A method for thermal diffusivity measurement in fluids

E. Marn, E. Hernndez-Rosales, A. M. Mansanares, R. Ivanov, J. B. Rojas-Trigos, and A. Caldern

Citation: Review of Scientific Instruments 84, 104903 (2013); doi: 10.1063/1.4824191
View online: http://dx.doi.org/10.1063/1.4824191
View Table of Contents: http://scitation.aip.org/content/aip/journal/rsi/84/10?ver=pdfcov
Published by the AIP Publishing

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:
129.210.19.225 On: Wed, 04 Dec 2013 23:05:54

REVIEW OF SCIENTIFIC INSTRUMENTS 84, 104903 (2013)

A method for thermal diffusivity measurement in fluids

E. Marn,1,a) E. Hernndez-Rosales,1 A. M. Mansanares,2 R. Ivanov,3 J. B. Rojas-Trigos,1
and A. Caldern1
1
Instituto Politcnico Nacional, Centro de Investigacin en Ciencia Aplicada y Tecnologa Avanzada,
Legara 694, Colonia Irrigacin, C. P. 11500, Mxico D. F., Mxico
2
Instituto de Fsica Gleb Wataghin, Universidade Estadual de Campinas, UNICAMP, Cx. P. 6165,
13083-970 Campinas, SP, Brazil
3
Facultad de Fsica, Universidad Autnoma de Zacatecas, Calz. Solidaridad Esquina Paseo de la
Bufa s/n, C. P. 98060, Zacatecas, Zac., Mxico

(Received 9 July 2013; accepted 20 September 2013; published online 8 October 2013)
A technique is proposed for thermal diffusivity measurement in fluids. It is based on the Angstrom
method, but with excitation of thermal waves by electromagnetic energy absorption and pyroelectric
detection. The good agreement between measured thermal diffusivity of air and some test liquids
with literature values shows the validity of the method. It is free of some limitations of conventional
photopyroelectric technique with length scanning because it is free of moving parts inside the sample
and because it avoids problems associated with the non-parallelism between thermal wave generator
surface and sensor. It does not require any data normalization procedure or special sample preparation.
2013 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4824191]
I. INTRODUCTION

The photothermal techniques are a group of experimental methods based on the measurement of heat generated in a
given sample.1 In one of the simplest and ancient approaches,
the so-called Angstrom method,2 an extreme of a solid rod is
heated periodically and the temperature oscillations with time
are measured at different distances from the heating source.
Neglecting border effects and for sinusoidal modulation of
heat, the temperature evolution with time at any point x of
the rod is given by a simple equation:
T = T0 exp(x/) cos(2f t x/ + 0 )
= A cos(2f t + ).

(1)

It has the same mathematical form as the equations describing a critically damped wave and it is therefore named a
thermal wave. In Eq. (1) t is the time, A = T0 exp (x/) is
the amplitude,  = x/ + 0 is a phase lag, is the samples thermal diffusivity, f is the modulation frequency, and
= (/ f)1/2 is the thermal diffusion length, i.e., the distance
from the samples surface at which the thermal wave amplitude decreases e-times from its value at the surface. Note that
both T0 and 0 are constants independent on the samples
length. It can be seen that the thermal diffusivity can be calculated straightforwardly from the slopes of the straight lines
resulting in ln(A) versus x and  versus x graphs, if f is well
known.
Although this method gives accurate results for extended, large solid samples (one practical application is the
measurement of the thermal diffusivity of soils using as a
heat source the daily and/or seasonal periodical oscillations
of ambient temperature3 ), its main disadvantage is the large
a) Author to whom correspondence should be addressed. Electronic mail:

emarinm@ipn.mx

0034-6748/2013/84(10)/104903/5/$30.00

volume of samples involved, which limits its application in a

general way.4 But in the case of liquids the photopyroelectric5
approach overcomes this limitation. In the photopyroelectric
technique the temperature oscillations are measured with a
pyroelectric sensor in contact with the sample. This technique
has gained considerable attention for the thermal characterization of materials, in particular of gases and liquids due to the
excellent thermal contact that is achieved between them and
the pyroelectric sensor. Different measurement configurations
have been proposed. One of the most useful experimental approaches is the so-called back detection configuration, which
is suitable for thermal diffusivity measurement.6 In this variant, the sample is sandwiched between a pyroelectric sensor
and a thermal wave source (it can be a periodical heated thin
metallic foil covering the surface in case of gaseous or liquid
samples). Measurements are often performed in a modulation
frequency scan mode but a samples length scan mode
avoids the use of normalization procedures to account for
the modulation frequency dependent instrumental function.7
This mode is very similar to the above described Angstrom
method, and has been realized before by stepwise changing
the distance between the sensor and the thermal wave source
in a range of few micrometers to millimeters.8 However,
for accurate measurements, good parallelism between both
sensor (usually a polyvinylidene-difluoride (PVDF) sensor
with some millimeters in diameter) and thermal waves source
(generally a heated metallic disc with approximately the same
dimensions as the sensor) must be guaranteed, which is difficult to be achieved in practice. Furthermore, in this dynamical
mode of measurement the homogeneity of the sample can
be affected due to local fluid displacements produced by
the moving parts of the experimental device. This can be
especially counterproductive in the case of inhomogeneous
samples such as colloids and other composites. Therefore,
in this work a method that overcomes these limitations is
described.

84, 104903-1

2013 AIP Publishing LLC

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:
129.210.19.225 On: Wed, 04 Dec 2013 23:05:54

104903-2

Marin et al.

FIG. 1. Cross-sectional schematic view of the measurement cell.

II. PROPOSED EXPERIMENTAL SETUP

The basic of our experimental setup is schematically

shown in Fig. 1. The thermal wave source is a 300 m thick
Silicon slab, whose surface is heated along a straight line by
a focused (using an optical fiber) intensity modulated diode
laser beam (Fiber Laser BWF-785-450E/55371, Edmund
Optics, 785 nm, 200 mW). In our set-up, the position of the
laser was maintained fixed and the measurement cell was
moved with respect to it (2 m resolution steps) using the
automated translation stage of an optical microscope (Optical
Measurement Inspection System RAM Optical Instrumentation 01-1001-01XYZ). Silicon has a high thermal diffusivity
(8.8 105 m2 /s),9 so that at the used modulation frequency
(18 Hz) it is thermally thin (the thermal diffusion length
defined above is much greater than the samples thickness)
and the thermal waves generated by the superficial absorption of light at one of its surfaces diffuse very fast to the
non-illuminated side, namely at the samples surface. Furthermore, the Si slab has smooth and plane surface in contact
with the liquid that guarantees the desired sample thickness
uniformity. The pyroelectric sensor is the active component
of a commercial Buzzer (BGD, Steren). It consists, as
demonstrated by Energy Dispersive Spectroscopy (Oxford
Pentafet X-ray detector coupled to a JEOL JSM-6390LV
Scanning Electron Microscope) of a PZT (lead zirconate
titanate) ceramic material (0.481 mm thick) on a copper
metal plate (0.323 mm) and with the other side (the surface
in contact with the sample in our experiments) covered with
a thin layer of silver. Both metals serve as electrical contact
electrodes. The stiffness of both Si slab and PZT sensor avoid
problems related with bending that can appear when thin
metallic sheets and PVDF pyroelectric foils are used. It is
assumed that at each point of incidence of the laser beam, i,
the corresponding samples length beneath it becomes Li .
The studied samples were dry air, distilled water, standard olive oil (pH = 6), ethylene glycol (J. T. Baker 99.90%),
and glycerin (J. T. Baker 99.90%). The measurement cell allows simple loading and replacement of the samples. It is
constructed in acrylic material and the volume of sample was

Rev. Sci. Instrum. 84, 104903 (2013)

0.2 ml. Thus, in addition to the above-mentioned advantages,

this small amount of sample can make the method very useful
for several applications related to thermal characterization of
liquids for which only very small quantities are available for
measurement (e.g., colloidal suspensions of nanosized particles, reported as potential candidates for thermal management
systems, for which the development of accurate thermal properties measurement methods has become essential10 ). The pyroelectric voltage, Vi = F Ti , is proportional to the average
temperature, Ti , on the sensor for each value of the sample
thickness Li , and its amplitude and phase can be measured using a Lock-in amplifier (SR-830, Stanford Research Systems)
synchronized at the light modulation frequency. The proportionality factor F, also called the instrumental transfer function, depends on the modulation frequency but it does not depend on the samples thickness. This is the reason of why no
normalization is necessary when scan-length mode measurements are performed. The uniformity of the PZT response
coming from the sensing area has been verified by previously scanning its surface with a laser beam and mapping the
generated PZT signal.

III. RESULTS AND DISCUSSION

A. Experimental results

Figure 2 shows typical curves of amplitude and phase of

the pyroelectric signal as a function of the samples length for
air and Fig. 3 for some test liquids. Measurements were performed at room temperature (26.6 C). The solid and dashed
lines are the best least squares fits according to Eq. (1). From
the slopes, p, thermal diffusivity is calculated as = f/p2 .
The thermal diffusivity mean values of at least 10 measurements are shown in Table I, together with literature reported
values for comparison. The relative measurement uncertainties are about 2%3%.

FIG. 2. Amplitude (circles) and phase (triangles) of the pyroelectric signal

as a function of samples thicknesses for air. Detector slope: a = 0.6. The
solid and dashed curves are the best least squares fits.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:
129.210.19.225 On: Wed, 04 Dec 2013 23:05:54

104903-3

Marin et al.

Rev. Sci. Instrum. 84, 104903 (2013)

FIG. 3. Amplitude (circles) and phase (triangles) of the pyroelectric signal as a function of samples thicknesses for some liquids. The solid and dashed curves
are the best least squares fits. (a) Distilled water (detector slope: a = 0.4); (b) olive oil (a = 0.2); (c) glycerin (a = 0.4); and (d) ethylene glycol (a = 0.3).

B. Theoretical aspects

Figure 4 depicts the geometry of the cell including the

heat source profile I (r ) at the upper surface of the sample,
which is related to temperature profile at the bottom of the
Si slab. A point-like heat source at the top of the Si would
be broadened at its bottom by, roughly, the thickness of the
slab. Therefore, for the sake of the analysis, let us consider
heat source at the upper surface of the sample (liquid or gas),
with size of the order of 300 m (Gaussian source with Rsource
= 300 m, for instance). The average temperature over the
entire sensor (slanted plane in Fig. 4), due to the heat source

distributed at the upper surface of the sample (z = 0), considering cylindrical symmetry (I(r), centered at the origin), is
given by

T  =

1
As Ad

 



T (r , r ) d 2 r 

I (r) d r
source

detector

(2)

with


T (r , r ) =

1
2 k



e s
s


(3)

TABLE I. Measured values of thermal diffusivity as determined from the amplitude and phase dependences on
samples thicknesses. Literature values are shown for comparison.
Sample

(amplitude) (m2 /s)

(phase) (m2 /s)

Air
Distilled water
Ethylene glycol
Glycerin
Olive Oil

(2.15 0.02) 105

(1.50 0.05) 107
(9.31 0.04) 108
(9.20 0.08) 108
(1.40 0.09) 107

(2.20 0.01) 105

(1.41 0.03) 107
(9.29 0.02) 108
(9.21 0.04) 108
(1.40 0.05) 107

(Literature) (m2 /s)

2.2 105
1.4 107
9.3 108
9.2 108
1.4 107

(Ref. 9)
(Ref. 9)
(Ref. 11)
(Ref. 9)
(Ref. 12)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:
129.210.19.225 On: Wed, 04 Dec 2013 23:05:54

104903-4

Marin et al.

Rev. Sci. Instrum. 84, 104903 (2013)

FIG. 4. Geometry of the cell including the heat source profile I (r ) at the
upper surface of the sample.

being the Green function which represents the temperature at

the detector point r = (x , y , z ) due to the point heat source
located at r = (x, y, z). Furthermore,
=

(1 + j )

(4)

is the complex thermal diffusion coefficient, k is the thermal

conductivity, and the distance s is given by
s = |r  r| = [(x  x)2 + (y  y)2 + (z )2 ]1/2

(5)

with
z = Li + ax  .
Here a = tan( ) is the detector slope (see Fig. 4). The areas As and Ad refer to the effective source area and the detector
area respectively, as follows:
 
1
As =
I (r)d 2 r,
(6)
I0
source
where I0 being the source intensity at its center;

d 2r 
Ad =

(7)

detector

(the detector is considered large enough for all purposes

described herein, so that border effects are neglected).
If the slope of the detector is very small (a 0), the detector plane becomes almost parallel to the source one. In this
case, the value of T remains the same when we interchange
r and r. Such operation, however, is equivalent to interchanging source and detector. In other words, T can be thought
as the average temperature oscillation produced on a detector
by a flat heat source at the sample upper surface. Of course,
this is a one-dimensional heat flow problem, and the signal
dependence with the thickness Li is straightforward given by
Eq. (1), as in the case of Angstrom experiment.
C. Numerical calculation

In order to evaluate the deviation introduced by the

slanted detector (a = 0), we carried out numerical calculations of Eq. (1) for several values of the detector slope, a.
Figure 5 shows the amplitude and phase of the calculated

FIG. 5. (a) Amplitude and (b) phase of the numerically calculated T for
a sample with thermal diffusivity equal to 1.5 107 m2 /s and modulation
frequency of 18 Hz. Several values of the detector slope were used (a = 0,
0.1, 0.2, 0.3, 0.4, and 0.5).

T for a sample with thermal diffusivity equal to 1.5 107

m2 /s, a value close to that of water. Modulation frequency was
set as 18 Hz, the same employed in our measurements. Numerical calculation was performed using appropriate source
and detector meshes, with self-consistent convergence criteria. As one can see, the behavior of both amplitude and phase
with the distance Li remains almost the same of those obtained using the one-dimensional theory. Similar result was
obtained for air thermal diffusivity. This is consistent with
the measured curves of Figs. 2 and 3 and allows us to use
the proposed method with high confidence. Such a result is
somehow expected, once the slanted detector averages over
distances smaller than Li in one of its half, and over distances
bigger than Li in the other half. If one expands the temperature function in the detector plane around its value at the center of the detector (x = 0, y = 0, z = Li ), the first order term
will cancel out from one half to the other, and the correction
will remain on the second order terms and higher. This explains the good agreement between the measurements and the
one-dimensional theory.
Both in the numerical calculation and in the theoretical
analysis of this section, we disregarded the role of thermal
wave reflection at the sample interfaces with adjacent media.
The more relevant reflection we should consider occurs in the
sample-detector interface. However, using the image method,
such a detector plane can be treated as a semi-transparent
mirror (of reflectance coefficient R), giving rise to an image
source which is completely symmetrical to the original source
at the top of the sample for each detector pixel/position.
Therefore, the consequence of the thermal wave reflection
is to change the amplitude by a constant factor, without

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:
129.210.19.225 On: Wed, 04 Dec 2013 23:05:54

104903-5

Marin et al.

affecting the behavior of the signal (amplitude and phase).

This is also an advantage of the proposed technique when
comparing with previous reported photopyroelectric approaches. Additional multiple thermal wave reflections are
less and less important, thus becoming negligible.

Rev. Sci. Instrum. 84, 104903 (2013)

The support from COFAA-IPN, Mexico, through the SIBE

and PIFI Programs, and CNPq is also acknowledged. Authors also dank E. Cedeo and G. Juarez-Gracia for help with
measurements.
1 D.

IV. CONCLUSIONS

In conclusion, the technique proposed here is a straightforward, accurate, and cheap method for thermal diffusivity
measurement. As mentioned above, the small sample amounts
required render this technique as a useful tool for thermal
characterization of liquids and gases but also of semi-liquids
such as pastes and gels. Because it is free of moving parts inside the sample, it is a convenient method for characterization
of non-homogeneous materials. Moreover, a compact and robust measurement cell can be devised, allowing the construction of a portable device useful for field measurements such
as those outlined in Ref. 13. Cells temperature can be also
controlled using for example Peltier elements so that temperature dependence on thermal diffusivity can be monitored
with applications for phase transitions detection and related
phenomena.
ACKNOWLEDGMENTS

This work was supported by research grants SIP-IPN1491, CONACyT-83289, and CONACyT-2011-01-174247.

P. Almond and P. M. Patel, Photothermal Science and Techniques in Physics and its Applications (Chapman and Hall, London,
1996), Vol. 10.
2 A. J. Angstrom, Ann. Phys. Lpz. 114, 513 (1861).
3 E. Marn, E. Jean-Baptiste, and M. Hernandez, Rev. Mex. Fis. E 52, 21
(2006).
4 J. B. Rojas-Trigos, A. Bermejo Arenas, and E. Marn, Eur. J. Phys. 33, 135
(2012).
5 M. Chirtoc and G. Mihilescu, Phys. Rev. B 40, 9606 (1989).
6 M. Chirtoc, I. Chirtoc, D. Bicanic, and J. Pelzl, J. Ferroelectr. 165, 27
(1995).
7 A. Mandelis and A. Matvienko, Photopyroelectric thermal-wave cavity
devices-10 years later, in Pyroelectric Materials and Sensors (Research
Signpost, Kerala, India), p. 61.
8 J. A. P. Lima, M. G. da Silva, M. S. O. Massunaga, E. Marn, H. Vargas,
and L. C. M. Miranda, J. Appl. Phys. 91, 5581 (2002).
9 Y. S. Touloukian, R. W. Powell, C. Y. Ho, and M. C. Nicolau, Thermal
Diffusivity (IFI, New York, 1973).
10 S. M. S. Murshed, K. C. Leong and C. Yang, Appl. Therm. Eng. 28, 2109
(2008).
11 E. R. G. Eckert and R. M. Drake, Jr., Analysis of Heat and Mass Transfer
(Hemisphere, New York, 1987).
12 D. Dadarlat, J. Gibkes, D. Bicanic, and A. Pasca, J. Food Eng. 30, 155
(1996).
13 E. Marn and H. Vargas, Recent developments in thermal wave interferometry, in Thermal Wave Physics and Related Photothermal Techniques:
Basic Principles and Recent Developments (Transworld Research, Kerala,
India, 2009), p. 99.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:
129.210.19.225 On: Wed, 04 Dec 2013 23:05:54