Accepted Manuscript

Title: Nano- and Microstructured model carrier surfaces to

alter dry powder inhaler performance
Author: Niklas Renner Hartwig Steckel Nora Urbanetz
Regina Scherlie
PII:
DOI:
Reference:

S0378-5173(16)31194-2
http://dx.doi.org/doi:10.1016/j.ijpharm.2016.12.052
IJP 16323

To appear in:

International Journal of Pharmaceutics

Received date:
Revised date:
Accepted date:

23-9-2016
21-12-2016
22-12-2016

Please cite this article as: Renner, Niklas, Steckel, Hartwig, Urbanetz, Nora,
Scherlie, Regina, Nano- and Microstructured model carrier surfaces to
alter dry powder inhaler performance.International Journal of Pharmaceutics
http://dx.doi.org/10.1016/j.ijpharm.2016.12.052
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Nano- and Microstructured model carrier surfaces to alter dry powder

inhaler performance

Niklas Rennera, Hartwig Steckelb ,Nora Urbanetzc, Regina Scherliea*

a

Department of Pharmaceutics and Biopharmaceutics, Kiel University, Grasweg 9a, 24118 Kiel,
Germany
b

Deva Holding A.S., Istanbul, Turkey

Daiichi-Sankyo, Pfaffenhofen, Germany

*Corresponding author rscherliess@pharmazie.uni-kiel.de Tel.: +49 431 8801340

Graphical Abstract

Abstract
The present study investigates the effect of different carrier surface modifications on the
aerosolisation performance and on the effective carrier payload of interactive blends for inhalation.
Two different active pharmaceutical ingredients (APIs) were used: Formoterol fumarate
dihydrate (FF) and budesonide (BUD). Blends were prepared with glass beads as model carriers
which have been subjected to mechanical surface modifications in order to introduce surface
roughness via treatment with hydrofluoric acid (HF) and/or milling with tungsten carbide (TC).
As far as effective carrier payload, in this study expressed as true surface coverage (TSC), is
concerned, surface modification had varying effects on blends containing BUD or FF.
Aerodynamic characterisation in vitro showed a significant decrease in respirable fraction for glass
beads treated with HF (40.2-50.1 %), due to the presence of clefts and cavities, where drug particles
were sheltered during inhalation. In contrast, grinding with TC leads to surface roughness on a nano
scale, ultimately increasing aerodynamic performance up to 20.0-38.1 %. These findings are true for
both APIs, regardless of their chemical properties.

Keywords: Surface modification, Glass beads, Roughness, Dry powder inhaler, impaction analysis

Chemical compounds studied in this article

Budesonide (PubChem CID: 5281004); Formoterol fumarate (PubChem CID: 53477580)

1. Introduction
Dry powder inhalers (DPIs) are an important tool in the treatment of respiratory diseases like asthma
bronchiale or chronic obstructive pulmonary disease (COPD). To reach the deeper parts of the lungs,
a micronised API in the size range of 1-5 m (aerodynamic particle size) is required (Bosquillon et al.
2001, Labris et al. 2003). Since those medicines are highly potent, the delivery of only a very low
dose (BUD: 200-800 g; FF: 6-12 g) is needed. However, small particle size and low quantity of
powder are generally critical parameters leading to poor flow properties, agglomeration and
3

deteriorated metering of dose. Therefore, the API is oftentimes blended with coarse carrier particles in
order to enhance powder flowability and dosing characteristics as well as dispersibilty (Le et al.,
2012).
The effectiveness with which the drug is delivered to its target site is influenced by a variety of factors
concerning the carrier. Previous studies have investigated, inter alia, the effect of particle size (Steckel
& Mueller, 1997; Guenette et al., 2009) and shape (Kaialy et al., 2011), carrier payload (Young et al.,
2011), amorphicity (Harjunen et al., 2002) and polymorphic forms (Kaialy and Nokhodchi, 2013).
Furthermore, factors unrelated to the powder blend itself impact the lung deposition profile. Design of
the inhalation device (Coates et al., 2004), humidity (Price et al, 2002) and patients inhalation
behaviour (Chew, 2000) also have to be taken into account. Another factor found to have a crucial
effect on lung deposition is the carrier surface topography/microstructure. It influences
interparticulate interactions between carrier and API and therefore drug detachment. The main
interaction forces between the two components in interactive mixtures are van der Waals forces
(Hickey et al., 1994) and triboelectric forces (Karner et al., 2014). On the one hand, the attractive
forces have to be sufficiently high to maintain adequate blend stability and reliable blend
homogeneity. Furthermore, a high level of attractive forces generally secures major carrier payload,
which has been shown to improve the respirable fraction (Young et al., 2011). Therefore this study
also investigated the effective carrier payload, expressed as the true surface coverage (TSC), on
aerodynamic performance.
However, these forces need to be low enough for drug detachment to take place during the inhalation
maneuvre. Since low drug delivery in DPIs is often due to insufficient separation of carrier and API
resulting from excessive adhesion forces, these have to be carefully balanced. So far, contrary results
concerning the effect of carrier surface topography, more precisely surface roughness, have been
published. While Chan et al. (2003) experienced an optimal drug detachment when using carriers
holding a rough surface, Zeng et al. (2000) and Iida et al. (2004) achieved an enhanced aerodynamic
performance with blends containing carrier particles with smoothed surfaces. These contradicting
results might be partly reasoned in an unclear definition of rough and smooth surfaces. Equally
important seems the fact that relevant variables concerning formulation and inhalation device
4

potentially interact with each other. Dickhoff et. al (2005) experienced this phenomenon discovering a
stronger influence of carrier payload on drug detachment for smoother carrier surfaces compared to
carriers with a substantial carrier roughness. Furthermore, de Boer et al. (2003) described a (slight)
increase in carrier residue when using carriers with rough surfaces compared to smooth particles at a
low flow rate of 30 L/min, while they found the effect of surface roughness being practically
negligible at 60 L/min.
This study aims to provide a systematic approach to this scientific question by introducing different
types of surface roughness on the carriers. Instead of the conventionally used lactose (Kou et al.,
2012) or mannitol, which is gaining in importance lately (Kaialy and Nokhodchi, 2015), glass beads
(GBs) in the size range of 400600 m were used as model carriers. Although GBs are very different
from typical carrier materials in terms of crystallinity, density and shape, the great advantage is the
fact, that they can easily be surface modified without changing their actual size or overall shape
providing the opportunity to investigate only the effect of that corresponding modification without
any changes of other carrier surface properties possibly influencing the result. This is beneficial, since
those factors have already been proven to effect interparticle interactions, as mentioned earlier.
Previous studies investigating the effect of carrier surface properties have mainly focused on
salbutamol sulfate as API (Littringer et al., 2012; Chan et al., 2003). In the present study, Budesonide
(BUD) served as hydrophobic model drug, while Formoterol fumarate (FF) was used as its more
hydrophilic counterpart. Both APIs were spray dried to create uniform particles with comparable size
and shape. By using carrier and API particles that are both spherical, it was possible to calculate the
amount of API needed for different carrier payloads (calculated as the amount needed to cover
different percentages of carrier surface), in this study expressed as calculated surface coverages
(CSC).
As previously mentioned, drug loading can also have a substantial effect on the aerodynamic
performance. Other authors attributed this observation to the presence of so called active sites,
where drug particles preferably bind to upon blending (Hersey, 1975; Staniforth, 1995). The presence
of these highly energetic spots might not only be ascribed to chemical properties on the carrier surface
but also to morphological characteristics (El-Sabawi et al., 2006).
5

This study was designed to investigate the impact of varying carrier surface topographies on the
deposition profile and fine particle fraction of two drugs routinely used for inhalation, which differed
in their physico-chemical properties. The obtained data should ultimately facilitate the tailoring of
interparticle interactions between carrier and API to reach optimal aerodynamic performance.

2. Materials and methods

2.1. Materials

Glass beads (SiLibeads Type S, x50 = 534.4 m 13.1 m) were provided by Sigmund Lindner
GmbH (Warmensteinach, Germany). Micronized budesonide was purchased from Minakem SAS
(Dunkerque, France). Micronised formoterol fumarate dihydrate was provided by Boehringer
Ingelheim (Ingelheim, Germany). Tungsten carbide (nominal grain size: 25 m) was provided by
Wolfram Bergbau und Huetten AG (St. Martin i.S./Austria). All other chemicals were of analytical
grade and have been purchased from common suppliers.

2.2. Modification of glass beads

Prior to modification, untreated glass beads (GB_UT) were cleaned with Piranha solution (3:7 =
H2O:H2SO4) followed by a standard clean (1:1:5 = H2O2:NH4OH:H2O). Afterwards, glass beads were
incubated with hydrofluoric acid (HF) for 10 min and thoroughly rinsed with purified water for
several times (HF10min). One batch each of these glass beads was then ground with tungsten carbide
(TC) in a Retsch PM 100 ball mill (Retsch GmbH, Haan, Germany) for four and eight hours,
respectively (HF+TC4h and HF+TC8h) . Additionally, untreated glass beads were treated with TC for
four and eight hours as well (TC4h and TC8h). TC remaining on glass surfaces was removed in an
ultrasonic bath by washing with purified water.

2.3. Scanning electron microscopy (SEM)

6

Samples were prepared by applying a thin layer of probe onto a carbon sticker which was glued on
top of a metal specimen holder. The samples were sputtered with gold atoms using a BAL-TEC SCP
050 Sputter Coater (Leica Instruments, Wetzlar, Germany) and analysed with a Zeiss Ultra 55 Plus
(Carl Zeiss NTS GmbH, Oberkochen, Germany) working at a voltage of 2 kV and equipped with a
SE-2 detector.

2.4. Atomic force microscopy (AFM)

Prior to analysis, glass beads were fixed onto a glass slide using a two-component adhesive (UHU,
Buehl, Germany). Determination of surface roughness as well as imaging was performed with a JPK
NanoWizard I (JPK Instruments AG, Berlin, Germany) equipped with an OMCLAC 160 TN-W2
cantilever (Olympus, Tokyo, Japan). For every type of glass bead 3 areas of 100 m2 (10 m x 10
m) on 3 different glass beads were measured with tapping mode imaging. From the resultant data,
carrier surface roughness, expressed as the root mean square roughness (Rrms), was determined using
Gwyddion 2.42 data analysis software (Gwyddion Open Source Software, supported by Czech
Metrology Institute, http://gwyddion.net/). Rrms was calculated based on equation 1, where n is the
number of data points and xi being the vertical distance of the ith data point from the mean value of
the corresponding scan line (Klapetek et al., 2012). In order to take the effect of overall curvature out
of the equation, it was subtracted from each AFM picture before the actual calculation.

equation 1

2.5. Spray drying of APIs

To generate spherical API particles, BUD and FF were dissolved in methylene chloride and methanol,
respectively, and processed under the conditions listed in Table 1. A Buechi B290 Mini Spray Dryer
7

(Buechi Labortechnik AG, Switzerland) equipped with a high performance cyclone and a B295 Inert
Loop with nitrogen as inert gas was used for the spray drying process. The final product was stored in
a desiccator over silica gel until further use.

2.6. Determination of particle size distribution

Particle size distributions of the spray dried products were investigated by the Sympatec Helium-Neon
Laser Optical System (HELOS, Sympatec GmbH, Clausthal-Zellerfeld, Germany) equipped with a
dry dispersion module RODOS/L at a dispersion pressure of 3.0 bar. Measurements were performed
in triplicate and data was evaluated with Windox 5 software based on Fraunhofer Enhanced
Evaluation.

2.7. Preparation of interactive mixtures

The API/carrier blends were prepared for calculated surface coverages (CSC) of 100 %, 50 % and
25 % based on area. Applying theoretical considerations and equations from Zellnitz et al. (2013), 15
g of glass beads and the corresponding amount of API were weighed into a stainless steel mixing
vessel via the double sandwich method. The actual blending was performed with a Turbula Blender
T2C (Willy A. Bachofen AG Maschinenfabrik, Muttenz, Switzerland) for 30 min at 20 rpm for BUD
containing blends. Blends containing FF were blended for 35 min at 20 rpm followed by 5 min at 30
rpm. Blend homogeneity was determined by taking a total of 10 samples of 150 mg from various
locations in the mixing vessel. The samples were dissolved in 20.0 mL of the appropriate solvent
(Table 3) and investigated via HPLC (see section 2.9).

2.8. Assessment of aerodynamic performance

Aerodynamic performance was investigated using the Next Generation Pharmaceutical Impactor
(NGI) (Copley Scientific, Nottingham, United Kingdom) described as Apparatus E in the European
8

Pharmacopoeia 8.0. Flow was set to 100 L/min and the Cyclohaler was used as inhalation device for
all experiments. Prior to every run each of the stages as well as the impaction plate of the preseparator
were coated with coating media consisting of Brij35 (15 %), ethanol (51 %) and glycerol (34 %) and
allowed to dry completely. 250 mg of the respective blends were filled into hard gelatin capsules of
size 3 (Capsugel, Colmar, France) manually using a spatula. For blends with a calculated surface
coverage of 100 % and 50 % the content of three capsules was released during one run, while for
blends with 25 % coverage 5 capsules were used in order to ensure an adequate amount of drug on
every stage for HPLC quantification. The inhalation device was washed with 10.0 mL of the
appropriate solvent and the capsule shells were washed with this solution afterwards. Throat and
preseparator were rinsed with 15.0 mL and 30.0 mL, respectively. API on the NGI stages was
dissolved in 5.0 mL each. Prior to inhalation, capsules and inhalation device were deionised using a
Static Line LC discharging bar (HAUG GmbH&Co.KG Leinfelden, Germany) to eliminate the effect
of electrostatic charge in the blends under investigation. All NGI experiments were carried out in
triplicate. Fine particle fraction (FPF) was calculated based on the emitted amount of API from NGI
experiments. The resultant data was evaluated using the CITDAS software 3.1 (Copley Scientific,
Nottingham, United Kingdom)

2.9. HPLC analysis

Budesonide quantification was carried out by HPLC on a Waters 600 system (Waters Corporation,
Milford, USA) and employed a LiChrospher RP-18 column (Merck KGaA, Darmstadt, Germany).
The mobile phase consisted of 75 %(v/v) methanol and 25 %(v/v) purified water. The flow rate and
column temperature were set to 1.0 mL/min and 20 C and peaks were detected at 220 nm. For
analysing FF samples a buffering system containing 2.34 g/L sodium octansulfonate and 1.38 g/L
NaH2PO4*H2O, pH adjusted to 3.2 with ortho-phosphoric acid 85 %, methanol and acetonitrile were
used as mobile phase in the ratio of 45:40:25 (v/v). Flow rate was adjusted to 1.2 mL/min and
detection wavelength was set to 214 nm. A Lichrospher RP Select B column was installed as
stationary phase. For both studies a sample volume of 100 L was injected. Prior to sample analysis a
9

calibration curve of seven points was created confirming linearity in the range of 0.5 g/mL to 100
g/ml. Each of these solutions was injected twice.
Data was subjected to analysis of variance (ANOVA) and Students t-test (MS Excel Office 2010),
with probability values of less than 0.05 considered to be statistically significant.

2.10.

Determination of true surface coverage

The true surface coverage (TSC) was calculated from NGI experiments by first subtracting the
recovered dose, being the total mass of API, from the total amount of interactive mixture weighed in
the capsules being used for one NGI experiment. Total recovered dose included the amount of API
remaining in capsule/device. Through division of the resulting mass and the mean mass of one glass
bead the total number of glass beads was calculated (I). Similar mathematical operations can be
applied for calculating the total number of API particles by dividing the recovered dose by the mass of
one API particle (II). Comparing the ratio of (I) to (II) to the corresponding ratio of carrier to API
particles originally weighed in for the pursued CSC provides the TSC for the corresponding blend.
Detailed information about the mass of a GB and a single API particle can be obtained from
supplementary methods.

3. Results and discussion

3.1. Spray drying of APIs

Spray drying of BUD produced particles with a median diameter (x50) of 2.9 m (Span = 2.42),
while FF resulted in slightly smaller particles (x50 = 2.7 m (Span = 2.09)).
Particle morphologies of the two different spray dried APIs are shown in their SEM scanning electron
micrographs (Fig.1). These reveal that spherical particles were formed, which is true for BUD as well
as FF.

10

Furthermore, both APIs were found to be in an amorphous state as confirmed by X-ray powder
diffraction analysis (data not shown here). Since recrystallization of amorphous materials is triggered
by relative humidity (rH) and temperature (Columbano et al., 2002), DVS measurements for the APIs
were performed (data not shown here). Thus, it was decided to conduct all measurements and blend
preparation in a climate chamber at 20 C and 30 %rH (FF) or 35 %rH (BUD). This procedure also
ensured comparable environmental conditions. Blends were stored in a desiccator over silica gel until
further use.
Since the two spray dried products were found to be broadly comparable in terms of particle size,
morphology (see Fig. 1) and particle size span, judging just by the chemical properties of the two
different APIs, BUD should provide an overall increased aerodynamic performance. This would be
expected due to its hydrophobic nature. In theory, adhesion forces between BUD and hydrophilic
glass bead surface are lower compared to the less hydrophobic FF.

Therefore, the following

considerations can be made: First, the overall TSC of blends containing BUD should be lower
compared to those containing FF, as the tendency of BUD to stick to the hydrophilic carrier should be
relatively low. Second, aerodynamic performance should be increased for BUD blends due to a
facilitated drug detachment.

3.2. Characterisation of untreated and modified GBs

Modification of carrier surfaces was investigated using two different methods. Fig. 2 shows SEM
micrographs of glass beads at different magnifications of 100x, 500x and 2500x. GB_UT present a
smooth surface with only few irregularities. The effect of HF treatment is visible even at 100x
magnification showing the microscale clefts and cavities that have been introduced to the GBs. Those
are of irregular shape, with some looking crescent-like, while others appear to be almost
hemispherical. Milling with TC, being among natures most robust materials (Mohs hardness: 9.5),
caused the occurrence of numerous submicron irregularities in contrast. By comparing results for
different milling times, it can be seen, that an eight hour treatment apparently intensified that effect as
11

the density of surface irregularities increased (Fig. 2, TC4h & TC8h). A combination of both
techniques introduced both microscale, HF induced, and nanoscale, TC milling induced, roughness
(Fig. 2, HF+TC4h and HF+TC8h). However, none of the modifications significantly altered the
overall shape or particle size (data not shown).
AFM was used as a more precise technique to display and quantify on nano scale roughness by
mapping the surface in tapping mode and calculating the Rrms value. Looking at Fig. 3, AFM pictures
confirm observations made from SEM micrographs. GB_UT proved to have a smooth surface with
occasional clefts and elevations leading to a Rrms value of 21.2 7.4 nm. At the same time, this might
explain the relatively high standard deviation observed in other examinations as those clefts are
located randomly across the GB surface. A four hours grinding step significantly increased the surface
roughness for TC4h to a Rrms value of 46.0 8.3 nm.
Moreover, an eight hour treatment enhanced the Rrms value even further to 64.2 9.3 nm (TC8h). This
increase in Rrms indicates that deeper cavities have been generated compared to TC4h. The Rrms value,
however, does not provide any data about their frequency of occurrence (see equation 1). Instead,
information about this matter can be taken from SEM pictures, where the GB surface appears to be
abraded to a greater extent areawise during an eight hour milling step (Fig. 2: V. TC4h (2500x) vs.
VI. TC8h (2500x)).
GBs which have been subjected to HF treatment exhibited strongly fluctuating values for surface
roughness between 190 nm and 609 nm depending on the frequency and depth of the clefts located on
the randomly picked scanned area. Therefore, no Rrms values are shown for HF10min. However,
measurements on the surfaces of HF+TC4h and HF+TC8h were conducted, namely on areas on the
GB surface, where no larger cavities were apparent leading to Rrms values of 44.7 10.8 nm and 74.5
12.0 nm and indicating that additional milling also adds a nanoscale roughness on top of the
microscale structure introduced by HF.

3.3. Determination of TSCs

12

As already mentioned, different blending protocols were applied for BUD and FF. This is
substantiated by preliminary trials showing that optimal blending parameters are highly API
dependent due to, i.e. cohesiveness of the spray dried powder.
TSCs were determined to investigate the effect of varying surface topographies on the effective drug
loading for different amounts of API weighed in. As seen in Fig. 4 (left) and Fig. 5 (left), a TSC of
100 % could not be reached. Furthermore, TSC was always lower than the corresponding CSC. The
missing fraction of API was found to partly cover the inner walls of the mixing vessel confirming that
not the entire drug could be attached to the carrier during blending. This applies to all types of glass
beads and is true for studies with FF as well as BUD. Additionally, an almost linear relationship was
established between CSC and TSC, which is shown in Fig. 4 (right) and Fig. 5 (right). This
demonstrates a clear correlation between the amount of API originally weighed in and the effective
fraction adhered to the carrier surface.
FF served as a rather hydrophilic model drug. Fig. 4 (left) showed highest resulting TSCs of blends
with unmodified glass beads at 57.5 %, 30.5 % & 14.9 % for 100 %, 50 % & 25 % CSC, respectively.
This observation can be explained by the physico-chemical properties of carrier and API. Free silanol
groups confer hydrophilicity to the GB surface, while FF is also rather hydrophilic. Therefore,
interaction between API and carrier can be based on hydrogen bonds resulting in relatively high
adhesion forces between carrier and glass bead for the combination of GB_UT and FF. It is assumed,
that as far as amorphous APIs are concerned, as in this study, hydrogen bonding groups are
statistically distributed without a preferential exposure to certain surface areas. Nonetheless, the
authors acknowledge potential change of this homogeneous distribution of functional groups over
time due to surface relaxation and mobility in the amorphous state. This alteration might occur despite
the storage of prepared blends at low relative humidity with silica gel as drying agent. Milling with
TC led to carrier surface roughness on a nano scale as verified by AFM. It is presumed, that this
results in a reduced contact area and consequently provides fewer contact points leading to a weaker
binding of API to the carrier surface. These considerations are in accordance with the lowest TSCs for
TC4h at 43.7 %, 23.7 % & 14.4 % and for TC8h at 38.1 %, 19.0 % & 11.1 % for 100 %, 50 % and 25
% CSC, respectively.
13

BUD was used as FFs hydrophobic counterpart. According to theory, BUD should in general bind to
the hydrophilic glass bead surface to a lower extent due to a decreased potential of interaction (mainly
van-der-Waals forces) compared to FF. This is confirmed when comparing TSCs of GB_UT + FF to
GB_UT + BUD (47.3 %, 27.9 % & 13.7 %) and HF10min + FF to HF10min + BUD (43.7 %, 25.1 %
& 14.4 %).
As seen in Fig. 5 (left), an increase in TSC was found for GBs, which have been subjected to an
additional tungsten carbide treatment (GB_UT vs. TC4h/TC 8h; HF10min vs. HF+TC4h/HF+TC8h).
As already mentioned, the introduction of a grinding step potentially decreases the contact area
between glass bead and spherical API particle. BUD particles might then be more likely to bind to the
hydrophilic glass bead surface. A prolonged milling time enhances this effect, as seen from an
increase in TSC. These tendencies can be observed for all three different coverage rates but are most
explicit for a CSC of 100 %.
When comparing results obtained for the two different APIs (TSCBUD vs. TSCFF), higher overall TSCs
were found for BUD. Although at first glance this might seem illogical considering chemical
properties of carrier and APIs, observations made during optimisation of the blending parameters
could contribute to a conclusive explanation: While for interactive mixtures containing BUD a
blending time of 30 min at 20 rpm proved to be sufficient to ensure blend homogeneity as well as an
adequate amount of API attached to the carrier surface, preparation of blends containing FF required
35 min at 20 rpm followed by 5 min at 30 rpm. These observations suggest that the ratio of cohesive
to adhesive forces is higher for FF leading to a higher amount of FF agglomerates not being adhered
to the carrier surface.
Findings demonstrate that interparticulate interactions are influenced by a variety of factors and in this
case not mainly by chemical surface properties.

3.4. Assessment of aerodynamic performance

14

Aerosolisation performance of interactive mixtures was determined using the NGI. The fine particle
fraction (FPF) was chosen as the suitable measure and comparative parameter for aeordynamic
performance.
Fig. 6 shows a significant decrease of respirable fraction of FF containing blends induced by applying
HF treatment to the GBs, when comparing blends containing GB_UT with an FPF of 25.6 0.9 %
(100 % CSC) to HF10min resulting in an FPF of 15.1 0.5 %. The same trend can be seen when
comparing HF+TC4h to TC4h and HF+TC8h to TC8h. Here again, an additional processing step (HF
treatment) significantly reduced the FPF (p< 0.01; p< 0.001). In contrast, milling with tungsten
carbide enhanced the FPF. Comparing HF10min, HF+TC4h and HF+TC8h, respirable fractions
increased from 15.1 0.5 % over 20.4 1.1 % to 23.6 0.2 %. The exact same trend was also
observed for TC4h and TC8h. FPFs of 28.18 0.2 % and 26.1 0.8 % were found, respectively
(p< 0.01).
Comparing the aerodynamic performance of blends containing BUD, the overall trends for blends
with different modified glass beads and FF were also present here. HF treatment led to a deteriorated
aerodynamic performance as GB_UT exhibited a respirable fraction of 19.5 1.4 %, while HF10min
only showed an FPF of 6.8 0.7 %. Surprisingly, the percent decrease was more pronounced
for BUD.
The introduction of surface roughness on nano scale by milling increased the FPF for HF+TC4h and
HF+TC8h from 10.4 0.6 % to 18.9 0.7 % as well as for TC4h and TC8h from 21.9 0.7 % to 24.4
0.6 %. Here, the same trend as for FF was observed which is quite unexpected as chemical
properties of the APIs differed remarkably.
In BUD as well as FF trials, FPFs varied depending on carrier surface coverage with higher FPFs
being observed at higher API content. This applied to every type of GB under investigation in the
present study. This observation might be explained in a coherent way for blends prepared containing
modified GBs. As seen in SEM pictures (Fig. 8 (left)), an accumulation of API particles in deep
cavities took place. The phenomenon of preferable binding to certain areas was first described by
Staniforth (1994) investigating lactose carrier particles and was attributed to high energy spots. This
theory was revised in 2014, proposing to define the activity of surface areas by their ability to shelter
15

API particles from removal forces during inhalation (Grasmeijer et al., 2014), directly linking drug
dispersion to site activity.
As already mentioned, API particles are initially deposited in those active sites during blending
leading to a higher percentage of API accumulated there at lower drug loadings compared to higher
loads where a large proportion of API is located elsewhere. Consequently, a positive correlation
between surface coverage and respirable fraction can be established (up to a certain extent). These
considerations also apply to GB_UT, since they hold surface irregularities as well.
When comparing graphs from Fig. 6 and Fig. 7, overall higher FPFs were found for blends containing
FF. Again, this is somewhat surprising if only chemical surface properties of the APIs are concerned.
Nevertheless, these observations are well in line with results from section 3.3., where overall lower
TSCs were found for FF compared to BUD.
As previously discussed, the different modified GBs had varying TSCs. It was also shown, that the
surface coverage had a significant impact on aerodynamic performance. In order to profoundly
compare the different interactive mixtures regarding their FPFs, a normalisation step was introduced.
TSCs were in the first step plotted against their corresponding FPFs. Fig. 9 displays the graph for
GB_UT and BUD representatively for all other blends. In a second step, the FPF for a TSC of 30 %
was calculated for each blend from this. 30% true surface coverage was a number being reached by all
API-carrier combinations despite the differences in drug adhesion observed. Thus, this value was
chosen for comparison as it was in the range of actual possible surface coverage for all blends.
Fig. 10 displays the resulting FPFs for blends containing FF and BUD, respectively. Comparing those
results to Fig. 7 and Fig. 8, the same trends for the different modified glass beads can be found, but
values can now be compared directly without considering different TSCs.
Introducing deep cavities (m range) via HF treatment drastically decreased the fine particle fraction
of FF as well as BUD for HF10min. The SEM pictures of the blend with BUD showed an
accumulation of API particles in those clefts (Fig. 8 (left)). During inhalation, these particles are
protected from the airstream leading to a persisting adhesion on the GB surface. Again, this
hypothesis is confirmed by SEM micrographs of glass beads collected from the preseparator after
inhalation routine. Fig. 8 (right) shows cavities still occupied with API particles. Despite the high
16

flow rate of 100 L/min, air stream was obviously not strong enough to detach the API particles
leading to a drastic decreases respirable fraction. This is equally true when comparing HF+TC4h to
TC4h and HF+TC8h to TC8h. Here again, an additional processing step of HF treatment significantly
reduces the FPF. Similar observations occurred in a study by Littringer et al. (2012). Here, the authors
compared aerodynamic performances of blends containing almost spherical mannitol particles to
those with carrier particles holding large indentions with a diameter of up to 50 m. They also found a
decrease in FPF for the latter, which was attributed to a substantial amount of API accumulated in
those indentions.
In contrast, milling with tungsten carbide provided surface roughness on a nano scale, as it was
confirmed by AFM data. This process led to a diminished contact area between carrier and API thus
reducing the potential for interaction between the two. Less interaction equals lower adhesion forces
leading to a facilitated drug detachment and consequently to an improved FPF. In addition, a
prolonged milling time enhanced that effect as it was proven by NGI trials. The exact same trend was
also observed for TC4h and TC8h. These findings are well in accordance with results obtained by
Kawashima et al (1998). They found a lower aerodynamic performance for smoothed lactose particles
compared to nanostructured carrier surfaces due to an increase in contact area ultimately enhancing
van der Waals attractive forces.
Although the two APIs are quite different in terms of chemical properties, the same trend as for FF
can be observed for BUD. Previous studies by Zellnitz et al. (2015) investigated the influence of
surface topography on the aerosolisation performance. Here, surface roughness was generated by
milling with quartz and plasma etching. Results obtained from that study showed a positive
correlation between surface roughness and respirable fraction. Apparently a decrease in contact area,
created through nano surface roughness leads to a facilitated drug detachment regardless of the
chemical properties of the API.
It needs to be pointed out, that this study focused on the Cyclohaler as inhalation device, where
deagglomeration/separation is mainly achieved through drag and lift forces (de Boer et al., 2006). API
particles located in deep cavities have been shown to remain mainly unaffected by those forces and
therefore are prevented from detaching. In contrast, in case of the Novolizer and the Turbuhaler,
17

redispersion is caused in particular by inertial forces, where surface roughness on a larger scale is not
hindering drug detachment. On the contrary, de Boer et al. (2003) illustrated even a possible positive
effect, inter alia, because of a reduction of press-on forces during blending.
Hence, present results addressing the optimization of aerodynamic performance by altering carrier
surface properties are most applicable to inhalation devices with similar redispersion mechanism as
the Cyclohaler, i.e. the Diskus. Furthermore, it can be concluded that in order to enhance
aerodynamic performance, one has not only to consider aspects concerning the actual interactive
mixture of carrier and drug, but also should place emphasis on the inhalation device itself.

4. Conclusion
The present study was designed to investigate the effect of carrier surface roughness on properties of
interactive blends for dry powder inhalation and their aerodynamic performance. In the course of this,
the presented techniques used for altering carrier surface proved to be suitable to create surface
roughness on a micro as well as nano scale. Results showed that roughness scale plays an important
role and has to be specified. Introducing roughness on micro-scale via HF treatment drastically
reduced the FPF attributed to the ability of created cavities to protect API particles from the airstream
during inhalation. In contrast, grinding with tungsten carbide enhanced the aerodynamic performance
which may be due to a reduction of contact area. Taking results from trials with both APIs into
account, it can be concluded, that the effect of surface roughness on aerodynamic performance is
largely independent from chemical properties of the API. This does not apply as far as surface
coverage is concerned. Here, different degrees of surface roughness have an opposite effect on the
TSC for BUD and FF.
Further studies are needed to elucidate the relationship between effective drug loading, interparticle
interactions and physico-chemical properties of the API. Therefore future steps will include
quantification of adhesion forces using AFM as well as determination of carrier surface energies.

Acknowledgements
18

The authors acknowledge the funding of this research project by the German research foundation
(DFG) within the priority program SPP 1486 Particles in Contact. The authors would also like to
thank the research group of Dr. Michael Kappl and Dr. Regina Fuchs (Max Planck Institute for
Polymer Research, Mainz) for providing technical support and equipment for AFM measurements and
Ann-Kathrin Muhs for her technical assistance in NGI trials.

References
de Boer, A.H., Hagedoorn, P., Gialtema, D., Goede, J., Kussendrager, K., Frijlink, H., 2003. Air classifier
technology (ACT) in dry powder inhalation Part 2. The effect of lactose carrier surface properties on the
drug-to-carrier interaction in adhesive mixtures for inhalation. Int. J. Pharm. 260, 201-216.
de Boer, A.H., Hagedoorn, P., Gialtema, D., Goede, J.,Frijlink, H., 2006. Air classifier technology (ACT) in dry
powder inhalation Part 4. Performance of air classifier technology in the Novolizer mulit-dose dry powder
inhaler. Int. J. Pharm. 310, 81-89.
Bosquillon, C., Lombry, C., Prat, V., Vanbever, R., 2001. Influence of formulation excipients and physical
characteristics of inhalation dry powders on their aerosolization performance. J. Controlled Release 70,
329339.
Chan, L., Lim, L., Heng, P., 2003. Immobilization of Fine Particles on Lactose Carrier by Precision Coating and
Its Effect on the Performance of Dry Powder Formulations. J. Pharm. Sci. 92, 975-984.
Chew, N., 2000. Effect of particle size, air flow and inhaler device on the aerosolisation of disodium
cromoglycate powders. Int. J. Pharm. 206, 7583.
Coates, M., Fletcher, D., Chan, H.-K., Raper, J., 2004. Effect of Design on the Performance of a Dry Powder
Inhaler Using Computational Fluid Dynamics. Part 1: Grid Structure and Mouthpiece Length. J. Pharm. Sci.
93, 28632876
Columbano, A., Buckton, G., Wikeley, P., 2002. A study of the crystallization of amorphous salbutamol
sulphate using water vapour sorption and near infrared spectroscopy. Int. J. Pharm. 237, 171-178.
Dickhoff, B.H., de Boer, A.H. , Lambregts, D., Frijlink, H.W., 2005. The interaction between carrier rugosity
and carrier payload, and its effect on drug particle redispersion from adhesive mixtures during inhalation.
Eur. J. Pharm. Sci. 56, 291-302

19

El-Sabawi, D., Edge, S., Price, R., Young, P.M., 2006. Continued Investigation Into the Influence of Loaded
Dose on the Performance of Dry Powder Inhalers: Surface Smoothing Effects. Drug Dev. Ind. Pharm. 32,
11351138.
Grasmeijer, F., Frijlink, H.W., Boer, A.H. de, 2014. A proposed definition of the 'activity' of surface sites on
lactose carriers for dry powder inhalation Eur. J. Pharm. Sci. 56, 102104.
Guenette, E., Barrett, A., Kraus, D., Brody, R., Harding, L., Magee, G., 2009. Understanding the effect of
lactose particle size on the properties of DPI formulations using experimental design. Int. J. Pharm. 380,
8088.
Harjunen, P., Lehto, V., Martimo, K., Suihko, E., Lankinen, T., Paronen P., Jrvinen, K., 2002. Lactose
modifications enhance its drug performance in the novel multipledose Taifun DPI. Eur. J. Pharm. Sci. 16,
313-321.
Hersey, J.A., 1975. Ordered mixing: A new concept in powder mixing practice. Powder Technol. 11, 41-44.
Hickey, A.J., Concession, n.M., von Oort, M.M., Platz, R.M.. Factors influencing the dispersion of dry powders
as aerosols. Pharm. Techn. 18, 58-64
Iida, K., Hayakawa, Y., Okamoto, H., Dajno, K., Luenberger, H., 2004. Effect of Surface Layering Time of
Lactose Carrier Particles on Dry Powder Inhalation Properties of Salbutamol Sulfate. Chem. Pharm. Bull.
52, 35035.
Kaialy, W., Alhalaweh, A., Velaga, S.P., Nokhodchi, A., 2011. Effect of carrier particle shape on dry powder
inhaler performance. Int. J. Pharm. 421, 12-23
Kaialy, W., Nokhodchi, A., 2013. Freeze-dried mannitol for superior pulmonary drug delivery via dry powder
inhaler. Pharm. Res. 30, 458477.
Kaialy, W., Nokhodchi, A., 2015. Dry powder inhalers: physicochemical and aerosolization properties of
several size-fractions of a promising alterative carrier, freeze-dried mannitol. European journal of
pharmaceutical sciences: official journal of the European Federation for Pharmaceutical Sciences 68, 5667.
Karner, S., Littringer, E.M., Urbanetz, N.A., 2014. Triboelectrics: The influence of particle surface roughness
and shape on charge acquisition during aerosolization and the DPI performance. Powder Technol. 262, 2229.
Kawashima, Y., Serigano, T., Hino, T., Yamamoto, H., Takeuchi, H., 1998. Effect of surface morphology of
carrier lactose on dry powder inhalation property of pranlukast hydrate. Int. J. Pharm 172, 179188.

20

Klapetek P., Necas D., Anderson C., 2012. Gwyddion user guide, 1126. http://gwyddion.net/download/userguide/gwyddion-user-guide-en.pdf
Kou, X., Chan, L.W., Steckel, H., Heng, P.W., 2012. Physico-chemical aspects of lactose for inhalation.
Advanced drug delivery reviews 64 (3), 220232.
Labiris, N.R., Dolovich, M.B., 2003. Pulmonary drug delivery. Part I: Physiological factors affecting therapeutic
effectiveness of aerosolized medications. Br. J. Clin. Pharmacol.56, 588599.
Le, V.N.P., Robins, E., Flament, M.P., 2012. Agglomerate behaviour of fluticasone propionate within dry
powder inhaler formulations. Eur J Pharm Biopharm 80, 596603.
Littringer, E.M., Mescher, A., Schroettner, H., Achelis, L., Walzel, P., Urbanetz, N.A., 2012. Spray dried
mannitol carrier particles with tailored surface properties--the influence of carrier surface roughness and
shape. Eur. J. Pharm. Biopharm. 82, 194204.
Price, R., Young, P., Edge, S., Staniforth, J., 2002. The influence of relative humidity on particulate interactions
in carrier-based dry powder inhaler formulations. Int. J. Pharm 246, 4759.
Staniforth, J.N., 1995. Performance-Modifying Influences in Dry Powder Inhalation Systems. Aerosol Sci.
Technol. 22, 346353.
Steckel, H., Mueller, B.W. 1997. In vitro evaluation of dry powder inhalers II: influence of carrier particle
size and concentration on in vitro deposition. Int. J. Pharm. 154, 31-37.
Young, P.M., Wood, O., Ooi, J., Traini, D., 2011. The influence of drug loading on formulation structure and
aerosol performance in carrier based dry powder inhalers. Int. J. Pharm. 416, 129135.
Zellnitz, S., Redlinger-Pohn, J.D., Kappl, M., Schroettner, H., Urbanetz, N.A., 2013. Preparation and
characterization of physically modified glass beads used as model carriers in dry powder inhalers. Int. J.
Pharm. 447, 132138.
Zellnitz, S., Schroettner, H., Urbanetz, N.A., 2015. Influence of surface characteristics of modified glass beads
as model carriers in dry powder inhalers (DPIs) on the aerosolization performance. Drug Dev Ind Pharm.
41,1710-7.
Zeng, X.M., Martin, G.P., Marriott, C., Pritchard, J., 2000. The influence of carrier morphology on drug
delivery by dry powder inhalers. Int. J. Pharm 200, 93106.

21

Figure legends

Fig. 1. SEM micrographs of spray dried FF (left) and BUD (right)

22

Fig. 2. SEM micrographs of untreated and modified glass bead surfaces (upper row 100x, middle row 500x and lower
row 2500x)

23

Fig. 3. Representative AFM surface images of untreated and different modified glass beads used in this study

24

Fig. 4. Left: TSCs for different modified glass beads in blends with FF, Right: CSC of the respective blends plotted
against the resulting TSC

25

Fig. 5. Left: TSCs for different modified glass beads and BUD Right: CSC plotted against the resulting TSC

26

Fig. 6. FPFs of blends containing different glass beads and FF for CSCs of 100%, 50% & 25%

27

Fig. 7. FPFs of blends containing different glass beads and BUD for CSCs of 100%, 50% & 25%

28

Fig. 8. SEM micrographs of blends of 50% CSC containing HF10min and BUD before (left) and after inhalation (right)

29

Fig. 9. Correlation of TSC and FPF for GB_UT with BUD

30

Fig. 10. Resulting FPFs at 30% TSC for untreated and modified GBs

31

Tables

Table 1. Spray drying conditions used for FF and BUD

Parameter
Inlet temperature [C]
Solid concentration
[% (m/V)]
Feed rate [ml/min]
Aspirator rate [m/h]
Spray gas flow [l/h]

BUD
120

FF
120

10

4.5

3.15
35
414

2.10
35
357

Table 2. Amounts of API corresponding to the different CSCs

Amount of API [% (m/m)]

BUD
FF
25 % CSC

0.26

0.36

50 % CSC

0.52

0.72

100 % CSC

1.03

1.45

Table 3. Solvents used for the different APIs

API
BUD
FF

Solvent
75 % Methanol
25% dd H2O
45 % dd H2O
40 % Methanol
15 % Acetonitrile

32