloromethane was supplied by Frutarom, chlo-
roform was supplied by Bio Lab Ltd., and PEI was obtained from the Westlake
Plastics Company.
In the second stage, this honeycomb PEI layer was coated with nickel using
(C)
(D)
(E)
Figure 2. (Continued.)
an electroless deposition technique. Electroless metal plating is a catalytic process

based on the use of a chemical reducing agent that promotes chemical reduction of
metal in solution and depositing the metal on the substrate surface. The technique
includes a sensitization step (treatment of the substrate with a mixture of SnCl2 and
HCl solutions for 2 min), an activation step (gold sputtering for 10 s, creating the
catalytic sites), and a Ni plating step (at 80◦ C for 2 min) according to the reaction
Ni2+ + 2H2 PO2 − + 2H2 O = Ni + 2H3 PO3 − + 2H+ .
Finally, surface B is formed by Ni layer with micrometric holes (Fig. 2B). We used
a BUEHLER® EdgeMet® Kit for surface coating by electroless deposition.
Surface C was prepared as follows: a low-density polyethylene (LDPE) film with
a thickness of 50 ± 10 µm was covered with a crimped textile surface; this sand-
wiched system was exposed to hot pressing at 95◦ C. The polyethylene substrate was
softened under the heating and replicated the textile morphology; and after cooling,
the substrate takes the form presented in Fig. 2C.
Surface D was polyester fabric textile which was cleaned from contamination
with ethyl alcohol and then with distilled water. An SEM image of surface D is
depicted in Fig. 2D.
Surface E was a natural rough object, namely a Feral Rock pigeon feather (pen-
nae), the wetting of which has been studied recently by the authors [40]. We re-
vealed that the water drop, when deposited on the feather, are supported by the
network formed by barbs and barbules depicted in Fig. 2E, demonstrating the pro-
nounced Cassie air trapping wetting regime [40].
The fraction of the liquid–solid interface in the Cassie model was established
with image processing technique, i.e. it was supposed that this interface corre-
sponded to the light part of images in Fig. 2A–E. The roughness parameter was
established from Atomic Force Microscopy (AFM) study for reliefs A, B, and with
optical microscopy for surfaces C and D. It was also supposed that the planar part
of the reliefs was ideally flat.
Droplets of distilled water of 2.5–50 µl volume were carefully deposited on all
types of rough substrates with a precise micro-dosing syringe. The substrate with
the drop was bound up with the moving part of a vibration generator, Frederiksen
2185.00, driven by a synthesized function generator, DS 345. A horizontal helium-
neon laser beam illuminated the droplet profile and gave its enlarged image on the
screen using a system of lenses (see Fig. 3). The contact angles were measured from
these projected images of the drop. Measurements were made on both sides of the
drop and averaged. A series of 10 experiments were carried out for every substrate.
The same optical projection technique was used to determine the contact angles
on flat surfaces (experimentally established angles are close but not equal to Young
ones; we will ignore this distinction below). On flat substrates made of extruded
polymers we established contact angles in a good agreement with the data supplied
by other authors, however on the nickel surface obtained by the above-outlined
Figure 3. Schematic of the vibration experiment.
process we obtained a surprisingly high contact angle of 70◦ in contrast to values

supplied by other groups (this will be discussed below in detail).
The vibration frequency used was 36 Hz; this frequency is sufficiently far from
eigenfrequencies of surface waves, leading to the varied wetting transitions dis-
cussed in our recent paper [41]. The amplitude was changed from 16 µm to 1.1 mm.
The drop was exposed to vertical vibration of increasing amplitude until the first
wetting transition took place, and the amplitude corresponding to this transition
was fixed. Thus, an intermediate (transition) stable wetting state was established.
Then the amplitude increased and the drop progressed to its final stable wetting
state.
2.2. Vertical Vibrations: Results and Discussion
2.2.1. Wetting States: Experimental Data and Their Interpretation
The values of the initial apparent contact angle (APCA) measured on the above
mentioned substrates in the rest state, θin , are summarized in Table 1. After exposing
water droplets to vibration of certain amplitude they undergo a wetting transition
(Fig. 4). An intermediate (transitional) contact angle corresponding to the wetting
transition onset can be measured, and it is denoted as θtr . A further increase in
the amplitude of vibration leads to a decrease in the APCA towards its final value
denoted as θfin (see Table 1). The transition and final wetting states are stable ones.
It could be easily recognized that θfin is much smaller than θtr . The decrease in
APCA was observed on surfaces A–D but not on surface E (the pigeon feather). The
increase in vibration amplitude could not lead to a wetting transition on surface E
and caused atomization of the drop. The contact angle hysteresis could be calculated
from the data presented in the Table 1 as θin − θfin .
It is reasonable to suggest that all three wetting states, namely, the initial hy-
drophobic state, intermediate and final states were observed in these experiments
are characterized by APCAs that decrease in the following order, θin > θtr > θfin .
The measured angles θin are listed in Table 1, together with calculated ones accord-
Table 1.
Measured and calculated angles (in degrees) corresponding to different wetting states
Sur- Angles Rough- Fraction Expl. Air Expl. Wenzel, Expl. Impregn. Critical
face on flat ness, of S–L initial trap- transi- calc
θWenzel final Cassie, contact
surface, f inter- APCA, ping tion calc
APCA θfin angle
θE ∗ face, θin Cassie, APCA, θfin θc
S calc
θCassie θtr
A 84 2 0.62 103 ± 2 110 98 ± 1.5 82 59 ± 2 65 74

B 70 1.6 0.47 94 ± 2 112 87 ± 1.5 57 58 ± 2 46 63
C 104 1.5 0.24 114 ± 3 145 98 ± 2 112 41 ± 2 46 55
D 80 3.1 0.36 118 ± 3 125 92 ± 2 57 40 ± 2 45 77
* The contact angles on flat surfaces are taken from Ref. [42] and our direct measurements.
Figure 4. Initial (A), transitional (B) and final (C) states of the drop.
ing to the well-known Cassie and Wenzel equations:

calc
cos θCassie = −1 + S (cos θE + 1), (1)
calc
cos θWenzel = f cos θE , (2)
calc
where θCassie calc
and θWenzel are the calculated apparent contact angles, S is the
fraction of the liquid–solid interface, and 1 − S is the fraction of the air-liquid
interface under the droplet, i.e., of the substrate area occupied by pores, and f is
the roughness of the surface, i.e., the ratio between the actual surface and its hori-
zontal projection. As usual, the Young contact angle θE is defined as
γSA − γSL
cos θE = , (3)
γ
where γSA , γSL and γ are the interfacial tensions for solid–air, solid–liquid and
liquid–air interfaces, respectively. It could easily be seen from Table 1 that the initial
wetting state for surfaces A and D is the Cassie air trapping state (compare 5th and
6th columns), whereas the initial wetting state on surface C is the Wenzel state
(compare 5th and 8th columns). The initial wetting state on surface E is the Cassie
air trapping state, as was already established in our previous work [40]. It could
readily be seen that the Cassie model overestimates the apparent contact angles.
This may be explained by the partial filling of pores with water [6].
Surface B needs separate discussion. The experimentally established angle on
a flat surface was 70◦ in contrast to results reported by Wang et al. (Ref. [43],
θE ≈ 20◦ ). We suggest that the relatively high value of the angle on a flat surface
results from electroless metal plating giving rise to an investigated substrate, which
is far less in purity than Ni substrates reported in Ref. [43]. However, only this high
value should be taken as an angle on a flat surface, because it is this electroless
plating process that was used for producing our honeycomb Ni surface. It could
be recognized that the initial wetting state at the surface B is also the Cassie air
trapping state. The partial filling of pores discussed in detail by Marmur in Ref. [6]
is also seen from the comparison of the experimental value with the two calculated
ones.
The intermediate state is problematic for interpretation for two reasons: (1) the
fraction of filled pores remains unknown and (2) when pores are filled, two different
wetting regimes are possible, namely, the Wenzel regime and the Cassie impregnat-
ing regime with water penetrating into the texture [1]. The Cassie impregnating
model yields for APCA:
calc
cos θfin = 1 − S + S cos θE . (4)
The penetration of a liquid film into the pores under the Cassie impregnating regime
is possible when the condition θE < θc is fulfilled, where θc is the critical contact
angle given by (see Ref. [1])
1 − S
cos θc = . (5)
f − S
The comparison of the angles on flat surfaces with the critical contact angles in
Table 1 shows that the initial static Cassie impregnating wetting regime is impos-
sible for all surfaces A, B, C and D. Indeed, as was already mentioned, the Cassie
impregnating regime is not observed as an initial state. However, in the course of
vibration, the local contact angle may reach the critical value in equation (5), as will
be shown below.
Identification of the transitional and final wetting regimes could be achieved from
a comparison of the angles predicted by equations (1), (2) and (4) with the exper-
imental data summarized in Table 1. The first conclusion to be deduced is that the
final wetting state for surfaces A, C, D is the Cassie impregnating wetting state and
not the Wenzel one as was supposed previously. The final wetting state for surface B
is definitely the Wenzel state.
It has to be emphasized that the transitional state was stable, as was already
discussed in Ref. [30]. However, in the cases of surfaces A and B the difference
between experimental initial and transitional angles is relatively small. This may
be interpreted as the existence of the mixed Cassie (air trapping) and the Wenzel
states in these cases (see also Ref. [6]). The calculated values of angles support this
assumption. A further increase in the amplitude of vibration leads to the evolution of
Figure 5. Qualitative description of the Gibbs free energy dependence on the apparent contact angle
for a droplet on a rough surface.
the drop towards the final wetting state, which is the Cassie impregnating wetting
one, except for surface B where it is the Wenzel state. A direct evidence for the
existence of the Cassie impregnating wetting regime for wetting on rough surfaces
was also obtained by the Environmental Scanning Electron Microscopy (ESEM)
study of the fine structure of the triple line [44]. After the decrease of the APCA
from θin to θtr , the development of the precursor film surrounding the drop becomes
possible and water penetrates the pores in the patterned surface in the vicinity of the
droplet [44]. Vibrations provide the drop with the necessary energy for penetration
into pores on the surface.
The observed wetting transitions occur under different scenarios: it is mixed
Cassie air trapping/Wenzel state − Cassie impregnating state transition for sur-
face A; mixed Cassie air trapping/Wenzel state − Wenzel state transition for sur-
face B: Wenzel state− Cassie impregnating state transition for surface C; and Cassie
air-trapping state − Cassie impregnating state transition for surface D. The interme-
diate wetting state on surface D is obviously not Wenzel; it is more likely the Cassie
impregnating regime with a partial filling of pores.
The diversity of wetting transitions’ scenarios could readily be understood if one
takes into account that the Gibbs free energy curve for a droplet on a real surface is
characterized by multiple minima (Fig. 5), as discussed in Refs. [30, 45]. It should
also be noted that the surfaces exist (e.g., surface E) where wetting transition is not
observed at all. The energy supplied by vibrations in this case is not sufficient to
surmount the potential barriers separating wetting states. Perhaps this is due to the
hierarchical structure of a feather, which enhances the stability of the Cassie (air
trapping) state [40, 46].
2.2.2. The Physical Mechanism of the Wetting Transition

The fact that under certain threshold amplitude of vibration a wetting transition oc-
curs looks trivial. What is really non-trivial is the quantitative information which
could be derived from the experiments with vibrating drops. This information al-
lows a distinction between the two mechanisms proposed for wetting transitions,
namely, the mechanism of nucleation in the droplet center [36], and the mechanism
that relates the wetting transition to the displacement of the triple line [29, 30, 37,
41]. In particular, a flat spot occupying almost all central part of the droplet contact
area did not change any of experimental results connected with vibration-induced
transitions [41]. If the Cassie (air trapping)–Wenzel transition follows the nucle-
ation mechanism, it is reasonable to suggest that when the critical pressure in the
drop is surpassed, a wetting transition takes place [26, 33, 36]. However, if the wet-
ting transition is due to the displacement of the triple line, the situation is more
complicated. Energy conservation yields:
Fdepin δr = Fpin δr +
Gprecursor +
Gs , (6)
where Fdepin δr is the work done by the “de-pinning” force resulting in the dis-
placement, δr, of the triple line, Fpin δr is the work done by the force that pins the
triple line to the substrate (actually it resembles the friction force, see Ref. [24]),
Gprecursor is the change of the energy of the precursor film surrounding the drop
(see Ref. [29]), and
Gs is the energy jump due to the area change in the liq-
uid/solid and liquid/air interfaces. It has to be emphasized that all the terms in
equation (6) are related to a unit length of the triple line. The reasonable rough
estimation of
Gprecursor is:

Gprecursor = (γ + γSL + P (h))H ≈ (γ + γSL + P (h))δr, (7)
where h and H are the thickness and width of the precursor film respectively (see
Fig. 6), P (h) is a term due to the disjoining pressure (h) = − dP dh (see Ref. [1]).
It is reasonable to suggest that the displacement, δr, of the triple line necessary for
the onset of a wetting transition, the width of the precursor film H and the character-
istic scaling length of the rough surface are of the same order of magnitude, namely
∼5 µm for surfaces A, B and ∼20 µm for surfaces C, D. This suggestion is justified
by our previous ESEM study of precursor films [44]. The use of the term “precursor
film” is problematic (for hydrophilic surface B the formation of the true precursor
film is natural, whereas hydrophobic materials according to de Gennes do not allow
formation of precursor films [1]); in our case, the surface is rough and formation of
the thin water film surrounding the drop reported in Ref. [44] is, perhaps, due to the
capillary suction. In one of our works [29] we neglected the energy connected with
the displacement of the triple line; it will be demonstrated that taking
Gprecursor
into account is essential for qualitative interpretation of experimental data.

Gs was estimated by Shanahan [47] as:
γ sin2 θin (2 + cos θin )(δr)2

Gs ≈ , (8)
2r
Figure 6. Schematic of the droplet surrounded by a precursor film.
where θin is the initial APCA and r is the contact radius (see Fig. 6 and Ref. [47]).
It could be easily recognized that
Gs is of the second order of magnitude in δr.
Equations (7) and (8), when substituted into (6), yield:
Fdepin = Fpin + γ + γSL + P (h) + O(δr) = const + O(δr), (9)
where O(δr) means terms of the second order of magnitude in δr. Equation (9)
predicts that the displacement of the triple line occurs under the constant de-pinning
force, related to a unit length of the triple line. The de-pinning force Fdepin was
calculated in our previous paper (see Ref. [29] and the Appendix) as:
pR
Fdepin = (2θ − sin 2θ ), (10)
4 sin θ
where p is the pressure in the drop, R is the radius of the drop (see Fig. 6), and θ is
the APCA. The pressure in the drop is given by:
2γ
p = pi + + ρgl, (11)
R
where pi is the pressure increase caused by oscillation (see Ref. [18]), l is the drop
height, and ρ is its density. All terms in equation (11) are compatible for 2.5–50 µl
volume droplets and the above-mentioned frequency and amplitudes of vibration.
Thus, for example, for the 2.5 µl drop the Laplace, inertia, and hydrostatic pressures
are respectively 150, 52 and 6 Pa whereas for the 50 µl drop these are 55, 3 and 18 Pa
[29]. The resulting pressure exerted on the substrate, given by formula (11), and the
force (10) acting on a unit length of the triple line, are depicted in Fig. 7A–D.
These data were calculated from experimental APCA, drop volumes and threshold
vibration amplitudes.
It is seen that, unlike the critical pressure (that should be constant under the
nucleation mechanism of transition [26, 33, 36]), which decreases markedly with
the drop size, the critical value of F = Fdepin , corresponding to the origin of the
drop movement, remains approximately constant. The critical values of F for re-
leasing droplets are 200 mN/m for surfaces A, B and 300 mN/m for surfaces C, D
(Fig. 7A–D). Taking into account these values and γSL ∼ 20 mN/m for polymers
[42], as well as γ = 72 mN/m, one can estimate a pinning force from equation (9)
as: Fpin + P (h) ∼ 100–200 mN/m. The disjoining pressure justifies the relatively
high threshold values of the de-pinning force established experimentally. Actually
(A)
(B)
Figure 7. Dependence of the critical pressure and force per unit length of the triple line on the drop
volume (A) relief A, (B) relief B, (C) relief C and (D) relief D.
the de-pinning force depends not only on the topography of the surface, but also
on the long-range interaction between the solid substrate and the liquid, as was dis-
cussed in Refs [20, 24]. It is important that the data presented in Fig. 7 and their
analysis concern only the first wetting transition.
3. Horizontal Vibrations
3.1. Experimental
For study of horizontal vibration we used polystyrene honeycomb surfaces similar
to those displayed in Fig. 2A (see Refs [38, 39]). We also studied vibration of the
drop deposited on honeycomb reliefs containing smooth spots (see Ref. [41]). The
(C)
(D)
experimental device depicted in Fig. 3 was used with the only change that the direc-
tion of vibration was horizontal. In the case of horizontal vibrations we concentrate
on the first wetting transition only.
3.2. Resonance Wetting Transition Under Horizontal Vibrations
In the previous section the frequency of vibrations was chosen far from eigen-
frequencies of the drop. Now we concentrate on the effects occurring in the vicin-
ity of eigenfrequencies. At first, let us consider drop vibration irrespective of the
Cassie–Wenzel transition. When the drop is vibrated both bulk and surface modes
are excited. For the rough estimation of the nth bulk eigenfrequency ωn of the
droplet, the well-known Rayleigh formula can be used:
n(n − 1)(n + 2)γ
ωn2 = , (12)
ρR 3
where γ and ρ are the surface tension and density of water respectively, R is the
radius of the drop already introduced above. For the hemispherical sessile drop, as
has been shown recently by Lyubimov and coworkers, eigenfrequencies depend on
the boundary conditions at the triple line, and for a freely sliding contact line they
coincide with the natural frequencies of spherical drop [48]. It could be seen that for
the water drop with R ∼ 1–3 mm, ρ = 103 kg/m3 , γ = 72 mN/m the fundamental
frequency (n = 2) is in the range 20–120 Hz. For a fixed triple line the formula
obtained in Ref. [49]

6γ h(θ )
ω1 = R −3/2 (13)
ρ(1 − cos θ )(2 + cos θ )
gives for bulk modes the frequencies in the interval 10–60 Hz for water droplets of
sizes of 1–3 mm. (The dependence of the geometrical factor h on the APCA (θ ) is
found in Ref. [49].) This range for bulk modes intersects with that of surface ones
(5–250 Hz), however, as will be shown below, in the case of horizontal vibration
only the surface modes are responsible for the Cassie–Baxter wetting transition.
With the exception of the mode with the lowest frequency, all modes considered in
this work correspond to a fixed triple line.
As a result of the horizontal drop vibration, waves appeared on the drop surface
(see Fig. 8) extending over the meridian stripe parallel to the vibration direction.
Quite surprisingly, these waves can be described in the framework of the same
Figure 8. Profiles of oscillating drops. The modes with j = 1 (A), 3/2 (B), 5/2 (C) and 7/2 (D).
Depressions on the drop surface are hidden by the front part of the drop.
simple approach developed by Noblin et al. for the vertical drop vibration [50].
According to this approach, the surface waves obey
2π π(j − 1/2)
qj = = , (14)
λj L
where λ/2 is half wavelength, equal to the mean distance between the wave nodes
(see Fig. 8), q is the corresponding wave vector, L is the half-perimeter of the max-
imal meridian cross section of the drop, j is an integer or half-integer. The authors
of Ref. [50] observed surface modes with a fixed drop triple line corresponding to
j = 2, 3, . . . and those with a mobile triple line described by equation (14) with
j = 3/2, 5/2, . . . .
Note that the mentioned modes generated by vertical vibration are fully axisym-
metrical. We have revealed also the modes which are antisymmetric relative to the
symmetry plane of the drop orthogonal to the vibration direction (see Ref. [41]).
To find the frequency of the modes, the authors of Ref. [50] have used the dis-
persion relation for 1D capillary-gravity waves on a liquid bath [51].

γ 3
ω = gq + q tanh(qb),
2
(15)
ρ
where the mean depth b of the droplet is
V
b= (16)
π(R sin θ )2
with V being the known drop volume. As mentioned above, formulae (14)–(15)
successfully describe also the antisymmetrical modes (half-integer j ) with a fixed
triple line [41]. The overall correspondence between observed and calculated eigen-
frequencies is satisfactory [41].
To elucidate the influence of the vibration frequency on this transition, the water
droplet was exposed to horizontal vibration of increased amplitude at various fre-
quencies until the wetting transition took place, and the amplitude corresponding
to the Cassie–Wenzel transition was measured when the APCA decreased approxi-
mately from 90◦ down to 60◦ . In Fig. 9 these critical amplitudes of exciting oscilla-
tions are plotted versus corresponding frequencies. An obvious resonance character
of the curve in Fig. 9 can be recognized. The curves obtained for honeycomb reliefs
comprising smooth spot coincide with those established for original surfaces. This
important observation supports the idea that the Cassie–Wenzel transition does not
occur via nucleation mechanism beginning from the center of the drop as discussed
recently by Ishino and Okumura [36] but is caused by the displacement of the triple
line as demonstrated in our previous papers [29, 30, 41].
In our opinion the aforementioned wetting transition is controlled by two fac-
tors, which are the inertia force and Laplace pressure. The minima of the curve in
Fig. 9 correspond well to the resonance frequencies of the surface waves. In this
case the Laplace pressure arising from the surface curvature is maximal and gives
rise to the transition. On the other hand, far from resonance frequencies the main
Figure 9. Threshold amplitude of the wetting transition caused by horizontal vibration. The drop
volume is 50 µl.
role is played by the inertia term, which is proportional to the amplitude and to the
squared frequency of oscillations [28]; therefore maxima of the amplitude of the
curve in Fig. 9 tend to descend with the frequency rise. For instance, the most dis-
tinct peaks at 50 Hz and 110 Hz which are well separated from adjacent resonances
correspond to the excitation amplitude values Ae = 460 µm and 105 µm, respec-
tively, and give the close values 140 and 150 mN/m of the inertia force per unit
length of the triple line. As it was already shown in Section 2.2.2, it is the force per
unit length of the triple line that governs the wetting transition. The obtained values
140 and 150 mN/m are comparable with the critical value of 200 mN/m for vertical
vibrations also reported in Section 2.2.2.
Note that the investigated transition is dynamic in its nature. Indeed, the vibrat-
ing surface contains not only convex parts but also concave ones where the Laplace
pressure is negative (see Fig. 8). It seems reasonable to suggest that the transition
happens when a crest of the wave is near the triple line that gives rise to the local in-
crease of the Laplace pressure leading to the triple line displacement and subsequent
wetting transition. This stresses the 1D character of the Cassie–Wenzel transition in
the case of the horizontal vibration of the drop as well.
4. Conclusions
We conclude that wetting transitions are more likely 1D than 2D affair, i.e., the
transition occurs when a threshold force acting on the triple line is exceeded. This
is true for both horizontal and vertical vibrations of drops deposited on artificial and
natural rough surfaces. This is also true for various scenarios of wetting transitions.
Various pathways of wetting transitions were observed. The lowest energy state
corresponds to a Cassie impregnating wetting regime, when the droplet finds itself
on a wet substrate, viewed as a patchwork of solid and liquid. The character of
wetting transition is explained by the multiple minima in the Gibbs free energy of
the droplet on the rough surface. Qualitative estimations of the pinning force are
presented for the various surfaces studied in this work.
Acknowledgements
This work was supported by the Israel Ministry of Immigrant Absorption. We are
grateful to Professor M. Zinigrad and Professor D. Aurbach for their generous sup-
port of our scientific activity. We are indebted to M. Erlich, T. Stein, A. Musin and
Ye. Bormashenko for their help in preparing this paper.
References
1. P. G. de Gennes, F. Brochard-Wyart and D. Quéré, Capillarity and Wetting Phenomena. Springer,
Berlin (2003).
2. H. Y. Erbil, Surface Chemistry of Solid and Liquid Interfaces. Blackwell Publishing, Oxford
(2006).
3. D. Quéré, Rep. Prog. Phys. 68, 2495–2532 (2005).
4. D. Quéré and D. M. Reyssat, Philos. Trans. Royal Soc. A 366, 1539–1556 (2008).
5. D. Quéré, Ann. Rev. Mater. Research 38, 71–79 (2008).
6. A. Marmur, Langmuir 19, 8343–8348 (2003).
7. A. B. D. Cassie and S. Baxter, Trans. Faraday Soc. 40, 546–551 (1944).
8. A. B. D. Cassie, Discuss. Faraday Soc. 3, 11–16 (1948).
9. A. Marmur and E. Bittoun, Langmuir 25, 1277–1281 (2009).
10. R. N. Wenzel, Ind. Eng. Chem. 28, 988–994 (1936).
11. L. Gao and T. J. McCarthy, Langmuir 23, 3762–3765 (2007).
12. C. W. Extrand, Langmuir 19, 3793–3796 (2003).
13. J. Bico, U. Thiele and D. Quéré, Colloids Surfaces A 206, 41–46 (2002).
14. G. Whyman, E. Bormashenko and T. Stein, Chem. Phys. Letters 450, 355–359 (2008).
15. F. E. Bartell and J. W. Shepard, J. Phys. Chem. 57, 455–458 (1953).
16. R. H. Dettre and R. E. Johnson, J. Phys. Chem. 69, 1507–1515 (1965).
17. D. Y. Kwok and A. W. Neumann, Colloids Surfaces A 161, 49–62 (2000).
18. E. Chibowski, Adv. Colloid Interface Sci. 103, 149–172 (2003).
19. S. Vedantam and M. Panchagnula, Phys. Rev. Lett. 99, 176102 (2007).
20. M. Nosonovsky, J. Chem. Phys. 126, 224701 (2007).
21. C. W. Extrand, Langmuir 19, 646 (2003).
22. W. Li and A. Amirfazli, J. Colloid Interface Sci. 292, 195–201 (2005).
23. W. Li and A. Amirfazli, Adv. Colloid Interface Sci. 132, 51–68 (2007).
24. V. V. Yaminsky, in: Apparent and Microscopic Contact Angles, J. Drelich, J. S. Laskowski and
K. L. Mittal (Eds), pp. 47–95. VSP, Utrecht (2000).
25. E. Bormashenko, Y. Bormashenko, G. Whyman, R. Pogreb, A. Musin, R. Jager and Z. Barkay,
Langmuir 24, 4020–4025 (2008).
26. A. Lafuma and D. Quéré, Nature Materials 2, 457–460 (2003).
27. T. N. Krupenkin, J. A. Taylor, T. M. Schneider and S. Yang, Langmuir 20, 3824–3827 (2004).
28. E. Bormashenko, R. Pogreb, G. Whyman, Ye. Bormashenko and M. Erlich, Appl. Phys. Letters
90, 201917-1–201917-2 (2007).
29. E. Bormashenko, R. Pogreb, G. Whyman, Ye. Bormashenko and M. Erlich, Langmuir 23, 6501–
6503 (2007).
30. E. Bormashenko, R. Pogreb, T. Stein, G. Whyman, M. Erlich, A. Musin, V. Machavariani and
D. Aurbach, Phys. Chem. Chem. Phys. 27, 4056–4061 (2008).
31. Z. Yoshimitsu, A. Nakajima, T. Watanabe and K. Hashimoto, Langmuir 18, 5818–5822 (2002).
32. G. McHale, S. Aqil, N. J. Shirtcliffe, M. I. Newton and H. Y. Erbil, Langmuir 21, 11053–11060
(2005).
33. B. Liu and F. F. Lange, J. Colloid Interface Sci. 298, 899–909 (2006).
34. P. Patricio, C. T. Pham and J. M. Romero-Enrique, European Physical J. E 26, 97–101 (2008).
35. J. T. Hirvi and T. A. Pakkanen, Surface Sci. 602, 1810–1818 (2008).
36. C. Ishino and K. Okumura, Europhysics Letters 76, 464–470 (2006).
37. J. F. Joanny and P. G. de Gennes, J. Chem. Phys. 81, 552–562 (1984).
38. E. Bormashenko, R. Pogreb, O. Stanevsky, Ye. Bormashenko, T. Stein, V.-Z. Gaisin, R. Cohen
and O. Gengelman, Macromol. Mater. Eng. 290, 114–121 (2005).
39. E. Bormashenko, R. Pogreb, O. Stanevsky, Ye. Bormashenko, T. Stein, R. Cohen, M. Nunberg,
V.-Z. Gaisin and M. Gorelik, Materials Letters 59, 2461–2464 (2005).
40. E. Bormashenko, Ye. Bormashenko, T. Stein, G. Whyman and E. Bormashenko, J. Colloid Inter-
face Sci. 311, 212–216 (2007).
41. E. Bormashenko, R. Pogreb, G. Whyman, Ye. Bormashenko and M. Erlich, Langmuir 23, 12217–
12221 (2007).
42. D. W. van Krevelen, Properties of Polymers, p. 875. Elsevier, Amsterdam (1997).
43. R. Wang, L. Cong and M. Kido, Appl. Surface Sci. 191, 74–84 (2002).
44. E. Bormashenko, Ye. Bormashenko, T. Stein, G. Whyman and R. Pogreb, Langmuir 23, 4378–
4382 (2007).
45. A. Marmur, Soft Matter. 2, 12–17 (2007).
46. M. Nosonovsky and B. Bhushan, Mater. Sci. Eng. R 58, 162–193 (2007).
47. M. E. R. Shanahan, Langmuir 11, 1041–1043 (1995).
48. D. V. Lyubimov, T. P. Lyubimova and S. V. Shklyaev, Physics of Fluids 18, 012101-1–012101-11
(2006).
49. F. Celestini and R. Kofman, Phys. Rev. E 73, 041602-1–041602-6 (2006).
50. X. Noblin, A. Buguin and F. Brochard-Wyart, European Physical J. E 14, 395–404 (2004).
51. L. Landau and M. Lifshitz, Fluid Mechanics, 2nd edn. Butterworth-Heinemann (1987).
Appendix: Force Acting on a Unit of a Triple Line (Equation (10))

In the spherical coordinate system (Fig. 10), the force acting on the small ring of
Figure 10. Schematic illustrating calculation of the force acting on the triple line.
the sphere characterized by the fixed angle θ is

dF = p dS = 2πR 2 p sin θ dθ ,
where p is the pressure inside the droplet, dS is the surface of the ring and R is the
radius of the droplet. The projection on the x axis equals dFx = dF sin θ . The total
force acting on the sphere from inside in the horizontal direction is:

θ
1
Fx = 2πR 2 p sin2 θ dθ = πR 2 p θ − sin 2θ .
0 2
The ring length is 2πR sin θ , thus per unit length
Fx pR
= (2θ − sin 2θ ).
2πR sin θ 4 sin θ
Length Scale Effects in Wetting of Chemically
Heterogeneous Surfaces
Neeharika Anantharaju a , Mahesh V. Panchagnula a,∗ and Srikanth Vedantam b

a
Department of Mechanical Engineering, Tennessee Technological University, Cookeville,
TN 38505, USA
b
Department of Mechanical Engineering, National University of Singapore, Singapore 117576
Abstract
Wetting of chemically heterogeneous surfaces is modeled using phase field theory. Phase field theory in-
volves constructing a coarse-grained free energy functional. A kinetic evolution equation is developed based
on this free energy functional, which allows the phase field to evolve following a variational derivative of
the free energy functional with respect to the phase field variable. Contact angle hysteresis is incorporated
by a modified kinetic parameter. Using this model, we demonstrate the effect of variation in the length
scale of a chemically heterogeneous surface as deviation from the Cassie theory for a surface with a finite
length scale. We demonstrate the realm of applicability of Cassie theory as being for surfaces where the
component materials are intricately alloyed such that they are not discernible on the length scale associated
with the thickness of the diffuse molecular interface. In this context, the interface between the wetted and
non-wetted regions is a diffuse interface suffering steep but continuous changes from wetted to non-wetted
regions. When the length scale associated with either of the component materials is greater than the inter-
face thickness, the specific arrangement of the materials affects the evolution and shape of the three-phase
contact line, and thereby affecting the macroscopic contact angle. These results are of importance to surface
designers for a range of applications, where targeted wetting characteristics can be achieved by suitably
alloying component materials.
Keywords
Contact angle hysteresis, wetting kinetics, Cassie theory
1. Introduction
Wetting of chemically heterogeneous surfaces is of importance in a wide range of

practical applications ranging from inkjet printing to biofluidic manipulation. The
equilibrium contact angle, θY on a pure smooth surface can be obtained theoretically
from minimizing the system Gibbs free energy. However, this is difficult to realize
empirically owing to a plethora of metastable states surrounding this globally stable
* To whom correspondence should be addressed. Tel.: 931 372 6143; e-mail: mvp@tntech.edu

54 N. Anantharaju et al.
state. The wettability of real surfaces is, therefore, empirically characterized by the
apparent contact angle subtended by the drop at the surface during advancement
and receding events, known as the advancing and receding contact angles. The dif-
ference between these two angles is also an important parameter that describes the
sessile drop mobility, referred to as contact angle hysteresis. The advancing, reced-
ing and equilibrium apparent contact angles on a smooth chemically heterogeneous
surface composed of pure component materials A and B is given by the Cassie
equation [1],
cos(θiC ) = f A cos(θiA ) + (1 − f A ) cos(θiB ). (1)
j
Here the combinations, θi (i = a, r, Y and j = A, B, C) refer to advancing, re-
ceding and equilibrium contact angles respectively for the pure materials A and B
and the composite surface C. Theoretically, only the equilibrium contact angle on
a composite surface can be related to the equilibrium contact angles of the pure
component surfaces using the principle of Gibbs free energy minimization. The
extension of the same qualitative relationship to advancing and receding contact
angles on composite materials is purely based on empirical observations. Some re-
ports of theoretical justification have also been published [2, 3]. This extension has
met with mixed results with some researchers claiming success while others claim-
ing a deviation from this relationship. Secondly, the Cassie equation (1) for the
equilibrium contact angles assumes that the pure materials A and B are intricately
alloyed on the chemically heterogeneous surface. In other words, any elemental area
on the heterogeneous surface dA contains f A dA of material A and (1 − f A ) dA of
material B, for dA approaching the limit of zero. This clearly implies that no length
scale can be assigned to the surface chemical heterogeneity. Any real surface is
bound to have a finite length scale associated with the heterogeneity (for example,
in the case of droplet confinement for inkjet printing). It is, therefore, important
to understand the implication of the finiteness of the length scale associated with
the chemical heterogeneity on the advancing and receding angles on the composite
surface in order to design such surfaces for targeted applications.
In the current study, we have created hydrophilic and hydrophobic specimens
using silicon and silanized silicon wafers. The silane specimen was rendered hy-
drophobic by a silanization process outlined by Oner and McCarthy [4]. The ad-
vancing (receding) contact angle of a water drop on the specimen was measured
using a dynamic contact angle analyzer [5] as the asymptotic angle obtained while
liquid is quasistatically injected into (withdrawn from) the drop employing the cap-
tive needle technique [6]. The recently developed constitutive phase field model [7,
8] was employed to correlate with the experimental results.
2. Phase Field Theory

The phase field model is a simple phenomenological means of describing any kind
of transformation. It can be used in any situation where two (or more) distinct con-
Phase Field Model of Wetting Hysteresis Applied to Cassie Theory 55
tiguous regions undergo a transformation (through boundary motion) — provided

the transformation is driven by the minimization of some global extensive property
(e.g., the total free energy). This general nature of the model has led to it being
employed in diverse fields ranging from superconducting phase transitions [9], so-
lidification and melting [10], grain growth [11], solid–solid phase transformations
[12, 13] and vesicle membranes [14, 15].
Thus the phase field approach has three important constituents: (1) A continuous
parameter (phase field) to describe the distinct regions, (2) a coarse-grained free en-
ergy functional of the parameter field to reflect the total free energy of the system,
and (3) a kinetic evolution law for the phase field parameter to drive the system to
minimize the total free energy. The phase field variable describes the two distinct
regions by taking specific values in the regions (say, 0 and 1). The rigorous process
of constructing the coarse-grained free energy functional, which is the main input in
the phase-field formalism, involves removing the microscopic degrees of freedom
and integrating them out such that only larger scale macroscopic degrees of freedom
remain. In practice, however this process can only be carried out for relatively sim-
ple systems. It is much easier to determine the general structure of the free energy
functional directly from the known phase diagram, which is the approach adopted
here. A detailed discussion of the phase field theory applied to dynamical interface
problems can be found in Ala-Nissila et al. [16].
In order to develop a phase field model appropriate to wetting, we restrict our
attention to the solid surface and use the phase field parameter η to distinguish
between the wetted and nonwetted regions. The process of wetting or dewetting
takes place at the three-phase contact line which serves as the phase boundary. The
next step is to write the system free energy as a function of the phase field variable.
This function is required to enforce special equilibrium values of the phase field
variable to represent the wetted region by 1 and the nonwetted region by 0. This
can be done by considering a double-well quartic function of η, f (η), as
f (η) = γLV {8η4 − 2(Y + 8)η3 + (3Y + 8)η2 }, (2)
where Y = (f (1) − f (0))/γLV = cos θ − cos θY . The exact form of the above
function is not important provided it has equal minima at 0 and 1 when θ = θY .
Y represents the imbalance in the capillary forces and is the difference in heights of
the two minima at η = 0, 1. For θ < θY , the η = 1 well is metastable and the drop
attempts to retreat into a smaller wetted circle while aspiring to reach the Young’s
equilibrium condition. Whereas the η = 0 well is metastable when θ > θY with
the drop attempting to wet additional solid surface in order to reach the Young’s
equilibrium condition. Most importantly, the integral of f (η) over the entire solid
surface area yields the Gibbs free energy, G, of the system (see [8] for detailed
calculation).
G = γSL ASL + γLV ALV + γSV ASV , (3)
where γi , Ai are the interfacial energies and areas respectively of i = SL, LV, SV,
the solid–liquid, liquid–vapor and solid–vapor interfaces.
The total free energy, F , for the phase field model is written as

1
F= f (η) + λ|∇η| dA.
2
(4)
A 2
Here, λ is the gradient coefficient that is thermodynamically required to be pos-
itive, and is shown below to be related to the three-phase contact line tension. ∇η
identifies the interface between the wetted and dry regions. The gradient term pro-
vides a penalty for the presence of interfaces between regions of constant η. It has
been postulated that the line tension arises out of finite range molecular forces at
the triple line. Equation (4) is the phase field equivalent of the Gibbs free energy
expression (3) and includes the energy of the three-phase contact line.
In the Ginzburg–Landau framework of the phase field theory, the equilibrium so-
lution for the order parameter is obtained through a gradient flow evolution equation
of the form
δF ∂f (η)
β η̇ = − = λ∇ 2 η − , (5)
δη ∂η
where δ is a functional derivative and β(x, t, η, ∇η, η̇) > 0 is the kinetic coefficient
which is required to be positive for all admissible values of its arguments [17]. Note
that in the regions away from the three-phase contact line, where η = 0 or 1, both
terms on the right-hand side of equation (5) vanish and there is no evolution of η.
Thus the evolution of the drop footprint radius is only through the motion of the
contact line and the local material properties at the contact line determine the state
of the drop. This is physically consistent with thermodynamic energy minimization
principle that the change in free energy determines the state of the drop [18].
The kinetic coefficient, β, is given a special form in order to account for hystere-
sis

β = ξ H(|∇η|) + ω|η̇|m /|η̇|, ξ, ω > 0, (6)
where H(x) is the Heaviside function. The ξ and ω terms provide the rate-
independent and rate dependent hysteresis components, respectively. In the limit
of quasistatic motion, the hysteresis is contributed purely by ξ .
Equation (5) is solved numerically over a discretized square domain. For de-
scribing a smooth heterogeneous surface, the properties ξ and θY in the chemically
different regions are given appropriately different values. A simple central differ-
ence of the spatial terms and forward difference of the temporal terms are employed
in the solution of equation (5). An explicit Euler method is used to relax the initial
guess for the solution to the final equilibrium solution. The elegance of this method
lies in the simplicity, and ease of numerical implementation and computer program-
ming.

The phase field model discussed in the previous section requires only the advancing
and receding contact angles of the pure material surfaces as inputs [8]. Figure 1
(a)
(b)
Figure 1. Instantaneous drop contact angle versus (a) drop volume and (b) drop footprint radius. Points
A–D and P–S indicate events when either angle or drop footprint radius changes value (courtesy of
[8]).
shows the advancing and receding behavior of a drop on two pure materials under
consideration, viz. perfluoroalkoxy (PFA) and etched perfluoroalkoxy (ePFA) [19].
Figure 1(a) is a plot of contact angle versus drop volume while Fig. 1(b) is a plot of
the contact angle versus the drop footprint radius for PFA (loop ABCD) and ePFA
(loop PQRS) surfaces. It can be seen from Fig. 1(b) that from A to B (P to Q),
the contact angle increases at constant drop footprint radius until advancing angle
of PFA (ePFA) is reached. Further increase in the volume yields the asymptotic
advancing angle with increase in drop footprint radius from B to C (Q to R). At the
point C (R), when the volume is withdrawn from the drop, the drop contact angle
is observed to decrease at constant drop footprint radius from C to D (R to S) until
the receding angle of PFA (ePFA) is reached. Further decrease in volume yields
the asymptotic receding angle from D to A (S to P) with the drop footprint radius
decreasing.
Now consider a composite surface as shown in Fig. 2, composed of a 1.35 mm
diameter island of ePFA in a PFA substrate. The phase field theory that was vali-
dated on each of the two pure surfaces, PFA and ePFA, can be used to model the
sessile drop advancement and receding behavior on this composite surface with no
additional inputs. Figure 3 is a plot of the instantaneous contact angle versus drop
footprint radius. As can be seen from this figure, a drop that is initially in the just
receded state (A) entirely sessile on the ePFA surface will accommodate increase
Figure 2. Composite surface composed of an island of etched perfluoroalkoxy embedded in a perflu-

oroalkoxy substrate.
Figure 3. Instantaneous drop contact angle versus drop footprint radius. Points A–H indicate events
when either angle or drop footprint radius changes value (courtesy of [8]).
in volume by a gradual increase in contact angle while remaining on the same drop
footprint until a point B is reached where the contact angle equals the advancing an-
gle of ePFA. As drop volume increases further, the footprint radius increases with
the angle remaining constant until the drop reaches a footprint radius corresponding
to the boundary between the ePFA and PFA regions. At this point C, the drop foot-
print radius remains constant until the angle becomes equal to the advancing angle
of the material that the drop is about to wet. Further increase in volume is accommo-
dated by increasing the drop footprint radius. In the reverse direction, a reduction
in drop volume is accommodated in such a fashion that the drop footprint radius
would only decrease when the drop angle is equal to the receding angle of the ma-
terial that the drop is about to de-wet. See the receding path for the same substrate
EFGHA. In this fashion, the phase field model is able to capture the sessile drop
advancement and receding behavior on a chemically heterogeneous substrate with
only the advancing and receding angles of the pure constituent materials as inputs.
The phase field theory is further validated against experimental measurements
of advancing and receding contact angles on chemically homogeneous silicon and
silane surfaces. Figure 4 presents a plot of the instantaneous contact angle versus
drop footprint radius on pure silicon and silane surfaces. Good agreement can be
noticed between theory and experiment.
The phase field model is now applied to three classes of heterogeneities associ-
ated with a composite surface of a superhydrophobic material, A (advancing angle,
θa = 145◦ and receding angle, θr = 125◦ ) and a relatively hydrophilic material,
B (θa = 95◦ and θr = 83◦ ) in order to demonstrate the effect of the three-phase con-
Figure 4. Hysteresis loops for two surfaces with advancing (circles) and receding angles (triangles) of
plain silicon (open symbols) and plain silane (closed symbols) surfaces from phase field predictions
(solid line) and experimental measurements (symbols). Also shown are the images of an advanced and
receded drop.
tact line structure on the observed macroscopic contact angle behavior. First, we
consider a situation where the two materials are intricately alloyed at the surface
(referred to as Case I). We achieve this condition numerically by using a random
number generator to assign properties to each grid point. When the value of the
random number generated is less than a desired area fraction, the grid point is as-
sociated with the properties of material A; else, the properties of material B are
ascribed to the grid point. We use this set of simulations to demonstrate the effect
of the lack of a heterogeneity length scale [20]. Secondly, we consider squares of A
(the more hydrophobic of the two materials) of side, a, separated by a distance, b,
at centers and embedded in B (referred to as Case II). For this case, we consider
a change in the length scale, b, while maintaining the same area fraction of A,
f = 1 − a 2 /b2 . Finally, we consider a composite surface of squares of material B
embedded in A (referred to as Case III).
Figure 5 is a graph corresponding to the data for Cases I and II. The graph shows
the variation of the cosine of the advancing contact angle with area fraction of A.
The solid line represents the values estimated by the Cassie theory. The solid circle
symbols represent the data for a situation where the two materials are considered
to be intricately alloyed. The data for different length scales of the centre spac-
ing between the squares (b = 16 µm, 32 µm and 64 µm) are represented by the
solid symbols. It can be observed from the graph that for the first case where the
materials are intricately alloyed at the surface, there is no discernible length scale
associated with each material [21]. This case, therefore, is consistent with the intent
Figure 5. Cosine of advancing contact angle versus fraction of material A for three different length
scales with squares of A in a substrate of material B. Here the solid line represents the data for pre-
dictions from the Cassie theory and the solid symbols corresponds to the data for different length
scales.
of Cassie’s derivation and hence this theory holds true. The surfaces with differ-
ent center spacings show deviation from the Cassie theory signifying length scale
effect, but showing negligible variations with the varying length scale at constant
area fraction. This is explained by the fact that the squares of the more hydrophobic
material A act as pinning locations causing the contact line to become more tortu-
ous. The strength of such pinning does not seem to depend on the center spacing
between the squares and hence the data for the different center spacing values, b,
show no deviation from each other.
Figure 6 shows the variation of the cosine of the advancing contact angle with
area fraction for Case III. The solid line again represents the predictions based on
the Cassie theory while the open symbols represent the data for varying length
scales (b = 16 µm, 32 µm and 64 µm). The data show negligible deviation from
the Cassie theory for smaller length scale of 16 µm and an increase in the devia-
tion with the corresponding increase in the length scale at constant area fraction.
This is qualitatively different from Case II since the squares of the relatively less
hydrophobic material B do not attempt to pin the three-phase contact line. There-
fore, the contact line remains relatively circular. However, in contrast with the data
in Fig. 5 for Case II, as the center spacing between adjacent squares increases, the
data show increasing deviation from Cassie theory. Similar asymmetric hysteresis
was also empirically observed by Priest et al. [22].
From the above observations, it can be summarized that the phenomenon of
“asymmetric” hysteresis can be observed on a chemically heterogeneous surface
Figure 6. Cosine of advancing contact angle versus fraction of material A for three different length
scales with squares of B in a substrate of material A. Here the solid line represents the data for pre-
dictions from the Cassie theory and the open symbols corresponds to the data from different length
scales.
where either the advancing or the receding contact angle deviates from the Cassie
curve for a surface of same area fraction and length scale. According to the Cassie
theory, the chemical heterogeneity in the surface is such that a small elemental
area, dA, of the surface contains f A dA of material A and (1 − f A ) dA of material
B [1, 23]. This can be achieved for the case where the two materials were considered
to be intricately alloyed. The effect of the length scale would be absent in such a
case. Thus we see agreement between the advancing angle data from Case I (closed
circle symbols) and Cassie theory (solid line) as can be seen from Fig. 5. But many
real surfaces have finite length scales associated with the chemical heterogeneity.
Fabrication of a composite surface as squares of one material in another brings in
this effect of the length scale, thus exhibiting a deviation from the Cassie curve.
Further, it may be noted that all the data points presented were obtained on surfaces
where the characteristic length scale associated with the heterogeneity was at least
ten times smaller than the drop footprint radius (R ≈ 1 mm). This implies that on the
length scale of the drop, the surface can be considered to be nearly “homogenized”
heterogeneous surface. However, a significant deviation from Cassie theory is no-
ticed even for small characteristic length scale of the heterogeneity (b/R ≈ 0.016).
Figure 7 shows the contact line structure for two surfaces, both of the same area
fractions; Fig. 7a is a substrate made of squares of material A in B while Fig. 7b is
for a substrate with squares of B in A. The nature of the modification of a nearly
circular contact line owing to the local variations in surface energy can be observed
for the composite surfaces in these two cases. This variation in the contact line
structure is carried over into macroscopic contact angle observations. Thus, for two
composite surfaces with the same global area fraction of A, we observe that the
nature of the contiguous material has an effect on the macroscopic contact angle.
The underlying physical principle that we have discovered, viz. deviation from
Cassie theory for a finite length scale, is analogous to the principle of estimating
the viscosity of binary Newtonian liquid mixtures. When the two fluids are per-
fectly miscible, the mixture viscosity is well estimated by a weighted mixture law
of the component properties [24]. However, when the two fluids are partially misci-
ble or immiscible (for example, air–water two-phase systems), a discernible length
(a) (b)
Figure 7. Shape of the triple line for (a) A in B, (b) B in A at constant area fraction of material A.
scale can be associated with the heterogeneity, due to which classical mixture laws
are violated. In order to predict the viscosity of such a mixture, other “internal”
variables (for example, the air bubble average diameter) need to be included in the
model to account for mixture morphology, similar to the current situation.
The Cassie equation (1) which was derived from the Gibbs free energy mini-
mization principle states that the drop assumes a shape where there is a change in
free energy per unit change in the wetted area. As the drop advances, the three-
phase contact line encounters metastable energy wells and could remain trapped
in one of such wells. This mechanism has been proposed to explain the difference
between the advancing angle and the thermodynamic equilibrium angle. However,
the metastable well associated with chemical heterogeneity has to be based on the
local change in surface energy at the contact line as a change in wetted area. This
change in free energy is, in turn, related to the local area fraction as opposed to
the global area fraction obtained from the geometric arrangement of the squares.
Therefore, the local area fraction that the contact line experiences as it advances is
apt for the calculation of the expected contact angle. Similarly, the area fraction that
is pertinent to the receding angle calculation is that obtained near or just inside the
three-phase contact line.
4. Conclusions
We studied wetting hysteresis on a smooth, chemically heterogeneous surface and
investigated the effect of the length scale associated with this heterogeneity. The
advancing and the receding measurements are observed to be significantly different
depending on the fabrication procedure of the two materials into a heterogeneous
surface. We propose that this asymmetric hysteresis depends on the distinct wetting
behavior closer to the three-phase contact line of the drop. The work provides in-
sight into the significance of the fabrication procedure and the arrangement of the
component materials, taking into account the length scale effect, for understanding
wetting of chemically heterogeneous surfaces.
References
1. A. B. D. Cassie, Disc. Faraday Soc. 3, 11–16 (1948).
2. G. Whyman, E. Bormashenko and T. Stein, Chem. Phys. Lett. 450, 355–359 (2008).
4. D. Oner and T. J. McCarthy, Langmuir 16, 7777–7782 (2000).
5. N. Anantharaju, M. V. Panchagnula and S. Vedantam, Langmuir 23, 11673–11676 (2007).
6. B. He, J. Lee and N. A. Patankar, Colloids Surfaces A 248, 101–104 (2004).
7. S. Vedantam and M. V. Panchagnula, Phys. Rev. Lett. 99, 176102 (2007).
8. S. Vedantam and M. V. Panchagnula, J. Colloids Interface Sci. 321, 393–400 (2008).
9. V. L. Ginzburg and L. D. Landau, Zh. Eksp. Teor. Fiz. 20, 1064 (1950).
10. L. Granasy, T. Pusztai and J. A. Warren, J. Phys. — Condensed Matter 16, R1205–R1235 (2004).
11. S. Vedantam and B. S. V. Patnaik, Phys. Rev. E 73, 016703 (2006).
12. E. Fried, Continuum Mech. Thermodynam. 9, 33–60 (1997).

13. E. Fried and G. Grach, Arch. Rational Mech. Anal. 138, 355–404 (1997).
14. T. Biben, K. Kassner and C. Misbah, Phys. Rev. E 72, 041921 (2005).
15. T. Biben and C. Misbah, Phys. Rev. E 67, 031908/1–031908/5 (2003).
16. T. Ala-Nissila, S. Majaniemi and K. Elder, Lecture Notes in Physics 640, 357–388 (2004).
17. E. Fried and M. E. Gurtin, Physica D72, 287 (1994).
18. M. Nosonovsky, Langmuir 23, 9919–9920 (2007).
20. M. V. Panchagnula and S. Vedantam, Langmuir 23, 13242 (2007).
21. T. Cubaud and A. Fermigier, J. Colloids Interface Sci. 269, 171–177 (2004).
22. C. Priest, R. Sedev and J. Ralston, Phys. Rev. Lett. 99, 026103 (2007).
24. P. N. Shankar and M. Kumar, Proc. Royal Soc. A 444, 573–581 (1994).
Effects of Water Adsorption on Silicon Oxide Nano-asperity
Adhesion in Ambient Conditions
David B. Asay, Anna L. Barnette and Seong H. Kim ∗

Department of Chemical Engineering, The Pennsylvania State University, University Park,
PA 16802, USA
Abstract
As the size of the contact decreases from the macro-scale to the nano-scale, the measured contact forces
become more sensitive to the molecular adsorption from gaseous environments. This paper discusses the
effects of water adsorption on the adhesion force measured with atomic force microscopy (AFM) for the
single-asperity contact between silicon oxide surfaces. As relative humidity (RH) increases, the adhesion
force measured with AFM initially increases, reaches a maximum, and then decreases at high RH. Tradition-
ally, this RH dependence has been attributed to the capillary force caused by the liquid water condensation
at the AFM tip–surface contact. However, the capillary force based on the liquid water property alone does
not seem to be enough to explain the observed magnitude of the RH dependence. In this paper, experimental
evidences are described to show correlations between the solid-like structured water layer on the silicon
oxide surface observed with vibrational spectroscopy and the large RH dependence of the nano-asperity
silicon oxide adhesion measured with AFM.
Keywords
Water adsorption, silicon oxide, capillary, adhesion
1. Introduction
In a humid ambient, water adsorbs onto virtually all surfaces except highly hy-
drophobic surfaces. As a consequence, interfacial water on solid surfaces plays
important roles in engineering systems such as tribology, biomaterials, and nano-
technology as well as in natural systems such as biology, meteorology, and geology
[1–34]. In many cases, understanding water structure at an interface is key to de-
termining wetting phenomena and their chemical activities [5]. On certain mineral
surfaces, strongly bound hydration layers show a remarkable lubrication property
for atomically smooth surfaces (such as mica) [6, 7]. In contrast, the water layers
adsorbed on silicon oxide surfaces are not good lubricants for asperity contacts [8].
* To whom correspondence should be addressed. E-mail: shkim@engr.psu.edu

66 D. B. Asay et al.
The adsorbed water layers can induce tribochemical reactions which enhance wear
of silicon oxide [9–12].
The effects of water adsorption in humid environments are manifested in small-
scale devices such as microelectromechanical systems (MEMS). The first MEMS
devices containing pin joints, gears and slides were constructed in the late 1980’s
[13]. Since then, a number of MEMS devices with various mechanical functions
have been designed and constructed [14, 15]. Typical MEMS devices are fabri-
cated with silicon-based materials using microfabrication technologies developed
in semiconductor industries. Once the device is exposed to air, the surface becomes
oxidized forming a native silicon oxide layer. The mechanical parts involved in
these MEMS devices have intrinsically high surface-to-volume ratio due to their
small scale size. Therefore, gravitational body forces become insignificant and
surface properties such as adhesion and friction are dominant factors in device op-
eration [1]. Although these devices were designed to have minimal adhesion and
friction forces, motion of these small-scale devices was possible only with high
actuation voltages. When water is adsorbed on these device surfaces, then the ad-
hesion force can significantly increase to a level which exceeds the force that can
be generated by the actuation voltages [13–15]. Such an occurrence can cause the
device to fail to operate. Even though the device surfaces are coated with hydropho-
bic layers to prevent water adsorption, these coatings wear off easily under repeated
sliding stresses and the newly exposed bare substrate surfaces become suscepti-
ble to water adsorption [16, 17]. If one boosts the actuation voltage and forces
the device to operate under high adhesion conditions in a humid ambient, the de-
vice surface starts showing severe wear in the sliding contact regions [18]. The
wear behavior associated with interfacial water is notorious for silicon oxide sur-
faces.
In order to elucidate the fundamental aspects of water adsorption effects on adhe-
sion of solid contacts involved in micro-scale devices, it is important to understand
the nano-asperity contacts. For example, in MEMS devices, real mechanical con-
tacts typically consist of a few nanometer-scale asperities [1]. For the silicon oxide
nano-asperity contact in humid environments, it is often observed that as relative
humidity (RH) increases, the adhesion force initially increases, reaches a maxi-
mum, and then decreases at high RH [19–27]. This complicated RH dependence
has traditionally been attributed to changes in the capillary force due to conden-
sation of liquid water between two solid surfaces. Numerous theoretical modeling
papers have been published in the literature to explain the relative humidity depen-
dence of the nano-asperity adhesion using the properties of liquid water. Elegant
mathematical modeling approaches were able to partially reproduce the relative hu-
midity dependence; but this does not mean that the influence of the adsorbed water
layer thickness and structure can be ignored from theoretical considerations [22–
27]. This review discusses the molecular origin of this discrepancy and shows a
correlation between the structure of the adsorbed water layer observed with vibra-
Effects of Water Adsorption 67
tional spectroscopy and the adhesion force measured with atomic force microscopy
(AFM) [28–31].
2. AFM Measurements of RH Dependence of Silicon Oxide Nano-asperity

Adhesion
The nano-asperity adhesion force of the silicon oxide surface was measured with
AFM in a controlled humidity environment. Details on the experimental conditions
are given in [29]. In brief, the silicon oxide surface was chemically cleaned so that
water wetted it completely. The water contact angle was lower than 5◦ , indicating
that the surface was fully covered with silanol (Si–OH) groups [32, 33]. Six differ-
ent AFM tips with an average curvature of 15 ± 5 nm were cleaned with UV/ozone
and used for force–distance measurements. The adhesion force was measured from
the pull-off point of the retraction part of the force–distance curve obtained in a
given RH.
Figure 1 displays the AFM pull-off force as a function of RH at room temper-
ature. On average, the nano-asperity silicon oxide adhesion force increases from
40 nN to 105 nN as RH increases from 0 to ∼30%, becomes relatively constant
within the RH region from 30 to 50%, and then decreases from 100 nN to 20 nN
as RH increases from 50 to 90%. Similar RH dependences of the nano-asperity
silicon oxide adhesion force have also been reported [20–25]. Although there are
some discrepancies in the magnitude and shape of the RH dependence among these
reports due to differences in sample preparation and cleanliness, the general trend
is consistent with the one shown in Fig. 1. This is due to the adsorption of water on
Figure 1. Adhesion force measured with AFM for hydrophilic silicon oxide surfaces measured at
room temperature. The average AFM tip curvature was 15 ± 5 nm [29].
the silicon oxide surfaces from the vapor phase. Theoretical models predicting the
water adsorption effect on the nano-asperity adhesion force will be discussed in the
following section.
3. Theoretical Approach to Calculate the Force Due to Meniscus Pressure of

the Capillary
In humid environments, water can condense to form a meniscus at or around the

nano-asperity contact (Fig. 2a). This can take place even though the water partial
pressure in the gas phase is much lower than the saturation vapor pressure of water
at a given temperature. This is because the effective saturation vapor pressure de-
creases in a nano-scale gap as found at the peripheral region of the AFM tip–surface
contact point [34]. Conventionally the force due to the pressure difference across
the capillary meniscus is described as follows:
FMeniscus = −PLaplace · Area = −PLaplace × πRA

2
, (1)
where PLaplace is related to the surface tension of the condensed liquid, γ , and
two principal radii of the meniscus, cross-sectional (RA ) and meridional (r), under
equilibrium condition:

1 1
PLaplace = γ + . (2)
RA r
Since this expression has two unknowns (RA and r) for one independent equation,
the degree of freedom is +1, i.e., it cannot be solved exactly without further infor-
mation such as assumptions or independent relationships. For macro-scale contacts,
one can assume that the tip curvature (Rtip ) is much larger than the cross-sectional
radius of the meniscus (RA ) which is, in turn, much larger than the axial curvature
(a) (b)
Figure 2. Schematics of two models describing the capillary necking between the AFM tip and the
substrate due to the condensed water. (a) represents the traditional view for the non-wetting solid
surfaces and (b) is shown for the toroidal approximation for the case where both the tip and substrate
surfaces are covered with the adsorbed water layer.
of the meniscus (r) [35]. Then, we can approximate PLaplace using the following
expression:

1 1 γ
PLaplace = γ + ≈ (since RA r). (3)
RA r r
This Laplace pressure acts on an area πRA 2 ≈ 2πR d (where d = height of the
tip
meniscus). If RA Rtip , then d ≈ 2r · cos θ (where θ = contact angle of the liquid
on the substrate surface). Using these approximations, the meniscus force due to
the Laplace pressure (equation (1)) can be expressed as:
FMeniscus ≈ 4πRtip γ cos θ. (4)
However, the assumption of Rtip RA r is not valid for nano-asperity contacts

such as AFM tips with R ∼ 15 nm [20, 22, 25]. For clean silicon oxide surfaces,
the contact angle is almost zero, giving rise to cos θ ≈ 1 and FMeniscus ≈ 4πRtip γ ,
which does not have any RH dependence.
The +1 degree of freedom problem can be resolved if one uses an experimen-
tally obtainable information. Figure 2b redraws the AFM tip–surface contact region
with a toroidal approximation. The main difference from the previously discussed
model is that both tip and substrate surfaces are covered with an adsorbed water
layer with the same thickness which is a function of RH, h(RH). This is because
both surfaces (tip and substrate) were cleaned and highly hydrophilic. The toroidal
approximation for the meniscus curvature is not accurate; but it does provide a rea-
sonably good approximation [36, 37]. With these assumptions, we can introduce a
simple geometric relationship among Rtip , RA , r, and h(RH):

RA = 2 Rtip · r + Rtip · h(RH) − r. (5)
When the condensed liquid in the meniscus is in equilibrium with the vapor phase,
the Laplace pressure can be expressed with the Kelvin equation:

1 1 Rg T P
PLaplace = γ + = ln , (6)
RA r V Psat
where Rg is the ideal gas constant. Note that PLaplace is determined by the relative
vapor pressure of water (P /Psat × 100% = RH), the molar volume of the condensed
water phase (V ), and the ambient temperature (T ). This, in turn, determines RA and
r for the liquid meniscus with a certain γ value. Now we have two equations {equa-
tions (5) and (6)} and three unknowns {RA , r, h(RH)}. Among the three unknowns,
the adsorption isotherm thickness, h(RH), can be experimentally measured. Once
this variable is determined, then equations (5) and (6) can be solved simultaneously
for RA and r. With these two solutions, one can calculate the capillary force due to
the meniscus pressure using equation (1).
4. Adsorption Isotherm Thickness of Water Adsorbed on Clean Hydrophilic

Silicon Oxide
In order to measure the adsorption isotherm thickness of water on the clean, hy-
drophilic silicon oxide surface, we utilized attenuated total reflection infrared (ATR-
IR) spectroscopy. Details of this experiment were previously described in [28].
A silicon ATR crystal coated with amorphous SiO2 layer was cleaned and exposed
to the controlled humidity argon flow. The IR beam travels through the ATR crystal
and is totally reflected at the crystal surface. Upon total reflection, an evanescent
electric field penetrates into the adsorbate and vapor phase space. Its penetration
depth is ∼270 nm at 3000 cm−1 . In this shallow penetration depth, there are some
gas-phase water molecules; but their density is not high enough to be detected in
ATR-IR. So, only the adsorbed water layer is detected without interference from
the gas phase species.
Figure 3a displays the O–H stretching (3000–3650 cm−1 ) and H–O–H bend-
ing (1600–1700 cm−1 ) regions of the adsorbed water molecules as a function of
RH. The data clearly show that the adsorbed water peak intensities increase as RH
increases. The equilibrium adsorption thickness is determined by comparing the
ATR-IR intensity of the adsorbed species with the IR absorption of bulk water on
the ATR crystal [28]. The adsorption isotherm thickness determined from the in-
tensity of the 1640 cm−1 peak is plotted in Fig. 3b. When the adsorption isotherm
thickness was calculated from the area of the OH stretching peak or by comparing
with theoretical spectra simulated with the optical constant of bulk water, it was
found within ∼20% error from the one calculated from the H–O–H bending vi-
bration intensity [30]. The thickness of water adsorbed on amorphous SiO2 in the
monolayer unit can be estimated by dividing the measured thickness by 2.82 Å, the
mean van der Waals diameter of water. The average adsorption isotherm thickness
determined from the ATR-IR intensity follows the typical type-II isotherm curve —
(a) (b)
Figure 3. (a) ATR-IR spectra for the adsorbed water layer on the hydrophilic SiO2 surface as a func-
tion of RH. (b) Adsorption isotherm thickness calculated from the intensity of the H–O–H bending
vibration [28].
fast growth of the thickness in the low humidity region (RH < 20%) and slow
growth in the mid-RH region followed by fast growth again in the near-saturation
region (RH > 80%). The data are in good agreement with previously reported data
for hydrophilic surfaces [38, 39].
5. Capillary Force Due to the Meniscus and van der Waals Interactions
Since we now know the adsorption isotherm thickness, h(RH), we can calculate two
principal radii of the meniscus at a given RH, RA (RH) and r(RH), using equations
(5) and (6). This allows us to calculate the capillary force contribution due to the
meniscus pressure using equation (1). The dotted line in Figure 4 is the force due
to the meniscus pressure (FMeniscus ) as a function of RH for 5% < RH < 95%. It
can clearly be seen that FMeniscus increases as RH increases from the dry condition,
reaches a maximum at RH ∼ 20%, and then decreases as RH increases further.
Whenever two solid objects are in close proximity, there are van der Waals in-
teractions not just inside the contact region but also outside the contact. In the
macro-scale, the van der Waals interaction outside the contact region is negligi-
ble compared to the interactions involved in the contact. In contrast, the van der
Waals interaction cannot be ignored at the nanoscale. One can estimate the van der
Waals force (FvdW ) using the Derjaguin approximation: FvdW = 4πRtip γ . For the
RH region where the silicon oxide is fully covered with water (at least a monolayer
or thicker), the surface tension value used in the Derjaguin approximation should be
that of the adsorbed water layer, not of silicon oxide. The surface tension of the ad-
Figure 4. Comparison of FMeniscus and FMeniscus +FvdW with the experimentally measured adhesion
force.
sorbed water layer could be estimated from those of ice and liquid water [29]. Note
that this simple approximation does not take into account the fact that in the center
of the contact region, there is also a continuous liquid neck. In order to consider
the exact situation, one has to calculate the surface tension force around the liquid
neck circumference and the van der Waals force should be integrated for the area
outside the liquid neck [36]. But, this difference is much smaller than FMeniscus ; so,
the Derjaguin approximation can give a reasonable estimate for these interactions.
The dashed line in Fig. 4 shows the sum of FvdW and FMeniscus . The sum of these
two contributions qualitatively follows the general trend observed experimentally
for the RH dependence of the nano-asperity silicon oxide contact in humid environ-
ments. At RH > 90%, the FvdW + FMeniscus term agrees well with the experimental
values. When the low RH region is extrapolated to the dry condition, it approaches
the experimentally observed value. These results imply that the model represents
the experimental situation fairly well when the capillary force is negligible. How-
ever, the predicted magnitude of the RH dependence significantly underestimates
the experimental results between RH 10% and 80%.
If the theory based on the assumption that the adsorbed layer behaves similar to
the bulk liquid cannot explain the experimentally observed RH dependence quanti-
tatively, then this assumption must be invalid and should be improved or modified.
On the clean, hydrophilic silicon oxide surface, there are a large number of silanol
groups exposed to the interface. These surface-anchored hydroxyl groups will par-
ticipate in hydrogen bonding with the adsorbed water molecules. This can influence
the structure or arrangement of water molecules in the adsorbed layer and make it
different from the bulk water. The improved theoretical model should include this
aspect.
6. Structure of the Water Layer Adsorbed on Silicon Oxide as a Function of

RH
In IR spectroscopy, the structure of the water layer adsorbed on a silicon oxide sur-
face can be deduced from the O–H stretching vibration peak position and shape.
Water in the condensed phase can form a varying degree of hydrogen bond net-
work. When water molecules are completely self-associated as in crystalline ice,
each water molecule has four hydrogen bonds with its nearest neighbors forming
a tetrahedral arrangement [40]. The O–H stretching vibration in the crystalline ice
structure appears at ∼3220 cm−1 with a full-width-at-half-maximum (FWHM) of
∼200 cm−1 [41–46]. Above the freezing temperature, liquid water is the thermo-
dynamically stable structure for the condensed water phase at ambient conditions.
In liquid water, the average number of hydrogen bonds per water molecule is lower
than three and the O–H stretching vibration appears at ∼3400 cm−1 with a FWHM
of ∼300 cm−1 [41, 46, 47]. When the hydroxyl group in a water molecule is not
involved in hydrogen bonding, the O–H stretching peak appears at ∼3640 cm−1
with a very narrow FWHM [48]. Thus, the position of the OH vibrational peak is
(a) (b)
Figure 5. (a) O–H stretching vibration region showing the solid-like (vertical dotted line) and liquid
(vertical dashed line) components in the adsorbed water layer as a function of RH. (b) Deconvolution
of the total isotherm thickness into the solid-like and liquid components [28].
very sensitive to the degree of hydrogen bonding and has been used as an indicative
tool to study the structure of water [49].
Figure 5a shows again the O–H stretching vibration region of the adsorbed water
layer detected with ATR-IR as a function of RH and the peak positions of the bulk
liquid water and ice (dotted and dashed lines, respectively). From these data, it
can clearly be seen that the O–H stretch peak of the water layer adsorbed at low
RH is centered at ∼3230 cm−1 and grows with RH. As RH increases above 30%,
a second peak grows at ∼3400 cm−1 and becomes dominant at RH larger than
60%. These spectral changes indicate that the structure of the adsorbed water layer
formed at low RH is different from the bulk liquid water. Based on the vibration
peak features, it looks more like ice. But there is no other evidence to claim that the
structure of the adsorbed water layer on silicon oxide surface at room temperature
is exactly the same as the homogeneous bulk ice formed at temperature below the
freezing point. Therefore, it would be best to call this structure “solid-like”. In the
literature, one could find this layer being called ice-like, hydration layer, structured
water, etc.
The broad O–H stretch peak can be deconvoluted into these two liquid and
solid-like components to find the relative distribution of the solid-like and liquid
structures in the adsorbed water layer. Figure 5b displays the solid-like and liquid
water component thicknesses along with the total adsorption isotherm thickness.
At low RH region (below 30%), the solid-like structure is dominant and grows up
to ∼3 molecular layers. The thickness increases fast initially; then its growth rate
slows down. In the medium RH range (30–60%), the liquid structure starts growing
slowly although the solid-like structure is still dominant. In this region, the total
thickness increases almost linearly with RH from 3 to 4 monolayers. In the high

RH region (above 60%), the solid-like structure does not grow any more and only
the liquid structure grows rapidly with increase of RH.
One can also investigate the orientation of water molecules in the adsorbed layer
by measuring polarization-dependent vibration peak intensities [30]. In this exper-
iment, the adsorbed water layer is probed with s- and p-polarized IR beams. The
s-polarized beam contains the electric field parallel to the surface. The p-polarized
beam contains both surface parallel and perpendicular components. Since the IR
transition dipole interacts with the electric field in the electromagnetic radiation (IR
beam in this case), the relative intensity ratio of these two polarized IR absorbances
can be used to calculate the angle of the IR transition dipole moment with respect
to the surface normal direction. This ratio is called the dichroic ratio (DR):
absorbance in s − polarization
DR ≡ .
absorbance in p − polarization
In our experimental geometry, DR varies continuously from zero (which is the case
where the IR transition dipole is aligned along the surface normal) to 1.1 (where
the IR transition dipole is aligned parallel to the surface).
Figure 6a compares the ATR-IR spectra obtained with the s- and p-polarizations.
It is easily noticeable that at 5% RH the solid-like water peak at 3220 cm−1 is much
larger in the p-polarization spectrum than in the s-polarization spectrum, yielding a
low DR value. The dichroic ratio estimated for the solid-like water structure peak at
this relative humidity is quite low. In contrast, although it is small, the liquid water
structure peak (3400 cm−1 ) is much more prominent in the s-spectrum at 5% RH
than in the p-spectrum, thus yielding a high DR value.
Figure 6b plots the DR values of the peaks corresponding to the solid-like and
liquid structures as a function of RH. At 5% RH, a DR of 0.4 ± 0.1 was observed
(a) (b)
Figure 6. (a) Polarization-dependent ATR-IR spectra of the adsorbed water layer on the hydrophilic
SiO2 surface as a function of RH. (b) Dichoric ratio of the solid-like (3220 cm−1 ) and liquid
(3400 cm−1 ) structures in the adsorbed water layer [30].
for the solid-like water peak. This low DR value would suggest that the average IR
transition dipole moment was oriented ∼35◦ from the surface normal. An average
dichroic ratio of around 0.7–0.8 was observed for the solid-like water at RH above
50%. For the liquid water, a DR of ∼1 was observed at 5% RH. This high DR value
implies that the average IR transition dipole moment is oriented ∼75◦ from the
surface normal. Again, as the relative humidity is increased, the average dichroic
ratio of 0.7–0.8 is observed for the liquid water structure. Theoretical calculations of
the adsorbed water spectra predict that randomly oriented ice and water structures
will give a DR of 0.7–0.8 [30]. Thus, these results indicate that the water molecules
in the thick layers formed at high RH are randomly oriented while their orientations
deviate from the random orientation in the thin layer formed at low RH.
Another supporting evidence for the deviation from the liquid water structure can
be found in the thermodynamics of the water adsorption process on SiO2 [31]. The
isosteric heat of adsorption, qst (θ ), at a given thickness (θ ) can be calculated from
the isotherm data obtained at various temperatures using the Clausius–Clapeyron
equation. In the low RH region where the solid-like water structure growth domi-
nates, the qst (θ ) value is ∼60 kJ/mol at ∼6% RH, which is much larger than the
heat of vaporization of liquid water (44 kJ/mol). This value is even higher than
the heat of sublimation of ice water (50 kJ/mol). The qst (θ ) value decreases grad-
ually as the adsorbed water layer thickness increases and becomes 43–44 kJ/mol
when the thickness is ∼8 Å. Note that above this thickness, the growth of liq-
uid structure dominates (Fig. 5b). The isosteric entropy of adsorption, S ads (θ ), can
also be calculated from thermodynamic equilibrium relationship between the gas-
phase chemical potential and the heat of adsorption. This calculation shows that
the decrease in the entropy becomes larger as the qst (θ ) increases at low RH. This
trend could be attributed to a more ordered structure due to the increased struc-
tural ordering in the adsorbed layer formed at low RH (Fig. 6). The deviation of
these thermodynamic functions of the structured water layers from the liquid water
values is more prominent on the clean hydrophilic SiO2 surface than on a partially-
methylated hydrophobic surface.
7. Approximation of the Solid-Like Water Contribution to the Nano-asperity

Adhesion
As discussed in the previous sections, the adsorbed water layer on the clean hy-
drophilic silicon oxide surface does not behave like pure liquid water. The O–H
stretching vibration features in ATR-IR spectra reveal that there is the solid-like
structure with some preferential orientation of water molecules very near the sur-
face. Thus, it would be more realistic to model the adsorbed water layer on the
silicon oxide surface at room temperature to be a mixture of the solid-like and liq-
uid structures of water. Their relative amounts should follow the trend shown in
Fig. 5b. Since the solid-like structure is caused by the interactions with the silanol
(a) (b)
Figure 7. (a) Schematic of the adsorbate structure in the continuous neck between the AFM tip and
the substrate at the moment of AFM tip pull-off. (b) Comparison of FSolid + FMeniscus + FvdW with
the experimentally observed RH dependence of the silicon oxide nano-asperity adhesion force [29].
groups at the surface, it is reasonable to assume that the solid-like structure is at the
bottom and the liquid structure grows on top of it [28].
When the adhesion force is measured with the AFM tip on this surface, it can be
assumed that the tip is released from the adsorbed water layer to the free-standing
position, not directly from the solid surface. Therefore, the condition just before the
snap-off process is more likely to be a solid–adsorbate–solid contact system with
the adsorbate layer being in equilibrium with the gas phase. Figure 7a schematically
describes this case where the solid-like structure exists at the center of the capillary
neck between the tip and the substrate. The cross-sectional area of the solid-like
structure region in the neck will increase as RH increases from 0%. As the thickness
of the liquid structure increases at higher RH, the contact area between the solid-
like layers decreases and eventually the liquid property of water governs the pull-off
behavior. In order to model the ATM tip pull-off from this system at the molecular
level, the physical property of the solid-like water should be known; however, this
information cannot be determined experimentally nor is readily available in the
literature. In this situation, one can assume that the solid-like water layer behaves
like ice and estimate the force required to break the solid-like structure contact
present at the center of the neck between the tip and the substrate.
The force required for spontaneous rupture of the solid-like region, FSolid , can
be approximated from the work of cohesion (W ) using the following relationship:
dW 2γice
Asolid
FSolid = ≈ . (7)
dz
z
Here
Asolid is the cross-sectional area of the solid-like structure region at the
moment of AFM tip pull-off and
z is the critical distance over which the solid
contact ruptures. For simplicity, let us assume that the AFM tip is separated from
the adsorbed layer when the bottom of the tip is at the position (z) equal to the
adsorption isotherm thickness, h(RH). Simple geometrical relations give
Aice as
follows:

Aice (h(RH)) = π(h(RH) + 2Rtip )(2hice − h(RH)). (8)
When RH < 45%, hice = h. For RH > 45%, hice is fixed at ∼9.4 Å. Based on
theoretical simulations of spontaneous cleavages [50],
z can be set to be the equi-
librium distance between water molecules:
z ≈ 21/6 σ with σ = Lennard–Jones
parameter for water potential (3.15 Å) ´ [51]. The final expression for F
solid as a
function of h(RH) is written as:
2πγice (h(RH) + 2Rtip )(2hice − h(RH))
FSolid (h(RH)) = . (9)
21/6 σ
Figure 7b plots the sum of FSolid , FMeniscus , and FvdW as a function of RH. The
simple theoretical calculation based on the model containing both the solid-like
and liquid structures in the adsorbed water layer reproduces fairly well the shape
and magnitude of the experimental RH dependence of adhesion. One might notice
that the simulation appears to slightly overestimate the adhesion force at RH >
70%. This might be due to the oversimplification of fixing hice to the h value at
RH 45% and ignoring the stretching effect of the liquid layer present at high RH
during the AFM tip retraction [29]. If the capillary neck of the adsorbed lalyer is
stretched further than the equilibrium isotherm thickness on a flat surface, then the
tip is snapped to the free-standing position farther away from the surface. In this
case, the contribution of the solid-like structure will be smaller than the FSolid value
calculated with equation (9). Overall, the simple model containing both the solid-
like and liquid structures in the adsorbed water layer appears to capture the essence
of the nano-asperity contact in a humid ambient. Further theoretical improvements
of the model are still needed to take into account the exact nature and property of the
solid-state structure, the roughness of the tip and substrate surfaces, the geometric
effects of the AFM tip shape, etc. [52, 53].
8. Conclusions
The structure and isotherm thickness of the adsorbed water layer plays an impor-
tant role in nano-asperity contacts. Previous theoretical calculations of the capillary
force were based on the liquid property of water and the geometry of the AFM tip.
These studies were able to explain, to some degree, the magnitude and shape of
the relative humidity dependence of the nano-asperity silicon oxide adhesion in a
humid ambient; however, the magnitude and exact shape of the relative humidity
dependence were not fully understood yet. This might be due to the oversimplifi-
cation of the water structure in the adsorbed layer. The vibrational spectroscopic
investigations reveal that the adsorbed water layer on the silicon oxide surface cov-
ered with silanol (Si–OH) groups contains both the solid-like and liquid structures.
The solid-like structure exists especially at low relative humidities and can grow
up to 3–4 monolayers before the liquid like structure becomes dominant at high
relative humidities. The simple model taking into account this solid-like structure
can explain the large relative humidity dependence of the nano-asperity adhesion
force measured for hydrophilic silicon oxide surfaces which are completely wetted
by water.
Acknowledgement
The authors acknowledge financial support from the National Science Foundation
(Grant Nos CMS-0408369 and CMMI-0625493).
References
1. S. H. Kim, D. B. Asay and M. T. Dugger, Nano Today 2, 22 (2007).
2. E. A. Vogler, Adv. Colloid Interface Sci. 74, 69 (1998).
3. H. R. Pruppacher and J. D. Klett, Microphysics of Clouds and Precipitation. Kluwer Academic
Publishers, Dordrecht (1997).
4. W. Stumm, L. Sigg and B. Sulzberger, Chemistry of the Solid–Water Interface: Processes at the
Mineral-Water and Particle-Water Interfaces in Natural Systems. Wiley, New York (1992).
5. P. G. de Gennes, Rev. Mod. Phys. 57, 827 (1985).
6. U. Raviv and J. Klein, Science 297, 1540 (2002).
7. U. Raviv, P. Laurat and J. Klein, Nature 413, 51 (2001).
8. J. M. Helt and J. D. Batteas, Langmuir 21, 633 (2005).
9. F. Katsuki, K. Kamei, A. Saguchi, W. Takahashi and J. Watanabe, J. Electrochem. Soc. 147, 2328
(2000).
10. M. L. W. van der Zwan, J. A. Bardwell, G. I. Sproule and M. J. Graham, Appl. Phys. Lett. 64, 446
(1994).
11. W. Maw, F. Stevens, S. C. Langford and J. T. Dickinson, J. Appl. Phys. 92, 5103 (2002).
12. D. B. Asay and S. H. Kim, Langmuir 23, 12174 (2007).
13. L.-S. Fan, Y. C. Tai and R. S. Muller, IEEE Trans. Electron Devices 35, 724 (1988).
14. L.-S. Fan, Y. C. Tai and R. S. Muller, Sensors Actuators 20, 41 (1989).
15. W. S. N. Trimmer and K. J. Gabriel, Sensors Actuators 11, 189 (1987).
16. D. B. Asay, M. T. Dugger and S. H. Kim, Tribol. Lett. 29, 67 (2008).
17. D. M. Tanner, J. A. Walraven, L. W. Irwin, M. T. Dugger, N. F. Smith, W. M. Miller and S. L.
Miller, in: Proc. of IEEE International Reliability Physics Symposium, San Diego, CA, pp. 189–
197 (1999).
18. S. L. Miller, M. S. Rodgers, G. LaVigne, J. J. Sniegowski, P. Clews, D. M. Tanner and K. A.
Peterson, in: Proc. IEEE International Reliability Physics Symposium, San Diego, CA, pp. 17–25
(1999).
19. M. Binggeli and C. M. Mate, Appl. Phys. Lett. 65, 415 (1994).
20. L. Xu, A. Lio, J. Hu, D. F. Ogletree and M. Salmeron, J. Phys. Chem. B 102, 540 (1998).
21. B. Bhushan and C. Dandavate, J. Appl. Phys. 87, 1201 (2000).
22. X. Xiao and L. Qian, Langmuir 16, 8153 (2000).
23. R. Jones, H. M. Pollock, J. A. S. Cleaver and C. S. Hodges, Langmuir 18, 8045 (2002).
24. D. L. Sedin and K. L. Rowlen, Anal. Chem. 72, 2183 (2000).
25. M. Y. He, A. S. Blum, D. E. Aston, C. Buenviaje, R. M. Overney and R. Luginbuhl, J. Chem.

Phys. 114, 1355 (2001).
26. G. Vigil, Z. H. Xu, S. Steinberg and J. Israelachvili, J. Colloid Interface Sci. 165, 367 (1994).
27. H.-J. Butt, Langmuir 24, 4715–4721 (2008).
28. D. B. Asay and S. H. Kim, J. Phys. Chem. B 109, 16760 (2005).
29. D. B. Asay and S. H. Kim, J. Chem. Phys. 124, 174712 (2006).
30. A. L. Barnette, D. B. Asay and S. H. Kim, Phys. Chem. Chem. Phys. 10, 4981 (2008).
31. D. B. Asay, A. L. Barnette and S. H. Kim, J. Phys. Chem. C 113, 2128 (2009).
32. R. N. Lamb and D. N. Furlong, J. Chem. Soc. — Faraday Trans. I 78, 61 (1982).
33. L. T. Zhuravler, Colloids Surfaces A 173, 1 (2000).
34. A. W. Adamson, Physical Chemistry of Surfaces, 5th edn, Chapter 16. John Wiley & Sons, New
York, NY (1990).
35. J. N. Israelachvili, Intermolecular and Surface Forces, Chapter 14. Academic Press, Orlando, FL
(1985).
36. H.-J. Butt, M. Farshchi-Tabrizi and B. Kappl, J. Appl. Phys. 100, 024312 (2006).
37. D. B. Asay, M. P. de Boer and S. H. Kim, J. Adhesion Sci. Technol. To appear.
38. P. A. Thiel and T. E. Madey, Surf. Sci. Rep. 7, 211 (1987).
39. D. Beaglehole and H. K. Christenson, J. Phys. Chem. 96, 3395 (1992).
40. J. J. Yang, S. Meng, L. F. Xu and E. G. Wang, Phys. Rev. B 71, 035413 (2005).
41. P. A. Thiel, F. M. Hoffmann and W. H. Weinberg, J. Chem. Phys. 75, 5556 (1981).
42. B. Mate, A. Medialdea, M. A. Moreno, R. Escribano and V. J. Herrero, J. Phys. Chem. B 107,
11098 (2003).
43. M. Nakamura, Y. Shingaya and M. Ito, Chem. Phys. Lett. 309, 123 (1999).
44. B. W. Callen, K. Griffiths and P. R. Norton, Phys. Rev. Lett. 66, 1634 (1991).
45. G. E. Ewing, J. Phys. Chem. B 108, 15953 (2004).
46. Q. Du, E. Freysz and Y. R. Shen, Science 264, 826 (1994).
47. T. A. Weber and F. H. Stillinger, J. Phys. Chem. 87, 4277 (1983).
48. W. Gan, D. Wu, Z. Zhang, R.-R. Feng and H. F. Wang, J. Chem. Phys. 124, 114705 (2006).
49. D. A. Schmidt and K. Miki, J. Phys. Chem. A 111, 10119 (2007).
50. J. B. Pethica and A. P. Sutton, J. Vac. Sci. Technol. A 6, 2490 (1988).
51. M. W. Mahoney and W. L. Jorgensen, J. Chem. Phys. 112, 8910 (2000).
52. M. P. Goertz, J. E. Houston and X.-Y. Zhu, Langmuir 23, 5491 (2007).
53. J. Jiang, G. Schatz and M. Ratner, Phys. Rev. Lett. 92, 085504 (2004).
Work of Wetting Associated with the Spreading of
Sessile Drops
C. W. Extrand ∗
Entegris, Inc., 3500 Lyman Boulevard, Chaska, Minnesota 55318, USA
Abstract
In this theoretical study, the work done on a sessile drop during spreading was estimated. It was found that
the energy required to stretch the contact line is much greater than the energy needed to stretch the air–
liquid interfacial area. The model shows that wetting energies are relatively small for large contact angles,
but increase dramatically as contact angles tend towards zero. For a given drop volume, more work is needed
to spread higher surface tension liquids than lower surface tension ones. Similarly, larger drops require more
energy to spread than smaller ones. The work of wetting estimated here for sessile drops is comparable to
energies from other wetting geometries, such as capillary bridge and sphere tensiometry. This work provides
theoretical support for the experimental observation that interactions at the contact line dominate the wetting
behavior of spreading sessile drops.
Keywords
Wetting, contact angle, sessile drop, work of wetting, wetting energies
1. Introduction
Wetting of solids by liquids is fundamental to many natural and industrial processes

[1–4]. In nature, wetting is central to repellency of plant leaves and bird feathers,
the ability of insects to walk on water, capillary rise and transpiration in trees, just to
name a few. In industry, wetting plays an important role in coating, adhesion, heat
transfer, drying, separations, filtration and repellency, as well as gauging cleanli-
ness of surfaces. An extensive wetting literature is a testament to its importance.
Hundreds of papers have been published on this topic in the last five years. Since
2003, more than 250 articles with “wetting” in their title have appeared in Langmuir
alone.
The sessile drop method is the most commonly used means of evaluating wetta-
bility. It is depicted in Fig. 1. A small drop of volume V is extruded from a syringe
or pipette and deposited on a solid horizontal surface. The drop spontaneously
* Tel.: 1-952-556-8619; e-mail: chuck_extrand@entegris.com

82 C. W. Extrand
Figure 1. A small liquid drop spreads over a horizontal solid surface. (a) Side view of the small
spherical liquid drop of volume (V ) prior to deposition. (b) Side view of the drop spreading on the
surface. (c) Side view and (d) plan view of the drop after spreading has ceased. S is the air–liquid
interfacial area and C is the length of the contact line.
spreads. As it does, the contact angle (θ ) decreases to a critical advancing value

(θa ), where spreading ceases. For smooth, flat surfaces, θa values typically fall be-
tween 120◦ and 5◦ , depending on the chemical composition of the liquid and solid.
Roughening of surfaces can extend this range. Surfaces with specially designed
structures can show extreme wettability. Super-wetting surfaces can approach the
lower limiting value of 0◦ . At the other end of the spectrum, the apparent contact
angles of super-lyophobic surfaces can nearly reach the upper limit of 180◦ .
Somewhat ironically, even though spreading of sessile drops is arguably the most
widely studied wetting phenomenon, there is no universal agreement on the under-
lying physico-chemical mechanisms. The most commonly employed construct for
analyzing the spreading of liquid drops is attributed to Thomas Young. In Young’s
paper of 1805 [5], there is no equation, but a description of a balance of forces or
interfacial tensions at the contact line. Since that time, clever scientists have used
energy concepts developed many decades after the publication of Young’s seminal
paper to derive a thermodynamic version of the mechanical force balance at the
contact line,
γl cos θ = γsl − γs , (1)
where γl is the liquid surface tension, γs is the solid surface tension and γsl is the
liquid–solid interfacial tension. Modern derivations of the Young’s equation [2, 6,
7], as well as derivations of the closely related Wenzel and Cassie equations [8,
9] for rough or structured surfaces, usually assume that change in interfacial areas
largely controls wetting.
Work of Wetting Associated with the Spreading of Sessile Drops 83
While thermodynamic analysis of interfacial areas may be useful for developing

theoretical models that describe various wetting phenomena, Pease stated more than
60 years ago that interactions at the contact line, not the contact area, drive wet-
ting [10]. Simple experiments have demonstrated that the liquid–solid interfacial
area under sessile drops has little or no influence [11–13]. There are other exam-
ples that support the supposition that forces at the contact line dominate interfacial
interactions. Contact angles measured with the same liquid–solid pair generally are
geometry independent [14–18]. If the composition and roughness of surfaces are
identical, measurements of contact angles from immersed plates, fibers or capillary
tubes are the same as those from sessile drops. Similarly, large sessile drops that are
distorted by gravity give the same contact angle as small spherical ones. If liquid is
withdrawn from sessile drops, the shape of the gas–liquid interface almost always
is distorted before the contact line budges. Small particles floating on the surface
of liquids can distort the gas–liquid interface to create a meniscus with little or no
movement of the contact line.
If indeed the interactions in the immediate vicinity of the contact line dictate how
far a liquid will spread, then it is anticipated that the energy required to expand the
contact line of a spreading drop should be much greater than that associated with
expansion of gas–liquid or liquid–solid interface area. Thus, in this study, the work
done on a liquid during spreading of sessile drops was estimated and compared to
wetting energies from previously published experimental findings.
2. Analysis
Consider the spreading of a sessile drop shown in Fig. 1. Assume the solid surface
is smooth, homogeneous and rigid. The extent of spreading is determined by a
competition between adhesive and cohesive forces. Molecular interactions between
the liquid and solid in the vicinity of the contact line cause the liquid to spread.
These spontaneous interactions stretch the liquid, pulling it down onto the solid.
Contractile forces arising from the liquid surface tension (γl ) try to minimize its
surface area and contact line. At the point spreading ceases, the drop exhibits an
advancing contact angle, θa .
If brought into close proximity and if sufficiently compliant, it is a natural ten-
dency of materials to spontaneously bond [19]. The universal forces that cause the
tip of an atomic force microscope to spontaneously jump into contact with an adja-
cent surface are the same physico-chemical forces that drive wetting [20].
Figure 2 depicts a nanoscopic view of the contact line, where the molecules of
the liquid are represented as white circles and those of the solid as black circles. In
wetting, these attractive forces operate in a small zone (a few nanometers or less)
just in front of the contact line. This region is depicted in Fig. 2 by cross-hatching.
Once a liquid–solid bond has formed, the molecules at the liquid–solid interface are
no longer in play. (This is in contrast to the more traditional view that movement of
liquid molecules from the bulk to liquid–solid interface drives wetting.) It should
84 C. W. Extrand
Figure 2. A side view of the contact line. (a) A macroscopic view of the contact line of a sessile drop.
(b) A nanoscopic view of the contact line. The white circles represent molecules of the liquid and the
black circles represent molecules of the solid.
also be noted that the spontaneous interactions between a liquid and solid that drive
spreading are not included in this analysis; the change in the free energy of solid
surfaces due to spreading of liquids has been analyzed elsewhere, assuming wetting
can be treated as an adsorption process [21]. In the analysis that follows, the work
done on the liquid is examined.
2.1. Lengths and Areas of Spreading Drops
Assume that the liquid drops considered here are sufficiently small that gravity
does not distort their shape, but not so small that “line tension” [22–24] affects
their contact angle. If a drop retains spherical proportions as it spreads, then simple
expressions can be derived that describe its contact circumference (C) and its gas–
liquid interfacial area (S) in terms of volume of the drop (V ) and contact angle (θ ).
If drop volume remains constant during spreading, then the contact circumference
(C) can be computed,

sin θ
C = 2(3π ) V
2 1/3 1/3
. (2)
(2 − 3 cos θ + cos3 θ )1/3
Similarly, the gas–liquid interfacial area (S) can be calculated as [25]

(1 − cos θ )
S = 2(9π)1/3 V 2/3 . (3)
(2 − 3 cos θ + cos3 θ )2/3
2.2. Work Associated with Spreading Drops
The total work done in spreading (wt ) of a sessile drop is a combination of the
work to stretch the contact line (wc ) as well as the work to stretch the gas–liquid
interface (ws ),
wt = wc + ws . (4)
Note that the total work of spreading contains only two terms. The solid sur-
face has been ignored, because the focus here is the work done on the liquid. The
liquid–solid interfacial area is not explicitly included either. However, expansion of
the contact line during spreading creates liquid–solid interfacial area. The conse-
quences of this intentional omission of the work associated liquid–solid interfacial
area are discussed later.
The work done in stretching the contact line (wc ) can be estimated by integrating
the force at the contact line (fc ) as the circumference length (C) of the spreading
drop expands,
wc = fc · ∂C. (5)
Similar forms of this simple expression have been used in textbook derivations of
the Laplace equation [26] as well as in analyses of sphere tensiometry [27].
The contact line force, which acts in the plane of the solid surface, is the product
of the liquid surface tension (γl ) and C [28],
fc = γl · C. (6)
By combining equations (5) and (6), we obtain a differential equation describing
work of wetting at the contact line in terms of γl and C,
wc = γl · C · ∂C. (7)
It is preferable to frame the expression for work in terms of contact angles. To do
so, equation (7) for C is differentiated with respect to θ ,
∂C = 2(3π2 )1/3 V 1/3 (8)

cos θ sin4 θ
× − · ∂θ,
(2 − 3 cos θ + cos3 θ )1/3 (2 − 3 cos θ + cos3 θ )4/3
and then incorporated into equation (7) along with equation (2) to give,
wc = 4(3π2 )2/3 γl · V 2/3 (9)

θa
sin θ cos θ sin5 θ
× − · ∂θ.
π (2 − 3 cos θ + cos3 θ )2/3 (2 − 3 cos θ + cos3 θ )5/3
Integrating equation (9) from θ = π (180◦ ) to θ = θa yields the work of wetting at
the contact line for the spreading of a sessile drop (wc ) in terms of θa , γl and V ,

(1 + cos θa ) sin2 (θa /2)
wc = 8π(18π) γl · V
1/3 2/3
. (10)
(8 − 9 cos θa + cos 3θa )2/3
The work to stretch the gas–liquid interface can be determined from the following
expression [2],
ws = γl · ∂S. (11)
By differentiating equation (3) with respect to θ and combining with equation (11),
we obtain the following equation,
ws = 2(9π)1/3 γl · V 2/3 (12)

θa
sin θ 2(1 − cos θ ) sin3 θ
× − · ∂θ,
π (2 − 3 cos θ + cos3 θ )2/3 (2 − 3 cos θ + cos3 θ )5/3
86 C. W. Extrand
that can be integrated from π to θa to obtain the work of wetting done on the gas–
liquid interface during the spreading of a sessile drop (ws )

sin2 (θa /2) 1
ws = 8(18π) γl · V
1/3 2/3
− . (13)
(8 − 9 cos θa + cos 3θa )2/3 (16)2/3

3.1. Lengths and Areas of Spreading Drops
Figure 3 shows dimensionless values of contact circumference (C) and gas–liquid
interfacial area (S) as functions of contact angle (θ ), where C has been divided
by V 1/3 and S by V 2/3 . As a small drop of constant volume begins to spread, C
increases gradually from θ = 180◦ to 90◦ and then rises asymptotically as θ tends
toward 0◦ . In contrast, the gas–liquid interfacial area initially decreases. Values fall
from S/V 2/3 = 4.84, reach a minimum at θ = 90◦ and then begin to climb. S/V 2/3
reaches it original value of S/V 2/3 = 4.84 again at θ = 42◦ . As θ tends toward 0◦ ,
S/V 2/3 values also increase asymptotically. At θ = 10◦ , the surface area of a drop
is more than 2 12 times its original area, S/V 2/3 = 11.87.
3.2. Forces at the Contact Line
One of the underlying suppositions of this paper is that a “line tension” exists along
the three-phase contact line. It is not the minuscule line tension suggested by the
modified Young equation [22–24] that is intended to explain and adjust for vari-
ations in the contact angles of very small drops, which reportedly has a value of
10−11 –10−6 N [23]. Rather, it is much larger force and it is directly related to the
surface tension of the liquid, γl [28]. This “line tension” or contact line force is the
product of γl and C, per equation (6). For a small water drop (V ≈ 1 µl), it is on the
order of 10−4 –10−1 N.
Figure 3. Contact circumference (C) and gas–liquid interfacial area (S) and as a function of contact
angle (θ ).
While some may find this idea novel for sessile drops, it is a concept that is
employed widely in other wetting geometries. The driving force for capillary rise in
a tube is the vertical component of this contact line force, which is the product of γl
and length of the contact line on the inner diameter of the tube [2]. The capillary
force in sphere tensiometry or sphere flotation is also the vertical component of the
contact line force, again the product of γl and the length of the contact line [29–31].
3.3. Work Associated with Spreading Drops
Since work is a form of energy, the SI units associated with the work of spreading
are Joules (J). On an absolute basis, wc , ws and wt all increase with the surface
tension of the liquid (γl ) and the volume (V ) of the drop. All else being equal,
a greater surface tension means a larger force will be required to stretch the contact
line and gas–liquid interface of a spreading drop. Consequently, more work will be
expended in spreading higher γl liquids than liquids with lower γl . Similarly, the
larger the drop, the more energy is required to stretch the contact circumference
and gas–liquid interface.
It is customary to report the work affiliated with capillary phenomena in terms
of energy per area, such as mJ/m2 [2]. Both wc and ws can be framed as energy per
area by simply dividing both side of equations (10) and (13) by V 2/3 . These ener-
gies can be further normalized into a dimensionless form by dividing wc and ws by
γl and V 2/3 . Figure 4 shows these dimensionless works of wetting (wi /γl V 2/3 ) plot-
ted against advancing contact angles (θa ), where the subscript c represents the work
associated with stretching of the contact line, the subscript s symbolizes the work
associated with stretching of the gas–liquid interface and the subscript t, the total
work.
Consider the work associated with stretching of the contact line, wc /γl V 2/3 . En-
ergy must be put into the liquid over the entire range of θa values. Thus, values of
wc /γl V 2/3 (as well as wc /V 2/3 and wc values) are all positive. Work at the contact
line increases modestly over the range of lyophobic contact angles and then rises
dramatically for lyophilic values, θa < 30◦ .
The work of stretching the gas–liquid interface (ws ) follows a different path. As a
drop spreads and θa decreases, ws /γl V 2/3 values initially decrease from zero, which
implies a spontaneous process. Values fall to a minimum of ws /γl V 2/3 = −1.0
at θa = 90◦ and then climb back up to ws /γl V 2/3 = 0 at θa = 42◦ . As θa tends
toward 0◦ , ws /γl V 2/3 becomes positive and begins to rise steeply where θa < 15◦ .
If the gas–liquid interface completely controlled wetting of sessile drops, it could be
argued that no smooth surface would ever exhibit contact angles greater than 90◦ .
Figure 4 also includes the sum of the dimensionless work to stretch both the con-
tact line and the gas–liquid interfacial area, wt /γl V 2/3 from equation (4). The work
done at the contact line completely dominates the work done at the free surface of
the liquid. The contribution of the interfacial area to the total work of spreading
is so small that wt /γl V 2/3 is barely distinguishable from wc /γl V 2/3 . Over most of
the range of contact angles, ws /γl V 2/3 values were less than 15% of wt /γl V 2/3 ;
88 C. W. Extrand
Figure 4. Dimensionless work of wetting (wi /γl V 2/3 ) associated with the spreading of a sessile
drop versus advancing contact angle (θa ), where the subscript c represents the work associated with
stretching of the contact line, the subscript s symbolizes the work associated with stretching of the
gas–liquid interface and the subscript t, the total work. Work of wetting values were calculated from
equations (10), (13) and (4).
between θa = 70◦ and 25◦ , they amounted to less than 5%. Thus, these findings
support the supposition that the energy associated with stretching of the contact
line dominates the spreading of sessile drops.
On an absolute basis, wt values depend on drop size. For a 1 µl water drop, the
magnitude of the total work of wetting extends from wt = 500 nJ for hydrophobic
poly(tetrafluoroethylene) (PTFE) (θa = 112◦ ) to wt = 9.6 µJ for hydrophilic quartz
(θa ≈ 5◦ ). If the volume of the water drop is decreased by an order of magnitude to
V = 0.1 µl, then wt for PTFE is less, 110 nJ. If the volume drop is larger, say 10 µl,
then wt is larger, 2340 nJ.
Ample evidence exists that movement of molecules from the bulk to the liquid–
solid interface does not play a prominent role in wetting. Capillary rise in a tube [2],
retention of liquid drops by inclined surfaces [32] and flotation of particles [33] all
can be explained by balancing the capillary forces due to the liquid surface tension
acting at the contact line against gravitational forces acting on the mass of the liq-
uid. None of these requires inclusion of interactions between the liquid and wetted
solid. Estimates of liquid surface tension from du Noüy ring measurements do not
require knowledge of the interfacial area [2]. Nor do estimates of contact angles
from tensiometry measurements, such as Wilhelmy plate. Likewise, the strength of
capillary bridges between particles as well as the shape and maximum size of pen-
dent drops can be quantified by tallying the forces arising from the liquid surface
tension at the contact line and the gas–liquid interface [7, 34]. The wetted area does
not enter into the model presented here either. (Had the work associated with the
creation of liquid–solid interface been included, its contribution would have been
rather small. This point is discussed further in the Appendix.)
3.4. Estimated Work of Wetting from Sessile Drops for Real Surfaces
Some wetting energies were calculated using the proposed model. Table 1 lists
advancing contact angles and work of wetting values for various liquid–polymer
combinations. Experimental values of the liquid surface tension (γl ) and advancing
contact angles (θa ) from sessile drop measurements were taken from the literature
[1, 35, 36]. Work of wetting values were calculated from equations (10) and (13).
Again, note that ws values were relatively small (and negative) as compared to wc
values for these smooth lyophobic polymers. Dimensionless values of the total work
of wetting (wt /γl V 2/3 ) for these surfaces range between 7 and 22. The correspond-
ing area-based energies (wt /V 2/3 ) span more than 600 mJ/m2 , from 510 mJ/m2 to
1140 mJ/m2 .
Consider the first series of data in Table 1 for a variety of liquids on poly
(tetrafluoroethylene) (PTFE). PTFE is the quintessential non-stick material and is
best known as Teflon® . It is used widely in corrosive environments due of its broad
resistance to chemical attack. Because of its inertness, PTFE is expected to interact
weakly and non-preferentially with liquids. Even though surface tensions of this
series of liquids vary by 250% from the most polar to the least polar and contact
angles ranged from 112◦ to 51◦ , the area-based work of wetting (wt /V 2/3 ) differs
only a little from one liquid to the next. The values of total work of wetting for
PTFE lie within a narrow band between 510 mJ/m2 and 630 mJ/m2 . The two non-
polar liquids, diiodomethane and hexadecane, had slightly larger wt /V 2/3 values
than the polar liquids. This makes sense from the perspective of the PTFE surface.
The old adage that “likes attract likes” applies here. With a chemical nature closer
to PTFE than the other liquids, diiodomethane and hexadecane might be expected
to interact a bit more strongly.
Next, let us turn to the second set of data in Table 1 where work of wetting has
been calculated for water on a series of polymers surfaces. The polymers range from
the relatively non-polar PTFE and polyethylene (PE) to the more polar polystyrene
(PS) and poly(ethylene terephthalate) (PET). As interactions between the polar
water and the polymer surfaces increase, lower contact angles are observed. Ac-
cordingly, wt /V 2/3 values for water are more than double, from 510 mJ/m2 for
PTFE to 1140 mJ/m2 for PET.
The polymers surfaces listed in Table 1 are relatively lyophobic. Work of wet-
ting values were also estimated for a lyophilic surface, such as that of silica or
quartz. If free of organic contamination, these silanol-laden surfaces are usually
water wettable. Assuming that θa ≈ 5◦ , the work of wetting for water is estimated
to be wt /V 2/3 = 9.6 J/m2 .
3.5. Comparison of Energies from Other Wetting Geometries
Other wetting geometries show comparable energies. Similar to the method em-
ployed in solid mechanics [37], the work done in the formation of capillary bridges
can be estimated from the area under experimental force–deflection curves. Work of
wetting values extracted from capillary force measurements for small water drops
90
Table 1.
Contact angles and calculated work of wetting values for various liquid–polymer combinationsa
Liquid γl Polymer θa wc /γl V 2/3 wc /V 2/3 ws /γl V 2/3 ws /V 2/3 wt /γl V 2/3 wt /V 2/3
(mN/m) (◦ ) (mJ/m2 ) (mJ/m2 ) (mJ/m2 )
Water 73 PTFE 112 7.79 570 −0.873 −64 6.92 510

−0.937 −59
C. W. Extrand
Glycerol 63 PTFE 105 9.08 570 8.14 510
Formamide 56 PTFE 91 11.8 660 −1.00 −56 10.8 610
Diiodomethane 51 PTFE 85 13.1 670 −0.99 −50 12.1 620
Ethylene glycol 48 PTFE 87 12.70 610 −0.99 −47 11.7 560
Hexadecane 28 PTFE 51 22.79 640 −0.38 −11 22.4 630
Water 73 PTFE 112 7.79 570 −0.87 −64 6.92 510
Water 73 PE 95 11.0 800 −0.99 −72 10.0 730
Water 73 PS 84 13.3 970 −0.99 −72 12.3 900
Water 73 PET 71 16.5 1200 −0.88 −64 15.6 1140
a Experimental values of the liquid surface tension (γ ) [1, 35] and advancing contact angles (θ ) [36] from sessile drop measurements were taken from
l a
the literature. Work of wetting values were calculated from equations (10), (13) and (4).
between two flat hydrophobic surfaces have magnitudes of tens of nJ [38]. If these
values are normalized, they yield wt /γl V 2/3 values that are roughly equivalent to
those calculated here for the spreading of sessile drops. Immersion of a sphere into
a liquid is analogous to the spreading of a sessile drop. As liquid engulfs the sphere,
the contact line becomes longer and, accordingly, the capillary force increases.
Force–deflection curves from tensiometry measurements with small spheres also
show comparable work of wetting values [27].
4. Conclusions
The work done in the spreading of sessile drops was estimated. It was found that the
energy required to stretch the contact line is much greater than the energy needed
to stretch the air–liquid interfacial area. Wetting energies are relatively small for
large advancing contact angles, but increase dramatically as contact angles tend
towards zero. Higher γl liquids require more work to spread than lower γl ones.
Similarly, larger drops require more work than smaller ones. The work of wetting
values estimated here are comparable to those from other wetting geometries. Thus,
it seems that the energy associated with stretching of the contact line does indeed
dominate over the energy associated with stretching of interfacial areas.
Acknowledgements
I thank Entegris management for supporting this work and allowing publication.
Also, thanks to M. Acevedo, L. Monson, S. I. Moon, S. Moroney, J. Pillion and
T. King for their suggestions on the technical content and text as well as F. Gadala-
Maria for telling me about Faraday’s juvenile lectures.
References
1. S. Wu, Polymer Interface and Adhesion. Marcel Dekker, New York (1982).
2. A. W. Adamson, Physical Chemistry of Surfaces, 5th edn. Wiley, New York (1990).
3. J. C. Berg (Ed.), Wettability. Marcel Dekker, New York (1993).
4. P. C. Hiemenz and R. Rajagopalan, Principles of Colloid and Surface Science, 3rd edn. CRC
Press, Boca Raton, FL (1997).
5. T. Young, Phil. Trans. Roy. Soc. (London) 95, 65–85 (1805).
6. B. W. Cherry, Polymer Surfaces. Cambridge University Press, New York (1981).
7. J. N. Israelachvili, Intermolecular and Surface Forces. Academic Press, New York (1992).
10. D. C. Pease, J. Phys. Chem. 49, 107–110 (1945).
11. F. E. Bartell and J. W. Shepard, J. Phys. Chem. 57, 455–458 (1953).
14. R. E. Johnson, Jr. and R. H. Dettre, in: Surface and Colloid Science, E. Matijević (Ed.), pp. 85–
153. Wiley, New York (1969).
92 C. W. Extrand
15. R. E. Johnson, Jr., R. H. Dettre and D. A. Brandreth, J. Colloid Interface Sci. 62, 205–212 (1977).
16. M. Morra, E. Occhiello and F. Garbassi, J. Colloid Interface Sci. 149, 84–91 (1992).
17. J. E. Seebergh and J. C. Berg, Chem. Eng. Sci. 47, 4468–4470 (1992).
18. L. M. Lander, L. M. Siewierski, W. J. Brittain and E. A. Vogler, Langmuir 9, 2237–2239 (1993).
19. M. Faraday (Ed.), The Forces of Matter, Delivered before a Juvenile Auditory at the Royal Insti-
tution of Great Britain during the Christmas Holidays of 1859–1860. Lecture II — Gravitation–
Cohesion. P. F. Collier and Son, New York (1909–1914).
22. R. J. Good and M. N. Koo, J. Colloid Interface Sci. 71, 283–292 (1979).
23. A. Marmur, J. Colloid Interface Sci. 186, 462–466 (1997).
24. R. David and A. W. Neumann, Langmuir 23, 11999–12002 (2007).
26. P. W. Atkins, Physical Chemistry, W. H. Freeman, San Francisco (1982).
27. O. Pitois and X. Chateau, Langmuir 18, 9751–9756 (2002).
28. V. R. Gray, Chemistry and Industry, 969–978 (1965).
29. G. D. Yarnold, Proc. Phys. Soc. 58, 120–125 (1946).
30. A. D. Scheludko and D. Nikolov, Colloid Polym. Sci. 253, 396–403 (1975).
31. R. Gunde, S. Hartland and R. Mader, J. Colloid Interface Sci. 176, 17–30 (1995).
32. K. Kawasaki, J. Colloid Sci. 15, 402–407 (1960).
33. H. J. Schulze, Physico-Chemical Elementary Processes in Flotation. Elsevier, Amsterdam (1983).
34. Y. I. Rabinovich, M. S. Esayanur and B. M. Moudgil, Langmuir 21, 10992–10997 (2005).
35. R. C. Weast (Ed.), Handbook of Chemistry and Physics. CRC, Boca Raton, FL (1992).
36. J. R. Dann, J. Colloid Interface Sci. 32, 302–320 (1970).
37. J. M. Gere and S. P. Timoshenko, Mechanics of Materials. PWS-Kent, Boston (1984).
38. E. J. De Souza, L. Gao, T. J. McCarthy, E. Arzt and A. J. Crosby, Langmuir 24, 1391–1396 (2008).
Appendix
Contact area and hypothetical work of the contact area. The liquid–solid interfacial
area (A) of a small, sessile drop also can be calculated from V and θ ,

sin2 θ
A = (9π) V 1/3 2/3
. (A1)
(2 − 3 cos θ + cos3 θ )2/3
By differentiating equation (A1) with respect to θ , combining with the following
equation,
wa = γsl · ∂A, (A2)
and then integrating from π to θa , an expression is obtained for estimating the work
of wetting done at the liquid–solid interface during the spreading (wa ),

sin2 θa
wa = 2(18π) γsl · V
1/3 2/3
. (A3)
(8 − 9 cos θa + cos 3θa )2/3
Figure 5 shows the liquid–solid interfacial contact area and the associated work
of wetting in dimensionless form, A/V 2/3 and wa /γsl V 2/3 . Both follow the same
Figure 5. Liquid–solid interfacial contact area (A) and the associated work of wetting (wa ).
path. At θa = 180◦ , both A/V 2/3 and wa /γsl V 2/3 = 0. As θa decreases with greater
spreading, A/V 2/3 and wa /γsl V 2/3 increase modestly until θa ≈ 80◦ , then rise as-
ymptotically as θa tends toward 0◦ .
It is the supposition of this study that stretching of the contact line dominates
the work done on the liquid during spreading. Stretching of the contact line cre-
ates liquid–solid interfacial area. Thus, including both wc and wa in the total
work of wetting (wt ) would have amounted to double counting. That said, if the
liquid–solid interfacial work of wetting (wa ) had been included, its contribution
to wt would have been rather insignificant as compared to work done in stretch-
ing the contact line (wc ). Over the full range of θa values, all else being equal,
wa /γsl V 2/3 ≈ 0.16wc /γl V 2/3 . Moreover, since the magnitude of γsl is usually
much less than γl for a given liquid–solid pair, the magnitude of wa would be further
diminished relative to wc .
Effect of Line Tension on Work of Adhesion for
Rock–Oil–Brine Systems
Dayanand Saini and Dandina N. Rao ∗

Craft & Hawkins Department of Petroleum Engineering, Louisiana State University, Baton Rouge,
LA 70803-6417, USA
Abstract
In a recent experimental study, the concept of line tension was evaluated for rock–oil–brine systems
at reservoir conditions to characterize the rock–oil adhesion interactions. The study concluded that line
tension-based modification of the Young’s equation provides a more realistic description of rock–oil adhe-
sion interactions at reservoir conditions. Based on these experimental findings, it is suggested that the effect
of line tension should be incorporated in the calculation of the work of adhesion for better understanding of
the effect of rock–oil adhesion interactions on oil recovery from petroleum reservoirs. In this present study,
we report the computed work of adhesion for different rock–live oil–brine systems after including the effect
of line tension.
The new equation for the work of adhesion suggests that the conventional approach of determining the
work of adhesion for rock–live oil–brine systems underestimates the impact of rock–oil adhesion interac-
tions on residual oil recovery. The new method of computing the work of adhesion indicates that the work of
adhesion is much stronger than predicted conventionally, and that more and more work needs to be exerted
in order to displace the residual oil from pore space as its saturation decreases.
Keywords
Line tension, rock–oil adhesion, wettability, oil–water interfacial tension, equilibrium contact angle, dy-
namic contact angle, water advancing contact angle, work of adhesion
Nomenclature
θ∞ Equilibrium contact angle

cos θ∞ Cosine of equilibrium contact angle
θd Dynamic contact angle, measured in drop volume reduction experiment
cos θd Cosine of dynamic contact angle, measured in drop volume reduction ex-
periment
* To whom correspondence should be addressed. Tel.: 225-578-6037; Fax: 225-578-6039; e-mail:

dnrao@lsu.edu

96 D. Saini and D. N. Rao
θA The equilibrium water advancing contact angle {Obtained from dual-drop–

dual-crystal (DDDC) technique}
σ Line tension
r Radius of contact line
γOW Free energy of oil–water interface or oil–water interfacial tension
γSO Free energy of solid–oil interface or solid–oil interfacial tension
γSW Free energy of solid–water interface or solid–water interfacial tension
WSOW Work of adhesion in solid–liquid–liquid (rock–oil–brine) systems
Abbreviations
EOR Enhanced oil recovery
S–L–L Solid–liquid–liquid
S–L–V Solid–liquid–vapor
IFT Interfacial tension
DDDC Dual-drop–dual-crystal
DSA Drop shape analysis
1. Introduction
Rock–oil adhesion interactions play a vital role in oil recovery from petroleum
reservoirs. The extent of adhesion of oil to rock surface determines the recovery
efficiency of any enhanced oil recovery (EOR) process. One of the objectives of
these processes involving injection of external fluids should be to overcome the
rock–oil adhesion for increased oil production. It can be achieved by two ways:
1. by displacing the oil (e.g. by water flooding); and 2. by interacting with reservoir
fluids and rock surface with consequent displacement of oil (e.g. by CO2 injection,
surfactant flooding). Understanding of rock–oil adhesion phenomenon is important
in the success of any EOR process because it enables the development of means
to overcome this rock–oil adhesion in order to achieve substantial increase in oil
recoveries.
In petroleum engineering, “wettability” is used to describe the overall effect of
different surface and interfacial (rock–fluids and fluid–fluid) interactions in petro-
leum reservoirs. These surface and interfacial interactions affect the movement of
different fluids within the reservoir. The Young’s equation is widely used for char-
acterizing reservoir wettability in terms of measurable quantities, i.e., oil–water
interfacial tension and contact angle. A detailed treatment of concept and charac-
Line Tension 97
terization of contact angles in solid–liquid–liquid (S–L–L) systems is presented by

Rao [1].
In the case of S–L–L systems, these surface and interfacial interactions, as de-
scribed by “wettability”, are comprised of two components, namely, spreading and
adhesion. Adhesion is defined as the molecular force exerted across an interface
of contact between unlike liquids or solids that resists interfacial separation [2].
The necessary work required to attain this interfacial separation is expressed by the
work of adhesion. For rock–oil–brine systems, due to lack of any standard method
to measure rock–oil adhesion interactions, at both ambient and reservoir conditions,
qualitative adhesion tests are used to evaluate the effect of rock–oil adhesion inter-
actions on oil recovery. The procedure and results of such qualitative adhesion tests
have been reported in the literature [2–4].
In our recent experimental work [5], the concept of line tension-based modifica-
tion of the Young’s equation was evaluated for rock–oil–brine systems at reservoir
conditions of elevated pressure and temperature to characterize the rock–oil ad-
hesion in S–L–L (rock–oil–brine) systems. Linear relationship found in our study
between the line tension and the adhesion number [5] substantiates that line ten-
sion correlates well with the adhesion phenomenon in rock–oil–brine systems. The
present study attempts to estimate the work of adhesion for several rock–live oil–
brine systems at reservoir conditions after including the line tension effect, instead
of computing the work of adhesion using conventional approach which ignores the
extent of rock–oil adhesion interactions present in the system.
2. Work of Ahdesion in Rock–Oil–Brine Systems

Adhesion, a naturally occurring phenomenon, is of practical importance in EOR
processes. Displacement of oil through pore space in the rock matrix is analogous to
separation of an oil drop from a solid surface in presence of water. In this case, two
new unit interfaces i.e. oil–water and solid–water are formed and solid–oil interface
is eliminated. The work required for creating these two new unit interfaces while
vanishing one of the existing unit interface is defined as “work of adhesion”. Rao
and Maini [4] discussed the rock–oil adhesion, the work of adhesion for rock–oil–
brine systems and their impact on reservoir mechanics in detail.
For rock–oil–brine system, the work of adhesion, WSOW is expressed as:
WSOW = (γSW + γOW ) − γSO or

WSOW = γOW − (γSO − γSW ). (1)
In equation (1), oil–water interfacial energy or interfacial tension (γOW ) is the only
parameter that can be measured conveniently. The other two terms can be expressed
in terms of measurable parameters using the Young’s equation. The Young’s equa-
tion provides an equilibrium relationship between different interfacial energies and
contact angle.
Figure 1. Concept of contact angle for S–L–L (rock–oil–water) systems.
For rock–oil–brine systems, the Young’s equation is:

(γSO − γSW ) = γOW cos θ∞ . (2)
Here, θ∞ is the equilibrium contact angle or the Young’s contact angle. A schematic
description of the equilibrium relationship as described by the Young’s equation is
shown in Fig. 1.
Combining equation (1) and equation (2), we obtain:
WSOW = γOW (1 − cos θ∞ ). (3)
Equation (3) does provide the basic equation for the work of adhesion involving
measurable quantities, the oil–water interfacial tension (IFT) and equilibrium con-
tact angle (θ∞ ), which can be measured for rock–oil–brine systems at elevated
pressure and temperature conditions using an optical cell apparatus.
3. Work of Ahdesion and Modifed Young’s Equation for Rock–Oil–Brine

Systems
Equation (3) is particularly valid for strongly water-wet systems. In qualitative
adhesion tests [2, 3], for strongly water-wet cases, oil drop can be completely sepa-
rated from rock surface without leaving any oil behind by sucking the fraction of oil
drop volume using an injector needle in several steps. The schematic representation
of this situation is shown in Fig. 2.
In case of weak or no rock–oil adhesion interactions, on stepwise reduction of
drop volume, the dynamic contact angle (θd ) subtended by the remaining oil drop
to the rock surface exhibits a small deviation from the equilibrium contact angle
(θ∞ ) value for a significant change in contact radius. The length of contact radius
is reduced to zero when no fraction of original oil drop is left on the rock surface in
stepwise reduction of drop volume. In this situation, the dynamic contact angle (θd )
does not differ significantly from the equilibrium water advancing contact angle
(θA ), representative of true wettability state of the system i.e. the extent of rock–
Line Tension 99
Figure 2. Schematic representation of oil drop separation from rock surface in strongly water-wet
S–L–L (rock–oil–water) systems at reservoir conditions.
oil adhesion interactions present in the system, as observed in case of water-wet

systems [5].
However, in other cases, ranging from intermediate-wet to strongly oil-wet sys-
tems, when volume of oil drop is reduced in steps, oil drop may adhere partially
to rock surface leaving a fraction of the original oil drop on the rock surface. This
situation is shown schematically in Fig. 3.
In the presence of strong rock–oil adhesion interactions, on stepwise reduction of
drop volume, the dynamic contact angle (θd ) subtended by the remaining oil drop to
the rock surface exhibits a large deviation from the equilibrium contact angle (θ∞ )
value for a small change in contact radius. The dynamic contact angle (θd ) in such
cases differs significantly from the equilibrium contact angle (θ∞ ) and approaches
towards the equilibrium water advancing contact angle (θA ), as observed in the
cases of intermediate-wet and oil-wet systems [5].
In such situations, equation (3) is inadequate as it does not take into account the
strong rock–oil adhesion interactions present in the system.
In case of very strong rock–oil adhesion interactions, contact radius does not
change at all (because of pinning of the contact line) thus leaving a fraction of
original oil drop back on the rock surface. It happens due to the lack of sufficient
force required to move the contact line in such drop-volume reduction experiments.
In such cases, the exhibited dynamic contact angle does not satisfy the definition of
Figure 3. Schematic representation of oil drop separation from rock surface in intermediate-wet to
oil-wet S–L–L (rock–oil–water ) systems at reservoir conditions.
the water advancing contact angle. Hence, they are being referred to as “dynamic
contact angle” throughout this paper.
Our recent experimental work [5] shows that the effect of rock-adhesion inter-
actions can be characterized more accurately using line tension-based modified
Young’s equation. Experimental determination of line tension by evaluating the
drop size dependence of dynamic contact angle using modified Young’s equation
for various S–L–L and S–L–V systems has been reported in the literature [5–8].
A schematic representation of line tension concept is shown in Fig. 4. For rock–
oil–brine systems, line tension, a thermodynamic property of the system, can be
defined as the work necessary to reduce isothermally a unit length of contact line
[5]. The modified Young’s equation describes the drop size dependence of the dy-
namic contact angle (θd ) and accounts for its effect on the equilibrium of the system
in terms of line tension.
For rock–oil–brine systems, line tension-based modified Young’s equation can
be written as:
σ
γOW cos θd = (γSO − γSW ) − . (4)
r
Combining equation (2) and equation (4), we have:
σ
cos θd = cos θ∞ − (1/r), (5)
γOW
Line Tension 101
Figure 4. Concept of line tension for S–L–L (rock–oil–water) systems.
Figure 5. Drop size dependence of dynamic contact angle (θd ) for different rock–live oil–brine sys-
tems at reservoir conditions (10.34 MPa, 114.4◦ C).
where the equilibrium contact angle θ∞ is defined by equation (2) and θd is the
dynamic contact angle and r is the contact radius. Equation (5) is the generalized
form of modified Young’s equation.
Plots of cosine of the dynamic contact angle (cos θd ) as a function of recipro-
cal of contact radius (1/r) for different rock–oil brine systems investigated in a
recent experimental work [5] are shown in Fig. 5. As evident from Fig. 5, the lin-
ear relationship described by equation (5) can be used to determine the value of
line tension in S–L–L systems of petroleum engineering interest. On the basis of
these findings of our recent experimental work [5], it seems reasonable to use line
tension-based modified Young’s equation while determining the work of adhesion
for rock–oil–brine systems.
Hence a new equation to compute the work of adhesion can be derived after
including the line tension effect on the work of adhesion for rock–oil–brine systems.
Replacing the term (γSO − γSW ) in equation (1) with the terms defined in equations
(2) and (4), we obtain:
σ
WSOW = γOW (1 − cos θd ) − . (6)
r
Equation (6) is the new equation for determining the work of adhesion after incor-
porating line tension effect on the work of adhesion for rock–oil–brine systems of
petroleum engineering interest.
4. Experimental Procdure
The description of the apparatus and its components as well as the experimental pro-
cedure to determine line tension in rock–oil–brine systems at reservoir conditions
using representative reservoir fluids are reported elsewhere [5]. The experimental
measurement of line tension consists of three steps: 1. Measurement of oil–water
interfacial tension by pendant drop technique, 2. Drop size dependence of dy-
namic contact angle (θd ) by sessile drop technique, and 3. Dual-Drop–Dual-Crystal
(DDDC) technique for characterization of system’s wettability by measuring the
equilibrium water advancing contact angle (θA ). A high-pressure high-temperature
optical cell was utilized to conduct reservoir condition experiments for measuring
all the necessary parameters mentioned in steps 1, 2 and 3.
In recently reported experimental work [5], stock-tank oil from a selected oilfield
in Louisiana and its reservoir brine composition was used to prepare live oil and
synthetic reservoir brine for representing actual reservoir fluids. All reservoir condi-
tion experiments were conducted at 10.34 MPa and 114.4◦ C (reservoir temperature)
for depiction of rock–oil adhesion phenomenon at actual reservoir conditions.
For measuring the drop size dependence of dynamic contact angle (θd ), a large
sessile drop of live oil is formed on the rock crystal surface in brine-filled opti-
cal cell and is maintained for 18–24 h at reservoir conditions to attain equilibrium
between all three phases (rock, oil and brine). The contact angle measured at equi-
librium conditions is the equilibrium contact angle (θ∞ ) for the given system. Then,
the size of oil drop is reduced in several steps by withdrawing small volumes of oil
back into the needle at regular intervals of 15 minutes to determine the movement in
contact radius and corresponding dynamic contact angle. The images of sessile drop
are captured at each volume reduction step and later are analyzed using Drop Shape
Analysis (DSA) technique for precise measurement of dynamic contact angle, drop
volume and contact radius.
Line Tension 103
The process of volume reduction is repeated until either oil drop is detached
from rock surface or leaves some portion of drop behind on the rock surface. As
drop size is reduced gradually by withdrawing a small volume of oil through the
needle, drop volume also changes. The volume of newly formed smaller oil drop is
measured using DSA technique and this instantaneous volume of smaller oil drop
is then divided by the volume of initial large sessile drop to obtain the drop volume
ratio.
Based on observations and conclusions of our recent work [5] and collected
experimental data, the work of adhesion was computed for selected rock–live oil–
brine systems at reservoir conditions using equation (6).

The present study utilizes reported results from our recent work [5] and collected
experimental data for computing the work of adhesion as defined by equation (6).
These experimental data include oil–water interfacial tension (γOW ), equilibrium
contact angle (θ∞ ), change in contact radius (r) with drop volume and the corre-
sponding dynamic contact angle (θd ). The published results [5] of measured line
tension values at reservoir conditions for the systems included in present study are
shown in Table 1.
Reported average value of oil–water IFT (23.58 mN/m) at reservoir conditions
for selected live oil–synthetic brine system [5] was utilized in calculations. The col-
lected data on contact radius (r) and the corresponding dynamic contact angle (θd )
along with the drop volume ratio at each volume reduction step and computed work
of adhesion using equation (6) for specific rock–live oil–brine systems investigated
are presented in Tables 2–4.
A series of captured images of sessile drops at different drop size reduction steps
for different rock–live oil–brine systems are shown in Fig. 6.
Plots of θd vs drop volume ratio for different rock–live oil–brine systems are
shown in Fig. 7.
The reported results from reservoir condition DDDC tests [5] conducted for
determining the wettability of investigated systems are presented in Table 5. For
chosen live oil–synthetic brine system, at reservoir conditions, the quartz surface
exhibited a strongly water-wet behavior with a small θA (24◦ ), dolomite surface
Table 1.
Published line tension values [5] at reservoir conditions
(10.34 MPa, 114.4◦ C) for the systems included in the present study
System Line tension, σ (N)
Quartz–live oil–synthetic brine 2.21E−09

Dolomite–live oil–synthetic brine 2.36E−07
Calcite–live oil–synthetic brine 5.55E−07
Table 2.
Contact radius (r) and dynamic contact angle (θd ) data and computed work of adhesion using equation
(6) for quartz–live oil–brine system at reservoir conditions (10.34 MPa, 114.4◦ C; θ∞ = 24◦ , initial
aging time of large sessile drop = 24 h)
Vol. red. Dynamic cos θd Cont. rad., Drop vol. Work of adhesion
step No. contact angle, r (m) ratio∗ WSOW (J/m2 ),
θd (◦ ) equation (6)
1 24 0.9135 3.03E−03 0.85 2.04E−03

2 24 0.9135 2.31E−03 0.73 2.04E−03
3 24 0.9135 1.84E−03 0.48 2.04E−03
4 24 0.9135 1.28E−03 0.36 2.04E−03
5 24 0.9135 8.02E−04 0.24 2.04E−03
6 40 0.7660 5.51E−04 0.15 5.51E−03
* Drop volume ratio = (Volume of current sessile drop/Volume of initial sessile drop).
Table 3.
(6) for dolomite–live oil–brine system at reservoir conditions (10.34 MPa, 114.4◦ C; θ∞ = 28◦ , initial
1 32 0.8480 3.97E−03 0.89 3.52E−03

2 34 0.8290 3.97E−03 0.81 3.97E−03
3 45 0.7071 3.97E−03 0.39 6.85E−03
4 51 0.6293 3.78E−03 0.18 8.68E−03
5 79 0.1908 3.20E−03 0.11 1.90E−02
Table 4.
(6) for calcite–live oil–brine system at reservoir conditions (10.34 MPa, 114.4◦ C; θ∞ = 32◦ initial
1 49 0.6561 3.08E−03 0.92 7.93E−03

2 65 0.4226 3.00E−03 0.83 1.34E−02
3 92 −0.0349 2.94E−03 0.55 2.42E−02
4 110 −0.3420 2.90E−03 0.35 3.15E−02
5 162 −0.9511 2.57E−03 0.08 4.58E−02
Line Tension 105
Figure 6. Images of sessile drops captured during drop size reduction tests for different rock–live
oil–brine systems at reservoir conditions (10.34 MPa, 114.4◦ C).
Figure 7. θd vs drop volume ratio plots for different rock–live oil–brine systems at reservoir conditions
(10.34 MPa, 114.4◦ C).
Table 5.
Reported results [5] from DDDC tests at reservoir conditions (10.34 MPa, 114.4◦ C) for different
rock–live oil–brine systems
System Equilibrium Equilibrium water Wettability of the

contact angle advancing system
θ∞ (◦ ) contact angle
θA (◦ )
Quartz–live oil–synthetic brine 24 28 Water-wet

Dolomite–live oil–synthetic brine 28 82 Intermediate-wet
Calcite–live oil–synthetic brine 32 154 Oil-wet
Table 6.
Computed work of adhesion using conventional approach [equation (3)] for different rock–live oil–
brine systems at reservoir conditions (10.34 MPa, 114.4◦ C)
System Equilibrium cos θ∞ Work of adhesion

contact angle WSOW (J/m2 ),
θ∞ (◦ ) equation (3)
Quartz–live oil–synthetic brine 24 0.9135 2.04E−03

Dolomite–live oil–synthetic brine 28 0.8829 2.76E−03
Calcite–live oil–synthetic brine 32 0.8480 3.58E−03
showed an intermediate-wet nature with θA value of 82◦ and calcite surface was
strongly oil-wet exhibiting a large θA (154◦ ).
The work of adhesion was also calculated using conventional approach i.e. equa-
tion (3). In those calculations, equilibrium contact angle (θ∞ ) values, obtained from
DDDC technique, were utilized. For different rock–live oil–brine systems, com-
puted work of adhesion values using equation (3) are given in Table 6.
The computed values of the work of adhesion after incorporating line tension
effect were compared (Tables 2–4) with the values obtained using conventional
approach (Table 6). For this, values of the work of adhesion at the smallest drop
volume ratio for each case (last column in Tables 2, 3 and 4) were used and were
compared with the values given in Table 6. The comparison results are given are
shown in Fig. 8. The computed work of adhesion at each drop volume reduction
step was plotted against the drop volume ratio as shown in Fig. 9.
Based on the definition of line tension (Section 3) and its direction in rock–oil–
water systems (Fig. 4), and if appropriate values of line tension, oil–water IFT and
contact radius are considered, it can be shown numerically that the computed work
of adhesion using equation (6) would be significantly higher than computed using
equation (3), although the line tension term is subtracted from contact angle term
in equation (6). The drop size dependence of dynamic contact angle as described
by line tension-based modified Young’s equation (equation (5)) renders the work of
Line Tension 107
Figure 8. Comparison of work of adhesion values computed with and without line tension effect for
different rock–live oil–brine systems at reservoir conditions.
Figure 9. Work of adhesion vs drop volume ratio plots for different rock–live oil–brine systems at
reservoir conditions (10.34 MPa, 114.4◦ C).
adhesion larger (equation (6)) compared to the work of adhesion computed without
including the line tension effect (equation (3)).
Figure 8 clearly demonstrates that conventional approach of determining the
work of adhesion underestimates the impact of rock–oil adhesion on residual oil
recovery. The computed work of adhesion after considering the effect of rock–oil
adhesion interactions in terms of line tension indicates the presence of stronger

rock–oil adhesion interactions in the system than predicted conventionally. In the
case of an oil-wet system, the work of adhesion is found to be 13 times higher after
inclusion of line tension effect.
The study of drop size dependence of dynamic contact angle is of practical
importance to the process of oil displacement in the reservoir. In dynamic situ-
ation, a certain amount of work is required to overcome the rock–oil adhesion
interactions depending on the extent of rock–oil adhesion interactions present in
the system. Conventional approach (equation (3)) fails to take into account the ex-
tent of rock–oil adhesion interactions. However, the line tension-included modified
Young’s equation for the work of adhesion (equation (6)) as proposed in the present
study takes into account the extent of rock–oil adhesion interactions, hence pro-
vides a realistic estimate of the necessary work required to overcome the rock–oil
adhesion interactions for mobilizing the residual oil.
It is evident from Fig. 9 that the work of adhesion as computed using equation
(6) increases with decrease in drop volume ratio as the dynamic contact angle ap-
proaches towards the equilibrium water advancing contact angle for the particular
rock–oil brine system. Specially, for strongly oil-wet systems, if the drop volume
is further reduced, then in the extreme situation, the dynamic contact angle may
exceed the equilibrium water advancing contact angle (measured using DDDC tech-
nique) and can attain a value very close to 180◦ , indicating the formation of oil film
on the rock surface. However during this process, the contact radius may not change
any more because of pinning of the contact line on the rock surface. In such a situ-
ation, at low drop volumes, with constant value of oil–water interfacial tension and
a limiting value of line tension (steep slope of cos θd vs 1/r line), for a specific
rock–oil–brine system, the work of adhesion would exceed that computed using the
equilibrium water advancing contact angle and limiting value of line tension.
The varying slope of the work of adhesion versus drop volume ratio plots for
quartz, dolomite and calcite surfaces indicates that more work is required to remove
oil from an oil-wet system due to the formation of oil film on the rock surface
compared to a water-wet system where no such oil film is formed. Higher values
of the work of adhesion at lower drop volume ratios as shown in Fig. 9 indicate
that more and more work needs to be exerted in order to displace the residual oil as
its saturation decreases. This situation may be of practical importance in evaluating
the use of certain chemicals for reservoir wettability alteration.
6. Conclusions
The effect of line tension on the work of adhesion is evaluated for different rock–
live oil–brine systems at reservoir conditions of elevated pressure and temperature
using representative live oil and reservoir brine. The present study provides an im-
proved equation for more realistic estimate of the work of adhesion in rock–live
oil–brine systems at reservoir conditions. The results of this study clearly demon-
Line Tension 109
strate that the conventional approach does not take into account the extent of rock–
oil adhesion interactions hence appears to be unable to predict the work of adhesion
correctly and thus underestimates the effect of rock–oil adhesion interactions on oil
recovery. In the case of intermediate-wet and oil-wet systems, the magnitude of the
work of adhesion is much larger than predicted by the conventional approach. This
study indicates that more and more work needs to be exerted in order to displace
the residual oil as its saturation decreases.
Acknowledgements
This work has been made possible by the Louisiana Board of Regents research grant
(LEQSF 2006-09-RD-B-03) and Louisiana Economic Development Assistantship
award of Louisiana State University. The donation from Louis C. and Denise R.
Soileau, IV for the renovation of L. C. Soileau, III Rock–Fluids Interactions Lab-
oratory, LSU is gratefully acknowledged. The authors thank MST Conferences
organizers for the invitation to present this work at the 6th International Sympo-
sium on Contact Angle, Wettability and Adhesion.
References
1. D. N. Rao, in: Contact Angle, Wettability and Adhesion, Vol. 3, K. L. Mittal (Ed.), pp. 191–210.
VSP, Utrecht (2003).
2. D. N. Rao, M. Girard and S. G. Sayegh, Interfacial phenomenon in miscible gas processes, SPE
paper 19698, Presented at the 64th Annual Technical Conference & Exhibition held in San Antonio,
TX (October 1989).
3. M. Valat, H. Bertin and M. Robin, in: Advances in Core Evaluation III: Reservoir Management,
P. F. Worthington and C. Chardaire-Rivière (Eds), pp. 387–409. Taylor & Francis, London (1992).
4. D. N. Rao and B. B. Maini, Impact of adhesion on reservoir mechanics, Paper-CIM-93-65, pre-
sented at the 44th Annual Technical Meeting of the Petroleum Society of CIM held in Calgary,
Alberta, Canada (May 1993).
5. D. Saini, Y. Zheng and D. N. Rao, Line tension-based modification of Young’s equation for
rock–oil–brine systems, SPE Paper 113321, presented at the 2008 SPE Improved Oil Recovery
Symposium held in Tulsa, OK, USA (April 2008).
6. J. Drelich, Colloids Surfaces A 116, 43–54 (1996).
7. D. Li and A. W. Neumann, Colloids Surfaces 43, 195–206 (1990).
8. A. Amirfazli and A. W. Neumann, Adv. Colloid Interface Sci. 110, 121–141 (2004).
Is the World Basic? Lessons from Surface Science
K. L. Mittal a and Frank M. Etzler b,∗

a
P.O. Box 1280, Hopewell Junction, NY 12533-1280, USA
b
Boehringer-Ingelheim Pharmaceuticals, 900 Ridgebury Road, Ridgefield, CT 06877, USA
Dedicated to Prof. Emil Chibowski on the occasion of his 65th birthday and published in ANNALES
(Universitatis Mariae Curie-Skłodowska, Lublin-Polonia),
LXIII, Sectio AA (Chemia), 1–16 (2008).
Abstract
In this work the authors review various definitions of acids and bases. Furthermore, the application of these
theories to surface and adhesion science is discussed. It is clear that application of various acid–base theories
may not always lead to consistent conclusions on the acid–base character of surfaces. Possible reasons for
the inconsistency include the nature of the scale, the choice of reference points, and the use of either poor or
inconsistent statistical procedures in addition to experimental difficulties. The present authors recommend a
future careful study and comparison of the various acid–base theories when applied to surface and adhesion
science. The limited studies already preformed suggest this activity would be worthwhile.
Keywords
Definitions of acids and bases, acid–base scales, acid–base theories, surface free energy, contact angle
1. Introduction
From antiquity chemists and alchemists realized the fundamental importance of
acid–base reactions and interactions in a variety of chemical and physico-chemical
processes. The development of molecular based and quantitative definitions of acid-
ity and basicity has not been fully completed despite a couple of centuries of work
by notable scientists.
Lavoisier [1] is responsible for the first molecular based definition of acids. He
incorrectly thought acids must contain oxygen. Indeed, the name oxygen refers to
the ability to make acids. Lavoisier was not aware of hydrohalic acids or indeed
the presence of hydrogen in sulfuric or nitric acids as they were prepared from the
gases not containing hydrogen. Justus von Liebig [1] around 1838 suggested that
acids must contain hydrogen. Von Liebig’s theory [2] differes from modern thought
* To whom correspondence should be addressed. Tel.: 203-798-5445; e-mail: frank.etzler@boehringer-

ingelheim.com

112 K. L. Mittal and F. M. Etzler
in a number of respects. Arrhenius [1] suggested that the ability to produce hy-
drogen or hydroxide ions etablished a compound as an acid or base, respectively.
Sorensen [3–5], in 1909, established the pH scale and its measurement both electro-
chemically and by the use of indicators. The pH scale is the first popular quantitative
scale for acidity. A competing and similar scale was the “Rational” scale proposed
by Giribaldo (see, for instance, Bates [6]). Bates [6] has discussed more fully the
development and use of pH scales. The usual pH scale applies exclusively to dilute
aqueous solutions. Similar scales can be and have been used in other protic solvents.
Relating pH scales for various solvents is at best difficult.
In 1923 Brønsted and Lowry [1] developed a theory of acids and bases based
on the ability of molecules to donate or accept protons (Hydrogen ions). Also in
1923 Gilbert Lewis [1] devised a scale which is based on the ability of compounds
to accept or donate electrons. The Brønsted and Lowry definition can be subsumed
into the Lewis definition. The Lewis definition serves as a basis for many of acid–
base theories used in surface science such as those advanced by Drago [7] and
Gutmann [8].
Pearson and Parr (see Pearson [9]) have more recently (when compared to the
work of Lewis) explored the concept of Hard–Soft Acids and Bases (HSAB).
Morra [10] compared the results of two methods of investigating the acid–base
properties of solids. His results suggest that the assessment of acid–base character
of surfaces may be different depending on the acid–base scale chosen.
In this work, the authors review various acid–base scales used in the literature to
assess the acid–base character of surfaces. Experimental approaches used to assess
the acid–base character of solids are also reviewed. This work provides an assess-
ment of the current state-of the-art.
2. Acid–Base Scales Used in Surface Science

2.1. Overview
Several approaches have been used to study the acid–base character of solids. Gen-
erally they employ either pH type scales, or Lewis acid–base scales to assess the
acid–base character. pH scales are linear and continuous with acid character lying
at one end of the scale and base character at the other end. Materials are said to
be basic or acidic based on the scale value relative to the neutral value (typically
pH = 7). Bates [6] has reviewed in detail the development of pH scales and their
determination by electrochemical and indicator methods.
Various Lewis acid–base scales have also been used for assessment of acid–base
character. In general, two scales have been used — one to assess acidity and the
other to assess basicity. Materials may thus have both acid and base properties.
These two scales may be based on different criteria and may not be of the same
range of values.
Drago [7] has proposed a scale where a molecule must be considered as either
an acid or a base but cannot have amphoteric character.
Lessons from Surface Science 113
2.2. pH Scales
The pH scale for dilute aqueous solutions was suggested by Sorensen [3–5] in 1909.
For aqueous systems the scale runs from 0–14 with 7 being neutral. pH scales for
other protic solvents and mixed solvents have also been suggested. The range of pH
values in these other solvents differs as to the points of neutrality and the length of
the scale. There are thus many pH scales. Comparison of pH values determined on
one scale to those determined on another scale is at best difficult as they do not share
a common standard state, range of values or neutral point. Bates [6] has discussed
the comparison of various pH scales extensively.
Fowkes [11] has reviewed various wet chemical methods for determining the
acidity of surfaces. Methods employing pH type scales are discussed further below.
2.2.1. pH Indicator Dyes
pH indicator dyes adsorbed on surfaces have been used to assess surface acidity
[11]. An acidity scale can be defined as follows:

[B]
H0 = pKa + log (1)
[BH + ]
[B]
where pKa is the value for the indicator dye in aqueous media. The ratio [BH +]
may be determined spectrophotometrically. Titration of the surface with different
bases is known to give different responses so that a unique acidity scale cannot be
found. Fowkes [11] more fully reviews this topic.
2.2.2. Isoelectric pH
Bolger [12] has proposed that the use of isoelectric pH in water could be used to
assess surface acidity. For monopolar acids and bases, such as silica and aluminum
oxide, this approach seems appropriate. For amphoteric materials like iron oxide the
use of a single number fails to tell the complete story. A single value assessment
of acid or base character appears to be an inadequate descriptor of the acid–base
character of surfaces. (Also see Chehimi et al. [13].)
2.3. Hard and Soft Acids
Pearson [9] has proposed scales for acidity based on the numerical values of equilib-
rium constants (K) of nucleophilic reactions. In this model acids (A) and bases (B)
are considered to be either hard or soft. Hard acids have positive charge, are difficult
to reduce and have small size. Soft acids are easy to reduce, may not be charged and
are of large size. Hard bases (nucleophiles) are often negatively charged, are diffi-
cult to oxidize and have small size. Soft bases are easy to oxidize, do not need to
have negative charge and are large size.
According to Pearson
log K = SA SB + σA σB , (2)
where S is a hardness factor and σ is a softness factor. Lee [14] has, in particular,
contributed to the use of HSAB theory.
HSAB theory introduces the use of separate scales for acidity and basicity. Com-
pounds are classified as acids or bases but cannot have amphoteric properties.
2.4. Drago
In a manner not dissimilar to HSAB theory Drago [7] proposed that the heat of
acid–base adduct formation can be expressed as:
−
HAB = (EA EB + CA CB ), (3)
where E and C are susceptibilities for electronic and covalent interactions. Fowkes
[11, 15–20], in particular, has exploited Drago’s model to determine acid–base char-
acter of polymer and metal surfaces using calorimetric heats of adsorption. Oldak
and Pearson [21] have recently commented on the determination of Drago constants
using infrared spectroscopy.
For common materials E and C values are in the range of ∼1–10. Eb values,
however, only range from ∼0.5 to 1.7. The lengths of the acid and base scales thus
may not be of equal ranges.
2.5. Gutmann
Gutmann [8] also proposed a scale for estimation of
HAB . His equation is
(AN · DN)
−
HAB = , (4)
100
where AN is referred to as the acceptor number and DN the donor number. DN is
defined by the enthalpy of interaction between SbCl5 and a base. Thus
DN = −
H (SbCl5 : base). (5)
The acceptor number, AN, is defined as the [31] P NMR shift when triethylphos-
phine is dissolved in the acid of interest. AN is defined to be 0 when hexane is used
and 100 when SbCl5 in 1,2 dichloroethane is used. Riddle and Fowkes [22] realized
that the [31] P NMR shift had a significant contribution from van der Waals inter-
actions. These authors suggested an alternate scale, AN ∗ , with units of kcal/mole.
Here
AN ∗ = 0.228(AN − AN d ), (6)
AN and AN d , the dispersion or van der Waals component, have units of ppm.
For common tested materials the values of AN ∗ and DN range from approxi-
mately 0 to 38, indicating that the scales are of nearly equal lengths.
3. Surface Science Models Incorporating Acid–Base Concepts

3.1. Fowkes
Earlier Fowkes [23] suggested that surface free energy could be expressed as a
summation of components. Over the past few decades several authors have pro-
posed acid–base scales to determine surface free energy. These authors include
Fowkes; van Oss, Chaudhury and Good (see van Oss [24]) as well as Chang [25].
Each of these approaches considers the surface free energy to be the summation of
Lifshitz–van der Waals interactions and Lewis acid–base interactions. They differ
with respect to how Lewis acid–base interactions are considered. (Also see Etzler
[26].)
Fowkes considers:
Wa = WaLW + WaAB , (7)
where Wa is the work of adhesion and the superscripts LW refers to Lifshitz–van der
Waals interactions and AB refers to Lewis acid–base interactions. WaAB is expressed
as follows:
WaAB = −f · N ·
HaAB , (8)
where N is the number of adsorption sites per unit area and

∂ ln WaAB −1
f = 1− (9)
∂ ln T
and
f ≈ 0.2, . . . , 1.0(10). (10)
When using the Fowkes approach some authors have taken f as unity although this
does not seem to be a good approximation [27]. Because f and N are generally not
known, direct calculations of the work of adhesion are often not made. Determi-
nation of
H AB for multiple probe liquids on a given solid together with models
by Drago [7] or Gutmann [8, 28] can be used to assess the acid–base nature of the
surface (also see [22]).
The Fowkes approach is particularly suitable for use with data (
H AB ) collected
using either IGC or flow microcalorimetry.
3.2. van Oss, Chaudhury and Good
According to the van Oss, Chaudhury and Good model [24, 29] the Lewis acid–base
component is modeled as follows
γiAB = 2(γi+ γi− )1/2 , (11)
where γ + is the Lewis acid parameter and γ − the Lewis base parameter. van Oss,
Chaudhury and Good further choose
γi+ = γi− ≡ 0 (12)
for alkanes, methylene iodide and α-bromonaphthalene which presumably interact
only through Lifshitz–van der Waals interactions. For water
γH+2 O = γH−2 O ≡ 25.5 mJ/m2 . (13)
Based on these above numerical choices, γ + and γ − have been experimentally
determined for a variety of liquids. van Oss [24] has compiled and reviewed the
determination of these values (also see Etzler [26]).
The range of γ − values for known liquids is 0–77, while the range of γ + values
is 0–25 with most being less than 10. The idea of the neutrality of pure water in this
scale is like the pH scale. It also appears that the acid and base scales may not be of
equal range.
3.2.1. Della Volpe
Della Volpe [30] has criticized the choices of van Oss and co-workers. Della Volpe
suggests that γ LW = 26.25, γ + = 48.5 and γ − = 11.2 mJ/m2 for water. Della Volpe
also suggests that the γ LW estimate made earlier by Fowkes [31] is incorrect. The
neutrality of water must be questioned in general because ordinary water contains
dissolved carbon dioxide. A different choice of acid and base parameters is math-
ematically permitted and Della Volpe argues that his choice yields acid and base
parameters for solid surfaces that are more consistent with chemical intuition.
Whether or not Della Volpe’s scale offers advantages over that of van Oss and
co-workers has not been resolved. van Oss [32] claims that the choice of numerical
values for acid and base parameters is arbitrary and unimportant. Della Volpe [30]
claims that his choice of parameters is more consistent with chemical intuition and
is more appropriate for statistical reasons as well.
3.3. Chang–Chen
The Chang–Chen model [25, 33] for interfacial free energy is largely based on
the same principles which govern the van Oss, Chaudhury and Good model. Both
models treat Lifshitz–van der Waals interactions in the same way. Calculation of the
surface free energy components requires the knowledge of the same experimental
data. The two models, however, differ in the way that Lewis acid–base interactions
are modeled.
In the Chang–Chen model
WaLW = WaL = 2(γ1LW γ2LW )1/2 = P1L P2L , (14)
where
PiL = (2γiL )1/2 (15)
PiL is the dispersion parameter. The superscript L is equivalent to LW.
Like the van Oss, Chaudhury and Good model the acid–base interaction is mod-
eled using two parameters. These parameters, Pia and Pib , are referred to as principal
values. The acid–base work of adhesion can be represented using the following re-
lation:
WaAB = −
GAB
a = −(P1 P2 + P1 P2 ),
a b b a
(16)
where
GAB a b
a is the acid–base free energy of adhesion. Tabulated Pi and Pi values
are substituted into equation (16) such that the work of adhesion is maximized and
the free energy of adhesion is minimized.
The acid–base character of a material is determined by the sign of Pia and Pib . If
Pia = Pib = 0 then the material is neutral (or non-polar). If Pia and Pib are both pos-
itive then the material is monopolar acidic and if both negative then the material is
monopolar basic. If Pia and Pib are of opposite sign then the material is amphoteric.
4. Experimental Methods Used to Evaluate Acid–Base Character of Surfaces

4.1. Contact Angles
Contact angle measurements have often been used for the determination of solid
surface free energies. Acid–base character can be determined using the models pro-
posed by van Oss, Chaudhury and Good as well as those by Chang or Della Volpe.
Contact angles can be obtained using sessile drops, Wilhelmy plate method,
wicking or much less popular heats of immersion [26]. The use of ASDA (Ax-
isymmetric Drop Shape Analysis) is particularly useful for determining the quality
of contact angle. Chibowski [34], Dalal [35] and Etzler [26] have separately com-
mented on the importance of obtaining quality contact angle data. Carre [36] has
recently commented on the effect of probe liquid choice on the calculated value
of the van Oss, Chaudhury and Good parameters. Chibowski [37, 38] has recently
commented on the application of different models for determination of surface free
energies from contact angle data. It is likely that much of the literature data are
inappropriate for determination of surface free energies.
The following criteria must be met in order for contact angle data to be used for
calculation of surface free energies.
1. Young’s equation is valid.
2. Probe liquids are pure compounds. Mixtures are likely to exhibit selective sur-
face adsorption. See Adamson [39], for instance.
3. The surface free energy of materials is independent of the test conditions at
fixed T and P (i.e. the surface free energy of a solid is not modified by the probe
liquid).
4. γl > γs or more generally cos(θ ) > 0.
5. γsv is not influenced by probe liquids.
Chibowski [40] has recently discussed some of the experimental issues in deter-
mining good contact angle data.
Contact angle titrations can also be performed. Contact angles of aqueous solu-
tions will vary with pH depending on the ionization state of the surface.
4.2. Inverse Gas Chromatography (IGC)
The Fowkes approach is most often used with IGC. Either Drago or Gutmann para-
meters are thus generated. Figure 1 illustrates how IGC may be implemented. The
free energies of adsorption near infinite vapor dilution of a series of alkanes are de-
termined. These free energies of adsorption show a linear relation with the number
of carbons in the probe alkane. Figure 1 follows the model of Dorris and Gray [41]
which is often employed. From the slope of the alkane data γ LW can be determined.
For the adsorption of materials with acid–base properties the
GAB is determined
Figure 1. RT ln Vn versus a(γLLW )1/2 . Squares: alkane liquids. Triangles: other liquids. Slope of linear
fit to alkane liquids gives γsLW and vertical displacement of triangular points from linear fit gives Isp .
Isp (
Gab ) is the free energy resulting from acid–base interactions.
from the difference between the measured value and that of an alkane at the same
value of a(γLLW )1/2 . Here a is molecular area. This difference is illustrated by the
vertical line in Fig. 1. If
GAB is known for at least three temperatures then
H AB
can be calculated using the usual thermodynamic relations. If
H AB is known for
several acid–base probes then the Gutmann or Drago parameters can be determined.
Similarly from
GAB van Oss acid–base parameters could be determined.
The acid–base component values are often larger than they might be using con-
tact angle data. This is presumably so as data are taken for a very low surface
coverage, thus adsorption largely occurs on high energy sites. Acid–base charac-
terization may differ from that determined via contact angle data for experimental
reasons.
4.3. XPS
XPS (ESCA) [42] employs the photoelectric effect to determine the energy of emit-
ted electrons. The energy of the emitted electrons is specific to the element from
which the electrons are emitted. Furthermore, the oxidation state of the element can
be discerned. XPS has been used extensively to understand the chemical composi-
tion of surfaces.
Shifts of the emitted electron energies (i.e. binding energies) have also formed
the basis of an acid–base scale for metal oxides [13].
The scales are:

O = BE(O1s ) − 530 eV (17)
and

M = BE(M2p ) − BE(M02p ), (18)
where BE is the binding energy and O1s refers to the oxygen 1s orbital and M2p
the metal 2p orbital. The superscript zero refers to the metallic state. No superscript
refers to the oxidized state. The sum
O +
M has been found to correlate with
isoelectric point measurements. XPS has been used with other chemical markers as
well.
4.4. SPM
The scanning probe microscopy [13] can be used to measure the adhesion force
between the probe and a surface. The SPM probe may be chemically modified with
specific chemical groups giving it specific acid–base or other chemical properties.
The measured adhesion forces have been found to depend on pH, surface charge
and acid–base properties in the expected way.
4.5. Colloid Titration
The point of zero charge (pH units) or the isoelectric point (zeta potential equal
to zero) can be determined and appear to be a measure of surface acidity [11, 43].
Such values have been found to correlate with Gutmann donor numbers [43].
4.6. Indicator Dye Adsorption
As discussed above, indicator dyes may be adsorbed on surfaces. The ionization
state of the dye can be used as an indicator of surface acidity. The Hammett acidity
function H0 defined above has been found to correlate with isoelectric points [11,
43].
4.7. Calorimetry
Heats of adsorption can be conveniently used with the models of Drago and Gut-
mann and have been exploited by Fowkes [19]. Heats of adsorption, however, do
not always correlate with isoelectric points [43].
5. Current Issues
5.1. What is the Best Scale for Determination of Acid–Base Character?
From the above discussion, it is clear that the concept of acids and bases arose
early in the history of chemistry and has been extensively exploited by chemists.
The definition of acids and bases has also evolved over the past 250 years or so.
Several quantitative scales for acidity and basicity have been suggested and many
are discussed above. For about 100 years, the Sorensen pH scale based on Brønsted–
Lowry theory has found particular usefulness for understanding the behavior of
aqueous systems.
Acid–base scales in modern use are of a few types:

1. pH [6] based scales are of the first type. The scales are continuous with low
numbers reflecting acids and high numbers bases. Isoelectric point and indicator
dye adsorption measurements are of this type. The pH scale is most frequently used
with dilute aqueous systems. It possible to construct other pH scales in other protic
solvent systems but the measurements on one scale cannot be readily translated to
results on another scale. pH scales are, therefore, not unique; many scales exist.
2. Drago’s [7] model is based on HSAB theory. As in the case of pH measure-
ments materials are characterized as being either acids or bases. Two parameters, E
and C, are required to specify the acid or base properties of a material.
3. Various Lewis acid–base scales have been constructed. These include the
scales proposed by Gutmann [8] and by van Oss, Chaudhury and Good (see van
Oss [24]). The modifications to van Oss et al.’s scale proposed by Della Volpe are
also included. For models based on Lewis acid–base theory, two independent scales
— one for acidity and one for basicity — exist. A given compound can have am-
photeric properties in that it can be assigned a value on each of the two scales. The
idea of neutrality is more complex. Hydrocarbons are usually assigned a value of
zero on each of the two scales and thus can be thought of as neutral. A material with
equal non-zero values on the two scales is more difficult to understand in terms of
neutrality. The values assigned for acidity and basicity on each scale are based on
different criteria and thus the ratio of the acid value to base value can vary signifi-
cantly. Both Morra [10] and Della Volpe [30] have discussed this point. At this time
the “holiness” of one scale over another is unclear [44]. It is, however, clear that if
the ranges of values of the two scales differ from one another an apparent imbalance
of acidity versus basicity could be perceived. The lengths of the two scales in the
van Oss, Chaudhury and Good model do not appear to be equal (see for instance
[13]) leading some to conclude that the materials are most often more basic than
acidic, sometimes in contradiction to usual chemical intuition.
4. The Chang–Chen [25] model is also based, perhaps somewhat more loosely,
on Lewis acid–base theory. Like other such models, two parameters describe the
acid–base character of a material. In this instance, however, the two parameters are
each allowed to take on both positive and negative values. Bases are considered
to have negative parameter values. Uniquely, the Chang–Chen model allows for
base–base and acid–acid repulsive contributions to WaAB .
At the present moment it is unclear which scale most accurately reflects nature.
5.2. Experimental Issues
The use of contact angles to determine the acid–base properties requires that the
measured contact angles meet each of the required assumptions listed above. Earlier
work by Kwok [45], Etzler [44] and Dalal [35] suggests that these assumptions
may not always be met, leading to incorrect acid–base parameters. Della Volpe,
furthermore, suggests that the set of probe liquids must be carefully chosen so that
suitable condition numbers for curve fitting are obtained. IGC leads to high values
of acid–base parameters presumably because data are taken at low surface coverage
where only high energy sites are occupied.
A comprehensive discussion of various aspects of curve fitting and other statisti-
cal issues related to the determination of surface free energy components seems to
be missing. Della Volpe [30] and Bialopiotrowicz [46, 47], however, have discussed
some issues in this regard.
Particular attention should be paid to the linearization of equations (transform-
ing coordinates so that linear plots can be constructed) done to calculate slopes
and intercepts. While determining slopes and intercepts by least squares is easy
to perform and is good for illustration, linearization has unintended consequences
on the resulting calculations because standard errors are often not constant for all
y-axis (vertical axis on graph) values. Modern mathematical software makes such
linearization unnecessary. It is statistically more robust not to perform such lin-
earizations.
With particular regard to the van Oss, Chaudhury and Good model several cal-
culation methods have been used.
As stated earlier by Gardner [48] the equation below,
Wa = γl [1 + cos(θ )] = 2(γlLW γsLW )1/2 + 2(γl+ γs− )1/2 + 2(γl− γs+ )1/2 (19)
may be solved in several ways. Some methods allow for γ + and γ −
to have negative
values. Other methods require both these values to be positive as suggested by van
Oss [24]. In a given paper, the author’s choice of calculation method is frequently
unclear to the reader. A non-linearized approach would only allow the acid and

H AB ∗
base parameters to be positive. Similarly for Gutmann’s model DNads vs. AN DN will
be linear with a slope equal to Kd and intercept of Ka . Here Kd is the solid base
parameter and Ka the solid acid parameter. Statistical errors may sometimes cause
Ka to be negative. A non-linearized approach to fitting will always result in positive
values for both Kd and Ka .
Standardized and statistically sound practices must be adopted by all future in-
vestigators.
5.3. Is the World Basic?
For purposes of investigating the acid–base character of solid surfaces, the van Oss,
Chaudhury and Good approach has become popular. As pointed out by, for instance,
Morra [10], the use of this model suggests that many materials are predominantly
basic (γ − > γ + ). Is the van Oss, Chaudhury and Good model the best choice for
the assessment of acid–base character of surfaces?
In this paper various scales for characterization of the acid–base character of ma-
terials have been discussed. It is clear that the van Oss, Chaudhury and Good model
provides some good insights into the acid–base character of surfaces. It is equally
clear that the acid–base scale chosen by these authors is not obviously more holy
than others. Indeed, water is considered to be neutral, a choice necessary only for
Sorensen’s pH scale in aqueous media. Indeed the generally smaller values of γ +
for probe liquids may be significantly responsible for the apparent basic character
of all materials.
van Oss [32] argues that the choice of numerical values for γ − and γ + for wa-
ter is unimportant. In striking contrast, Della Volpe [30] and Lee [49–51] suggest
that the choice is in fact important for both scientific and statistical reasons. These
authors also propose alternate scales. Morra [10], in particular, suggests that acid–
base scales must be consistent with chemical intuition and, furthermore, suggests
that van Oss’ scale appears to be inconsistent with chemical intuition.
With perhaps a few exceptions, a careful and consistent application of sound
statistical principles to the fitting of various parameters does not appear to be the
general practice. A consensus on sound scientific and statistical approaches must
be reached.
It is clear that despite more than 250 years of investigation into the nature of acids
and bases a quantitative description of this property of matter has not been fully or
universally achieved. Both experimental and theoretical issues affect the application
to surface and adhesion science. The authors encourage others to improve the state
of the art through the design of critical experiments.
References
1. L. M. Miessler and D. A. Tar, Inorganic Chemistry. Pearson Prentice Hall, New York (1991).
2. H. L. Finston and A. C. Rychtman, A New View of Current Acid–Base Theories. John Wiley, New
York (1982).
3. S. P. L. Sorensen, Compt. Rend. Trav. Lab. Carlsberg 8, 1 (1909).
4. S. P. L. Sorensen, Biochem. Z. 21, 201 (1909).
5. S. P. L. Sorensen, Biochem. Z. 21, 131 (1909).
6. R. G. Bates, Determination of pH: Theory and Practice. John Wiley, New York (1964).
7. R. S. Drago, G. C. Vogel and T. E. Needham, J. Amer. Chem. Soc. 93, 6014 (1971).
8. V. Gutmann, The Donor–Acceptor Approach to Molecular Interaction. Plenum, New York (1978).
9. R. G. Pearson, J. Amer. Chem. Soc. 85, 3533 (1963).
10. M. Morra, J. Colloid Interface Sci. 182, 312 (1996).
11. F. M. Fowkes, in: Industrial Applications of Surface Analysis, Advances in Chemistry Series, Vol.
199, L. A. Casper and C. J. Powell (Eds), p. 69. American Chemical Society, Washington, DC
(1982).
12. J. C. Bolger and A. S. Michaels, in: Interface Conversion, P. Weiss and G. D. Cheevers (Eds),
Chapter 1. Elsevier, New York (1969).
13. M. M. Chehimi, A. Azioune and E. Cabet-Deliry, in: Handbook of Adhesive Technology, 2nd edn,
Chapter 5, A. Pizzi and K. L. Mittal (Eds). Marcel Dekker, New York (2003).
14. L. H. Lee, in: Fundamentals of Adhesion, L. H. Lee (Ed.), p. 1. Plenum, New York (1991).
15. F. M. Fowkes, J. Colloid Interface Sci. 28, 493 (1968).
16. F. M. Fowkes, J. Adhesion 4, 155 (1972).
17. F. M. Fowkes, J. Adhesion Sci. Technol. 1, 7 (1987).
18. F. M. Fowkes, D. W. Dwight, D. A. Cole and T. C. Huang, J. Non-Cryst. Solids 120, 47 (1990).
19. F. M. Fowkes, Y. C. Haung, B. Shah, M. J. Kulp and T. B. Lloyd, Colloids Surfaces 29, 243 (1988).
20. F. M. Fowkes, K. L. Jones, G. Li and T. B. Lloyd, Energy and Fuels 3, 97 (1989).
21. R. K. Oldak and R. A. Pearson, J. Adhesion Sci. Technol. 21, 775 (2007).
22. F. L. Riddle and F. M. Fowkes, J. Amer. Chem. Soc. 112, 3259 (1990).
23. F. M. Fowkes, J. Phys. Chem. 66, 382 (1962).
24. C. J. van Oss, Interfacial Forces in Aqueous Media. Marcel Dekker, New York (1994).
25. F. Chen and W. V. Chang, Langmuir 7, 2401 (1991).
26. F. M. Etzler, in: Contact Angle, Wettability and Adhesion, Vol. 3, K. L. Mittal (Ed.), p. 219. VSP,
Utrecht, The Netherlands (2003).
27. M. D. Vrbanac and J. C. Berg, in: Acid–Base Interactions: Relevance to Adhesion Science and
Technology, K. L. Mittal and H. R. Anderson, Jr. (Eds), p. 67. VSP, Utrecht, The Netherlands
(1991).
28. V. Gutmann, A. Steininger and E. Wychera, Montash. Chem. 97, 460 (1966).
29. R. J. Good, in: Contact Angle, Wettability and Adhesion, K. L. Mittal (Ed.), p. 3. VSP, Utrecht,
The Netherlands (1993).
30. C. Della-Volpe and S. Siboni, in: Acid–Base Interactions: Relevance to Adhesion Science and
Technology, Vol. 2, K. L. Mittal (Ed.), p. 55. VSP, Utrecht, The Netherlands (2000).
31. F. M. Fowkes, Ind. Eng. Chem. 56(12), 48 (1964).
32. C. J. van Oss, in: Acid–Base Interactions: Relevance to Adhesion Science and Technology, Vol. 2,
K. L. Mittal (Ed.), p. 173. VSP, Utrecht, The Netherlands (2000).
33. W. V. Chang and X. Qin, in: Acid–Base Interactions: Relevance to Adhesion Science and Tech-
nology, Vol. 2, K. L. Milttal (Ed.), p. 3. VSP, Utrecht, The Netherlands (2000).
34. E. Chibowski, Adv. Colloid Interface Sci. 98, 245 (2002).
35. E. N. Dalal, Langmuir 3, 1009 (1987).
36. A. Carre, J. Adhesion Sci. Technol. 21, 961 (2007).
37. E. Chibowski and K. Terpilowski, J. Colloid Interface Sci. 319, 505 (2008).
38. H. Radelczuk, L. Holysz and E. Chibowski, J. Adhesion Sci. Technol. 16, 1547 (2002).
39. A. W. Adamson, Physical Chemistry of Surfaces, 5th edn. John Wiley, New York (1990).
41. G. M. Dorris and D. G. Gray, J. Colloid Interface Sci. 77, 353 (1980).
42. T. L. Barr, Modern ESCA: The Practice and Principles of X-Ray Photoelectron Spectroscopy.
CRC Press, Boca Raton, FL (1994).
43. C. Sun and J. C. Berg, Adv. Colloid Interface Sci. 105, 151 (2003).
44. F. M. Etzler, in: Contact Angle, Wettability and Adhesion, Vol. 4, K. L. Mittal (Ed.), p. 216.
VSP/Brill, Leiden (2006).
45. D. Y. Kwok, T. Gietzelt, K. Grundke, H.-J. Jacobasch and A. W. Neumann, Langmuir 13, 2880
(1997).
46. T. Bialopiotrowicz, J. Adhesion Sci. Technol. 21, 1557 (2007).
47. T. Bialopiotrowicz, J. Adhesion Sci. Technol. 21, 1539 (2007).
48. D. J. Gardner, S. Q. Shi and W. T. Tze, in: Acid–Base Interactions: Relevance to Adhesion Science
and Technology, Vol. 2, K. L. Mittal (Ed.), p. 363. VSP, Utrecht, The Netherlands (2000).
49. L.-H. Lee, Polym. Mater. Sci. Eng. 81, 391 (1999).
50. L.-H. Lee, Polym. Mater. Sci. Eng. 80, 304 (1999).
51. L.-H. Lee, J. Adhesion 76, 163 (2001).
Part 2
Wettability Control/Modification
Reversible Transition from Hydrophobicity to
Hydrophilicity of Photon Responsive Surfaces:
From Photochromic Molecules to Nanocrystals
Athanassia Athanassiou ∗
NNL—National Nanotechnology Laboratory of CNR-INFM, Via per Arnesano, 73100 Lecce, Italy
and IIT—Italian Institute of Technology, Via Morego 30, 16152 Genova, Italy
Abstract
The work presented here deals with an exclusive use of light to alter reversibly the wettability of surfaces.
This fascinating property is due to the involvement of photoresponsive materials. Two categories of mate-
rials are investigated: organic photochromic molecules and inorganic colloidal nanocrystals. In the case of
the photochromic molecules, the wetting characteristics of surfaces of polymers doped with such molecules
can be tuned when irradiated with laser beams of properly chosen photon energy. Their hydrophilicity is
enhanced upon UV-laser irradiation, and the process is reversed upon green laser irradiation. The mecha-
nism involved in these transformations is the photoinduced isomerization between the isomeric forms of
the photochromic molecules that have different polarities. Structuring of the photochromic polymeric sur-
faces enhances significantly both the initial hydrophobicity of the system and the light-induced wettability
variations of the surfaces. In the case of the inorganic nanocrystals we examine different substrates coated
with organic-capped TiO2 nanorods. Such nanorod-based coatings exhibit a surface transition from a highly
hydrophobic state to a highly hydrophilic one under selective UV-laser irradiation. This behavior is reversed
under long dark storage. The mechanism involved is a progressive increase in the degree of surface hydrox-
ylation of TiO2 upon UV irradiation. The surfactant molecules that cover the nanorods appear to undergo
simultaneous conformational changes without suffering any significant photocatalytic degradation.
Keywords
Reversible wettability, polymers, photochromic molecules, TiO2 nanorods, laser light
1. Introduction
The fabrication of “smart” functional and responsive surfaces with controlled and
reversible wetting characteristics has stimulated intense research efforts, due to
their importance in both fundamental studies and practical applications, such as in
biosensors, controllable drug delivery, microfluidic devices, intelligent membranes,
multifunctional coatings, and self-cleaning surfaces [1–17].
* E-mail: athanassia.athanassiou@unile.it

128 A. Athanassiou
The wettability control of solid surfaces has been attempted by various means
of surface tailoring, and special attention has been focused recently on the de-
velopment of surfaces that can alter reversibly their wetting properties when il-
luminated by photons of specific energy. A class of strategies for the design of
light-responsive solid substrates with dynamically modifiable surface properties
involves photoisomerizable molecules that can modify their conformation and/or
polarity when illuminated with light of appropriate wavelength. The effect is called
photochromism and is defined as a reversible transformation of chemical species,
induced by electromagnetic radiation, between two states (isomer forms) having
light absorption bands in distinctively different regions [18]. Some characteristic
examples follow: Thymine-terminated self-assembled monolayers attached on gold
surfaces undergo reversible photo-dimerization when exposed to UV radiation of
appropriate wavelength, which, in turn, leads to reversible changes in the water
contact angle [19, 20]. Photosensitive azobenzene chromophores undergo a photo-
induced cis-trans isomerization that causes a different orientation of the dipole
moment of the azobenzene, thus, affecting the substrate surface energy. The azoben-
zene units can be employed as an end-group of a monolayer [21, 22] or they can be
attached as side groups to polymer chains [23, 24]. The photochromic molecules
of the spiropyran family have been utilized in the form of a monolayer [25, 26] or
as additives within a polymer matrix [4, 27, 28] in order to affect the surface wet-
tability. Since the wettability modulation achieved with these systems is inherently
small, techniques such as increase of nano- and micro-scale surface roughness of
such surfaces have been used for the amplification of the wettability changes.
Another approach towards switchable wettability changes exploits the reversible
property of some semiconductor oxides, such as TiO2 [5–12, 29], ZnO [13], WO3
[30] and V2 O5 [31], to exhibit increased hydrophilicity upon band-gap photoexci-
tation. Although the related mechanistic aspects are still a matter of intense debate,
this conversion is generally believed to be initiated by photogenerated holes that
create oxygen vacancies at the semiconductor surface. These defects are then able
to promote dissociative adsorption of atmospheric water, which ultimately leads to
an increase in surface hydroxylation. Under dark ambient conditions, a slow recov-
ery of the starting properties occurs, as molecular oxygen replaces the UV-grafted
hydroxyl moieties [5–11, 13, 30–33]. Among the aforementioned materials, TiO2
has been the most studied one, mainly due to its photocatalytic activity upon UV
illumination. The photocatalytic properties of TiO2 surfaces, combined with their
unique capability to convert under UV light from a hydrophobic to a highly am-
phiphilic state (i.e., simultaneously hydrophilic and oleophilic), have paved the way
to the development of inorganic TiO2 -based films with simultaneous antireflective,
antibacterial, self-cleaning and antifogging behavior [34–39].
For the case of the wettability of substrates based on photochromic molecules
the present paper offers a review of some of our findings on the reversible photo-
induced wettability changes of polymer surfaces doped with photochromic spiropy-
ran (SP) molecules. SP molecules absorb in the UV spectral region, and upon
From Photochromic Molecules to Nanocrystals 129
Scheme 1. The two stereoisomers spiropyran (left) and merocyanine (right).
irradiation in this region, they convert to their colored merocyanine (MC) isomers
by the photochemical cleavage of the C–O bond in the SP ring and the consequent
ring opening [40] as shown in Scheme 1. It is the enhanced dipolar moment of the
MC stereoisomers as compared to the SP ones [41] that potentially leads to the en-
hancement of hydrophilicity. The isomerization process is reversible, with the MC
being converted back to the SP upon irradiation with visible light [42]. This property
is retained when the photochromic molecules are incorporated within macromolec-
ular matrices [43–45].
In particular in this work we demonstrate how the hydrophilicity of the SP-doped
polymeric surfaces is enhanced upon UV-laser irradiation when the embedded non-
polar SP molecules convert to their polar MC isomers and how this process is
reversed upon irradiation with green laser pulses. Moreover, we show that, when
the roughness of the doped polymer surfaces is changed by micropatterning us-
ing soft molding lithography or interferometric photopolymerization lithography,
the light-induced wettability variations of the structured surfaces can be enhanced
compared to those on flat surfaces. On the top, surface tuning between different hy-
drophilic and hydrophobic conditions can be achieved by changing the topological
parameters of the introduced patterns. This enhancement should be attributed to the
influence of the controlled surface roughness [46–48] together with the reversible
macroscopic volume changes induced [49] to the imprinted gratings by the photoi-
somerism of the SP, which additionally contributes to the wettability changes, due
the light-induced photochromic interconversions.
In the case of the UV-switchable wettability of TiO2 substrates we present a
review on a bottom-up strategy we follow to create photoswitchable TiO2 coat-
ings that exhibit large and reversible wettability changes upon selective UV-laser
irradiation and dark storage cycles. Our approach relies on the use of surfactant-
capped anatase TiO2 nanorods (NRs) as the building blocks for the fabrication of
hybrid inorganic/organic thin films. These films are used as coatings for various
substrates, such as ITO, silicon, and different polymer surfaces, thus determining
the ultimate wettability of these materials. The prepared oxide thin-film coatings
consist of closely packed arrays of laterally oriented TiO2 domains that expose
well-defined light-sensitive crystal facets. Such NR-based films exhibit a reversible
surface transition from a highly hydrophobic (water angle ∼ 110◦ ) to a highly hy-
drophilic condition (water contact angle ∼ 20◦ ) under excitation with pulsed laser
light at 355 nm. By combining structural and spectroscopic techniques, a mech-
130 A. Athanassiou
anism is identified, according to which the UV-induced hydrophilicity correlates

with a progressive increase in the degree of surface hydroxylation of the TiO2
NR components. Interestingly, the wettability changes are not accompanied by
any noticeable photocatalytic degradation of the NR capping surfactants, which
can be explained by the combined effect of the periodic nature of the irradia-
tion regime and of the nanocrystal morphology. The organic ligands on the oxides
are instead assumed to undergo conformational rearrangements in response to the
light-driven surface reconstruction. Upon prolonged storage in the dark, ambient
oxygen removes the newly implanted hydroxyl groups from the TiO2 surfaces and
the ligand conformations are again modified, such that the films can recover their
native hydrophobic properties. We also present the deposition of uniform, crys-
talline and optically transparent thin films from an aqueous suspension of TiO2
NRs of different sizes on a variety of substrates, and we demonstrate that the pre-
pared coatings exhibit different hydrophobicity depending on the NRs size, always
demonstrating the fully reversible hydrophobic–hydrophilic behavior upon laser
pulsed ultraviolet irradiation and dark storage cycles. Finally we describe the real-
ization of hybrid organic/inorganic surfaces that have been engineered by exploiting
photo-lithographically tailored SU-8 (SU-8 2010, MicroChem, Newton, MA) poly-
mer patterns as templates for accommodating close-packed thin films of colloidal
anatase TiO2 NRs, which respond to UV light as described above. The prepared
SU-8/TiO2 substrates offer a dual micro-/nano-scale roughness, arising from the su-
perimposition of the surfactant-capped inorganic nanocrystals onto the micrometer-
sized polymer patterns. Such suitably combined architectural and chemical surface
design enables achievement of UV-driven transitions from a highly hydrophobic
to a highly hydrophilic condition, with excursions in water contact angle (WCA)
of more than 100◦ . A dark storage period of a few weeks allows the native hy-
drophobicity of the surfaces to be restored. To our knowledge this is the first report
combining the light-induced and reversible wettability changes of inorganic oxides
with micrometer-scale patterned polymeric systems, results that may be especially
promising in microfluidics [50–54].
2. Experimental Part
2.1. Spiropyran-Based Surfaces
2.1.1. Photochromic Dopant
The photochromic dopant used was the 1 ,3 -dihydro-1 ,3 ,3 -trimethyl-6-nitrospiro
[2H-1-benzopyran-2,2 -(2H)-indole], also known as 6-NO2 BIPS, which belongs
to the family of spiropyrans (SP). It has a 3-D structure (Scheme 1). Initially it is
transparent in the visible range of spectrum, and absorbs in the ultraviolet (UV). By
irradiation with UV light, it is converted to its isomeric form, merocyanine (MC),
through the photochemical cleavage of the C−O bond in the spiropyran ring, as
shown in Scheme 1. The MC has a new absorption band in the visible range of
the spectrum and can exist, in principle, in eight stereoisomers, with respect to the
Scheme 2. The photochromic dopant spiropyran and the stable forms of its isomer merocyanine.
three partial double bonds (the cis isomers: TCT, TCC, CCT, CCC, and the trans
isomers: TTT, TTC, CTC, CTT (Scheme 2)) [40]. The cis isomers are metastable,
whereas the trans isomers, shown in Scheme 2, are the stable ones and have planar
structures [40–42]. This isomerization process is reversible. Thus, upon irradiation
with visible light of the colored MC forms, the molecules are converted back to the
SP form.
2.1.2. Host Polymers
The first polymer used was the poly(ethyl methacrylate-co-methyl acrylate) [P(EM-
co-MA)] (average MW ∼ 100 000) (Aldrich) having a glass transition tempera-
ture (T g ) of 48◦ C. It was doped with 5.0% by wt. SP dopant. For the prepara-
tion of the flat films, solutions of 95.0 wt.% of the polymer [P(EMA)-co-P(MA)]
and 5.0 wt.% of the photochromic molecule 6-NO2 BIPS were prepared in toluene.
Certain volume of this solution was spin-coated onto glass substrates, to produce
homogeneous films. The second polymer used in this work was an acrylate-based
photo-polymer. It consists of three basic components: A sensitizer dye, the Eosin Y
(2-, 4-, 5-, 7-tetrabromofluorescein disodium salt), an amine photo-initiator, the
N-methyldiethanolamine (MDEA), and a multifunctional acrylatemonomer, the
Pentaerythritol triacrylate (PETIA), which forms the backbone of the polymer
network. This system is particularly sensitive in the spectral region from 450
to 550 nm. For the preparation of the flat films, a solution of the photocurable sys-
tem with 0.3% by wt. of the SP molecules was spin coated onto glass substrates.
Before spin-coating, the glass substrates were cleaned in a sonic bath with acetone
132 A. Athanassiou
for 10 min, and with 2-propanol for 10 min more. Then, the glass substrates were
re-fluxed for 30 min in a mixed solution of (3-mercaptopropyl) trimethoxysilane :
H2 O : 2-propanol (3 ml : 5 ml : 250 ml), rinsed with 2-propanol, and dried at 100◦ C
for 30 min in oven. The polymeric layers were finally formed with photopolymer-
ization after their exposure for 6 min to a highly coherent cw laser beam (532 nm)
of 2.0 W power.
2.1.3. Patterning Methods
For the preparation of the pattern on the SP/[P(EM-co-MA)] substrates, nanopat-
terning by soft lithography was used. The surface of the samples was patterned
using soft molding [55], which had been previously demonstrated to be a powerful
approach to pattern functional organic materials [56]. More specifically, the mas-
ter structures (gratings with a period of about 1.3 µm) were fabricated onto quartz
by either photo- or electron-beam lithography, and subsequent reactive ion etch-
ing. Next their elastomeric replicas were realized by poly(dimethylsiloxane) (Dow
Corning Sylgard 184) according to a standard replica molding procedure and placed
onto the flat spin-cast photochromic polymer films under their own weight. Then
the system was heated at 55◦ C, a temperature higher than the glass transition tem-
perature of the P(EMA)-co-P(MA) films (Tg = 48◦ C). After the thermal cycle, the
replica was easily peeled off from the SP-polymer substrate, on which the pattern
was faithfully transferred. The soft molding was carried out in nitrogen atmosphere
to avoid deterioration of the photochromic molecules upon heating. It is based on
the capillarity effect that drives a polymer, or an appropriate chosen blend of or-
ganics, to penetrate into the recessed features of an elastomeric element previously
templated by a suitable master. A low polymer viscosity favors such a penetration.
The proper increase of temperature reduces the viscosity of the organic compound,
because of increased molecular mobility, thus allowing filling of the pattern in a
few minutes.
The surface of the photopolymerized SP/PETIA films was patterned using inter-
ferometric lithography. The same solution of the photocurable system as described
above was spin coated onto the previously photopolymerized film. Then, a periodic
pattern was formed onto the surface of the second layer, deposited on the first one,
after exposure for 5 s to two interfering cw laser beams (532 nm) of 1.0 W power
each. The sample was washed in a bath of deionised water for 4 hours, in order
to develop the grating. After the complete development of the pattern, the sample
was dried with a nitrogen flux. The gratings formed had a period of 2.5 µm. The in-
terferometric lithography technique described here is ideal for creating periodic
structures onto SP-containing substrates, since the SP molecules do not absorb in
the spectral region from 450 to 550 nm where the photocurable system is sensitive.
2.1.4. Experimental Description
The sessile drop geometry was selected for determining the contact angle of water
on the solid SP/P(EMA)-co-P(MA) and SP/PETIA substrates. The method is based
on the principle that the profile of the sessile drop of a liquid is governed by a force
balance between the surface/interfacial tension as well as the contact angle of the
liquid on the substrate. The advantage of this method for the determination of con-
tact angles is due to the fact that use is made of the whole drop profile and not of just
the contact points with the substrate surface; the actual value of the contact angle is
then extracted from the data and is not subject to the influence of possible impurities
at the drop edges at the surface. An automated home-made contact angle machine
was used to determine the contact angle based on a collection of digital images
of sessile drops. A drop (3 µl) of distilled, deionized Millipore water (18.2 M)
was formed from a capillary tip, and was detached gently from the tip upon the
substrate of interest. The atmosphere around the drop was rich in water vapor in
order to achieve minimum evaporation of the droplet. The reversible wetting prop-
erties of the samples are examined upon UV and visible pulsed laser irradiation.
After absorption measurements on the polymer samples doped with SP before and
after irradiation (Fig. 1), we chose the appropriate lasers needed for the SP to MC
conversion and for the MC to SP recovery. The lasers chosen for the irradiation of
the samples were a XeCl laser operating at 308 nm, 30 ns pulse duration (Lambda
Physik, EMG 201 MSC), and a Nd:YAG laser, operating at the second harmonic,
λ = 532 nm, 5 ns pulse duration (B. M. Industries, Series 5000). The laser beams
were focused onto the surface of the samples at an area of 3.5 × 3.5 mm2 . After
the measurement of the initial contact angle, the drop was removed and the flat or
patterned sample was irradiated with 50 UV pulses of specific fluence, depending
on the sample. The new contact angle of a water drop was measured after the UV
irradiation. The sample was then again dried and irradiated with 200 green laser
Figure 1. Absorption spectra of a [P(EM-CO-MA)] film doped with the SP photochromic molecules
before and after UV irradiation.
134 A. Athanassiou
pulses of specific fluence, depending on the sample. The new contact angle of a
water drop was measured after the green irradiation.
2.2. TiO2 NR-Based Surfaces
2.2.1. Materials
All chemicals were of the highest purity available and were used as received. Ti-
tanium tetraisopropoxide (Ti(OPri )4 or TTIP, 97%), titanium tetrachloride (TiCl4 ,
99.999%), trimethylamine N-oxide dihydrate ((CH3 )3 NO · 2H2 O or TMAO, 98%),
oleic acid (C17 H33 CO2 H or OLAC, 90%), 1-octadecene (C18 H36 or ODE, 90%),
oleyl amine (C17 H33 NH2 or OLAM, 70%), poly(methyl/methacrylate) (PMMA,
MW = 120 000), and polystyrene (PS, MW = 88 000) were purchased from Aldrich.
Silicon wafers were purchased from Jocam (Milano, Italy). SU-8 2010 was pur-
chased from MicroChem (Newton, MA, USA). All solvents used were of analytical
grade and were also purchased from Aldrich. Water was bidistilled (Millipore Q).
2.2.2. Synthesis of TiO2 Nanorods (NRs)
The nanocrystal synthesis was carried out under nitrogen flow using a standard
Schlenk line set-up. Anatase TiO2 NRs with an average diameter of ∼3–4 nm and
a mean length of ∼10–25 nm were obtained by low-temperature TMAO-catalyzed
hydrolysis of TTIP [55]. Typically, 10 mmol of TTIP were dissolved in 70 g of de-
gassed oleic acid (OLAC) and the resulting solution was then reacted with 3.5 ml
of an aqueous 2M TMAO solution at 100◦ C for 72 h. Larger TiO2 NRs, i.e., with
an average diameter of ∼3–5 nm and a mean length up to ∼30–40 nm, were syn-
thesized by aminolysis of titanium oleate with oleyl amine. In a typical preparation,
6 g of ODE, 26 mmol of OLAM, and 2 mmol of OLAC were carefully degassed at
120◦ C for 20 min, after which the mixture was put under N2 flow. Then, 2 mmol
of TiCl4 were added at 40◦ C and the flask was heated to 290◦ C for 30 min. By
varying the concentrations of the chemicals used it was possible to tune the size
of the prepared NRs. Next, the TiO2 NRs were separated from their growing mix-
ture by 2-propanol addition and were subsequently subjected to repeated cycles of
re-dissolution in toluene and precipitation with acetone to wash out any possible
OLAC surfactant residuals. Finally, optically clear TiO2 stock solutions in toluene
were prepared and stored under ambient conditions.
2.2.3. Preparation of SU-8 Masters
SU-8 micropatterns were prepared on silicon wafer substrates by photolithography.
Due to the formation of a ubiquitous oxide surface layer under ambient condi-
tions, the substrates are henceforth referred to as SiO2 /Si. First, the substrates were
cleaned by ultrasonication in acetone for 10 min and in 2-propanol for 10 min, after
which they were dried with a nitrogen airflow. SU-8 was processed according to a
procedure reported in the manual of MicroChem Corporation [60]. Each substrate
was spin-coated with the polymer in order to obtain the desired thickness of about
25 µm. The patterning was carried out by using a Karl Suss MJ B3 mask-aligner
with UV illumination and a photomask containing the patterns. Post-exposure bake
(PEB) was performed on a hot plate to achieve ultimate crosslinking of the resist,
which was then allowed to cool down in order to improve adhesion of SU-8 to the
substrate. The wafers were then developed by immerging them into a SU-8 devel-
oper and 2-propanol.
2.2.4. Preparation of TiO2 Coatings
The TiO2 NR coatings were fabricated under ambient laboratory conditions as fol-
lows. All the substrates, including the SU-8 patterns, were dipped into a 0.4 M
toluene TiO2 solution, using a dip coater system (NIMA, Nima Technology, Coven-
try, UK). The sample was dipped up to fifty times and withdrawn with a rate of
1 cm/min. Then the as-prepared films were gently dried with a nitrogen flow and
stored at ambient conditions.
2.2.5. Irradiation Experiments
The TiO2 coatings were irradiated with the third harmonic wavelength (355 nm) of
a pulsed Nd:YAG laser, with pulse duration of 3 ns, a repetition rate of 10 Hz, and
energy density of 5 mJ/cm2 . The total duration of the irradiation experiments, which
was required to detect the largest wettability changes, was found to be 120 min. This
period corresponds to 72 000 laser pulses, and hence, to an actual interaction time
of UV photons with the TiO2 samples of only 0.216 ms. Additional irradiation did
not influence the wettability of the prepared films any further.
2.3. Characterization Techniques
2.3.1. Contact Angle Measurements
Contact angle characterization was performed on the substrates by the sessile drop
method using a CAM200-KSV instrument (equipped with a digital camera for ac-
quiring magnified images of the microdroplets). Bidistilled water was used for these
tests and dispensed by means of a microsyringe. Typically, the droplet volume
was 3 µl. The WCA value was the average of 10 measurements recorded on dif-
ferent neighboring surface locations. The estimated error in a single measurement
was ±1◦ . Measurements were carried out on uncoated and then TiO2 -coated sub-
strates before irradiation and immediately after subjecting the samples to a known
number of laser pulses, as well as after storing the samples in the dark under ambi-
ent conditions (relative humidity 30–40%).
2.3.2. Transmission Electron Microscopy (TEM)
Low-magnification TEM images of TiO2 NRs were recorded with a JEOL JEM
1011 microscope operating at an accelerating voltage of 100 kV. The samples for
TEM analyses were prepared by dropping a dilute solution of the NRs dissolved
in toluene or chloroform onto carbon-coated copper grids and then allowing the
solvent to evaporate. Then the grids were immediately transferred to the TEM mi-
croscope.
2.3.3. Scanning Electron Microscopy (SEM)
Low-resolution SEM characterization of TiO2 coatings deposited onto silicon sub-
strates was performed with a RAITH 150 EBL instrument. Typically, the images
136 A. Athanassiou
were acquired at low accelerating voltages (less than 5 kV) using short exposure
times.
2.3.4. Atomic Force Microscopy (AFM)
AFM measurements were carried out using an Asylum Research MFP3D instru-
ment, operating in contact mode with standard silicon probes CSG10 (NT-MDT,
Zelenograd, Moscow, Russia). The nominal tip apex diameter was ∼10 nm, the tip
length was ∼4 µm, and the nominal cantilever spring constant was ∼0.1 N/m.

3.1. Spiropyran-Based Surfaces
The wettability of polymeric samples doped with SP molecules was investigated
using water drops on the prepared samples and measuring the water contact angle
(see Experimental Part). The apparent contact angle of a drop on a flat substrate,
θY , formed when the liquid is in contact with a solid surface in static equilibrium
with its vapor, is determined by Young’s equation
γLV cos θY = γSV − γSL , (1)
where γLV , γSV and γSL represent the interfacial tensions at the boundaries between
the liquid (L), vapor (V) and solid (S).
Figure 2 shows characteristic images of water drops, residing on the SP-doped
polymeric substrate before irradiation, after irradiation with 50 UV-laser pulses,
energy density 40 mJ/cm2 , and after subsequent irradiation with 200 green pulses,
energy density 45 mJ/cm2 . These are shown together with the average values of
the contact angles of drops residing on the initial, the UV-irradiated, and the green-
irradiated flat photochromic polymeric surfaces. The values are taken from drops
on a number of different samples of the same composition. The contact angle values
after the UV and green irradiations are limited to the first two UV-green irradiation
cycles to ensure limitation of photooxidation, which may cause degradation of the
photochromic molecules. The maximum difference between the average contact an-
gles measured in the experiment on the flat surfaces shown in Fig. 2 is 7 ± 1◦ . When
the SP-doped P(EM-CO-MA) sample is irradiated with UV pulses, the SP mole-
cules convert to their merocyanine (MC) isomers and the surface becomes colored.
Since the MC stereoisomers exhibit an enhanced dipolar moment compared to that
of the SP isomers [41], the surface of the sample becomes more hydrophilic and,
thus, the contact angles of the water drops decrease. Subsequent irradiation of the
sample with green laser pulses converts the molecules back to the initial SP form,
and thus, the surface becomes more hydrophobic, resulting again in an increase of
the contact angles. The number and the energy density of the UV pulses used in the
experiment presented in Fig. 2 ensure a complete conversion of the SP molecules
to their MC isomeric form, according to spectroscopic studies. The number and the
energy density of the green pulses ensure maximum recovery of the contact angles
(a)
(b)
Figure 2. (a) Characteristic images of water drops residing on the initial, the UV-irradiated, and
the green-irradiated flat surfaces of 5% SP–95% P(EMA)-co-P(MA) samples. (b) Average water
contact angle values on the initial, the UV-irradiated, and the green-irradiated flat surfaces of 5%
SP–95% P(EMA)-co-P(MA) samples (circles), and average water contact angle values on the ini-
tial, the UV-irradiated, and the green-irradiated flat surface of a P(EMA)-co-P(MA) sample without
photochromic molecules (squares).
Figure 3. 2-D (left) and 3-D (right) atomic force microscopy images of nanoimprinted grating formed
on the surface of the 5% SP–95% P(EMA)-co-P(MA) sample. The x and y axes are presented on
different scales.
of the water drops. Additional green laser pulses do not influence the wettability of
the surfaces any further. For comparison reasons, the average values of the contact
angles of water drops on a flat polymer surface without SP molecules are also pre-
sented in Fig. 2. It is obvious that laser irradiation does not have any impact on the
contact angles since the light-induced interconversions of the doped photochromic
molecules are exclusively responsible for the changes in the wetting properties.
Using the soft molding technique, surface patterns of gratings were achieved on
the SP-doped polymers (Fig. 3). After the patterning of the surface, the contact an-
138 A. Athanassiou
(a)
(b)
Figure 4. (a) Characteristic photographs of water drops situated on the initial, the UV-irradiated,
and the green-irradiated nanoimprinted surface of a 5% SP–95% P(EMA)-co-P(MA). (b) Average
contact angle values of water drops situated on the initial, the UV-irradiated, and the green-irradiated
nanoimprinted surface of different 5% SP–95% P(EMA)-co-P(MA) samples having an initial contact
angle of >105◦ (up-triangles) and an initial contact angle of <105◦ (down-triangles), and average
contact angle values of water drops situated on the initial, the UV-irradiated, and the green-irradiated
nanoimprinted surface of a P(EMA)-co-P(MA) sample without photochromic molecules (squares).
gle measurements were performed on the formed surface relief gratings of period
1.3 µm. The initial values of the water contact angles on the patterned surfaces are
always greater than on the flat surfaces and they reach θr ∼ 116◦ . Their variations
depend on structural variations of the imprinted gratings. In Fig. 4 are shown the
water contact angle values onto a patterned surface before irradiation, after UV ir-
radiation with 50 pulses, and after subsequent green irradiation with 200 pulses.
The differences between the contact angles measured in the experiments performed
on patterned surfaces vary between 11◦ and 24◦ . Comparing these results with the
ones for the flat surfaces, it is obvious that the surface patterning enhances signif-
icantly the light-induced wettability variations. The values of the contact angles of
the drops residing on the gratings of 1.3 µm period are always greater than those of
the drops residing on the flat surfaces, as clearly shown in Fig. 4. This effect can be
explained by a hypothesis which belongs to Cassie and Baxter, describing the effect
of roughness on the wetting of a surface, and it proposes that air can remain trapped
below the drops [57]. Since the drop is situated partially on air, and partially on the
stripes of the grating, the surface exhibits an enhanced hydrophobic behavior. Thus,
in the Cassie–Baxter model, the apparent contact angle that liquids form onto rough
or microtextured surfaces is larger than those onto flat surfaces of the same surface
chemistry. The apparent contact angle of a drop on a rough substrate θr is related to
Young’s contact angle θY , determined on a flat surface of the same nature, and is an
average between the value on air (180◦ ) and on the flat surface (θY ), given by:
cos θr = −1 + fs (1 + cos θY ) (2)
where, fs is the solid fraction of the surface in contact with the liquid (fs is dimen-
sionless and smaller than unity).
The significant enhancement of the light-induced wettability variations in the
case of the patterned surfaces is attributed to the light-induced structural changes on
the nanoimprinted grating. Atomic force microscopy (AFM) measurements show
clearly the reversible macroscopic changes of the gratings following the UV-green
irradiation. Figure 5 presents the AFM images of one grating at the second irradia-
tion cycle, before, after UV, and after green irradiation. Particularly, UV irradiation
induces a decrease in the width of the stripes by 13% and a subsequent increase
of the distance between the two stripes. A small decrease is also observed in the
period of the grating, whereas after irradiation with visible light the values recover
very close to the initial one [58]. In the AFM images of some of the prepared grat-
ings, as shown in Fig. 5, we observe a dip separating each stripe in two equal parts.
Our soft imprinting technique mainly relies on the capillarity that allows the viscous
polymer to spontaneously fill the vertical channels constituted by the recessed fea-
Figure 5. 2-D (left) and 3-D (right) AFM images that show the reversible dimensional changes of a 5%
SP–95% P(EMA)-co-P(MA) grating after UV and green irradiation. After UV irradiation a decrease
in the width of the stripes is observed, whereas after green irradiation the dimensions of the stripes are
almost recovered.
140 A. Athanassiou
tures of elastomeric elements, given that the wetting results in lowering the overall
free energy. As a consequence, due to the possible incomplete filling of such re-
cessed regions by the photochromic blend, double-crest features can be obtained in
the imprinted polymers, initially piling up in the regions adjacent to the protruding
areas of the mold and forming dips in the central part of the growing capillarity
features. This dip is useful for the morphological analysis of the patterned surfaces,
since it makes much clearer the demonstration of the volume changes upon alter-
nating UV and visible irradiation by AFM.
UV irradiation is responsible for the dimensional reduction of the stripes of the
gratings, whereas after green irradiation they recover their initial dimensions. As the
stripes of the pattern get narrower the water drop penetrates deeper in the channels
of the UV irradiated pattern, increasing the liquid–solid interfacial area. According
to (2) the contact angle of the water drop decreases when the fraction of the solid
which is in contact with the liquid (fs ) increases. When the stripes recover they
previous volume, the drop returns to its previous condition. Therefore, the structural
changes that occur to the pattern following alternating laser irradiation seem to
contribute further to the wettability changes due to the photochromic transformation
of SP to MC and back.
The reversible light-induced macroscopic deformations of the nanoimprinted
gratings are connected with the phototransformation of the photochromic dopants.
In particular upon UV irradiation, formation of aggregates between different MC
stereoisomers with zwitterionic character occurs, causing density fluctuations in
the polymer matrix. For these density fluctuations to be reduced, short scale motion
of the polymer chains may occur causing macroscopic contraction of the grating
stripes. Next, upon green irradiation, the formation of aggregates is less favored
until it stops due to the return of the MC molecules to the SP form which does not
form aggregates. When MC molecules return to SP, the polymer chains are forced
to return to their initial positions, and thus, the sample recovers to its initial volume.
This mechanism is analytically described in a work by Athanassiou et al. [49] and
is graphically described in Scheme 3.
Next we checked the wettability changes on the photopolymerized SP/PETIA
substrate. This substrate is more hydrophilic than the SP/P(EMA)-co-P(MA).
The water drops were stabilized a few minutes after their positioning on the flat or
the patterned surface. Therefore, the contact angle values presented in Fig. 6 were
taken 10 min after the drops were placed on the surface. The initial contact angle
on the flat surface was between 55◦ and 60◦ . In Fig. 6 are presented the values of
the contact angles of drops situated on a flat surface, before and after multiple UV
and green irradiations. After irradiation with 50 UV-laser pulses of energy density
40 mJ/cm2 , due to the SP to MC conversion the surface became more hydrophilic
resulting in the decrease of the contact angle. Subsequent irradiation with 200 green
pulses of energy density 45 mJ/cm2 did not cause the hydrophobicity of the system
to recover to its previous value, but it further increased its hydrophobic behavior.
Therefore, green irradiation steadily enhances the hydrophobicity of the surface af-
Scheme 3. Graphical representation of the mechanism responsible for the dimensional reduction of the
SP/P(EMA)-co-P(MA) samples upon UV-laser irradiation.
Figure 6. () Average contact angle values of water drops situated on the initial, the UV irradiated,
and the green irradiated patterned surface of a SP/PETIA sample, (•) average contact angle values
of water drops situated on the initial, the UV irradiated, and the green irradiated flat surface of a
SP/PETIA sample.
ter each irradiation cycle. This can be due to the fact that apart from the MC to
SP transformation, the green pulses might also initiate further polymerization of
the material, since it is photocurable upon green irradiation. The degradation of the
photochromic molecules upon light irradiation limits the lifetime of the system to
142 A. Athanassiou
(a)
(b)
Figure 7. (a) Interferometric setup for the formation of surface relief gratings of SP/PETIA of period
2.5 µm (b).
only a few cycles. The contact angle values decrease 2◦ to 4◦ upon UV irradiation
and increase 4◦ to 8◦ upon green irradiation. For the patterned samples, shown in
Fig. 7b, the interferometric setup (Fig. 7a) formed surface relief gratings of period
2.5 µm, almost double the period of the gratings prepared with soft molding. In
this case the initial values of the water contact angles on the patterned surfaces are
always smaller than on the flat surfaces. In Fig. 6 the water contact angle values
are presented onto a patterned surface before irradiation, after UV irradiation with
50 pulses (energy density 40 mJ/cm2 ) and after subsequent green irradiation with
200 pulses (energy density 45 mJ/cm2 ). The values were taken 10 minutes after the
drop was placed on the grating. The light-induced wettability variations were sig-
nificantly enhanced upon alternating laser irradiation, compared to the flat surfaces,
like in the case of the soft molding lithography.
Another hypothesis on the effect of roughness on the wetting of a surface, pro-
posed by Wenzel [59], explains why the values of the contact angles of the drops
located on the gratings of 2.5 µm period are always smaller than the ones on the flat
surfaces, as clearly shown in Fig. 6. The Wenzel model proposes that roughness in-
creases the liquid–solid interfacial area, and thus hydrophilicity or hydrophobicity
is enhanced geometrically. In our case the fact that the hydrophilic flat surface be-
comes more hydrophilic after patterning means that the water drops penetrate into
the grooves of the grating. This happens because the gratings have period higher
than that of the gratings formed with soft molding lithography, in combination with
the fact that the initial flat surface is more hydrophilic than the one of the P(EMA)-
co-P(MA) polymer. The apparent contact angle of a drop on the rough substrate θr
is related to Young’s contact angle θY , according to the following equation:
cos θr = r cos θY , (3)
where the surface roughness r is defined as the ratio of the actual wetted surface
area to the geometric surface area of the substrate. The latter is the surface as mea-
sured on the plane of the interface. (r is a number larger than unity which equals to
unity for flat surfaces.) In this case, the significant enhancement of the light-induced
wettability variations of the patterned surfaces is due to the fact that the water–solid
interfacial area is increased. Therefore, the water molecules are in contact with an
increased number of photochromic molecules, which undergo the transformations
from the non-polar SP to the polar MC and back, and they are responsible for the
wettability changes.
3.2. TiO2 NR-Based Surfaces
3.2.1. Structural Characterization of the TiO2 NRs
In this work we use organic-capped TiO2 nanorods (NRs) with tailored geometric
parameters. They are synthesized by employing oleic acid (OLAC) as the pro-
moter of anisotropic growth. After the purification procedures, the NRs are fully
dispersible in nonpolar media (such as chloroform, toluene, hexane) due to their
hydrophobic organic coating, and provide optically clear solutions. Figure 8 il-
lustrates the main morphological–structural details of the as-prepared colloidal
TiO2 nanocrystals. Low-magnification Transmission Electron Microscopy (TEM)
images (Fig. 8 panels a and b) show that the nanoparticles are uniform and unidirec-
tionally elongated, with average short and long axes dimensions tunable between
∼3–5 nm and ∼10–40 nm, respectively. High-resolution phase-contrast TEM
(HRTEM) investigations along with the relevant Fast Fourier Transform (FFT)
analyses (Fig. 8 panels c and d) demonstrate that the NRs are single-crystalline
TiO2 tetragonal anatase nanostructures, exhibiting an elongation in the 0, 0, 1 di-
rection, in agreement with the corresponding X-ray diffraction pattern (see Fig. 9).
Detailed HRTEM examinations of NRs viewed under different zone axes reveal that
they expose stepped longitudinal sidewalls dominantly made of crystallographically
equivalent (011)/(101) facets, while their apexes terminate exclusively with (001)
facets. More rarely (010) type surfaces can also be observed (Fig. 8 panel d). This
144 A. Athanassiou
Figure 8. (a) and (b) Low-magnification TEM images of TiO2 NRs synthesized by the hydrolytic
and the nonaqueous approaches, respectively. (c) and (d) HR-TEM images of portions of individual
nanorods viewed along their [1, 0, 0] zone axes. The dotted yellow lines marks the relevant facets that
enclose the nanorods under the observed orientations. In the corresponding insets, FFT analyses of
the respective images are reported.
peculiar morphological–structural arrangement reflects the consecutive alignment

of truncated octahedral bi-pyramidal units, which actually represent equilibrium
shape building blocks for TiO2 anatase, as derived by the Wulff construction and
experimentally observed under a variety of solution-phase reaction environments
[60]. Under our synthesis conditions, unidirectional nanocrystal growth is believed
to be promoted by the stronger adsorption of the OLAC surfactant molecules to
the TiO2 facets associated with the NR longitudinal sidewalls, which lowers their
surface energy value below that of other facets.
By changing the parameters in the preparation of the NRs their average length
can be further modulated from 50 to 200 nm as it can be seen in the TEM images
presented in Fig. 10a–c. In particular, the shorter NRs continue to be characterized
by a rod-like profile, while the longer ones exhibit curvature and an arrow-like
terminations.
3.2.2. Fabrication and Morphological Characterization of TiO2 NR-Films
We present the deposition of uniform, crystalline and optically transparent thin
films from an aqueous suspension of TiO2 NRs of different sizes on a variety of
Figure 9. Typical XRD pattern of the as-prepared colloidal TiO2 NRs along with the standard pattern
of anatase TiO2 .
Figure 10. Low-resolution TEM images of TiO2 NRs of different mean lengths: (a) 50 nm;
(b) 100 nm; (c) 200 nm.
146 A. Athanassiou
Figure 11. (a) High magnification SEM image of a TiO2 thin film (∼300 nm in thickness) prepared
by dip coating from an aqueous suspension of TiO2 NRs (about 4 nm in width and 40 nm in length)
on undoped (100) silicon substrate. (b) AFM topographic image corresponding to the same sample.
(c) Schematic representation of the NR arrangement on the substrates after the dip coating procedure.
substrates. The TiO2 NR-coatings have been fabricated under ambient conditions
by transferring the TiO2 NRs floating at the water/air interface onto the desired
substrates by a simple sequential dipping/withdrawal procedure. Many substrates,
such as ITO, silicon, poly (methyl methacrylate) (PMMA), and polystyrene (PS),
have been successfully used with comparable results. The dipping/withdrawal se-
quence is repeated up to five times to increase the thickness of the coating, that
varies from ∼50 to ∼300 nm. Finally the samples are dried under vacuum at room
temperature. Figure 11 demonstrates the surface features of a coating achieved by
applying the aforementioned fabrication method to a silicon substrate, starting from
the suspensions with the shortest available NRs. Scanning Electron Microscopy
(SEM) and Atomic Force Microscopy (AFM) images demonstrate a dense and uni-
form substrate coverage with almost negligible density of cracks over areas as large
as several squared micrometers. The high-resolution SEM inspection (Fig. 11a)
reveals that the films embody closely packed arrays of NRs, most of which are
preferentially accommodated in a roughly parallel orientation with respect to the
substrate. Accordingly, AFM analysis (Fig. 11b) shows nanometer-scale surface
roughness, consistent with the organization of NRs in compact multilayered struc-
tures arranged almost horizontally relative to the substrate. The self-organization
of the NRs in the thin-film coatings can be understood in terms of surface tension
balance, attained during the deposition process applied. When a small volume of a
concentrated organic solution of TiO2 NRs is spread onto the surface of water and
the solvent is allowed to evaporate, hydrophobic interactions among the surfactant-
coated NRs are progressively enhanced due to the increasing TiO2 concentration
in the organic phase. These attractive forces drive the NRs to segregate in close-
packed superstructures decreasing in this way the overall surface energy of the
system [61, 62]. The rod-like morphology of the TiO2 nanocrystals promotes their
preferential lateral packing, which maximizes the strength of inter-NR interactions,
while minimizing the TiO2 /water, and the TiO2 /air interfacial tensions [61, 62]. The
high cohesive strength among neighboring NRs guarantees that the obtained super-
structures maintain their arrangement intact upon transfer to the desired substrate
by the dipping/withdrawal step, as shown schematically in Fig. 11c. The minimum
spacing among neighboring NRs should be ultimately dictated by the sterical hin-
drance of the ligands on the NR facets. Nevertheless, even assuming that such NR
arrangement is repeated across the entire coating thickness, some packing irregular-
ities are expected to be present in the films (e.g., voids due to missing NR building
blocks, or NRs deviating from the parallel orientation of the arrays relative to the
underlying substrate). Moreover, the surfactant shell around the NRs is likely to
be incomplete. Taken together, these arguments suggest that the TiO2 NR coatings
can naturally possess an inner nanoporous structure [36], which does not, however,
affect their optical quality. All the NR films present similar surface and optical char-
acteristics independently of the substrates used, which remain unaltered upon UV
irradiation.
Figure 12 demonstrates the surface features of coatings achieved starting from
suspensions of NRs with different lengths, by applying the aforementioned fabrica-
tion method. The NRs of different lengths are deposited on silicon substrates and
the produced films are shown under the same magnification in the SEM images. It
can be observed that the shorter NRs show greater degree of lateral alignment on
the surface compared to the films prepared by the longer and less regularly shaped
NRs. The latter exhibit increased surface roughness, due to the various orientations
with respect to the substrate, as well as formation of aggregates with flower-like
morphology (Fig. 12c). The NRs in all films appear closely packed, and although
the degree of packaging of NRs is similar for all the samples, the films formed by
the smaller NRs appear to be more uniform and without cracks.
3.2.3. Light-Induced Reversible Wettability Changes

3.2.3.1. TiO2 NR-Coatings on Flat Surfaces The wettability properties of the NR-
films have been evaluated by water contact angle (WCA) measurements following
repeated cycles of UV irradiation and dark storage under ambient conditions. These
results, summarized in Fig. 13 for a TiO2 coating on a Si substrate, are similar
for all the substrates investigated. The as-prepared films exhibit a stable WCA of
110 ± 1◦ , indicating that the TiO2 NR-coatings are highly hydrophobic. The contact
angle values did not change for samples stored in the dark for months, which signi-
148 A. Athanassiou
Figure 12. SEM images of TiO2 NR-films deposited onto a silicon substrate starting from NRs sus-
pensions of different mean lengths: (a) 50 nm; (b) 100 nm; (c) 200 nm.
fies negligible surface contamination by atmospheric species under our laboratory

conditions.
Remarkable wettability changes are induced upon pulsed laser irradiation at
λ = 355 nm, with a 3 ns pulse duration, energy density 5 mJ/cm2 , and 10 Hz rep-
etition rate, for 120 min (corresponding to a net illumination time of 0.216 ms).
The UV wavelength employed guarantees selective band-gap excitation of the TiO2
component, leaving the OLAC capping molecules unaffected, since the latter ab-
sorbs below ∼320 nm. At the same time, the low repetition rate of 10 Hz ensures
negligible heat production within the film, which may in fact influence the kinetics
of chemical changes and/or of molecule adsorption on the TiO2 surface [63, 64]. In
Fig. 13a are shown pictures of water drops on the NR-coated surfaces before and
after UV irradiation. Upon UV irradiation a decrease in the WCA down to ∼20◦ is
seen. Therefore, under selective pulsed UV excitation, the TiO2 NR-coatings con-
vert from highly hydrophobic to highly hydrophilic.
Subsequent prolonged storage of the samples in the dark gradually re-establishes
the initial hydrophobic character of the as-prepared TiO2 NR-films. A moderate
hydrophobic condition (WCA ∼ 70◦ C) is already restored in the first 3 days, after
which the recovery rate reduces with time, so that the original WCA of ∼110◦ is
achieved again after few weeks. The process of reversible change of the WCA can
be carried out over several cycles of irradiation/dark storage (Fig. 13b), proving that
the NR coatings exhibit excellent reversibility in their wettability properties without
being affected by the photochemical fatigue which is, in fact, the main limitation of
photoresponsive molecules [61, 62].
(a)
(b)
Figure 13. (a) Pictures of water drops on a TiO2 NR-coating before and after irradiation with UV light.
(b) Reversible changes of WCA upon cycles of UV irradiation and dark storage on TiO2 NR-coatings.
Comparing the wetting behavior of NR-coatings deposited on silicon substrates

using NRs of different lengths, it is obvious that the starting hydrophobicity of the
coatings prepared with the longer NRs is enhanced by more than ten degrees com-
pared to the one of the short NRs. Figure 14 demonstrates the WCA measurements
performed on the different TiO2 NR-coatings deposited on silicon substrates, be-
fore irradiation, immediately after the UV irradiation, and after the dark storage.
The increased hydrophobicity of the longer NR-coatings is due to the aforemen-
tioned increased roughness of these surfaces demonstrated in the SEM images of
Fig. 12.
3.2.3.2. TiO2 NR-Coatings on Patterned Surfaces To increase the above pre-
sented differences in the WCA induced by UV irradiation of TiO2 NR-coatings, we
followed an approach where we combine the nanoroughness of the NRs with the
microroughness of the underlying surface. Such carefully designed surface archi-
tecture enables achievement of highly hydrophobic surfaces (mimicking the “lotus
effect”), that become very hydrophilic after UV irradiation due to the use of TiO2
NRs. For this purpose we used as substrate a polymer on which we realized pat-
terns of pillars by photopolymerization. The polymer SU-8 was chosen for the
realization of the patterned surfaces for its strong mechanical and chemical resis-
tant properties towards the common organic solvents. Moreover, it is commonly
150 A. Athanassiou
(a)
(b)
Figure 14. (a) Pictures of water drops on a TiO2 NR-coating made of NRs with 50 nm length before
and after irradiation with UV light. (b) Reversible changes of WCA upon a cycle of UV irradiation
and dark storage on TiO2 coatings made with NRs of different sizes. The x-axis is shown in log scale.
used in microfluidic structures [50–54], and it has been applied already success-
fully in the fabrication of super-hydrophobic surfaces using arrays of pillars. Low
magnification SEM images of two characteristic samples used in this work as sub-
strates before their coverage by the colloidal anatase TiO2 NRs, are shown in Fig. 15
(panels a and b). Pillars with the same dimensions (42 µm × 42 µm) but different
spacing were chosen in order to investigate wettability behaviors connected to spe-
cific surface geometries.
The wettability properties of SU-8 patterns were first investigated as a function of
their geometric parameters. We discuss here structures featured by the same pillar
size (42 µm × 42 µm in width, 25 µm in height), and different inter-pillar spacing,
the latter ranging from 14 µm to 77 µm.
The contact values of water droplets positioned on the micropatterned substrates
are increased compared to the corresponding SU-8 flat surface (WCA ≈ 80◦ ).
As their inter-pillar spacing is enlarged, the WCAs decrease monotonically from
118.0◦ ± 1◦ to 82.0◦ ± 1◦ , as demonstrated in Fig. 16. In the same figure, this
behavior is compared to the ones predicted by the models of Cassie–Baxter [57]
(a)
(b)
Figure 15. Low-magnification top-view SEM images of pillared SU-8 patterns fabricated onto
SiO2 /Si wafers, each comprising equally sized 42 µm × 42 µm pillars that are spaced from each
other by different distances: (a) 14 µm; (b) 28 µm.
and Wenzel [59], that describe the wetting behavior of rough surfaces by invoking
different mechanisms, and were analyzed above.
As already described the Cassie–Baxter’s model assumes partial wetting of the
patterned surface, due to the trapping of air underneath the droplet at the recessed
regions of the surface. For the calculation of the theoretical WCA values we con-
sider θY = 80◦ , as found experimentally for a flat SU-8, and we use for the cal-
culation of the fraction of the solid surface on which the drop is residing, fs , the
equation [65, 66]:
1
fs = , (4)
(b/a + 1)2
where a is the width, and b the spacing of the pillars, respectively.
On the other hand, the Wenzel’s model [59] assumes that the water seeps within
the recessed regions of the surface, until a complete wetting condition is eventually
152 A. Athanassiou
achieved. We calculated the theoretical values according to this model by applying

equation (5) for the estimation of the roughness factor r of the pillared surface [65,
66]:
4fs
r =1+ , (5)
(a/H )
where fs is the factor in the Cassie–Baxter model, and H is the height of the pillars.
In the case of the patterned pillars used in this work, the calculation of the WCA
following the Wenzel’s model is quite approximate, since it assumes that the water
drop resides always on SU-8, whereas a part of it is in contact with the silicon at
the bottom of the structures which is usually more hydrophilic than SU-8.
It is clear from Fig. 16 that the monotonic decrease of the measured WCA from
118.0◦ ± 1◦ to 82.0◦ ± 1◦ for increasing inter-pillar spacing is in contrast with the
trend predicted by both theoretical models. On the one hand, within the frame of
the Cassie–Baxter model, the air fraction of the surface (i.e., 1 − fs ), on which the
water droplet is deposited, should scale with the inter-pillar distance, increasing the
WCA values. On the other hand, the Wenzel’s picture describes that the interfacial
contact area between the water and the SU-8 substrate with θY < 90◦ decreases with
increasing inter-pillar spacing, explaining the increase of the WCA values. Careful
examination of our data reveals that the WCA values associated with the SU-8
micropatterns with the shortest spacings (14, 21 µm) are close to the theoretical
values predicted by the Cassie–Baxter picture. Under such circumstances, the liquid
can reasonably be assumed to be situated mainly on the top of the pillars without
penetrating into the recessed regions. On the contrary, the WCA value measured on
the sample characterized by the largest inter-pillar distance (77 µm) is close to the
Figure 16. Comparison between WCA values determined experimentally and those calculated theo-
retically by applying Cassie–Baxter’s and Wenzel’s models, for the SU-8 pillared structures (42 µm ×
42 µm) with different inter-pillar distances. The WCA recorded for a flat SU-8 substrate is also re-
ported.
value predicted by the Wenzel model, indicating that water is able to wet these types
of microstructures completely. Finally, for the patterns with intermediate spacings
(28, 35 and 42 µm), the WCA takes values in between those calculated on the
basis of these two theoretical models, which suggests that in these cases a partial
wetting could actually occur on the microstructured surfaces. In summary, these
results allow us to reasonably assume that as the inter-pillar spacing becomes larger
than a critical size threshold (about ∼20 µm), the wetting mechanism progressively
switches from a Cassie–Baxter’s to Wenzel’s type, crossing an intermediate regime
where water seeps into the pillars wetting only partially their vertical walls [67].
Subsequently, the surface of the SU-8 patterns was coated with TiO2 NRs in or-
der to increase the hydrophobicity of the SU-8 patterned substrate, and eventually
to render the wettability of the resulting organic/inorganic hybrid substrates switch-
able by UV-light. The high-magnification SEM inspection of the top surfaces of
the pillars confirms a continuous and crack-free NR coverage on such regions over
areas as large as several squared micrometers Fig. 17 (panel b). After the dipping
procedure in the NRs suspension the contact angle rose for all the patterned sub-
strates to values close to 140 ± 1◦ , enhancing in this way the hydrophobicity of
the starting materials. The surface geometry and the wettability are demonstrated
schematically in Fig. 17 (panels a and b), where are shown SEM images from the
top of an uncoated and of a coated pillar, and pictures of the water drops on both
samples. These results indicate that the dipping treatment of the pillars induces
extra-roughness to the substrates, which leads to the increase of the WCA values
[67].
Notably, the increase in WCA measured after NR addition ranges from about
20◦ up to 40◦ for patterns with progressively larger inter-pillar distances, which
suggests a clear synergy between the micro- and nano-scale surface structuring in
the modification of the wetting properties of the TiO2 -functionalized substrates.
In Fig. 18 is also shown a comparison between the WCA values measured on the
Figure 17. (a) and (b) High-magnification SEM images of a bare and a TiO2 –NR-coated top surface
of a pillar, respectively. In the corresponding top-right insets, the images of a water droplet in contact
with such surfaces are reported, for the respective cases. It is apparent that upon coating the substrates
with TiO2 enhance their hydrophobic character.
154 A. Athanassiou
Figure 18. Experimental WCA values on the SU-8 pillared (42 µm × 42 µm) microstructures be-
fore (•), and after TiO2 coating (). Moreover, theoretical WCA values on the pillars after TiO2
coating following the Cassie–Baxter’s model (), as a function of the inter-pillar spacing are also
shown. The WCA values recorded for an uncoated and TiO2 -coated flat SU-8 substrate are also re-
ported.
NR-covered pillars and the theoretical values predicted by the Cassie–Baxter model
(equation (2)). Applying in equation (2): θY = 96◦ , which is the WCA measured on
the flat SU-8 surface covered by TiO2 NRs, it is demonstrated that the experimental
values are greater than the theoretical ones, except for the one measured on the
structure with the largest spacing of 77 µm. Therefore, with an exception only in the
case of the 77 µm inter-pillar spacing, we can safely assume that the water droplets
are located on the top of the patterned pillars after their coverage with the TiO2 NRs.
The achieved dual scale roughness seems to enable the trapping of more numerous
air pockets underneath the water drop, thus preventing the liquid from seeping into
the void spaces among the pillars. This effect is similar to that observed on the lotus
leaves, on which the combined effect of micro-/nano-scale morphological features
is ultimately responsible for their superhydrophobic properties.
The discrepancy between the experimental and the theoretical WCA values on
the NR-functionalized SU-8 pillars can be ascribed to possibly enhanced nanor-
oughness induced on the SU-8 pillars compared to the one induced on the flat SU-8
after the dipping/coating procedure. The plausible change in the nanoroughness
would differentiate the overall nature of the surfaces in the two cases. Therefore,
the use of the WCA on the NR-covered flat SU-8 as the Young angle (θY = 96◦ )
would introduce an error in the theoretical calculations, since Cassie and Baxter de-
fine as the Young angle the WCA measured on the corresponding flat surface with
the same physicochemical characteristics. In fact, the actual Young angle would
exhibit a higher value due to the increased surface nanoroughness.
Figure 19. Variations in the WCA values on pillared samples with different inter-pillar distances
covered by TiO2 NRs after UV irradiation.
In order to alter the wettability of the afore-described surfaces covered with TiO2
NRs, UV light irradiation was employed by a pulsed laser source that, as previously
demonstrated, does not induce any photochemical damage to the surfactant mole-
cules of the NRs [61, 62]. The third harmonic (355 nm) of a Nd:YAG laser source
was employed at energy density of 5 mJ/cm2 , for 120 min (72 000 pulses) which
ensured the highest wettability changes. In Fig. 19 are reported the variations of
the WCA on all the samples after the UV irradiation. Upon UV irradiation, the sur-
faces become very hydrophilic, exhibiting a decrease in their WCA more than 100◦ .
The photoinduced hydrophilic character of our samples should be exclusively at-
tributed to the wettability behavior of the TiO2 upon UV irradiation.
It is worth remarking that the light-driven WCA excursions measured on the
micropillared structures are larger than those found for the corresponding flat sub-
strate (
WCA ≈ 65◦ ), demonstrating the beneficial effect of the micrometer-scale
texturing on the ultimate degree of hydrophilicity achievable. In this respect, it is
conceivable that the droplet-spreading over the substrate is facilitated by its front
experiencing the highly hydrophilic TiO2 film located both on the top and on the
side of the SU-8 pillars. The WCA values are expected to be smaller for narrower
inter-pillar spacing since the interfacial area of contact between the drop and the
highly hydrophilic surface is larger. Nevertheless, the results presented in Fig. 19
show that the lower WCA after the UV irradiation of the surfaces is achieved for
the pillar arrangement with the bigger spacing. A very plausible explanation is that
the large spacing between the SU-8 structures facilitates the functionalization of the
TiO2 NRs, to form uniform films on the top and the side of the SU-8 pillars as well
as on the recessed SiO2 /Si regions. On the contrary, as the inter-pillar spacing gets
narrower it is likely that the process of nanocrystal self-assembly may be inhibited
156 A. Athanassiou
Figure 20. Reversible wettability changes during alternating cycles of 120-min pulsed UV illumina-
tion and 40-day storage in the dark under ambient conditions for the flat and the patterned SU-8.
resulting in sparsely packed NR-films at the sides and the bottom of the pillars.
Thus, it is expected that after the UV irradiation of the different pillar arrangements
the most hydrophilic surfaces would be those covered with more densely packed
TiO2 NRs, i.e. with the larger-spaced pillars.
Subsequently to the light stimulation, prolonged storage of the samples in the
dark under ambient conditions reverts them back to their initial hydrophobic condi-
tion, close to that encountered in the freshly prepared TiO2 -coated SU-8 substrates.
Interestingly, the average recovery process is generally faster for the micropat-
terned samples (t1 ∼ 25 days) than for the flat surface counterpart (t2 ∼ 40 days).
The process of the reversible change of the WCA can be carried out over several
cycles of irradiation/dark storage, as demonstrated in Fig. 20, proving that the NR
coatings exhibit excellent reversibility in their wettability properties without being
affected by any significant photochemical fatigue [67].
Figure 21 shows the WCA on the structure of 35 µm-spaced pillars that can be
used as a characteristic example to summarize all the above results. Starting with a
surface of SU-8 having a WCA ∼ 80◦ (Figure 21a, I), we managed to increase its
hydrophobicity to ∼100◦ by micropatterning of the surface with pillars of specific
geometries (Fig. 21b, II). The next step was to cover the patterned surface with
TiO2 NRs producing a hybrid organic-inorganic structure that exhibits combined
micro and nano-roughness responsible for its superhydrophobic character (∼140◦ )
(Fig. 21c, III). Finally, this surface can be inverted into highly hydrophilic (∼30◦ )
by UV irradiation (Fig. 21d, IV).
3.2.3.3. Mechanism of UV-Induced Hydrophilicity on TiO2 NR-Coatings FT–IR
spectroscopy was used to investigate the behavior of the films exposed to UV ir-
radiation in order to understand the mechanism of the significant reduction of the
Figure 21. Summary of the overall procedure for the preparation of the hybrid SU-8/TiO2
NR-substrates with enhanced light-induced reversible wettability.
Figure 22. Water Contact angle values (a), and FT–IR spectra (b) of TiO2 NR-coatings during their
irradiation with UV-laser pulses.
WCA, presented above and shown again in Fig. 22a. As demonstrated in Fig. 22b,
a considerable increase in the hydroxyl stretching signal centered at 3300 cm−1 is
detectable, that is attributable to surface Ti–OH moiety [32–34]. The higher hy-
droxylation degree of the surface allows spreading of the water droplets, explaining
the low WCA values measured after irradiation. After dark storage for two months,
the occurrence of dehydroxylation process on the TiO2 surface is evidenced by a
clear decrease in the Ti–OH stretching band (Fig. 23b). As a macroscopic evidence
of these chemical modifications the WCA almost recovers its initial value as shown
in Fig. 23a.
Moreover, it can be observed that the stretching signals of alkyl chain of the
OLAC surfactants undergo a quite significant decrease upon UV irradiation. These
signals recover their intensity after the dark storage. This recovery cannot be due to
the adsorption of any organic compounds present in the environment, as confirmed
by our control experiments. We ascribe these reversible changes in the infrared ab-
sorption to conformational modifications of the surfactant molecules. Therefore, we
assume that only negligible photocatalytic degradation of the surfactant molecules
which surround the NRs occurs under our irradiation experiments.
158 A. Athanassiou
Figure 23. Water Contact angle values (a), and FT–IR spectra (b) of TiO2 NR-coatings during dark
storage, following UV irradiation, for different time periods.
Scheme 4. Representation of the increased coordinated hydroxyl groups on the TiO2 NR-coatings, after UV
irradiation, that promote the adsorption of water molecules, making the coating very hydrophilic. The rep-
resentation is shown on a flat substrate.
These experiments are in perfect agreement with the mechanistic pathways that
have been proposed in the literature to explain the hydrophobic to hydrophilic con-
version of a TiO2 surface. The coordinated hydroxyl groups promote adsorption
of further water multilayers, making the surfaces highly hydrophilic (Scheme 4).
The organized NR-arrays onto the surfaces naturally expose (011)/(101) anatase
facets that are particularly effective in the light-induced hydrophilicity mechanism
of TiO2 [5–11, 61, 62]. In addition, under the specific irradiation conditions de-
scribed in the Experimental Section the surfactant molecules anchored to the NR
surfaces are not subjected to any noticeable photocatalytic degradation [61, 62].
Therefore, the TiO2 NR-films are expected to have alternating very hydrophilic and
quite hydrophobic domains, the former related to the introduced –OH groups and
the latter associated with surfactant-protected TiO2 areas. Such surface arrange-
ments could lead to nanocapillary infiltration, allowing water droplets to spread out
over such TiO2 -based coatings [61, 62]. This mechanism ultimately ensures the
highest wettability excursion, and guarantees full recovery of the initial hydropho-
bic state in the dark [61, 62].
The hydroxylated surface is energetically metastable, since after dark storage,
the atmospheric oxygen replaces the hydroxyl groups on the TiO2 surface, finally
leading to the recovery of the initial hydrophobic state (Scheme 5). The observed re-
Scheme 5. Representation of the decreasing number of the coordinated hydroxyl groups on the TiO2 NR-
coatings after dark storage, that causes recovery of the coating to its previous hydrophobic state. The repre-
sentation is shown on a patterned substrate.
versible wettability changes are similar for TiO2 NR coatings deposited on bare ITO
as well as on polymeric substrates. The polymer coating does not suffer from any
UV driven modification, such as bond cleavage or photodegradation, as revealed by
control experiments performed on pure polymers.
4. Summary and Conclusions

We present organic and inorganic materials with a common fascinating property:
The reversible change of their surface wettability upon light irradiation. The ma-
terials investigated were organic photochromic molecules, and inorganic colloidal
nanocrystals. In the former case, when such molecules are incorporated into the
polymers, the wetting characteristics of the polymeric surfaces change when irra-
diated with laser beams of properly chosen photon energy. Their hydrophilicity is
enhanced upon UV-laser irradiation, and the process is reversed upon visible laser
irradiation. The mechanism involved in these transformations is the photoinduced
isomerization between the isomeric forms of the photochromic molecules that have
different polarities. In the latter case of the inorganic nanocrystals we focused on
organic-capped TiO2 nanorods that were used as coatings on several substrates.
These coatings exhibit a surface transition from a highly hydrophobic state to a
highly hydrophilic one under selective UV-laser irradiation, and the behavior is re-
versed under dark storage. The mechanism involved is a progressive increase in the
degree of surface hydroxylation of TiO2 upon UV irradiation. The surfactant mole-
cules that cover the NRs appear to undergo simultaneous conformational changes
without suffering any significant photocatalytic degradation.
Acknowledgements
The author would like to thank all individuals who had worked and are still working
on this project:
From NNL-National Nanotechnology Laboratory of CNR-INFM, Lecce, Italy:
Mr Gianvito Caputo, Mrs Barbara Cortese, Dr Despina Fragouli, Dr Concetta No-
bile, Dr Elisa Mele, Dr Luana Persano, Dr Dario Pisignano, Dr P. Davide Cozzoli,
Prof. Giuseppe Gigli and Prof. Roberto Cingolani.
From IESL-FORTH, Crete, Greece: Dr Maria Lygeraki, Dr Maria Farsari, Prof.
Costas Fotakis and Prof. Spiros Anastasiadis.
160 A. Athanassiou
References
1. X. M. Li, D. Reinhoudt and M. Crego-Calama, Chem. Soc. Rev. 36, 1350–1368 (2007).
2. X. J. Feng and L. Jang, Adv. Mater. 18, 3063–3078 (2006).
3. T. L. Thompson and J. T. Yates, Chem. Rev. 106, 4428–4453 (2006).
4. A. Athanassiou, M. I. Lygeraki, D. Pisignano, K. Lakiotaki, M. Varda, E. Mele, C. Fotakis, R. Cin-
golani and S. H. Anastasiadis, Langmuir 22, 2329–2333 (2006).
5. R. Wang, K. Hashimoto, A. Fujishima, M. Chikuni, E. Kojima, A. Kitamura, M. Shimohigoshi
and T. Watanabe, Nature 388, 431–432 (1997).
6. R. Wang, K. Hashimoto, A. Fujishima, M. Chikuni, E. Kojima, A. Kitamura, M. Shimohigoshi
and T. Watanabe, Adv. Mater. 10, 135–138 (1998).
7. R. Wang, S. Nobuyuki, A. Fujishima, T. Watanabe and K. Hashimoto, J. Phys. Chem. B 103,
2188–2194 (1999).
8. A. Nakajima, K. Shin-ichi, T. Watanabe and K. Hashimoto, J. Photochem. Photobiol. A: Chem.
146, 129–132 (2001).
9. A. Nakajima, K. Shin-ichi, T. Watanabe and K. Hashimoto, Langmuir 16, 7048–7050 (2000).
10. N. Sakai, A. Fujishima, T. Watanabe and K. Hashimoto, J. Phys. Chem. B 107, 1028–1035 (2003).
11. X. Feng, J. Zhai and L. Jiang, Angew. Chem. Int. Ed. 44, 5115–5118 (2005).
12. D. S. Kommireddy, A. A. Patel, T. G. Shutava, D. K. Mills and Y. M. Lvov, J. Nanosci. Nanotechol.
5, 1081–1087 (2005).
13. X. Feng, L. Feng, M. Jin, J. Zhai, L. Jiang and D. Zhu, J. Am. Chem. Soc. 126, 62–63 (2004).
14. D. A. LaVan, T. McGuire and R. Langer, Nature Biotechnol. 21, 1184–1191 (2003).
15. M. E. Napier and J. M. Sesimone, Polymer Reviews 47, 321–327 (2007).
16. B. Zhao, J. S. Moore and D. J. Beebe, Science 291, 1023–1026 (2001).
17. R. Blossey, Nature Mater. 2, 301–306 (2003).
18. J. C. Crano and R. J. Guglielmetti (Eds), Organic Photochromic and Thermochromic Compounds.
Plenum Press, New York (1999).
19. S. Abbot, J. Ralston, G. Reynolds and R. Hayes, Langmuir 15, 8923–8928 (1999).
20. N. Lake, J. Ralston and G. Reynolds, Langmuir 21, 11922–11931 (2005).
21. K. Ichimura, S.-K. Oh and M. Nakagawa, Science 288, 1624–1626 (2000).
22. C. Radüge, G. Papastavrou, D. G. Kurth and H. Motschmann, Eur. Phys. J. E 10, 103–114 (2003).
23. G. Möller, M. Harke and H. Motschmann, Langmuir 14, 4955–4957 (1998).
24. A. Bobrovsky, N. Boiko, V. Shibaev and J. Stumpe, J. Photochem. Photobiol. A: Chem. 163,
347–358 (2004).
25. R. Rosario, D. Gust, M. Hayes, F. Jahnke, J. Springer and A. Garcia, Langmuir 18, 8062–8069
(2002).
26. R. Rosario, D. Gust, A. A. Garcia, M. Hayes and J. L. Taraci, J. Phys. Chem. B 108, 12640–1264
(2004).
27. A. Athanassiou, M. Varda, E. Mele, M. I. Lygeraki, D. Pisignano, M. Farsari, C. Fotakis, R. Cin-
golani and S. H. Anastasiadis, Appl. Phys. A 83, 357–356 (2006).
28. E. Mele, D. Pisignano, M. Varda, M. Farsari, G. Filippidis, C. Fotakis, A. Athanassiou and R. Cin-
golani, Appl. Phys. Lett. 88, 203124 (2006).
29. N. Stevens, C. I. Priest, R. Sedev and J. Ralston, Langmuir 19, 3272–3275 (2003).
30. S. Wang, X. Feng, J. Yao and L. Jiang, Angew. Chem. Int. Ed. 45, 1264–1267 (2006).
31. H. S. Lim, D. Kwak, D. Y. Lee, S. G. Lee and K. Cho, J. Am. Chem. Soc. 129, 4128–4129 (2007).
32. N. Sakai, R. Wang, A. Fujishima, T. Watanabe and K. Hashimoto, Langmuir 14, 5918–5920
(1998).
33. N. Sakai, A. Fujishima, T. Watanabe and K. Hashimoto, J. Phys. Chem. B 105, 3023–3026 (2001).
34. M. Miyauchi, A. Nakajima, K. Hashimoto and T. Watanabe, Adv. Mater. 12, 1923–1927 (2000).
35. A. Fujishima and X. Zhang, C. R. Chimie 9, 750–760 (2006).
36. D. Lee, M. F. Rubner and R. E. Cohen, Nano Lett. 6, 2305–2312 (2006).
37. W. Y. Gan, S. W. Lam, K. Chiang, R. Amal, H. Zhao and M. P. Brungs, J. Mater. Chem. 17,
952–954 (2007).
38. X.-T. Zhang, O. Sato, M. Taguchi, Y. Einaga, T. Murakami and A. Fujishima, Chem. Mater. 17,
696–700 (2005).
39. X. Zhang, A. Fujishima, M. Jin, A. V. Emeline and T. Murakami, J. Phys. Chem. B 110, 25142–
25148 (2006).
40. N. P. Ernsting and T. Arthen-Engeland, J. Phys. Chem. 95, 5502–5509 (1991).
41. Y. Abe, R. Nakao, T. Horii, S. Okada and M. Irie, J. Photochem. Photobiol. A: Chem. 95, 209–214
(1996).
42. A. K. Chibisov and H. Görner, J. Photochem. Photobiol. A: Chem. 105, 261–267 (1997).
43. G. Smets, Adv. Polym. Sci. 50, 17–44 (1983).
44. A. Tork, F. Boudreault, M. Roberge, A. M. Ritcey, R. A. Lessard and T. V. Galstian, Appl. Optics
40, 1180–1186 (2001).
45. H. Bouas-Laurent and H. Durr, Pure Appl. Chem. 73, 639–665 (2001).
46. Z. Yoshimitsu, A. Nakajima, T. Watanabe and K. Hashimoto, Langmuir 18, 5818–5822 (2002).
47. A. Lafuma and D. Quéré, Nature Mater. 2, 457–460 (2003).
48. N. A. Patankar, Langmuir 19, 1249–1253 (2003).
49. A. Athanassiou, M. Kalyva, K. Lakiotaki, S. Georgiou and C. Fotakis, Adv. Mater. 17, 988–992
(2005).
50. M. H. Park, Y. J. Jang, H. M. Sung-Suh and M. M. Sung, Langmuir 20, 2257–2260 (2004).
51. S. Natarajan, D. A. Chang-Yen and B. K. Gale, J. Micromech. Microeng. 18, 045021 (2008).
52. S. A. Soper, S. M. Ford, S. Qi, R. L. McCarley, K. Kelly and M. C. Murphy, Anal. Chem. 72,
642–643 (2000).
53. J. Zhang, M. B. Chan-Park and S. R. Conner, Lab-on-a Chip 4, 646–653 (2004).
54. J. Hobley, V. Malatesta, R. Millini, L. Montanari and W. O. N. Parker, Phys. Chem. Chem. Phys.
1, 3259 (1999).
55. K. Y. Suh, Y. S. Kim and H. H. Lee, Adv. Mater. 13, 1386–1389 (2001).
56. D. Pisignano, L. Persano, G. Gigli, R. Cingolani, F. Babudri, G. M. Farinola and F. Naso, Appl.
Phys. Lett. 84, 1365–1367 (2004).
58. D. Fragouli, L. Persano, G. Paladini, D. Pisignano, R. Carzino, F. Pignatelli, R. Cingolani and
A. Athanassiou, Adv. Funct. Mater. 18, 1617–1623 (2008).
59. R. N. Wenzel, Ind. Eng. Chem. 28, 988 (1936).
60. X. Chen and S. S. Mao, Chem. Rev. 107, 2891–2959 (2007).
61. G. Caputo, C. Nobile, R. Buonsanti, T. Kipp, L. Manna, R. Cingolani, P. D. Cozzoli and
A. Athanassiou, J. Mater. Sci. 43, 3474 (2008).
62. G. Caputo, C. Nobile, T. Kipp, L. Blasi, V. Grillo, E. Carlino, L. Manna, R. Cingolani, P. D. Coz-
zoli and A. Athanassiou, J. Phys. Chem. C 112, 701–714 (2008).
63. M. Takeuchi, K. Sakamoto, G. Martra, S. Coluccia and M. Anpo, J. Phys. Chem. B 109, 15422–
15428 (2005).
64. M. Takeuchi, G. Martra, S. Coluccia and M. Anpo, J. Phys. Chem. C 111, 9811–9817 (2007).
65. L. Zhu, Y. Y. Feng, X. Y. Ye and Z. Y. Zhou, Sensors Actuators A, 130, 595–600 (2006).
162 A. Athanassiou
66. R. D. Narhe and D. A. Beysens, Langmuir 23, 6486–6489 (2007).

67. G. Caputo, B. Cortese, C. Nobile, M. Salerno, R. Cingolani, G. Gigli, P. D. Cozzoli and
A. Athanassiou, Adv. Funct. Mater. DOI: 10.1002/adfm.200800909 (2009).
Femtosecond Laser-Induced Surface Structures on Platinum
and Their Effects on Surface Wettability and Fibroblast
Cell Proliferation
E. Fadeeva a,∗ , S. Schlie a , J. Koch a , B. N. Chichkov a , A. Y. Vorobyev b and

Chunlei Guo b
a
Laser Zentrum Hannover e.V., Hollerithallee 8, 30419 Hannover, Germany
b
The Institute of Optics, University of Rochester, Rochester, NY 14627, USA
Abstract
In this work, we performed a systematic study of femtosecond laser-induced surface structures on platinum
and their effects on both hydrophobicity and fibroblast cell proliferation. Our finding is that the femtosec-
ond laser-induced surface structures suppress the fibroblast cell proliferation. This finding provides a way
for controllably inhibiting fibroblast proliferation that is important in a variety of biomedical applications.
Another finding is that there is a clear correlation between the femtosecond laser-induced hydrophobicity
and cell growth: the higher the hydrophobicity, the lower the cell proliferation. This indicates that wettability
tests can be a suitable technique for the prognosis of cell response to textured biomaterials.
Keywords
Laser manufacturing, metal surface treatment, surface topography, contact angle, fibroblast
1. Introduction
Since conventional biomaterial surfaces often do not have all desired properties for
in vivo applications, functionalization methods enabling a control over cell behav-
iour on an implant surface are a matter of increasing interest. One example is the
cochlea implant whose function is to restore hearing in deaf patients by electrical
stimulation of the auditory nerve [1]. The problem here is postoperative fibrotic tis-
sue accumulation that forms an additional insulating layer on electrode surface and
negatively affects the implant function. The major cell types of the connective tissue
are fibroblasts. The functionality of the cochlea implant could be clearly improved
by controlling fibroblast adhesion on the implant surface. Here, one very attractive
approach could be applying surface topographies for functionalization.
* To whom correspondence should be addressed. Fax: +49 511 2788-100; e-mail: e.fadeeva@lzh.de

164 E. Fadeeva et al.
Previously, specific cell responses to both nanometer and micrometer sized sur-
face structures have been reported. It has been shown that surface topographies
influence cell morphology, orientation, adhesion, and proliferation [2–5]. However,
until now there has been no commonly accepted approach that can predict the ef-
fect of certain topographies on cell behaviour. On the contrary, the effects of surface
topography on wetting characteristics are well studied. The influence of roughness
on the wettability is predicted by two well-established models developed by Wen-
zel [6] and Cassie and Baxter [7]. The Wenzel model presuming complete surface
wetting gives the relation cos W = r cos . Here, W is the measured contact
angle on a given structured surface, r is the ratio of the total structured surface
area to the corresponding flat surface area, and is the contact angle measured on
a flat, non-structured surface of the same material. Since r > 1, both hydropho-
bicity and hydrophilicity of the surface are enhanced by this roughness. In the
Cassie–Baxter model, presuming incomplete wetting, the contact angle is given
by cos CB = −1 + f (cos + 1), where cos CB is the measured contact angle
on the structured surface, is the contact angle on the flat, non-structured surface,
and f < 1 is the fraction of the solid in direct contact with the liquid drop. In this
way, the resulting macroscopic contact angle on textured solid can be increased
and thus hydrophobic properties of the solid surface can be amplified. Furthermore,
there are some theoretical predictions showing that certain topographies can render
any hydrophilic surface with non-zero microscopic contact angle into non-wetting
superhydrophobic surface [8]. Provided that chemical changes of the material are
excluded, the wettability alteration is just a characteristic of the surface topography.
For the prognosis of the effect of a huge variety of possible structures on cell re-
sponse, it is very attractive to use such well-studied topography characteristic as the
wettability.
For studying correlations between topography-induced wettability changes and
cell response, one needs a technique that provides structuring of biomaterials with-
out both chemical alteration and surface contamination. For this purpose, laser
surface structuring is one of suitable techniques. It has previously been demon-
strated that laser processing of biomaterials has several advantages over other
methods, namely low surface contamination, low mechanical damage, and con-
trollable surface texturing of biocomponents with complicated geometries [9, 10].
In the past, the surface structures on biomaterials have been fabricated using long-
pulse lasers [10–13]. With the advent of femtosecond lasers, it has recently been
shown that ultrashort pulsed laser processing has advantages over long-pulse laser
processing due to higher precision, reduced heat-affected zone, and a larger vari-
ety of surface structures [14–20]. In our research, we take advantages of ultrashort
laser materials processing by using a femtosecond laser for surface structuring
of platinum and study effects of femtosecond laser-induced surface structures on
both hydrophobicity and fibroblast cell proliferation. We found that femtosecond
laser-induced surface structures significantly reduce the fibroblast proliferation that
provides a way for producing biomaterial surfaces unfavorable for cell growth. Fur-
Femtosecond Laser-Induced Surface Structures on Platinum 165
thermore, we found a clear correlation between the wettability of structured surfaces

and fibroblast proliferation, namely, the higher the hydrophobicity, the lower the
cell proliferation. Based on this finding, we conclude that the wettability tests can
be used for the prognosis of cell response to fabricated surface structures. Our study
shows that femtosecond laser surface structuring is a promising technique for con-
trolling cell proliferation in a variety of biomedical applications.
2. Experimental
2.1. Materials
In our experiments, we used samples with a size of 5 × 5 × 0.25 mm3 prepared from
commercial rolled platinum foils with a purity of 99.99% (Goodfellow, Ltd). Prior
to femtosecond laser surface treatment, all samples were cleaned with methanol.
Untreated platinum plates and glass slides were used as references.
2.2. Femtosecond Laser Treatment of Platinum Samples
For producing surface structures, we employed an amplified Ti:sapphire laser sys-
tem that generates 65-fs laser pulses with a pulse energy of about 1 mJ at a maxi-
mum pulse repetition rate of 1 kHz and a central wavelength of λ = 0.8 µm. Surface
structures were fabricated using a direct femtosecond ablation technique [20]. We
textured a surface area with a diameter of 4 mm by raster scanning a femtosecond
laser beam across the sample. The laser beam was polarized and focused normally
onto a platinum sample mounted vertically on an X–Y motorized translation stage.
An achromatic lens with a focal distance of 20 cm was used for focusing the laser
beam. The scanning direction was horizontal (X-direction) and parallel to the po-
larization direction. In all experiments, the scanning speed in the X-direction was
1 mm/s and the translation step along the Y-direction was 100 µm. To measure the
incident pulse energy, a certain fraction of the incident light was split off by a beam
splitter and measured with a pyroelectric joulemeter. The uncertainty in this mea-
surement was estimated to be 5%. To obtain different femtosecond laser-induced
structural features, we varied laser fluence F from 0.084 J/cm2 up to 7.7 J/cm2 and
pulse repetition rate f from 10 to 1000 Hz. The laser fluence of the incident light
was varied by varying the distance between the focusing lens and the sample.
In our study, we fabricated two types of surface structures. The first type of
structures is periodic surface gratings that we produced using the phenomenon of
laser-induced periodic surface structures (LIPSS). For long laser pulses, this phe-
nomenon has been known for a long time [21–25] and it has recently been shown
that LIPSS can also be produced on metals using femtosecond laser pulses at near-
ablation threshold values of the laser fluence [26]. The second type of fabricated
structures is a combination of random nano- and micro-roughness that we produced
at laser fluence noticeably above the ablation threshold value. By varying laser ab-
lation conditions, we prepared twelve Pt samples (#1–12) structured with various
femtosecond laser-induced periodic structures (FLIPSS) and twelve samples struc-

tured with various combinations of random nano- and micro-roughness (samples
#13–24). The topography of the structured samples was studied using a scanning
electron microscope (SEM). In our experiment, all the surface structures were pro-
duced in air at a pressure of 1 atmosphere. Laser processing of a platinum sample
in air may alter its surface elemental composition due to possible laser-induced
chemical reactions. To clarify this point, we performed a comparative study of the
elemental composition of both treated and untreated platinum samples using an en-
ergy dispersive X-ray analysis and found that there was no significant change in the
elemental composition between the treated and untreated platinum surfaces.
2.3. Contact Angle Measurements
To investigate the effect of femtosecond laser-induced surface topographies on wet-
ting characteristics, we performed contact angle measurements using a video-based
optical contact angle measuring system (OCA 40 Micro, DataPhysics Instruments
GmbH, Germany). To measure the water contact angle, θM (θM — contact angle
found by measurement), we used the sessile drop method that determines the ad-
vancing contact angle in a few seconds from the moment when a water drop is
brought into contact with the surface. A water drop with approximately 6 µl vol-
ume was gently placed on the structured surface using a computer-controlled stage.
A sideview photograph was taken and used to automatically measure the contact
angle. All experiments were carried out under normal atmospheric conditions and
at an ambient temperature of 20◦ C. Each experimental result is an average of at
least five individual measurements.
2.4. Fibroblast Proliferation Experiments
For studying the influence of femtosecond laser-induced surface structures on cell
behaviour, we examined proliferation profiles of fibroblasts grown on the pre-
pared platinum samples. Non-structured platinum samples and glass slides served
as references. The samples were placed in 24-well plates filled with 2 ml Dul-
becco’s Modified Eagle’s Medium (Sigma, Taufkirchen, Germany) with the ad-
dition of penicillin/streptomycin (100 U/ml and 10 mg/ml, respectively), particin
(0.5 mg/ml) and supplemented with 10% fetal calf serum. An average cell density
of 13.9 × 105 cells/ml was seeded out. Then, the plates were placed into a cell cul-
ture incubator (Heraeus, Hanau, Germany), where a 95%:5% air:CO2 atmosphere
and 80% humidity were maintained. After a cultivation time of 48 h, the cells were
trypsinized and counted. For a better comparison between experiments, the cell
density was normalized to the seeding density at the beginning of the cell growth
(at 0 h time) and given in percent.

In our experiment, FLIPSS were found to be generated on platinum in the laser
fluence range of 0.084–0.35 J/cm2 and typical examples of FLIPSS produced at the
Figure 1. SEM micrographs of femtosecond laser-induced periodic surface structures fabricated

on platinum by laser treatment at F = 0.35 J/cm2 and different repetition rates: (a) f = 50 Hz;
(b) f = 200 Hz; (c) f = 500 Hz; (d) f = 1000 Hz.
same laser fluence but various pulse repetition rates are shown in Fig. 1. Usually,
LIPSS generated by long laser pulses [21–25] have a period close to the incident
laser wavelength and are oriented perpendicularly to the polarization of the incident
light. However, femtosecond laser-induced periodic surface structures exhibit two
distinctive features. First, the FLIPSS are covered with nanoroughness. Second, the
FLIPSS period (576 nm in Fig. 1(c)) is much shorter than the free-space laser wave-
length (λ = 800 nm). As can be seen in Fig. 1, FLIPSS become more pronounced
with increasing the number of overlapping laser pulses. The water contact angle
measurements show that structuring with FLIPSS turns an originally hydrophilic
flat platinum surface (with the water contact angle of θM = 81◦ ) into a hydropho-
bic one; moreover, the surfaces structured with more pronounced FLIPSS exhibit
more hydrophobic behavior (see values of the measured water contact angle, θM , in
Fig. 1). Therefore, by varying both laser fluence and number of overlapping laser
pulses, one can vary structural features of FLIPSS and thus control the contact
angle. In our experiment that included a group of twelve samples (#1–12) with dif-
ferent FLIPSS structures, the water contact angle varied from 81◦ to 129◦ .
In our study, we found that generation of FLIPSS ceases quickly as the laser flu-
ence increases above 0.35 J/cm2 and in the range of 0.45 J/cm2 < F < 7.7 J/cm2 ,
Figure 2. SEM micrographs of femtosecond laser-induced random nano- and micro-structures fabri-
cated on platinum by laser treatment at F = 7.7 J/cm2 and different repetition rates: (a) f = 200 Hz;
(b) f = 500 Hz; (c) f = 1000 Hz.
a combination of random nano- and micro-roughness becomes a dominant struc-

tural feature. As the laser fluence increases above 8.1 J/cm2 , nanoroughness begins
to vanisch and microroughness becomes a dominant surface topography. By vary-
ing laser processing conditions, we prepared twelve samples (#13–24) with various
combinations of random nano- and micro-roughness. Examples of this type of sur-
face structures produced at F = 7.7 J/cm2 are shown in Fig. 2. It is seen in Fig. 2(c)
that the surface has a porous structure consisting of nano- and micro-scale cavities,
nanoprotrusions, and microscale aggregates that appear to be formed by liquid-like
coalescence of nanoparticles. We note that on the surface of our samples there is
some amount of deposited nanoparticles due to re-deposition of a fraction of ab-
lated material back onto the sample. The water contact angle measurements show
that all the samples (#13–24) textured with this type of surface structure are more
hydrophobic than those with FLIPSS. The water contact angles for this group of
samples were measured to be in the range of 129–158◦ . We note that in our experi-
ment the transition from FLIPSS to pure random roughness at laser fluence slightly
above 0.35 J/cm2 results in a smooth change of the wettability. This leads us to be-
lieve that the FLIPSS periodicity does not have any specific effect on the wettability
behavior.
Due to the fact that in our experiment the observed changes in wettability orig-
inate only from the topography, the increase of the water contact angle can be
explained by incomplete wetting of the surface structure. This means that the wa-
ter drop does not follow the solid topography but rests on a heterogeneous surface
consisting of solid and gaseous phases. Such incomplete wetting has been proposed
in [8] for producing roughness-induced non-wetting on originally hydrophilic sur-
faces. In other words, the topography-dependent increase of the water contact angle
in our experiment reflects the reduction of solid surface area being in contact with
the water drop.
In our study, all the structured samples (#1–24) were tested in vitro for fibroblast
proliferation. Since fibroblasts are strongly grown on glass slides, we used glass
slides as references. Non-structured platinum samples have been found to have
Figure 3. Fibroblast proliferation versus measured water contact angle on structured platinum sam-
ples.
comparable strong cell growth results. In our tests, all the structured samples ex-
hibited a decreased fibroblast growth as compared with non-structured samples.
Since we found no chemical changes in all structured samples, the only possible
reason is the fabricated surface structures.
Furthermore, we found a clear correlation between the water contact angle and
fibroblast proliferation: the higher the hydrophobicity, the lower the proliferation
(see Fig. 3). Taking into account that hydrophobicity enhancement is induced by
reducing the solid surface area being in contact with the water drop, one can con-
clude that, similar to a water droplet, a cell will also have a reduced solid surface
area available for contacting and adhesion. As is well known, proliferation, dif-
ferentiation, and organization of the cytoskeleton are only possible if cells attach
to a biomaterial surface [27, 28]. The adhesion, mediated by adhesion receptors
within the cell membrane and ligands from the extracellular matrix, can be influ-
enced by biomaterial properties; therefore poor adhesion to the surface manifests
itself in a decreased cell growth. Previously, it has been demonstrated that effects
of cell growth, attachment, and spreading can be controlled by substrate patterning
with micro- and nanosized cell-adhesive islands separated by non-adhesive regions
[29, 30]. We believe that similar effects occur on our samples patterned with fem-
tosecond laser-induced structures that are ‘seen’ by the cells as adhesive islands
separated by non-adhesive areas.
In summary, studied femtosecond laser-induced surface structures significantly
reduce the fibroblast proliferation that provides a way for producing biomaterial
surfaces unfavorable for cell growth. Furthermore, our study shows a clear corre-
lation between the wettability of structured surfaces and fibroblast proliferation,
namely, the higher the hydrophobicity, the lower the cell proliferation. Although
the behavior of a water drop only roughly simulates the cell response to structured
surface, our results show that wettability tests can be used for the prognosis of cell
response to femtosecond laser-textured biomaterials.
4. Conclusions
The major results of our detailed study of femtosecond laser-induced surface struc-
tures on platinum and their effects on both hydrophobicity and fibroblast cell pro-
liferation are as follows. (1) We find that the femtosecond laser-induced surface
structures suppress the fibroblast cell proliferation. This finding provides a way
for controllably inhibiting fibroblast proliferation that is important in a variety
of biomedical applications. (2) We show that there is a clear correlation between
the femtosecond laser-induced hydrophobicity and cell growth: the higher the hy-
drophobicity, the lower the cell proliferation. This finding leads us to conclude that
wettability tests can be a suitable technique for the prognosis of cell response to
textured biomaterials.
Acknowledgements
This work was supported by The German Research Foundation, DFG SFB599
“Sustaintable Bioresorbing and Permanent Implants of Metallic and Ceramic Ma-
terials”, DFG Transregio Project TR37, and The National Science Foundation in
the US.
References
1. U. Reich, P. P. Mueller, E. Fadeeva, B. N. Chichkov, T. Stoever, T. Fabian, T. Lenarz and G. Reuter,
J. Biomed Mater. Res. Part B 87B, 146–153 (2008).
2. P. Clark, P. Connolly and G. R. Moores, J. Cell Sci. 103, 287–292 (1992).
3. X. F. Walboomers and J. A. Jansen, Odontology 89, 2–11 (2001).
4. C. C. Berry, G. Campbell, A. Spadiccino, M. Robertson and A. S. G. Curtis, Biomaterials 25,
5781–5788 (2004).
5. M. J. Dalby, M. O. Riehle, D. S. Sutherland, H. Agheli and A. S. G. Curtis, Biomaterials 25,
5415–5422 (2004).
8. S. Herminghaus, Europhys. Lett. 52, 165–170 (2000).
9. A. Gaggl, G. Schultes, W. D. Müller and H. Kärcher, Biomaterials 21, 1067–1073 (2000).
10. G. Petö, A. Karacs, Z. Pászti, L. Guczi, T. Divinyi and A. Joób, Appl. Surface Sci. 186, 7–13
(2002).
11. A. Joób-Fancsaly, T. Divinyi, A. Fazekas, C. Daroczi, A. Karacs and G. Petõ, Smart Mater. Struct.
11, 819–824 (2002).
12. C. Hallgren, H. Reimers, H. D. Chakarov, J. Gold and A. Wennerberg, Biomaterials 24, 701–710
(2003).
13. A. Karacs, A. Joob-Fancsaly, T. Divinyi, G. Peto and G. Kovach, Mater. Sci. Eng. C 23, 431–435
(2003).
14. B. N. Chichkov, C. Momma, S. Nolte, F. von Alvensleben and A. Tünnermann, Appl. Phys. A 63,
109–115 (1996).
15. A. Semerok, C. Chaleard, V. Detaille, J. L. Lacour, P. Mauchin, P. Meynadier, C. Nouvellon,
B. Sallé, P. Palianov, M. Perdrix and G. Petite, Appl. Surface Sci. 138–139, 311–314 (1999).
16. V. Margetic, A. Pakulev, A. Stockhaus, M. Bolshov, K. Niemax and R. Hergenröder, Spectrochim.
Acta B 55, 1771–1785 (2000).
17. R. Le Harzic, N. Huot, E. Audouard, C. Jonin, P. Laporte, S. Valette, A. Fraczkiewicz and R. For-
tunier, Appl. Phys. Lett. 80, 3886–3888 (2002).
18. J. Koch, F. Korte, T. Bauer, C. Fallnich, A. Ostendorf and B. N. Chichkov, Appl. Phys. A 81,
325–328 (2005).
19. Y. P. Meshcheryakov and N. M. Bulgakova, Appl. Phys. A 82, 363–368 (2006).
20. A. Y. Vorobyev and C. Guo, Optics Express 14, 2164–2169 (2006).
21. D. C. Emmony, R. P. Howson and L. J. Willis, Appl. Phys. Lett. 23, 598–600 (1973).
22. Z. Guosheng, P. M. Fauchet and A. E. Siegman, Phys. Rev. B 26, 5366–5381 (1982).
23. J. E. Sipe, J. F. Young, J. F. Preston and H. M. van Driel, Phys. Rev. B 27, 1141–1154 (1983).
24. S. E. Clark and D. C. Emmony, Phys. Rev. B 40, 2031–2041 (1989).
25. A. M. Bonch-Bruevich, M. N. Libenson, V. S. Makin and V. V. Trubaev, Opt. Eng. 31, 718–730
(1992).
26. A. Y. Vorobyev, V. S. Makin and C. Guo, J. Appl. Phys. 101, 034903 (4 pages) (2007).
27. F. G. Giancotti and E. Ruoslahti, Science 285, 1028–1032 (1999).
28. E. A. Clark and R. Hynes. J. Biol. Chem. 271, 14814–14818 (1996).
29. M. Arnold, E. A. Cavalcani-Adam, R. Class, J. Blümmel, W. Eck, M. Kantlehner, H. Kessler and
J. P. Spatz, ChemPhysChem 5, 383–388 (2004).
30. S. C. Chen, M. Mrksich, S. Huang, G. M. Whitesides and D. E. Ingber, Science 276, 1425–1428
(1997).
A Novel Design of Water- and Oil-Repellent Surface
Modifier Having Double-Fluoroalkyl Groups
Tokuzo Kawase ∗ , Shinsuke Ohshita, Chihiro Yoshimasu and Tasuo Oida

Department of Chemistry and Materials Technology, Kyoto Institute of Technology, Matsugasaki,
Sakyo-ku, Kyoto 606-8585, Japan
Abstract
The repellency of fluoropolymer finishes depends on the structures of fluorocarbon segment and nonfluori-
nated segment of the molecule, the orientation of the fluorocarbon tail, and the distribution and the amount
of the fluorocarbon moiety on the surface. In this work, based on the new strategy for introduction of double
semifluorinated segments as side groups, an effective and powerful water- and oil-repellent fluoropolymer
was developed.
First, a diester of 2,2-bis(hydroxymethyl)-propanoic acid using C8 F17 (CH2 )10 COOH was prepared, where
two semifluorinated segments are end-capped on trimethylene chain. Effect of double semifluorinated
structure on the orientation of the fluorocarbon tail was evaluated from its monolayer behavior (surface
pressure–area measurement). As expected, a rigid monolayer of the acid was formed on water even at ele-
vated temperatures.
Next, the methacrylate monomer having two (double) semifluorinated segments using HEMA was syn-
thesized and polymerized. Based on contact angle measurements of decane and water on the cast film of
PMMA containing this double semifluorinated polymer, the surface was found to be highly water- and oil-
repellent even at the minimum addition amount of double semifluorinated polymer (0.01 wt%/PMMA). The
effectiveness was evaluated on the basis of Cassie equation, and the alignment of double semifluorinated
groups on the surface was also discussed.
Keywords
Double semifluorinated group, fluorinated polymer, surface modification, contact angle, surface free en-
ergy, repellency, Cassie equation
1. Introduction
The wetting of solid surfaces is a very important phenomenon and, as widely ac-
cepted, is governed by both the chemical composition and the geometrical structure
of the surface. In the development of polymers with low surface energies, it is com-
mon to introduce a fluorinated group onto the polymer backbone. For example,
* To whom correspondence should be addressed. Tel. and Fax: +81-75-724-7515; e-mail: kawase@kit.ac.jp

174 T. Kawase et al.
acrylic and methacrylic polymers having fluorinated ester side groups are commer-
cially available as typical low surface energy coating materials.
In the field of wetting, super-hydrophobicity, namely water contact angle higher
than 150◦ , has attracted much interest for both fundamental research and practical
applications, such as non-sticking of snow. As well referred, Barthlott and Neinhuis
studied the relation between hydrophobicity and the topograhpical structures on the
surface of a plant leaf [1], and Ball also reported that super-hydrophobicity of a
lotus leaf is based on surface roughness and presence of wax [2]. In essence, super-
hydrophobicity can be achieved by a combination of chemical hydrophobization
and physical roughening of the surfaces.
In general, super-hydrophobic surfaces have been produced mainly in the follow-
ing two ways. One is to create a highly rough structure on a hydrophobic surface,
and the other is to modify a rough surface with materials having low surface free
energy such as fluoropolymers. In the latter cases, to obtain super-hydrophobic
surfaces, coating with low surface energy materials such as fluorosilane is often
necessary. In fact, many super-hydrophobic surfaces have been developed, e.g.
by solidification of melted alkylketene dimer [3], plasma polymerization/etching
of polypropylene (PP) in the presence of poly(tetrafluoroethylene) (PTFE) [4],
plasma-enhanced CVD of trimethylmethoxysilane [5], anodic oxidation of alu-
minum and modification with fluorinated monoalkylphosphate [6], and so on.
With respect to fluoroalkylsilanes, Kawase et al. reported the synthesis of novel
fluoroalkyl end-capped silane oligomers, Rf–[CH2 CHSi(OMe)3 ]n –Rf (n = 2, 3),
and clarified that various substrates including glass [7], cellulose, poly(ethylene
terephthalate) (PET) [8] etc. were much more effectively modified by these
oligomeric silanes than by the corresponding single-type fluoroalkyl silane. Kawase
et al. also synthesized a wide range of fluoroalkyl end-capped oligomeric sur-
factants, and studied their surface properties and concluded that the end-capped
structure of fluoroalkyl groups was essential for the superior property of these sur-
factants [9]. This means that the performance of fluoroalkyl groups to lower the
surface free energy would be effective when two fluoroalkyl groups can come in
close contact.
Quite similar behavior has been shown for gemini surfactants [10]. They are
known to show higher efficiency in decreasing the surface tension of water than
the corresponding monomeric surfactants, and achieve one to two orders of mag-
nitude lower cmc’s and smaller surface tensions at their cmc’s (γcmc ) than the
corresponding monomeric surfactants. From the viewpoint of molecular structure,
gemini surfactants are made up of two identical amphiphilic moieties having the
structure of conventional (monomeric) surfactants connected by a spacer group at
the level of, or in close vicinity to, the head groups. Therefore, two hydrophobic
tail groups can easily come close, nearly in contact, and a rigid monolayer can be
formed, which has also been confirmed by the surface pressure–area (π –A) mea-
surements.
Double Fluoroalkylated Surface Modifier 175
In this work, referring to the structural feature of gemini surfactants, we have

designed a quite effective and powerful water- and oil-repellent fluoropolymer by
introducing the new concept of double semifluorinated segments aligned closely.
2. Experimental
2.1. Materials and Instruments
All the reagents and solvents were purchased from Wako Pure Chemical Indus-
tries. Water was distilled using the GS-2000 apparatus (Advantec Co. Ltd., Japan).
1 H and 19 F NMR spectra were recorded in CDCl solution on a Brucker AC 300
3
spectrometer. Infrared spectra were recorded on a Nicolet Avatar 370 DTGS FTIR
spectrometer.
Surface tension measurements were carried out on a Shimadzu ST-1 surface ten-
siometer at 25◦ C. The surface pressure was measured by an automated Langmuir
film balance apparatus FSD-220 (USI Systems, Japan). Contact angles were mea-
sured using an S150 contact angle meter (Kyowa Interface Science Co. Ltd., Japan).
2.2. Synthesis of Heptadecafluorooctylalkanoic Acid (C8 F17 (CH2 )n COOH,
n = 2, 3, 4, 6, 10)
According to Brace [11], heptadecafluorooctyl iodide (C8 F17 I) was reacted with
ω-olefinic acid methyl esters (H2 C=CH–(CH2 )n−2 –COOCH3 , n = 2, 3, 4, 6, 10)
in the presence of 2,2 -azobisisobutyronitrile (AIBN) for n = 2, 3 or di-tert-butyl
peroxide (DTBPO) for n = 4, 6, 10 at 80◦ C or 120◦ C, respectively, for 8 h un-
der nitrogen atmosphere. Without any purification, the mixture was dissolved in
ethanol, to which activated zinc was added, and a small amount of acetic acid was
also added. The mixture was heated to 70◦ C under vigorous stirring for 5 h. After
filtration, ethanol was removed and the residue was extracted with diethyl ether.
The organic layer was washed with NaHCO3 -aq and brine, dried over Na2 SO4 , and
concentrated to afford heptadecafluorooctylalkanoic acid methyl ester.
The ester was dissolved in ethanol and heated to 60◦ C, followed by the addition
of KOH (5 equiv.). After 15 min, H2 O was added and stirring was continued at 80◦ C
for 5 h. The reaction mixture was neutralized with 1N–HCl and concentrated. The
residue was extracted with diethyl ether and dried over Na2 SO4 and concentrated
to afford heptadecafluorooctylalkanoic acid, 1 (Scheme 1), in good yields. In the
cases of n = 6, 10, recrystallization was carried using hexane.
2.3. Synthesis of 2,2-Bis(heptadecafluorooctylundecanoyloxymethyl)propanoic
Acid
According to Xiang et al. [12], heptadecafluorooctylundecanoyl chloride pre-
pared from heptadecafluorooctylundecanoic acid 1e (Scheme 2) and SOCl2 was
reacted with benzyl 2,2-bis(hydroxymethyl)propanoate in ethyl acetate in the pres-
ence of N ,N -dimethylaminopyridine (DMAP) at room temperature to afford ben-
zyl 2,2-bis(heptadecafluorooctylundecanoyloxymethyl)propanoate almost quanti-
Scheme 1.
Scheme 2.
tatively. To a solution of this benzyl ester in ethanol-THF (5:1), Pd(OH)2 /C

was added. The heterogeneous solution was stirred overnight at ambient tem-
perature under hydrogen. After filtration and concentration, the residue was re-
crystallized from hexane to afford the target double fluoroalkyl type acid, 2,2-
bis(heptadecafluorooctylundecanoyloxymethyl)propanoic acid 2 (Scheme 2).
Similarly, double alkyl type acid, 2,2-bis(octadecanoyloxymethyl)propanoic
acid, was prepared using stearoyl chloride.
2.4. Synthesis of Double Fluoroalkyl Type Monomer
First, double semifluorinated type acid 2 (Scheme 3) was converted to its acid chlo-
ride using SOCl2 . A solution of acid chloride in ethyl acetate was added dropwise to
a solution of 2-hydroxyethylmethacrylate (HEMA) and DMAP in ethyl acetate and
the reaction mixture was stirred overnight. After filtration, the organic layer was
washed with aq NaHCO3 . After dried over Na2 SO4 , ethyl acetate was removed
under reduced pressure to afford double fluoroalkyl type monomer 3.
Under similar conditions, single fluoroalkyl type acrylate monomer 5 (Scheme 4)
was prepared using acid chloride of 1e (Scheme 2) and HEMA.
Scheme 3.
Scheme 4.
2.5. Polymerization
To a solution of semifluorinated acrylate monomer 3 (Scheme 3) in 2-butanone,
AIBN (10 mol%) was added as an initiator. The mixture was stirred and refluxed
for 8 h under nitrogen atmosphere. The reaction mixture was concentrated to af-
ford the double semifluorinated type homopolymer 4 (Scheme 3). Similarly, single
semifluorinated type homopolymer 6 (Scheme 4) was prepared. GPC measurements
were carried out using a Shodex GPC-101 system consisted of an HPLC pump, a re-
fractive index detector (RI-100), and Shodex KD-806M column. The mobile phase
employed was DMF (1.0 ml/min) at 40◦ C.
2.6. Preparation of Cast Film
The solution of poly(methyl methacrylate) (PMMA) was prepared in acetone (0.5 g
in 10 ml). Semifluorinated polymers were dissolved into the PMMA solution to
adjust the additive concentration at 1.0 wt% or 0.01 wt% with respect to PMMA.
Then, the solution was spread over a Teflon petri dish and acetone was vaporized at
ambient temperature.
2.7. Surface Pressure–Area Measurement
Sample solution was prepared by dissolving each acid (20 µmol) in 10 ml benzene.
A drop of sample solution (10 µl) was placed on aqueous HCl subphase (0.01 M)
in a Teflon trough with an available surface area of 100 mm × 171 mm using a
syringe. After standing for 30 min at controlled temperature, the surface area was
compressed by moving Teflon barrier at the constant speed of 0.1 mm/s, and surface
pressure was recorded using the platinum Wilhelmy plate on the Langmuir film
balance apparatus FSD-220 (USI Systems, Japan).
2.8. Contact Angle Measurement
To determine the surface free energies of cast film of PMMA, contact angles of
water and decane were directly measured using Kyowa contact angle meter S150.
The cast film was held on an Al plate and a 2 µl liquid drop was placed on the
surface of the film. Measurements were repeated for at least 5 drops. The standard
deviation in measurements was less than 3◦ . All measurements were done at 25◦ C.
2.9. Determination of Surface Free Energies
According to Fowkes [13], the total surface free energy, γ , is assumed to be the sum
of contributions from different intermolecular surface forces, namely, the dispersion
(d) and polar components (p), and the geometric mean expression for the dispersion
attraction forces:
γ = γ d + γ p, (1)
γAB = γA + γB − 2(γAd γBd )1/2 . (2)
Owens and Wendt [14] also adapted the geometric mean expression for the polar
attraction forces. Thus, the work of adhesion WA would be expressed as:
p p
WA = γL (cos θ + 1) = 2(γSd γLd )1/2 + 2(γS γL )1/2 . (3)
p
As the values of γL , γLd and γL for water and decane are known, it is possible to
p
determine the values of γSd and γS using contact angles.

It has been well accepted that the wettability of monolayer-coated surfaces is de-
termined by the nature and packing of the outermost (or terminal) atoms in the
exposed surface. Commercially available fluorinated ester side group containing
acrylic and methacrylic polymers are known to have a problem that reconstruc-
tion at the surface often occurs in contact with polar liquids such as water, and
this limits the practical application of these materials. To prevent surface recon-
struction, several approaches have been suggested. Bain and Whitesides studied
self-assembled monolayers (SAMs) of alkanethiols on gold to create a uniformly
organized trifluoromethyl (–CF3 ) array on the surface [15]. Such a uniformly orga-
nized trifluoromethyl (–CF3 ) array would allow a surface with the lowest possible
surface free energy, but this SAM technique would be impractical for large-scale
applications and also for coatings of polymers. Chaudhury and Whitesides also
suggested an alternative approach to use the semifluorinated materials, which can
be expected to form well-ordered monolayer film and avoid SAM techniques to
produce uniform –CF3 surfaces [16]. On the other hand, Wang et al. reported that
a film of poly(styrene-b-semifluorinated side chain) block copolymer showed an
extremely low critical surface tension (8 mN/m) because its fluorinated groups are
located in a highly ordered liquid crystalline smectic B phase on the surface and the
–CF3 end groups are arranged in a hexagonally organized layer [17]. This would
be ideal also for semifluorinated side groups to form liquid crystalline (LC) phases,
which would effectively freeze the surface structure and prevent surface reconstruc-
tion processes. Bearing these strategies in mind, we developed quite effective and
powerful water- and oil-repellent fluoropolymers by introducing the new concept
of double fluoroalkyl segments aligned closely.
3.1. Semifluorinated Alkanoic Acid (Single Chain Type)
Shafrin and Zisman reported wetting characteristics of various heptadecanoic acid
derivatives having perfluoroalkyl groups in the terminal 17-position [18]. Based on
the critical surface tension, the effect of the fluorinated segment on wettability was
discussed in terms of the number of fluorine-substituted carbon atoms (Cn F2n+1 )
in the hydrophobe. They concluded that perfluoroheptyl segment (n = 7) was suffi-
ciently long enough to shield the hydrocarbon segment from the external boundary.
In addition, Genzer et al. [19] made NEXAFS measurements on the orientation of
semifluorinated alkanethiols, F(CF2 )n (CH2 )16−n SH in the form of self-assembled
monolayers (SAMs) on gold surface, and reported a similar effect of the number
of CF2 groups, where the intensity of the 1s → σ ∗ transition for the C–F bond in
NEXAFS spectra increased with increasing n and leveled off for n 7. Therefore,
also considering the availability of perfluoroalkyl iodide, we employed heptade-
cafluorooctyl group (C8 F17 –) as perfluoroalkyl segment of semifluorinated alkanoic
acids in this work.
As shown in Scheme 1, Brace prepared a wide range of semifluorinated alkanoic
acids by a three-step synthesis of fluoroalkyl radical addition reaction of perfluo-
roalkyl iodide with ω-olefinic fatty acid methyl ester, reduction of resulting iodide
and followed by hydrolysis [11]. We also prepared 5 semifluorinated alkanoic
acids 1 with a terminal C8 F17 – segment, C8 F17 –(CH2 )n –COOH (n = 2, 3, 4, 6, 10),
by following Brace’s method with some modification. To distinguish from the
next double semifluorinated alkanoic acid, these semifluorinated alkanoic acids are
named as single chain type acids.
Figure 1. Effect of –(CH2 )– number of semifluorinated alkanoic acids 1 (Scheme 1) on surface ten-
sion–concentration curves at 25◦ C, pH = 13 (0.1 M KOH).
3.2. Surface Tension Measurement on Single Chain Type Acid

As all single chain type acids are also considered to act as a surfactant, surface
tension measurements were carried out on 0.1 M KOH solutions and the effect of
the number of –CH2 – on surface properties was studied.
As can be seen in Fig. 1, all acids gave clear cmc values in the order of n =
6, 4, 10, 3, 2 and all the γcmc values were lower than 20 mN/m. These γcmc values
are quite comparable to that of perfluorooctanoic acid as suggested by Shafrin and
Zisman [18]. In addition, the molecular occupied areas in the adsorbed monolayers
were calculated from the surface excess, , determined from the slope of the γ −
log C curves below the cmc. Interestingly, all slopes were similar and the molecular
occupied areas were around 0.5 nm2 .
3.3. Surface Pressure–Area (π –A) Measurement on Single Chain Type Acid
Judging from the cmc and γcmc values, C8 F17 –(CH2 )6 –COOH 1d seemed to be
the most appropriate for single chain type ester group. But, as surface tension re-
flects only the adsorbed monolayer at equilibrium with the solution, surface tension
measurements would not always be enough to evaluate the efficiency of lowering
the surface free energy when arranged on the solid surface. Therefore, the π –A
measurements on single chain type acid monolayers on water were also carried out
using 0.01 M aq-HCl subphase to avoid dissolving of acids as much as possible,
and the effect of the number of –CH2 – on packing of hydrophobic groups at the
air/water interface was studied. Figure 2 shows the π –A isotherms on water at two
different temperatures.
Figure 2. Effect of temperature on π –A curves of semifluorinated alkanoic acids 1. Subphase: 0.01 M

HCl.
Figure 2a (left) and b (right) show the results measured at 15◦ C and 5◦ C, respec-
tively. In the isotherms at 15◦ C, except for C8 F17 (CH2 )10 –COOH 1e, only a single
molecular occupied area was observed (approximately 0.5 nm2 ), which is in accor-
dance with the value obtained from surface tension measurements. On the contrary,
C8 F17 (CH2 )10 –COOH afforded two occupied areas of 0.46 nm2 and 0.33 nm2 .
Moreover, in the measurements at 5◦ C, similar two occupied areas were observed
for all acids, which means a two-stage monolayer formation would proceed even
for the acid having smaller n at lower temperature.
As the fluorinated segments are very rigid and have a larger cross-sectional area
than that of hydrocarbon segments, this two-stage monolayer formation behavior
can be explained as follows: At first, monolayer formation would occur by packing
of only fluoroalkyl segments, but the interaction between hydrocarbon segments
would not be strong and thus only a single molecular occupied area was observed.
But, with increasing the length of hydrocarbon segments or lowering the temper-
ature, the interaction between hydrocarbon segments would become sufficiently
significant, and the second packing of hydrocarbon segments would occur to afford
the second molecular occupied area. This explanation may be in good agreement
with Shafin and Zisman’s model [18]. When the measurements were done at 25◦ C
(data not shown), similar behavior was observed. Therefore, we fixed the number
of –CH2 – to 10 in the following.
3.4. Double Semifluorinated Alkanoic Acid (Double Chain Type)
As Xiang et al. [12] proposed, multiply attached semifluorinated groups would
cover a large area per attachment site and protect the surface, and form stable
hydrophobic surfaces. We also entered in the molecular design of multiply semi-

fluorinated polymers. As a first step, the double chain type acid 2 (Scheme 2) was
synthesized to introduce multiple semifluorinated groups on acrylate monomer.
By the way, we have also been studying the synthesis and surfactant properties
of gemini surfactants, in which two identical hydrophobic groups are connected
by a spacer group –(CH2 )m –. As the outcome, it has been clarified that the sur-
factant properties of gemini surfactants depend highly on the spacer length, in
other words on the packing behavior of the two hydrophobic groups. Using a
molecular model, we searched the most suitable spacer length for the introduc-
tion of two semifluorinated segments as ester group. As a result, to align two
semifluorinated segments in parallel, the number m of the spacer group –(CH2 )m –
should be more than 1, namely two semifluorinated segments are end-capped on
trimethylene chain. It is quite interesting that our designed double chain type acid
is the same as the semifluorinated monodendron prepared by Xiang et al. [12]. We
also prepared the double chain type acid 2 using C8 F17 –(CH2 )10 –COOH 1e and
2,2-bis(hydroxymethyl)propanoic acid as shown in Scheme 2. To examine the va-
lidity of our molecular design, a double hydrocarbon type acid was also prepared
using stearic acid.
3.5. Surface Pressure–Area (π –A) Measurement on Double Chain Type Acid
As 2,2-bis(octadecanoyloxymethyl)propanoic acid 2 (Scheme 2) was insoluble in
an alkaline solution unlike single chain type acids, π –A isotherms of double chain
type acids were measured to study the effect of double chain structure on the surface
properties. Results are shown in Fig. 3.
Figure 3. Effect of temperature on π –A curves of double chain acids. (a) Double semifluorinated
type 2. (b) Double hydrocarbon type.
Figure 4. Comparison of π –A curves of single and double chain acids at 25◦ C. (a) Palmitic acid,
(b) single semifluorinated acid 1e, (c) double hydrocarbon acid, and (d) double semifluorinated acid 2.
Values shown with arrows mean the molecular occupied areas.
The surface pressure increased sharply at molecular area around 1 nm2 as ex-
pected (lift-up area) and the formation of a solid monolayer film evidently occurred
from the beginning, and the final surface pressure reached up to 60 mN/m, which
means the monolayers of double semifluorinated type acids are quite robust. Similar
solid film formation was also observed for the double hydrocarbon type. However,
when comparing these two acids, especially with respect to the effect of measure-
ment temperature, double semifluorinated type 2 was not influenced by varying
the measurement temperature, but hydrocarbon type was influenced and the lift-
up area became larger with increasing temperature, which probably means that the
interaction between two hydrocarbon chains would decrease due to the molecular
movement with increasing temperature. In the case of 2, it may be explained that
fluoroalkyl segment is quite rigid and –(CH2 )10 – segment may be long enough to
maintain strong interaction at elevated temperature.
Figure 4 summarizes the results of π –A isotherms at 25◦ C for palmitic acid
(hydrocarbon single chain acid), C8 F17 –(CH2 )10 –COOH (semifluorinated single
chain acid), double hydrocarbon type acid, and semifluorinated double chain type
acid. Using these results, we calculated each value of the molecular occupied area
and compared the molecular area occupied by double chain acids with single chain
ones. Results are summarized in Fig. 5.
In case of the hydrocarbon series, double hydrocarbon type has the mole-
cular occupied area of 0.55 nm2 and it is larger than twice of the value of
palmitic acid (0.21 nm2 ). As a reference, the result of the tartaric gemini
[bis(palmitoyloxy)tartaric acid], which has also two long hydrocarbon chains
Figure 5. Comparison of molecular occupied areas. (a) Palmitic acid, (b) bis(palmitoyloxy)tartaric
acid (gemini), (c) double hydrocarbon acid, (d) single semifluorinated acid, and (e) double semifluo-
rinated acid.
(C15 H31 –), is also given. The molecular occupied area of this tartaric gemini
(0.44 nm2 ) was almost twice of palmitic acid. This indicates that alkyl chains in
tartaric gemini would be tightly packed. However, the value of the double hydro-
carbon chain type was larger than that of tartaric gemini. Thus, it can be reasonably
considered that the two alkyl groups on both ends of C–C–C chain would be some-
what separated even in the solid monolayer film.
On the other hand, in the semifluorinated series, single chain type acid 1e and
double chain type acid 2 showed occupied areas of 0.46 nm2 and 0.91 nm2 , re-
spectively. The latter area is almost twice as much as that of the former. It is quite
consistent with the close packing of adjacent fluoroalkyl segments in a solid mono-
layer film. Therefore, to arrange two semifluorinated segments close and parallel,
the smallest methylene chain number should be 3 as predicted from molecular
model.
3.6. Acrylate Polymer with Double Semifluorinated Side Groups
To extend this strategy, we further devised a polymer having two semifluorinated

groups per monomer unit as a candidate for developing novel water- and oil-
repellent agents. As shown in Scheme 3, the double chain type acid was converted
to its acid chloride as usual and reacted with 2-hydroxyethyl methacrylate (HEMA).
This monomer 3 was polymerized using AIBN as initiator in MEK under nitrogen
atmosphere for 8 h to afford homopolymer 4 having double semifluorinated side
groups almost quantitatively, whose molecular weight was ca. 12000 (polymer-
ization degree = 8.5, Mw /Mn = 2.0 from GPC analysis). Similarly, as shown in
Scheme 4, a polymer having single semifluorinated side group 5 was also prepared
from 1e and HEMA.
Unfortunately, double semifluorinated polymer 4 could not afford any cast-film,
while the corresponding single semifluorinated polymer 6 could. At the present, it
is considered that the double semifluorinated polymer 4 would have considerably
high glass transition temperature owing to the formation of LC structure of semi-
fluorinated side groups.
3.7. Effect of Double Semifluorinated Side Groups
Very often, small amounts of fluorinated polymer can be mixed as additives with
another polymer such as polystyrene (PS), PMMA, etc., and the polymer surface
properties are dramatically altered. We investigated the effect of double semifluo-
rinated side groups on the surface properties of polymer by preparing cast films of
PMMA which contained single and double semifluorinated polymers as additives.
First, PMMA cast films were prepared with addition of 1.0 wt% of semifluori-
nated polymers 4 and 6, and contact angle measurements were carried out using
both decane and water. Results were summarized in Table 1.
As expected, though the original PMMA surface was not oil-repellent, but with
adding both single-semifluorinated polymer 6 and double-semifluorinated polymer
4 to PMMA (1.0 wt%), the surfaces were found to be modified to highly water-
(θW > 110◦ ) and oil-repellent (θD > 60◦ ), irrespective of the number of semifluo-
rinated side groups. Recently, Grampel et al. [20] reported the advancing contact
angles of water (θW ) and hexadecane (θHD ) on PTFE (θW = 111.6◦ , θHD = 43.5◦ )
and C8 F17 CH2 CH2 SH SAMs on gold (θW = 117.2◦ , θHD = 75.5◦ ), which are
comparable to those of PMMA cast film with added semifluorinated polymer at
1.0 wt%. We also calculated the surface free energies of modified PMMA surface
using these angles, and found that the dispersion component γ d and the polar com-
ponent γ p were 16 and 1.5 mJ/m2 , respectively, which are also well comparable to
those of PTFE and C8 F17 CH2 CH2 SH SAMs on gold. Though we measured only
static contact angles, this result shows that the surface of PMMA cast film would
be fully covered by fluoroalkyl segments, but no difference in the adsorbing power
Table 1.
Results of contact angles of water and decane on PMMA cast films
Additive Contact angle (degree)

Polymer wt(%)/PMMA Decane Water
– – 0 70.6 (1.4)
Double chain type, 4 1.0 67.3 (2.1) 111.1 (2.0)
Single chain type, 6 1.0 64.1 (1.9) 113.0 (2.6)
Double chain type, 4 0.01 50.4 (2.7) 104.3 (2.0)
Single chain type, 6 0.01 7.0 (1.2) 73.7 (1.8)
Values in parentheses are standard deviations of 5 samples.
between single- and double-semifluorinated polymers could be determined at an

addition level of 1.0 wt% to PMMA.
Next, on assuming that double semifluorinated side groups would interact inter-
and intramolecularly to arrange a closer packing than single side groups, we also
measured contact angles of decane and water on PMMA cast films with lower
added amount of semifluorinated polymers 4 and 6. The minimum amount can be
estimated using the molecular occupied area of double semifluorinated groups as
follows: In the cast film, in order to cover both upper and bottom surfaces with
semifluorinated groups, twice the area of Teflon petri dish was divided by the mole-
cular occupied areas (0.46 nm2 for single side groups and 0.91 nm2 for double side
groups) to obtain the minimum numbers of semifluorinated groups. Consequently,
as a stock solution of PMMA (0.5 g/10 ml acetone) was used, the minimum addi-
tive amount was estimated as 0.01 weight % of PMMA for both semifluorinated
polymers.
As also shown in Table 1, the PMMA surface was modified as expected to be
water- and oil-repellent with the addition of double semifluorinated polymer 4 even
only at 0.01 weight %, while, in the case of single semifluorinated polymer 6, the
surface was nearly the same as the original PMMA. This result would reasonably
support our assumption of monolayer coating of double semifluorinated polymer,
which might also be due to the formation of liquid-crystalline structure as reported
by Xiang et al. [12], but details are uncertain.
3.8. Efficiency of Surface Monolayer Coating
Though PMMA surface with the addition of double semifluorinated polymer 4 at
0.01 wt% was undoubtedly modified to be highly water- and oil-repellent, contact
angles of decane and water were smaller than those on PMMA surface with the
addition at 1.0 wt%. To clarify the modification mechanism, the degree of surface
coverage by double semifluorinated polymer was evaluated using the Cassie equa-
tion (4) [21] assuming that the modified PMMA surface consisted of 2 kinds of
surfaces, double-semifluorinated polymer monolayer and unmodified PMMA sur-
face, and the modified surface would be nearly flat and smooth but heterogeneous
as shown in Fig. 6.
cos θ = A1 cos θ1 + A2 cos θ2 , (4)
where θ is the apparent (measured) contact angle, θ1 the true contact angle of
surface-1, θ2 the true contact angle of surface-2, A1 and A2 are the area ratios
of surface-1 and surface-2, respectively, and A1 + A2 = 1.
In calculation of A1 and A2 , with respect to the true contact angles θ1 and θ2 ,
we also assumed that PMMA surface would be completely covered with semifluo-
rinated groups at the additive amount of 1 wt% single- and double-semifluorinated
polymers, and θ1 would be very close to the contact angle obtained on this modified
PMMA surface.
As listed in Table 2, we obtained the values of surface coverage by semifluori-
nated groups, A1 = 0.6 for decane and A1 = 0.84 for water. On the contrary, as
Figure 6. Schematic illustration of PMMA surface with the addition of 0.01 wt% double semifluori-
nated polymer 4. Rf: double-semifluorinated polymer monolayer.
Table 2.
Results of surface coverage of surface treated with fluo-
ropolymers 4 and 6 at an addition level of 0.01 wt% to PMMA
Polymer Decane Water

A1 A2 A1 A2
Double chain type, 4 0.60 0.40 0.84 0.16

Single chain type, 6 0.01 0.99 0.07 0.93
suspected, the PMMA surface with the addition of single semifluorinated polymer
6 at 0.01 wt% was nearly the same as the original PMMA probably because 6 might
have a better compatibility with PMMA than 4 and no phase separation occurred to
form monolayer film.
The observed difference in the two coverage values for decane and water can
be explained as shown in Fig. 7. According to Genzer et al. [22], semifluorinated
groups are tilted to the surface normal, and in the case of polystyrene (PS)-b-
polyisoprene (PI) copolymer with C8 F17 –(CH2 )9 –CO– group attached to the PI
block, average tilt angle was ∼45◦ . In our case, double semifluorinated side groups
would be arranged parallel to each other and tilted by about 40–45◦ to the sur-
face normal (Fig. 7). On contact with decane, only fluoroalkyl segment (Rf) would
contribute to oil repellency and show the surface coverage of 0.6. On contact with
water, however, as hydrocarbon segment (–(CH2 )10 –) can also contribute to wa-
ter repellency, double semifluorinated side groups would totally act as hydrophobic
groups and the surface coverage became larger (0.84).
Figure 7. Schematic illustration of configuration of double semifluorinated side groups on the PMMA
surface at an addition level of 0.01 wt%. Rf: fluoroalkyl segment, wavy line: –(CH2 )n – segment.
4. Conclusion
To develop an effective and powerful water- and oil-repellent fluoropolymer, we
have designed a new concept of double semifluorinated segments as side groups
on polymers. Referring to the structural feature of gemini surfactants, where two
hydrophobic tail groups can easily come close, nearly in contact, and a rigid mono-
layer can be formed, double semifluorinated alkanoic acid (double chain type) was
prepared from 2,2-bis(hydroxymethyl)-propanoic acid and C8 F17 (CH2 )10 COOH.
As expected, a rigid monolayer formation of double semifluorinated acid was con-
firmed by surface pressure–area (π –A) measurement. The molecular occupied area
of double semifluorinated acid, 0.91 nm2 , was twice as much as that of single acid,
0.46 nm2 . It is quite consistent with the close packing of adjacent semifluorinated
segments in a solid monolayer film.
Next, the methacrylate monomer having double semifluorinated groups was syn-
thesized using double chain type acid and HEMA, and polymerized. Based on
contact angle measurements of decane and water on the cast film of PMMA contain-
ing this double semifluorinated polymer, it was made clear that the PMMA surface
was modified to be highly water- and oil-repellent even at the minimum addition
amount of double semifluorinated polymer (0.01 wt%/PMMA) determined from
the molecular occupied area of double semifluorinated acid. Effectiveness of sur-
face modification was evaluated on the basis of Cassie equation, and the alignment
of double semifluorinated groups on the surface was explained in terms of tilt angle
of semifluorinated side groups in polymer monolayer film.
In conclusion, our strategy for the introduction of double semifluorinated seg-
ments as side groups enabled the monolayer coating to attain effective and powerful
water- and oil-repellency by fluorochemicals.
References
1. W. Barthlott and C. Neinhuis, Planta 202, 1–8 (1997).
2. P. Ball, Nature 400, 507–508 (1999).

3. T. Onda, S. Shibuichi, N. Satoh and K. Tsujii, Langmuir 12, 2125–2127 (1996).
4. J. P. Youngblood and T. J. McCarthy, Macromolecules 32, 6800–6806 (1999).
5. Y. Wu, H. Sugimura, Y. Inoue and O. Takai, Chem. Vap. Deposition 8, 47–50 (2002).
6. T. Tsujii, T. Yamamoto, T. Onda and S. Shibuichi, Angew. Chem. Int. Ed. Engl. 36, 1011–1012
(1997).
7. T. Kawase, T. Fujii, M. Minagawa, H. Sawada, T. Matsumoto and M. Nakayama, J . Adhesion Sci.
Technol. 10, 1031–1046 (1996).
8. T. Kawase, M. Yamane, T. Fujii, M. Minagawa and H. Sawada, J. Adhesion Sci. Technol. 11,
1381–1397 (1997).
9. H. Sawada and T. Kawase, in: Recent Research Developments in Macromolecules Research,
Vol. 4, Part II, S. G. Pandalai (Ed.), pp. 229–246. Research Signpost, Trivandrum, India (1999).
10. R. Zana and J. Xia, in: Gemini Surfactants: Synthesis, Interfacial and Solution-Phase Behavior,
and Applications, R. Zana and J. Xia (Eds), Surfactant Science Series, Vol. 117. Marcel Dekker,
New York (2004).
11. N. O. Brace, J. Org. Chem. 27, 4491–4498 (1962).
12. M. Xiang, X. Li, C. K. Ober, K. Char, J. Genzer, E. Sivaniah, E. J. Kramer and D. A. Fisher,
Macromolecules 33, 6106–6119 (2000).
14. D. K. Owens and R. C. Wendt, J. Appl. Polym. Sci. 13, 1741–1747 (1969).
15. C. D. Bain and G. M. Whitesides, Angew. Chem. 101, 522–528 (1989).
16. M. K. Chaudhury and G. M. Whitesides, Langmuir 7, 1013–1025 (1991).
17. J. Wang, G. Mao, C. K. Ober and E. J. Kramer, Macromolecules 30, 1906–1914 (1997).
18. E. G. Shafrin and W. A. Zisman, J. Phys. Chem. 66, 740–748 (1962).
19. J. Genzer, E. Sivaniah, E. J. Kramer, J. Wang, H. Korner, M. Xiang, K. Char, C. K. Ober, R. A.
Bubeck, D. A. Fischer, M. Graupe, R. Colorado, Jr., O. E. Shmakova and T. R. Lee, Macromole-
cules 33, 6068–6077 (2000).
20. R. D. Grampel, W. Ming, A. Gildenpfennig, J. Laven, H. H. Brongersma, G. de With and R. Linde,
Langmuir 20, 145–149 (2004).
21. A. B. D. Cassie, Discuss. Farady Soc. 3, 11–21 (1949).
22. J. Genzer, E. Sivaniah, E. J. Kramer, J. Wang, H. Korner, M. Xiang, K. Char, C. K. Ober, B. M.
DeKoven, R. A. Bubeck, M. K. Chaudhury, S. Sambasivan and D. A. Fischer, Macromolecules
33, 1882–1887 (2000).
Toward Superlyophobic Surfaces
Weihua Ming a,b,∗ , Boxun Leng b,c , Rik Hoefnagels b , Di Wu b , Gijsbertus de With b and
Zhengzhong Shao c
a
Nanostructured Polymers Research Center, Materials Science Program, University of New
Hampshire, Durham, NH 03824, USA
b
Laboratory of Materials and Interface Chemistry, Eindhoven University of Technology,
P.O. Box 513, 5600 MB Eindhoven, The Netherlands
c
Key Laboratory of Molecular Engineering of Polymers of Ministry of Education,
Advanced Materials Laboratory, Department of Macromolecular Science, Fudan University,
Shanghai 200433, China
Abstract
We report on our recent endeavor in developing superlyophobic surfaces, on which both water and hexade-
cane demonstrate contact angles greater than 150◦ and, even more importantly, roll-off angles for 20-µl
droplets of less than 10◦ . Our superlyophobic surfaces are based on a multilength-scale structure, either
from a raspberry-like topography or from particle-containing woven fabrics. The multilength-scale rough-
ness plays a major role in rendering the surfaces superoleophobic (the liquid droplets are in the Cassie
wetting state), especially in terms of obtaining low hexadecane roll-off angles. In addition, surface perfluo-
rination has proven to be essential in achieving superoleophobicity.
Keywords
Superlyophobicity, superhydrophobicity, superoleophobicity, multilength-scale roughness, surface perfluo-
rination
1. Introduction
There has been substantial interest in developing superhydrophobic surfaces in both

academia and industry due to their potentially wide applications, including self-
cleaning property with respect to water. In Nature, there are many elegant examples
of perfectly designed surface wettability. For instance, the leaves of the sacred lotus
flower, known as “rising out of muddy water, untainted”, demonstrate self-cleaning
property, which is due to a combination of a proper surface chemistry and a pe-
culiar topographic feature based on dual-scale roughness: the coarse-scale rough
structure is about 10–20 µm in size whereas the finer structure on top of the coarse
* To whom correspondence should be addressed. E-mail: W.Ming@unh.edu

192 W. Ming et al.
structure is in the range of a few hundred nanometers [1–3]. There are many other
interesting superhydrophobic surfaces, such as legs of the water strider [4] and the
gecko’s feet [5]. Surface roughness at a dual- or multi-length scale has shown to be
very effective in generating this surprising nonwetting behavior, especially for ob-
taining low water roll-off angles, which has inspired many biomimetic approaches
to obtain artificial superhydrophobic surfaces [6–14]. On the other hand, the ma-
jority of the reported superhydrophobic surfaces are not super oil-repellent; it is
highly desirable for superhydrophobic surfaces to be also oil repellent to maintain
their superhydrophobicity. For instance, in an industrial or household environment
a superhydrophobic surface with poor oil repellency can be easily contaminated
by oily substances, which, in turn, will compromise its surface superhydropho-
bicity. Therefore, superlyophobic (or superhygrophobic [15]) surfaces combining
superhydrophobic and superoleophobic properties are desirable for practical appli-
cations [16].
Despite extensive investigations on superhydrophobic surfaces, studies on super-
oleophobic surfaces with high repellency against liquids with a low surface tension
(<35 mN/m) have been rather limited so far. Oil repellency has been examined
on various perfluorinated, superhydrophobic surfaces [17–27], with reported static
contact angles (CAs) for benzene, hexadecane or rapeseed oil in the range of 135–
160◦ . No receding contact angles or roll-off angles for oil droplets were reported
on most of the studied surfaces [17–26], so it is difficult to judge whether these
surfaces are truly superoleophobic. For instance, despite a high static CA (160◦ )
for rapeseed oil on a superhydrophobic surface made of surface-fluorinated carbon
nanotubes [24], the rapeseed oil droplet remained pinned to the surface (indicating
very low receding CA) when the sample was tilted, underlying the importance of
achieving high receding CAs (thus, low contact angle hysteresis and low roll-off
angles). Only those surfaces with high CAs (>150◦ ) and low roll-off angles for oil
droplets can be regarded as truly superoleophobic surfaces.
Very recently, truly superoleophobic surfaces have been achieved on the basis of
microhoodoo [16, 28] and nanonail [29] structures, as exemplified by low contact
angle hysteresis for probe liquids of low surface tension (<30 mN/m) like octane
and ethanol. In both cases, the key to obtaining true superoleophobicity is the re-
entrant surface structure, thus ensuring the entrapment of air beneath the top solid
surface and preventing the transition from the Cassie–Baxter state to the Wenzel
state [16, 18, 28, 29]. However, the fabrication of these superoleophobic surfaces
involves lithography and etching steps, which may limit their practical applications.
In this contribution, we will summarize our recent efforts in developing su-
perlyophobic (both superhydrophobic and superoleophobic) surfaces, including
raspberry-like surfaces and particle-containing textiles, both containing dual- or
multilength-scale surface structures.
Toward Superlyophobic Surfaces 193
2. Lyophobicity on Raspberry-Like Surfaces

2.1. Superhydrophobic Surfaces from Raspberry-Like Particles
Inspired by the lotus leaf structure, we prepared epoxy-based polydimethylsilox-
ane (PDMS)-surface-modified, superhydrophobic films with dual-scale hierarchi-
cal structure originating from well-defined raspberry-like particles, as shown in
Scheme 1 [6]. First, a conventional cross-linked film based on epoxy-amine system
was prepared with unreacted epoxy groups available for further surface grafting.
Second, amino-functionalized raspberry-like silica particles (core particles: 700 nm
in diameter; shell particles: 70 nm) were then chemically deposited onto the epoxy
films to generate a dual-scale surface roughness. Finally, a layer of monoepoxy-
end-capped PDMS was grafted onto the raspberry-like particles to render the film
surface hydrophobic.
The advancing water contact angle (CA) on a smooth epoxy-based film (surface
modified with PDMS) was 107 ± 2◦ , with a CA hysteresis (CAH) of about 40◦ .
When only large silica particles (700 nm) or only small particles (70 nm) were
deposited (surface also modified with PDMS) on the epoxy-amine film, the water
advancing CA increased to 150◦ , but at the same time the CAH also increased sig-
nificantly to ∼60◦ . In a sharp contrast, on the superhydrophobic films containing
raspberry-like particles (Fig. 1b), the advancing CA for water was about 165◦ and
the roll-off angle of a 10-µl water droplet was <3◦ (Fig. 2) [6]. The superhydropho-
bic surface demonstrated self-cleaning property with respect to water, similar to the
lotus leaf.
We used atomic force microscopy (AFM) to examine the surface topography
of the superhydrophobic films. As shown in Fig. 1b, the topographic image of the
superhydrophobic film clearly shows a dual-scale structure: the micrometer-level
structure can be ascribed to the large core particles (700 nm) of the raspberry-
like particles, while on each of these microparticles there is a finer structure at
a sub-micrometer level due to the 70-nm shell nanoparticles [6]. Obviously, the
topographic feature of the raspberry-like particles was completely preserved in the
superhydrophobic film. The dual-scale surface structure mimics that of a lotus leaf.
Scheme 1. Preparation of superhydrophobic films based on raspberry-like particles [6].

194 W. Ming et al.
(a) (b) (c)
Figure 1. (a) A raspberry fruit with a naturally occurring dual-scale structure and (b) the surface
topography (AFM) of our nature-inspired, superhydrophobic films from raspberry-like particles [6],
which to some extent resembles (c) the surface morphology (SEM) of the lotus leaf.
Figure 2. Roll-off of a 10-µl water droplet on our superhydrophobic film as the film was tilted from
the horizontal position to 1.8◦ .
(a) (b)
Scheme 2. Schematic illustration of the air/water interfacial area on a polymer surface covered by particles:
(a) no particle embedment; and (b) particle partially embedded into the polymer matrix.
2.2. Mechanically Robust Raspberry-Like Surfaces from Layer-by-Layer Particle

Deposition
For the films we discussed in Section 2.1, an obvious drawback is their poor me-
chanical robustness, due to the fact that there would be only point contacts between
the raspberry-like particles and the polymer films (despite the presence of covalent
bonding between them). To improve mechanical robustness of the superhydropho-
bic films, we used the layer-by-layer particle deposition approach and obtained
films with particles partially embedded into the films (Scheme 2b), in addition to
the covalent bonding between the particles and the polymer matrix. The partially
embedded particles would demonstrate similar effect on the surface wettability on
the basis of the following rationale: there is no difference in terms of air/water con-
tact area fraction between the case (a) and the case (b) in Scheme 2 as long as
water does not have direct contact with the polymer surface. In other words, what
is important to the liquid repellency of a structured surface is not the volume of the
entrapped air, but the air/liquid interfacial area. (The entrapped air volume, on the
other hand, may have impact on the stability of the superhydrophobic state, which
will not be discussed here.)
The layer-by-layer particle deposition approach we have adopted is as follows.
First, a layer of a partially pre-cured epoxy was prepared (the epoxy we used here
was Epikote 1004 from Shell Chemical, which is a solid at room temperature).
Then, a single layer of epoxy-modified silica microparticles (∼800 nm in diameter)
was spin-coated on top of the epoxy film. After that, the film was fully cured at an el-
evated temperature, leading to completely cross-linked epoxy films with micropar-
ticles partially embedded. Next, amine-modified silica nanoparticles (∼50 nm)
were deposited onto the monolayer of microparticles. The particle-covered film
was then treated with a SiCl4 toluene solution, which resulted in cross-linking
between silica micro- and nanoparticles and also among nanoparticles to ensure
mechanical robustness of the films. In the end, the films were chemically modified
with 1H ,1H ,2H ,2H -perfluorodecyltrichlorosilane (RfSi) in toluene (4 v%) at 0◦ C
for 3 h.
As shown in Fig. 3, when the epoxy film was fully cured before the 800-nm
particle deposition, silica particles were completely on top of the polymer coating
(Fig. 3d), and there was no particle embedment. On the other hand, if the epoxy
was not pre-cured at all before the particle deposition, silica particles became com-
pletely embedded into the polymer matrix (image not shown), and the film surface
was quite smooth and would not have a significant effect on the surface wettability.
Neither of these two films was suitable to develop robust superhydrophobic sur-
faces. When the epoxy conversion during the pre-curing step was 44%, despite that
the particles protruded out (Fig. 3a), the individual particles appeared to be com-
pletely covered by the epoxy film, and the surface roughness may not be sufficient
to lead to superhydrophobicity. In contract, at epoxy conversions of 57% and 65%
during the pre-curing step, about 40% (Fig. 3b) and 20% (Fig. 3c), respectively, of
each individual particle was embedded into the polymer matrix. The morphology
of these two films resembled the scenario we depicted in Scheme 2b.
We used the pull-off test [30]1 to examine the mechanical stability of the films
with particles 40% and 20% partially embedded in the polymeric matrix, and found
1 In brief, a sample was first sputter-coated with a 10-nm thick gold layer and then glued to a pull-off
stud (stainless steel, d = 8 mm) by an epoxy glue (cured for 24 h at room temperature), followed by cutting
around the stud. The as-prepared sample was then clamped in a TesT 810 tensile instrument and connected to
the load cell via a long cable to make sure the force application angle was 90◦ at any moment. Subsequently,
the clamped sample was moved downwards at a constant speed until fracture occurred. The tests were
performed in air at room temperature using a tensile machine crosshead speed of 1 mm/min.
196 W. Ming et al.
(a) (b)
(c) (d)
Figure 3. SEM images of epoxy films with silica microparticles embedded into the films to various
degrees. The epoxy films were pre-cured at different epoxy conversions: (a) 44%, (b) 57%, (c) 65%,
and (d) full epoxy conversion, before the particle deposition, which was followed by complete curing
at 100◦ C.
that with the 40% partial embedment none of particles could be pulled out, indicat-
ing the film was mechanically robust. In comparison, about 2% of the particles were
indeed pulled out of the film with the 20% partial embedment. On the film with par-
ticles 40% partially embedded, we deposited a second layer of silica nanoparticles
(50 nm), followed by SiCl4 and RfSi treatments. The SEM images of the film in
Fig. 4 demonstrate evidently a dual-scale surface roughness for the film, which is
similar to the morphology obtained directly from raspberry-like particles and mim-
ics the lotus leaf structure. The mechanical robustness of this superhydrophobic
surface was also confirmed by the pull-off test. The water advancing CA on the
film was 156 ± 0.5◦ with a CA hysteresis of about 4◦ ; both values were close to the
CA data for the dual-scale surfaces described in Section 2.1.
2.3. Oleophobicity on Raspberry-Like Surfaces
To examine the oleophobicity of the dual-scale structured surfaces, we used mix-
tures of water and ethanol as probe liquids. Water and ethanol are completely
miscible at all fractions, allowing us to tune surface tension of the mixture from
22.1 mN/m for pure ethanol, at 20◦ C, to 72.8 mN/m for pure water, as shown in
Fig. 5.
(a) (b)
Figure 4. SEM images of a superhydrophobic film made from the layer-by-layer particle deposition
approach: (a) side view and (b) top view. The insert shows the photo of a 10-µl water droplet on the
film.
Figure 5. Surface tension of water/ethanol mixtures as a function of ethanol volume fraction at 20◦ C.
We first examined the contact angles of the water/ethanol mixtures on PDMS-

based superhydrophobic surfaces. As shown in Fig. 6, although the advancing CA
was greater than 140◦ for the interrogating liquids with a surface tension larger than
35 mN/m, a small CAH (<5◦ ) was observed only for those interrogating liquids
with a surface tension larger than 60 mN/m. Most oleophilic liquids have a surface
tension less than 40 mN/m, thus the PDMS-based superhydrophobic surfaces are
not really oleophobic, particularly as judged from the standpoint of contact angle
hysteresis.
Instead of the PDMS-modified surface, we then examined perfluoroalkyl-
modified superhydrophobic surfaces. We used RfSi to modify the raspberry-like
surfaces that contained surface amino groups. The advancing CA for pure water on
this film was 168◦ with a CAH as low as 2◦ (Fig. 7). For the water/ethanol mixtures
with surface tension greater than 35 mN/m, the CA was above 160◦ and the CAH
was less than 10◦ , as shown in Fig. 7, indicating that the surface-perfluorinated
superhydrophobic surfaces may also be highly oleophobic. The film was then sub-
198 W. Ming et al.
Figure 6. Contact angles and CAH values of water/ethanol mixtures on PDMS-based superhydropho-
bic surfaces. The lines serve as a guide to the eye only.
Figure 7. Contact angles and CAH values of water/ethanol mixtures on perfluoroalkyl-based super-
hydrophobic surfaces. The lines serve as a guide to the eye only.
jected to interrogation by hexadecane (a hydrophobic liquid with a surface tension

of 27.5 mN/m at 20◦ C) and sunflower oil (surface tension: 33.0 mN/m at 20◦ C).
The contact angle for a 5-µl sunflower oil droplet was 132◦ (Fig. 8b). The static
hexadecane CA was 125◦ for a 5-µl droplet (Fig. 8a), significantly greater than that
(typically 80◦ ) on smooth perfluoroalkyl-modified surfaces (for instance [31, 32]).
However, the roll-off angles for both hexadecane and sunflower oil on the superhy-
drophobic surface appeared to be very high (>60◦ ). Therefore, we can only deem
the surface oleophobic, not yet superoleophobic.
(a) (b)
Figure 8. Images of 5-µl droplets of (a) hexadecane and (b) sunflower oil on the surface-fluorinated
superhydrophobic raspberry-like surface.
3. Superlyophobic Textiles
The intrinsic roughness of woven textiles can be considered as ‘local surface cur-
vature’ (analogous to a re-entrant structure) [16, 28, 29]. Combining this local
curvature with silica particles, it is feasible to generate dual- and multilength-scale
surface structures, possibly leading to superlyophobic surfaces.
3.1. Superhydrophobic Textiles
We have prepared superhydrophobic cotton textiles by covalently introducing a
layer of silica particles onto the fiber surface and subsequent surface hydrophobi-
zation [33]. Our particle-based approach is schematically shown in Scheme 3:
silica microparticles were first covalently bonded to the cotton fibers via in situ
Stöber reaction [34] due to the abundant hydroxyl groups in cellulose; amino
groups were then introduced to the particle surface via the reaction with 3-
aminopropyltriethoxysilane (APS); and finally monoepoxy-functionalized PDMS
was used to react with the amino groups to hydrophobize the fiber surface. After
the modification, normally hydrophilic cotton was turned superhydrophobic, which
exhibited a static water CA of 155◦ for a 10-µl droplet. The roll-off angle of water
droplets depended on the droplet volume, ranging from 7◦ for a droplet of 50 µl to
15◦ for a 10-µl droplet. It should be noted that due to the stick-out of fibers from the
cotton fabric, the measurement of contact angles was often not straightforward be-
cause of the difficulty in determining the baseline of the water droplet. Also because
Scheme 3. Procedure to make a superhydrophobic cotton with PDMS modification.

200 W. Ming et al.
of the protruding fibers, it was difficult to obtain accurate values for advancing and
receding water contact angles, therefore only static CAs were reported. The rela-
tively high roll-off angles may be partially due to the protruding fibers in the cotton
fabric, which have some elasticity and thus can exert forces on the water droplet
[35].
We also extended this modification strategy to synthetic textiles such as polyester
fabrics. To render the polyester fabric surface reactive, a UV/ozone treatment was
performed for 30 min to generate surface hydroxyl groups on the polyester before
the incorporation of silica particles and PDMS. There was no visible macroscopic
change after the modification steps. However, the UV/ozone treatment should not
last too long; otherwise, too much oxidation and degradation would lead to yellow-
ing of the fabric. As shown in Fig. 9, the silica particles were uniformly distributed
along the polyester fiber surface, and a dual-scale structure was obtained after the
particle incorporation. The polyester fabric was also turned superhydrophobic: the
static and roll-off angles for 10-µl water droplets were 149 ± 3◦ and 12 ± 1◦ , re-
spectively. The polyester fabric showed water CA data similar to the cotton fabric,
but the roll-off angle was relatively smaller, probably due to the absence of the pro-
truding fibers in the cotton fabric (the different woven structures may also lead to
the different surface wettability).
3.2. Superoleophobic Textiles
It is known that PDMS is hydrophobic but not oleophobic. The PDMS-modified
fabric can indeed be completely wetted by hexadecane and sunflower oil, which
have surface tensions of 27.5 and 33.0 mN/m at 20◦ C, respectively. To render a tex-
tile sample oleophobic, the PDMS modification is not sufficient. Therefore, a per-
fluoralkylsilane (1H ,1H ,2H ,2H -perfluorodecyltrichlorosilane, RfSi) was used for
the surface modification to render the fabric oleophobic.
The modification procedure is illustrated in Scheme 4a: immediately after the co-
valent incorporation of microparticles onto the fiber surface, perfluorooctyl groups
were grafted onto the particle surface (also to the fiber surface via the reaction with
(a) (b)
Figure 9. SEM images of particle-covered polyester fabric: (a) low magnification and (b) high mag-
nification.
Scheme 4. Procedures to make superoleophobic cottons with surface perfluorination: (a) with a monolayer
of particles and (b) with a dual-layer of particles.
(a) (b)
Figure 10. SEM images for cotton fabrics containing (a) a single layer of silica microparticles and
(b) dual layers of silica micro-/nano-particles (800 nm/160 nm). The inserts show the images of 10-µl
(a) and 5-µl (b) hexadecane droplets on the two samples after being perfluorinated. The white line in
(a) is the droplet baseline from which the contact angle was determined.
the remaining hydroxyl groups in cellulose). A typical SEM image of the modified
cotton fabric is shown in Fig. 10a. On the perfluorooctyl-modified cotton fabric,
the static CA and the roll-off angle for a 15-µl sunflower oil droplet were 140 ± 2◦
and 24 ± 2◦ , respectively [33]. When hexadecane was used as the probe liquid, its
static CA was about 137◦ for a 10-µl droplet (insert in Fig. 10a), which is signif-
icantly greater than that on smooth perfluoroalkyl-modified surfaces. The roll-off
angle for the same droplet was 34 ± 5◦ . Obviously, the surface-fluorinated cotton
fabric was turned highly oleophobic. However, the sample cannot be deemed truly
superoleophobic, since the roll-off angle for hexadecane was still too large.
We further introduced a raspberry-like dual-scale structure onto the woven cotton
fibers, leading to a triple-size surface structure [36]. As shown in Scheme 4b, sil-
ica microparticles (diameter: ∼800 nm) were first in situ generated and covalently
202 W. Ming et al.
bonded to the cotton fibers. After treatment with APS and hydrochloric acid, the
surface charge was turned positive due to the protonation of amino groups. Nega-
tively charged silica nanoparticles (∼160 nm) were then electrostatically adsorbed
onto the fiber surface, leading to a triple-scale structure: raspberry-like particles
bonded to the woven fabric (Fig. 10b). The obtained rough structure was stabilized
by SiCl4 cross-linking, followed by surface modification with RfSi. It was observed
that the silica nanoparticles also covered part of the fiber surface not covered by mi-
croparticles. This may be considered as beneficial for the roughened surface, similar
to the lotus leaf on which nanostructured protrusions cover both micropapillae and
the lower part of the leaf [37].
The addition of nanoparticles did not change much the static CA (all above 150◦ )
for water droplets, compared to the sample with only microparticles. However, the
roll-off angles decreased substantially; for instance, for a 20-µl water droplet, the
roll-off angle decreased from 12 ± 4◦ for the sample with only microparticles to
5 ± 1◦ for the sample also containing nanoparticles. Our results clearly indicate
that the samples with additional silica nanoparticles have demonstrated better water
repellency than those with only microparticles [33, 35].
For C16 H34 droplets, the incorporation of silica nanoparticles to the cotton fab-
rics led to much higher static CA: 152 ± 2◦ for 5-µl droplets. The increase was also
apparent in the insert images in Fig. 10. More significantly, the roll-off angles de-
creased dramatically when nanoparticles were incorporated onto the fiber surface:
for 20-µl droplets, roll-off angles as low as 9◦ were observed. The roll-off angles
demonstrated strong dependency on the droplet volume: roll-off angles for 10-µl
and 5-µl droplets were about 15◦ and 30◦ , respectively, likely owing to the elastic
nature of the cotton fiber as previously discussed. With the high hexadecane and
water repellency, we can conclude that these samples are both superhydrophobic
and superoleophobic. We examined how much our modified cotton fabrics “hated”
both water and hexadecane: the sample was first submerged into these two liquids;
when the sample was lifted out of these two liquids, the liquid contact line retreated
very rapidly from the top edge of the modified cotton, and the samples were not
wetted by either liquid at all.
Two distinct models, the Wenzel [38] and Cassie–Baxter [39] models, have been
extensively used to explain the roughness effect on the contact angles of liquids
on a roughened surface. The Wenzel model describes the wetting regime in which
the liquid penetrates into the roughened surface (intimate contact between the liq-
uid and the solid surface always exists), which usually leads to high contact angle
hysteresis. In contrast, in the Cassie wetting regime there is air trapped between
the solid surface and liquid, which results in much smaller hysteresis. When a liq-
uid droplet sits on a liquid-repellent cotton textile surface, the wetting behavior can
be described by the Cassie–Baxter equation [39], cos θCB = fls cos θ0 − flv , where
θCB is the observed CA on a rough surface, θ0 is the intrinsic CA on the corre-
sponding smooth surface (θ0 is 120◦ for water and 80◦ for C16 H34 , respectively,
on a perfluoroalkyl surface), fls is the liquid/solid contact area divided by the pro-
jected area, and flv is the liquid/vapor contact area divided by the projected area.
This equation has been modified to account for the local surface roughness on the
wetted area, as follows [33, 40, 41]: cos θCB = rf f cos θ0 + f − 1, where f is the
fraction of the projected area of the solid surface wetted by water (flv = 1 − f )
and rf is the surface roughness factor of the wetted area (rf 1). By introduc-
ing a layer of microparticles onto the fiber surface, rf can indeed be increased
for a PDMS-modified cotton fiber, leading to superhydrophobic cotton [33]. With
the surface perfluoroalkyl modification, the cotton fabric was turned to be highly
oleophobic [33], as demonstrated by a C16 H34 CA of 137◦ , but not superoleopho-
bic. It has become obvious that by incorporating a second layer of nanoparticles
onto the microparticle-covered fiber, rf can be further increased to an even higher
level, which allows the wetting by C16 H34 to be in the Cassie regime, as clearly
manifested by the high static CA and the small roll-off angle for C16 H34 [36].
The creation of a triple-length-scale roughness (the woven fiber, microparticle, and
nanoparticle) has proven to be essential to achieving superoleophobicity.
4. Conclusions
In summary, we have successfully obtained superlyophobic surfaces on the basis of
multilength-scale structures, such as a dual-/triple-scale raspberry-like surface and
cotton textiles with a dual-scale nano/microparticle structure incorporated onto the
woven fabrics. Our superlyophobic surfaces were completely nonwettable by both
water and hexadecane, which showed both high contact angles and low roll-off
angles. Mechanically robust surfaces have been successfully obtained via a layer-
by-layer particle deposition approach. Superhydrophobic surfaces can be easily
achieved by using a dual-scale structure, in combination with the surface hydropho-
bization with PDMS. To achieve superoleophobicity, the presence of a multilength
(>2)-scale surface roughness has proven to be essential for this particle-based ap-
proach, especially in terms of low roll-off angles. Our particle-based approach is
not limited to the use of silica particles; our approach can be easily extended to
using functional polymeric microparticles and nanoparticles, which can be readily
synthesized by dispersion and emulsion polymerization, respectively, as building
blocks to prepare multilength-scale structured surfaces.
Acknowledgements
Part of the work was financially sponsored by DSM Research. We thank China
Scholarship Council for supporting BL’s stay at Eindhoven.
References
1. G. E. Fogg, Nature 154, 515 (1944).
3. S. Herminghaus, Europhys. Lett. 52, 165–170 (2000).
204 W. Ming et al.
4. X. Gao and L. Jiang, Nature 432, 36 (2004).

5. K. Autumn, Y. A. Liang, S. T. Hsieh, W. Zesch, W. P. Chan, T. W. Kenny, R. Fearing and R. J.
Full, Nature 405, 681–685 (2000).
6. W. Ming, D. Wu, R. van Benthem and G. de With, Nano Lett. 5, 2298–2301 (2005).
7. W. Han, D. Wu, W. Ming, J. W. Niemantsverdriet and P. C. Thuene, Langmuir 22, 7956–7959
(2006).
8. P. Roach, N. J. Shirtcliffe and M. I. Newton, Soft Matter 4, 224–240 (2008).
10. X. J. Feng and L. Jiang, Adv. Mater. 18, 3063–3078 (2006).
11. T. Sun, L. Feng, X. Gao and L. Jiang, Acc. Chem. Res. 38, 644–652 (2005).
12. N. J. Shirtcliffe, G. McHale, M. I. Newton, G. Chabrol and C. C. Perry, Adv. Mater. 16, 1929–1932
(2004).
13. L. Zhai, F. C. Cebeci, R. E. Cohen and M. F. Rubner, Nano Lett. 4, 1349–1353 (2004).
14. H. Y. Erbil, A. L. Demirel, Y. Avci and O. Mert, Science 299, 1377–1380 (2003).
16. A. Tuteja, W. Choi, M. L. Ma, J. M. Mabry, S. A. Mazzella, G. C. Rutledge, G. H. McKinley and
R. E. Cohen, Science 318, 1618–1622 (2007).
17. J. Zimmermann, M. Rabe, G. R. J. Artus and S. Seeger, Soft Matter 4, 450–452 (2008).
18. L. L. Cao, T. P. Price, M. Weiss and D. Gao, Langmuir 24, 1640–1643 (2008).
19. H. Yan, K. Kurogi and K. Tsujii, Colloids Surfaces A 292, 27–31 (2007).
20. Y. Tian, H. Q. Liu and Z. F. Deng, Chem. Mater. 18, 5820–5822 (2006).
21. M. Nicolas, F. Guittard and S. Geribaldi, Angew. Chem. Int. Edit. 45, 2251–2254 (2006).
22. A. Nakajima, M. Hoshino, J. H. Song, Y. Kameshima and K. Okada, Chem. Lett. 34, 908–909
(2005).
23. H. Yabu, M. Takebayashi, M. Tanaka and M. Shimomura, Langmuir 21, 3235–3237 (2005).
24. H. J. Li, X. B. Wang, Y. L. Song, Y. Q. Liu, Q. S. Li, L. Jiang and D. B. Zhu, Angew. Chem. Int.
Edit. 40, 1743–1746 (2001).
25. S. Shibuichi, T. Yamamoto, T. Onda and K. Tsujii, J. Colloid Interface Sci. 208, 287–294 (1998).
26. K. Tsujii, T. Yamamoto, T. Onda and S. Shibuichi, Angew. Chem. Int. Edit. Engl. 36, 1011–1012
(1997).
27. D. Wu, R. J. Vrancken, B. G. H. van Loenen, R. A. T. M. van Benthem, G. de With and W. Ming,
Polym. Mater. Sci. Eng. 97, 418–419 (2007).
28. A. Tuteja, W. Choi, G. H. McKinley, R. E. Cohen and M. F. Rubner, MRS Bull. 33, 752–758
(2008).
29. A. Ahuja, J. A. Taylor, V. Lifton, A. A. Sidorenko, T. R. Salamon, E. J. Lobaton, P. Kolodner and
T. N. Krupenkin, Langmuir 24, 9–14 (2008).
30. D. Wu, Nature-inspired superlyophobic surfaces, Ph.D. dissertation, Eindhoven University of
Technology (2007).
31. W. Ming, M. Tian, R. D. van de Grampel, F. Melis, X. Jia, J. Loos and R. van der Linde, Macro-
molecules 35, 6920–6929 (2002).
32. L. van Ravenstein, W. Ming, R. D. van de Grampel, R. van der Linde, G. de With, T. Loontjens,
P. C. Thuene and J. W. Niemantsverdriet, Macromolecules 37, 408–413 (2004).
33. H. F. Hoefnagels, D. Wu, G. de With and W. Ming, Langmuir 23, 13158–13163 (2007).
34. W. Stöber, A. Fink and E. Bohn, J. Colloid Interface Sci. 26, 62–69 (1968).
35. T. Wang, X. G. Hu and S. J. Dong, Chem. Commun., 1849–1851 (2007).
36. B. Leng, Z. Shao, G. de With and W. Ming, Langmuir 25, 2456–2460 (2009).
37. L. Feng, S. H. Li, Y. S. Li, H. J. Li, L. J. Zhang, J. Zhai, Y. L. Song, B. Q. Liu, L. Jiang and D. B.
Zhu, Adv. Mater. 14, 1857–1860 (2002).
41. N. J. Shirtcliffe, G. McHale, M. I. Newton and C. C. Perry, Langmuir 21, 937–943 (2005).
The Porosity and Wettability Properties of Hydrogen Ion
Treated Poly(tetrafluoroethylene)
Hernando S. Salapare III ∗ , Gene Q. Blantocas, Virginia R. Noguera and

Henry J. Ramos
Plasma Physics Laboratory, National Institute of Physics, University of the Philippines, Diliman,
Quezon City 1101, Philippines
Abstract
The porosity and wettability properties of hydrogen ion treated poly(tetrafluoroethylene) (PTFE) materials
are related using contact angle, scanning electron microscopy (SEM) and ellipsometry techniques. PTFE
samples are irradiated using a low-energy hydrogen ion shower (LEHIS) produced by a Gas Discharge Ion
Source (GDIS). The plasma discharge current (Id ) is varied at intervals of 1 mA. Results show that treatment
using lower Id improved the hydrophobic property of the PTFE material with water contact angle increasing
from 102◦ to 119◦ . It also becomes less porous as indicated by the increase in the index of refraction,
decrease in optical transmittance, and increased fissures and striations in the SEM images. Opposite effects
are observed for higher Id .
Keywords
Poly(tetrafluoroethylene) (PTFE), porosity, low-energy hydrogen ion shower (LEHIS), gas discharge ion
source (GDIS), wettability, contact angle, ellipsometry, scanning electron microscopy (SEM)
1. Introduction
Surface modification of polymer materials has been of great interest to many re-
searchers for the past years because of its importance in the fields of materials
science, electronics and biomedical physics [1, 2]. Surface modification technology
allows for the change and improvement of the property of a material, consequently
making the processed material more useful in various aspects [3–6]. Different
characterization techniques that determine the change in the surface of the mate-
rial include X-ray photoelectron spectroscopy and atomic force microscopy, which
provide surface profiles of the polymers [7, 8]. Other characterization techniques
involve ellipsometry and fluorophotometry in which the dielectric function and
* To whom correspondence should be addressed. Tel.: +63-2-9204475; Fax: +63-2-9280296; e-mail:

hsalapare@nip.upd.edu.ph

208 H. S. Salapare III et al.
structure of the material can be obtained [9]. The most common technique to as-
sess the surface modification of a material is through contact angle measurements.
Modified porosity and wettability of surfaces of polymer materials find useful-
ness in orthopedic science [10], tissue engineering [11], immobilization of lipase
[12–15], cell culture [16] and drug delivery [16].
The material being modified in this study was poly(tetrafluoroethylene) (PTFE),
commonly known as Teflon. It was the chosen material because it is the simplest
organic polymer that is analogous to polyethylene. Altering the wettability of PTFE
is of great importance because of its applications in biosensors, hemo-compatible
materials, immobilized enzymes, microelectromechanical systems (MEMS) and
electronics [17–24]. PTFE materials’ wettability is altered to either hydrophilic or
hydrophobic, depending on their intended application [25].
In this study, PTFE materials were treated using a low-energy hydrogen ion
shower (LEHIS) system. The effect of the treatment on the porosity and wetta-
bility of the material was investigated using contact angle measurements, scanning
electron microscopy (SEM) and ellipsometry techniques.
2. Methodology
2.1. Experimental Device and Procedures
The PTFE tape of size 1 cm × 2 cm was wrapped around a 2 × 2 cm2 stainless steel
plate holder. The surface of the clean samples was blow-dried to prevent the forma-
tion of moisture on the surface. It was not subjected to ultrasonic cleaning because
of the fragile condition of the samples. PTFE was then irradiated using a low-energy
hydrogen ion shower (LEHIS) from a gas discharge ion source (GDIS) system. Fig-
ure 1(a) shows the schematic diagram of the experimental setup and Fig. 1(b) shows
the schematic diagram of the GDIS. It has a compact discharge region of volume
0.8 cm3 and an exit aperture of 2.0 mm in diameter. The extraction and focusing
electrodes are grounded to ensure a diffused ion shower configuration. The GDIS
fits a standard 70 mm knife-edge flange coupled to the diagnostic chamber whose
volume is about 2400 cm3 . A 10 cm-diameter oil diffusion pump coupled to an
8 m3 /h rotary pump evacuates the system. Complete details of the facility and its
operation are described in [3, 4] and [26, 27].
Pirani and ionization gauges monitor the pressure inside the chamber. The base
pressure was ∼1.0 × 10−6 Torr. The total hydrogen gas filling pressure was kept
at 3 mTorr for all experimental runs. Plasma was produced when a potential differ-
ence, Vd , was applied across the discharge region. The PTFE samples were placed
on a holder positioned 70 mm downstream from the entrance port of the processing
chamber. This position was determined to give maximum ion current density [3].
The irradiation time was set to 30 minutes. Treatment conditions are summarized
in Table 1.
Porosity and Wettability Properties 209
(a) (b)
Figure 1. Illustrations of the experimental setup. (a) Schematic diagram of the overall facility.
(b) Schematic diagram of the GDIS.
Table 1.
Summary of experimental parameters
Treated Irradiation Discharge Discharge

group time voltage, current,
(min) Vd (kV) Id (mA)
1 30 1.0 1.0
2 30 1.3 2.0
3 30 1.4 3.0
2.2. Characterization
2.2.1. Ion Beam Detection
The ion species were determined using a cast steel mass spectrometer. The mass
spectrometer was then replaced by a Faraday cup system, which consisted of an
isolated brass disk of 2.0 cm diameter and was connected to an electrometer. The
Faraday cup was used to determine the total ion current density [26].
The treated and untreated samples were subjected to contact angle measurements
using an Intel® Play™ QX3™ Computer Microscope. The absorption of the water
droplet in the sample was recorded at one frame per 5 seconds. Deionized water
was used to prevent surface contamination which may distort the wetting dynamics
of the samples.
The video of the absorption of water was processed using AVS DVD Media
Player to obtain the images of the absorption every 5 seconds. ImageJ was used to
measure the contact angle from each frame of the absorption of deionized water.
Contact angle was then plotted against absorption time to show the dynamics of
the water droplet absorption on the surface of the sample. Three replicates for each
experimental condition were considered.
2.2.3. Ellipsometry Characterization
A VASE M-44 ellipsometer (J. A. Woolam Co., USA) was used to determine the
transmission of light through the sample and index of refraction. The transmittance
and the index of refraction were considered at 595 nm wavelength. This was done
to compare the results of the optical transmittance with a previous study [3].
2.2.4. Scanning Electron Microscopy
The substrates were coated with gold using a JEOL (JFC-1100) Fine Coat Ion Sput-
ter unit to avoid accumulation of electrons on the surface of the material because of
its fragile state and because the sample is a non-conducting polymer. A Leica S440
Scanning Electron Microscope (SEM) was used to determine the structure of the
surface of the material.

3.1. Ion Beam Characterization
The typical hydrogen ion peaks detected by the cast steel mass spectrometer are
shown in Fig. 2(a). The current intensity (nA) of the hydrogen beam is plotted
against the scanning magnetic field (G). Figure 2(b) shows the ion current density
for different discharge conditions, registering a high of 0.25 A/m2 and a low of
0.06 A/m2 . Monoatomic H+ ions dominate the H+ 2 ions of the beam produced by
the gas discharge ion source for discharge currents of 1 mA, 2 mA and 3 mA.
Hence, there is reason to believe that the monoatomic ions play an active role in the
surface modification process.
3.2. Contact Angle Measurement
Rates of decrease of the contact angles for some representative samples are shown
in Fig. 3. Wettability is quantified by fitting the wetting model used in [3, 26] to
actual data. The model is expressed mathematically as
dθ
= −kθ, (1)
dt
where θ is the contact angle between the supporting solid surface and the tangent
to the drop-shape of the liquid, and k being the change rate constant or the quantity
that describes the angle’s rate of decrease in units of per second. The baseline value
for k is that of the control (i.e., all k-values of the treated samples are measured and
interpreted against the k-value of the control). Rising values of k signify increas-
ing affinity to water while samples with diminishing values are less wettable, viz.,
compared to the control. Small variations in k are extremely significant as these will
indicate whether a particular sample is moisture absorbent or not. The R-squared
(a)
(b)
Figure 2. (a) Typical mass spectra of LEHIS at varying plasma discharge currents. (b) Plot of the ion
current density at different discharge currents.
method shows that the experimental data correlate well with the wetting model as
indicated by the high values of R 2 (98% fit) for all experimental conditions. Com-
pared to the control (k = 0.0065/s), samples from group 1 show lower values of k
(∼0.0054/s), while samples from groups 2 and 3 show higher values (k ∼ 0.0089/s
and 0.016/s, respectively). These results indicate that in relation to the control, sam-
ples treated with higher discharge currents became hydrophilic (θ < 90◦ ) while
samples treated using low discharge current showed improved hydrophobicity as
Figure 3. Typical numerical constructs of equation (1) (wetting model) fitted against empirical data.
The time rate equation of the wetting model is made to fit actual contact angle data. Calculations of
the goodness index of fit employing the R-squared method show 98% match between experiment
and theory. The wetting dynamics of PTFE is aptly captured by the model. PTFE is known to be
hydrophobic, thus the k-value of this material (control) is used as reference to interpret the wettability
characteristics of the treated samples. Lower k-values denote longer moisture absorption time and
higher k-values signify faster absorption vis-à-vis the control. Group 1 treated with low-energy beams
has a lower k-value indicating improved hydrophobicity. In contrast, groups 2 and 3 consisting of
samples exposed to higher energy ion showers came up with higher k-values indicative of specimens
with hydrophilic qualities.
the initial contact angle increased from 110◦ to about 120◦ . By convention, wettabil-
ity property is established through the initial contact angle θ (t = 0). The 90◦ angle
separates the hydrophilic and hydrophobic regimes [28]. Wetting occurs at θ < 90◦
and non-wetting if θ > 90◦ . This technique however is inadequate in that when
a liquid droplet is placed on a surface, liquid penetration as well as spreading occur.
The contact angle is never static; it will recede over time in accordance with equa-
tion (1). A material may be hydrophobic initially [θ (t = 0) > 90◦ ], however due to
liquid spreading, it will eventually cross over to the hydrophilic regime (θ < 90◦ ),
eluding classification as to whether it is hydrophobic or otherwise. Describing the
wettability characteristic of a material based on its k-value is straightforward and
more inclusive as this method takes into account the total recession time of the liq-
uid on the surface. Theoretically based on equation (1), a truly hydrophobic material
has k = 0 and a hydrophilic material may have its k as high as unity.
Table 2.
Values of discharge current, optical transmittance, and index of refraction
Group Discharge Optical Index of

current, Id transmittance refraction
(mA) (%)
Control 82.6 1.19

1 1 69.1 1.22
2 2 95.9 1.10
3 3 98.3 1.07
The cause for the increased hydrophilicity of the samples can be attributed to the
increased surface roughness [29–32] and porosity of the material. Hydrophobiza-
tion by ion irradiation is not yet fully understood and further experiments are being
conducted to explain the phenomenon. This result is in correlation with the work of
Blantocas et al. on the ion irradiation of wood samples [32].
3.3. Ellipsometry Measurements
Table 2 summarizes the results obtained from the ellipsometry technique. It shows
the optical transmittance and the index of refraction of the samples at 594 nm wave-
length.
The optical transmittance investigates material porosity by measuring the trans-
mittance of light through the sample. Transmittance may be regarded as a porosity
index of the sample. A porous material tends to have a high transmittance value and
at the same time is also likely to have high water absorbency. In this sense, transmit-
tance may be regarded as an indicator of material wettability. In Table 2, samples
irradiated by low-energy ion shower (Id =1 mA) exhibit the least transmittance,
meaning the material is also the least porous. Samples from group 1 give the lowest
transmittance of 69.1%, and also are the least wettable, having the lowest k-value.
The opposite effect is seen for samples irradiated by higher energy beams. Samples
from group 3 have the highest transmittance of 98.3% and are the most wettable as
well.
The index of refraction of samples treated with low discharge current is higher
than of those samples treated with high discharge current for the 594 nm wave-
length. High index of refraction indicates that the density of the material is also
high [33]. This would imply that the samples treated with low discharge current
have a high density as compared with samples treated with high discharge current.
3.4. SEM Characterization
Figure 4 shows SEM images of the untreated and treated samples. All SEM images
were recorded at 20k× magnification. Figure 4(a) shows the surface of samples
from group 1 subjected to lower energy ion shower (Id = 1 mA) that are much
smoother than the rest of the specimens. Samples from groups 3 and 4 subjected
Figure 4. (a) SEM image of a representative sample from the control group, (b) typical SEM im-
age from the group processed at low discharge current, Id = 1.0 mA, (c) typical SEM image from
the group processed at Id = 2 mA, and (d) typical SEM image from the group processed at higher
discharge current, Id = 3 mA.
to higher energy ion showers (Id 2 mA, 3 mA) show surfaces that are rough and
have striations and fissures as seen in Fig. 4(c) and (d). The surface deterioration of
group 3 looks more pervasive than that of sample 2, possibly due to higher discharge
current. It becomes clear from this figure vis-à-vis wettability and porosity parame-
ters that generally a surface becomes more wettable and optically transmissive as
it roughens. Roughening by high-energy ion irradiation leads to surfaces becoming
more hydrophilic and porous. The observed striations and fissures on the surface ac-
count for their higher optical transmittance. This finding is in consonance with the
work of Tzeng et al. on the roughening of Teflon samples using wet processes [30].
Conversely, low-energy ion irradiation results in relatively smooth, hydrophobic
surfaces. PTFE materials become thinner when bombarded by high flux density
ion showers [3]. Porosity can also be analyzed in relation with the thickness of the
material. The optical transmittance T in % as described by the Beer–Lambert law
is
T = P /P0 = e−εbc , (2)
where P0 and P are the powers of the incident and transmitted light, respectively,
ε is the molar absorptivity of the absorber, b is the path length of the sample which
is also the thickness, and c is the concentration of the absorbing species in the ma-
terial. Thinner samples result in higher transmission of light. Cross-sectional SEM

images of the samples are reported in [3]. The density and the index of refraction
of the material changed due to the air gaps created by the striations as seen from
the SEM images. The produced striations and fissures indicate that the material
becomes more porous when irradiated by higher energy ion showers.
4. Conclusions
The porosity and wettability properties of hydrogen ion treated PTFE materials
were evaluated using contact angle, ellipsometry and SEM techniques. Low dis-
charge LEHIS treatment produces hydrophobic, smooth, less porous surfaces while
high discharge LEHIS produces hydrophilic, rough, optically transmissive, more
porous surfaces.
Acknowledgements
The financial support of the project from the Department of Science and Technology
(DOST) — Philippine Council for Advanced Science and Technology Research and
Development (PCASTRD) is gratefully acknowledged.
References
1. K. L. Mittal (Ed.), Polymer Surface Modification: Relevance to Adhesion, Vol. 4. VSP/Brill, Lei-
den (2007).
2. M. Strobel, C. S. Lyons and K. L. Mittal (Eds), Plasma Surface Modification of Polymers: Rele-
vance to Adhesion. VSP, Utrecht (1994).
3. H. S. Salapare III, G. Q. Blantocas, V. R. Noguera and H. J. Ramos, Appl. Surface Sci. 255,
2951–2957 (2008).
4. H. J. Ramos, J. L. C. Monasterial and G. Q. Blantocas, Nucl. Instrum. Meth. Phys. Res. B 242,
41–44 (2006).
5. M. S. Shen, D. M. Hudson and I. H. Loh, in: The Encyclopedic Handbook of Biomaterials and
Bioengineering: Part A, D. L. Wise, D. J. Trantolo, D. E. Altobelli, M. J. Yaszemski, J. D. Gresser
and E. R. Schwartz (Eds), p. 865. Marcel Dekker, New York (1995).
6. C. M. Chan, T. M. Ko and H. Hiraoka, Surface Sci. Rep. 24, 3 (1996).
7. P. G. de Gennes, Rev. Mod. Phys. 57, 827–863 (1985).
8. J. Israelachvili, Intermolecular and Surface Forces. Academic Press, New York (1992).
9. R. Mohammadi, W. H. Finlay, W. Roa and A. Amirfazli, Proceedings of the International Confer-
ence on MEMS, NANO and Smart Systems, Banff, Alberta, Canada (2003).
10. J. Glodek, P. Milka, I. Krest and M. Keusgen, Sensors Actuators B 83, 82–89 (2002).
11. J. Chakraborty, S. D. Sarkar, S. Chatterjee, M. K. Sinha and D. Basu, Colloids Surfaces B: Bioin-
terfaces 66, 295–298 (2008).
12. C. Erisken, D. M. Kalyon and H. Wang, Biomaterials 29, 4065–4073 (2008).
13. S. Gao, Y. Wang, T. Wang, G. Luo and Y. Dai, Bioresource Technol. 100, 996–999 (2009).
14. L. F. Zhang, R. Sun, L. Xu, J. Du, Z. C. Xiong, H. C. Chen and C. D. Xiong, Mater. Sci. Eng. C
28, 141–149 (2008).
15. L. Safinia, N. Datan, M. Höhse, A. Mantalaris and A. Bismarck, Biomaterials 26, 7537–7547
(2005).
16. M. T. Khorasani, H. Mirzadeh and Z. Kermani, Appl. Surface Sci. 242, 339–345 (2005).
17. T. Białopiotrowicz and B. Jańczuk, Appl. Surface Sci. 201, 146–153 (2002).
18. W. Prissanaroon, N. Brack, P. J. Pigram. P. Hale, P. Kappen and J. Liesegang, Synth. Metals 154,
105–108 (2005).
19. V. N. Vasilets, G. Hermel, U. Konig, C. Wener, M. Muller, F. Simon, K. Grundke, Y. Ikada and
H. J. Jacobasch, Biomaterials 18, 1139–1145 (1997).
20. J. M. Li, M. J. Singh, P. R. Nelson, C. M. Hendricks, M. Itani, M. J. Rohrer and B. S. Cutler,
J. Surgical Res. 105, 200–208 (2002).
21. M. Crombez, P. Chevallier, R. C. Gaudreault, E. Petitclerc, D. Mantovani and G. Laroche, Bioma-
terials 26, 1402–1409 (2005).
22. P. Chevallier, R. Janvier, D. Mantovani and G. Laroche, Macromol. Biosci. 5, 807–810 (2005).
23. E. T. Kang, K. L. Tan, K. Kato, Y. Uyama and Y. Ikada, Macromolecules 29, 6872–6879 (1996).
24. R. K. Y. Fu, Y. F. Mei, G. J. Wan, G. G. Siu, P. K. Chu, Y. X. Huang, X. B. Tian, S. Q. Yang and
J. Y. Chen, Surface Sci. 573, 426–432 (2004).
25. H. Y. Kwong, M. H. Wonga and Y. W. Wong, Appl. Surface Sci. 253, 8841–8845 (2007).
26. G. Q. Blantocas, H. J. Ramos and M. Wada, Rev. Sci. Instrum. 75, 2848–2853 (2004).
27. G. Q. Blantocas, H. J. Ramos and M. Wada, Jpn. J. Appl. Phys. 45, 8525–8530 (2006).
28. J. M. Goddard and J. H. Hotchkiss, Prog. Polym. Sci. 32, 698–725 (2007).
29. S. Guruvenket, M. Kumath, S. P. Vijayalakshmi, A. M. Raichur and G. M. Rao, J. Appl. Polym.
Sci. 90, 1618 (2003).
30. G. S. Tzeng, H. J. Chen, Y. Y. Wang and C. C. Wan, Surface Coatings Technol. 89, 108–113
(1997).
31. F. Garbassi, M. Morra and E. Occhiello, Polymer Surfaces: From Physics to Technology, Chap-
ter 4. Wiley, New York (1994).
32. G. Q. Blantocas, P. E. R. Mateum, R. W. M. Orille, R. J. U. Ramos, J. L. C. Monasterial, H. J.
Ramos and L. M. T. Bo-ot, Nucl. Instrum. Meth. Phys. Res. B 259, 875–883 (2007).
33. F. Wooten, Optical Properties of Solids. Academic Press, New York (1972).
Part 3
Superhydrophobic Surfaces
Superhydrophobicity: Localized Parameters and
Gradient Surfaces
G. McHale ∗ , S. J. Elliott, M. I. Newton and N. J. Shirtcliffe

School of Science and Technology, Nottingham Trent University, Clifton Lane,
Nottingham NG11 8NS, UK
Abstract
The use of Cassie and Baxter’s equation and that of Wenzel has been subject to some criticism of late. It has
been suggested that researchers use these equations without always considering the assumptions that have
been made and sometimes apply them to cases that are not suitable. This debate has prompted a reconsid-
eration of the derivation of these equations using the concept of parameters for the Wenzel roughness and
Cassie–Baxter solid surface fractions that are local to the three-phase contact lines. In such circumstances,
we show the roughness and Cassie–Baxter solid fractions depend not only on the substrate material, but
also on which part of the substrate is being sampled by the three-phase contact lines of a given droplet. We
show that this is not simply a theoretical debate, but is one which has direct consequences for experiments
on surfaces where the roughness or spatial pattern varies across the surface. We use the approach to derive
formulae for the contact angle observed on a double length scale surface under the assumption that the
small-scale features on the peaks of larger scale features are either wetted or non-wetted. We also discuss
the case of curved and re-entrant surface features and how these bring the Young’s law contact angle into
the formula for roughness and the condition for suspending droplets without penetration into the surface. To
illustrate the use of local parameters, we consider the case of a variation in Cassie–Baxter fraction across a
surface possessing a homogeneous hydrophobic surface chemistry and discuss the conditions (droplet vol-
ume, surface hydrophobicity, gradient in superhydrophobicity and contact angle hysteresis) under which a
droplet may be set into motion. We show that different contact angles on each side of a droplet of water
placed on such a surface can generate sufficient lateral force for the droplet to move towards the region of
the surface with the lowest contact angle. Using an electrodeposited copper surface with a radial gradient
in superhydrophobicity we exemplify these ideas by showing experimentally that droplets enter into self-
actuated motion and accumulate in the centre of the surface where the wettability is higher. In principle,
paths can be defined and water droplets can be collected by creating such gradients in superhydrophobicity
through changes in the lateral topography of the surface.
Keywords
Superhydrophobicity, Cassie–Baxter, Wenzel, wetting, contact angle
* To whom correspondence should be addressed. Tel.: +44 115 8483383; Fax: +44 115 9486636; e-mail:
glen.mchale@ntu.ac.uk

220 G. McHale et al.
1. Introduction
Superhydrophobic surfaces constitute one class of wetting problems in which

topography amplifies the effect of surface chemistry [1]. In the superhydropho-
bic case, water repellence is emphasized to create extremely high contact angles
and low contact angle hysteresis. Surfaces with high contact angles and high con-
tact angle hysteresis, and surfaces that create superspreading [2, 3], superwetting
or hemiwicking [4, 5] use similar topographic principles, but these latter surfaces
emphasize the spreading tendency of a liquid on a given substrate material. Many
methods exist for creating superhydrophobic surfaces and these have been reviewed
by a range of authors [6–9]. It is often the case that their properties are discussed
with reference to the Cassie–Baxter and Wenzel models [10–13] and, in particular,
with reference to a very specific type of surface composed of flat-topped vertical
posts [14]. This has led to a common description of superhydrophobicity as due to
a droplet behaving as if it were sitting on a bed-of-nails (a fakir’s carpet). Whilst
it is certainly true that a droplet is supported by the surface protrusions below the
entirety of the solid surface within the wetted perimeter of a droplet, this simplified
view appears to have caused confusion with regards to the definition of the Cassie–
Baxter solid fraction and Wenzel roughness parameter [15–17]. With the maturing
of the field of superhydrophobicity, such that many materials and methods are now
available to create surfaces, it is important that topographic amplification of wet-
ting occurring locally at the three-phase contact lines, both at the perimeter of the
droplet and below the droplet within that perimeter, are understood. One potential
area of application for superhydrophobic surfaces is droplet transport and here well-
developed concepts of wettability gradients and actuating forces are required [18,
19].
In Section 2, wetting on defect and composite surfaces is discussed and this
leads on, in Section 3, to a consideration of the implications for understanding the
Wenzel and Cassie–Baxter models. The principal outcome is to emphasize that
the Cassie–Baxter fraction and Wenzel roughness parameter are defined locally to
the three-phase contact lines [16]; similarly the relevant Young’s law contact angle
in these two formulae relates to the local surface chemistry. In Section 4, we discuss
a number of more complex cases, including two-length scales, re-entrant surfaces
and wetting on spherical beads; spherical beads provide an example of a system
for which roughness may become a function of the Young’s law contact angle. In
Section 5, we consider how varying the wettability across a surface by changing the
Cassie–Baxter fraction with position might be used to generate motion of droplets.
Finally, in Section 6, we exemplify some of these ideas by briefly presenting an ex-
ample of a surface where we created a radial gradient in superhydrophobicity using
electrodeposition of copper so that droplets roll to a central location. Furthermore,
we suggest that more complex patterns could be used to define paths and tracks for
droplet transport [1, 16].
Localized Parameters and Gradient Surfaces 221
2. Wetting on Defect and Composite Surfaces

A fundamental question in wetting is whether processes are local to the three-phase
contact lines or whether the entireties of the various interfaces need to be taken into
account. First consider measuring advancing and receding contact angles by using
a syringe inserted into the apex of a sessile droplet. This causes a slight distortion of
the liquid–vapor interface and so prevents axisymmetric drop shape analysis based
on the full profile of the droplet. An alternative method, used in studies of possi-
ble line tension effects [20, 21], is to fill the droplet by delivering liquid through a
hole in the substrate (Fig. 1a). However, the existence of the hole now means the
droplet sits on a composite surface of the solid substrate and a central area which is
the liquid-filled hole. The presumption is that provided the droplet contact area is
wider than the hole, this does not alter the contact angles measured at the perimeter
of the droplet. Now imagine that as the droplet grows in volume it encounters suc-
cessive changes in surface chemistry from one with a low contact angle to one with
a higher contact angle, i.e. a surface with radial rings of differing surface chem-
istry. This experiment has been reported in the literature (e.g. [22]) and the contact
angles measured, both advancing and receding, are those that would be expected
for a droplet entirely on a surface with the same surface chemistry as that surface
chemistry where the perimeter rests upon [23, 24]. Our expectation is that when the
droplet perimeter reaches the boundary from a lower contact angle to higher con-
tact angle ring, the droplet perimeter will stop advancing and the contact angle will
increase until suddenly a rapid advance will occur across part of the higher contact
angle ring (see also Ref. [22]). From these considerations it should, therefore, be
clear that contact angles are local to the three-phase contact lines; this is a long held
view within the literature on contact angles (see Ref. [25]). The phrase “three-phase
contact lines” has been chosen carefully to be plural to emphasize that there may
be more than one such three-phase contact line for a droplet. For example, if in
Fig. 1 the tube feeding liquid into the droplet from below was empty of liquid, but
was narrow enough and sufficiently hydrophobic that liquid from the droplet did
Figure 1. (a) Measurement of advancing contact angle, θ , for a droplet on a smooth flat surface (r is
the droplet contact radius), and (b) two equilibrium configurations for a droplet of a fixed volume
on a smooth flat hydrophilic surface characterized by a Young’s law contact angle θ2S , with a central
hydrophobic defect characterized by a Young’s law contact angle θ1S (r1 and r2 are the droplet contact
radii for the two droplet configurations).
not penetrate into it, then a three-phase contact line would exist below the droplet;
this would be an additional three-phase contact line disconnected from the one at
the droplet’s external perimeter on the substrate.
Now consider a circular hydrophobic defect with, say, θ1S = 110◦ , within a more
hydrophilic region having, say, θ2S = 70◦ , where both contact angles are due to a
homogeneous surface chemistry within their respective regions. For a given volume
of droplet, the total surface free energy can be calculated. For some volumes, such
as V = 1 × 10−9 m3 , we find that provided the central defect is within a certain size
range, two stable configurations that satisfy a minimum in the surface free energy
can exist [16]. The first of these is with a droplet sitting entirely on the hydrophobic
defect and the second is with the droplet perimeter sitting entirely on the more
hydrophilic region. Thus, not only is the Young’s law contact angle local to the
three-phase contact line important, but the initial state is also important because it
will determine what configurations close to that initial state can be sampled by the
three-phase contact line to reach an equilibrium.
3. Wenzel and Cassie–Baxter Equations with Local Parameters

We now consider the analogous situation of a rough defect within a smooth surface
area with a droplet sitting entirely on the defect and maintaining contact with the
surface at all points beneath itself (Fig. 2a). The contact angle, θW , for this situation
is described by the Wenzel equation [8],
cos θW = r cos θe , (1)
where θe is the Young’s law contact angle and r is the Wenzel roughness factor,
which is often defined to be the ratio of actual area to planar projection of area of
the substrate. This implies that roughness is a property of the substrate alone, which
is clearly not the case, since that would then imply that the extent of smooth area
completely remote from the vicinity of the droplet would determine the roughness
factor. Restricting the definition to the area beneath the droplet would give a def-
inition r = Awetted /πrc2 where rc is the planar contact radius. However, this then
Figure 2. Droplet of contact radius rc with its contact line on (a) a rough patch of radius ri within a
smooth flat surface of radius ro , and (b) on the smooth flat surface surrounding the rough patch. All
surfaces are characterized by a Young’s law contact angle θe .
produces a definition that seems contradictory to the situation discussed for a sin-
gle defect entirely encompassed by a droplet since it implies that the interior away
from the three-phase contact line matters. The analogous case is Fig. 2b whereby the
rough patch is entirely within the wetted area and the droplet perimeter lies entirely
on the external smooth area; the rough portion of the surface does not influence
the contact angle which is given by Young’s law for the smooth area in the vicinity
of the droplet three-phase contact line. Thus it is the surface areas in the proxim-
ity of the three-phase contact line, x, that matter and the appropriate definition of
roughness is,

Awetted (x)
r(x) = , (2)

Ac (x)
where
Awetted is a small change in wetted area that can be sampled by a three-
phase contact line and
Ac = 2πrc
rc is the planar projection of that change.
More precisely, it is the first derivative in the wetted area with respect to the planar
projection of area evaluated at the contact line. Thus, roughness is not an average
property of the substrate or even of the substrate below the droplet, but is local to the
three-phase contact line(s) where changes can occur, i.e. r = r(x) where x is used
to indicate that values are taken at the three-phase contact line. It also emphasizes
that the initial state of the droplet from the deposition matters because it defines the
part of the substrate surface that the three-phase contact line initially samples.
Similar arguments can be made for a surface composed of areas of two different
surface chemistries (Fig. 3). Provided the patchwork of different surface chemistries
is entirely encompassed by the droplet (Fig. 3b), only the surface chemistry of the
area on which the three-phase contact line sits matters as far as the contact angle
is concerned. If the three-phase contact line can sample the two different surface
chemistries, the droplet contact angle, θCB , is defined by the Cassie–Baxter formula,
cos θCB = f1 cos θ1 + f2 cos θ2 , (3)
where θ1 and θ2 are the Young’s law defined contact angles on the two surface types
and the f1 and f2 are normally defined as the area fractions. The term area fraction
could be taken to mean fi = Ai /(A1 + A2 ), where Ai are the areas associated with
Figure 3. Cases (a) and (b) show two possible arrangements for droplets with their contact lines
entirely on a patch possessing one type of surface chemistry. The existence of several patches with
two different surface chemistries within the wetted area does not alter the contact angle, which is
determined by the surface chemistry at the contact line.
each type of surface chemistry, but if that were to be the case we would be implying
that the Cassie fractions are global properties of the substrate or of the substrate
area beneath the droplet. Essentially, the hole for the syringe in Fig. 1a would mat-
ter. Considering the case of Fig. 3b, whereby the patchwork of the two types of
surfaces is entirely encompassed within the wetted portion of the surface and the
three-phase contact line sits entirely on the external area which is uniformly of one
type of surface chemistry, the droplet contact angle is determined by the Young’s
law contact angle for that surface. Thus, it is the surface areas in the proximity of
the three-phase contact line, x, that matter and the appropriate definition of Cassie
fraction is,

Ai (x)
fi (x) = , (4)

AT (x)
where
Ai (x) is the change in wetted area of type i that can be sampled by a small
three-phase contact line change
AT (x) =
A1 (x) +
A2 (x). Thus, the Cassie
fractions are not average properties of the substrate or even of the substrate below
the droplet, but are local to the three-phase contact line(s) where changes can occur,
i.e. fi (x) where x is used to indicate that values are taken at the three-phase contact
line. As with the rough surface case, it also emphasizes that the initial state of
the droplet from the deposition matters because it defines the part of the substrate
surface that the three-phase contact line can initially sample as it progresses towards
equilibrium.
The fundamental conclusion from these considerations is that the Wenzel rough-
ness parameter, the Cassie fractions and the Young’s law contact angles in the
Wenzel and Cassie–Baxter formulae are local to the three-phase contact line(s) and
not global properties of the substrate or of the area beneath the droplet, i.e.
cos θW (x) = r(x) cos θe (x), (5)
cos θCB (x) = f1 (x) cos θ1 (x) + f2 (x) cos θ2 (x), (6)
where the functional dependence on the location of the three-phase contact line is
given by the (x) notation. When the Cassie–Baxter formula is applied to a surface
consisting of a simple flat-topped post-type structure, so that θ1 = θe , θ2 = 180◦ ,
f1 (x) = f (x) and f2 (x) = (1 − f (x)), equation (6) can be written as,
cos θCB (x) = f (x) cos θe (x) − (1 − f (x)). (7)
While these conclusions have been justified by comparison to a surface with a de-
fect, it is also possible to more rigorously derive these conclusions (see Ref. [16]).
Arguments about wetting using surfaces visualized using a cartoon, such as
Fig. 3a, do not fully represent the 3-D situation in which a droplet exists. If we
translate the axial symmetry for the droplet to the surface, the surface becomes a
set of concentric rings with a droplet at its centre. The small change
A(x) that
a three-phase contact line can sample is then unlikely to cover one period of the
surface and unless the droplet’s three-phase contact line is at the edge of a ring, the
small change will only sample one type of surface. The Cassie–Baxter formula can
be applied, but one of the fi is unity and the other vanishes, resulting in a contact
angle given by Young’s law for that surface type. Use of the local version of the
Cassie–Baxter formula (equation (6)), therefore, involves an assumption that the
small change
A(x) is, on average, equivalent to one period of the surface. For a
droplet on a surface that does not have a symmetry in how the patches of different
surface types are distributed and for a droplet much larger than the patch size and
visibly demonstrating an approximately circular three-phase contact line, this might
be a reasonable assumption [16] although there is no proof that these conditions
alone are sufficient. Drops of comparable size to the patch sizes and ones displaying
facets may not be well described by equation (6). In addition, these considerations
do not capture the full complexity that occurs between the interplay of local sur-
face feature shape, distribution and organization, and the advancing and receding
contact angles important for the design of practical liquid-shedding surfaces [26].
4. Dual Length Scales and Re-entrant Surfaces

Derivations of the results in the previous section using minimisation of changes
in surface free energy for a 2-D flat-topped post-type structure have previously
been presented [16]. In this section, we begin by using the same approach to con-
sider the slightly more complicated cases of multiple length scales. Consider a
post-type surface, but with the tops of the posts themselves possessing a post-type
structure. The three-phase contact line may advance across one period,
Ap (x)
of the large-scale post structure, characterised by a local Cassie fraction fL (x) =
p p p
ATop (x)/(
ATop (x) +
ABottom (x)) where the subscript L indicates large-scale
structure, the superscript p indicates planar projections of areas and the “Top” and
“Bottom” refer to the large-scale post structure.
4.1. Top-Filled Case
Focussing on the wetting of the smaller scale structure on the top of each large
post, we can imagine the idealized situation in which either it is completely wetted
Figure 4. Changes in surface free energy as a contact line advances over a two length scale structure
assuming liquid penetrates the small-scale structure at the top of posts, but does not penetrate the
larger scale structure. The roughness at the top of posts is characterized by a Wenzel parameter, rS ,
and the larger scale structure is described by a Cassie fraction, fL .
or there is no penetration by the liquid. For the first case, corresponding to the tops
being filled, the surface free energy change,
F (x), is

F (x) = (γSL − γSV )rS (x)fL (x)
Ap (x)
+ γLV (1 − fL (x))
Ap (x) + γLV cos θ
Ap (x), (8)
where the γij are the interfacial tensions and rS (x) is the local small-scale roughness
at the top of the posts (indicated by the subscript S on the roughness parameter).
Setting this surface free energy change to zero and using the usual definition of the
Young’s law contact angle, cos θe = (γSV − γSL )/γLV , we find that the observed
contact angle, θObs , is given by,
cos θObs (x) = rS (x)fL (x) cos θe (x) − (1 − fL (x)). (9)
4.2. Top-Empty Case

Now consider the second case, whereby the small-scale roughness at the top of each
post of the larger scale structure is not penetrated by the liquid. Essentially, the top
of each post in the large-scale structure corresponds to a Cassie–Baxter state rather
than a Wenzel state and this is characterized by a solid fraction fS (x), where the
subscript S indicates the Cassie fraction for the small-scale structure at the top of a
post. The surface free energy change is then,

F (x) = (γSL − γSV )fS (x)fL (x)
Ap (x)
+ γLV [(1 − fL (x)) + fL (x)(1 − fS (x))]
Ap (x)
+ γLV cos θ
Ap (x). (10)
Setting this surface free energy change to zero and using the usual definition of the
Young’s law contact angle and grouping terms, we find that the observed contact
angle is given by
cos θObs (x) = fL (x)[fS (x) cos θe (x) − (1 − fS (x))] − (1 − fL (x)). (11)
Figure 5. Changes in surface free energy as a contact line advances over a two length scale structure
assuming liquid does not penetrate either the small-scale structure at the top of posts or the larger scale
structure. The small-scale structure at the top of posts is described by a Cassie fraction parameter, fS ,
and the larger scale structure is described by a Cassie fraction, fL .
4.3. Transformation Formulae

Equations (9) and (11) represent different assumptions about the wetting state of the
tops of the posts in the large scale structure. The formulae can be related back to the
single length scale formulae (equation (5) and equation (7)) by thinking in terms of
successive transformations due to the two length scale structures. In the case of the
top-filled situation, the Young’s law contact angle for the smooth solid, θe (x), is
first transformed to a local Wenzel contact angle, θW (x), using the local roughness
factor, rS (x), existing at the top of the posts. Subsequently this local Wenzel contact
angle is transformed via the Cassie–Baxter formula (equation (7)) using the solid
fraction, fL (x) for the large scale structure and the Wenzel contact angle in place of
the Young’s law contact angle, i.e.
r(x) fL (x)
θe (x) −→ θW (x) −→ θObs . (12)
In the case of the top-empty situation, the Young’s law contact angle for the smooth
solid, θe (x), is first transformed to a local Cassie–Baxter contact angle, θCB (x),
using the local Cassie fraction for the small scale structure, fS (x), existing at the
top of the posts. Subsequently this local Cassie–Baxter contact angle is transformed
via the Cassie–Baxter formula (equation (5)) using the solid fraction, fL (x), for
the large scale structure and the Cassie–Baxter contact angle from the small-scale
structure in place of the Young’s law contact angle, i.e.
fS (x) fL (x)
θe (x) −→ θCB (x) −→ θObs . (13)
More complex wetting situations including roughness at the base of the large-scale
structure could be considered using this approach based on minimising surface free
energy changes. For example, having roughness at the base of large posts in a situ-
ation whereby tops of posts are completely wetted, but the large scale structure is a
Cassie–Baxter state, would not lead to any change to equation (12). However, if the
second transformation was a Wenzel one, a roughness at the base of the large scale
structure would have an effect.
4.4. Curved Features and Re-entrant Shapes
Superhydrophobic effects for curved surfaces have a long history because of the ori-
gin of the subject in the water repellency of fibers and textiles. Indeed, the original
Cassie–Baxter formula was based on cylindrical shapes and used a formula similar
to equation (9) with both a roughness factor and a Cassie-fraction [10]. One differ-
ence is that the roughness factor, which really represents the extra length of wetted
area compared to a planar projection, is related to how far down a curve is wetted
and this itself depends on the local Young’s law contact angle. Thus, the roughness
factor should be written as r = r(x, θe (x)) indicating a dependence on the surface
chemistry (and liquid since that can also result in a different Young’s law contact
angle) as well as the physical location. The impact of this on the wetting of spherical
beads has been considered in the literature [27, 28] and is essentially an application
of the transformation law given by equation (12). A single layer of beads is able to
suspend a liquid even when the Young’s law contact angle approaches 0◦ . If there
are multiple layers and the beads are hexagonally close-packed, a liquid will im-
bibe into the bead pack under capillary forces once the contact angle falls below
50.8◦ , but not before unless pressure or gravity is considered as a driving force.
This contact angle corresponds to the point at which the penetrating front of the
liquid touches a bead from the next layer below [29]. The use of superhydrophobic
ideas to describe soil, when regarded as bead packs, was first proposed in Ref. [27]
and more recently a complete theoretical analysis based on equation (12) and an ex-
perimental comparison have been carried out [28]. Recognition of the importance
of inward (re-entrant) curves, such as observed with bead packs, in the ability of
superoleophobic surfaces to suspend liquids with Young’s law contact angles sig-
nificantly below 90◦ has recently been reported in [30].
5. Gradient Superhydrophobic Surfaces

It has long been known that on surfaces with variations in surface chemistry droplets
move towards regions of lower wettability (see Ref. [31] and references therein).
In an earlier report we suggested that lateral variation in topography to create a
variation in superhydrophobicity should also generate droplet motion even when the
surface chemistry was homogeneous [1, 32]. A number of reports have considered
this problem experimentally with varying levels of success [32–35] and there has
been at least one attempt to model this theoretically [33]. Lithographic approaches
have tended to lead to droplets that are unable to move unless energy is inputted
via, e.g., vibration [33], but such surfaces have been shown to be useful in inducing
asymmetric rebounds from impacting droplets [37].
The basic concept of lateral gradient forces in superhydrophobicity is shown
schematically in Fig. 6. In this case, the horizontal spacing between features is
progressively decreased across a surface so that the contact angles at the left-hand
and right-hand sides of the droplet, θL , and θR , differ by a small amount. The driving
Figure 6. The concept of inducing motion by a gradient in superhydrophobicity achieved using a vari-
ation in lateral scale topography. The magnitude of the driving force, given by γLV (cos θL − cos θR ),
where γLV is the liquid–vapor surface tension and θL and θR are the observed contact angles on the
left- and right-hand sides of the droplet, must overcome contact angle hysteresis.
force per unit length of contact line is then γLV (cos θR − cos θL ) and, assuming the
surface chemistry is homogeneous, this can be written using the local form of the
Cassie–Baxter equation as,
Force/length = γLV [f (xR ) − f (xL )][1 + cos θe ]. (14)
Since xR = xo + rc and xL = xo − rc , where xo is the centre coordinate of the droplet
in the contact plane and rc is the contact diameter which is small, the local Cassie
fractions can be expanded in terms of the gradient in Cassie fraction,

df
Force/length = 2γLV rc (1 + cos θe ) . (15)
dx x=xo
If we assume that our droplet is a spherical cap with a small contact area we can
write

2rc ≈ 2R 2f (xo )(1 + cos θe ), (16)
where f (xo ) is the average Cassie fraction between the left-hand and right-hand
sides of the droplet. Hence, the force per unit length of contact line becomes

3/2 df
Force/length = γLV 2R 2f (xo )(1 + cos θe ) . (17)
dx x=xo
For a droplet to move, this force needs to exceed the force arising from contact angle
hysteresis. There are a range of possible models relating contact angle hysteresis to
the Cassie fraction including one based on the Cassie–Baxter formula [38] and a
more fundamental defect based model due to Joanny and de Gennes [39, 40]. In the
defect based model for contact angle hysteresis, the pinning force per unit length
is given by the combination −γLV f (x) loge f (x) and so it scales with the density
of defects multiplied by a logarithmic correction. Using this with equation (17) and
rearranging, we find the following condition for the gradient in Cassie fraction for
motion,
√
df − f (xo ) loge f (xo )
> √ . (18)
dx x=xo 2 2R(1 + cos θe )3/2
This derivation is far from sophisticated and does not take into account shape
changes around the entire contact perimeter, which can be expected to change
overall constants, but it does provide an attempt to understand a number of fac-
tors preventing or initiating motion. For example, if the average Cassie fraction,
f (xo ), is made smaller so that the droplet is in a stronger superhydrophobic state,
the gradient in Cassie fraction needed to initiate motion can be smaller. Similarly,
a lower gradient is needed when the spherical radius, R, is larger. Larger volume
droplets roll more easily, not because of gravity, but because the contact radius, rc ,
is larger. Increasing the Young’s law contact angle also reduces the need for larger
gradients in the Cassie fraction.
6. An Example Gradient Surface

Considerations in the previous sections suggest self-actuated motion and definition
of paths should be possible simply by varying the superhydrophobicity through
topographic control and without changing surface chemistry. In this section we
provide one simple experimental example of such a surface. If a widely spaced
superhydrophobic surface texture is surrounded by a more narrowly spaced super-
hydrophobic surface texture a drop should experience a force so that it tends to roll
onto the area with the more closely spaced texture provided contact angle hysteresis
can be overcome (Fig. 6). The variation in lateral spacing will lead to a patterning of
the effective surface free energy and hence can be used to define regions and paths
on the surface [1]. A similar approach could be used with the Wenzel equation,
but here we focus only on the Cassie–Baxter situation. To investigate this exper-
imentally we modified a previously reported electrodeposition method that uses
diffusion limited aggregation to create a fractally rough superhydrophobic copper
surface because it can produce surfaces having an exceptionally low contact angle
hysteresis [41]; stabilising even very small droplets on these surfaces is difficult.
We also investigated square post based surfaces, but were unable to achieve self-
actuated motion of droplets.
Our copper-based surface was produced using a mechanical cantilever device
with two small DC motors to rotate and elevate the substrate whilst it was half im-
mersed in a copper electroplating solution (as described in Ref. [41]). Samples were
produced using different combinations of starting substrate material, anode mater-
ial, power supply, rotation speed and elevation rate. This approach produced copper
plates with a radial gradient in superhydrophobicity (θ ∼ 115◦ in the centre to sig-
nificantly higher than 160◦ at the edge) so that drops would roll to the centre and
pool, thus providing a water collection plate. This approach was highly success-
ful with small droplets released from a hydrophobised needle of a microsyringe at
the edge of the plate, always starting a self-initiated roll to the centre (Fig. 7 is an
image sequence showing a small droplet of water being deposited and rolling to
the centre). When droplets were released so that they skirted the central part of the
plate, they underwent several transits back and forth until they came to rest at the
centre of the plate. We characterised our surfaces by contact profilometry and SEM
imaging and established that the change in vertical height of the surface from edge
Figure 7. Self-initiated rolling of a water droplet on a copper surface possessing a gradient in super-
hydrophobicity. There is no significant overall change in average substrate height from the edge to the
centre of the copper plate.
Figure 8. (a) Surface topography profile measured along different radial lines, (b) variations of contact
angle hysteresis (y-axis) across sample from centre to edge (x-axis) estimated using radial view and
tilting of table tangential to radial direction.
to centre (a 2 cm distance) was below 25 µm (Fig. 8a). The contact angle hysteresis
was difficult to quantify across the sample, but to provide an estimate we tilted the
surface radially such that a droplet could be stabilised at locations radially from the
centre and we then measured the advancing and receding angles using a side profile
view tangential to the radial direction (contact angle hysteresis is shown in Fig. 8b).
We also estimated the tilt angle needed to prevent a roll and this was less than 1◦ at
the edge of the plate.
7. Conclusion
In this work we have emphasized that the Wenzel roughness parameter and the
Cassie solid surface fractions are quantities local to the three-phase contact line and
are not global parameters of the substrate or even of the substrate beneath a droplet.
We have shown how the surface free energy derivations that underpin this view can
provide transformation formula for multiple length scales and re-entrant surfaces
based upon the Wenzel and Cassie–Baxter formulae. We have considered how the
local view of the Cassie fraction can provide a condition for self-actuated droplet
motion and have shown that a simple surface with a gradient in superhydrophobicity
can initiate and direct droplet motion. More complex paths for use within droplet
microfluidics could be designed using these principles and similar ideas could be
used for droplets in Wenzel configurations and for hemi-wicking liquids.
Acknowledgements
The financial support of the UK Engineering & Physical Sciences Research Council
(EPSRC) and Dstl under grant GR/S34168/01 is gratefully acknowledged.
References
1. G. McHale, N. J. Shirtcliffe and M. I. Newton, Analyst 129, 284–287 (2004).
2. G. McHale., N. J. Shirtcliffe, S. Aqil, C. C. Perry and M. I. Newton, Phys. Rev. Lett. 93, Art. 036102
(2004).
3. N. J. Shirtcliffe, G. McHale, M. I. Newton, G. Chabrol and C. C. Perry, Adv. Mater. 16, 1929–1932
(2004).
5. D. Quéré, Physica A 313, 32–46 (2002).
6. X. J. Feng and L. Jiang, Adv. Mater. 18, 3063–3078 (2006).
7. P. Roach, N. J. Shirtcliffe and M. I. Newton, Soft Matter 4, 224–240 (2008).
9. S. H. Kim, J. Adhesion Sci. Technol. 22, 235–250 (2008).
11. R. E. Johnson, Jr. and R. H. Dettre, in: Contact Angle, Wettability and Adhesion, Adv. Chemistry
Series No. 43, pp. 112–135. Amer. Chem. Soc., Washington, DC (1964).
13. R. N. Wenzel, J. Phys. Colloid Chem. 53, 1466–1467 (1949).
14. M. Callies and D. Quéré, Soft Matter 1, 55–61 (2005).
15. L. C. Gao and T. J. McCarthy, Langmuir 23, 3762–3765 (2007).
16. G. McHale, Langmuir 23, 8200–8205 (2007).
17. M. V. Panchagnula and S. Vedantam, Langmuir 23, 13242–13242 (2007).
18. R. B. Fair, Microfluidics Nanofluidics 3, 245–281 (2007).
19. S. Y. Teh, R. Lin, L. H. Hung and A. P. Lee, Lab on a Chip 8, 198–220 (2008).
20. D. Li and A. W. Neumann, Colloids Surfaces 43, 195–206 (1990).
21. D. Q. Li, Colloids Surfaces A 116, 1–23 (1996).
22. M. A. Rodriguez-Valverde, F. J. M. Ruiz-Cabello and M. A. Cabrerizo-Vilchez, Adv. Colloid
Surface Sci. 138, 84–100 (2008).
25. P. G. de Gennes, Rev. Mod. Phys. 57, 827–863 (1985).
26. L. C. Gao and T. J. McCarthy, Langmuir 24, 9183–9188 (2008).
27. G. McHale, M. I. Newton and N. J. Shirtcliffe, Eur. J. Soil Sci. 56, 445–452 (2005).
28. J. Bachmann and G. McHale, Eur. J. Soil Sci. In press (2009).
29. N. J. Shirtcliffe, G. McHale, M. I. Newton, F. B. Pyatt and S. H. Doerr, Appl. Phys. Lett. 89,
Art. 094101 (2006).
30. A. Tuteja, W. Choi, M. L. Ma, J. M. Mabry, S. A. Mazzella, G. C. Rutledge, G. H. McKinley and
R. E. Cohen, Science 318, 1618–1622 (2007).
31. S. Daniel, S. Sanjoy, J. Gliem and M. K. Chaudhury, Langmuir 20, 4085–4092 (2004).
32. G. McHale, S. J. Elliott, D. L. Herbertson, N. J. Shirtcliffe and M. I. Newton, Passive and active
actuation of droplet motion, Paper presented at EU COST Action P21 meeting, March 29th 2007,
University of Granada, Spain.
33. A. Shastry, M. J. Case and K. F. Böhringer, Langmuir 22, 6161–6167 (2006).
34. J. Zhang and Y. Han, Langmuir 23, 6136–6141 (2007).
35. J. Zhang and Y. Han, Langmuir 24, 796–801 (2008).
36. C. Sun, X.-W. Zhao, Y.-H. Han and Z.-S. Ge, Thin Solid Films 516, 4059–4063 (2008).
37. M. Reyssat and D. Quéré, On “fakir” Drops, Paper presented at the Fifth International Symposium
on Contact Angle, Wettability and Adhesion, June 21st–23rd 2006, Toronto, Canada.
38. G. McHale, N. J. Shirtcliffe and M. I. Newton, Langmuir 20, 10146–10149 (2004).
39. J. F. Joanny and P. G. de Gennes, J. Chem. Phys. 81, 552–562 (1984).
40. J. F. Joanny and P. G. de Gennes, C. R. Acad. Sci. Paris Ser. II 299, 279–283 (1984).
41. N. J. Shirtcliffe, G. McHale, M. I. Newton, G. Chabrol and C. C. Perry, Langmuir 21, 937–943
(2005).
Design of Superhydrophobic Paper/Cellulose Surfaces via
Plasma Enhanced Etching and Deposition
Balamurali Balu, Jong Suk Kim, Victor Breedveld and Dennis W. Hess ∗
School of Chemical and Biomolecular Engineering, Georgia Institute of Technology,
311 Ferst Drive, Atlanta, GA 30332-0100, USA
Abstract
Superhydrophobicity has been achieved on different paper surfaces via plasma enhanced etching and film
deposition. The effects of fiber types and paper making parameters on the superhydrophobic behavior were
studied. Achievement of superhydrophobic behavior depends on the formation of nano-scale features on
the paper fibers established by selective etching of the amorphous domains in cellulose. Despite different
fiber types and paper making processes, superhydrophobicity can be attained provided that plasma etching
can occur on the fiber surface to create nano-scale features. Plasma processing conditions that allow the
design of superhydrophobic paper or cellulose surfaces with specific adhesion properties are described. The
significance of water drop volume on contact angle measurements and thus on characterization and analysis
of superhydrophobic behavior of heterogeneous, porous paper substrates is discussed as well.
Keywords
Superhydrophobic, paper, cellulose, plasma, fibers, contact line
1. Introduction
Cellulose is an inexpensive biopolymer which is abundantly present in nature.
Nevertheless, its inherent hydrophilic nature restricts direct use in a number of in-
dustrial applications such as printing, packaging and construction [1]. Hence, the
hydrophilic fibers are often treated to make them hydrophobic via a process com-
monly referred to as “sizing” [2, 3]. For more than two hundred years, rosin (resin
obtained from pine trees) based internal sizing agents have been added to pulp slur-
ries to yield hydrophobic paper surfaces [3]. In recent decades, there have been
significant developments in this field through advances in synthetic and polymer-
based sizing agents [1, 2, 4–6]. An alternative approach for the internal sizing is
external sizing where only the surface is coated with the sizing chemicals. Un-
til recently, these methods (both external and internal) had generated hydrophobic
*To whom correspondence should be addressed. Tel.: (404) 894-5922; Fax: (404) 894-2866; e-mail:
dennis.hess@chbe.gatech.edu

236 B. Balu et al.
surfaces [7–10], but were not able to impart superhydrophobic (advancing contact
angle (CA) > 150◦ ) characteristics to surfaces that are necessary to display extreme
water repellency.
Superhydrophobicity can be achieved only by a unique combination of low
surface energy and surface roughness. We recently reported the fabrication of su-
perhydrophobic paper surfaces (advancing CA ∼ 166.7◦ ± 0.9◦ ) using an external
sizing method — surface modification via plasma processing [11]. Various other
methods for the fabrication of superhydrophobic paper surfaces have also been re-
ported recently [12–14]. However, these methods used a solvent based approach in
at least one of the steps of fabrication. On the other hand, the plasma processing
method that we have demonstrated is a vapor phase, solvent free, external sizing
method. Moreover, the films formed using plasma deposition are mechanically ro-
bust because they are covalently bonded to the fibers. Indeed, adhesion between
the coating and fibers is stronger than the adhesion between the fibers [11]. It
was shown that superhydrophobic behavior on the paper or cellulose substrates re-
sulted from the combination of a low surface energy fluorocarbon film deposited by
plasma polymerization over cellulose fibers and roughness of these fibers on two
separate length scales, i.e. on the nano- and micro-scale. While paper substrates
have inherent micro-scale roughness as a result of the highly cross-linked web of
cellulosic fibers, the nano-scale roughness was created by uncovering the crystalline
domains on the cellulose fibers via oxygen plasma etching [11].
For superhydrophobic surfaces, the advancing contact angle (CA) defines the
shape of a static drop, while the contact angle hysteresis (difference between the
advancing and receding CA) defines the strength of adhesion of the drop. Hence,
a superhydrophobic surface may yield different levels of adhesion for water drops
(low water repellency to extreme water repellency) depending on the CA hysteresis
values [15–21]. Clearly, an understanding of this fundamental aspect of superhy-
drophobicity is critical because it is related directly to the dynamics of water drops
in contact with such substrates. Superhydrophobicity can be classified into two
categories depending on the CA hysteresis values [11, 22]: (1) “roll-off” super-
hydrophobicity (advancing CA > 150◦ , hysteresis < 10◦ ) and (2) “sticky” super-
hydrophobicity (advancing CA > 150◦ , hysteresis > 10◦ ). As the names suggest,
even for superhydrophobic surfaces, water drops can either roll off or stick to the
substrate, depending on the CA hysteresis.
We have reported the fabrication of both “roll-off” and “sticky” superhydropho-
bic paper surfaces using plasma etching and deposition [11]. Recently, we described
a methodology to tune the hysteresis (adhesion) between the two extreme behaviors
and control the wetting mechanisms responsible for these behaviors [22]. The tun-
ability in hysteresis was obtained by controlled formation of nano-scale features via
selective plasma etching of the cellulose fiber surfaces. During paper manufactur-
ing, the choice of fiber source and process conditions are chosen to meet the desired,
application-specific properties. Based on our previous work, it is reasonable to ex-
pect that different types of paper, with different fiber sources and fiber treatments,
Superhydrophobic Paper/Cellulose Surfaces 237
might display significant differences in micro-scale roughness (defined by the fiber

web structure) and in the evolution of nano-scale roughness due to etching. Both
of these factors are critical in determining their superhydrophobic behavior. In this
contribution, we, therefore, investigate the impact of variations in fiber type and
paper making process on the superhydrophobic properties of paper surfaces. The
experiments also provide insight into the appropriate window of plasma processing
conditions that enable design and fabrication of superhydrophobic paper surfaces
for different applications. Finally, we discuss the effect of water drop size on contact
angle and contact line geometry as observed in superhydrophobicity measurements
on heterogeneous porous substrates such as paper.
2. Experimental
2.1. Paper Substrates
Five types of paper substrates were used for superhydrophobicity studies as de-
scribed in Table 1. Handsheets (H, S, HS) were fabricated following the TAPPI
method T205 sp-02 with southern hardwood kraft (Alabama River Pulp Co., Per-
due Hill, AL) and/or southern softwood kraft (International Paper Co., Riegelwood,
NC). Both of the fiber types (hardwood and softwood) were refined to the same level
for a freeness value of ∼500 prior to the paper forming process. Commercial copy
paper substrates, “Premium white copy paper”, were obtained from local Office De-
pot. Commercial paper towels, SCOTT® High Capacity Hard Roll Towels (product
code 01000) were manufactured by Kimberly-Clark.
2.2. Plasma Etching/Deposition
A 6-inch parallel plate rf (13.56 MHz) plasma reactor was used for plasma etching
and deposition sequences; substrates were heated to 110◦ C. Details of the reactor
configuration and operational procedures for the treatment of paper substrates can
be found elsewhere [11]. Experimental conditions for oxygen etching and fluoro-
carbon (pentafluoroethane (PFE)) film deposition are listed in Table 2.
Table 1.
Paper substrates used for superhydrophobicity studies
Substrate designation Description
H Handsheet (100% hardwood fibers)

S Handsheet (100% softwood fibers)
HS Handsheet (50% hardwood–50% softwood)
CP Copy paper
PT Paper towel
238 B. Balu et al.
Table 2.
Plasma reactor parameters for etching and deposition steps
Parameters Etching Deposition
Gas Oxygen Pentafluoroethane (PFE) and Argon (carrier gas)

Flow rate 75 sccm 20 sccm (PFE) and 75 sccm (Argon)
Pressure 0.55 Torr 1 Torr
Power 10 W 120 W
Treatment time 0–60 min 2 min
2.3. SEM Investigation

SEM micrographs were obtained using a LEO scanning electron microscope
(model 1530) at an acceleration voltage of 5 kV or 10 kV depending on the damage
induced by the electron beam on the paper surfaces. Prior to SEM studies, paper
substrates were sputter coated (EMS 350; Electron Microscopy Sciences, Hatfield,
PA) with a thin film of gold (∼15 nm).
2.4. Water Contact Angle Measurements
For standard contact angle measurements, a 4 µl water drop was used. Advanc-
ing and receding contact angles were measured by moving the substrate left to right
with respect to the drop, following the methodology discussed by Gaudin et al. [23].
This method yields improved statistically averaged CA values relative to measur-
ing CAs at individual locations independently, since a larger area of the substrate is
scanned during measurement. However, for sticky substrates with receding CA less
than 10◦ , this method could not be used because the drop breaks before the receding
CA is attained; this limitation has been discussed in detail [23]. Our previous work
[11] reported the receding CA observed during drop breakup which is not the true
receding CA. To overcome this limitation inherent in Gaudin’s method, we used the
standard “volume decrement” method for sticky superhydrophobic substrates for
which drop breakup was observed. Further details regardingthe above contact angle
measurement methods can be found elsewhere [22]. For our sticky superhydropho-
bic substrates, measurement of the advancing CA is also complicated. Interaction
of the water drop with a “sticky” superhydrophobic paper surface is described by
the Wenzel regime on a micrometer scale. The Wenzel regime is characterized by a
very high hysteresis and therefore many closely placed metastable states [24]. This
means that a range of advancing contact angles are possible depending on the force
used to press the drop against the substrate. Since no quantitative measure of force
was possible with the goniometer used, a slight variability in the advancing CA
values was observed in the current studies relative to the CA values published previ-
ously [11]. For example, the advancing CA of sticky superhydrophobic HS reported
in our Langmuir article [11] and this manuscript are 140◦ ± 1.7◦ and 159.4◦ ± 7.7◦ ,
respectively.
2.5. Microscopic Imaging of the Contact Line

Drops with appropriate volumes were dispensed onto sticky superhydrophobic pa-
per (HS) surfaces that had been attached to microscopic slides. The existence of a
superhydrophobic contact angle made it impossible to obtain a clear microscopic
image of three-phase contact lines due to the lensing effect of the drops. Hence, the
water drops were allowed to evaporate until a hydrophilic contact angle was ob-
served and the contact line was then imaged with a Leica microscope (DM4500 B)
using a 10× objective.

3.1. Effects of Paper Making Parameters on Achievement of Superhydrophobicity
As discussed in the Introduction, the superhydrophobicity imparted to paper sub-
strates results from the combination of a low surface energy film and two-scale
roughness (nano-scale and micro-scale). The nano-scale roughness originates from
the protrusion of crystalline domains on fiber surfaces after removal of the sur-
rounding amorphous domains via selective plasma etching [11, 22]. On the other
hand, the micro-scale roughness is determined by the topography of the paper fibers,
in particular the fiber size and mesh size of the cellulose web. In this study we ex-
plore two key paper making parameters that may affect the micro- and nano-scale
roughness and thus the resulting superhydrophobicity of paper substrates: (1) fiber
source and (2) paper making technology.
3.1.1. Effects of Fiber Type
Cellulose paper is typically produced from hardwood fibers, softwood fibers, or a
combination of the two. This classification of cellulose fibers is based on the trees
from which they are obtained: hardwood fibers come from angiosperm trees (e.g.,
American yew, Common juniper, Douglas fir), and softwood fibers originate from
gymnosperm trees (e.g., wild plum, peach, pear) [25, 26]. Both fiber types have
approximately the same chemical composition: cellulose (40–50%), hemicellulose
(25–35%) and lignin (20–35%), but there is a significant difference in physical di-
mensions [25]. Softwood fibers are usually larger than hardwood fibers roughly by
a factor of two as shown in Table 3. Considering these facts, we expect that: (1) dif-
ferent cellulose fiber types would show differences with regards to the evolution of
nano-scale roughness during etching (exposure of crystalline domains) and (2) the
different fiber sizes will impact differently the micro-scale roughness of the paper
Table 3.
Typical dimensions of hardwood and softwood fibers [25]
Fiber type Fiber length, mm Fiber width, µm
Hardwood 1.0–1.5 16–22

Softwood 3.0–3.7 27–38
240 B. Balu et al.
surfaces. We have shown previously that both length scales contribute to superhy-
drophobicity [11]. In order to investigate the role of fiber type in more detail, we
fabricated handsheets from three different combinations of hardwood and softwood:
100% hardwood (H), 100% softwood (S) and 50% hardwood–50% softwood (HS).
Other than the origin of the fibers, all procedures for handsheet fabrication were the
same.
Figure 1a–c show high and low magnification SEM images of untreated hand-
sheets for 100% softwood, 100% hardwood and a 50–50% hardwood/softwood
mixture, respectively. The larger size of softwood fibers in comparison with hard-
wood fibers is confirmed by the SEM images of Fig. 1a–c. In addition, it was
confirmed by SEM images (not shown) that a thin film of PFE (∼100 nm) deposited
on unetched handsheets did not alter the roughness (either micro- or nano-scale)
of the handsheet surface (H, S, HS). This PFE deposition without oxygen etching
yielded “sticky” superhydrophobic properties for all three handsheets with the fol-
lowing advancing and receding CAs: H (CAadv /CArec ) — 154.3◦ ±1.9◦ /12.5◦ ±5◦ ,
S (CAadv /CArec ) — 149◦ ± 2.5◦ /8.5◦ ± 5◦ and HS (CAadv /CArec ) — 159.4◦ ±
7.7◦ /9.65◦ ± 5.8◦ . The fact that the advancing and receding CAs are similar for all
three handsheets confirms that the differences in micro- and nano-scale roughnesses
due to variations in fiber types do not significantly affect the “sticky” superhy-
drophobic behavior.
In subsequent experiments, handsheets were etched in an oxygen plasma for
different durations before depositing the PFE film. Figure 2 displays plots of ad-
vancing and receding CAs for the different handsheets as a function of oxygen
etching time. The figure shows the transition from “sticky” to “roll-off” superhy-
drophobicity (contact angle hysteresis < 10◦ ) after ∼30 minutes of etching for all
substrates. The curves in Fig. 2 overlap, showing that the rate of change of advanc-
ing and receding CAs, which is closely connected to the evolution of nano-scale
features, was indistinguishable for the three handsheets within experimental error.
Indeed, there were no noticeable differences between the sizes of the nano-scale
features formed on etched hardwood and softwood fibers (SEM images not shown).
These results provide evidence that there is no significant difference between the
nano-scale features formed on fibers of different types. In conclusion, different fiber
type does not affect the superhydrophobic behavior provided that the paper making
procedures are constant.
3.1.2. Effects of Paper Making
The pulping process and the paper machine configuration vary from mill to mill
in order to optimize paper properties for specific applications [25]. The process in-
volves the following steps: after wood chips are pulped and bleached, the paper web
is formed in the paper machine, after which it undergoes a variety of mechanical
treatments (pressing, drying and calendering) before being collected on a large roll
[27]. All these steps of the paper making process ultimately affect the roughness of
the final paper surface. Each paper mill uses a unique set of paper making proce-
dures and sequences depending on the application of the final paper product. Our
Figure 1. High (left) and low (right) magnification SEM images of laboratory handsheets made with
(a) 100% hardwood (H), (b) 100% softwood (S), (c) 50–50% hardwood and softwood (HS), and two
commercial paper samples, (d) copy paper (CP) and (e) paper towel (PT). Scale bars correspond to
40 µm (high magnification) and 400 µm (low magnification).
focus in this study does not involve a comprehensive investigation of the large num-
ber of parameters invoked in paper making and their effect on superhydrophobicity.
Rather, we have selected two different paper types (apart from the laboratory-made
handsheets) that were fabricated for unique and distinct applications: (1) a commer-
cial copy paper (CP) which is moderately hydrophobic to yield good printability
242 B. Balu et al.
Figure 2. Plots of advancing CA and receding CA of handsheets (H, S, HS) with respect to oxygen
plasma etching time for 2 min PFE deposition (∼100 nm).
and (2) a paper towel (PT) which is extremely hydrophilic to provide high absorp-
tivity. The copy paper and paper towel also represent two extremes of porosity and
hence micro-scale roughness. Finally, the copy paper contains a significant amount
of filler particles which are of similar size to the nano-scale features formed during
oxygen plasma etching. Our intent in this part of the work is to explore the superhy-
drophobic properties of copy paper and paper towels in order to obtain insight into
the effect of paper-making parameters on superhydrophobicity.
The SEM images in Fig. 1d and e show high and low magnification SEM images
of untreated CP and PT, respectively. Of these two samples, the copy paper is most
similar to the handsheets; the main difference is the presence of (inorganic) filler
particles on the fiber surface (shown in Fig. 1d). The paper towels have a noticeably
more porous surface with very loosely cross-linked fibers, since these substrates
are designed for superior absorption properties. From the SEM images it is evident
that these substrates have very different surface roughness values prior to plasma
treatment.
The untreated copy paper displayed an advancing CA ∼ 79.15◦ ± 3.37◦ , which
confirmed its moderately hydrophobic behavior. For the untreated paper towel, the
water drop was absorbed into the paper within one second; therefore CA values
could not be measured. After deposition of a thin film of PFE (without oxygen
etching), the CP and PT substrates yielded different superhydrophobic behavior
than that of the HS substrate as shown in Fig. 3. The difference in receding CA
values between the samples can be attributed to differences in the micro- and nano-
scale roughness that result from the distinct processing conditions in the paper mills
(evident from the SEM images in Fig. 1). The advancing and receding CA values
for CP, which is most similar to the HS handsheet with regard to fiber composi-
tion, are analogous to the values obtained for HS. However, the PT showed a very
different receding CA relative to those for HS and CP. The increased values of the
Figure 3. Plots of advancing and receding CAs of handsheet (HS), copy paper (CP), paper towel-top
side (PT-top) and paper towel-bottom side (PT-bottom), after 2 min PFE deposition (∼100 nm) and
without oxygen etching.
receding CA (decreased CA hysteresis) of the PT can be attributed to the increased

micro-scale roughness resulting from the increased porosity of this substrate. In ad-
dition, the PT showed different superhydrophobic behaviors on the two sides of the
substrate (labeled PT-top and PT-bottom in Fig. 3). Although the SEM images did
not reveal a significant difference between the two sides, we believe that the distinct
CA values are due to the different roughness scales generated on the felt side and
wire side of the paper during the manufacturing process, usually referred to as “two
sidedness of paper” [10, 25]. The copy paper did not show a difference in super-
hydrophobic behavior between the top and bottom sides, which is expected since
the applications of copy paper require that it has the same physical and chemical
properties on both sides.
The paper substrates (CP, PT-top and PT-bottom) were subsequently etched in an
oxygen plasma environment for different durations prior to PFE deposition. The ad-
vancing and receding CAs of these substrates with respect to oxygen etching times
are shown in Fig. 4. It is evident from Fig. 4 that “roll-off” superhydrophobic be-
havior could be obtained for all samples tested, in spite of significant differences
in paper making methods. Indeed, the nano-scale roughness established by oxy-
gen etching, which is responsible for the “roll-off” superhydrophobic behavior, was
similar for all papers (SEM images not shown).
In conclusion, the difference in CA hysteresis between various paper samples
(Fig. 3) results in differences in the adhesion of water drops on these substrates. This
demonstrates that by control of the paper making processes, adhesion of water drops
on a superhydrophobic paper surface can be tuned. Also, after the paper substrates
are etched, the formation of nano-scale roughness dominates the superhydrophobic
behavior, thereby leading to more similar wettability for all tested paper substrates.
Although these experiments do not represent a comprehensive study of the array
244 B. Balu et al.
(a)
(b)
Figure 4. Plots of advancing CA and receding CA of copy paper (CP) (a) and paper towel (PT-top and
PT-bottom) (b) with respect to oxygen plasma etching time for 2 min PFE deposition.
of paper making parameters, they do provide a general picture of the effects of

these parameters on superhydrophobicity as established by our plasma treatment
process. Furthermore, we conclude that, provided the fibers can be etched to cre-
ate nano-scale features, superhydrophobicity can be imparted on any paper surface
irrespective of the fiber origin or paper making technique.
3.2. Design of Superhydrophobic Paper Surfaces by Optimizing Fiber Type and
Plasma Processing Conditions
Longer softwood fibers are usually responsible for paper strength, while shorter
hardwood fibers are predominantly responsible for the paper shininess because
of reduced roughness. Our experiments indicated that oxygen etching ultimately
reduces the shininess of the paper by creating nano-scale roughness, so that the
presence of hardwood fibers no longer provides enhanced optical properties in the
etched handsheets. Therefore, we believe that the fabrication of superhydrophobic
paper based on softwood fibers is the most desirable approach because of the ex-
pected increased physical strength.
For longer PFE deposition times than are shown in the preceding figures
(e.g., 15 minutes), roll-off superhydrophobic behavior can only be achieved after
60 minutes etching (data not shown), while for 2 min PFE deposition times roll-off
is observed after much shorter etching times (∼30 minutes) [22]. These results are
due to smoothing of the topography of the roughened surface that occurs during
the deposition of a thicker PFE film. Since prolonged oxygen etching damages the
fiber surfaces, it has a significant negative impact on the strength of the paper, which
is undesirable. Therefore, it is most desirable to obtain roll-off superhydrophobic-
ity at reduced etch times. We expect that an optimum PFE thickness exists: thick
enough to retard the absorption of water, yet thin enough to prevent smoothing of
the morphology created by oxygen etching. Our results suggest that a ∼100 nm
film obtained from a 2 min PFE deposition is a near-optimum thickness to achieve
a roll-off superhydrophobic paper surface with good physical properties. Of course,
optimizing the paper making process to tune the micro-scale roughness may offer
an additional degree of freedom for the design of superhydrophobic paper surfaces.
3.3. Effect of Drop Size on Contact Angle and Edge Geometry
Measurement of contact angles on rough surfaces is more complex when the drop
size is comparable to the roughness length scale of the substrate. Advances in drop
dispense technologies have made it possible to vary the dispensed volume of a water
drop in a controlled manner from a few picoliters to a few microliters. Here we
present results that describe the significance of the water drop size when measuring
CAs on paper surfaces.
Figure 5 shows the contact line established by water drops of four different vol-
umes on a HS handsheet with CA characteristics shown in Fig. 3. Solid lines were
drawn along the three-phase contact lines to highlight the contact line geometry.
Clearly, the contact line is more distorted by the topography of the fiber network
for smaller drops (Fig. 5a and b) than for larger drops (Fig. 5c and d).
In order to determine the effect of contact line distortion on the measurement of
CA, we varied the drop volume from 0.1 µl to 16 µl, and measured the advancing
CA values. Figure 6 shows the advancing CA with respect to drop volume for HS
substrates etched for three different etching durations. Figure 7 shows the images
of water drops corresponding to a 30 min etched HS (“roll-off” superhydrophobic)
and 0 min etched HS (“sticky” superhydrophobic). From Fig. 6 it can be concluded
that the advancing CA increases up to a volume of ∼2 µl. This suggests a lower
limit of drop volume that should be used to measure CA on superhydrophobic paper
surfaces. On the other hand, the upper limit depends on the angular resolution of
the CA goniometer. As the drop volume increases, the drop flattens due to gravity,
which makes it difficult to locate exactly the three-phase contact point from a side
view of the drop. This limitation can greatly affect the accuracy of CA values, which
is consistent with the observation of a slight decrease in the advancing CA values for
246 B. Balu et al.
Figure 5. Contact lines formed by 0.1 µl (a), 0.2 µl (b), 4 µl (c) and 8 µl (d) water drops on a 2 min
PFE deposited (without etching) HS substrate. Scale bars correspond to 160 µm.
Figure 6. Plots of advancing CA with respect to drop volume for oxygen etched (0, 10 and 30 min)
and PFE deposited (2 min) handsheet (HS) surfaces.
the 16 µl drop (Fig. 6). Hence, we suggest that in order to mitigate ambiguity in CA
values when measuring contact angles on porous, heterogeneous substrates such as
paper, it is important to select drop sizes that (1) are larger than the length scale
Figure 7. Photographs of advancing CA for different drop volumes for “roll-off” superhydrophobic
(0 min oxygen etched and 2 min PFE deposited) and “sticky” superhydrophobic (30 min oxygen
etched and 2 min PFE deposited) handsheet (HS) surfaces.
248 B. Balu et al.
of fibers (to avoid the distortion of the contact line by fiber web) and (2) provide
sufficient image resolution for the goniometer to identify the three-phase contact
line.
4. Conclusions
The effects of fiber type and paper making parameters on the creation of super-
hydrophobic paper surfaces were studied. The different fiber types and the paper
making techniques do not affect the superhydrophobicity provided that the fibers
can be etched to create the necessary nano-scale surface features. Paper made from
softwood fibers is likely to be more suitable for superhydrophobic applications
because of improved physical properties with this fiber type, in particular paper
strength. A PFE film of ∼100 nm represents a near-optimum thickness to obtain
superhydrophobicity. The importance of water drop volume in the measurement of
CAs on superhydrophobic surfaces fabricated on heterogeneous and porous sub-
strates such as paper has been discussed.
Acknowledgements
The authors thank Dr. Ashwini Sinha (Praxair) for kindly donating the PFE gas,
Yonghao Xiu (Georgia Tech) for support with CA measurements. B. B. thanks the
Institute of Paper Science and Technology (IPST) at Georgia Tech for fellowship
support.
References
1. N. Yang and Y. L. Deng, J. Appl. Polym. Sci. 77, 2067–2073 (2000).
2. Y. Ishida, H. Ohtani, S. Tsuge and T. Yano, Anal. Chem. 66, 1444–1447 (1994).
3. F. Wang and H. Tanaka, J. Appl. Polym. Sci. 78, 1805–1810 (2000).
4. K. Asakura, M. Iwamoto and A. Isogai, J. Wood Chem. Technol. 25, 13–26 (2005).
5. M. J. Lindstrom and R. M. Savolainen, J. Dispersion Sci. Technol. 17, 281–306 (1996).
6. T. Yano, H. Ohtani, S. Tsuge and T. Obokata, Analyst 117, 849–852 (1992).
7. S. M. Mukhopadhyay, P. Joshi, S. Datta, J. G. Zhao and P. France, J. Phys. D — Appl. Phys. 35,
1927–1933 (2002).
8. S. Vaswani, J. Koskinen and D. W. Hess, Surf. Coat. Technol. 195, 121–129 (2005).
9. A. G. Cunha, C. S. R. Freire, A. J. D. Silvestre, C. P. Neto, A. Gandini, E. Orblin and P. Fardim,
Biomacromolecules 8, 1347–1352 (2007).
10. H. T. Sahin, S. Manolache, R. A. Young and F. Denes, Cellulose 9, 171–181 (2002).
11. B. Balu, V. Breedveld and D. W. Hess, Langmuir 24, 4785–4790 (2008).
12. H. Yang and Y. Deng, J. Colloid Interface Sci. 325, 588–593 (2008).
13. D. Nystrom, J. Lindqvist, E. Ostmark, A. Hult and E. Malmstrom, Chem. Commun., 3594–3596
(2006).
14. S. H. Li, S. B. Zhang and X. H. Wang, Langmuir 24, 5585–5590 (2008).
15. Y. Yao, X. Dong, S. Hong, H. Ge and C. C. Han, Macromol. Rapid Commun. 27, 1627–1631
(2006).
16. A. Winkleman, G. Gotesman, A. Yoffe and R. Naaman, Nano Lett. 8, 1241–1245 (2008).
17. D. Quere, M. J. Azzopardi and L. Delattre, Langmuir 14, 2213–2216 (1998).
18. K. S. Liao, A. Wan, J. D. Batteas and D. E. Bergbreiter, Langmuir 24, 4245–4253 (2008).
19. Y. B. Li, M. J. Zheng, L. Ma, M. Zhong and W. Z. Shen, Inorg. Chem. 47, 3140–3143 (2008).
20. R. Di Mundo, F. Palumbo and R. d’Agostino, Langmuir 24, 5044–5051 (2008).
21. S. Boduroglu, M. Cetinkaya, W. J. Dressick, A. Singh and M. C. Demirel, Langmuir 23, 11391–
11395 (2007).
22. B. Balu, J. S. Kim, V. Breedveld and D. W. Hess, J. Adhesion Sci. Technol. 23, 361–380 (2009).
23. A. M. Gaudin, A. F. Witt and T. G. Decker, Trans. of the Society of Mining Engineers of AIME
226, 107–112 (1963).
24. A. M. Schwartz and F. W. Minor, J. Colloid Sci. 14, 584–597 (1959).
25. H. Goyal, http://www.paperonweb.com (2007).
26. G. Fewless, http://www.uwgb.edu/BIODIVERSITY (2006).
27. A. W. C. Company, http://www.chesterton.com/industries/process.asp?industry=24&process=51
(2008).
Superhydrophobic Aluminum Surfaces: Preparation Routes,
Properties and Artificial Weathering Impact
M. Thieme a,∗ , C. Blank a , A. Pereira de Oliveira b , H. Worch a , R. Frenzel c ,

S. Höhne c , F. Simon c , H. G. Pryce Lewis d and A. J. White d
a
Technische Universität Dresden (TUD), Institute of Materials Science, D-01062 Dresden, Germany
b
TUD, now at: Universidade Estadual de Campinas UNICAMP, Faculdade de Engenharia Química,
Cidade Universitária “Zeferino Vaz”, Av. Albert Einstein 500, Campinas, SP, Brazil
c
Leibniz Institute of Polymer Research Dresden (IPF), Hohe Str. 6, D-01069 Dresden, Germany
d
GVD Corporation, 45 Spinelli Place, Cambridge, MA 02138, USA
Abstract
Among the materials that can be treated in order to impart superhydrophobic properties are many orig-
inally hydrophilic metals. For this, they must undergo a sequential treatment, including roughening and
hydrophobic coating. This contribution presents various preparation routes along with various characteriza-
tion methods, such as dynamic contact angle (DCA) measurements, scanning electron microscopy (SEM)
and spectroscopic techniques (FT–IRRAS, XPS, EIS).
Micro-rough surfaces of pure and alloyed aluminum were generated most easily by using a modified
Sulfuric Acid Anodization under Intensified conditions (SAAi). This produces a micro-mountain-like ox-
ide morphology with peak-to-valley heights of 2 µm and sub-µm roughness components. Additionally,
micro-embossed and micro-blasted surfaces were investigated. These micro-roughened initial states were
chemically modified with a solution of a hydrophobic compound, such as the reactive fluoroalkylsilane
PFATES, the reactive alkyl group containing polymer POMA, or the polymer Teflon® AF. Alternatively,
the chemical modification was made by a Hot Filament Chemical Vapor Deposition (HFCVD) of a PTFE
layer. The latter can form a considerably higher thickness than the wet-deposited coatings, without detri-
mental leveling effects being observed in comparison with the original micro-rough surface. The inherent
and controllable morphology of the PTFE layers represents an important feature. The impacts of a standard-
ized artificial weathering (WTH) on the wetting behavior and the surface-chemical properties were studied
and discussed in terms of possible damage mechanisms. A very high stability of the superhydrophobicity
was observed for the fluorinated wet-deposited PFATES and Teflon® AF coatings as well as for some of the
PTFE layer variants, all on SAAi-pretreated substrates. Very good results were also obtained for specimens
produced by appropriate mechanical roughening and PTFE coating.
*
To whom correspondence should be addressed. Tel.: 0049 351 463-36461; Fax: 0049 351 463-33207;
e-mail: michael.thieme@tu-dresden.de

252 M. Thieme et al.
Keywords
Aluminum, superhydrophobicity, anodic oxidation, micro-embossing, wet-deposited coatings, PTFE lay-
ers, artificial weathering
1. Introduction
The phenomenon of superhydrophobicity (SH, earlier designated as ultrahydropho-
bicity) has received much attention for more than a decade by numerous research
groups, extending the crucial and exciting investigations of Barthlott and Neinhuis
[1]. SH as defined by water contact angles (CA) of more than 150◦ , a negligible hys-
teresis (the difference of the advancing and receding CAs, θa and θr , respectively)
and extremely low droplet roll-off angles is based on the interplay of morpholog-
ical and surface-chemical properties, which lower the surface free energy down to
very small values. From the literature it can be seen that the micro-roughness has in
most cases more than one lateral/transversal component covering micro- and sub-
micro dimensions [2–4]. Some researchers postulated a fractal character [5]. With
regard to the structural properties of superhydrophobic surfaces, the spectrum cov-
ers layered structures with self-assembled monolayers (SAMs) of water-repellent
compounds to polymer layers of considerable thickness as well as intrinsically hy-
drophobic polymers with suitable morphology. It has been stated that the surface
free energies of molecular groups rank according to CH2 > CH3 > CF2 > CF3 [6].
Fluorine-substituted organic compounds are, therefore, generally preferred for im-
parting a surface with SH. Moreover, they are characterized by the exceptionally
high strength of the C–F bond as well as by high chemical and biological inert-
ness [7]. There are literature surveys that reflect the state-of-the-art of science and
technology in the field of SH in great details [8–10]. A survey of superhydrophobic
aluminum is given elsewhere [11]. Because Al with its native oxide has a hy-
drophilic character, it must undergo sequential roughening and coating treatment
steps in order to obtain superhydrophobic properties.
The far-reaching commercial possibilities of the so-called Lotus-effect® are, at
present, only being seldom utilized. This is, in part, caused by the use of other
advanced technologies for self-cleaning (as with photocatalytically acting superhy-
drophilic glass [12]). Moreover, a major reason for this is the fact that SH is inti-
mately associated with the uppermost surface layers. Thus, SH is generally sensitive
to mechanical impacts, e.g. from handling, as well as from the (photo-)chemical at-
tack, i.e. from weathering.
This contribution looks at novel preparation routes as well as the behavior of
superhydrophobic Al material after artificial weathering. The surfaces were pre-
pared by both micro-roughening and chemical modification. In addition to our
anodization approach using sulfuric acid [13, 14], we present the novel variants
of elevated-temperature micro-embossing and micro-blasting as purely mechani-
cal ways for achieving a suitably roughened surface (cf. [11]). Other roughening
variants developed [13, 15–17] are not considered here. For the subsequent coating
Superhydrophobic Aluminum Surfaces 253
step, two novel compounds of quite different nature were successfully applied for
imparting SH and tested for their weathering behavior: i — the reactive polymer
poly(octadecene-alt-maleic anhydride) (POMA), which can be grafted onto previ-
ously deposited chitosan (N-amino-2-desoxy-β-D-glycopyranose, Chs), which acts
as an anchor and, moreover, which was found to lead to strengthening effects of
the anodic oxide [11, 18, 23], and ii — poly(tetrafluoroethylene) (PTFE), which
was deposited by hot filament chemical vapor deposition (HFCVD) [19–21] with
various thicknesses and morphologies. Further, perfluoroalkylethyltriethoxysilane
(PFATES) and [3-(2-aminoethyl) aminopropyl] trimethoxysilane plus Teflon® AF
(AS/TAF) [13, 15] were compared to these layers.
Mechanical properties of the produced systems as obtained from micro- and
nano-hardness measurements as well as from gentle abrasion tests are reported in
Ref. [11].
2. Experimental
2.1. Materials and Preparation Methods
2.1.1. Substrate Materials

Sheets (26 × 38 × 1 mm3 ) of Al Mg1 (AA 5005) were the main substrate material,
with analytical grade Al 99.95 (Merck) and pure aluminum 99.5 (AA 1050) used
for comparison. Using the Brinell hardness test (ISO 6506-1:2005), the follow-
ing hardness values were determined: 52 HBW 2.5/31.25 for Al Mg1, 35 HBW
2.5/15.625 for Al 99.5. Rod specimens (5 mm in diameter) of Al MgSi0.5 (AA
6060) were used for EIS testing (see below).
2.1.2. Sulfuric Acid Anodization under Intensified Conditions, SAAi

The electrolyte solution was a mixture of sulfuric acid and aluminum sulfate (start-
ing concentration 0.1 mol/l) with a total sulfate concentration of 2.3 mol/l. The an-
odization was carried out at (40 ± 1)◦ C, 30 mA/cm2 for 1200 s [13, 14]. The con-
ditions of the usual anodization procedure (SAAu) were <20◦ C, 15 mA/cm2 , and
1200 s. The anodization followed an initial etching treatment using 1 mol/l NaOH
(600 s) with subsequent neutralizing in 1 mol/l HNO3 (60 s).
2.1.3. Intermediate Deposition of Chitosan

Chitosan was deposited either by cathodic precipitation from a 1 wt.% solution in
1 vol.% acetic acid, pH = 3.8 at −5 mA/cm2 for 40 s (Chs-e) or by immersion in
the same solution for 1800 s (Chs-i) [18].
2.1.4. Micro-embossing under Annealing Conditions, ME

The embossing operation was done at 350◦ C and 120 MPa using a laser-structured
SiC tool (stamp diameter 18 mm), having hexagonally arranged cone-shaped holes
(spaced at intervals of 23 µm and about 25 µm deep).
2.1.5. Micro-blasting, MB
Corundum powders (grit 600, 800, 1000, 1200) were applied as a fine beam at 6 bar
pressure under manual control.
2.1.6. Wet-Chemical Dip Coating Treatments
i — Poly(octadecene-alt-maleic anhydride) (POMA) was applied as a 0.1 wt.% so-
lution in acetone at ambient temperature for 30 min, followed by a vacuum-drying
step. ii — Perfluoroalkylethyltriethoxysilane (PFATES) was applied in tert-butyl
methyl ether (2 vol.%) for 180 min in the presence of 0.1 vol.% of triethylamine as
a catalyst; iii — N-(2-aminoethyl)-3-aminopropyltrimethoxysilane was applied in
a 10 vol.% ethanolic solution for 180 min, followed by spin-coating (5000 min−1 ,
30 s) with a solution of the copolymer Teflon® AF (1 wt.% in FC 75 solvent)
(AS/TAF). All steps were followed by rinsing and annealing at 120◦ C for 1 h.
2.1.7. HFCVD
The process consists of thermal decomposition (>150◦ C) of the precursor hexaflu-
oropropylene oxide, which is associated with radical formation, and polymerization
to PTFE. The deposition thicknesses used were 50, 250, 500 and 1000 nm. Besides
the standard coating conditions a number of other variants were employed, such as
post-deposition annealing, deposition at elevated substrate temperature, and lower
pressure conditions during deposition.
2.2. Artificial Weathering
The weathering procedure was carried out in two ways: (i) Normal weathering ex-
posure (WTH) for 360 h, comprising of continuous xenon-arc irradiation (filtered
corresponding to day-light spectral distribution at a black-standard temperature of
55◦ C) and a cyclic sequence of shower wetting (18 min) and drying at relative air
humidity of 60–80% (102 min), and (ii) special weathering exposure using irradia-
tion, but excluding moisture (WTH-L) (Xenotest Alpha; Atlas, Chicago, IL).
2.3. Characterization
2.3.1. Dynamic Contact Angle (DCA) Measurement
DCA measurements were made at five different positions on each specimen using
a maximum droplet size of 30–50 µl. The contact angle data were averaged from
about 20 successive measurements during advancing and receding (DSA 10, Krüss/
Germany).
2.3.2. Scanning Electron Microscopy (SEM)
The images were taken at an acceleration voltage of 2 keV in a DSM 982 Gemini
equipment (Zeiss/Germany).
2.3.3. Fourier-Transform Infrared Reflection-Absorption Spectroscopy
(FT–IRRAS)
The spectra were recorded using an FTS 2000 instrument (Perkin-Elmer/Germany)
over the frequency range of 550–4000 cm−1 as averages of 256 individual spectra
measured at four positions on each sample. The analyzed spot had a diameter of ca.
100 µm.
2.3.4. X-Ray Photoelectron Spectroscopy (XPS)
The analysis utilized monochromatic Al Kα radiation, charge compensation and
step widths of 0.3 eV for survey spectra, or 0.02 eV for high-resolution spectra
(Axis Ultra, Kratos/UK). The scale was calibrated using the C1s binding energy of
saturated hydrocarbons, which was set at 285 eV. The maximum information depth
for the C1s peak was about 10 nm [24–26].
2.3.5. Electrochemical Impedance Spectrometry (EIS)
Spectra were recorded in the frequency range of 100 kHz–0.5 mHz using a
0.133 mol/l phosphate buffer test solution pH = 6.0 and an IM 6 instrument (Zah-
ner/Germany). Sheet specimens were tested using an O-ring cell (effective area
0.25 cm2 ). Alternatively, rod specimens were used in a three-electrode cell arrange-
ment with a concentric platinum net counter electrode and a Haber–Luggin capil-
lary (cf. [22]). The depth of immersion was 40 mm. The measurements were made
at least twice for each of the selected sample states.
3. Results
3.1. Substrate Surfaces
Irrespective of the material employed, the SAAi pretreatment leads to a specific
morphology of the oxide layer produced which is characterized by an irregularly
ordered mountain-like structure showing typical top-to-valley and lateral distances
of about 2 µm each (Fig. 1a). This structure is produced more uniformly on the pure
Al as compared to the technical Al substrates. At higher magnifications a sub-µm
fibre-like roughness is also observed (Fig. 1b).
In contrast, surfaces treated by the usual SAAu method have a more or less flat,
rippled morphology (Fig. 1c). For details of layers formation, their structures and
compositions see Refs. [13, 14, 22].
Chitosan can be deposited onto anodized aluminum from a diluted acetic acid
medium by means of a cathodic process, which causes interfacial alkalization and,
Figure 1. SEM images of different anodized surfaces; (a) SAAi-treated Al 99.95; (b) SAAi-treated Al
Mg1; (c) SAAu-treated Al Mg1.
Figure 2. SEM images of specimen surfaces following a cathodic chitosan deposition on SAAi-treated
surfaces; (a) Al Mg1, SAAi + Chs-e; (b) Al99.5, SAAi + chitosan deposition at higher pH, for larger
current density and for longer duration.
Figure 3. EIS spectra for SAAi, SAAi + Chs-e and SAAi + Chs-i in phosphate buffer; rod specimens;
Bode plot: modulus of impedance, |Z| and phase angle, φ vs. frequency, f .
hence, deprotonation of the previously produced polycations of the type R–NH+ 3

[11, 18]. For optimized process conditions, the organic material is homogeneously
precipitated and practically cannot be seen in SEM micrographs (Fig. 2a). For too
high solution pH, current density and duration, inhomogeneous precipitation occurs
(Fig. 2b). Additionally, cone-like microscopic defects form, probably due to the
concurrent hydrogen formation and bubble expansion at the metal–oxide interface.
EIS measurements indicated that defects were present independent of the ac-
tual manner of cathodic chitosan deposition. As Fig. 3 shows for the optimized
Chs-e deposition conditions, the curve of the impedance modulus log |Z| vs. log f
is markedly shifted to lower values for f < 1 kHz in comparison with the origi-
nal anodized state SAAi. This observation is indicative of the formation of a more
porous oxide structure. On the contrary, specimens that were merely immersed in
chitosan solution (SAAi + Chs-i) gave practically the same impedance spectrum as
SAAi. An analogous situation was found for SAAu-treated specimens.
Figure 4. SEM images of the laser-structured SiC embossing die (a; top view) and of mechanically
structured Al Mg1 surfaces (b–d; specimens tilted in SEM); (b) ME; (c) ME + MB (grit 1200);
(d) ME + MB (grit 600).
As an alternative to the anodic route of micro-roughening, micro-embossing was

employed at elevated temperatures. Figure 4a, b shows the SiC embossing die with
its regular array of laser-formed cavities and the embossed Al Mg1 metal surface,
respectively. The protrusions of the latter have a shape and arrangement which are
very similar to patterns found on the lotus leaf. The die removal did not cause
damage to either the metal nor to the tool. It should be noted that the hardness
of Al Mg1 is relatively high. Therefore, it was necessary to employ an elevated-
temperature embossing technique, in which both the tool and the sheet sample were
heated. At ambient temperatures the pattern was not completely transferred. How-
ever, ambient-temperature embossing was found to be suitable for Al 99.5, which
has about 2/3 of the hardness of Al Mg1.
An additional blasting treatment gave the surface a uniform roughness, but also
caused abrasion and deformation of the protruding bumps. The latter was more pro-
nounced for the rather coarse 600 grit powder compared with 1200 grit (Fig. 4c, d).
3.2. Coated Surfaces
The grafted hydrophobizing polymer POMA forms very thin films, comparable
to the wet-deposited PFATES and AS/TAF coatings. Thus, the underlying micro-
mountain-like morphology (Fig. 1b) is fully preserved in these cases.
For the HFCVD-generated PTFE layers, the microscopic shape is noticeably
different, because of the inherent morphological properties of the deposits. The spe-
cific morphology of the coating is more pronounced with increasing thickness (50–
1000 nm) and, moreover, it is dependent on the deposition conditions employed.
Figure 5. SEM images for different treatments and corresponding DCA data (θa //θr ); (a) SAAi + PTFE
(standard coating, 1000 nm), fractured specimen showing its outermost surface (upper part) and the
fractured area (lower part), fractured PTFE layer marked by arrows; (b) SAAi + PTFE (annealing
type 1, 500 nm); (c) ME + MB (grit 600) + PTFE (standard coating, 1000 nm); (d) Smooth metallic
substrate + PTFE (standard coating, 1000 nm); (a–c: specimens tilted by 35◦ , d: top view, b: acceler-
ation voltage 10 kV).
While the standard coating conditions produce a shape with very small protrusions
of about 0.1 µm height (Fig. 5a, d), other deposition regimens produce interpenetrat-
ing flakes of 0.2–0.4 µm in diameter (Fig. 5b). The cryo-fractured sample of Fig. 5a
shows that the PTFE layer follows the substrate’s oxide surface profile, where the
new micro-profile is slightly more rounded than that of the oxide. The real coating
thickness in the case of the micro-rough SAAi substrate can be derived likewise.
Fig. 5c shows the situation for a standard-coated, ME/MB-pretreated substrate.
Nearly all the investigated combinations of roughening and coating treatments
led to superhydrophobic properties with CAs of around 150◦ and a generally neg-
ligible hysteresis. The entire wetting results can be seen in Table 1. In Fig. 5a–d
the corresponding DCA data are displayed. The data demonstrate that various
roughening pretreatments have different impacts on the SH, in particular on the
receding angles. In detail, SH is not preserved when i — SAAi is replaced by SAAu,
ii — 600 blasting grit is replaced by 1200 grit, or iii — the micro-blasting step is
completely omitted. These examples emphasize that a sufficient degree of rough-
ness of the substrate is definitely necessary in order to achieve SH.
When there was no roughening pretreatment, as in the case of a mere PTFE
standard coating on a smooth sheet (Fig. 5d), then the receding angle was dra-
matically reduced down to less than 100◦ . The advancing angle was also affected
((144 ± 3)◦ ). This means that the specific morphology of thicker hydrophobic
Table 1.
Data compilation for the as-coated states and after the artificial weathering exposure (WTH); col-
umn 1: sequence of treatments (details, coating thickness) [number of specimens included]; columns 2
and 3: wetting properties according to DCA measurements stated as CA averages ± standard devia-
tions for single specimens or CA spans for several specimens of the same type; column 4: carbon to
fluorine elemental ratios acc. to XPS (single specimens)
Contact angles Contact angles c(F) / c(C)

Treatment (θa / ◦ // θr / ◦ ) (θa / ◦ // θr / ◦ ) (before →
(as coated) (after WTH) after WTH)
SAAi + Chs + POMA [3] 153 // 152 superhydrophilic –

SAAi + PFATES [2] 153 // 152 152–154 // 148–150 1.8 → 1.6
SAAi + AS/TAF [2] 152 // 151–152 154 // 151–152 1.4 → 1.7
SAAi + PTFE (std. coat., 250–1000 nm) 151–152 // 148–151 142–153 // 130–147 2.1 → 1.9
[6]
SAAi + Chs + PTFE (std. coat., 250– 151–152 // 149–151 133–150 // 99–130 –
1000 nm) [4]
SAAi + PTFE (annealing type 1, 500 nm) [2] 151–152 // 151 152 // 147–148 2.0 → 2.0
SAAi + PTFE (annealing type 2, 500 nm) [1] 153 ± 1 // 151 ± 1 152 ± 2 // 142 ± 2 2.1 → 2.1
SAAi + PTFE (elevated substrate temp., 151 ± 1 // 150 ± 1 152 ± 1 // 150 ± 1 2.2 → 2.1
500 nm) [1]
SAAi + PTFE (lower pressure, 500 nm) [1] 152 ± 1 // 151 ± 1 154 ± 1 // 141 ± 1 1.8 → 2.0
SAAu + PTFE (std. coat., 50–1000 nm) [3] 151–153 // 140–144 127–134 // 47–70 2.1 → 1.6
Pickled metallic substrate + PTFE (std. 153 ± 1 // 149 ± 1 137 ± 4 // 79 ± 2 –
coat., 500 nm) [1]
Smooth metallic substrate + PTFE (std. 144 ± 3 // 94 ± 6 112 ± 4 // 73 ± 11 –
coat., 1000 nm) [1]
ME + PTFE (std. coat., 1000 nm) 151 ± 3 // 147 ± 5 154 ± 1 // 149 ± 1 –
MB (grit 1200) + PTFE (std. coat., 1000 nm) 151 ± 1 // 146 ± 2 127 ± 11 // 84 ± 27 –
[1]
ME + MB (grit 1200) + PTFE (std. coat., 151 ± 2 // 146 ± 2 150 ± 1 // 146 ± 3 –
1000 nm) [1]
MB (grit 600) + PTFE (std. coat., 1000 nm) 157 ± 1 // 155 ± 1 151 ± 2 // 141 ± 7 –
[1]
ME + MB (grit 600) + PTFE (std. coat., 155 ± 1 // 153 ± 1 152 ± 1 // 146 ± 1 –
1000 nm) [1]
PTFE coatings is insufficient to achieve superhydrophobic properties. On the other

hand, the advancing angle is markedly higher than the value of 108◦ stated for
smooth PTFE material [21].
EIS measurements yielded results for SAAu + PTFE (standard coating) spec-
imens that were quite similar to different wet-deposited, thin-film coatings [22].
Increasing PTFE thickness leads to higher impedance levels of the ‘plateaus’ at
intermediate frequencies in the log |Z| − log f curves (Fig. 6). However, the resis-
tance of the anodic barrier layer, expressed by the impedance level in the sub-mHz
region, is the largest in the entire system. Measurements on SAAi-based sheet spec-
Figure 6. EIS spectra for SAAu, SAAu + PTFE (std. coat., 50 nm), and SAAu + PTFE (std. coat.,
1000 nm) in phosphate buffer; sheet specimens; Bode plot: modulus of impedance, |Z| and phase
angle, φ vs. frequency, f .
Figure 7. FT–IRRAS absorbance spectra for different samples; (a) SAAi + Chs-e + POMA;
(b) SAAi + PFATES; (c) SAAi + AS/TAF; (d) SAAi + PTFE (std. coat., 500 nm); (e) SAAi + PTFE
(annealing type 1, 500 nm); (f) SAAi + PTFE (lower pressure deposition, 500 nm); (g) MB + PTFE
(std. coat., 1000 nm).
imens could not be satisfactorily made because of an inconsistent effective area

during immersion due to capillary effects.
The chemical properties of the coating–substrate systems were investigated by
means of FT–IRRAS and XPS. The infrared spectra were highly reproducible for
a particular sample so that only one spectrum for each is displayed in Fig. 7.
The POMA-modified surface showed C–H stretch bands at 2851 and 2923 cm−1
(Fig. 7, curve a) indicating the presence of long alkyl chains, which are respon-
sible for diminished surface free energy. The presence of C–O and C=O bonds
is indicated by the small bands at 1700–1770 cm−1 . For the F-containing coat-
ing compounds the typical C–F stretch vibrations were recorded most clearly with
PTFE coating on ME/MB substrates (Fig. 7, curve g). The two bands at 1150 and
1205 cm−1 (shoulder at 1260 cm−1 ) are in agreement with literature data [19, 20].
For SAAi-pretreated specimens the positions of these bands deviate significantly.
Additionally, the absorbance pattern in this region slightly varies for the different
PTFE types (Fig. 7, curves d–f), where the coating generated at lower pressure
shows a small deviation. When the thickness of the standard coating is varied,
the band at about 1175 cm−1 remains constant in contrast to the band beyond
1200 cm−1 , which undergoes a shift. This results from a superposition with a band
resulting from the oxide substrate.
For the application of XPS to specimens with a rough surface it should be noted
that the real take-off angles, and, hence the information depth, vary locally. This is
especially true for SAAi-based samples with their steep micro-profile. According to
the findings, the C1s high resolution spectra reveal a more or less complex structure,
which results from the respective structure and binding situations of the different
coating compounds analyzed. For the thin wet-deposited Chs + POMA coating the
C1s spectrum of the composite layer (Fig. 8b) shows a dominant component peak
(A) which is mainly due to POMA’s octadecyl groups. The two component peaks C
and D, which are typical for the C–O–(H, C) and O–C–O (acetal) groups of chitosan
(Fig. 8a), respectively, are strongly diminished. This indicates that the chitosan layer
is completely covered by POMA. Amide and imide groups, formed during the re-
action between chitosan and POMA, were identified as the cause of the component
peaks E and G [23]. The situation is similarly complex for the cases of PFATES
and AS/TAF (Fig. 8c, d). The fluorosilane PFATES contains groups such as –CH2 –,
–CF2 – and –CF3 , whereas the carbon atoms of the duplex film AS/TAF are bound
to the hetero elements silicon, nitrogen, oxygen (C–O–C), and fluorine (–CF2 –,
–CF3 ). The C–F bonds are characterized by high binding energies Eb 292 eV;
they correspond to the component peaks Y and Z. It should be noted that in the
cases of thin coatings the measured high oxygen contents of about 30 at.% do not
come from the coatings alone, but also from the oxidized Al oxide. This is con-
firmed by the detection of Al (ca. 9 at.%).
For specimens with different 500 nm thick PTFE coatings the F/C ratios were
higher than in the cases of the F-containing wet-deposited coatings (Table 1). The
ratios were in the range 1.8–2.2, i.e. near the theoretical value for PTFE. In the
case of bulk PTFE material a ratio of 2.1 was determined. Most of the C1s spectra
are dominated by the peak at 292 eV, which is attributable to the –CF2 – units in
the polymer chains (Fig. 8e). The lower pressure coating variant is characterized
by a considerably higher proportion of –CF3 bonds (Fig. 8f). Moreover, there is
a noticeable contribution of carbon with a lower binding energy. This is consistent
with the considerable oxygen content of about 13 at.% (it generally does not exceed
2 at.%). These findings indicate that the lower pressure formation conditions result
in marked deviations from the regular PTFE composition. It is expected that the
compound trifluoroacetyl fluoride CF3 CFO, which is one of the products formed
Figure 8. C1s high-resolution ×PS spectra of SAAi-based specimens with different coatings;
(a) SAAi + Chs; (b) SAAi + Chs + POMA; (c) SAAi + PFATES; (d) SAAi + AS/TAF; (e) SAAi + PTFE
(std. coat., 500 nm); (f) SAAi + PTFE (lower pressure deposition, 500 nm).
by the thermal decomposition of hexafluoropropylene oxide [20], still plays a role

in this type of polymer coating.
3.3. Coated Surfaces Followed by Artificial Weathering
After having undergone the WTH exposure, the specimens revealed no visual alter-
ations. Moreover, the SEM examination showed practically unchanged morpholog-
ical properties.
The wetting behavior of the exposed samples, however, gave a different picture.
As the CA data clearly document (Table 1, third column), degradation phenomena
were observed, the degree of which was influenced by the respective treatments:
i — Practically no changes in the CAs were observed for PFATES, AS/TAF,
PTFE (annealing type 1), and PTFE (elevated substrate temperature), all on SAAi
substrates. The same findings are true also for ME + PTFE (std. coat.), and
ME + MB (different grit sizes) + PTFE (std. coat.).
Figure 9. FT–IRRAS absorbance spectra for different samples before and after artificial weathering;
(a, b) SAAi + AS/TAF + WTH (for b); (c, d) SAAi + PTFE (std. coat., 500 nm) + WTH (for d);
(e, f) SAAi + PTFE (elevated substrate temperature, 500 nm) + WTH (for f).
ii — Moderate changes in the wetting behaviour with receding angles of about

140◦ were observed for SAAi + PTFE (std. coat., annealing type 2 and lower pres-
sure coating), and MB-600 + PTFE (std. coat.). Here, the advancing angles still
remained at the SH level.
iii — Considerable worsening occurred with SAAi + Chs-e + PTFE (std. coat.),
where also the advancing angle decreased.
iv — A dramatically worsened behavior was observed for SAAi + Chs + POMA
(becoming completely hydrophilic), SAAu + PTFE (std. coat.), pickled substrate +
PTFE (std. coat.), smooth metal + PTFE (std. coat.), and MB (grit 1200) + PTFE
(std. coat.).
It follows from the DCA measurements that the behavior after the WTH exposure
of the specimens with a PTFE standard coating was noticeably variable in spite of
the same initial surface chemistry. Specimens with only low roughness appear to
undergo a more pronounced SH degradation during WTH compared to the rougher
specimens. The reason is not yet clear.
An attempt was made to relate the actual wetting properties and their changes
to the corresponding surface chemistry findings. FT–IRRAS revealed that the
C–F-related region at 1150–1200 cm−1 was not influenced by WTH as seen from
the spectra in Fig. 9 for different layer systems, despite the different impact of ex-
posure on their wetting behavior. However, the content of water in all the specimen
types was slightly higher than before the exposure.
XPS is known to be more sensitive to changes in the outermost surface, which
governs the wetting behavior. For SAAi-based specimens covered with PFATES or
AS/TAF, where there were no significant changes in the wetting properties follow-
ing WTH, interesting features were detected by XPS. In the case of PFATES, the
findings reveal a very high stability of this coating compound under the conditions
of the exposure (Table 1, Fig. 10a, b). However, for the AS/TAF coating, an increase
Figure 10. C1s high-resolution ×PS spectra for artificially weathered, SAAi-based specimens with
different coatings; (a, b) SAAi + PFATES + WTH (for a) or + WTH-L (for b); (c, d) SAAi + PTFE
(std. coat., 500 nm) + WTH (for c) or + WTH-L (for d); (e, f) SAAi + PTFE (enhanced substrate
temperature, 500 nm) + WTH (for e) or + WTH-L (for f).
in the F/C ratio was accompanied by a decrease of the low-energy components of

the carbon signal, whereas the F-bound carbon signal remained at a high level. This
shows that the primary aminosilane coating component, which is free of fluorine,
underwent vast degradation in the course of the exposure.
For PTFE standard coatings the weathering impact was found to be generally
higher according to the XPS measurements. As mentioned above, at least some
of the SAAi-based specimens underwent noticeable drops in the receding angles
(Table 1) such that SH was not fully preserved in these cases. Figure 10c indicates,
for a SAAi + PTFE (std. coat.) specimen, that the proportion of electropositively
bound carbon (low binding energy) has increased as a result of the WTH exposure
(cf. Fig. 8e). This finding is associated with an increase in the oxygen content, which
represents some newly generated side groups or breaking of the polymer backbone.
For SAAi + PTFE (elevated substrate temperature) + WTH, which was found
to preserve very high CAs, XPS measurements indicated that the proportion of
electropositively bound carbon remained lower than for the standard coating type
PTFE (Fig. 10c, e). Thus, XPS was found to be a valuable tool in relating the
wetting behavior with the elemental composition of the uppermost surface layer.
An additional weathering experiment excluded moisture so that only a dry light
irradiation took place over 360 h (WTH-L). The advancing contact angles were
found to be nearly the same as measured after the regular WTH procedure, but the
receding CAs were drastically diminished and, generally, showed a higher scatter.
In the best case, values of (150 ± 3)◦ //(119 ± 10)◦ were obtained for SAAi + PTFE
(elevated substrate temperature, 500 nm) after WTH-L. Despite these findings, the
C1s high resolution spectra for the PTFE-coated specimens were generally very
similar to those after regular WTH (exposure cf. spectra pairs of Fig. 10). These
findings suggest that the various coating compounds might have been affected by
the dry exposure in different ways and that the damaging mechanism without mois-
ture may be different from that for regular WTH conditions, where the water is
expected to influence the actual degradation. In light of the XPS findings, the wors-
ened wetting properties (especially lowered θr ) after WHT-L might be explained
by small local coating defects, which are responsible for local pinning during the
receding of the wetting triple line in the course of the dynamic CA measurement.
4. Conclusions
In order to investigate the effects of preparation differences on superhydrophobic-
ity (SH) and to judge the weathering stability, a variety of roughening pretreatments
(electrochemical and mechanical) and water-repellent coatings (wet-deposited thin
films and PTFE films generated by hot-filament chemical vapor deposition) were
considered and tested. The chemical stability was investigated by employing a stan-
dardized artificial weathering test.
Within the experimental conditions, noticeable influences on the wetting prop-
erties of the coated systems were found to depend on the manner of roughening.
Superhydrophobicity was achieved in those cases, where the pretreatment generated
a suitable degree of sub-micro-roughness and micro-roughness components, e.g.
with the anodization route SAAi or for micro-embossed plus micro-blasted surfaces.
The usual anodization SAAu and mere etching caused worsened water-repellent
properties, although there was a contribution from the inherent micro-roughness
of the HFCVD-produced PTFE films.
For artificial weathering exposure, a very high stability of superhydrophobic-
ity was observed for the fluorinated wet-deposited PFATES and AS/TAF coatings
as well as for PTFE deposited at an elevated substrate temperature, all on SAAi-
pretreated substrates. Very good results were also obtained for specimens produced
by appropriate mechanical roughening in combination with PTFE coatings. As a
rule, deteriorating water-repellent properties were associated with a decrease in the
XPS-derived fluorine concentration and the F/C ratio as well as with an increase
of the oxygen concentration. The AS/TAF duplex film underwent decomposition of
the aminosilane component as a result of the weathering. The weathering stabil-
ity of the PTFE standard coating was found to be better on the SAAi substrates as
compared to those treated by SAAu or etching.
Acknowledgements
This work was, partly, supported by a grant from the Saxon State Ministry of Sci-
ence and Fine Arts (Saechsisches Staatsministerium fuer Wissenschaft und Kunst,
SMWK). One of us (A.P.) appreciates the support of the German Academic Ex-
change Service (DAAD). The micro-embossing and micro-blasting methods were
developed by Mr. T. Burkhardt and Mr. J. Engelmann (FhG-IWU Chemnitz). The
experimental contributions of Mrs. K. Galle (TUD) and Mrs. B. Schneider (IPF)
are also greatly acknowledged.
References
2. D. Öner and T. J. McCarthy, Langmuir 16 (20), 7777–7782 (2000).
3. Y. Yu, Z.-H. Zhao and Q.-S. Zheng, Langmuir 23, 8212–8216 (2007).
4. M. Nosonovsky and B. Bhushan, Ultramicroscopy 107, 969–979 (2007).
5. S. Shibuichi, T. Yamamoto, T. Onda and K. Tsujii, J. Colloid Interface Sci. 208, 287–294 (1998).
6. W. A. Zisman, in: Contact Angle, Wettability and Adhesion, Adv. Chem. Ser. No. 43, pp. 1–51.
American Chemical Society, Washington, DC (1964).
7. M. Pagliaro and R. Ciriminna, J. Mater. Chem. 15, 4981–4991 (2005).
8. M. Callies and D. Quéré, Soft Matter 1, 55–61 (2005).
9. M. Ma and R. M. Hill, Current Opinion Colloid Interface Sci. 11, 193–202 (2006).
10. J. Wang, Y. Yu and D. Chen, Chinese Science Bulletin 51, 2297–2300 (2006).
11. C. Blank, M. Thieme, V. Hein, H. Worch, T. Burkhardt, R. Frenzel, S. Höhne, H. Pryce Lewis
and A. J. White, in: Aluminium Alloys — Their Physical and Mechanical Properties, J. Hirsch,
B. Skrotzki and G. Gottstein (Eds), pp. 2131–2138. Wiley-VCH, Weinheim (2008).
12. R. Wang, K. Hashimoto and A. Fujishima, Nature 388, 431–432 (1997).
13. M. Thieme, R. Frenzel, S. Schmidt, H. Worch, F. Simon and K. Lunkwitz, Adv. Eng. Mater. 3,
691–695 (2001).
14. M. Thieme, R. Frenzel, V. Hein and H. Worch, J. Corrosion Sci. Eng. 6, Paper 47 (2003).
15. R. Frenzel, S. Schmidt, V. Hein, M. Thieme and F. Simon, in: Verbundwerkstoffe, H. P. Degischer
(Ed.), pp. 489–493. Wiley-VCH, Weinheim (2003).
16. C. Blank, R. Frenzel, V. Hein, B. Schmidt, F. Simon, M. Thieme, K. Tittes and H. Worch, in:
Praktische Metallographie, Sonderband, Vol. 36, G. Petzow (Ed.), pp. 491–496. Werkstoffinfor-
mationsges, Frankfurt (2004).
17. K. Tittes, B. Schmidt, C. Blank, V. Hein, H. Worch, F. Simon and R. Frenzel, in: GdCh-
Monographie, Vol. 32, J. Besenhard and J. Russow (Eds), pp. 176–184. GdCh, Frankfurt (2004).
18. C. Blank, V. Hein, M. Thieme, H. Worch, S. Höhne and F. Simon, in: Praktische Metallographie,
Sonderband, Vol. 39, G. Petzow (Ed.), pp. 175–182. Werkstoffinformationsges, Frankfurt (2007);
C. Blank, S. Höhne, M. Thieme, H. Worch, R. Frenzel, F. Simon and V. Hein, Pending patent
application at Deutsches Patent- und Markenamt, Munich, 11.09.2007.
19. K. K. S. Lau, J. A. Caulfield and K. K. Gleason, Chem. Mater. 12, 3032–3037 (2000).
20. K. K. S. Lau, S. K. Murthy, H. G. P. Lewis, J. A. Caulfield and K. K. Gleason, J. Fluorine Chem.
122, 93–96 (2003).
21. K. K. S. Lau, J. Bico, K. B. K. Teo, M. Chhowalla, G. A. J. Amaratunga, W. I. Milne, G. H. McKin-
ley and K. K. Gleason, Nano Lett. 3, 1701–1705 (2003).
22. M. Thieme and H. Worch, J. Solid State Electrochem. 10, 737–745 (2006).
23. S. Höhne, R. Frenzel, A. Heppe and F. Simon, Biomacromolecules 8, 2051–2058 (2007).
24. D. Briggs, Characterization of Surfaces. Pergamon, Oxford (1989).
25. M. P. Seah and W. A. Dench, Surface Interface Anal. 1, 2 (1979).
26. G. Beamson and D. Briggs, High Resolution XPS of Organic Polymers, The Scienta ESCA 300
Database. Wiley, Chichester (1992).
Aqueous and Non-aqueous Liquids on Superhydrophobic
Surfaces: Recent Developments
Michele Ferrari
CNR-National Research Council — IENI, Institute for Energetics and Interphases, via De Marini 6,
16149 Genova, Italy
Abstract
In this paper recent developments regarding design and preparation of superhydrophobic substrates and the
wetting behavior of water based solutions, engineered and non-aqueous liquids in contact with such surfaces
are summarized, considering application in a range of basic research and industrial fields.
The combination of highly water repellent surfaces with engineered liquids is of great interest in opening
new trends in liquid handling and manipulation, especially as regards to small volumes.
Reference data and related studies are still not sufficient to cover the several aspects of ultralyophobicity,
in particular in those fields where the research deals with specific problems related to compatibility or
solubility, to name a few. The exploitation of these studies in switching between wetting states has been
a topic of some investigations, while only very few reports are available on immiscible liquids.
Keywords
Superhydrophobicity, wetting, non-aqueous liquids
1. Introduction
In this paper an overview about the behavior of water-based solutions and non-
aqueous liquids in contact with surfaces or coatings with extreme water repellency
is provided. The terms superhydrophobic or ultrahydrophobic refer to low energy
surfaces with a water contact angle (CA) greater than 150◦ and since the introduc-
tion of the idea of the Lotus effect [1, 2] to enhance and exploit the self-cleaning
properties of solid surfaces, many researchers have introduced plenty of techniques
to obtain superhydrophobic surfaces, most of them being based on controlling the
roughness or topography of low energy surfaces (Fig. 1).
The birth of new disciplines like biomimetics indicates the strong attention given
to the exploitation of structural properties of plant or animal surfaces: starting from
micro or submicroscopical observations, the characterization has led to a more and
more increased “market” of structures with specific tailored features. The combina-
Tel.: +39 010 6475723; Fax: +39 010 6475700; e-mail: m.ferrari@ge.ieni.cnr.it

270 M. Ferrari
Figure 1. AFM image of a FAS-TEOS superhydrophobic surface showing nanoscale roughness.
tion of such highly water repellent surfaces with engineered liquids is and will be of
great interest in opening new trends in liquid handling and manipulation, especially
as regards to small volumes.
In particular, recent developments regarding design and preparation of superhy-
drophobic coatings or substrates will be considered here in terms of their applica-
tions in a wide range of basic research and industrial fields.
The importance of the studies regarding non-aqueous liquid–solid interactions
so far has not been investigated in detail; therefore, the literature available has been
reviewed focusing on application aspects of ultrahydrophobic and ultralyophobic
states.
2. Superhydrophobic States: Learning from Nature

The first lesson coming from nature deals with the well-known self-cleaning proper-
ties of lotus leaves, which have been under investigation for their particularly strong
water repellency due to a unique structure with a highly hydrophobic character.
This structure was recently studied in detail by Marmur [3, 4] by electron mi-
croscopy, showing a 2D roughness in leaf appendixes of several micrometer in size
and covered with small wax crystals.
Since the time of seminal works of Wenzel [5, 6] and later of Cassie and Bax-
ter [7], investigation of birds and insects wings or beetle shields [8, 9], together with
different kinds of plants leaves, has led to the key conclusion that such a strong hy-
drophobic effect arises from combining a micro- and nano-patterned surface with
a particular geometry with a coating of natural organic compounds of different na-
ture [1].
Aqueous and Non-aqueous Liquids on Superhydrophobic Surfaces 271
The models of Wenzel and Cassie–Baxter have been utilized to interpret the
roughness effect on the wettability properties of a solid surface.
The Wenzel’s approach assumes the liquid to fill the space between the pro-
trusions on the surface, linking the apparent contact angle θ and thermodynamic
contact angle θ as
cos θ = r cos θ, (1)
where r (roughness factor) is the ratio between the true surface area and its hori-
zontal projection.
According to Cassie and Baxter the surface traps air in the hollow spaces of the
rough surface, and the superhydrophobicity can be interpreted as follows:
cos θ = fLS cos θ − fLV , (2)
where fLS is the fraction of liquid area in contact with the solid and fLV is the
fraction of liquid area in contact with the trapped air (fLS + fLV = 1).
The two states can be distinguished by the contact angle hysteresis. At larger
hysteresis they can be regarded as belonging to the Wenzel’s one, while at smaller
hysteresis values, the Cassie–Baxter approach is applicable: the surface can be re-
garded as composed by a pillar-like structure supporting the liquid and reducing the
available area.
This combination of micrometer and nanometer-scale roughness with low sur-
face energy in a non-homogeneous wetting regime (Cassie–Baxter regime) led
Onda and co-workers [10] to create “artificial” superhydrophobic surfaces.
In order to apply the water repellent properties of duck feathers for engineer-
ing new materials featuring such non-wetting behavior, the microstructure of the
feathers was investigated by Liu et al. [11] with a scanning electron microscope
(SEM) via a method based on a different magnification stages procedure. The re-
sults showed that superhydrophobic behaviour of duck feathers was the result of the
combination of this particular structure and the presence of the preening oil.
A novel method, based on surface solution precipitation (SSP) was introduced to
simulate the feather microstructure on textile substrates using chitosan as building
blocks, and then the textile substrates were further modified with a silicone com-
pound to lower the surface energy. Highly water repellent properties were observed
in textiles showing bionic superhydrophobic surfaces prepared on soft substrates by
a simple procedure involving flexible biopolymeric materials.
3. Superhydrophobic Surfaces and Drop Manipulation

Handling small volumes of liquids has provided a great stimulus for designing
appropriate surface topography. Surface roughness is known to be crucial for su-
perhydrophobicity, however, new approaches are required to expand the concept of
surface roughness, such as hierarchical roughness in more complex mechanisms of
wetting, CA hysteresis and wetting regime transition.
272 M. Ferrari
A simple casting method was used by Hou et al. [12] for preparing a polypropy-
lene/methylsilicone based superhydrophobic surface. The surface microstructure
could be tuned by varying the ratio of polypropylene and methylsilicone result-
ing in different surface features. The wetting behaviour of the as-prepared surface
was investigated. A polypropylene monolithic material was also prepared and its
superhydrophobicity was still retained when the material was cut or abraded. The
as-prepared material can also be used to separate some organic solvents from wa-
ter.
A different application of bio-inspired superhydrophobic surfaces was found by
Nosonovsky and Bhushan [13] for adhesion reduction between micro/nanoelectro-
mechanical systems (MEMS/NEMS) components.
The authors suggest that the dissipation mechanisms is related to hierarchical
roughness and may lead to self-organized criticality.
To obtain the desired surface roughness, various techniques like chemical and
plasma etching, laser treatment, chemical vapour deposition and electrodeposition,
dipping and spraying, have been used together with surface modification by me-
chanical methods or photopatterning [14].
As suggested by Extrand [15] and Sedev et al. [16], roughened surfaces obtained
by micropatterning, machining or etching show higher values of CA due to inhibi-
tion of the liquid by the rough surface in such a way that drop spreading is hindered
by the edges of the grooves.
In addition, a systematic study on different rough surfaces with a well-defined
surface chemistry was the topic of Spori et al. [17] where water CA measurements
were performed for a better understanding of liquids in contact with rough surfaces
ranging from sandblasted glass slides to sandblasted titanium. In particular, they
found that photolithographically fabricated golf-tee shaped micropillars (GTMs)
showed Cassie-type hydrophobicity, even in the presence of hydrophilic surface
chemistry.
Despite the hydrophilic nature of the rough surfaces, CAs are shifted to more
hydrophobic values because of pinning effects, unless roughness or surface energy
are such that capillary forces become significant, leading to complete wetting.
The observed hydrophobicity is thus not consistent with the well-known Wenzel
equation. The authors show that surface chemistry does not influence the pinning
strength of the surface if capillary forces or air pockets are not involved. By plotting
CAs on rough versus flat surfaces, the authors described the pinning strength by the
intercept of the plots at fixed surface chemistry.
Electrowetting has also been recently employed [18] for the micromanipulation
of a liquid droplet which can be picked up and handled by controlling the wetting
property between the liquid and the substrate itself (Fig. 2). The authors provide
a precise analysis through a numerical method of the process behind the rupture of
the liquid bridge between the conical gripper and the substrate. The authors found
it possible to control the efficiency of micromanipulation in different conditions
of CAs between the liquid and the gripper taking into account the distribution ra-
Figure 2. Micromanipulation by electrowetting of a liquid droplet which can be picked up and handled
by controlling the wetting between the liquid and the substrate.
tio between the droplet volume retained by the substrate and the whole volume
of the liquid droplet during the rupture. An optimal efficiency was attained which
was supported by a theoretical analysis that helped to find the best conditions and
parameters for the micromanipulation process experimentally demonstrated on the
standard probe of an AFM.
Another interesting application of such a technique for nanomaterials can be
found in Brunet et al. [19] where experiments on drop impact impalement (Wenzel
state) and electrowetting were performed to compare the wetting properties of su-
perhydrophobic silicon nanowires. In this paper the authors provide a comparison
between the resistance to impalement by electrowetting and drop impact.
From this study it becomes evident that there is a proportional, direct relationship
between the increase of the length and density of nanowires and the thresholds for
drop impact, observing an increase of the electrowetting irreversibility while the
CA hysteresis after impalement decreases. The threshold to impalement of such
a surface results to be up to three times higher than most of the surfaces or coatings
tested providing a good preservation of the reversibility.
The design of hydrophobic materials and coatings with tailored superhydropho-
bic features of the surface finds theoretical support in Boinovich and Emelya-
nenko [20] where the authors discuss the possibilities of the formation of ordered
textures with high CAs on the surfaces of hydrophobic materials. They provide
an analysis of the necessary conditions for thermodynamic stability of the het-
erogeneous wetting regime of such surfaces facing the problems of ageing and
degradation of superhydrophobic coatings. The authors give examples of the use
of superhydrophobic materials for industrial applications.
274 M. Ferrari
4. Superhydrophobic Surfaces and Aqueous Mixtures of Organic Liquids

Basic research investigations on exploiting biosurfaces in contact with liquids can
be found in the work by Fang et al. [21] where the wetting properties of distilled
water and methanol solution on the wings of butterflies were studied by CA mea-
surements. The scale structures of the wings were observed using scanning electron
microscopy and the influence of the structure scale on the wettability was investi-
gated.
Results show strong hydrophobic behaviour in numerous species with CAs
greater than 150 degrees. In some cases the CA of distilled water on the wing
surfaces varies from 134 to 159 degrees. The wing surfaces of some species are
not only hydrophobic but also resist wetting by methanol solution with 55% con-
centration. Because of the structure features (spindle-like and pinnule-like shapes)
only two species with these shapes cannot resist wetting, showing large difference
compared to the other species. Spreading/wetting on the wing surfaces of differ-
ent species was observed for a large concentration range (from 70% to 95%) of
methanol solutions. After contact with methanol solution for 10 min, the wing sur-
face showed an enhanced capacity against wetting by distilled water with a CA
increase up to two degrees.
Because of the lack of adequate literature in this field, authors like Shirt-
cliffe [22], have focused on the topic of interaction between non-aqueous liquids
and highly water repellent surfaces. In this work they studied the switching behav-
ior from superhydrophobic to hydrophilic properties by varying physico-chemical
parameters. In particular, they investigated aqueous solution of ethanol in contact
with porous superhydrophobic substrates finding that in low surface tension condi-
tions such liquids could enter surface grooves.
For superhydrophobic surfaces Rao et al. [23] also studied porous substrates for
industrial purposes. They showed how the surface tension could be the reason for
the transition from the Cassie state to the Wenzel state. Even without performing
CA measurements of pure liquids, it was reported that with lower surface tension
the non-aqueous pure liquids, but not the pure water, penetrated the surface grooves.
Investigations on water–alcohol mixtures were reported also by Fujita et al. [24].
They measured high advancing angles (above 150◦ ) for highly hydrophobic pat-
terned surfaces with water and water–glycerol mixtures, but below 10◦ for ethanol
(γLV = 22.3 mN/mm), confirming the Cassie–Wenzel transition with the decrease
in surface tension.
Organic and inorganic coatings were studied by Shibuchi and co-workers [25,
26]. They first explored the potential offered by fractal surface produced sponta-
neously by solidification from the melt of alkylketene dimer, AKD (a kind of wax)
and obtained highly water repellent substrates showing a CA on the order of 174◦
with drops rolling off the surface at a small tilt angle.
The CAs of some water/1,4-dioxane mixtures on the fractal and the flat AKD
surfaces were determined and a decrease in CA with increase of dioxane fraction
was observed. The surface tension of the liquid changes from 36 mN/m for pure 1,4-
dioxane to 72 mN/m for pure water, depending on the concentration of 1,4-dioxane.

As a function of the fractal parameters they also studied the wetting properties of
the mixtures at different water contents, finding low CAs (15–20◦ ) for the lowest
water ratio (20%). It has been demonstrated by this work that the fractal concept is
a powerful tool to develop some novel functional materials, for example which can
act as a switch from the Cassie to the Wenzel mode.
In [26] the authors investigated an aluminum substrate roughened first by elec-
trochemical etching and then hydrophobized by a fluorination treatment. The alu-
minum was processed by anodic oxidation obtaining a highly water wettable sur-
face, which was then modified to ultra water repellent (CA of 160◦ for water
droplets) by treatment with perfluorooctyltrichlorosilane.
In this study they reported super water- and oil-repellent surfaces made by
exploiting the fractal structure of the surface. Fluorinated monoalkyl phosphates
were used as hydrophobizing agent for the aluminum substrate to obtain super oil-
repellent surfaces. They found a CA of about 150◦ for rapeseed oil (surface tension
of about 35 mN/m) on the super oil-repellent surface where droplets rolled off the
surface without significant attachment. In other cases for oils with a surface tension
greater than 23 mN/m, CAs greater than 120◦ on the super-oil-repellent surfaces
were found. Shibuichi et al. explained their results in terms of fractal dimension
and influences of chemical structure on smooth and rough surfaces.
In [27] several organic liquids of different chemical nature characterised by low
surface tension were used to study the superlyophobic behavior of nanostructured
surfaces. Among the wide variety of tested liquids were alcohols (aliphatic and
alicyclic, short chained), water–alcohol mixtures, aromatic hydrocarbons, ethers,
esters, and silicone oils. These liquids formed on these surfaces droplets with high
mobility, low hysteresis, with related CAs larger than 131◦ . When no voltage was
applied, these surfaces showed, in the initial state, CAs as high as 150◦ for liquids
with surface tensions ranging from ethanol to water. Once applied, the electrical
voltage induced a transition from the superlyophobic state to wetting, whose nature
was investigated both experimentally and theoretically by the authors. These results
can be considered a promising method for manipulating liquids on the microscale,
in fact showing, for the first time, dynamically tunable surfaces, such as nanonails,
capable of driving a transition from a remarkable superlyophobic behavior to almost
complete wetting.
Mohammadi et al. [28] studied pure liquids and surfactant solutions on super-
hydrophobic substrates featuring a rough microstructure produced as in [26] by
spontaneous formation of AKD crystals. As in the paper of Shibuichi et al. [26],
the advancing CAs showed a discontinuous drop with pure liquids at surface ten-
sion values of about 45 mN/m.
In the paper by Chen et al. [29] the preparation of both ultrahydrophobic and ul-
tralyophobic surfaces using several techniques is described. Plasma polymerization
of a fluoroacrylate on poly(ethylene terephthalate) (PET) produced surfaces with
high CAs >170◦ and hysteresis of about 1◦ . Argon plasma etching of polypropy-
276 M. Ferrari
lene in the presence of poly(tetrafluoroethylene) also produces surfaces with similar

CAs but larger hysteresis of 3◦ .
By compressing spherical particles of poly(etrafluoroethylene) (PTFE) of submi-
crometer diameter range, superhydrophobic surfaces were prepared which showed
high CAs and low CAH for water, methylene iodide and hexadecane (CAs of about
140◦ ), in particular the latter result being of great value if compared with previous
data [30] on smooth surfaces. Even considering a standard interpretation concern-
ing a Wenzel state behaviour in this case the authors observe an ultralyophobic
feature. A helpful explanation can be found in [31] in which the geometry of the
surface is regarded as responsible for high CAs despite the low surface tension of
the liquids.
The role of CA hysteresis in characterizing lyophobicity instead of looking at
the maximum CA must be emphasized, as these surfaces are usually rough at
the micrometer and submicrometer scales, and water drops roll easily from all of
them.
The authors in [29] also report about smooth ultralyophobic surfaces prepared
by silanization of silicon wafers. These surfaces exhibit much lower CAs but little
or no hysteresis, and droplets of water, hexadecane and methylene iodide slide off
easily on them. This behaviour has been interpreted with the liquid nature and the
flexibility of the monolayers assuming that droplets in contact with them experience
very low energy barriers. Their conclusion is that topography of the roughness is
important in controlling the continuity of the three-phase contact line and thus the
hysteresis.
5. Non-aqueous Liquids at Superhydrophobic Surfaces

Studies regarding the wetting of smooth hydrophobic and ultrahydrophobic sur-
faces by non-aqueous pure liquids or mixtures are only limited.
Egatz-Gómez et al. [32] follow a microfluidics approach for a faster and more
flexible control over drop movement. They describe a method to control drop mo-
tion on superhydrophobic surfaces by means of magnetic fields operating under the
surface (Fig. 3). In this way they can move liquid nano-drops prepared with low
percentage of paramagnetic particles (0.1% weight) relatively fast coalescing with
a static drop.
Recently new engineered liquids have been under investigation by Bormashenko
et al. [33] for their unique properties in microfluidics. A microfluidic device based
on ferrofluidic marbles has been described.
In particular, a study on the motion on flat polymer substrates containing fer-
rofluidic marbles, prepared by dispersing nanopowder of poly(vinylidene fluoride)
and γ -Fe2 O3 , evidences the behaviour of such a fluid in presence of a magnetic
field. The sliding of ferrofluidic drops on superhydrophobic surfaces was studied
after activation of the marbles by means of an external magnetic field. It is shown
that drop radius influences the drop displacement with a linear dependence on the
Figure 3. Motion of a nano-drop containing low percentage of paramagnetic particles controlled by

means of a magnetic field operating under the superhydrophobic surface.
threshold magnetic force evidencing the role played by the processes at the contact
line.
In [34] the present author shows how the increased topography, from a simple
polymeric fluorine-based coating to a mixed nanoparticles–polymer, increases the
CA in a water–hexane system with a jump of almost 40◦ , where water drops roll
off from the surfaces without sliding (Fig. 4a–c). In combination with surfactant
solutions with different oil solubilities, these systems offer the opportunity of wet-
ting control as a switching effect from a Cassie–Baxter to a Wenzel state that can
be effectively reversed to superhydrophobic behaviour by exploiting the surfactant
distribution between the liquid phases.
6. Conclusions
After more than two decades, the developments regarding design and preparation of
superhydrophobic coatings or substrates are ongoing and seem to give new insights
in terms of a wide range of basic research and industrial application fields. The
combination of such highly water repellent surfaces with engineered liquid is and
will be of great interest in opening new trends in liquid handling and manipulation,
especially regarding small volumes.
Moreover, the behaviour with water, water-based solutions, or organic solvents,
despite the enormous potential applications, has not been studied adequately and
the related studies are still not sufficient to cover the several aspects of ultralyopho-
bicity, in particular in fields, like microfluidics, where one has to deal with specific
problems related to compatibility or solubility, to name a few.
278 M. Ferrari
(a)
(b)
(c)
Figure 4. Water droplet in hexane on glass (a), on fluorinated polymer coated glass (b), on mixed
nanoparticle-fluorinated polymer coated glass (c).
The exploitation of these studies in switching between wetting states has been
the topic of some investigatons, while very few studies are available on immiscible
liquids, and it is clear that topography and a given surface chemistry play important
roles in controlling this transition.
References
1. W. Barthlott and C. Neinhuis, Planta 202, 1 (1997).
2. C. Neinhuis and W. Barthlott, Ann. Botany London 79, 667 (1997).
3. A. Marmur, Langmuir 19, 8343 (2003).
4. A. Marmur, Langmuir 20, 3517 (2004).
5. R. N. Wenzel, Ind. Eng. Chem. 28, 988 (1936).
6. R. N. Wenzel, J. Phys. Colloid Chem. 53, 1466 (1949).
7. A. B. D. Cassie and S. Baxter, Trans. Faraday Soc. 40, 546 (1944).
8. T. Wagner, C. Neinhuis and W. Barthlot, Acta Zool. 77, 213 (1996).
9. L. Zhai, M. C. Berg, F. C. Cebeci, Y. Kim, J. M. Milwid, M. F. Rubner and R. E. Cohen, Nano
Lett. 6, 1213 (2006).
10. T. Onda, S. Shibuichi, N. Satoh and K. Tsujii, Langmuir 12, 2125 (1996).
11. Y. Y. Liu, X. Q. Chen and J. H. Xin, Bioinspiration & Biomimetics 3, 046007 (2008).
12. W. X. Hou, B. Mu and Q. H. Wang, J. Colloid Interface Sci. 327, 120 (2008).
13. M. Nosonovsky and B. Bhushan, Scripta Materialia 59, 941 (2008).
14. S. H. Kim, J. Adhesion Sci. Technol. 22, 235 (2008).
15. C. W. Extrand, Langmuir 20, 5013 (2004).
16. R. Sedev, R. Fabretto and J. Ralston, J. Adhesion 80, 497 (2004).
17. D. M. Spori, T. Drobek, S. Zurcher, M. Ochsner, C. Sprecher, A. Muehlebach, N. D. Spencer and
D. Nicholas, Langmuir 24(10), 5411 (2008).
18. B. Bhushan and X. Ling, J. Phys. Condens. Matter 20, 485009 (2008).
19. P. Brunet, F. Lapierre, V. Thomy, Y. Coffinier and R. Boukherroub, Langmuir 24, 11203 (2008).
20. L. B. Boinovich and A. M. Emelyanenko, Uspekhi Khimii 77, 619 (2008).
21. Y. Fang, G. Sun, Q. Cong, G. H. Chen and L. Q. Ren, J. Bionic Eng. 5, 127 (2008).
22. N. J. Shirtcliffe, G. McHale, M. I. Newton, C. C. Perry and P. Roach, Mater. Chem. Phys. 103,
112 (2007).
23. V. Rao, N. D. Hegde and H. J. Hirashima, J. Colloid Interface Sci. 205, 124 (2007).
24. M. Fujita, H. Muramatsu and M. Fujihira, J. Appl. Phys. 44, 6726 (2005).
25. S. Shibuichi, T. Onda, N. Satoh and K. Tsujii, J. Phys. Chem. 100, 19512 (1996).
26. S. Shibuichi, T. Yamamoto, T. Onda and K. Tsujii, J. Colloid Interface Sci. 208, 287 (1998).
27. A. Ahuja, J. A. Taylor, V. Lifton, A. A. Sidorenko, T. R. Salamon, E. J. Lobaton, P. Kolodner and
T. N. Krupenkin, Langmuir 24, 9 (2008).
28. R. Mohammadi, K. Wassink and A. Amirfazli, Langmuir 20, 9657 (2004).
29. W. Chen, A. Y. Fadeev, M. C. Hsieh, D. Öner, J. Youngblood and T. J. McCarthy, Langmuir 15,
3395 (1999).
30. D. Y. Kwok and A. W. Neumann, Adv. Colloid Interface Sci. 81, 167 (1999).
31. S. Herminghaus, Euro. Phys. Lett. 52, 165 (2000).
32. A. Egatz-Gómez, S. Melle, A. A. García, S. A. Lindsay, M. Márquez, P. Domínguez-García,
M. A. Rubio, S. T. Picraux, J. L. Taraci, T. Clement, D. Yang, M. A. Hayes and D. Gust, Appl.
Phys. Lett. 89, 034106 (2006).
280 M. Ferrari
33. E. Bormashenko, R. Pogreb, Y. Bormashenko, A. Musin and T. Stein, Langmuir 24, 12119 (2008).
34. M. Ferrari, in: Contact Angle, Wettability and Adhesion, Vol. 5, K. L. Mittal (Ed.), pp. 295–308.
VSP/Brill, Leiden, The Netherlands (2008).
Part 4
Surface Free Energy and Relevance
of Wettability in Adhesion
Comparison of Apparent Surface Free Energy of Some
Solids Determined by Different Approaches
Emil Chibowski ∗ and Konrad Terpilowski

Department of Physical Chemistry-Interfacial Phenomena, Faculty of Chemistry,
Maria Curie Sklodowska University, 20-031 Lublin, Poland
Abstract
Four different approaches to determination of solid surface free energy (van Oss et al.’s (LWAB), Owens
and Wendt’s (O–W), Chibowski’s contact angle hysteresis (CAH) and Neumann’s equation of state (EQS))
were examined on glass, silicon, mica and poly(methyl methacrylate) (PMMA) surfaces via measurements
of advancing and receding contact angles. Sessile drop and tilted plate methods were employed to measure
the contact angles of probe liquids water, formamide and diiodomethane. The results obtained show that
on a given solid the advancing contact angle is slightly larger and the receding one smaller if measured by
tilted plate method. Hence, the resulting hysteresis is larger than that from the contact angles measured by
sessile drop. The calculated (apparent) surface free energy is the greatest if determined from O–W equation.
Unexpectedly, EQS fails for weakly polar polymer PMMA surface, giving significantly lower value of the
calculated energy. In rest of the tested systems LWAB, CAH and EQS approaches give comparable results
for the apparent surface free energy of the tested solids. A hypothesis is put forward that using a probe
liquid only apparent surface free energy of a solid can be determined because the strength of interactions
originating from the solid surface depends on the strength of interactions coming from the probe liquid
surface.
Keywords
Solids, surface free energy, different approaches
1. Introduction
Despite numerous papers published dealing with solid surface free energy determi-
nation, it is still an open problem. Not only regarding the theoretical approach, but
also regarding which measured contact angle on a real solid surface is appropriate
for use in the Young equation [1–4]. A critical evaluation of the approaches and
the resulting equations used for calculations of surface tension components and/or
solid surface free energy was published by Lyklema [5], who pointed out a weak
thermodynamic basis underlying the derivations and concluded: “So there is a rea-
*To whom correspondence should be addressed. Tel.: +48-81-537 56 51; Fax: +48-81-533 33 48; e-mail:
emil.chibowski@umcs.pl

284 E. Chibowski and K. Terpilowski
son to continue using equations like equation (16). However, there is no reason
to over-interpret equation (16) by empirically adding extra terms (like acid–base
interactions), let alone combining γ values with contact angle data, using an ad-
ditional empirical expression to obtain solid–liquid interfacial tensions. . .”. The
equation mentioned above is that used for the first time by Fowkes [6, 7], in which
for the dispersion interactions Berthelot’s rule was applied:
γ12 = γ1 + γ2 − 2(γ1d γ2d )1/2 , (1)
where subscripts 1 and 2 denote two phases, for example solid and liquid, and
superscript d means dispersion interactions. Using this equation and measuring n-
alkane/water interfacial tensions Fowkes determined the dispersion interaction for
water to be 21.8 mN/m, and hence the resulting polar (now interpreted as acid–base
hydrogen bonding [8–10]) interaction to be 72.8 − 21.8 = 51.0 mN/m. These values
are still commonly accepted.
Keeping in mind Lyklema’s [5] criticism, one would ask whether there is any
sense in calculating surface free energy of a solid using known approaches, and the
interpretation of measured (apparent) contact angles is even more problematic. On
the other hand, knowledge of solid surface free energy is a valuable quantity for
prediction of many interfacial processes, especially those depending on wetting. In
our opinion, such calculations can still be useful, but one should be aware that the
calculated values are apparent ones, similarly like the measured contact angles on
real surfaces, both the advancing and receding, are also mostly apparent. It is also
worth mentioning a review article published recently by Etzler [11], in which the
theoretical approaches and experimental determination of solid surface free energy
are reviewed together with some other ‘supporting’ methods that “can shed consid-
erable insight into the nature of the surfaces”, with 137 references.
In this paper we would like to compare the surface free energy results for several
solid surfaces evaluated from the known approaches, being aware of the problems
associated with their use. In one of the approaches both advancing and receding
contact angles are utilized. The contact angles were measured on glass, silicon,
mica and PMMA by sessile drop and tilted plate methods using three probe liquids
water, formamide and diiodomethane. The surfaces investigated were molecularly
flat. We were interested in finding whether these two methods of the advancing
and receding contact angle measurements gave compatible results. In the published
literature there are not many papers in which comparison of the energy values has
been done for the same solids and using the contact angles measured by different
methods. The results should answer the question whether all known approaches are
suitable to satisfactorily evaluate the apparent surface free energy of an investigated
solid.
Comparison of Apparent Surface Free Energy 285
2. Brief Descriptions of the Approaches Used

2.1. Owens and Wendt Approach
One of the approaches for solid surface free energy determination which is still
found in the published papers is that of Owens and Wendt [12]. It is based on
extended Fowkes’ equation (1) and Young–Dupré equation (2):
WA = γl (1 + cos θ ), (2)
where WA is the work of adhesion of probe liquid (l) to solid surface (s) and θ is
the liquid advancing contact angle. Hence:
p p
γl (1 + cos θ ) = 2(γsd γld )1/2 + 2(γs γl )1/2 , (3)
where the superscript d means dispersion interactions and p means polar interac-
tions, which can be considered as dipole–dipole or hydrogen bonding. To determine
p
the components γsd and γs of solid surface free energy, contact angles of apolar
(diiodomethane) and polar (water or formamide) liquids have to be measured and
then the two equation (3) obtained can be solved simultaneously, thus obtaining the
components and the total surface free energy as the sum of the two components.
2.2. Neumann’s Equation of State Approach
It is a well known equation, which has been criticized in the literature [13, 14], but
it is also used for solid surface free energy determination, especially of polymeric
materials [15, 16]. This equation is expressed as follows:

γs −β(γ −γ )2
cos θ = −1 + 2 e l s , (4)
γl
where β = 0.000125 (m2 /mJ)2 is an experimental constant. In this approach no
surface free energy components are considered.
2.3. Lifshitz–van der Waals Acid–Base (LWAB) Approach of van Oss, Chaudhurry
and Good
These authors [8–10] in the late eighties of the past century offered a new theoretical
approach to describe to surface and interface free energy, where free energy of a sur-
face i is expressed as a sum of two components, apolar Lifshitz–van der Waals γiLW
and polar Lewis acid–base γiAB , which is determined by two parameters electron
donor γi− and electron acceptor γi+ . In most cases the acid–base interaction is due
to hydrogen bonding. They assumed that γiAB could be expressed as the geometric
mean of γi− and γi+ parameters and included the dipole–dipole, and dipole–induce
dipole (if present) into the γiLW component in which the principal interaction is
London dispersion. In fact, experimentally only this latter interaction is determined
from the contact angle of an apolar liquid, mostly diiodomethane. Thus for a solid
surface its total surface free energy can be written as:
γstot = γsLW + γsAB = γsLW + 2(γs− γs+ )1/2 (5)
and for solid/liquid interfacial free energy the equation reads:

γsl = γs + γl − 2(γsLW γlLW )1/2 − 2(γs+ γl− )1/2 − 2(γs− γl+ )1/2 . (6)
Next, taking into account equation (2), and the definition of work of adhesion WA :
WA = γs + γl − γsl (7)
and the advancing contact angles θa one obtains:
γl (1 + cos θa ) = 2(γsLW γlLW )1/2 + 2(γs+ γl− )1/2 + 2(γs− γl+ )1/2 . (8)
Then three equations of type (8) can be solved simultaneously and the solid total
surface free energy can be obtained from equation (5) if contact angles of three
probe liquids are measured and their surface tension components are known. Van
Oss et al. [8–10] arbitrarily assumed for water at room temperature equal acidic and
basic interactions, i.e., γl− = γl+ = 25.5 mN/m, while γlLW = 21.8 mN/m. Then the
components of surface tension for other probe liquids were determined via contact
angle and/or interfacial tension measurements. Using this approach in most cases
the electron donor interaction γs− is determined to be much greater than γs+ , which
has caused heated discussion and other values for probe liquids surface tension
components [17–21] have been suggested. Besides other debatable issues, both in
Owens and Wendt’s [12] and van Oss et al.’s [8–10] approaches it is assumed that
the strength of the given kind of interactions from the solid side is the same indepen-
dent of the strength of probe liquid interactions, which, it seems, can be questioned.
Anyway, at present van Oss et al.’s [8–10] procedure is often used for evaluation of
a solid surface free energy.
2.4. Contact Angle Hysteresis (CAH) Approach
Recently Chibowski [3, 17–19] derived an equation for evaluation of total surface
free energy of a solid from advancing θa and receding θr contact angles of only one
probe liquid:
γl (1 + cos θa )2
γstot = . (9)
(2 + cos θr + cos θa )
Thus the evaluated surface free energy of a given solid depends, to some extent,
on the kind of probe liquid used. However, if several probe liquids are used the
averaged value (arithmetic mean) of the solid surface free energy agrees perfectly
with the mean value of total surface free energy determined from LWAB approach
if several triads of probe liquids are used for the contact angle measurements and
the calculations [18]. If one considers that the equilibrium contact angle θe is that
when no hysteresis appears, i.e. θa = θr = θe , then from equation (9) it results that
[19]:
γl WA
γstot = (1 + cos θe ) = . (10)
2 2
With this assumption, if the equilibrium contact angle of this liquid is known in-
deed, the solid surface free energy equals half of the work of adhesion of this probe
liquid to the solid surface [19].
2.5. Evaluation of Equilibrium Contact Angle from Tadmor’s Equation
In addition to experimental procedures to determine the equilibrium contact angle
[3], it can also be evaluated from the measured advancing and receding contact
angles as derived by Tadmor [1, 4].

−1 a cos θa + r cos θr
θo = cos , (11)
a + r
where according to the author’s denotations: θo , θa and θr are the equilibrium, ad-
vancing and receding contact angles, respectively, and a and r are defined as
follows:
1/3 1/3
sin3 θa sin3 θr
a ≡ , r ≡ .
(2 − 3 cos θa + cos3 θa ) (2 − 3 cos θr + cos3 θr )
Equation (11) has been derived from a combination of Young and Wenzel equations
and recognizing that the equilibrium contact angle results from the global energy
minimum in the system [1, 4]. The advancing and receding contact angles result
from pinning of the three-phase contact line, resisting motion of the drop. He as-
sumed that the resistance to the motion out for advancing drop was just equal to
the resistance to the motion in of the receding drop. This is because both of these
resistances are due to the three-phase contact line pinning to the similar protrusions
[1]. In other words, the irregularities on the surface are isotropic with respect to
their nature and distribution [1, 4]. However, this is rather a weak assumption in
the derivations. We will calculate the equilibrium contact angles from equation (11)
and then use them in calculations of the solid surface free energy both from LWAB
(equation (8)) and CAH (equation (10)) approaches.
3. Experimental
3.1. Materials
Contact angles were measured on 20 mm × 30 mm plates of glass (Comex, Poland),
silicon (Semiconductor Co., Czech Republic), mice sheets and PMMA. The glass
and silicon plates were washed successively in methanol and acetone in an ultra-
sonic bath, then rinsed with Milli-Q water and dried at 100◦ C. The mica sheets used
for the measurements were detached from a larger mineralogical specimen right be-
fore the measurements. The PMMA plates after removing the protective foil were
washed for 15 min in 20% aqueous methanol solution in an ultrasonic bath, then
rinsed with Milli-Q water. All the plates before using them for the experiments were
kept in a desiccator at room temperature. The AFM images and the histograms of
the surfaces are shown in Fig. 1.
(a)
(b)
(c)
Figure 1. 3D AFM images (left) of glass (a), silicon (b), mica (c), and PMMA (d) surfaces and
topography of the surfaces. Rrms roughness and the average heights of the surfaces are given in the
figures (right).
As the probe liquids we used: water from Milli-Q system, formamide (Fluka,
>99%) and diiodomethane (POCh Co., Poland, p.a.). Their surface tension and its
components, as well as the molecule’s volume and its diameter (calculated from
(d)
Table 1.
Surface tension and its components in mN/m of the probe liquids used and volume (Å3 ) and diameter
(Å) of the molecules
Liquid γL γLLW γL+ γL− γLAB Vmolec. , Å3 d, Å
Water 72.8 21.8 25.5 25.5 51.0 29.9 3.8

Formamide 58.0 39.0 2.28 39.6 19.0 66.1 5.0
Diiodomethane 50.8 50.8 0 0 0.0 133.9 6.3
molecular volume of the liquid and Avogadro’s number and assuming spherical
shape of the molecule) are presented in Table 1.
3.2. Methods

Contact angles were measured using a contact angle meter (GBX France) equipped
with a video camera with sessile drop and tilted plate methods. In the sessile drop
method 6 µl droplet of the probe liquid was gently deposited on the surface by an
automated deposition system and advancing contact angle was measured. Then 1 µl
of the liquid was withdrawn into the syringe and the receding contact angle was
measured.
3.2.2. AFM Images

10 mm × 10 mm size plates were investigated using a Nanoscope (Veeco, USA)
atomic force microscope (AFM) with standard silicon tip and contact mode. The
images were obtained at room temperature and in open atmosphere. Then the rough-
ness of the surfaces were analyzed using WSxM 4.0, Develop 8.0 software [20].

4.1. Topography of the Surfaces
Figure 1a–d shows 3D (top view, zero angle inclination) AFM images (left) and
histograms (right) of the investigated surfaces. As can be seen the most flat surface
is that of silicon (Fig. 1b) whose average roughness height is 0.43 nm and the Rrms
(root mean square roughness) roughness amounts to 0.12 nm. The silicon surface
roughness distribution is quite symmetrical around 0.43 nm and it ranges between
0–0.8 nm. The greatest height, 4.91 nm (Rrms = 0.88 nm) is found for the PMMA
surface, and it is distributed between 2 to 8 nm (Fig. 1d). One would expect mica
surface to be a very flat one. But, as seen in Fig. 1c in the 3D image there are
small patches on the smooth large mica sheet surface. Hence, the average height is
2.35 nm (Rrms = 0.72) and the protrusion height ranges from 0 up to 5 nm, but 80%
of them lie between 1.2 and 3.5 nm. As for the glass plate surface (Fig. 1a), the
average height of the protrusions is 2.59 nm (Rrms = 0.95 nm) and higher protru-
sions than their average height are seen in the histogram in significant amounts. The
protrusions can affect contact angle hysteresis of the probe liquids, i.e. the bigger
the height and less dense their distribution the larger the contact angle hysteresis
[21–23]. But, this may only occur if the contact angle is measured for the same liq-
uid on the same solid surface but possessing different roughnesses. However, Öner
and McCarthy [24] show that the contact angle of water on silane-modified silicon
surface (dimethyldichlorosilane (DMDCS) possessing 2–32 µm2 square posts, with
a height of 40 µm, the contact angle hysteresis is almost similar 26–35◦ , while it is
only 5◦ on the flat surface. But, the hysteresis increases sharply to 58◦ if the post
size is 64 µm2 and again drops down to 36◦ if the posts are 128 µm2 . Moreover,
the same authors [24] found that on the surface with 16 × 16 µm or 32 × 32 µm
square posts, with heights between 20 and 120 µm, water contact angle hysteresis
was similar. These contact angles were well above 100◦ and even up to 176◦ . The
protrusions of the surfaces studied here by us are of only few nanometers scale,
therefore the surfaces can be regarded as molecularly flat. As can be seen in Table 1
the largest molecular volume is for diiodomethane 134 Å3 , while that of water is
only 30 Å3 , and that of formamide 66 Å3 . Therefore, one can expect that water can
much easily penetrate between the protrusions than diiodomethane or formamide.
On the other hand, the differences in the diameter of the molecules, if considered
as spheres, are not as great (Table 1). However, it should be kept in mind that the
liquid molecule penetration depends on its surface tension as well as on whether
any polar interactions (e.g. acid–base) occur from the solid side.
4.2. Advancing and Receding Contact Angles of the Probe Liquids
The advancing and receding contact angles of the probe liquids measured on the
tested solid surfaces by sessile drop and tilted plate methods are shown in Figs 2–5.
Moreover, the bars representing the advancing contact angles measured by both
methods are marked with horizontal solid lines which represent the values of
Figure 2. Advancing θa and receding θr contact angles of probe liquids on glass surface measured by
sessile drop (open bars) and tilted plate (filled bars) methods. Dashed lines on the advancing contact
angle bars show equilibrium contact angles calculated from equation (11). The vertical lines on the
bars show standard deviations.
Figure 3. Advancing θa and receding θr contact angles of probe liquids on silicon surface measured by
equilibrium contact angles calculated from Tadmor’s equation (11). As can be seen,
generally the standard deviations of the measured contact angles by both methods
are within ±2◦ , which is typical. On the same surface, the advancing contact angle
measured by the tilted plate is in most cases several degrees greater than that mea-
sured by the sessile drop. The exception is for water on mica (Fig. 4) and formamide
Figure 4. Advancing θa and receding θr contact angles of probe liquids on mica surface measured by
Figure 5. Advancing θa and receding θr contact angles of probe liquids on PMMA surface measured
by sessile drop (open bars) and tilted plate (filled bars) methods. Dashed lines on the advancing contact
on silicon (Fig. 3), where the advancing contact angle by the tilted plate is slightly
smaller. The respective receding contact angles measured by tilted plate are gener-
ally smaller (except for formamide on mica, Fig. 4) than those measured by sessile
Table 2.
Contact angle hysteresis of the probe liquids determined on the
investigated surfaces by sessile drop and tilted plate methods
Surface Hysteresis, degrees;

sessile drop/tilted plate
Liquid
Water Formamide Diiodomethane
Glass 11.2/16.9 7.9/14.4 9.8/17.5

Silicon 10.5/17.0 13.1/14.8 7.1/16.7
Mica 5.4/7.5 4.2/6.1 5.3/14.3
PMMA 13.4/17.4 14.8/19.5 6.5/11.5
drop. This means that the contact angle hysteresis is greater if determined by tilted
plate method. The actual values of the contact angle hysteresis measured by the two
methods are listed in Table 2. Interestingly, irrespective of the kind of investigated
surface and its roughness, the greatest differences between the extent of hysteresis
found by sessile drop and tilted plate methods appear for diiodomethane. That is,
for an apolar probe liquid interacting practically by the dispersion force only, which
amounts to γl ≈ γld = 50.8 mN/m and this is the highest value among the probe liq-
uids used, i.e. γld = 21.8 mN/m for water and γld = 39.0 mN/m for formamide
(Table 1). Hence, it can be speculated that this kind of interaction is principally
responsible for the differences in the contact angle hysteresis values measured by
these two methods. The hysteresis is considered to be due to the pinning and de-
pinning energies [1, 4, 24–31]. On the other hand, the diiodomethane molecule is
the largest, its volume is more than four times that of water and two times that of
formamide (Table 1). Hence, one would expect Cassie’s wetting mechanism and
the smallest contact angle hysteresis for this liquid. The smallest hysteresis for the
three probe liquids if measured by tilted plate method is found on mica surface, ex-
cept for diiodomthane (Table 2). From the above discussion, data in Tables 1 and 2,
and Fig. 1a–d, it can be concluded that there is no direct correlation between the
surface roughness, surface tension of the probe liquid and contact angle hysteresis,
although some other general features are observed.
4.3. Apparent Surface Free Energy of the Investigated Solids
Taking the measured contact angles (Figs 2–5) and applying the above described
approaches (equations (3), (4), (8), (9)) the apparent surface free energies of the
tested surfaces were calculated. They are shown in Figs 6–9. In the figures are also
marked arithmetic mean values of the energies calculated from the contact angles
measured by sessile drop and tilted plate methods and using the four approaches
(the solid lines). As is seen, for a particular surface the surface free energy cal-
culated from Owens and Wendt’s equation (equation (3)) is always the highest.
Figure 6. Apparent surface free energy of glass calculated from four different approaches as indicated
in the figure. The horizontal dashed lines show arithmetic mean values from the four approaches and
the solid lines represent the mean values excluding O–W approach.
Figure 7. Apparent surface free energy of silicon calculated from four different approaches as indi-
cated in the figure. The horizontal dashed lines show arithmetic mean values from the four approaches
and the solid lines represent the mean values excluding O–W approach.
Therefore, in Figs 6–9 are also marked the arithmetic mean values calculated by
neglecting the energy values obtained from this equation.
The equation of state (equation (4)) fails for PMMA surface (Fig. 9) giving the
energy lower by ca. 8 mJ/m2 than the mean value from the four approaches (or
6.5 mJ/m2 without O–W value). The values of the surface free energy plotted in
Figs 6–9 calculated from contact angle hysteresis (CAH) approach are the averaged
Figure 8. Apparent surface free energy of mica calculated from four different approaches as indicated
in the figure. The horizontal dashed lines show arithmetic mean values from the four approaches and
the solid lines represent the mean values excluding O–W approach.
Figure 9. Apparent surface free energy of PMMA calculated from four different approaches as indi-
cated in the figure. The horizontal dashed lines show arithmetic mean values from the four approaches
and the solid lines represent the mean values excluding O–W approach.
values (arithmetic means) of those determined separately from water, formamide

and diiodomethane contact angles hystereses for the given surface. It is worth not-
ing that the mean values of the surface free energy calculated from contact angles
measured by sessile drop and tilted plate methods are very close. Only for glass the
difference is 4 mJ/m2 and a higher value is calculated from the contact angles mea-
sured by sessile drop. For mica, irrespective of the approach used and contact angle
measurement method, the determined surface free energy values are the most con-
sistent (excluding O–W’s value) (Fig. 8). Taking into account the Rrms and average
height values for the investigated surfaces (Fig. 1a–d) and the surface free energy, it
can be stated that for mica no direct correlation is observed between these two para-
meters and its consistent surface free energy values (Fig. 8). The same is true for the
difference between the mean values of the surface free energy for glass determined
from sessile drop and tilted plate methods (Fig. 6). The smoothest surface is that of
silicon, while roughness of mica surface is comparable to that of glass. On the other
hand, the average height of PMMA surface is the largest, and for this surface the
values of surface free energy calculated are most scattered (Fig. 9). Similarly, there
is no clearly seen relation between the extent of the contact angle hysteresis and the
surface roughness.
4.4. Surface Free Energy of the Surfaces Calculated from Equilibrium Contact
Angles
Finally, the equilibrium contact angles calculated from Tadmor’s equation (equation
(11), Figs 2–5) were employed to determine the apparent surface free energy of the
investigated solids from CAH approach (equation (9)). The results are shown in
Fig. 10. The calculated apparent surface free energy for a given solid is practically
the same irrespective of the ‘origin’ of the equilibrium contact angle of the probe
liquid used (sessile drop or tilted plate). Consequently, also the ‘averaged’ values of
the apparent surface free energy (arithmetic mean of the values calculated from wa-
ter, formamide and diiodomethane equilibrium contact angles, marked by dashed
horizontal lines in Fig. 10) are practically the same. From the CAH approach the
highest apparent surface free energy values are those calculated from water and the
lowest those calculated from diiodomethane contact angles. A different behavior
is observed for PMMA surface, which possesses apolar γsLW interaction and only
relatively weak electron donor γs− interactions. Figure 11 shows the solid surface
free energy components calculated from LWAB approach (equation (8)) using the
equilibrium contact angles calculated from the advancing and receding contact an-
gles measured with a sessile drop. In case of PMMA apolar γsLW interaction is 48
mJ/m2 , which is the highest among the four solids studied, but its electron donor
parameter γs− is the smallest one, 20 mJ/m2 , and there is no electron acceptor in-
teraction γs+ [32–34]. These weak acid–base interactions might be the reason that
the apparent values of PMMA surface free energy calculated from CAH approach
and the equilibrium contact angles are similar, irrespective of the probe liquid used,
Fig. 10.
Figure 10. Apparent surface free energy of the investigated solids calculated from equilibrium contact
angles of water (W), formamide (F) and diiodomethane (D) (CAH approach, equation (10)). The
horizontal dotted lines show averaged apparent surface free energy values, which are arithmetic means
of the values calculated separately from the three probe liquids. Open bars — the equilibrium contact
angles were calculated from the advancing and receding contact angles measured by sessile drop
method and the filled bars — those measured by tilted plate method.
Figure 11. Surface free energy components of the solids calculated from equilibrium contact angles
obtained from the contact angles measured by sessile drop method and using LWAB approach.
5. Summary and Conclusions

Four different solid surfaces (glass, silicon, mica and PMMA) and two methods
for the advancing and receding contact angle measurements of three probe liquids
(water, formamide and diiodomethane) were used. Then four different approaches
for the determination of the surface free energy were employed in calculations of
the apparent surface free energy. The results show that:
– Higher values of the contact angle hysteresis are obtained if tilted plate method
is applied, which is a result of generally slightly higher advancing and lower
receding contact angles if measured by tilted plate than by sessile drop method.
– The highest values of calculated total surface free energy from the advancing
contact angles are obtained from Owens and Wendt’s equation.
– CAH, LWAB and equation of state (EQS) give comparable values of the total
surface free energy, except for PMMA surface for which EQS fails, leading to
significantly lower value.
– Both techniques for the advancing and receding contact angles measurements
can be applied for the apparent surface free energy determination from CAH ap-
proach. Comparable values of surface free energy are then obtained. The same
is true if the surface free energy is determined via equilibrium contact angles
calculated from Tadmor’s equation.
– The results obtained lead to the conclusion that only apparent surface free
energy and its components can be determined from the advancing or advanc-
ing/receding contact angles of probe liquids.
– It is hypothesized that the strength of interaction originating from the solid
surface may vary depending on the strength of the same kind of interaction orig-
inating from the liquid surface. In other words, surface free energy of a solid as
determined via contact angle is not an absolute value, but it varies depending
on the kind of probe liquid used.
Acknowledgement
Financial support by the Polish Ministry of Science and Higher Education, project
N204 130435 is gratefully acknowledged.
References
1. R. Tadmor, Langmuir 20, 7659 (2004).
2. P. Letellier, A. Mayaffre and M. Turmine, J. Colloid Interface Sci. 314, 604 (2007).
4. R. Tadmor and P. S. Yadav, J. Colloid Interface Sci. 317, 241 (2008).
5. J. Lyklema, Colloids Surfaces A 156, 413 (1999).
6. F. M. Fowkes, J. Phys. Chem. 66, 382 (1962).
8. C. J. van Oss, R. J. Good and M. K.Chaudhury, J. Colloid Interface Sci. 111, 378 (1986).
9. C. J. van Oss, R. J. Good and M. K. Chaudhury, Langmuir 4, 884 (1988).
10. C. J. van Oss, M. K. Chaudhury and R. J. Good, Chem. Rev. 88, 927 (1988).
11. F. M. Etzler, in: Contact Angle, Wettability and Adhesion, Vol. 3, K. L. Mittal (Ed.), pp. 219–264.
12. D. K. Owens and R. C. Wendt, J. Appl. Polym. Sci. 13, 1741 (1969).
13. I. D. Morrison, Langmuir 5, 540 (1989).
14. B. Janczuk, J. M. Bruque, M. L. Gonzalez-Martin, J. Moreno del Pozo, A. Zdziennicka and
F. Quintana-Gragera, J. Colloid Interface Sci. 181, 108 (1996).
15. C. A. Ward and A. W. Neumann, J. Colloid Interface Sci. 49, 286 (1974).
16. H. Tavana, G. Yang, C. M. Yip, D. Appelhans, H. Zschoche, K. Grundke, M. L. Hair and A. W.
Neumann, Langmuir 22, 628 (2006).
17. E. Chibowski, in: Contact Angle, Wettability and Adhesion, Vol. 2, K. L. Mittal (Ed.), pp. 265–288.
20. http://www.nanotec.es.
21. J. Long, M. N. Hyder, R. Y. M. Huang and P. Chen, Adv. Colloid Interface Sci. 118, 173 (2005).
22. B. Krasovitski and A. Marmur, Langmuir 21, 3881 (2005).
23. E. Bormashenko, R. Pogreb, G. Whyman and M. Erlich, Langmuir 23, 6501 (2007).
24. D. Öner and T. J. McCarthy, Langmuir 16, 7777 (2000).
25. H. Kusumaatmaja and J. M. Yeomans, Langmuir 23, 6019 (2007).
26. D. Quéré, A. Lafuma and J. Bico, Nanotechnology 14, 1109 (2003).
27. L. Gao and T. J. McCarthy, Langmuir 22, 6234 (2006).
28. L. Gao and T. J. McCarthy, Langmuir 22, 2966 (2006).
29. G. McHale, Langmuir 23, 8200 (2007).
30. P. M. Harder, T. A. Shedd and M. Colburn, J. Adhesion Sci. Technol. 22, 1931 (2008).
31. G. Whyman, E. Bormashenko and T. Stein, Chem. Phys. Lett. 450, 355 (2008).
32. C. J. van Oss, R. F. Giese and W. Wu, J. Adhesion 63, 71 (1997).
33. E. Chibowski, A. Ontiveros-Ortega and R. Perea-Carpio, J. Adhesion Sci. Technol. 16, 1367
(2002).
34. H. Radelczuk, L. Hołysz and E. Chibowski, J. Adhesion Sci. Technol. 16, 1547 (2002).
Surface Free Energy of Viscoelastic Thermal
Compressed Wood
Marko Petrič a,∗ , Andreja Kutnar b , Borut Kričej a , Matjaž Pavlič a ,

Frederick A. Kamke c and Milan Šernek a
a
University of Ljubljana, Biotechnical Faculty, Dept. of Wood Science & Technology,
Jamnikarjeva 101, SI 1000 Ljubljana, Slovenia
b
University of Primorska, Primorska Institute for Natural Sciences and Technology, Muzejski trg 2,
SI 6000 Koper, Slovenia
c
Oregon State University, Department of Wood Science and Engineering, 119 Richardson Hall,
Corvallis, OR 97331-5751, USA
Abstract
Surface free energy and its components of a viscoelastic thermal compressed (VTC) wood were evaluated by
the Lifshitz–van der Waals/acid–base approach and the results compared to those calculated by the Owens–
Wendt–Rabel–Kaelble method. We observed the decrease of the total surface free energy and its dispersion
component as a consequence of VTC densification process. The surface free energy values depended on the
evaluation method used, so one should consider them only relatively to assess the influence of the densi-
fication process. In contrast to increased contact angles of water on VTC wood, the phenol-formaldehyde
adhesive exhibited better wetting on VTC wood than on the untreated substrate.
Keywords
Poplar wood, viscoelastic thermal compression, Lifshitz–van der Waals/acid–base approach, surface free
energy, phenol-formaldehyde adhesive, contact angle
1. Introduction
Wood is a valuable renewable resource, used in numerous applications. However, it

has some less desirable properties, such as vulnerability against insect and fungi
attack, dimensional instability, relatively low resistance against weathering, etc.
There are many ways to overcome the negative characteristics of wood. One pos-
sibility is to produce new wood-like materials using wood just as a raw material.
For instance, there are numerous reports on the production, characteristics and ap-
plications of chemically or thermally-treated wood (e.g. [1] or [2]). In general,
*To whom correspondence should be addressed. Tel.: +386 1 257-2285; Fax: +386 1 257-2285; e-mail:
marko.petric@bf.uni-lj.si

302 M. Petrič et al.
modified wood exhibits improved dimensional stability and increased resistance

against wood pests. The poor mechanical properties of low-density wood can be
modified and improved by wood densification processes. So, Kamke and Sizemore
[3] developed a method for wood densification using a viscoelastic thermal com-
pression (VTC) process.
All new wood-like materials are used in the same way as normal, unprocessed
wood. This means that they should allow surface finishing with commercial wood
coatings as well as bonding with common wood adhesives. However, surface char-
acteristics (surface free energy, surface morphology) of wood-based materials could
be different from those of untreated wood and might affect bonding and surface fin-
ishing. Consequently, it is necessary to study the influence of exposure of wood to
heat, chemicals, compression, etc., on its wettability characteristics and surface free
energy.
Wettability can be determined by contact angle measurements. The Wilhelmy
plate method was shown to be especially appropriate for such a heterogeneous and
porous material as wood [4]. There are various approaches to determine the wood
surface free energy and its components. The most common ones are those of Zis-
man, Owens–Wendt–Rabel–Kaelble and van Oss–Chaudhury–Good [5–7].
The objective of the present study was to evaluate the influence of viscoelastic
thermal compression of poplar wood on its surface free energy and its components,
as calculated by the Lifshitz–van der Waals/acid–base approach and to assess the
wettability of VTC wood by phenol-formaldehyde (PF) adhesive.
2. Experimental
Low density hybrid poplar wood (Populus deltoides × Populus trichocarpa), from
a plantation in Northwest Oregon, was densified using the VTC process, as already
described [8]. The VTC process uses a combination of steam, heat and mechanical
compression. The process begins with exposure of the test specimen to saturated
steam pressure at 175◦ C, and continues with the venting phase and compression
to the target thickness. After final compression, the temperature of the platens is
raised from 175◦ C to 200◦ C in order to allow relaxation of the remaining stresses,
and to stabilize the VTC product. The densification process finishes with cooling
under compression to 100◦ C. The duration of the VTC process was 15 min. VTC
specimens with three different degrees of densification (63%, 98% and 132%) were
prepared. The control and VTC specimens were examined without further surface
preparation. All tested surfaces were one year old. The PF adhesive for contact
angle measurements was a commercial product, supplied by The Georgia-Pacific
Resins, Inc. (Decatur, GA, USA) (viscosity 330 mPa s, about 50% solid matter).
After conditioning to equilibrium at 20◦ C and 65% RH, the control samples had a
moisture content (MC) of 12%, whereas the VTC specimens had 7% MC.
The apparent advancing contact angles of water (distilled in our laboratory), for-
mamide (99 + %, supplied by Kefo Slovenia), diiodomethane (99 + %, supplied by
Surface Free Energy of VTC Wood 303
Kefo Slovenia) and the PF adhesive were determined by the Wilhelmy plate method
(K100 Krüss Processor Tensiometer). The measurements with the wood plates (10
mm × 24 mm × (2.5 or 4) mm, longitudinal × tangential-radial × tangential-
radial directions) were performed at the following conditions: 20◦ C, RH 40–50%,
and immersion velocity of 12 mm/min. Each test specimen was immersed along the
length 24 mm (tangential-radial direction) into the liquid, and then withdrawn from
it. The determined surface free energy was, therefore, an average of surface free
energies on the tangential-radial direction. If the measurement was done on another
direction, the determined surface free energy could be different.
The surface free energy of control and VTC wood (γtot ) and its components
(the dispersion (or LW, Lifshitz–van der Waals) component γSV LW , the acid–base
component γSV AB with its electron-accepting and electron-donating parameters γ +

SV
−
and γSV ) were evaluated by the Lifshitz–van der Waals/acid–base approach from
contact angles of water, formamide and diiodomethane. The calculation procedure
can be found elsewhere, for example in [9].

3.1. Surface Free Energy of VTC Wood
As can be seen from Table 1 [8], modification of poplar wood by the VTC densifi-
cation process affects its surface wettability. The densification process increases its
hydrophobic character.
It appears that the hydrophobic character of the VTC wood samples is a con-
sequence of the chemical changes in the wood due to hydrothermal conditioning
during the densification process and the related reduction in the surface free energy
of densified wood, as suggested by Jennings et al. [10]. In addition, it has to be
taken into account that the contact angle can also be influenced by the wood surface
morphology [11–14]. The control wood surface is rougher and has greater porosity
than the VTC wood, which could result in lower contact angle of the test liquids on
the control wood. Furthermore, the VTC process significantly alters the morphol-
ogy of the VTC wood [15]. Densification to a higher degree is achieved by a larger
Table 1.
Average (n = 10) advancing contact angles on poplar wood, ob-
tained with the different test liquids (the standard deviation is
shown in parentheses) [8]
Contact angle (◦ )
Sample Water Formamide Diiodomethane
Control 71.6 (7.6) 38.7 (19.6) 5.4 (9.2)

VTC 63% 90.9 (5.7) 79.2 (6.6) 48.5 (6.5)
VTC 98% 93.6 (10.3) 75.4 (3.7) 60.9 (7.0)
VTC 132% 93.1 (6.2) 81.8 (5.0) 58.0 (5.17)
reduction in the void spaces of the wood. Therefore, it was expected that the contact
angles of the test liquids would depend on the degree of densification. However, the
results revealed that the degree of densification had only minor and limited effect
on the contact angle of the VTC wood.
Indeed, the surface free energy of VTC densified poplar wood decreased from
the original value of 53.3 mJ/m2 to 26–30 mJ/m2 due to the densification process,
as calculated by the Lifshitz–van der Waals/acid–base approach (Table 2). The to-
tal surface free energy of the control wood was relatively close to the data in the
literature [16]. The decrease in the surface free energy of the VTC wood was a
consequence of the increased hydrophopic character of heat-treated wood due to
hemicellulose’s degradation and reorganization of the lignocellulosic polymeric
components of wood [9]. Apparently, a larger reduction in the void spaces of
the wood due to higher degree of densification did not affect the surface free en-
ergy.
For all groups of specimens, the Lifshitz–van der Waals (γSVLW ) component was
the major surface free energy component. It should be noted that when the sur-
face free energy of the VTC 63% and VTC 132% was evaluated, the square root
+
of the electron-acceptor component γSV became negative, which is a consequence
of the mathematical formalism of the LW-AB theory. Shalel-Levanon and Marmur
[17] analyzed the validity and accuracy of the LW-AB equation from a mathemat-
ical point of view. Their study determined the conditions for an infinite number of
+ −
solutions and for no mathematical solution, when γSV < 0 and γSV < 0. An infi-
nite number of solutions were obtained if liquids with similar properties were used.
The allowed range of contact angles that results in physical meaningful solutions
(positive values of square roots of the surface free energy components of the solid)
was determined. Furthermore, the research found that the allowed range becomes
significantly narrower when the chosen liquids have more similar properties.
Table 2.
Surface free energy and its components (in mJ/m2 ) of the control
and VTC poplar wood specimens obtained by the Lifshitz–van
der Waals/acid–base approach
LW AB + −
Sample γSV γSV γSV γSV γtot
Control 50.6 2.75 5.4 0.4 53.3

VTC 63%* 35.1 5.55 7.1 1.1 40.7 (29.6)
VTC 98% 28.1 0.33 3.2 0.0 28.4
VTC 132%* 29.7 3.81 6.4 0.6 33.6 (25.9)
* The corrections were used in the evaluation. The values in
parentheses were obtained using the proposal of de Meijer et al.
+
[16], the negative γSV value was considered as a negative contri-
bution to the total surface free energy.
Experimental determination of surface free energy of solids and the problem of

negative square roots has been discussed in many studies [17–22]. Della Volpe and
Siboni [18] proposed the best-fit approach. The method considers large sets of liq-
uid/solid pairs with a corresponding large system of algebraic equations which is
solved by means of a best-fit algorithm. Furthermore, Greiveldinger and Shana-
han [19] inverted the method and used the assumed solid surface characteristics
to “redeliver” surface free energy parameters of probe liquids. Based on the ob-
tained inconsistencies, which indicates inadequacies in the mathematical treatment
of the LW-AB method, this study suggested that theoretical developments were re-
quired.
Michalski et al. [20] discussed negative square roots in the LW-AB method and
interpreted them based on the literature review. The study disagrees with the con-
+
sideration of Good and van Oss [21] that a negative value γSV was allowable, if
the total surface free energy was positive. Also, the study did not find suitable the
suggestion of Good [22], who affirmed that the negative values can be considered
as an artefact and simply assume that they are zero. Therefore, they chose to handle
these electron-acceptor parameter data simply by taking them as the square of the
negative square root.
+
Also in the present study, the γSV was taken as the square of the negative square
root (which always yields positive values). De Meijer et al. [16] considered this
+
negative γSV value as a negative contribution to the total surface free energy and
their proposal was adopted also in our calculations. The surface free energies ob-
tained using the proposal of de Meijer et al. [16] are shown in parentheses in Ta-
ble 2.
Kutnar et al. [8] reported data on the surface free energy of VTC wood, ob-
tained by the Owens–Wendt–Rabel–Kaelble (OWRK) approach using contact angle
data of formamide and diiodomethane. Similarly to the results presented here, they
stated that the densification process considerably lowered the total surface free en-
ergy of the VTC specimens. Both the dispersion and the polar components of the
surface free energy were higher in the case of the control wood than the VTC wood.
Previous studies on the surface free energy of heat treated wood have, however, re-
ported a slight increase in the dispersion component after heat treatment [9, 10].
The decrease of the dispersion component of the surface free energy of the VTC
wood is most likely the result of surface ageing, as mentioned earlier.
When the surface free energy data in Table 2 are compared with the values
obtained by the OWRK approach [8], we can see some differences in calculated
surface free energies of the VTC 63% and VTC 132% specimens. When the pro-
posal of de Meijer et al. [16] was adopted higher OWRK values of 35.5 mJ/m2 and
+
29.8 mJ/m2 , respectively, were reported. On the contrary, when the negative γSV
values were not considered as a negative contribution to the total surface free en-
ergy, the OWRK values were lower. On the other hand, the determined surface free
energies of the control and VTC 98% specimens were independent of the evaluation
theory used. Furthermore, the γSVLW was found to be the same as γ d determined by
SV
OWRK theory for all groups of specimens, which is due to the same primary data
used and the structure of the equations. The observed differences highlight the dif-
ferent extents of validity of different theories, applied to calculate the surface free
energy and its components. We strongly believe that in the case of such a heteroge-
neous and porous material as wood, one should not take the calculated values as the
absolute ones. Wood exhibits enormous variability due to the differences in its com-
position [23]. Hence also the magnitude of surface free energy and its components is
highly variable. De Meijer et al. [16] presented an overview of the literature on the
surface energy data obtained for various wood species and methods. The reported
variations in the surface free energy data can be explained by the complex nature
of the wood surface, which is the result of porous structure, surface roughness and
chemical heterogeneity of wood. Furthermore, the contact angle measurements are
influenced by the water adsorbed onto the cell wall, which is always present in
significant amounts, depending on the wood species and the relative humidity of
the environment [16]. It is much more important, therefore, to study and interpret
the changes in surface free energy as a consequence of the process to which the
specimens were subjected.
Additionally, there is a problem in experimental determination of the equi-
librium contact angle that results from minimum free energy in the investigated
solid/liquid/air system at constant conditions. The equilibrium contact angle, which
is expressed by the Young equation, should lie between the advancing and reced-
ing contact angles [24–26]. However, the utilization of the Wilhelmy method on
wood determines only the advancing contact angles, while the receding contact an-
gles are usually zero due to the liquid soaking into the wood. Rodriguez-Valverde
and coworkers [27], who measured contact angles on stone and wood, stated that
a sessile drop on a real surface shows different contact angle values due to its lack
of symmetry or to the loss of volume by capillary action. That is why they con-
ducted measurements using the dynamic technique, the so-called captive bubble
method. Wålinder [28] also showed that, because of the intricate nature of wood
(e.g. the porosity, hygroscopicity, anatomic complexity, heterogeneity, and the ex-
tractive content) and its surface, wetting measurements on wood may be inherently
difficult. The direct measurement of the contact angle of a drop deposited on the
wood surface is according to Wålinder an unsatisfactory approach from both ex-
perimental and statistical points of view. But he showed a successful application of
the Wilhelmy method. To conclude this part of our discussion, we believe that it is
correct to use dynamic contact angles instead of the static (thermodynamic) ones
for surface free energy assessment of wood, for the purpose of comparison within
the same set of wood samples.
3.2. Contact Angle of PF Adhesive on VTC Wood
Since the wettability study with the Wilhelmy plate method showed that the VTC
wood surface exhibited hydrophobic behaviour, the concern rose regarding poten-
tial increase of contact angle of certain adhesive types. Therefore, we measured
Table 3.
Multiple range test results with 95% LSD for the ad-
vancing contact angles of the PF adhesive on the con-
trol and VTC wood (the standard deviation is shown in
parentheses)
Count Mean (◦ )
VTC 132% 5 126.3 (3.0)
VTC 98% 5 127.8 (3.0)
VTC 63% 5 126.3 (5.6)
Control 5 133.7 (7.8)
Contrast Difference ± Limits
Control–VTC 63% 7.45* 7.08

Control–VTC 98% 5.91 7.08
Control–VTC 132% 7.47* 7.08
VTC 63%–VTC 98% −1.53 7.08
VTC 63%–VTC 132% 0.026 7.08
VTC 98%–VTC 132% 1.56 7.08
* Denotes a statistically significant difference.
contact angles of the liquid PF adhesive in dependence of the degree of densifi-

cation of VTC poplar wood (Table 3). The multiple range test with the 95% LSD
procedure did not indicate statistically significant differences in the contact angles
among the VTC wood specimens with different degrees of densification. However,
the contact angle on the control wood was significantly higher than that obtained
on the VTC 63% wood and the VTC 132% wood. The results could be the conse-
quence of the contact angle measurements of the high viscous PF adhesive, where
it is difficult to assess the equilibrium. The air trapped in pores could account for
high values of contact angles with PF on the control wood. Since the pores of the
VTC wood are reduced, the effect of air trapped was less pronounced in the contact
angle measurements on the VTC wood surface.
Contrary to our expectation, based on the increase of water contact angles on
VTC wood, a slight decrease of contact angles of the PF adhesive was observed.
This is a promising and interesting result, opening possibilities of using the adhe-
sive, just as in the case of normal, unprocessed wood. Although heat-treated wood is
a different material than the VTC wood we used, it is interesting to note that a sim-
ilar observation with waterborne wood coatings on heat-treated wood was reported
in [29]. Contact angles of water on surfaces of oil heat-treated Scots pine wood were
substantially higher than on unmodified specimens. On the contrary, various water-
borne acrylic stains revealed much lower contact angles on the modified samples
than on the untreated ones. The study [29] did not elucidate the phenomenon, but
only stated that due to complex formulations of commercial wood finishes further
investigations were needed to clarify the observation mentioned. Thus we should
continue additional studies on wetting of VTC wood.
4. Conclusions
Viscoelastic thermal compressed poplar wood exhibited increased hydrophobicity
in comparison with the unprocessed wood. Calculations of the surface free energy
and its components of VTC wood, by the Lifshitz–van der Waals/acid–base ap-
proach, revealed a decrease in both the total surface free energy and its dispersion
component as a result of VTC densification process. In contrast to increased contact
angles of water on VTC wood, the phenol-formaldehyde adhesive exhibited better
wetting on VTC wood than on the untreated substrate.
References
1. H. Epmeier, M. Westin and A. O. Rapp, Scandinavian J. Forest Res. 19, 31–37 (2004).
2. R. M. Rowell, Wood Mater. Sci. Eng. 1, 29–33 (2006).
3. F. A. Kamke and H. Sizemore, US Patent No. 7,404,422 (2008).
4. M. E. P. Walinder and I. Johansson, Holzforschung 55, 21–32 (2001).
5. M. Gindl, G. Sinn, W. Gindl, A. Reiterer and S. Tschegg, Colloids Surfaces A 181, 279–287
(2001).
6. E. Chibowski and R. Perea-Carpio, Adv. Colloid Interface Sci. 98, 245–264 (2002).
7. M. Zenkiewicz, Polym. Test. 26, 14–19 (2007).
8. A. Kutnar, F. A. Kamke, M. Petrič and M. Sernek, Colloids Surfaces A 329, 82–86 (2008).
9. P. Gérardin, M. Petrič, M. Petrissans, J. Lambert and J. J. Ehrhrardt, Polym. Degrad. Stab. 92,
653–657 (2007).
10. J. D. Jennings, A. Zink-Sharp, C. E. Frazier and F. A. Kamke, J. Adhesion Sci. Technol. 20, 335–
344 (2006).
11. D. J. Gardner, Wood Fiber Sci. 28, 422–428 (1996).
12. G. I. Mantanis and R. A. Young, Wood Sci. Technol. 31, 339–353 (1997).
13. J. Bico, U. Thiele and D. Quere, Colloids Surfaces A 206, 41–46 (2002).
14. D. E. Packham, Int. J. Adhesion Adhesives 23, 437–448 (2003).
15. A. Kutnar, F. A. Kamke and M. Sernek, Wood Sci. Technol. 43, 57–68 (2009).
16. M. de Meijer, S. Haemers, W. Cobben and H. Militz, Langmuir 16, 9352–9359 (2000).
17. S. Shalel-Levanon and A. Marmur, J. Colloid Interface Sci. 262, 489–499 (2003).
18. C. Della Volpe and S. Siboni, J. Colloid Interface Sci. 195, 121–136 (1997).
19. M. Greiveldinger and M. E. R. Shanahan, J. Colloid Interface Sci. 215, 170–178 (1999).
20. M. Michalski, J. Hardy and B. J. V. Saramago, J. Colloid Interface Sci. 208, 319–328 (1998).
21. R. J. Good and C. J. van Oss, in: Modern Approaches to Wettability: Theory and Applications,
M. E. Schrader and G. I. Loeb (Eds), Chapter 1. Plenum Press, New York (1992).
22. R. J. Good, in: Contact Angle, Wettability and Adhesion, K. L. Mittal (Ed.), p. 3. VSP, Utrecht
(1993).
23. B. J. Zobel and P. van Buijten, Wood Variation: Its Causes and Control. Springer-Verlag, Berlin
(1989).
24. A. Marmur, Adv. Colloid Interface Sci. 50, 721–141 (1994).
25. A. Marmur, Colloids Surfaces A 136, 209–215 (1998).
26. E. Chibowski and K. Terpilowski, J. Colloid Interface Sci. 319, 505–513 (2008).
27. M. A. Rodriguez-Valverde, M. A. Cabrerizo-Vilchez, P. Rosales-Lopez, A. Paez-Duenãs and
R. Hidalgo-Alvarez, Colloids Surfaces A 206, 485–495 (2002).
28. M. Wålinder, Wetting phenomena on wood. Factors influencing measurements of wood wetta-
bility, Ph.D. Thesis, Department of Manufacturing Systems Wood Technology and Processing,
KTH-Royal Institute of Technology, Stockholm (2000).
29. M. Petrič, B. Knehtl, A. Krause, H. Militz, M. Pavlič, M. Pétrissans, A. O. Rapp, M. Tomažič,
C. Welzbacher and P. Gérardin, JCT Res. 4, 203–206 (2007).
Modification of Sugar Maple Wood Board Surface by
Plasma Treatments at Low Pressure
V. Blanchard a,b , B. Riedl b,∗ , P. Blanchet a,b and P. Evans c

a
FPInnovations — Forintek Division, 319 rue Franquet, Québec (QC) G1P 4R4, Canada
b
Université Laval — Centre de Recherche sur le Bois, Pavillon G.-H. Kruger,
Québec (QC) G1K 7P4, Canada
c
Centre for Advanced Wood Processing, 2900-2424 Main Mall, Vancouver (BC) V6T 1Z4, Canada
Abstract
The Canadian wood products industry is going through hard times. To face up to competition from emerg-
ing economies and substitution products, it needs to be innovative and develop the next generation of wood
products. Plasma technology could be used to improve wood surface properties, making them less sensitive
to environmental conditions (moisture, water, temperature, UV light). Thus, wood properties could be sig-
nificantly different between species or even amongst different samples of the same species. Over the last few
years, plastic and textile industries have begun experimenting with plasma technology to activate surfaces,
mainly to improve coating/substrate adhesion. The literature on potential applications of plasma treatment
to wood surfaces is very limited, however. This report describes the results of an exploratory study on the
effect of plasma treatments on sugar maple wood board surfaces using different gases and mixtures (N2 ,
H2 , O2 and Ar) at different pressures (13.3–665 Pa). Water wettability of the wood and adhesion between
waterborne polyurethane coatings and wood were also studied. The results show that it was possible, un-
der certain conditions, to significantly increase coating/wood adhesion (by 30–50%). This improvement was
correlated to the wood surface energy increase. The processes used were not directly transferable to industry
(treatment time, vacuum process), but they can easily be adapted at reasonable costs.
Keywords
Surface modification, plasma, adhesion, sugar maple, wettability
1. Introduction
In addition to the current world economic crisis, the Canadian wood industry faces
competition from emerging economies (China, Brazil, etc.) and substitution prod-
ucts (poly(vinyl chloride), aluminum). It therefore needs to innovate to remain
competitive at the global level. High technology such as plasma treatment might
* To whom correspondence should be addressed. Tel.: 418 656 2437; Fax: 418 656 5262; e-mail:
bernard.riedl@sbf.ulaval.ca

312 V. Blanchard et al.
fully transform this industry by imparting new functionalities to wood and enhanc-
ing its durability and surface properties.
Plasma is defined as an ionized gas whose global charge is neutral. It is often
considered the fourth state of matter, and constitutes more than 99% of visible mat-
ter in the universe. It is made up of electrons, radicals and ions capable of reacting
with matter, thus modifying the surface and structure of treated substrates. It can
generate low or high temperatures depending on the way it is activated and the
power applied, hence the reference to cold (or non-thermal) and thermal plasmas.
This wide temperature range allows for various laboratory and industrial applica-
tions, such as hospital tool sterilization, surface modification, thin film deposition,
chemical synthesis, surface chemical analysis and more.
Over recent years, atmospheric plasmas have been developed that are now avail-
able for industry use. This technique is already used in the plastic and glass in-
dustries to improve adhesion between coating and substrate by modifying substrate
surface energy or/and grafting new chemical functions. The textile industry has
been developing new surface techniques based on plasmas in order to obtain fibers
with better resistance to industrial washing cycles [1].
So far, only a few studies have demonstrated the potential of plasma technology
as a means to alter wood surfaces and modify surface energy [2]. However, some
applications have been investigated. Podgorski and coworkers [3, 4] studied the
influence of various plasma treatments on several fir species. They showed that coat-
ing adhesion could be improved under specific conditions. Rehn et al. [5] showed
that the fracture strength of glued black locust (Robinia pseudoacacia) could be
increased and coating delamination reduced. The fire and moisture resistance of
different Philippine species was improved after hydrogen plasma treatment by Blan-
tocas et al. [6]. Evans et al. [7] investigated the impact of plasma treatment on the
wettability and glue bond strength of four eucalypt wood species. Finally, Wolken-
hauer and coworkers studied several properties of wood after plasma treatments
[8–11]. In 2009, they demonstrated [8] that a plasma treatment of wood surfaces
was superior to sanding for increasing surface energy. Other, more complete, re-
sults (XPS, Confocal Raman spectroscopy, etc.) have been obtained and will be
published shortly [12].
2. Plasma Basics
A plasma is created by the application of high energy to a gas to reorganize the

electronic structure of species contained in order to create excited species, radicals
and ions. The energy needed to achieve it can be thermal or carried by electric
current or electromagnetic radiation [13, 14].
Non-thermal plasmas in Direct Current (DC) or Radio Frequency (RF) dis-
charges are generally created in closed discharge vessels using electrodes or coils.
The characteristics of the resulting plasma depend on either voltage or current ap-
Sugar Maple Wood Surface Modification by Plasmas 313
plied and frequency of alternating current. Generally, three kinds of discharges are
used to generate plasmas: DC glow, RF inductive and RF capacitive.
In most cases, the mechanism leading to plasma development is independent of
the type of discharge used to initiate the gas ionization reaction. The infinitesimal
amount of free electrons present in any gas (due to UV and/or cosmic radiation)
is excited under a magnetic or electric field, causing collisions with neutral mole-
cules. The ionization of molecules is followed by the release of valence electrons
which cause new collisions to form ions, radicals or “metastable” particles. Plasma
is generated when the amount of electrons is high enough to form visible radiation.
A DC glow discharge results from the application of a direct current between
two electrodes to create an electric field that accelerates free electrons to initiate
collisions. When the amount of free electrons in the plasma is sufficient, secondary
electrons begin to be released from the cathode by electron bombardment. Plasma
energy is thus increased. The development of electric arcs must be avoided, as they
make the plasma unstable with this type of discharge. For this reason, the operation
requires a resistor in series, or it has to be interrupted for a short period of time to
prevent arcing.
An inductive discharge is obtained through the application of an alternating cur-
rent to a conductor that creates an electric field. The magnetic field thus induced
generates an electric field that accelerates free electrons within the gas, thereby ini-
tiating the plasma. This type of plasma is obtained through the use of a coil or wire
located either outside or inside the plasma volume. The unit used for this study is
shown schematically in Fig. 1.
Figure 1. Schematic of inductive plasma apparatus used.

Figure 2. Schematic of RF capacitive plasma apparatus used.
A capacitive discharge results from the application of an alternating current be-

tween two electrodes in order to obtain a potential difference and create an electric
field to excite free electrons. A plasma is thus generated as previously described.
A reactor of this kind can be used to deposit thin films by magnetron sputtering
plasma. In simple terms, an inorganic or organic target is placed on a cathode sur-
rounded by a magnetic field. As a result, electrons are confined around the target
and can intensively bombard it until they pull up target atoms that are then ionized
and deposited on the substrate located under the plasma cloud. Ions are not affected
by a magnetic field and store kinetic energy from collisions with electrons. The
unit used to carry out RF capacitive plasma treatments and magnetron sputtering is
presented schematically in Fig. 2.
A conventional RF system capable of generating capacitive and inductive dis-
charges consists of a generator combined with an impedance matching network and
a reactor with electrodes. The impedance matching network should be such that
reactor impedance matches discharge impedance, so that peak efficiency may be
obtained and reflected power minimized.
3. Materials and Methods
3.1. Substrate
Sugar maple (Acer saccharum) boards were selected. For each parameter series
tested, all the specimens came from the same board. Wood being an organic material
sensitive to its growth environment, its properties tend to vary slightly from one
board to another. Each series consisted of eight specimens, each 4 mm × 85 mm ×
100 mm in thickness, length and width respectively.
Table 1.
Plasma types and parameters studied
Plasma gas∗ Type of Pressure Number of

discharge (Pa) treatments∗∗
H2 O Glow 13.3 3
Ar RF inductive 13.3, 133 and 665 1/1/2
Ar RF capacitive 133 2
N2 RF inductive 13.3, 133 and 665 2/1/2
N2 RF capacitive 133 3
Ar/N2 RF inductive 13.3, 133 and 665 1/1/3
N2 /O2 RF inductive 13.3, 133 and 665 1/1/1
N2 /H2 RF inductive 13.3, 133 and 665 1/1/1
N2 /H2 RF capacitive 133 1
* Mixes were 1:1.
** For plasma treatments studied under different pressures (x, y and z), the number of treatments
per pressure is indicated as a/b/c. For each treatment two replicates were used.
3.2. Coating
A waterborne UV curable polyurethane/polyacrylate resin was employed. The com-
ponents and their formulations were supplied by our industrial partner (Société
Laurentides Inc., Canada). The coating was applied with a square applicator to a
thickness of about 60 µm, and then dried in two steps. It was first dried at 60◦ C for
10 minutes for water evaporation and polyacrylate initial cross-linking. And then, it
was exposed to UV radiation to cross-link the polyurethane (Mercury lamp/UVA =
5300 J/m2 ) and complete the curing process.
3.3. Plasma Sources
Three types of vacuum plasma sources were employed in this study: DC glow
plasma, RF inductive plasma, and RF capacitive plasma, as described above. All
plasma treatments were performed at 150 W with three different exposure times:
0.5, 10 and 20 min. Plasmas were generated from different gases and mixes as
summarized in Table 1. These mixes were used at 13.3, 133 and 665 Pa pressures.
3.4. Contact Angle
Substrate wettability towards water was quantified before and after treatments with
a First Ten Ångstroms (FTA) 200 Dynamic Contact Angle Analyzer. Contact angles
were measured immediately after droplet deposition using the sessile drop method.
Droplet volume was 70 µl.
3.5. Adhesion Characterization
Coating adhesion to wood surfaces was measured in accordance with the ASTM
D4541 test method. The technique involves gluing stainless dollies with an epoxy
resin on painted surfaces and allowing them to dry for 24 hours before pulling
them off. The force necessary to pull off the coating was measured with a Positest
AT adhesion tester (DeFelsko — USA). Three pull-off tests per specimen were
performed for each of the three plasma treatment times.
3.6. Penetration Characterization
Coating penetration in wood was measured by Scanning Electron Microscopy

(SEM) at Laval University. This technique was recently replaced by Raman con-
focal spectroscopy (not presented here).
4. Experimental Protocol
The different parameters series used for this study are summarized in Table 1. Every
experiment was performed at 150, 180 or 200 W for 0.5, 10 and 20 min under partial
vacuum. Each series consisted of 8 treatments: one control, one vacuum treatment
for 20 minutes to investigate the influence of vacuum on coating adhesion, and 2
plasma treatments for each plasma exposure time.
The different steps involved in the experimental procedure are described below:
1. Preparation and sanding of specimens (150-grit sandpaper). For all tests, wood
was sanded immediately before treatment.
2. Measurement of water contact angle before plasma treatment.
3. Measurement of coating adhesion before plasma treatment.
4. Vacuum treatment for 20 min at 13.3, 133 or 665 Pa depending on parameters
selected for the plasma treatment.
5. Plasma treatment for 0.5, 10 or 20 min at 13.3, 133 or 665 Pa depending on
parameters selected.
6. Measurement of water contact angle after plasma treatment. As the equipment
to measure contact angles was not located close to the plasma chamber, the
specimens were conditioned under partial nitrogen vacuum or partial vacuum
in an inert atmosphere to avoid contamination.
7. Coating application, drying at 60◦ C for 10 min and then exposure to UV radia-
tion.
8. Measurement of coating/wood adhesion after plasma treatment.
9. Measurement of coating penetration after plasma treatment (only for glow dis-
charges).

5.1. Coating Adhesion to Wood Before and After Treatment
The efficiency of RF inductive plasma treatment was strongly related to vacuum
level and gas flow in the closed vessel. As, in this particular case, the plasma was
not directly in contact with the substrate, two parameters might have contributed to
collisions between excited particles and the substrate: gas flow and particle mean
free path. To achieve the desired effect, particles must have collided with the sub-
strate before recombination of excited species.
At pressures of 13.3 Pa and 133 Pa (Figs 3–5), the gas flow rate was relatively
low (1.67 × 105 m3 /s). Consequently, collisions were mostly driven by the particle
mean free path, which increases with the vacuum level (the lower the pressure, the
greater the mean free path). This likely explains why treatments at 13.3 Pa were
more effective.
At a pressure of 13.3 Pa, adhesion seemed to increase with exposure time (Figs
3 and 4). The time necessary to perform these experiments was very long, so they
were abandoned in favor of RF inductive and capacitive discharges, which proved
faster in most of experiments. The best treatments were found to be those using Ar
and N2 /H2 .
At a pressure of 133 Pa, the resulting adhesion values could be grouped accord-
ing to two different time ranges: 0.5–10 min and 10–20 min (Fig. 5). In the first
region, adhesion decreased, whereas it rose in the latter case. At this pressure, the
plasma treatment was less effective due to the smaller particle mean free path, so
that fewer active molecules collided with the wood surface. In addition, wood ex-
Figure 3. Adhesion improvement as a function of glow plasma exposure time at 13.3 Pa. Values were
calculated by reference to an untreated control specimen (Pref ). Pi represents the force needed to pull
off the coating after “i” minutes of plasma exposure.
Figure 4. Adhesion improvement as a function of RF inductive plasma exposure time at 13.3 Pa.
Values were calculated by reference to an untreated control specimen (Pref ). Pi represents the force
needed to pull off the coating after “i” minutes of plasma exposure. Two experiments were conducted
with N2 — they are identified as N2 and N2 bis in the legend.
Figure 5. Adhesion improvement as a function of RF inductive plasma exposure time at 133 Pa.
needed to pull off the coating after “i” minutes of plasma exposure.
Figure 6. Adhesion improvement as a function of RF inductive plasma exposure time at 665 Pa.
needed to pull off the coating after “i” minutes of plasma exposure. Two experiments were conducted
with N2 and Ar — they are identified as N2 /N2 bis and Ar/Ar bis in the legend. Three tests were
conducted with Ar/N2 mixes — they are identified as Ar/N2 , Ar/N2 bis and Ar/N2 ter in the legend.
tractives are likely to have migrated to the wood surface during the vacuum process,
which would have had a negative impact on adhesion. After 10 minutes, adhesion
of the coating increased. One possible explanation for this would be that the amount
of active molecules colliding with the wood surface was insufficient to compensate
for the previous effect.
At a pressure of 665 Pa (Fig. 6), the gas flow was increased (to about 8.33 ×
106 m3 /s) and mostly drove collisions. Under these conditions, the inductive treat-
ments were more successful than at low pressure. The plasma is likely to have
degraded the extractives present on the wood surface, creating active sites (radicals,
ions, metastables) and new functional groups, thus enhancing coating adhesion.
Only in the RF capacitive treatments was the vacuum level important, as the
substrate was positioned inside the plasma. The density of both ions and electrons
varies with the vacuum level. For this series of experiments, the results showed
significant improvements in adhesion (up to 27% for Ar plasmas). The results are
presented in Fig. 7.
For plasma generated by a glow discharge, adhesion increased greatly, up to
50% for a 20-minute treatment. The substrate was also positioned in the plasma.
This method provided the best adhesion results (see Fig. 7).
5.2. Measurement of Contact Angles Before and After Plasma Treatments
Measurements were only performed for RF capacitive and inductive plasma treat-
ments of 20-minute duration. The results are summed up in Table 2.
Figure 7. Adhesion improvement as a function of RF capacitive plasma exposure time at 133 Pa.
needed to pull off the coating after “i” minutes of plasma exposure.
Table 2.
Contact angle of distilled water and coating adhesion results for wood before and after plasma treat-
ments for 20 minutes
Contact angle Adhesion improvement

(◦ ) (%)
Wood before treatment 81 0

N2 — 133 Pa — capacitive 31 27
Ar — 133 Pa — capacitive 67 25
Ar/N2 — 133 Pa — capacitive 51 –
Ar — 665 Pa — inductive 76 0
N2 — 665 Pa — inductive 51 21
N2 /H2 — 665 Pa — inductive∗ 38 18
Ar/N2 — 665 Pa — inductive∗ 75 8
* Mixed in a 1:1 ratio.
Water wettability improved significantly following N2 RF capacitive and N2 /H2

RF inductive plasma treatments. Contact angles generally tended to decrease after
plasma treatments.
For inductive discharges, the best results were obtained in the presence of N2
(except with Ar/N2 mixtures). This may be due to N atoms being grafted onto the
wood surface and forming new chemical functions, which might explain the wa-
ter wettability increase. The same mechanism is not very likely to have occurred
with Ar atoms because of their very stable electronic structure (rare gas). In this
Figure 8. SEM picture (30 kV — X1000) — coating/untreated wood surface.
case, we assume that Ar atoms created ions and radicals on the wood surface while
modifying its structure. Further work is needed to validate these hypotheses.
For capacitive discharges, as the wood specimens were positioned inside the
plasma clouds, the atom grafting is unlikely to have occurred except to a small
extent, as ion and/or electron bombardments would prevent or limit grafting.
Finally, these results show a correlation between wettability and adhesion im-
provement. The best adhesion results were obtained with surfaces exhibiting the
lowest contact angles (highest water wettability). Given that the sphere of activity
of attraction forces is similar to molecular distances, intimate contact between coat-
ing and substrate is a necessary condition for good adhesion [2, 15–17]. Indeed, the
bond strength between coating and wood is maximized when the best wettability
has been achieved.
5.3. Wood Surface Characterization by SEM
The wood surfaces were characterized after glow plasma treatments. Figures 8
and 9, respectively, represent pictures of the coating/wood interface before and after
glow-discharge plasma treatment.
These pictures show that plasma treatments tended to smooth out surfaces and
facilitate coating penetration. This indicates that adhesion improvements observed
after plasma treatment also had a mechanical origin.
6. Conclusion
This exploratory study demonstrated that plasma technology could be used to im-
prove the adhesion of coatings to wood surfaces. For glow discharge plasmas,
Figure 9. SEM picture (30 kV — X1000) after 20 minutes H2 O glow plasma treatment at 13.3 Pa.
adhesion was increased by up to 50%, whereas RF capacitive plasmas showed that

improvements of 30% were feasible.
Adhesion improvement was directly correlated with increased water wettability,
and adhesion improved as contact angle decreased. In addition, SEM observations
showed that such improvements could also be related to surface topography, i.e.
surface smoothing and penetration depth.
Surface analyses should be carried out to provide a better understanding of coat-
ing adhesion mechanisms after plasma treatments. The chemical structure of the
wood surface should be analyzed by X-ray photoelectron spectroscopy, and rough-
ness should be studied by AFM and profilometry.
Acknowledgements
We acknowledge the support of Natural Resources Canada through its Value to
Wood and Transformative Technologies Programs, FPInnovations — Forintek Di-
vision and Canada Economic Development through its regional program in Quebec.
References
1. R. Morent, N. De Geyter, J. Verschuren, K. De Clerk, P. Kiekens and C. Leys, Surface Coatings
Technol. 202, 3427–3449 (2008).
2. F. S. Denez, L. E. Cruz-Barba and S. Manolache, in: Handbook of Wood Chemistry and Wood
Composites, R. M. Rowell (Ed.), pp. 447–474. Taylor and Francis, Boca Raton, FL (2005).
3. L. Podgorski, C. Boustas, F. Schambourg, J. Maguin and B. Chevet, Pigment Resin Technol. 31(1),
33–40 (2002).
4. L. Podgorski, B. Chevet, L. Onic and A. Merlin, Int. J. Adhesion Adhesives 20, 102 (2000).
5. P. Rehn, A. Wolkenhauer, M. Bente, S. Foorster and W. Violet, Surface Coatings Technol. 174–
175, 515 (2003).
6. G. C. Q. Blantocas, P. E. R. Mateum, R. W. M. Orille, R. J. U. Ramos, J. L. C. Monasterial, H. J.
Ramos and L. M. T. Bo-Ot, Nucl. Instrum. Meth. Phys. Res. B 259, 875 (2009).
7. P. D. Evans, M. Ramos and T. Senden, in: Proc. European Conference on Wood Modification,
Cardiff, UK, pp. 123–132 (2007).
8. A. Wolkenhauer and G. Avramidis, Int. J. Adhesion Adhesives 29, 18–22 (2009).
9. A. Wolkenhauer, G. Avramidis, H. Militz and W. Viöl, Holzforschung 62, 472–474 (2008).
10. A. Wolkenhauer, G. Avramidis and W. Viöl, Holz als Roh- und Werkstoff 66, 143–145 (2008).
11. A. Wolkenhauer, G. Avramidis, Y. Cai, H. Militz and W. Viöl, Plasma Processes Polymers 4,
470–474 (2007).
12. V. Blanchard, P. Blanchet and B. Riedl, Wood Fiber Sci., Submitted (2009).
13. C. Tendero, C. Tixier, P. Tristian, J. Desmaison and P. Leprince, Spectrochim. Acta Part B 61, 2
(2006).
14. H. Conrads and M. Schmidt, Plasma Sources Sci. Technol. 9, 441 (2000).
15. G. Elbez and D. Bentz, in: Le collage du bois (Gluing Wood), CTBA (Ed.). Paris, France (1991).
16. B. M. Collette, Wood Sci. Technol. 6, 1–42 (1972).
17. K. L. Mittal, Polym. Eng. Sci. 17, 467–473 (1977).
The Effect of a Plasma Pre-treatment on the Quality of Flock
Coatings on Polymer Substrates
Thomas Bahners a,∗ , Gerald Hoffmann b , Jürgen Nagel c , Eckhard Schollmeyer a

and Arne Voigt b
a
Deutsches Textilforschungszentrum Nord-West e.V., Adlerstr. 1, 47798 Krefeld, Germany
b
Technische Universität Dresden, Institut für Textil- und Bekleidungstechnik, Mommsenstraße 13,
01062 Dresden, Germany
c
Leibniz-Institut für Polymerforschung e.V., Hohe Straße 6, 01069 Dresden, Germany
Abstract
Flock coating is a widely used process to create a textile-like texture on substrates of various shapes and
materials. In the process, flock fibers — short fibers typically 1–3 mm long — are oriented and accelerated
towards the substrate by means of an electric field. Impacting fibers are stuck to the substrate surface by
an appropriate adhesive. Primary quality criteria are adhesion of the flock fibers to the adhesive, and also
the so-called flock density, i.e. number of fibers per unit area, and evenness. The influential physical and
chemical factors refer to interfacial adhesion, but also charging effects by the impacting fibers. The system
presently under investigation is based on aliphatic polyamides as material for a molded car component,
hot-melt adhesive, and flock fibers. Experiments reported here refer to the application of an air plasma pre-
treatment of the polyamide (PA) substrate, mainly in order to increase the adhesion of the hot-melt layer.
It was found that the plasma treatment affects the polar energy of the PA surface with a related increase
in wettability due to a reduction of C–C and C–H bonds and an increase of carboxylic groups. Surface
carbonization occurred at higher plasma doses. The effect on hot-melt adhesion was rather small, however
two types of failures were observed in these experiments, either due to insufficient adhesion of the hot-melt
or due to a break of one of the PA plates with the bond still intact. The characterization of flock coatings on
these samples showed no effect on flock fiber adhesion in pull-out as well as on abrasion resistance, but an
increased flock density was observed. This is assumed to be due to enhanced dissipation of charges by the
conductive water layer adsorbed on the substrate surface.
Keywords
Flock coating, air plasma, contact angle, hot-melt adhesive, charge dissipation
* To whom correspondence should be adressed. Tel.: +49 (0)2151 843-156; Fax: +49 (0)2151 843-143;
e-mail: bahners@dtnw.de

326 T. Bahners et al.
1. Introduction
Flock coating is a widely used process to create a textile-like texture on substrates
of various shapes and materials. In the process, flock fibers — short fibers typically
1–3 mm long — are oriented and accelerated towards the substrate by means of an
electric field. Impacting fibers are stuck to the substrate surface by an appropriate
adhesive (Fig. 1). The technique is applied to products as diverse as textiles, plastic
car interior components, floor coverings or furniture with the objective to provide
decorative, but also functional effects, e.g., to reduce friction [1–5]. Substrate ma-
terials range from polymers to metals as well as ceramics.
Primary quality criteria are adhesion of the flock fibers to the adhesive layer, but
also the so-called flock density, i.e. number of fibers per unit area, and evenness.
These criteria are influenced by physical and chemical factors, namely by interfacial
adhesion between adhesive and substrate as well as between fibers and adhesive,
and by charging effects by the impacting fibers.
Present developments, especially with regard to car parts, aim at easy-to-recycle
single-material systems [3, 4], i.e. substrate, adhesive and flock fibers based on iden-
tical polymer chemistry. The system presently under investigation is comprised of
a molded car component, hot-melt adhesive, and flock fibers all based on aliphatic
polyamides (PA) [4]. The adhesive, e.g. a co-polymer of PA6 and PA12, is often
modified with black carbon or with ions, e.g. Li+ , solvated in a poly(ethylene ox-
ide) (PEO) matrix forming an inter-penetrating network (IPN) [6–8].
One aspect in this study was the application of an air plasma pre-treatment of the
PA substrate, mainly in order to increase hot-melt adhesion and interface conductiv-
ity. The application of air plasmas as a means to increase wettability and adhesion
is well known (see e.g. [9–17]). While most papers deal with low pressure plasma
Figure 1. Schematic of the flock coating process.

Improved Flock Coating by Plasma Treatment 327
applications or variants of the corona treatment, versatile and mobile plasma units
working under ambient atmospheric pressure as, e.g., the air jet plasma described
by Lommatzsch et al. [18] are favorable for the process in question.
The scope of the experiments reported here was to study the effect of a simple
air plasma treatment using a mobile unit carrying the electrodes of the discharge
setup (‘electrode head’) on the surface properties of model components made of
polyamide (PA) 6. The plasma treatment was characterized with regard to its effect
on water contact angle and surface energy of the samples, on the adhesion of a
hot-melt adhesive and on the quality of a flock coating on these samples.
2. Experimental
2.1. Samples
For the experiments, rectangular plates were manufactured at the Leibniz-Institut
für Polymerforschung e.V. (IPF) from PA 6. Samples destined for surface analyses
were 1 cm × 5 cm (1 mm thick), samples for surface conductivity measurement
were 12 cm × 12 cm (3 mm thick), and samples for flock coating experiments were
8 cm × 8 cm (3 mm thick).
2.2. Plasma Treatment
The plasma treatment was performed in air using a system with a mobile electrode
head housing both electrodes (3D-Treater by Ahlbrandt GmbH, Lauterbach, Ger-
many). This choice was based on the possibility to move either a planar sample
along the electrode head or, alternatively, move the electrode head above compo-
nent surfaces of more complex shape.
The PA plates which served as substrate in the experiments reported here were
fitted to a stage and moved through the discharge zone. The intensity of the plasma
treatment (plasma dose) was varied by varying the speed of the moving stage (10 or
20 mm/s) and the number of passes through the plasma zone (1 to 10). The actual
exposure was estimated on the basis of the dimension of the burning plasma on the
sample surface (“foot” of the discharge), which was about 25 mm in the direction
of sample movement. The electrode head was held at a distance of 25 mm from the
sample throughout.
In all cases, the samples were wiped with ethanol before the plasma treatment in
order to remove production residuals.
2.3. Surface Characterization
The surface properties of the PA plates — as-received and following plasma treat-
ment — were characterized in terms of surface energy, chemical composition,
surface topography and surface electric properties.
Surface energies were determined by applying test inks of defined surface energy
(supplied by Ahlbrandt GmbH, Lauterbach, Germany) and by measuring the (static)
contact angles of de-ionized water and methylene iodide. The polar and dispersion
components of the surface energy were calculated from the contact angles using the
Owens and Wendt [19] equation

p p
γl (1 + cos ) = 2 · γld · γsd + γl · γs , (1)
where is the measured contact angle, γ the total and γ p and γ d the polar and
dispersion components of the surface energy. Indices l and s denote liquid and
solid, i.e. PA. The relevant values for the chosen test liquids are γ p = 52.2 mJ/m2
and γ d = 19.9 mJ/m2 for water and γ p = 2.6 mJ/m2 and γ d = 47.4 mJ/m2 for
methylene iodide [20].
Surface chemical compositions of the treated samples were determined by X-ray
photoelectron spectroscopy (XPS). The measurements were performed externally
by the Deutsches Wollforschungsinstitut e.V. (DWI) in Aachen, Germany. Surface
topography was studied by tapping mode atomic force microscopy (AFM) using a
Dimension3100/NanoScope IIIa (Veeco, St. Barbara, CA, USA).
Surface conductivity was measured according to DIN EN 1149-1 using a setup
as sketched in Fig. 2. Following application of a defined potential between disk and
ring electrodes, the resulting current is measured with an electrometer. A second
way to characterize electrostatic properties of the samples was to measure charge
dissipation after a defined surface charging by a corona discharge using a JCI 155v5
Charge Decay Test Unit in combination with a JCI Charge Measuring Sample Sup-
port (JCI John Chubb Instrumentation, Cheltenham, UK) (Fig. 3). After the artificial
charging, the electric field generated by residual surface charges is measured using
an electric field mill as a function of time. Characteristic measures of charge relax-
ation are the time elapsed until the measured field decreases to 1/e and 1/10 of its
initial value.
Figure 2. Electrode setup for the determination of electrical surface conductivity according to DIN
EN 1149-1. The conductivity is determined by applying a potential U and measuring the resulting
current I (U ).
Figure 3. Test setup for the measurement of the dissipation of artificially induced surface charges. The
sample to be investigated is charged with a corona discharge and moved into the detection field of the
electric field mill which records the electric field originating from the surface charges as a function of
time. An air curtain serves to prevent random charging.
2.4. Textile-Physical Testing

Hot-melt adhesion was characterized by bonding two overlapping PA plates and
determining the shear strength of the bond. For the bonding of plates, the PA copoly-
mer Griltex 1500A (EMS Chemie, Domat, Switzerland) was used as the adhesive.
Characteristic data of the hot-melt adhesive are melting point Tm = 80◦ C, glass
temperature Tg = 60◦ C, and viscosity at 130◦ C η130◦ C = 300 Pa s. To achieve a
good bond, the plates were pressed with 300 kPa and kept at 120◦ C for 4 min.
Microscopic and visual assessments of the surfaces where the bond failure had oc-
curred served to indicate interfacial or cohesive failure.
Flock coating was performed at the Institut für Textil- und Bekleidungstechnik,
Technische Universität Dresden (ITB) using laboratory equipment. The experi-
ments reported in this paper were conducted again using Griltex 1500A as the
adhesive which was applied with varying thickness (0.1 and 0.2 mm) and process
temperature (room temperature (RT), 40, 60 and 80◦ C). The stated process temper-
atures refer to the temperature of the adhesive, not the substrate. Flock fibers were
3 mm long PA 6.6 fibers. The quality of the flock coating was characterized by flock
fiber adhesion, abrasion resistance, and flock fiber density. It should be noted that
the measured adhesion of the flock fibers must be assumed to characterize mainly
the adhesion of the fibers to the adhesive.
The flock fiber adhesion was determined by application of an industrial test
procedure developed by Maag Flockmaschinen GmbH (Gomaringen, Germany;
www.maag-flock.com). For the test, a cylindrical mould of 10 mm diameter is
pressed on the sample and filled with adhesive (Lowmelt Q 3192 by Henkel, Hei-
delberg, Germany). Great care is taken to guarantee that the adhesive protrudes into
the fiber flock and does not contact the hot-melt layer as sketched in Fig. 4. The
(A) (B)
Figure 4. Test procedure for the characterization of the adhesion of flock fibers to the hotmelt
layer (A). The cylindrical mould is pressed on the sample and filled with adhesive (Lowmelt Q 3192 by
Henkel, Heidelberg, Germany) that just protrudes into the fiber flock. Exemplary force–displacement
curve as the ensemble of mould, test adhesive and adhering flock fibers is pulled from the sample (B).
ensemble of mould, test adhesive and adhering flock fibers is then pulled from the
sample and the force recorded.
Flock abrasion resistance was measured by pressing a steel chisel of defined
geometry on the sample and moving it in an oscillating motion (Fig. 5). The chisel
is moved in strokes of 40 mm at a rate of 60 double strokes per minute with a total
number of strokes of 400. Abrasion resistance is characterized by the weight lost
during the process.
Flock fiber density, i.e. number of flock fibers per unit area, is determined from
mass increase after flocking and calculating the individual fiber mass from the
known geometry and gravimetric density ρ (1.15 g/cm3 for PA).
3. Results
Water contact angles measured on the PA plates as-received and following the air
plasma treatment are shown in Fig. 6. In the graphs shown in Fig. 6A, the intensity
of the plasma treatment is characterized by the number of passes of the samples,
i.e. plates, through the plasma zone at two different speeds of the sample stage. In
Fig. 6B, the plasma dose is characterized by the exposure time. It has to be noted
that the exposure time was calculated on the basis of an estimated 25 mm length of
the plasma zone. As could be expected from an air plasma treatment, the contact
angle drops significantly from 62◦ to 42◦ even at a rather low plasma dose (1 pass
through the plasma zone at 10 mm/s). A saturation value of approximately 30◦ is
reached after 6 passes at 10 mm/s. If we assume the stated length of the active
plasma zone l = 25 mm and calculate the total exposure time t from t = N · l/v
for each treatment with N passes and sample speed v, the results from experiments
(A) (B)
(C)
Figure 5. Test procedure for the characterization of the abrasion resistance of the flock coating (A)
and photographs of samples with good (B) and poor (C) abrasion resistance.
run at different speeds can be scaled to a combined curve giving contact angle as a
function of exposure time as is seen in Fig. 6B.
From the contact angles measured with water and methylene iodide, surface en-
ergies were calculated using the Owens and Wendt equation and summarized in
Table 1. The resulting values of polar and dispersion components of the surface en-
p
ergy indicate that the plasma treatment only affects the polar component γs of the
p
PA surface. It is worth mentioning that the measured value of γs for the untreated
sample (9.8 mJ/m2 ) is in good agreement with values published in the literature
(10.7 mJ/m2 ), while the dispersion component deviates by more than 20% [20].
X-ray photoelectron spectroscopy (XPS) was performed to detect changes in sur-
face chemistry. As could be expected from an air plasma treatment, the overall
carbon content decreased from 78 to 73 and 62% and the overall oxygen content
increased from 11 to 15 and 24% following plasma treatments of 2.5 and 5 s dura-
tions (Table 2). The highly resolved carbon and oxygen spectra (Table 3) show that
these changes are due to a reduction of C–C and C–H bonds at the sample surface
(signal at 285 eV) and an increase of carboxylic groups characterized by the signals
at 289 eV (C–OOH bond) and at 532 eV (C–O bond). Based on the results of the
XPS analyses, the scheme for the initial reaction induced by the plasma treatment
(A)
(B)
Figure 6. Water contact angles measured on the PA plates as-received and following the air plasma
treatment. The intensity of the plasma treatment is characterized (A) by the number of passes of the
samples through the plasma zone, and (B) by the exposure time calculated on the basis of an estimated
25 mm length of the plasma zone. Measurements were made with two different speeds of the sample
stage.
can be assumed to be
Table 1.
Polar and dispersion components of the surface energies calculated from water and methylene iodide
contact angles using the Owens and Wendt equation as a function of the exposure time to the air
plasma
water (◦ ) meth. iodide (◦ )

p
t (s) γs (mJ/m2 ) γsd (mJ/m2 ) γs (mJ/m2 )
0 (untreated) 62.0 ± 0.9 29.2 ± 1.3 9.8 46.3 56.1

5 36.1 ± 2.8 33.8 ± 1.2 23.9 44.2 68.1
10 33.0 ± 1.4 31.9 ± 1.8 25.9 45.1 71.0
15 37.1 ± 3.6 32.3 ± 1.7 23.4 44.9 68.3
20 44.9 ± 2.9 30.1 ± 1.7 19.3 45.9 65.2
25 38.4 ± 3.7 32.5 ± 1.4 22.7 44.8 67.5
Note: The stated contact angles were averaged over 5 sets of experiments and may differ from the
values shown in Fig. 6.
Table 2.
Chemical surface composition (in %) of air plasma treated PA plates accord-
ing to X-ray photoelectron spectroscopy (XPS): Total element concentration
C1s O1s N1s
As-received 78 11 11
Plasma 2.5 s 73 15 12
Plasma 5 s 62 24 9
Plasma 15 s 76 15 4
Table 3.
Chemical surface composition (in %) of air plasma treated PA plates according to X-ray photoelectron
spectroscopy (XPS): Individual bond concentration
C1s O1s
Binding energy (eV) 285.0 286.5 287.8 288.9 531 532
Bond C–C, C–H C–N (C–O) C=O, C–OOH C=O C–O

HN–C=O
As-received 53 13 11 9 1
Plasma 2.5 s 47 12 13 11 4
Plasma 5 s 41 10 10 2 9 15
Plasma 15 s 65 6 4 1 14 1
As the data in Table 2 show, surface carbonization occurs with further increasing
the plasma dose. This is thought to be due to debris from etching effects at high
plasma doses.
Figure 7. AFM micrographs of an as-received PA plate and plasma treated samples. For characteri-
zation of the topography in the scanned areas, roughness values Ra (arithmetic mean), Rq (geometric
mean), and A/Ao = ratio of actual surface to projected surface are given for each micrograph.
AFM topographic measurements (Fig. 7) showed that surface roughness — char-

acterized by the roughness values Ra (arithmetic mean) and Rq (geometric mean),
and ratio of actual surface to projected surface A/Ao — decreased following mod-
erate plasma treatments, but increased again with increasing exposure to the plasma,
which would be in accordance with plasma etching.
In order to study the effect of the plasma treatment on hot-melt adhesion, treated
plates were bonded with a certain overlap using the hot-melt Griltex 1500A, and
the resulting bond characterized with regard to shear adhesion. The overlap was
necessary for experimental reasons. The measurements showed that the shear force
increased by approximately 20% following a 5 s plasma treatment but a slight de-
crease occurred within experimental error, if the plasma dose was further increased.
Two types of failures were observed in these experiments, which are exemplified
by the photographs shown in Fig. 8. The left photograph shows shearing of the
bonded plates. In this case, failure occurs through either insufficient adhesion of the
hot-melt or cohesion (supposed to be independent of the plasma treatment). The
example shown in the right photograph failed through break of one of the PA plates
while the bond was still intact. A summary of the data including a statistic of failure
types is given in Table 4. It is interesting to note that in the case of samples which
were treated at the high plasma dose, the failure of 6 out of 8 tested bonds occurred
through the break of one of the PA plates rather than through insufficient hot-melt
adhesion as is the case of untreated samples or plasma treated with a comparatively
low dose. This could be attributed to cohesive failure of the plate, but also to exper-
imental effects, e.g. simultaneous bending stress, if one considers the rather large
cross section of the plates (80 × 3 mm2 ). It is, however, known that high plasma
Figure 8. Photographs of bond partners after the shear adhesion test showing two characteristic types
of failures, insufficient adhesion or cohesion of the hot-melt (left), or break of one of the PA plates
(right).
Table 4.
Results of the shear adhesion measurements. Besides the mean maximum shear strength, the table
also shows a statistic indicating the occurrence of adhesion or cohesive failure as exemplified by the
left photograph shown in Fig. 8
t (s) No. of measurements F max (N) No. of adhesion failures
0 (untreated) 9 266 ± 24 8 (89%)

5 8 322 ± 17 4 (50%)
10 8 310 ± 19 2 (25%)
doses can lead to a degradation of the surface of a polymer substrate. Ruppert et al.
[21] described the effect for the corona treatment of poly(ethylene terephthalate)
(PET), where a highly hydrophilic, but low molecular weight surface layer with
low mechanical and chemical resistance was produced at high doses. This would
be in agreement with the AFM topographic analyses described before.
Flock coating experiments were conducted again using Griltex 1500A as the
hot-melt adhesive and 3 mm long PA 6.6 flock fibers. The plasma pre-treatment
conditions were 5 s exposure time, with varying thickness of the hot-melt adhesive
and the process temperature. The quality of the flock coating was characterized by
flock fiber adhesion, abrasion resistance, and flock fiber density.
Results of flock fiber adhesion tests are summarized in Fig. 9 for samples with
different hot-melt layer thicknesses. The data show that the plasma treatment has
no influence on this criterion. It is important to remember the flock fiber adhesion
basically does not refer to the adhesion of the hot-melt to the (plasma treated) sub-
strate but to the adhesion of the flock fibers to the hot-melt layer, into which they
are embedded. Only if the adhesion of the fibers to the hot-melt is extremely high,
a failure at the hot-melt/PA interface, i.e. a pull-off of the hot-melt, could occur.
Obviously this is not the case in the presented pull-out tests. The most influential
parameter for a high fiber adhesion appears to be the process temperature which
(A)
(B)
Figure 9. Measured flock fiber adhesion to an ‘as-received’ and a plasma treated PA plate for adhe-
sive layers applied with a thickness of (A) 0.1 mm and (B) 0.2 mm and varying temperature of the
hot-melt. The adhesive used was Griltex 1500A and the plasma treatment was performed with two
passes through the plasma zone (equivalent to 5 s exposure time).
keeps the hot-melt in a viscous state, allowing a good penetration of the fibers into
this layer.
A similar correlation is found in the measurements of the abrasion resistance of
the flock, fibers. Again, the plasma pre-treatment of the substrate is of minor effect,
an optimization being mainly achieved through an increase of process temperature
(Fig. 10).
A third criterion for the quality of the flock coating is the so-called flock fiber
‘density’ which is quantified by the number of fibers per unit area. The density is de-
termined from mass increase of the sample (PA plate) after flocking and fiber geom-
etry and density. The measurements showed that the flock fiber density on plasma
treated substrates was 80% higher than on ‘as-received’ plates (270 fibers/mm2
compared to 150 fibers/mm2 ). This could be due to enhanced dissipation of charges
following the plasma treatment.
Figure 10. Flock abrasion resistance characterized by the mass loss following the abrasion test
sketched in Fig. 5 of an ‘as-received’ and a plasma treated PA plate for adhesive layer (Griltex 1500A)
with varying temperature of the hot-melt. The plasma treatment was performed with two passes
through the plasma zone (equivalent to 5 s exposure time).
In order to understand the observed effect, it has to be considered that impacting

flock fibers carry charges which — with increasing number of deposited fibers —
reduce the effective field strength of electrostatic setup. An important criterion for
flock fibers buildup, therefore, is the fast dissipation of residual charges. With the
given conductivities of polymer substrate and hot-melt, a conductive layer of water
adsorbed from the atmosphere has great influence on overall charge dissipation. As
substrate material, PA already has the advantage of a water take-up of up to 10%.
It may be assumed on this background that the plasma treatment increases water
take-up even more and enhances charge dissemination. Similar effects following a
plasma treatment of polymer powder were reported by Sharma et al. [22].
In order to verify this assumption, measurements of (static) surface conductivity
and (dynamic) charge relaxation were conducted. In both cases, the samples were
stored for over three days before the measurement to simulate transport and han-
dling of the PA plates employed in the flock coating experiments discussed before.
Also, the cleaning procedures before and after the plasma treatment were identical.
The measurement of (static) surface conductivity followed the specifications in
DIN EN 1149-1 using a setup of concentric ring and disk electrodes as sketched
in Fig. 2. Voltage was applied between the electrodes and the electric current
measured. The calculated surface resistances of ‘as-received’ and plasma treated
samples (exposure time 5 s) are given in Table 5. Basically, a decrease in the sur-
face resistance, averaged over 5 samples, was observed, but the significance of the
effect is doubtful given the high variation in the individual values. As a second
Table 5.
Static surface resistance of untreated (‘as-received’) and air plasma treated PA plates (the measure-
ments were performed under normal laboratory conditions)
Sample mean (G) min. value min (G) max. value max (G)
Untreated 666 ± 452 116 1466

Plasma treated (5 s) 515 ± 442 155 1465
Table 6.
Measurement of dissipation (relaxation) of artificially induced surface charges on untreated (‘as-
received’) and air plasma treated PA plates (the measurements were performed under normal labo-
ratory conditions). The charges were measured as a function of time using a field mill, the relaxation
times referring to residual charges amounting to 1/e and 1/10 of the initial value
Sample Relaxation time to 1/e (s) Relaxation time to 10% (s)
Untreated 4.17 ± 0.88 15.00 ± 4.00

Plasma treated (5 s) 1.06 ± 0.06 3.44 ± 0.21
Plasma treated (15 s) 0.92 ± 0.16 2.95 ± 0.53
method, the dynamic behavior of artificially induced charges was characterized by

measuring the electric field produced by surface charges as a function of time using
a field mill. The initial charging was achieved with a corona discharge (cf. Fig. 3).
The data in Table 6 give the elapsed times for the electric field produced by the
surface charges to drop to 1/e (36.8%) or to 1/10 (10%) of the initial value. It is
seen that these ‘relaxation’ times are 4–5 times shorter on plasma treated samples
(exposure times 5 and 10 s) than on the untreated ‘as-received’ material.
4. Summary and Conclusion

The scope of this work was to study the effect of a simple air plasma treatment using
a mobile electrode unit (electrode head) on the surface properties of components
made of polyamide (PA) 6 and on the quality of flock coating on these samples.
The wettability of the PA increased already after short plasma treatments with
exposure times of 5–10 s. Measurements of water and methylene iodide contact
angles and the calculation of surface energies in dependence on the parameters
of the plasma treatment indicate that the plasma treatment only affects the polar
p
component γs of the PA surface energy. XPS analysis shows that these changes are
due to a reduction of C–C and C–H bonds at the sample surface and an increase
of carboxylic groups characterized by the signals at 289 eV (C–OOH bond) and
at 532 eV (C–O bond). Surface carbonization occurred with further increasing the
plasma dose by increasing exposure time to 15 s and longer.
Hot-melt adhesion was characterized by bonding PA plates and measuring the

shear force to bond failure. The measurements showed that the shear force increased
by approximately 20% following a 5 s plasma treatment but decreased slightly, if
the plasma dose was further increased. In the case of samples which were treated
at the high plasma dose, the failure of 6 out of 8 tested bonds occurred through a
break of one of the PA plates rather than through insufficient hot-melt adhesion.
Experimental evidence for a cohesive failure of the PA plates is weak, however.
Lessons learned from flock coating experiments can be summarized as follows.
The plasma treatment has no effect on flock fiber adhesion to the hot-melt layer,
as shown by pull-out tests, as well as on abrasion resistance. These properties are
determined by the choice and application parameters of the hot-melt such as, e.g.,
layer thickness and processing temperature. The plasma treatment has a significant
effect on flock density, however. This is assumed to be due to enhanced dissipation
of charges deposited by impacting flock fibers, which otherwise would reduce the
effective field strength, following the plasma treatment. The ruling factor is the
take-up of water from the atmosphere which is enhanced by the plasma treatment
and produces a conductive water layer on the substrate surface.
Acknowledgements
The authors wish to acknowledge financial support by the DECHEMA Gesellschaft
für Chemische Technik und Biotechnologie e.V. in the framework of project AiF-Nr.
14578 BG (This support granted within the program Industrielle Gemeinschafts-
forschung (IGF) from resources of the Bundesministerium für Wirtschaft und Tech-
nologie (BMWi) via a supplementary contribution by the Arbeitsgemeinschaft In-
dustrieller Forschungsvereinigungen e.V. (AiF)). We would also like to thank Dr.
Robert Kaufmann of the Deutsches Wollforschungsinstitut e.V. (DWI) in Aachen,
Germany, for conducting XPS analyses and Andreas Janke Leibniz-Institut für
Polymerforschung e.V., Dresden, Germany, for conducting AFM analyses.
References
1. R. Lohauß, H. Kramer and E. Völkl, Melliand-Textilberichte 72, 32 (1991).
2. T. J. Witham, Industrial Textiles 5, 1 (1992).
3. J. Müller, Int. Textile Bulletin 41, 26 (1995).
4. J. Müller, Flock 28, 16 (2002).
5. B. K. Murugesh, Textile Asia 38, 37 (2007).
6. W. T. Whang, L. H. Yang and Y. W. Fan, J. Appl. Polym. Sci. 54, 923 (1994).
7. W. T. Whang and C. L. Lu, J. Appl. Polym. Sci. 56, 1635 (1995).
8. I. Delgado, J. Castillo, M. Xhacon and R. A. Vargas, Phys. Status Solidi (b) 220, 625 (2000).
9. B. Z. Jang, Composites Sci. Technol. 44, 333 (1992).
10. Q. Wang, A. Ait-Kadi and S. Kaliaguine, J. Appl. Polym. Sci. 45, 1023 (1992).
11. D. A. Biro, G. Pleizier and Y. Deslandes, J. Appl. Polym. Sci. 47, 883 (1993).
12. T. Bahners and E. Schollmeyer, Textil Praxis Int. 49, 422 (1994).
13. G. S. Sheu and S. S. Shyu, J. Adhesion Sci. Technol. 8, 531 (1994).

14. G. S. Sheu and S. S. Shyu, J. Adhesion Sci. Technol. 8, 1027 (1994).
15. S. R. Wu, G. S. Sheu and S. S. Shyu, J. Appl. Polym. Sci. 62, 1347 (1996).
16. T. Bahners, W. Best, J. Erdmann, Y. Kiray, A. Lunk, T. Stegmaier and N. Weber, Technische
Textilien 44, 147 (2001).
17. Y. Qiu, X. Shao, C. Jensen, Y. J. Hwang, C. Zhang and M. G. McCord, in: Polymer Surface
Modification: Relevance to Adhesion, Vol. 3, K. L. Mittal (Ed.), pp. 3–24. VSP, Utrecht (2004).
18. U. Lommatzsch, M. Noeske, J. Degenhardt, T. Wübben, S. Strudthoff, G. Ellinghorst and O. D.
Hennemann, in: Polymer Surface Modification: Relevance to Adhesion, Vol. 4, K. L. Mittal (Ed.),
pp. 25–32. VSP/Brill, Leiden (2007).
19. D. K. Owens and R. C. Wendt, J. Appl. Polym. Sci. 13, 1741 (1969).
20. Source: Lehrstuhl für Kunststofftechnik (LKT), University of Erlangen, Germany. Available at
www.lkt.uni-erlangen.de.
21. S. Ruppert, B. Müller, T. Bahners and E. Schollmeyer, Textil Praxis Int. 49, 614 (1994).
22. R. Sharma, S. Trigwell, M. K. Mazumder and R. A. Sims, in: Polymer Surface Modification:
Relevance to Adhesion, Vol. 3, K. L. Mittal (Ed.), pp. 25–37. VSP, Utrecht Boston (2004).
Adhesion Properties of Wood–Plastic Composite Surfaces:
Atomic Force Microscopy as a Complimentary Analysis Tool
Gloria S. Oporto a,∗ , Douglas J. Gardner a and David J. Neivandt b

a
Advanced Engineered Wood Composites (AEWC) Center, University of Maine, Orono,
ME 04469, USA
b
Department of Chemical and Biological Engineering, University of Maine, Orono, ME 04469, USA
Abstract
The adhesion properties of the individual components of wood–plastic composites (WPCs) are highly
relevant in determining their compatibility during processing, and in the case of WPC boards, adhesion
properties can help to determine their potential application as structural components. Preliminary results in-
dicate that WPCs have a low surface energy between 20 and 25 mJ/m2 . The application of forced air plasma
treatment (FAPT) appears to be an effective method for increasing WPC surface energy (over 40 mJ/m2 ),
and is also a relatively safe surface treatment process. To characterize WPC surfaces before and after FAPT
treatment, and to evaluate bonding effectiveness, chemical (XPS), mechanical (shear strength) and ther-
modynamic (surface energy) techniques have been used. In this paper, microscopic analysis using Atomic
Force Microscopy (AFM) has been proposed as a complimentary tool to evaluate the adhesion properties of
individual components comprising WPCs as well as of WPCs before and after FAPT. For the AFM analysis,
silicon tips were used to determine pull-off forces (adhesion forces) in contact mode. AFM measurements
were performed in air and water (HPLC grade) and the histograms of the adhesion forces were obtained
from a series of more than two hundred force curve measurements. Adhesion force in air resulted in signif-
icantly higher values than those obtained in HPLC grade water; electrostatic interactions and viscoelastic
properties of the surfaces appear to have a great contribution to this phenomenon, especially for the lubricant
component of WPCs.
Keywords
Wood–plastic composites, polypropylene, atomic force microscopy, adhesion forces, forced atmospheric
(air) plasma treatment
1. Introduction
Wood–plastic composites continue to show exponential growth in the marketplace.
Particularly in North America, these materials have grown from less than 1% in the
mid-90’s to over 10% in 2004 with growth projected by several studies to reach
* To whom correspondence should be addressed. Tel.: (1-207) 581-2117; Fax: (1-207) 581-2074; e-mail:
gloria.oporto@umit.maine.edu

342 G. S. Oporto et al.
more than 20% before the end of 2010 [1]. The next generation of these materi-
als has been forecast to be used for structural applications and, therefore, a more
thorough understanding of the interaction between the components comprising the
WPCs is highly relevant. In this paper, attention is focused on the surface chem-
istry of the WPC components and composites on the nanoscale using atomic force
microscopy (AFM).
AFM has been recognized as an effective and powerful tool to determine mor-
phological properties of surfaces and also in determining interfacial interaction
strength or adhesion forces between substrates and non- or functionalized tips [2–
10]. Very detailed information regarding adhesion force measurements has been
published [2, 3].
In AFM analysis, a fine pyramidal tip at the end of a reflective cantilever is
scanned back and forth over a substrate in a raster pattern. As the tip is deflected
by the sample, the cantilever also deflects, and the magnitude of the deflection is
registered by the change in direction of a laser beam that is reflected off the end
of the cantilever and detected by a photomultiplier array. Thus, a topographic im-
age of the surface is constructed. From the forces that the cantilever can experience
while it is approaching the surface, it is possible to measure both long-range and
short-range forces. Long-range forces (electrostatic) appear when the cantilever tip
is located several micrometers above the surface; conversely, shorter-range forces
(such as van der Waals or capillary forces) appear when the cantilever comes close
to the surface (nanometers or atomic distances). Once the cantilever has made con-
tact with the surface, it can be pushed into the surface with some force (in this
case the viscoelastic properties can be measured) and then, when the cantilever is
pulled away from the surface, adhesion forces can be measured. Figure 1 shows
Figure 1. Deflection–displacement curve during tip–surface interaction [5]. In zone A, the cantilever
is far from the surface and there is no interaction with the surface. During the approach toward (or
withdrawal from) the surface, the tip interacts with the sample and a jump-in (or jump-off) contact
occurs (zone B (for loading) and zone F (for unloading)).
Adhesion Properties of Wood–Plastic Composite Surfaces 343
Figure 2. Deflection–displacement curve including approaching and retracting parts. Three types of
hystereses can occur: In the zero force line (A), in the contact part (B) and during adhesion (C) [2].
a scheme of the interaction of an AFM tip and the surface (deflection–displacement

curve) [5]. Zone A of Fig. 1 represents the tip approaching the surface (non-contact
area), then the tip touches the surface (zone B) and stays in contact with the sur-
face (zones C, D and E). While the tip is in contact with the surface it will either
experience attractive or repulsive forces according to the properties of the surface
being analyzed (adhesion zone). The adhesion force is proportional, through the
spring constant of the cantilever (N/m), to the difference between the jump-off and
jump-in deflections observed in Fig. 1. Butt et al. [2] have proposed that during
a deflection–displacement measurement three kinds of hystereses can be present
(Fig. 2); these are: in the zero force line (zone A), in the contact part (zone B) and
in the adhesion zone (zone C). Hysteresis in the zero force line (A) is common when
the material measured is a liquid due to the hydrodynamic drag on the cantilever.
In the contact regime (B) the hysteresis is usually caused by plastic or viscoelastic
deformation of the sample; an elastic deformation should not lead to a hysteresis,
however, some authors have indicated that friction can also lead to hysteresis in
this zone. Finally, for zone C, several attempts have been made to determine the
different mechanisms for adhesion hysteresis. Some researchers have determined
that hydrogen bonds, cross-linking, and physical chain entanglements dominate the
adhesion hysteresis, while others concluded that viscoelasticity was the dominant
factor. According to Butt et al. [2], in the most general case, the adhesion force Fad
is a combination of the electrostatic force Fel , the van der Waals force FvdW , the
meniscus or capillary force, and forces due to chemical bonds or acid–base interac-
tions FChem . Thus Fad will be equal to Fel + FvdW + Fcap + FChem . In many of the
AFM studies on adhesion forces, conditions have been chosen such that the van der
Waals forces were expected to dominate.
In previous work it was demonstrated that a forced atmospheric plasma treatment
(FAPT) was effective for improving the wettablity of WPC boards [11]. Based on
the earlier findings, the overall goal of the present work was to determine qual-
itatively and quantitatively the adhesion force interactions (in air and in water)
between a typical silicon AFM tip and a WPC and its components, prior and post
FAPT.
2. Experimental
2.1. Treated Surfaces
WPC boards were prepared using 50.4 wt% pine wood flour (American Wood
Fibers, Schofield, WI, USA), 39.5 wt% polypropylene (BP Amoco, Houston, TX,
USA), 2.3 wt% Polybond 3200 coupling agent (Chemtura, Middlebury, CT, USA),
4.4 wt% TPW 113 lubricant (Struktol, Stow, OH, USA) and 3.4 wt% gray colorant
(Clariant, Lewiston, ME, USA). The raw materials were extruded using a Davis-
Standard Woodtruder™ WT94 twin-screw extruder to produce WPC boards ap-
proximately 2.5 m to 3 m (long), 14 cm (wide) and 4 cm (thick).
2.2. Forced Atmospheric (Air) Plasma Treatment (FAPT)
WPC boards and WPC components (polypropylene, coupling agent and lubricant)
were surface treated using a Lectro Treat III forced air plasma surface treater at
15 kV (LTIII, Lectro Engineering, St. Louis, MO, USA). Treatments were applied
to WPC boards and WPC components (all in solid form) using the highest length of
discharge projected from the gun head (6.25 cm). The treatments were performed
by passing the electrode over the surfaces for 10 s, 30 s, 60 s and 90 s.
2.3. Atomic Force Microscopy Measurements
An MFP-3D Atomic Force Microscope (Asylum, Santa Barbara, CA, USA) was
used to determine adhesion forces between the silicon AFM tips and WPC surfaces
prior and post treatment, and also between the AFM silicon tips and the main com-
ponents of the WPCs (polypropylene, lubricant and coupling agent). Wood was not
considered in this analysis since an appropriately smooth wood surface could not be
obtained. For all samples, topographic images were obtained first using AFM in tap-
ping mode with a cantilever spring constant of 42 N/m and a resonance frequency of
300 kHz. To determine adhesion forces between the silicon AFM tip and WPC sur-
faces, a calibration of the cantilever spring constant was performed. This calibration
was made on a clean and hard surface (mica) and consisted of a 2-step procedure:
(1) determine the slope of the contact region of a force curve to determine the sen-
sitivity of the cantilever (in nm/V); (2) perform a thermal tuning to determine the
resonance frequency of the cantilever. An algorithm was subsequently used to com-
pute the spring constant using the equi-partition theorem. Adhesion forces were
determined in the contact mode of the AFM with a set-point voltage of 0.4 V and
the correspondent curves (similar to that shown in Fig. 1) were registered. A his-
togram was constructed based on more than two hundred adhesion force curves and
the corresponding mean adhesion force and its standard deviation were determined.
To evaluate the effect of the environment, the experiments were conducted in air
and in water (HPLC grade). For experiments in water the same cantilever was used
(spring constant 42 N/m).
To perform adhesion force determination on individual components of WPCs
(polypropylene, lubricant and coupling agent), 0.2 g of each component was melted
separately (on a glass slide) in an oven to produce a relatively flat surface.
Figure 3 shows representative deflection–displacement curves for the lubricant

(measured in air) before and after FAPT. After the treatment, an increased hys-
teresis can be seen especially in the contact and adhesion zones. This hysteresis was
more pronounced when the surfaces were exposed to FAPT for a longer time period
(90 s). This hysteresis caused, as well, a considerable increase in the adhesion force
values as shown in the corresponding histograms and summary of adhesion forces
(Figs 5 and 11). The hysteresis appears to be related to the lubricant viscoelastic
properties and the main possible cause of this phenomenon is the temperature rise
on the surface during FAPT. It is known that during a FAPT treatment little heat is
added while treating the surfaces; however, especially for the lubricant WPC com-
ponent this little heat is enough for softening and producing viscoelastic changes
on its surface. For polypropylene and coupling agent, the hysteresis in the contact
and adhesion zones is less apparent (compared with lubricant) after FAPT (Fig. 4),
but a minor increase in the hysteresis is also observed for longer time periods of
FAPT. The melting temperatures for polypropylene and coupling agent used in this
research are approximately 170◦ C and 190◦ C respectively. However, the lubricant
possesses a lower melting temperature of about 90◦ C which makes this component
more susceptible to the effect of heat during FAPT.
Mean adhesion force values, and the corresponding standard deviations for all
WPC components (analyzed in air and in water) prior and post FAPT treatment
are presented in Figs 5–12. For the AFM analysis performed in air, high adhesion
force values were found and a high level of variability in the results was obtained as
well. The high variability can be partly attributed to the effect of long-range forces
(electrostatic interactions) and also, as mentioned before, because of viscoelastic
changes produced during FAPT. Variability in adhesion force values was reduced
when the adhesion force measurements were performed in water since no electro-
static effects are present in the water and fast reduction of the temperature on the
surface occurs, thus avoiding any possible softening. For the polypropylene compo-
nent, an increase in the adhesion forces after FAPT was obtained for measurements
both in air and in water. Similarly, the coupling agent and lubricant in air exhib-
ited an increase in the adhesion force; however, in water no effect from the FAPT
treatment was established, and even a reduction in the level of adhesion force was
found (Fig. 12). Based on these results and if we consider the analysis in water to be
(a)
(b)
(c)
Figure 3. Representative deflection–displacement curves for lubricant component in air. (a) Untreated;
(b) FAPT treated 10 s; (c) FAPT treated 90 s.
(a)
(b)
(c)
Figure 4. Representative deflection–displacement curves for polypropylene component in air. (a) Un-
treated; (b) FAPT treated 10 s; (c) FAPT treated 90 s.
(a)
(b)
(c)
Figure 5. Adhesion force histograms for lubricant component in air, before and after FAPT treatment.
(a) Untreated; (b) FAPT treated 10 s; (c) FAPT treated 90 s.
more sensitive to interactions between the tip and the surface, then we can conclude
that the FAPT treatment primarily affects the properties of the polypropylene resin
rather than of the lubricant and coupling agent.
(a)
(b)
(c)
Figure 6. Adhesion force histograms for lubricant in HPLC grade water before and after FAPT treat-
ment. (a) Untreated; (b) FAPT treated 10 s; (c) FAPT treated 90 s.
For WPC boards, mean adhesion force values, and the corresponding standard
deviations (analyzed in air and in water) prior and post FAPT treatments are pre-
sented in Figs 13–16. For WPC boards, the same general behavior as exhibited by
the specific components can be observed, i.e. the adhesion forces increase after the
FAPT treatment, and the adhesion forces are considerably smaller when the analysis
is performed in water. An important consideration that must be taken into account
(a)
(b)
(c)
Figure 7. Adhesion force histograms for polypropylene component in air, before and after FAPT
treatment. (a) Untreated; (b) FAPT treated 10 s; (c) FAPT treated 90 s.
is the fact that the analysis of WPC boards in air was performed using planed and
sanded surfaces (using abrasive papers of three different grit sizes, P60, P100 and
P220), i.e., higher wood–tip interaction might be expected in the final results; on
the contrary the analysis in water was performed using unplaned surfaces, i.e., it is
expected to have more resin and lubricant covering the surface. The present analysis
will be complemented by measurements in air on unplaned surfaces, since planed
surfaces are affected by swelling phenomenon when the analysis is performed in
water.
All the adhesion force magnitudes determined in this research are in the range of
the AFM capability, but, unfortunately, the adhesion force measurements cannot be
compared with previous results, as there are no known previous studies performed
on similar systems.
(a)
(b)
(c)
Figure 8. Adhesion force histograms for polypropylene component in HPLC grade water before and
after FAPT treatment. (a) Untreated; (b) FAPT treated 10 s; (c) FAPT treated 90 s.
For complementary information contact angle (advancing and receding) mea-

surements, surface energy determination, and X-ray photoelectron spectroscopy of
WPC components prior to and after FAPT treatment for 10 s and 90 s should be
performed. Contact angle analysis will determine the contribution of the dispersion
and acid/base components of the surface energy and XPS will determine changes
in surface chemistry of the WPC components.
(a)
(b)
(c)
Figure 9. Adhesion force histograms for coupling agent component in air, before and after FAPT
treatment. (a) Untreated; (b) FAPT treated 10 s; (c) FAPT treated 90 s.
4. Conclusions
• The use of a silicon AFM tip allows distinguishing different levels of adhesion
forces between treated surfaces of some of the individual components of WPC
and WPC itself.
(a)
(b)
(c)
Figure 10. Adhesion force histograms for coupling agent component in HPLC grade water before and
after FAPT treatment. (a) Untreated; (b) FAPT treated 10 s; (c) FAPT treated 90 s.
• Adhesion force in air resulted in significantly higher values than those obtained
in water. Electrostatic interactions and viscoelastic properties of the surfaces
Figure 11. Summary of the adhesion forces in air determined by contact mode AFM for untreated and
10 s, 30 s, 60 s and 90 s FAPT treatments.
Figure 12. Summary of the adhesion forces in HPLC grade water determined by contact mode AFM
for untreated and 10 s, 30 s, 60 s and 90 s FAPT treatment.
appear to have a great contribution to this phenomenon, especially for the lu-
bricant component of WPCs.
• Adhesion forces measured in the water indicate that FAPT treatment is more
effective on the polypropylene surface than on the lubricant or coupling agent
(a)
(b)
Figure 13. Adhesion force histograms for WPC boards in air, before and after FAPT treatment. (a) Un-
treated; (b) FAPT treated 30 s.
(a)
Figure 14. Adhesion force histograms for WPC boards in HPLC grade water before and after FAPT
treatment. (a) Untreated; (b) FAPT treated 30 s.
surfaces. A considerable reduction in the adhesion force resulted for lubricant

and coupling agent when the treatment was performed for 30 s or 60 s.
(b)
Figure 15. Adhesion forces in air for WPC boards determined by contact mode AFM prior and post
(10 s and 30 s) FAPT treatments.
• AFM is a promising method to evaluate adhesion forces between functionalized

probes (tips) and WPC individual components and, therefore, to establish the
specific interactions.
Acknowledgements
Funding for this work was provided by the United States Coast Guard, Con-
tract #DTCG32-03-C-R00013 “Advanced Engineered Lumber Pier and Retain-
ing Wall for CG Shore Facilities II”; Federal Highway Administration (FHWA),
Contract #DTFH61-06-C-0064 “The Structural Use of WPCs in Transportation
Applications”, and the National Science Foundation under Grant No. EPS-05-
Figure 16. Summary of the adhesion forces in HPLC grade water for WPC boards determined by
contact mode AFM prior and post (10 s and 30 s) FAPT treatments.
54545 “Investing in Maine Research Infrastructure: Sustainable Forest Bioprod-

ucts”.
References
1. J. Winandy, N. Stark and C. M. Clemons, in: Proc. of the 5th Global Wood and Natural Fibre
Composites Symposium, pp. A6-1–A6-9, Kassel, Germany (2004).
2. H.-J. Butt, B. Cappella and M. Kappl, Surface Sci. Reports 59, 1–152 (2005).
3. S. N. Magonov, Annu. Rev. Mater. Sci. 27, 175–222 (1997).
4. R. Mahlberg, H. E.-M. Niemi, F. S. Denes and R. M. Rowell, Langmuir 15, 2985–2992 (1999).
5. O. Noel, M. Brogly, G. Castelein and J. Schultz, European Polym. J. 40, 965–974 (2004).
6. S. E. Woodcock, W. C. Johnson and Z. Chen, Polymer News 29, 176–183 (2004).
7. P. Eaton, J. R. Smith, P. Graham, J. D. Smart, T. G. Nevell and J. Tsibouklis, Langmuir 18, 3387–
3389 (2002).
8. H. Dvir, J. Jopp and M. Gottlieb, J. Colloid Interface Sci. 304, 58–66 (2006).
9. H. Awada, G. Castelein and M. Brogly, Surface Interface Anal. 37, 755–764 (2005).
10. B. J. R. Thio and J. C. Meredith. J. Colloid Interface Sci. 314, 52–62 (2007).
11. G. S. Oporto, D. J. Gardner, D. J. Neivandt and G. Bernhardt, Composites Interfaces (2009). To
appear.
Wettability Behavior and Adhesion Properties of a
Nano-epoxy Matrix with Organic Fibers
W. H. Zhong a,∗ , Y. Fu a , S. Jana b,∗∗ , A. Salehi-Khojin b,∗∗ , A. Zhamu b,∗∗ and

M. T. Wingert b,∗∗
a
School of Mechanical and Materials Engineering, Washington State University, Pullman,
WA 99164, USA
b
Department of Mechanical Engineering and Applied Mechanics, North Dakota State University,
Fargo, ND 58108, USA
Abstract
Matrix materials used in advanced composites reinforced by organic fibers such as ultrahigh molecular
weight polyethylene (UHMWPE) should not only possess high performance, but also need to have good
wettability and adhesion properties with the fibers. We previously developed a nano-epoxy matrix with
0.3 wt% reactive graphitic nanofibers (r-GNFs) that showed significant enhancements in mechanical, ther-
mal, hygrothermal properties, and UV aging resistance. In this paper, the wetting behavior and the adhesion
properties of the nano-epoxy matrix with UHMWPE fibers are presented. The results indicate that the nano-
epoxy matrix is also the most effective in improving wettability and adhesion properties with the UHMWPE
fiber. The corresponding mechanisms of the interfacial adhesion performance improvement of the nano-
epoxy matrix with the UHMWPE fiber are elucidated. This work demonstrates that the nano-epoxy matrix
is a unified polymeric matrix with extraordinary wettability and adhesion properties for organic fiber re-
inforcements. It validates that the novel nano-epoxy matrices effectively improve the interfacial adhesion
between the matrix and the reinforcement and provide important candidates for high-performance fiber–
polymer composites in aerospace and other industry applications.
Keywords
Adhesion, wettability, epoxy, and ultra-high molecular weight polyethylene (UHMWPE)
1. Introduction
High performance fiber-reinforced polymer composite materials have been used
extensively in aerospace and wind energy structures in recent years. They consist
of high performance fibers and polymeric matrices, which are used as the principal
* To whom correspondence should be addressed. Tel.: (509) 335-7658; Fax: (509) 335-4662; e-mail:
Katie_Zhong@wsu.edu
** They were previously graduate students or post-doctorate research associates conducting the related
experiments in Dr. Zhong’s group.

360 W. H. Zhong et al.
load-carrying elements and stress-transferring medium, respectively [1, 2]. Epoxy

resins are widely accepted as matrices because of the excellent balance between
processability and ultimate properties as well as good adhesion to a large variety of
fiber reinforcements such as glass fibers, carbon/graphite fibers, and other inorganic
fibers [3]. Carbon fibers possess the highest specific strength (strength/density) and
specific modulus (modulus/density) among the various inorganic fibers, and thus
are increasingly in demand as fiber reinforcements in a widening set of applications
[4]. Due to this growing demand for carbon fibers there is the concurrent possibility
of resultant shortages of carbon fiber composite materials. Other industries such as
wind energy and automotive are just now switching from glass fiber composites to
carbon fiber composites for reduced weight and improved performance. Therefore,
lightweight, higher performance fiber alternatives for carbon fibers are increasingly
sought as solutions to the growing demands and shrinking supplies of carbon fibers.
Organic fibers have attracted extensive attention due to their excellent mechani-
cal and physical properties and can be used as reinforcements in ballistic and high
impact composite applications. In addition to the attractive mechanical properties
of organic fibers, albeit limited in their current forms, it should be noted that they
have a distinct advantage in key industries because they do not suffer from galvanic
corrosion when in contact with certain metals as carbon does. With organic fiber
composites, the corrosion threat is avoided, and the cost and weight benefits would
be enormous to the aerospace and other industries. For example, Kevlar® (DuPont,
Richmond, VA, USA) is an organic aramid fiber with a specific tensile strength five
times greater than conventional steel. Zylon® (Toyobo Co., Osaka, Japan) consists
of a rigid chain of molecules of poly(p-phenylene-2,6-benzobisoxazole) (PBO).
Spectra® (Honeywell, Morristown, NJ, USA) is one type of UHMWPE fibers,
an ultra-lightweight, high-strength polyethylene fiber. Spectra® and Zylon® pos-
sess excellent mechanical properties, high damage tolerance, non-conductivity and
flexibility, very high specific strength, modulus and energy-to-break, low moisture
sensitivity, and good UV resistance [4]. All these organic fibers have great poten-
tial as excellent alternatives for inorganic fibers used as reinforcements in advanced
composites if their interface strength with the structural matrices can be improved.
Typical molecular characteristics of these organic fibers include non-polar chem-
ical structure and crystalline domains. It is these structural characteristics that im-
part advanced mechanical properties to these fibers. Among these organic fibers,
UHMWPE fiber possesses a higher strength to weight ratio vis-à-vis other fiber
reinforcements. At the same time the chemical inertness of the UHMWPE fiber sur-
face may lead to a high degree of corrosion resistance; there are no sites to allow for
a concentrated attack on the surface. One limitation of UHMWPE fibers is the poor
fiber/matrix adhesion which leads to poor load transfer from matrix to fiber, disrupt-
ing reinforcement. The poor interfacial bonding results from the chemical inertness
of the reinforcement with no reactive surface sites available for interfacial bonding
and the low-energy fiber surfaces. The non-polar nature of the organic fibers leads
to insufficient interactions with most polymeric matrices to form a strong interfacial
Wettability Behavior and Adhesion Properties 361
bond. Therefore, it is necessary to study and develop a new technology to overcome

the critical interfacial weakness.
2. Theoretical Aspects of Wetting and State-of-the-Art Techniques for

Improving Interfacial Quality Between Organic Fibers and Resin Matrices
Good wetting is a prerequisite to achieve good adhesion between a fiber reinforce-
ment and a matrix. It is well known that the properties of an interface are governed
largely by the surface characteristics of the two components, viz. fiber reinforce-
ment and matrix. The adhesion properties between a liquid and a solid can be
evaluated by the following thermodynamic relationships, when the liquid–solid in-
teraction equilibrium approaches.
WS = γSV − γLV − γSL = γLV (cos θ − 1), (1)
WW = γSV − γSL = γLV cos θ,
WA = γSV + γLV − γSL = γLV (1 + cos θ ),
where, WS , WW , WA , and θ represent the work of spreading, wetting, adhesion, and
the equilibrium contact angle, respectively, during the liquid–solid contact.
From these equations, we can conclude that the work increases with the decrease
of the equilibrium contact angle θ . When the equilibrium contact angle is zero, the
components of work reach their maximum value. Therefore, the equilibrium contact
angle can be used to characterize the wetting behavior of the liquid–solid systems.
Additionally, the equilibrium contact angle can be described by Young’s equa-
tion (2), which can be derived from the interaction of a liquid droplet on a solid
surface shown in Fig. 1(a)
γSV − γSL
cos θ = , (2)
γLV
where, γSV , γSL , and γLV are the interfacial energies of the solid–vapor, solid–
liquid, and liquid–vapor interfaces, respectively. The equilibrium contact angles of
0◦ and 180◦ represent the two extreme states of complete wetting and complete
non-wetting, respectively.
From equation (2), we can see that good wetting requires that the difference be-
tween the solid–vapor surface energy (γSV ) and the solid–liquid interfacial energy
(γSL ) approaches the value of the liquid–vapor surface energy (γLV ). Therefore,
wetting is favored when the solid surface energy is high and/or the liquid surface
energy is low [5]. These theories result in two strategies to achieve improved wetta-
bility between a solid fiber and a resin matrix: one is to enhance the surface energy
of the fiber reinforcements by fiber surface treatments; the other is to reduce the
surface energy of the polymeric matrix by modifying its characteristics (Fig. 1(b)).
To our knowledge, almost all reported work on improvement in wetting and
adhesion at the fiber/matrix interface has focused on the study of surface treat-
ments or coatings for fiber surfaces. A variety of surface coating and modification
Figure 1. (a) Wetting of solid reinforcement surface by liquid resin. (b) Sketch of methods for increas-
ing wetting and adhesion.
techniques for fibers have been developed to improve interfacial bonding between
fibers and the epoxy matrices. However, for UHMWPE fiber, silane coupling agent
treatments (effective for glass fibers) and oxidation treatments (effective for car-
bon fibers) are not effective in improving the interfacial strength [6–11]. Therefore,
other methods have been developed to treat the UHMWPE fibers including nitrogen
ion implantation [12, 13], nitrogen plasma [13], fast atom beams [14], laser abla-
tion [15], chain disentanglement [16, 17], high power ion beam treatments [18], and
cold plasma [19]. Such approaches have shown improvement in interfacial proper-
ties, but inevitably degradation of the mechanical properties of the fibers occurs.
It has been reported that many of these treatments damage the chain structure of
the UHMWPE fiber and lead to the formation of amorphous hydrogenated car-
bon. With the strength and modulus so heavily dependent on the crystallinity of the
UHMWPE fiber [14, 15], these treatments leading to reduced crystallinity would
not be desired [20]. Additionally, ensuring a high degree of wetting and good in-
terfacial adhesion for the organic fibers/polymer matrix has been problematic. The
reasons for this include: (1) fiber surface treatment techniques not only incur an
extra cost for manufacturing composite structures, but also reduce intrinsic proper-
ties, such as the strength of the organic reinforcements; (2) most of the treatment
methods are not effective for the surface of organic fibers (e.g. UHMWPE fiber) as
compared to that of the inorganic fibers (carbon and glass fibers). There appears to
be a lack of effective means to alter the fiber surface without sacrificing its desirable
bulk properties. It is concluded that novel cost-effective methods for improving the
interfacial adhesion between the organic fibers and the polymer matrices are vital
to the full realization of their potential as structural materials [21] and that great
opportunity exists in developing novel matrices to solve the aforementioned prob-
lems.
3. Approach Involving Modification of Epoxy Matrices by Nanotechnology

Graphitic nanofibers (GNFs) are an attractive additive for fabricating reinforced
polymer materials with enhanced mechanical properties due to their commer-
cial availability, unusual atomic structure, and controllable graphene orientation
[22–26]. GNFs can be produced today in high volumes at low cost using metal
catalysts and either ethylene or carbon monoxide as sources of carbon [27, 28].
Each GNF structure presents a unique, reactive surface and atomic spatial arrange-
ment to an external polymer matrix, thus permitting a wide range of options for
promoting GNF/polymer covalent bonding. With the derivatized GNFs, using 3,4 -
oxydianiline (ODA) as linker molecule, several variants of GNF-ODA, have been
synthesized in Lukehart’s group [29]. Based on these GNF-ODA nanofibers, we
have fabricated reactive GNFs (r-GNFs), and thus a reactive nano-epoxy matrix
was developed, as shown in Fig. 2.
Instead of a simple physical mixture of nanofibers and epoxy resins, as is
common in many nanocomposites, this nano-epoxy is composed of unified su-
per molecules containing covalently bonded nanofiber–epoxy, in which covalent
bonds are formed chemically between the nanofibers and epoxy resin. Through
our series of previous studies [30–39], it has been found that, unlike some of the
more prevalent fiber surface treatments, our approach not only offers substantial
processing cost savings, but also significantly enhances the overall performance
of the composites with a complete retention of the exceptional performance of the
fiber reinforcements. Our previous experiments showed that nano-epoxy matrix had
greatly enhanced thermal and mechanical properties. Three-point bending experi-
mental results for the nano-epoxy matrix with 0.3 wt% r-GNFs showed that the
flexural strength, flexural modulus and breaking strain were increased by approxi-
mately 26%, 21% and 30% respectively versus the pure epoxy. Fracture toughness
increased by 40% for nanocomposite specimens with 0.5 wt% r-GNFs. Dynamic
mechanical analysis of nano-epoxy matrix with 0.3 wt% r-GNFs also exhibited a
significant enhancement in storage modulus (by 122%). The tensile tests on fiber-
reinforced nano-epoxy composites showed that nano-epoxy matrix with 0.3 wt%
of r-GNFs had the highest initial stiffness and ultimate tensile strength. Thermal
Figure 2. Structure of GNF-ODA and formation of an r-GNF.
analysis also showed that nano-epoxy matrix with 0.3 wt% graphitic nanofibers
increased the glass transition temperature by 14◦ C and reduced the coefficient of
thermal expansion compared to pure epoxy. The studies on the effects of hygrother-
mal exposure and UV irradiation confirmed that nano-epoxy matrix with 0.3 wt%
r-GNFs had the best aging resistance performance, i.e. the maximum UV resistance
and the lowest rate of moisture absorption.
In addition, we found that the nano-epoxy with reactive functional nanofibers,
r-GNFs, and a diluent led to decreased surface energy, and thus improvements in
interfacial adhesion between the continuous fibers and the nano-matrix [40]. There-
fore, it is believed that the nano-epoxy can be an attractive candidate as a high
performance composite matrix, and we have examined its wetting and adhesion
properties with UHMWPE fiber, a ubiquitous and highly attractive organic fiber.
In this work, the nano-epoxy matrix was prepared from reactive GNFs (r-GNFs)
through the reaction of the cut GNF-ODA nanofibers with butyl glycidyl ether
(BGE), which is also the sonication agent [41]. It was found that introduction of
BGE to r-GNFs in a fixed ratio, the so-called “liquid nano-reinforcement”, could
facilitate an easy dispersion and reaction of the r-GNFs with the epoxy. Therefore,
the nano-epoxy was produced by adding the liquid nano-reinforcement with a con-
trolled ratio of r-GNFs to BGE, into the epoxy resin. The details of preparation
of reactive nano-epoxy matrix can be found in [21, 39] and [41]. We now address
the research carried out on the wetting behavior and interfacial adhesion of the
nano-epoxy matrix to the organic fibers through choosing UHMWPE fiber as the
reinforcement and as well as the underlying interaction mechanisms. Wetting be-
havior of the nano-epoxy matrix on the UHMWPE fibers was evaluated by surface
tension, viscosity, and contact angle measurements. The efficiency of improvement
in the interfacial adhesion was characterized by the microbond and bundle pullout
tests. Correspondingly, mechanisms of improvements in interfacial adhesion are
also elucidated utilizing the energy dispersive X-ray technique.
4. Wetting Behavior of the Nano-epoxy Matrix

The wetting behavior of a reinforcement fiber surface by a polymer resin matrix can
be generally described by two processes: spreading of the polymeric matrix on the
fiber surface and penetration of the polymeric matrix into the fiber surface cavities.
The spreading of the polymeric matrix on the fiber is determined by viscosity and
contact angle of the polymeric resin. The penetration activity is mainly governed
by the viscosity of the polymeric matrix.
4.1. Viscosity
Viscosity is an important parameter for characterizing the resin–fiber interactions
and it affects the wetting behavior of epoxy matrix on the fiber surface. The less vis-
cous the matrix, the faster the spreading rate of the liquid drop on solid surfaces and
the shorter the time taken to reach the liquid–solid interaction equilibrium. Viscos-
ity of various nano-epoxy samples was measured by a viscometer (LVT, Brookfield)
[40]. The experimental results are given in Fig. 3.
As shown in Fig. 3, the diluted epoxy samples had a relatively greater reduction
of viscosity than other samples. In particular, the smaller decrease of viscosity for
the GNF-filled epoxy samples without the diluents, as compared to all the other
Figure 3. Decrease (%) of viscosity versus pure epoxy (nano-epoxy: epoxy + 0.3 wt%
r-GNFs + 1.8 wt% BGE; Epoxy + diluent (1.8 wt% BGE); Epoxy + GNFs (0.3 wt%)).
samples, confirmed that the BGE diluent played a key role in controlling the vis-
cosity of the epoxy matrix. The nano-epoxy samples exhibited only a slightly lesser
decrease (3%) than the corresponding diluted pure epoxy samples, but still obtained
a reasonably lower viscosity value than that of pure epoxy.
From the results of the viscosity measurement, it was evident that the nano-epoxy
matrices with the specific amount of BGE diluent effectively decreased the viscosity
with respect to pure epoxy matrix and had a similar viscosity value as the BGE
diluted epoxy matrix.
4.2. Contact Angles
Contact angles of the nano-epoxy samples on UHMWPE sheets were measured by

an FTA 125 instrument (manufactured by First Ten Ångstroms Co., Portsmouth,
VA). From the contact angle experiments, we observed a three-stage wetting be-
havior vs. time: (1) the rapid spreading stage, (2) the transition stage, and (3) the
static/stable stage [40]. For the nano-epoxy samples, the contact angles were less
than those of the pure-epoxy for all time intervals. It was found that the nano-
epoxy uniformly coated the UHMWPE surface in the first wetting stage (Fig. 4(a):
decreases in contact angle were 12%, 62% and 69% for the nano-epoxy matrices
with 0.2, 0.3 and 0.5% r-GNFs, respectively, vs. pure epoxy), and produced lower
constant contact angles in the final stable wetting stage (Fig. 4(b): decreases were
15%, 19% and 21% for the nano-epoxy matrices with 0.2, 0.3 and 0.5% r-GNFs,
respectively, vs. pure epoxy). Together these contributed to substantially enhanced
wettability of the nano-epoxy versus the pure epoxy on the UHMWPE material.
The low viscosity of the BGE was the fundamental cause for the reduction of con-
tact angles in nano-epoxy versus pure epoxy. Since the weight ratio of the r-GNFs to
BGE was 1:6, a higher concentration of nanofibers meant a higher volume of dilu-
ent. Therefore, the nano-epoxy with 0.5 wt% r-GNFs (therefore including 3.0 wt%
BGE) showed the least viscosity and lowest contact angle.
Spreading distance and coverage area of the matrix droplets were measured and
are shown in Figs 5 and 6 [42], respectively. The spreading rates (Fig. 4) signifi-
cantly increased with r-GNFs concentration versus pure epoxy, but higher amounts
of r-GNFs did not achieve the expected increase in spreading rates. For example,
spreading rates of 0.3 wt% r-GNFs increased by 50% with respective to 0.2 wt%
r-GNFs, but lowered by 7% with respect to 0.5 wt% r-GNFs. As shown in Figs
5 and 6, the nano-epoxy matrix initially showed the same major axis length and
coverage area of the droplets as the pure epoxy, but with the lapse of time, the
nano-epoxy matrix exhibited a significant increase in both parameters, increasing
by 2.5 times and 6 times respectively as compared to the pure epoxy. This indicated
the nano-epoxy matrix exhibited excellent spreading properties on the UHMWPE
fibers.
(a)
(b)
Figure 4. Wetting of UHMWPE sheet surface by different epoxy systems: (a) Increase in spreading
rate vs. pure epoxy; (b) Decrease in contact angle vs. pure epoxy (error < 5%).
5. Adhesion Properties of the Reactive Nano-epoxy Matrix

Interfacial adhesion properties of the nano-epoxy to UHMWPE fiber were studied
using a filament fiber microbond test and fiber bundle pullout test [43–45].
5.1. Microbond Test
Specimens were prepared by molding spherical droplets of the pure epoxy and
nano-epoxy matrices with diameters of 5 mm and 7 mm on a single UHMWPE
Figure 5. Spreading distance (represented by major axis length) vs. time [42].
Figure 6. Coverage area (represented by elliptic area of the drop) vs. time [42].
fiber surface. The upper surface of the droplets was secured to the vise plates as
schematically shown in Fig. 7 [43]. The droplets formed were cured at 120◦ C for
8 h. Microbond testing was then performed on a microbond apparatus at a crosshead
displacement rate of 1 mm/min. The pullout force was recorded by the load cell as
the interfacial debonding proceeded. A typical force–displacement curve from the
tests is shown in Fig. 8.
Figure 7. Schematic of a microbond test sample.
Figure 8. A typical force–displacement curve from microbond tests.
It was observed that in the initial stage (displacement < 1 mm), the fiber–matrix
interface remained intact, and the curves were linear. As the external load increased
to the critical value for crack initiation, debonding commenced with concomitant
crack propagation. When the external load increased to the ultimate debonding
force, the fiber–matrix interface was fully debonded and the fiber was pulled out
from the surrounding matrix. The corresponding dissipated energy is represented
by the area under the force–displacement curve. Specimens from the nano-epoxy
matrix had significantly higher crack initiation force and ultimate debond force than
those from pure epoxy matrix, indicating the effectiveness of the nano-epoxy in im-
proving fiber/matrix interface bonding.
The experimental results from force–displacement curves for microbond spec-
imens of the nano-epoxy with different concentrations of r-GNFs are shown in
Fig. 9. The maximum percent increases were found for the matrix containing
0.3 wt% r-GNFs. The percent increases for crack initiation force and ultimate
debonding force were approximately 40% and 32%, respectively, with respect to
those of the pure epoxy matrix.
Figure 9. Percent increase in various forces from force–displacement curves for specimens with dif-
ferent r-GNF concentrations with 5 mm and 7 mm diameter samples (error < 5%).
Figure 10. Percent increase in energies for specimens containing different r-GNF concentrations with
7 mm diameter samples.
A similar tendency, shown in Fig. 9, for specimens with different r-GNF concen-
tration levels, was found in Fig. 10 [43]. The debonding energy and friction energy
were calculated and shown. The maximum debonding energy and maximum fric-
tion energy were also obtained in specimens with 0.3 wt% r-GNF concentration and
increased by 65% and 38%, respectively over those in the pure epoxy specimens.
The results implied that the nano-epoxy matrix could promote adhesion im-
provement between the UHMWPE fiber and the epoxy matrix. Particularly, the
nano-epoxy with 0.3 wt% r-GNFs was the most efficient among all concentration
levels for improving the adhesion property of the epoxy matrix to the UHMWPE
fiber. The nano-epoxy matrix containing a higher amount of r-GNFs did not reach
the expected improvement in the adhesion property of the UHMWPE fiber and the
epoxy matrix.
5.2. Pullout Test
The bundle pullout test is recognized as a convenient method for evaluating the
interfacial bond strength of a fiber tow rather than a single filament. The load–
displacement curves from these pullout tests could be used to evaluate the adhesion
property of the UHMWPE fiber-reinforced nano-epoxy in composite forms. In
these tests, the UHMWPE fiber bundles, rather than a single fiber used in the mi-
crobond test, were used to prepare the test specimens. The fiber bundles were coated
with different matrices solutions and finally cured at 120◦ C for 4 h. Four types of
matrices (with 0, 0.2, 0.3 and 0.5 wt% r-GNFs) with the same embedded length
were designed for evaluating the interfacial properties between the UHMWPE fiber
and the nano-epoxy matrix [44, 45]. The embedded length (H) was 17 mm as shown
in Fig. 11. A typical load–displacement curve from pullout tests for pure epoxy
matrix and nano-epoxy matrix with 0.3 wt% r-GNFs is shown in Fig. 12. It was ap-
parent that there existed two important values, the initial and maximum debonding
loads, which were used to evaluate the adhesion properties between the fiber and
matrix.
The initial and maximum debonding loads with different concentration levels
of r-GNFs can be seen in Figs 13 and 14, respectively. The error is less than 2%.
It can be seen that the nano-epoxy matrix with 0.3 wt% r-GNFs had the highest
value among the debonding loads. The experimental results are in agreement with
the microbond tests. The nano-epoxy matrix with 0.3 wt% r-GNFs was the most
effective among the concentration levels of nanofibers for improving the adhesion
properties of the UHMWPE fiber and the epoxy matrix. This further confirmed
that in our case a higher concentration of r-GNF in the matrix did not necessarily
achieve better interfacial properties.
Figure 11. Schematic of samples for pullout tests where H is embedded length.
Figure 12. A typical load–displacement curve from pullout tests.
Figure 13. Initial debonding load vs. r-GNF concentration (error < 5%).
The similar conclusions from experimental results from both the microbond tests
and pullout tests verified that the nano-epoxy matrices with r-GNFs produced sig-
nificantly improved adhesion properties with the UHMWPE fibers over the other
epoxy matrix systems.
5.3. Mechanism of the Interfacial Adhesion of Nano-epoxy Matrix
To fully understand the significant improvement in the interfacial adhesion be-
tween the UHMWPE fiber and the nano-epoxy matrix, it is necessary to further
investigate the mechanisms of the effect of the nano-epoxy matrix on the inter-
facial adhesion properties in the UHMWPE–epoxy composites. Energy dispersive
X-ray (EDX) was used to analyze the elemental distribution in the interfacial region
[46]. Specimens of pure epoxy and nano-epoxy with 0.2, 0.3, and 0.5 wt% r-GNFs
Figure 14. Maximum debonding load vs. r-GNF concentration (error < 5%).
Figure 15. Schematic of the observed points in EDX investigations for UHMWPE fiber and
nano-epoxy matrix (“+” sign represents the observed points).
loadings were prepared. The cross section of the composite samples showing the
UHMWPE–matrix interface was observed by a Philips SEM with 20 kV acceler-
ation voltage under high vacuum. The oxygen content at the points of interest on
the specimen surface was examined by the EDX analyzer equipped with SEM. The
schematic of the chosen points at the interface boundary (about 5 µm to 10 µm) in
UHMWPE fiber composites is shown in Fig. 15. The corresponding oxygen con-
tents are listed in Table 1.
Both carbon and oxygen elements at both chosen points on the pure epoxy and
nano-epoxy sides were found and the oxygen content varied from 8 to 11% by
weight. In particular, the oxygen element was observed close to the fiber boundary
Table 1.
Oxygen content (%) of UHMWPE composites with different epoxy matrices from EDX investigations
Pure epoxy Nano-epoxy Nano-epoxy Nano-epoxy

(0.2 wt% r-GNFs) (0.3 wt% r-GNFs) (0.5 wt% r-GNFs)
Epoxy side 6.62–9.83 10.25 9.04 10.61

UHMWPE side 0 4.41 5.63 2.45
on the UHMWPE side. Far from this boundary, only carbon element was found. It
is known that the UHMWPE is composed of two elements, carbon and hydrogen,
while epoxy and nano-epoxy consist of carbon, hydrogen, and nitrogen elements.
Therefore, the presence of oxygen at the UHMWPE side must have resulted from
diffusion of oxygen atoms of the nano-epoxy matrix into the fiber, assuming no
possibility of sample oxidation. At the 5 µm depth into the fiber the EDX detected
negligible sample oxidation. However, the oxygen element was not found at the
UHMWPE side in the pure epoxy matrix specimen because high viscosity of pure
epoxy prevented diffusion of the epoxy matrix, while nano-epoxy matrix with lower
viscosity was sufficient to allow diffusion of oxygen into the UHMWPE fiber. Fur-
thermore, nitrogen element was not observed on the UHMWPE side of nano-epoxy
matrix specimen. This can be explained by the relatively larger size of r-GNFs,
inhibiting nitrogen diffusion through the interface into the UHMMPE side.
As listed in Table 1, the oxygen content on the UHMWPE side varied in different
nano-epoxy systems. The maximum amount of oxygen was observed for the nano-
epoxy matrix with 0.3 wt% r-GNFs. For the nano-epoxy with 0.2 wt% r-GNFs,
its higher viscosity (lower decrease in viscosity as shown in Fig. 3) over that with
0.3 wt% r-GNFs led to less diffusion of the oxygen into the UHMWPE side, but
this was not the case for nano-epoxy with 0.5 wt% r-GNFs. The possible reason
was that the extra addition of r-GNFs resulted in aggregation of nanofibers on the
rough surface and thereby inhibited the diffusion of oxygen from the nano-epoxy
matrix. Consequently, the higher diffusion capability led to stronger interlocking
at the interface and higher energies and larger debonding forces were required to
break the interlocking during adhesion tests. This explains why the highest values
of the maximum debonding forces and the debonding energies were manifest for the
nano-epoxy matrix with 0.3 wt% r-GNFs with respect to the other epoxy systems.
6. Conclusions
The studies of wetting behavior of the reactive nano-epoxy matrix on the fiber re-
inforcements found that nano-epoxy matrix with 0.3 wt% r-GNFs possessed lower
viscosity, contact angle, and better spreading properties with respect to the pure
epoxy matrix. The adhesion tests (microbond and pullout) between the UHMWPE
fiber and the epoxy matrix system also showed that the nano-epoxy matrix with
0.3 wt% r-GNFs had more significant improvement in the interfacial adhesion prop-
erties with the UHMWPE fibers than the pure epoxy matrix. The elemental analysis
from EDX in the interfacial region of the UHMWPE fiber–epoxy matrix system
provided a strong support for better wettability and interfacial adhesion of nano-
epoxy matrix. Mechanical interlocking and diffusion could also contribute to the
interfacial adhesion improvement of the UHMWPE fiber and the epoxy matrix.
In summary, the experimental results indicated that the nano-epoxy with an op-
timal concentration level of r-GNF significantly improved the adhesion properties
of the UHMWPE fiber and the epoxy matrix. As shown in the experiments, nano-
epoxy with 0.3 wt% r-GNFs was the most effective one for enhancing the interfacial
quality between the UHMWPE fiber and the epoxy matrix. In more general terms,
this new nano-epoxy matrix lays the foundation for further improving the fiber–
matrix interactions in composites by modification of the polymeric matrix.
Acknowledgements
Dr. W. H. Zhong gratefully acknowledges the support from NASA through grant
NNM04AA62G. This work is also partially supported by NSF through NIRT grant
0506531. Dr. W. H. Zhong also gratefully acknowledges Dr. Charles M. Lukehart
and Mr. Jiang Li (Vanderbilt University) for providing the derivatized graphitic
nanofibers for this work.
References
1. K. K. Chawla, Composite Materials: Science and Engineering, pp. 80–83. Springer-Verlag, New
York (1987).
2. C. Galiotis, Composites Sci. Technol. 48, 15–28 (1993).
3. A. M. Clayton, Epoxy Resins: Chemistry and Technology, pp. 885–928. Marcel Dekker, New York
(1988).
4. B. Z. Jang, Advanced Polymer Composites. ASM International, Materials Park, OH (1996).
5. K. L. Mittal, Polym. Eng. Sci. 17, 467–473 (1977).
6. J. Delmonte, Technology of Carbon and Graphitic Composites. Van Nostrand Reinhold, New York
(1980).
7. R. V. Subramanian and J. J. Jukubowski, Polym. Eng. Sci. 18, 590–600 (1978).
8. L. J. Broutman and B. D. Agarwal, Polym. Eng. Sci. 14, 581–588 (1974).
9. J. H. William, Jr. and P. N. Kousiounelos, Fibre Sci. Technol. 11, 83–88 (1978).
10. D. G. Peiffer, J. Appl. Polym. Sci. 24, 1451–1455 (1979).
11. R. C. Arridge, Polym. Eng. Sci. 15, 757–760 (1975).
12. J. S. Chen, S. P. Lau, Z. Sun, B. K. Tay, G. Q. Yu, F. Y. Zhu, D. Z. Zhu and H. J. Xu, Surface
Coatings Technol. 138, 33–38 (2001).
13. K. G. Kostov, M. Ueda, I. H. Tan, N. F. Leite, A. F. Beleto and G. F. Gomes, Surface Coatings
Technol. 186, 287–290 (2004).
14. T. Ujvari, A. Toth, I. Bertoti, P. M. Nagy and A. Juhasz, Solid State Ionics 141–142, 225–229
(2001).
15. L. Torrisi, S. Gammino, A. M. Mezzasalma, A. M. Visco, J. Badziak, P. Parys, J. Wolowski,
E. Woryna, J. Krasa, L. Laska, M. Pfeifer, K. Rohlena and F. P. Boody, Appl. Surface Sci. 227,
164–174 (2004).
16. Y. Cohen, D. M. Rein, L. E. Vaykhansky and R. S. Porter, Composites: Part A 30, 19–25 (1999).
17. Y. Cohen, D. M. Rein and L. E. Vaykhansky, Composites Sci. Technol. 57, 1149–1154 (1997).
18. A. N. Netravali, in: Proc. 50th Intl. SAMPE Symposium, Long Beach, CA (2005).
19. H. X. Nguyen, G. Riahi, G. Wood and A. Poursartip, in: Proc. 33rd Intl. SAMPE Symp., Anaheim,
CA (1988).
20. J. Wang and K. J. Smith, Jr., Polymer 40, 7261–7274 (1999).
21. W. H. Zhong, J. Li, L. R. Xu and C. M. Lukehart, Polym. Composites 26, 128–135 (2005).
22. T. Nemes, A. Chambers and R. T. K. Baker, J. Phys. Chem. B 102, 6323–6330 (1998).
23. A. Chambers, T. Nemes, N. M. Rodriguez and T. R. K. Baker, J. Phys. Chem. B 102, 2251–2258
(1998).
24. A. Chambers, N. M. Rodriguez and T. R. K. Baker, J. Mater. Res. 11, 430–438 (1996).
25. N. M. Rodriguez, A. Chambers and R. T. K. Baker, Langmuir 11, 3862–8266 (1995).
26. N. M. Rodriguez, J. Mater. Res. 8, 3233–3250 (1993).
27. J. Sandler, T. P. Shaffer, W. Bauhofer, K. Schulte and A. H. Windle, Polymer 40, 5967–5971
(1999).
28. Z. R. Yue, W. Jiang, L. Wang, H. Toghiani, S. D. Gardner and C. U. Pittman, Carbon 37, 1607–
1618 (1999).
29. L. R. Xu, V. Bhamidipati, W. H. Zhong, J. Li and C. M. Lukehart, J. Composite Mater. 38, 1563–
1582 (2004).
30. S. Jana, W. H. Zhong, J. J. Stone and Y. X. Gan, Mater. Sci. Eng. A445–446, 106–112 (2007).
31. A. Salehi-Khojin, S. Jana and W. H. Zhong, J. Nanosci. Nanotechnol. 7, 1–9 (2007).
32. S. Jana, B. R. Hinderliter and W. H. Zhong, J. Mater. Sci. 43, 4236–4246 (2008).
33. S. Jana, A. Zhamu, W. H. Zhong, Y. X. Gan and J. J. Stone, Mater. Manuf. Processes 23, 102–110
(2008).
34. S. Jana and W. H. Zhong, J. Composite Mater. 41, 2897–2914 (2007).
35. A. Zhamu, S. Jana, A. Salehi-Khojin, E. Kolodka, Y. X. Gan and W. H. Zhong, Composite Inter-
faces 14, 177–198 (2007).
36. S. Jana and W. H. Zhong, J. Appl. Polym. Sci. 106, 3555–3563 (2007).
37. S. Jana and W. H. Zhong, J. Mater. Sci. Lett. 43, 413–416 (2008).
38. A. Salehi-Khojin, S. Jana and W. H. Zhong, J. Mater. Sci. 42, 6093–6101 (2007).
39. A. Zhamu, M. Wingert, S. Jana, W. H. Zhong and J. J. Stone, Composites Part A 38, 699–709
(2007).
40. S. Neema, A. Salehi-Khojin, A. Zhamu, W. H. Zhong, S. Jana and Y. X. Gan, J. Colloid Interface
Sci. 229, 332 (2006).
41. W. H. Zhong, J. Li, L. R. Xu, J. A. Michel, L. M. Sullivan and C. M. Lukehart, J. Nanosci. Nano-
technol. 4, 794–802 (2004).
42. M. T. Wingert, Master thesis, North Dakota State University, Fargo, ND (2004).
43. A. S. Khojin, J. J. Stone and W.-H. Zhong, J. Composite Mater. 41, 1163 (2007).
44. S. Jana, A. Zhamu, W.-H. Zhong and Y. X. Gan, J. Adhesion 82, 1157 (2006).
45. A. Zhamu, W. H. Zhong and J. J. Stone, Composites Sci. Technol. 66, 2736 (2006).
46. S. Jana, G. Sui and W. H. Zhong, J. Adhesion Sci. Technol. 23, 1281 (2009).
Enhancing the Fiber–Matrix Adhesion in Woven Jute Fabric
Reinforced Polyester Resin-Based Composites
Abdullah A. Kafi and Bronwyn L. Fox ∗

Centre for Material and Fiber Innovation, Deakin University, Geelong, VIC 3217, Australia
Abstract
This study focused on current drawbacks associated with wet lay-up processes, polyester resin and jute
fabrics for jute/polyester based composite applications. Three different processes were studied to address
manufacturing drawbacks. The composites are based on a wet lay-up and Quickstep™ (QS) process; com-
bined atmospheric helium plasma, wet lay-up and Quickstep™ (PQS) process; and finally vacuum assisted
resin infusion (VARI) process. A systematic comparison with composites made by wet lay-up was carried
out. The optimal QS process cure cycle of 30 min curing time at 95◦ C was selected based on flexural and
mode-I fracture toughness properties. It was observed that the values of both fracture toughness and flexural
strength/modulus factors increased up to a 30 min cure cycle and decreased afterwards. After optimisation,
QS was proven to have a positive influence on fracture toughness without sacrificing flexural properties,
saving curing time by 50% when compared to wet lay-up. Composites made by vacuum assisted resin infu-
sion (VARI) provided the best flexural strength and modulus values which was due to highest fiber volume
fraction and lowest flexural deflection at break. It also showed a decrease in fracture toughness proper-
ties in comparison to wet lay-up and the QS composites. All of these phenomena were due to change in
fiber–matrix adhesion after various processing conditions.
Keywords
Jute, polyester, wet lay-up, Quickstep™, infusion, plasma
1. Introduction
The rising costs of synthetic fibres and stringent environmental legislations are forc-
ing composite manufacturers to seek an alternative reinforcement fibre from nature
[1]. Amongst the various natural fibers, jute fibers have drawn attention because
of their low cost, renewability and ecofriendly nature [2, 3]. The use of jute is ap-
pealing to the automotive industry because of its low specific gravity (1–1.45), high
specific modulus (19 GPa) and low cost [3]. The inherent disadvantages of using
* To whom correspondence should be addressed. Tel.: +61 3 5227 1015; Fax: +61 3 5227 1103; e-mail:
blfox@deakin.edu.au

378 A. A. Kafi and B. L. Fox
jute, however, are poor wettability, and poor fiber–matrix adhesion with low resis-
tance to moisture.
Jute is potentially a good candidate in combination with polyester resins for com-
posite industry as both are low cost commodities. Novel polyester resin systems
reinforced with fiberglass have been widely used in the construction, transporta-
tion, and electronic industries since their development in 1941 [4]. Polyester resins
contain C=C double bonds which are capable of free radical polymerization to form
a highly crosslinked network when these are reacted with styrene monomer in pres-
ence of a catalyst [5]. Styrene performs the vital role of cross-linking the molecular
chains of the polyester and reduces the viscosity of resin. However, emission of
styrene during the curing process can lead to health and workplace safety problems
[6]. In an effort to resolve styrene related health and safety issues much research
has been conducted on developing new polyester resin systems, improving ventila-
tion or air filtering systems and minimizing styrene vapor through closed moulding
systems [7].
Natural fibre-based thermoset composites have been produced through resin
transfer moulding (RTM) and compression moulding (CM) [8]. However, wet lay-
up is still the most widely used process in industry [7]. The labor-intensive wet
lay-up process remains a problem for unsaturated polyester or vinyl ester resin users
which emit styrene vapor as previously discussed. One way to control the emission
of styrene vapor is to install extraction chambers which require high setup and regu-
lar maintenance costs. For small and medium scale operations, styrene vapor can be
reduced by using low styrene content or modified polyesters, however these mea-
sures are not sufficient for larger scale operations. A way to reduce styrene emission
is to modify or replace the wet lay-up process with a potential alternative which will
add major benefits such as reduction of curing time, lowering of tooling/labor costs
and reduced pollution risk without sacrificing the desired composite properties.
A new initiative to cure natural fiber-based composites using Quickstep™ was
pursued because of its ability to significantly reduce cycle times [9], lower the vis-
cosity of the resin [10] as well as reduce possible exothermic effects associated
with the polyester resins. The Quickstep™ process, a patented fluid filled float-
ing mould technology of Quickstep Technology Pty Ltd, Western Australia, was
developed to manufacture lightweight advanced composites [11]. The technology
operates by rapid application of heat to the uncured laminate surface achieved by
close contact of a heat transfer fluid of high heat capacity and thermal conductivity.
The heat transfer fluid has an important secondary role in removal of heat generated
by the polymeric reaction during cure progression [12]. Rapid heating properties of
heat transfer fluid maintain precise control of the resin viscosity, which promotes
uniform resin dispersion and improved fiber wetting.
A potential way to uniformly modify the surface of natural fibres is atmospheric
pressure glow discharge plasma [13]. The operation is continuous and does not re-
quire vacuum systems therefore saving time, space and energy over low pressure
plasma processes [14]. Various chemical and energetic treatments or combinations
Enhancing the Fiber–Matrix Adhesion 379
of both [15] have been studied to modify natural fiber surfaces. Several relevant
studies have focused on the modification of fiber surfaces using low pressure plasma
for high performance composites; although little has been reported on the use of
atmospheric plasma for natural fibre based composite applications. This work inves-
tigated the effect of atmospheric plasma in combination with Quickstep™ process
on the composite properties.
Vacuum assisted resin infusion (VARI) has been used for polyester resins as an
alternative to wet lay-up since 1950 [16]. There are a number of variants of this
process all involving the use of vacuum driven resin infusion into pre-forms. VARI
has several advantages over conventional resin transfer moulding (RTM) as one
of the solid tools of RTM is replaced by a flexible polymeric film using low cost
vacuum bagging [7]. Because of the high tooling cost of resin transfer moulding
and uncontrolled styrene emission in hand lay-up, VARI not only ensures better
mechanical properties but also reduces exposure to styrene and resins.
The aim of this research was to study the effect of different manufacturing
processes on the properties of jute fiber-based polyester composite production. The
work was mainly focused on selecting the optimal manufacturing process condi-
tions in terms of curing time, fiber volume fraction as well as fracture mechanical
behavior with the aim of making high strength composites from jute fibers.
2. Experimental
2.1. Materials
The polymer matrix used was a commercial unsaturated polyester resin (ESCON
62-333) procured from Fiberglass International, Australia and unbleached woven
jute fabrics were sourced from Bangladesh Jute Mills Corporation. 1.5% cobalt
napthenate solution as promoter and 36% styrene as crosslinker were premixed with
ESCON. The curing agent was methyl ethyl ketone peroxide (MEKP) and the mold
release agent was poly(vinyl alcohol) (PVA) supplied by Fiberglass International,
Australia.
2.2. Fabric and Fibre Characteristics
320 g/m2 woven jute fabrics were selected for composite reinforcement. The yarns
had an average fineness of 270.41 (±18.25) tex in the warp direction and 242.21
(±19.22) tex in the weft direction. Tex is commonly reported as units of fibre linear
density expressed as mass per unit length of fibre. The yarn twist was 191 tpm (turns
per meter) in the warp direction and 167 tpm in the weft direction. Yarn twist was
measured according to ASTM D 1422-92 and yarn linear density was tested as per
ISO 2060:1994(E) standard.
2.3. Wet Lay-up Process
Jute fabrics were conditioned for 24 hours at 20 ± 2◦ C and 65 ± 2% relative humid-
ity before starting the wet lay-up process. Fabric samples of 200 mm squares were
Figure 1. Schematic representation of vacuum bagging.
cut out and a 3:1 resin to jute ratio was used for each case. Fabrics were placed alter-
natively at a 0/90 orientation in both warp and weft directions. The mould surface
was previously coated with a layer of PVA release agent at least 20 minutes prior
to setting of release film. The overall sequence of lay-up was as follows: A layer of
release film, sandwich structure made of resin and fabric, a layer of peel ply, and
breather on the top (Fig. 1). It was finally vacuum bagged at −85 kPa as shown in
Fig. 1 and cured at room temperature. A post-cure was conducted for an hour in a
fume cupboard at ambient temperature to reduce styrene emission.
2.4. Quickstep™ (QS) Process
Composites were hand laid-up, vacuum bagged and then cured using the Quick-
step™ process on a Quickstep QS5 (Quickstep Technology Pty Ltd, Australia)
plant. Two thermocouples were inserted on each side of the laminates before vac-
uum bagging (Fig. 1) to record resin temperature and to control the Quickstep™
cure cycle. The laminates were cured in a clamshell style mould with flexible blad-
ders as shown in Fig. 2. Four different cure cycles were used at constant dwell
temperature (95◦ C) and ramp rate (3◦ C/min) where dwell time was varied (e.g.
5 min, 30 min, 60 min and 90 min).
2.5. Plasma and Quickstep™ (PQS) Process
A Sigma International APC 2000 atmospheric pressure glow discharge plasma ma-
chine (Sigma Technologies International, USA) was used (Fig. 3). The fabric was
attached to the treatment roller and rotated past the plasma source. Treatments con-
sisted of 5, 25, 35, 50 and 100 revolutions of the treatment roller. The fabric was
exposed to 0.424 second of plasma treatment during each pass. The power of the
treatment plasma was 970 W at a frequency of 90 kHz and a helium treatment gas
rate of 14000 cm3 /min.
Figure 2. Schematic representation of Quickstep™ mould [12].
Figure 3. Schematic representation of atmospheric plasma setup.
2.6. Vacuum Assisted Resin Infusion (VARI) Process
The overall arrangement of infusion process is shown in Fig. 4. First, PVA release
agent was uniformly applied to the mould surface in both directions. Jute fabric
was fixed to the mould with a pressure sensitive tape as a first step of setting fabric
layers as a pre-form. Two layers of release peel ply were placed on the top of pre-
form and edges were made shorter on both sides as well as on the top of pre-form.
The bottom side of the peel ply was taken long enough to initiate resin flow from the
inlet. Another two short layers of peel ply were inserted underneath the bottom exit
end of pre-form so that excess resin could be directed towards small vacuum outlet
(Fig. 4). Afterwards two layers of nylon flow medium were fixed on the top of peel
ply alternatively in both warp and weft directions. In this case, edges were again
made shorter on top and both sides than peel ply. This created a low permeability
dead zone to force resin to the desired exit.
Figure 4. Schematic representation of vacuum assisted resin infusion.
2.7. Fiber Volume Fraction (FVF)

Fibre volume fraction was obtained from the fiber/resin volume ratio following
equation (1) as described by Heslehusrt [17] assuming volume of voids equal to
zero. The first step was to calculate fibre/resin volume ratio from the fibre/resin
weight ratio where the densities of both fibre and the resin systems were taken from
vendor data. The method is handy as it does not require any sample destruction [18]
or imaging [19].
1
Fiber volume fraction (FVF) = 1 − , (1)
1 + (Vfiber /Vresin )
where Vfibre /Vresin = fiber/resin volume ratio.
2.8. Mode-I Double Cantilever Beam (DCB) Tests
The mode-I DCB tests were performed in accordance with the protocol of the Eu-
ropean Structural Integrity Society using a Lloyd LR30K universal testing machine
with a 1 kN load cell [20]. Each specimen was cut into an average size of 20 mm in
width, 6 mm in thickness, and 172 mm in length with a starter film length of 65 mm
to initiate the crack during testing. The cut edges of each specimen were polished
before one side of the specimen was coated with white correction fluid. Two rec-
tangular shaped aluminum tabs were attached to the end of each specimen using an
instant adhesive supplied by Loctite. The mode-I critical energy release rate GIc was
calculated by the corrected beam theory (CBT) [20]. Both GIc -initiation and GIc -
propagation values were calculated from the nonlinear point of the load/extension
curve identified as the first point of resistance curve and maximum or plateau of
resistance curve (R-curve), respectively. At least five specimens were conditioned
in the test environment at 20 ± 2◦ C and 65 ± 2% relative humidity for 24 h prior to
testing.
2.9. Flexural Tests
Flexural strength (FS) and flexural modulus (FM) were determined using the 3-
point bending method as per ASTM D 790-84a at a crosshead speed of 2 mm/min
with a Lloyd LR30K universal tensile tester. All tests were conducted at 20 ± 2◦ C
and 65 ± 2% relative humidity allowing sample conditioning for 24 h prior to the
tests. At least six specimens were tested for each type. Flexural strength and modu-
lus factor were calculated from the ratio of flexural strength/modulus and deflection
at break.
3.1. Effect of Quickstep™ (QS) process
The effect of Quickstep™ (QS) process, i.e. composites made by both wet lay-up
and Quickstep™ processes, on the fracture mechanical properties of composites
was investigated and finally compared with wet lay-up only composites. Combina-
tion of the wet lay-up and Quickstep™ processes was utilized to address the present
drawbacks of wet lay-up process such as long curing times, low fiber volume frac-
tion, and poor fiber–matrix adhesion. Firstly, optimisation of the cure cycle time
was done based on flexural deflection at break (Db ), flexural strength (FS) and flex-
ural modulus (FM) factor and finally mode-I interlaminar fracture toughness (GIc ).
Four different cure cycles with 90, 60, 30, and 5 min dwell time at a fixed dwell
temperature of 95◦ C were used. Figure 5a shows the relationship between Db and
curing time, whereas Fig. 5b and c represent flexural strength (FS) and flexural
modulus (FM) factor with respect to curing time. Figure 5d shows the change in
fracture toughness behavior for different curing times. It was observed that Db de-
creased when curing time increased from 5 to 30 min (Fig. 5a). A further increase in
curing time increased the values of Db . This result correlates with the results shown
in Fig. 5b and c where both FS and FM factors increased sharply up to 30 min cur-
ing time and afterwards both properties decreased. Similar trends were observed in
crack resistance curve (Fig. 5d) where the 30 min curing time showed higher GIc -
initiation/propagation values compared to composites made by 60 min curing time.
However, both properties were significantly reduced for composites made by 5 and
90 min curing time. This is most likely due to poor fiber–matrix adhesion which re-
sults in low observed fiber bridging in case of 5 min curing time. For 90 min curing
time, the reason could be the decomposition of resin at high temperature and long
curing time. It was noted that mode-I double cantilever beam (DCB) specimens
made by wet lay-up process showed crack growth up to 10 mm crack length after
which it failed prematurely (Fig. 5d). This is most likely due to poor fiber–matrix
adhesion at the interface. However, the crack growth for QS cured DCB specimen
was up to 35 mm crack length without any failure.
(a)
(b)
(c)
Figure 5. Effect of Quickstep™ cure cycle (represented in terms of curing time) on (a) deflection at
break, Db , (b) flexural strength factor, FS/Db , (c) flexural modulus factor, FM/Db , (d) crack resis-
tance curve (R-curve), (e) GIc -initiation/propagation values of jute-based composites.
(d)
(e)
3.2. Effect of Combined Plasma, Wet Lay-up and Quickstep™ (PQS) Process
Although Quickstep was found to compliment wet lay-up in the previous section,
there is still scope to improve properties further as jute is inherently incompatible
with hydrophobic polyester resins [21]. The effect of combined plasma, wet lay-up
and Quickstep™ (PQS) process on the fracture mechanical properties of composites
was investigated and finally compared with QS and wet lay-up composites. First,
optimization of the plasma treatment time was carried out based on flexural strength
(FS) and flexural modulus (FM) factors (Fig. 6a and b) and finally compared with
composites made by wet lay-up and QS processes as shown in Fig. 7(c) and (d). It
has been observed that flexural strength (FS) factor improved up to 25 passes and
(a)
(b)
Figure 6. Optimization of plasma treatment time (represented in terms of number of plasma treatment
passes) based on (a) flexural strength factor, FS/Db , (b) flexural modulus factor, FM/Db of jute-based
composites.
flexural modulus (FM) factor up to 35 passes of plasma treatment and both prop-
erties decreased afterwards as shown in Fig. 6(a) and (b). This could be explained
as the removal of outermost weakly attached pectin layer from outermost jute fibre
surfaces after short plasma treatment time [22]. Both FS and FM factors decreased
when the internal structure was exposed (>35 passes). Composites made by PQS
process showed 24% and 11% decrease in fiber volume fraction (Vf ) values com-
pared to QS and the wet lay-up composites. It also showed a significant decrease
in GIc -initiation/propagation compared to wet lay-up process as shown in Fig. 7(e)
and (f). Further study is currently under way to find possible explanations.
(a)
(b)
(c)
Figure 7. Effect of different manufacturing processes on (a) fiber volume fraction, Vf , (b) deflec-
tion at break, Db , (c) flexural strength factor, FS/Db , (d) flexural modulus factor, FM/Db , (e) crack
resistance curve (R-curve), (f) GIc -initiation/propagation values of jute-based composites.
(d)
(e)
(f)
3.3. Effect of Vacuum Assisted Resin Infusion (VARI) Process

Both plasma treatment and Quickstep curing have been found to complement the
wet lay-up process, however, there were still significant styrene emission problems
remaining in the workplace as the matrix was unsaturated polyester. In this section,
vacuum assisted resin infusion (VARI) process was adopted with jute fiber-based
polyester composites to ensure environmental safety and further improvement in
fiber–matrix adhesion. A maximum fiber volume fraction (Vf ) with minimum de-
flections at break (Db ) was achieved for the VARI process as shown in Fig. 7(a)
and (b). Furthermore, both flexural strength (FS) and flexural modulus (FM) fac-
tors were found to be significantly improved for VARI compared to other processes
(Fig. 7(c) and (d)). The results indicated that composites made by VARI had lower
GIc -propagation values compared to composites made by QS and PQS processes.
However, GIc -propagation values were not considered here for comparison for
composites made by wet lay-up process as the specimen failed prematurely be-
fore the propagation values were calculated. VARI also showed 22% decrease and
223% increase in GIc -initiation values as compared to QS and PQS process, respec-
tively (Fig. 7f). Further study is currently underway to understand the actual reason
behind this change.
4. Conclusion
This work has dealt with three major issues: problems related to the wet lay-up
process, surface modification of jute fabric and finally styrene emissions related
to polyester resins. Three processing methods were applied and flexural/mode-I
properties were chosen as performance indicators as they are important in design-
ing structural materials. The use of Quickstep™ curing process after wet lay-up
of fabric significantly reduced the curing time and increased the values of fracture
toughness as well as fiber volume fraction without sacrificing the flexural properties
when compared to composites produced by wet lay-up. Atmospheric plasma treat-
ment prior to wet lay-up and Quickstep™ (QS) process was found as a potential
alternative to solve most common compatibility problems between jute and poly-
ester. Moreover, the vacuum assisted resin infusion (VARI) process was identified
as a potential alternative to the wet lay-up process with respect to workplace safety,
higher fiber volume fraction and improved flexural properties.
Acknowledgements
The authors would like to acknowledge especially Mr. August Deveth for his assis-
tance during infusion work. Moreover, we are also grateful to Mr. Chris J. Hurren
and Mr. John Walter Vella for their technical support on other parts of this research.
References
1. A. N. Netravali and S. Chabba, Mater. Today 6, 22–29 (2003).
2. D. Brosius, High Performance Composites Magazine, February 10 (2006).
3. A. K. Bledzki, O. Faruk and V. E. Sperber, Macromol. Mater. Eng. 291, 449–457 (2006).
4. J. Lu and R. P. Wool, Polym. Eng. Sci. 47, 1469–1479 (2007).
5. http://www.netcomposites.com/education.asp?sequence=9 [Cited 10th September, 2008].
6. www.plasticseurope.org [Cited 9th September, 2008].
7. C. Williams, J. Summerscales and S. Grove, Composites Part A 27, 517–524 (1996).
8. P. A. Sreekumar, K. Joseph, G. Unnikrishnan and S. Thomas, Composites Sci. Technol. 67, 453–
461 (2007).
9. B. Griffiths and N. Noble, SAMPE J. 40, 41–46 (2004).
10. L. W. Davies, R. J. Day, D. Bond, A. Nesbitt, J. Ellis and E. Gardon, Composites Sci. Technol. 67,
1892–1899 (2007).
11. M. D. Silcock, C. Garschke, W. Hal and B. L. Fox, Composite Mater. 41, 965–978 (2007).
12. http://www.quickstep.com.au/how-quickstep-works (2006) [Cited 10th September, 2008].
13. C. W. Kan, K. Chan, C. W. M. Yuen and M. H. Miao, Mater. Process. Technol. 83, 180–184
(1998).
14. Y. Qiu, C. Zhang, Y. J. Hwang, B. L. Bures and M. McCord, J. Adhesion Sci. Technol. 16, 99–107
(2002).
15. G. S. Oporto, D. J. Gardner, G. Bernhardt and D. J. Neivandt, J. Adhesion Sci. Technol. 21, 1097–
1116 (2007).
16. US Patent No. 2495640 (1950).
17. R. B. Heslehurst, Paper presented at Developments in Composites: Advanced, Infrastructural,
Natural, and Nano-composites Conference, UNSW, Sydney, Australia (2006).
18. ASTM D 3171-76 (1982).
19. M. T. Cann, D. O. Adams and C. L. Schneider, Composite Mater. 42, 447–466 (2008).
20. Protocols for the Interlaminar Fracture Testing of Composites, European Structural Integrity So-
ciety (ESIS), Delft (1993).
21. A. K. Mohanty, M. A. Khan and G. Hinrichsen, Composites Part A 31, 143–150 (2000).
22. A. Baltazar-Y-Jimenez, Green Chemistry 9, 1057–1066 (2007).
Scratch and Hydrophobic Properties of Al Alloy Surface by
Combination of Transparent Inorganic and Silane-Based
Coatings
A. R. Phani, P. De Marco and S. Santucci ∗

CASTI, CNR-INFM Regional Laboratory, Department of Physics, University of L’Aquila,
Via Vetoio, Coppito, L’Aquila 67100, Italy
Abstract
Transparent inorganic–organic coatings with embedded silica nanospheres followed by dip coating from
silane based hydrophobic solution have been deposited onto Al alloy surfaces as well as on sodalime glass
surfaces. Silica nanospheres embedded in the inorganic–organic (TEOS–GPTMS) solution were synthesized
by the Stöber process. The deposited coatings were annealed at 150◦ C for 3 h in air ambient for Al alloys and
100◦ C to 300◦ C for 3 h in air ambient in case of sodalime glass. The coated substrates were tested for their
scratch and hydrophobicity properties by employing linear scratch tester and contact angle measurements,
respectively. The coatings withstood a load up to 4 N indicating good scratch resistance property and showed
a water contact angle of 120◦ indicating hydrophobic nature of the top coated surface.
Keywords
Sol–gel process, scratch resistance, hydrophobicity, silica nanospheres, Al alloy, annealing
1. Introduction
It is well known that the hydrophobicity of a surface is strongly affected by its

chemical composition and topographical appearance. Considering only the chemi-
cal factor, contact angles of only around 120◦ can be obtained for materials with
lowest surface energy (6.7 mJ/m2 for a surface with regularly aligned closely
packed hexagonal layer of –CF3 groups). For achieving higher contact angles (su-
perhydrophobicity), a certain topography is required. For the topographical modifi-
cation of low surface energy materials, in the last years two main approaches have
been investigated. The first approach considers the structuring or texturing of the
low surface energy material itself during or after depositing it on a smooth sub-
strate through, e.g. varying the deposition conditions [1], post treatment with ion or
* To whom correspondence should be addressed. Tel.: +390862433037; Fax: +390862433033; e-mail:

sandro.santucci@aquila.infn.it

392 A. R. Phani et al.
electron beam, and plasma etching [2]. In other approach, the low surface energy
materials were deposited on a substrate with a certain structure or texture, which
was obtained by methods such as embossing or laser ablation [3], photolithogra-
phy [4] or deposition of rough interlayer by thin film technology. It was shown
that by combining the chemical structure (by using sol–gel technique) with nano-
topography (by using sputtering), one could achieve superhydrophobicity [5].
Many deposition techniques, including physical vapor deposition (PVD), chem-
ical vapor deposition (CVD) or plasma polymerization, have been used to produce
hydrophobic coatings on glass, silicon and other metallic substrates. Unlike PVD,
CVD or plasma polymerization techniques, with the sol–gel technique it is possible
to modify the surface of any material (e.g. glass, metals, polymers) with organic
polymers. This process is cost effective with easy operation at low or even at am-
bient temperature. One main advantage of this process is the possibility to coat
three-dimensional substrates (or irregular geometries) with high homogeneity in
film thickness and good adhesion.
Al alloys and glass are two materials widely employed in construction and
transportation sectors. Aircraft coatings on aluminum alloy substrates provide long
lasting corrosion and abrasion protection, in addition to decorative functions. In par-
ticular, European environmental regulations have suggested to eliminate the Cr(VI)
(hexavalent chromate) coatings on Al alloys used in aircraft applications. For the
last 5 years there has been intensive research to replace these hexavalent chromium
based conversion coatings as described in various reviews [6, 7].
Organically modified silicates (Ormosils) are hybrid inorganic–organic materials
formed through the hydrolysis and condensation of organically modified silanes
based on alkoxide precursors [8]. Ormosil coatings are of interest for corrosion
resistance because they provide the mechanical and chemical characteristics by a
proper combination of inorganic and organic networks, producing coatings that are
mechanically durable, flexible, scratch resistant and dense with excellent adhesion
to the metal (Al alloy) surfaces and, in particular, are functionally compatible with
organic polymer paint systems.
On the other hand, various organic–inorganic coatings have been developed to
functionalize glass surfaces to enhance hydrophobicity [9, 10]. However, only few
investigations have been devoted to the enhancement of the mechanical and hy-
drophobic properties on both Al alloys and glass surfaces [11]. In this short commu-
nication, silica nanospheres synthesized by the Stöber process and well dispersed
in ethanol solvent and mixed with previously synthesized Ormosil solution (TEOS
and GPTMS) have been coated onto Al alloy and sodalime glass surfaces followed
by silane based hydrophobic coating applied by simple dip coating process. The
effect of such combination of layers after annealing at desired temperatures on the
scratch and hydrophobic property has been investigated.
Scratch and Hydrophobic Properties of Al Alloy Surface 393
2. Experimental Procedure
Silica nanospheres synthesized by the Stöber process were annealed at 800◦ C for
3 h in air ambient in order to improve the mechanical properties. The obtained sil-
ica nanospheres were non-agglomerated with size ranging from 300 nm to 500 nm.
Calculated quantity of annealed silica nanospheres was dispersed in isopropanol
solvent using an ultrasonic bath for 15 min. The obtained silica nanospheres dis-
persion was used as a solvent for the Ormosil solution preparation. Ormosil so-
lution was prepared by a suitable combination of tetraethylorthosilicate (TEOS)
and 3-glycidoxypropyltrimethoxysilane (GPTMS) in order to obtain clear transpar-
ent sol–gel solution. On the other hand, the top hydrophobic coating solution was
synthesized by mixing tetraethylorthosilicate (TEOS) and hexadecyltrimethoxysi-
lane (HDTMS) in 1:0.2 mole ratio in the presence of ethanol solvent. The detailed
experimental preparation procedures for the Ormosil solution have been reported
elsewhere [12]. The Ormosil coating with embedded silica nanospheres was coated
onto Al alloys as well as onto sodalime glass by using a simple dip coating process.
After 8 dippings, the substrates were thermally treated at 80◦ C for 1 h and after
complete drying the top hydrophobic coating was applied by dip coating process.
Crystallinity was investigated by X-ray diffraction (Bruker: D5000) and mor-
phology of the silica nanospheres was observed by both atomic force microscopy
(Digital 500) and scanning electron microscopy (ZEISS) techniques. Scratch re-
sistance experiments were conducted to estimate the surface resistance toward
mechanical damage. The technique involves scratching with a diamond tip on the
sample under test. The tip, either a Rockwell C diamond or a sharp metal tip, is
drawn across the coated surface under either a constant or progressive load. At a
certain critical load the coating will start to fail. The critical load data are used to
quantify the adhesion properties of film–substrate combination, progressive load
(until the coating failed) was applied starting from 1 N to 5 N. In the present in-
vestigation KRUSS DAS 100 contact angle instrument was used to measure water
contact angles on the surface.

Figure 1 shows the XRD pattern of the silica nanospheres annealed at 800◦ C for
3 h in air ambient. Figure 2 shows the AFM image of silica nanospheres embedded
in Ormosil solution deposited on Al alloy and subsequently annealed at 150◦ C for
3 h in air. The layer formation of silica nanospheres remains unchanged before and
after mixing with Ormosil and thermal treatment, indicating that removal of organic
residues (solvent) has no effect on the film morphology.
The presence of multilayers of silica nanospheres was observed which could be
attributed to surface-tension induced forces occurring during the drying and dip
coating processes. The total thickness was estimated to be 2 µm. The image re-
veals the formation of silica nanospheres in the form of layered structure with two
different sizes (500 nm and 1000 nm). Figure 3 shows the SEM image of silica
Figure 1. X-ray diffraction pattern of silica nanospheres annealed at 800◦ C for 3 h in air ambient.
Figure 2. AFM image of the silica nanospheres embedded in Ormosil coated on Al alloy surface and
subsequently annealed at 150◦ C for 3 h in air.
Figure 3. SEM image of silica nanospheres annealed at 800◦ C for 3 h in air ambient.
nanospheres annealed at 800◦ C for 3 h prior to dispersing in isopropanol solvent.

The load applied (progressive) during the scratch resistance test on the Al alloy with
bilayer coating indicated the ability to resist up to a load of 4 N. No cracks (micro)
Figure 4. Profile of the scratch test and scratch test pattern after 4 N load applied on bilayer coatings
(Ormosil with embedded silica nanospheres and hydrophobic coating) on Al alloy surface annealed at
150◦ C.
Figure 5. Graph representing contact angle of water and Rrms surface roughness versus substrate
temperature.
were observed in the length of the scratch region. Figure 4 shows the profile of the
scratch region and the inset shows the optical image of the scratched path length.
Contact angle of the films deposited at room temperature and annealed at differ-
ent temperatures as well as substrate heated at various temperatures were measured
by spheroidal segment method using a contact angle measurement system (Circular
Curve Fit Option). The contact angle, θ , was derived from advancing contact angle,
θa , and receding contact angle, θr , by the following equation: cos θ = (θa + θr )/2 or
cos θ = [(γSV − γSL )/γLV ], were γSV , γSL and γLV are free energies of solid–vapor,
solid–liquid and liquid–vapor interfaces, respectively.
Water contact angle versus annealing temperature for the deposited films on Al
alloy substrates are presented in Fig. 5, in which is evident that the films on Al alloy
as deposited show water contact angle of 122◦ and as the temperature increases
from 100◦ C to 300◦ C the contact angle decreases from 122◦ to 23◦ . This could be
due to the change in the surface structure and chemical composition making the
surface layer from hydrophobic to hydrophilic.
Figure 6 shows the water droplet image on the film deposited onto Al alloy
substrate. The roughness of the film was measured by the AFM technique. The
roughness of the film (Rrms ) deposited on Al alloy substrates changes with anneal-
ing temperature from 2 ± 1 nm for xerogel film to 12 ± 1 nm for the film annealed
at 300◦ C.
It is worth mentioning here that the Al alloy substrates coated with bilayer coat-
ings also resisted salt spray test (corrosion resistance) (5 wt% NaCl, 35◦ C — ASTM
B571) for exposure period of 300 h without a single pit. The corrosion resistance
results have been discussed elsewhere [7].
Figure 7 depicts the photographic images of the films coated onto Al alloy and
glass substrates.
Figure 6. Picture of the water droplet during the contact angle measurements.
(a) (b)
Figure 7. Water droplet pictures on Al alloy (a) and glass (b) with bilayer coatings (Ormosil with
embedded silica nanospheres and hydrophobic coating) annealed at 150◦ C.
Conclusions
In conclusion, we have reported the preparation of transparent inorganic–organic
coatings with embedded silica nanospheres onto Al alloy surfaces as well as sodal-
ime glass surfaces. The deposited coatings were annealed at 150◦ C for 3 h in air
ambient for Al alloys and 100–300◦ C for 3 h in air ambient in case of sodalime
glass. The coated substrate exhibited good scratch resistance up to a load of 4 N
and hydrophobic property with water contact angle of 120◦ .
References
1. G. Cicala, A. Milella, F. Palumbo, P. Favia and R. D’Agostino, Diamond Related Mater. 12, 2020
(2003).
2. Y. Inoue, Y. Aoshimura, Y. Ikeda and A. Kohno, Colloids Surfaces B 19, 257 (2000).
3. D. Oner and T. J. McCarthy, Langmuir 16, 777 (2000).
4. T. Uelzen and J. Muller, Thin Solid Films 434, 311 (2003).
5. Y. Gerbig, A. R. Phani and H. Haefke, Appl. Surface Sci. 242, 251 (2005).
6. A. Nylund, Aluminum Trans. 2, 121 (2001).
7. R. L. Twite and G. P. Bierwagen, Prog. Org. Coat. 33, 91 (1998).
8. J. D. Mackenzie, J. Sol–Gel Sci. Technol. 1, 81 (1994).
9. N. Carmona, M. A. Villegas and J. M. Fernandez Navarro, Thin Solid Films 458, 121 (2004).
10. G. Helsch, E. Radlien and G. H. Frischat, J. Non-Cryst. Solids 256, 193 (2000).
11. J. Malzbender and R. W. Steinbrech, Surface Coat. Technol. 176, 165 (2004).
12. A. R. Phani and S. Santucci, Prog. Org. Coat. Accepted.

Mittal, K.L. - Contact Angle, Wettability and Adhesion Volume 6-VSP (2009)

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Mittal, K.L. - Contact Angle, Wettability and Adhesion Volume 6-VSP (2009)

Încărcat de

Drepturi de autor:

Formate disponibile

The topic of wettability (measured in terms of contact

angle) is of tremendous interest from both fundamental

Contact Angle, Wettability and Adhesion

mittal2.indd 1 10-06-2009 16:07:31

No claim to original U.S. Government works

International Standard Book Number-13: 978-9-00-418102-1 (eBook - PDF)

A Novel Design of Water- and Oil-Repellent Surface Modifier Having

Wettability Behavior and Adhesion Properties of a Nano-epoxy Matrix

This volume embodies the proceedings of the Sixth International Symposium

much. Then, it is a pleasure to express my sincere thanks to my colleague and dear

* Fax: 972-4-829-3088; e-mail: marmur@technion.ac.il

Contact Angle, Wettability and Adhesion, Vol. 6

2. The Contact Angle Lexicon

3. Equilibrium Contact Angle Theory

Figure 2. The Gibbs energy for a drop on an ideal solid surface.

Another way to account for this effect is to indeed consider σl to be a function

3.2. Contact Angles on Rough or Chemically Heterogeneous Surfaces

Real surfaces are usually rough or chemically heterogeneous, or both. On rough

4. Contact Angle Measurement and Interpretation

As mentioned in the Introduction, a most important application of CA measurement

scale is of the order of magnitude of a few micrometers, a drop of a few millimeters

M. E. R. Shanahan a,∗ and K. Sefiane b

* To whom correspondence should be addressed. E-mail: m.shanahan@lmp.u-bordeaux1.fr

Contact Angle, Wettability and Adhesion, Vol. 6

2. Simplified Considerations of Stick–Slip Evaporation

Since [dG/dR]R=Ro = 0, as Ro is the equilibrium value, we may evaluate the ex-

γ sin2 θo (2 + cos θo )(δR)2

Alternatively, we may use a Taylor expansion in powers of δθ , rather than δR as in

initial volume of 10 µl were deposited on the substrate at a controlled rate using a

4. Results and Discussion

where ρ is liquid density. For axisymmetric TL motion, the corresponding inertial

5. H. Y. Erbil and M. Dogan, Langmuir 16, 9267 (2000).

Edward Bormashenko ∗ , Gene Whyman and Roman Pogreb

* To whom correspondence should be addressed. E-mail: edward@ariel.ac.il

Contact Angle, Wettability and Adhesion, Vol. 6

(A) (B) (C)

Figure 1. Wetting regimes on rough surfaces.

an electroless deposition technique. Electroless metal plating is a catalytic process

Ni2+ + 2H2 PO2 − + 2H2 O = Ni + 2H3 PO3 − + 2H+ .

Figure 3. Schematic of the vibration experiment.

process we obtained a surprisingly high contact angle of 70◦ in contrast to values

A 84 2 0.62 103 ± 2 110 98 ± 1.5 82 59 ± 2 65 74

ing to the well-known Cassie and Wenzel equations:

2.2.2. The Physical Mechanism of the Wetting Transition

Figure 6. Schematic of the droplet surrounded by a precursor film.

3.2. Resonance Wetting Transition Under Horizontal Vibrations

Appendix: Force Acting on a Unit of a Triple Line (Equation (10))

the sphere characterized by the fixed angle θ  is

Neeharika Anantharaju a , Mahesh V. Panchagnula a,∗ and Srikanth Vedantam b

Wetting of chemically heterogeneous surfaces is of importance in a wide range of

Contact Angle, Wettability and Adhesion, Vol. 6

2. Phase Field Theory

tiguous regions undergo a transformation (through boundary motion) — provided

3. Results and Discussion

Figure 2. Composite surface composed of an island of etched perfluoroalkoxy embedded in a perflu-

12. E. Fried, Continuum Mech. Thermodynam. 9, 33–60 (1997).

David B. Asay, Anna L. Barnette and Seong H. Kim ∗

* To whom correspondence should be addressed. E-mail: shkim@engr.psu.edu

Contact Angle, Wettability and Adhesion, Vol. 6

2. AFM Measurements of RH Dependence of Silicon Oxide Nano-asperity

3. Theoretical Approach to Calculate the Force Due to Meniscus Pressure of

In humid environments, water can condense to form a meniscus at or around the

FMeniscus = −PLaplace · Area = −PLaplace × πRA

FMeniscus ≈ 4πRtip γ cos θ. (4)

the sphere characterized by the fixed angle θ is

However, the assumption of Rtip RA r is not valid for nano-asperity contacts