Langmuir 2002, 18, 5113-5119

5113

Simultaneous Determination of Toluene Diffusion

Coefficient in Air from Thin Tube Evaporation and Sessile
Drop Evaporation on a Solid Surface
H. Yildirim Erbil* and Yonca Avci
Department of Chemistry, Faculty of Sciences & Arts, Kocaeli University,
41300 Izmit, Kocaeli, Turkey
Received October 15, 2001. In Final Form: March 11, 2002

In this work, we simultaneously determined the diffusion coefficient of toluene vapor in air by following
the toluene evaporation in two different geometries: in a thin glass tube and a sessile drop on poly(tetrafluoroethylene) (Teflon) substrate in the same cell by inducing a quasi-steady liquid-vapor phase
transition. A glass cell whose walls were lined with activated carbon was used in both experiments.
Photography was applied to monitor the sessile drop evaporation, which occurred with constant contact
angle mode (approximately 43). When the restriction of evaporation space by the presence of the substrate
surface [f() factor] was considered in the sessile drop evaporation, the toluene vapor diffusion coefficient
in air was found to be close in both methods. The diffusion coefficient was found to vary linearly with the
temperature for both methods, and there is a 3.17 power dependence when the absolute temperatures are
used. However, this figure is higher than the 1.75-2.00 power dependence, which is generally used for
evaporating liquid vapors in the existing literature. In addition, an equation was derived showing the
decrease of the square of the height of the sessile drop should vary linearly with time and was confirmed
with the experimental results. Similarly, the decrease of the square of the contact radius of the drops was
found to vary linearly with time in the experiments as expected from the theory, which was previously
derived.

Introduction
The mechanism of the evaporation of sessile liquid drops
during contact angle () measurement is the subject of
many recent papers.1-15 The occurrence of liquid evaporation is inevitable unless the atmosphere in the immediate
vicinity of a sessile drop is saturated with the vapor of the
liquid and drop evaporation usually results in a decrease
of . Thus, a more complete understanding of how
evaporation influences the contact angle of the drop on a
solid surface in still air or in controlled atmospheric
conditions is very important in the wetting and surface
characterization processes. There are two modes of
evaporation of liquid drops on surfaces: that at constant
contact area with diminishing contact angle and that at
constant contact angle with diminishing contact area.
* Corresponding author: e-mail: yerbil@kou.edu.tr; Fax, 90 (262)
331 39 06.
(1) Picknett, R. G.; Bexon, R. J. Colloid Interface Sci. 1977, 61, 336.
(2) Birdi, K. S.; Vu, D. T.; Winter, A. J. Phys. Chem. 1989, 93, 3702.
(3) Birdi, K. S.; Vu, D. T. J. Adhes. Sci. Technol. 1993, 7, 485.
(4) Rowan, S. M.; Newton, M. I.; McHale, G. J. Phys. Chem. 1995,
99, 13268.
(5) Shanahan, M. E. R.; Bourges, C. Int. J. Adhes. Adhes. 1994, 14
(3), 201.
(6) Bourges-Monnier, C.; Shanahan, M. E. R. Langmuir 1995, 11,
2820.
(7) Erbil, H. Y.; Meric, R. A. J. Phys. Chem. B 1997, 101, 35, 6867.
(8) Meric, R. A.; Erbil, H. Y. Langmuir 1998, 14, 1915.
(9) Rowan, S. M.; McHale, G.; Newton, M. I.; Toorneman, M J. Phys.
Chem. 1997, 101, 1265.
(10) McHale, G.; Rowan, S. M.; Newton, M. I.; Banerjee, M. K. J.
Phys. Chem. B 1998, 102, 1964.
(11) Erbil, H. Y. J. Phys. Chem. B 1998, 102, 9234.
(12) Erbil, H. Y. J. Adhes. Sci. Technol. 1999, 13, 1405.
(13) Erbil, H. Y.; McHale, G.; Rowan, S. M.; Newton, M. I. J. Adhes.
Sci. Technol. 1999, 13, 1375.
(14) Erbil, H. Y.; McHale, G.; Rowan, S. M.; Newton, M. I. Langmuir
1999, 15, 7378.
(15) Erbil, H. Y.; McHale, G.; Newton, M. I. Langmuir 2002, 18,
2636.

Picknett and Bexon1 was the first to distinguish these

two modes. They developed a theory to predict the
evaporation rate and residual mass at any time in the life
of the drop based on the spherical cap geometry. Shanahan
and Bourges have also obtained the direct measurements
of the change of , h, and rb with time and have shown the
existence of four distinct stages in the evaporation process
in open air conditions.5,6 Constant contact area mode is
the dominating evaporation mode for water and many
other drops on most of the solids when the contact angle
is less than 90.4-9 McHale et al.10 monitored the evaporation of water droplets on poly(tetrafluoroethylene)
(Teflon) where > 90 and the evaporation is dominated
by an initial stage with constant contact angle and
decreasing rb. A new model based on diffusion across the
liquid-vapor interface has been developed which fits well
the experimental data and predicts a linear dependence
on time for the square of the contact radius.10 Erbil et
al.11-14 applied ellipsoidal geometry to drop evaporation
and derived a method to determine peripheral contact
angle from evaporation rate. More recently, Erbil et al.15
reported the evaporation of 1-butanol, toluene, n-nonane,
and n-octane drops on Teflon which occurs with approximately constant contact angle mode where < 90.
Video microscopy and digital image analysis techniques
were applied to monitor the drop evaporation. The square
of the contact radius of these drops was found to decrease
linearly with time for most of the cases.15 The high linearity
was found in all the experimental (drop volume)2/3-time
plots, and previous diffusion models were discussed and
compared for this special case in this publication. The
Picknett and Bexon model, which considers the restriction
of evaporation space by the presence of the substrate
surface and which was derived from the diffusion of the
exact spherical shape analysis, gave the best fit for all the

10.1021/la011557a CCC: $22.00 2002 American Chemical Society

Published on Web 05/29/2002

5114

Langmuir, Vol. 18, No. 13, 2002

four drops examined.15 In all of the above-mentioned

studies, the diffusion coefficient values of the vapor of the
evaporating liquid drops were taken from the published
literature and their air temperature dependence was
calculated by the well-known equation (D1/D2) ) (T1/T2)n
(T in kelvin) where the power factor n )1.75-2.00 was
used.16,17
In this work, we compared the toluene vapor diffusion
coefficient in air, which is determined simultaneously in
two different geometries, in a thin glass tube and a sessile
drop on Teflon substrate in the same cell, and also we
tested the validity of the above temperature dependence
equation of diffusion coefficient. There are only a few
publications on the experimental determination of the
temperature dependence of the vapor diffusion coefficient,
and these papers reported widely scattered results.16,17,23-30
Experimental determination of D at various temperatures
is also required to compare the use of diffusive evaporation
models of sessile drops when the restriction of evaporation
space by the presence of the substrate surface is considered. For this purpose, we monitored the evaporation of
a toluene sessile drop on a Teflon substrate in a cell whose
walls are lined with activated carbon to maintain zero
vapor concentration at the cell wall. We chose this
experimental setup because toluene sessile drop evaporates with the constant contact angle mode on Teflon,
which is comparatively easy to analyze.15 We also determined the diffusion coefficients by applying the thin glass
tube method in the same cell for various temperatures in
order to find the temperature dependence. Then it is
possible to discuss the diffusive evaporation models on a
firm experimental basis.
Theory

Erbil and Avci

tube and then to the free atmosphere above the open end
of the tube. The diffusion coefficient, D, is determined
from observations of the slow loss of liquid from the tube
at constant temperature and pressure. The thin tube is
a cylinder of uniform cross section and usually with the
approximate dimensions of 5-10 mm diameter and 1020 cm length in order to avoid convection currents.17,19
The height of the air column in the thin tube above the
liquid level, hair, is measured during the experiments. hair
will increase slowly during the course of diffusive evaporation, and thus we are not dealing with steady state" in
the strict sense of the word but only a quasi-stationary
state. Therefore, neither the concentrations themselves
nor the concentration gradients are exactly constant and
independent of time (dc/dt 0). However, since the rate
of evaporation is slow and the gas-vapor column in the
tube changes little in height, and even though there are
steady losses of liquid due to evaporation, the diffusion
path can be approximated as constant, and no serious
error is introduced if we use (dc/dt ) 0).19 The simple theory
for the thin tube method involves the usual assumptions
of constant pressure and temperature, constant D, onedimensional diffusion, axial symmetry, gas (air) insolubility in liquid, and the absence of convection effects. When
quasi-steady-state conditions are reached, the vapor
concentration gradients between the liquid level and the
tube outlet are assumed to be constant. The net flux in
the tube, which is nonzero, consists only of vapor and the
gas (air) in the tube is stagnant with zero flux. The thin
tube evaporation method has been extensively used by
other investigators to find the D values for liquid vaporgas mixtures.17,19,20
For a diffusive evaporation in still air, Ficks first law
states that17-20

Diffusion is a process, which leads to an equalization

of concentrations within a single phase. The laws of
diffusion connect the rate of flow of the diffusing substance
with the concentration gradient responsible of this flow.
When steady state is established, the concentration
gradient in one-dimensional diffusion (dc/dx) is constant
and there is no change of concentration with time (dc/dt
) 0) and this condition will lead a linear distribution of
concentration for linear flow. In a nonuniform mixture,
the diffusion coefficient, D, is a proportionality constant
between the molecular flux and the composition gradient
of a species. In gaseous diffusion, the molecular collisions
are very important because the individual molecules
themselves carry mass through the gas and the molecular
collisions control the overall transport rate. The forces of
interaction between the molecules in turn control the
collisions, and without collisions the mass transport rate
should be 104 times that the actual diffusion rate.19,20
Diffusion Coefficient Determination by the Thin
Tube Evaporation Method. Stefan was the first to
develop this method in 1873.18 In this method, the
evaporation rate of a liquid (toluene for our case), which
partially fills the lower part of a vertical tube whose other
end is open, is controlled by diffusion through the stagnant
gas (air for our case), which fills the rest of the tube. The
liquid vapor diffuses from the lower part to the top of the

where m is diffused mass (kg), t is time (seconds), A is

cross-sectional area (m2), D is diffusion coefficient (m2/s),
c is vapor concentration (kg/m3), x is distance (m), and
c/x is the concentration gradient. For a liquid placed in
a thin cylindrical tube, the evaporating liquid mass is
given as, (m ) FAhL), where F is the liquid density (kg/m3)
and hL is the height of the liquid column (m) between the
liquid level and the tube bottom. hL is different from hair,
which is the height of the air column in the tube. Both
heights can be calculated by subtracting one of them from
the total tube height, hT. Since hT is a constant, its
derivative with time dhT/dt ) 0, and this permits use of
the difference of hL and hair by time (decrease of hL and
increase of hair) interchangeably during the integration.
(In practice, if one fills the thin tube with the liquid up
to the open top, then it is possible to use hL ) hair
directly, but this will give somewhat erroneous diffusion
coefficient results before the establishment of the quasistationary state.) Then, the only variable during thin tube
evaporation is h, and eq 1 can be written as

(16) Perry, R. H.; Chilton, C. H., Eds., Chemical Engineers Handbook,

5th ed.; McGraw-Hill: New York, 1973.
(17) Marrero, T. R.; Mason, E. A. J. Phys. Chem. Ref. Data 1972, 1,
3.
(18) Stefan, J. Sitzber. Akad. Wiss. Wien. 1873, 68, 385.
(19) Jost, W. Diffusion in Solids, Liquids, Gases, 3rd Printing;
Academic Press Inc.: New York, 1960.
(20) Cussler, E. L. Diffusion Mass Transfer in Fluid Systems, 2nd
ed.; Cambridge University Press: Cambridge, 1999.

During this quasi-stationary state evaporation, toluene

vapor diffuses with constant velocity through the air
column of height, hair. The vapor concentration just above
the toluene-air interface is assumed to correspond to the
equilibrium toluene vapor concentration, cs, which is
determined by the vapor pressure at the liquid surface
temperature. Toluene vapor concentration decreases with

c
m
) -AD
t
x

D c
h
)t
F x

(1)

(2)

Diffusion Coefficient of Toluene

Langmuir, Vol. 18, No. 13, 2002 5115

the increase of the air column in the thin tube, and the
vapor is carried away by slight convection currents at the
upper open end, always present in the free atmosphere.
Then the toluene vapor concentration is assumed to be
zero at the outlet of the tube (co ) 0). The concentration
gradient is assumed to be constant throughout the whole
air column height in the tube and is given by

cs
cs - c o
c
=
=x
0-h
h

(3)

By combining eqs 2 and 3, one obtains

h D cs
)
t
F h

(4)

If the vapor and the gas are assumed to behave ideally,

then the concentrations can be expressed in terms of
partial pressures

cs ) PvsMw/RT

(5)

where Pvs is the saturated vapor pressure (N/m2) at the

measurement temperature, Mw is the molecular weight
(kg/mol), R is the gas constant (8.3143 J/(mol K)), and T
is the temperature (K). However, the column above the
liquid level inside the tube is filled with both air and
toluene vapor molecules and the vapor flux equation (eq
1) contains a term for air molecules moving to the opposite
direction too. Consequently, a correction term for the vapor
pressure from Stefans general diffusion theory should be
included in eq 5 if the vapor pressure of the liquid is not
very small17-20

Pv ) PT ln

PT
PT - Pvs

(6)

where PT is the total pressure of the ambient atmosphere,

Pv is the actual vapor pressure, and Pvs is the saturation
vapor pressure at the liquid surface temperature. By
combining eqs 4-6, one obtains

PT
h D MwPT
K
)
ln
)
t
F hRT
PT - Pvs
h

(7)

where

) (1 - cos )2(2 + cos ) ) 2 - 3 cos + cos3 (11)

The height of the spherical cap above the supporting solid
surface is related to the two radii and the contact angle
by

hd ) Rs(1 - cos )

Thus, a spherical cap-shaped drop can be characterized

by using only two of the above four parameters. To take
into account the restriction of vaporization space by the
horizontal solid surface by using the spherical cap
geometry, eq 1 is rearranged so that1,15

( )

4RsD
dVc
(cs - c) f()
)
dt
F

(8)

is a constant. By integrating eq 7, between the height

limits hi and h for time limits of 0 and t

h2 ) hi2 + 2Kt

(9)

is obtained. Then it is possible to calculate K and D from

the slope of an h2-t plot.
Spherical Cap Drops. The influence of gravity becomes negligible when a liquid drop is sufficiently small
and the drop forms a spherical cap shape. A spherical cap
shape can be characterized by four different parameters,
the drop height (hd), the contact radius (rb), the radius of
the sphere forming the spherical cap (Rs), and the contact
angle ().4-15 By geometry the relationships between the
two radii, the contact angle, and the volume of the spherical
cap (Vc) at any instant in time are

rb ) Rs sin

and

Rs ) (3Vc/)1/3

(10)

(13)

where cs is the concentration of vapor at the sphere surface

(at Rs distance) (g/cm3), c is the concentration of the vapor
at infinite distance which is assumed to be zero, and f()
is a function of contact angle of the spherical cap. There
are three different versions for the solutions of the above
general eq 13 in the literature.1,4-6,15 Using the analogy
between the diffusive flux and electrostatic potential,
Picknett and Bexon1 derived the exact solution by using
power series. Rowan, Newton, and McHale4 derived an
approximate but closed-form solution which gives good
results around ) 90, and Shanahan and Bourges5,6
also derived an approximate but closed-form solution. In
a previous publication, we showed that when applied to
sessile drop evaporation experiments, the Picknett and
Bexon model, which was derived from the diffusion of the
exact spherical shape analysis, gave the best fit for all the
drop evaporation results.15 Thus, this model was chosen
to be applied for this work too.
Picknett and Bexon1 converted the problem of determining the evaporation rate of a sessile drop to a problem
of evaluating the capacitance of an isolated conducting
body of the same size and shape as the drop as a equiconvex
lens. By symmetry, the evaporation rate of this equiconvex
lens is twice that of the sessile drop and the above eq 13
was rearranged as

where

DMwPT
PT
ln
K)
FRT
PT - Pvs

hd ) rb tan(/2) (12)

and

( )

( )

4RsDcs C
dVc
)
dt
2F
Rs

(14)

where C is the capacitance of the equiconvex lens. The

capacitance of an equiconvex lens can be obtained by using
stream functions, but it was proved more convenient to
apply Snows series solutions to relate contact angle to
capacitance by an empirical polynomial and Picknett and
Bexon1 expressed the resulting values so that: for 0.175
e e radians (10 e < 180)

C/Rs ) 0.00008957 + 0.6333 + 0.1162 0.088783 + 0.010334 (15)

Picknett and Bexon1 indicated that the differences between
the values obtained from finite series and those from
Snows formula are in all cases less than 0.2% and are
generally less than 0.1%. By combining eqs 13 and 14 one
obtains

f() )

( )

1 C
2 Rs

(16)

5116

Langmuir, Vol. 18, No. 13, 2002

Erbil and Avci

To perform the integration, the general eq 13 is written

in terms of Vc and

( )

( )

dVc
4D 3Vc
)
dt
F

1/3

(cs - c)f() ) KVc1/3f() (17)

where K ) (42/331/3D(cs - c))/F1/3 and is a constant

independent of time and volume. Equation 17 can be
readily integrated between Vci (initial) when t f 0 and Vc
when t f t

Vc2/3 ) Vci2/3 -

2
f()Kt
3

(18)

Then it is possible to calculate [Kf()] and D from the

slope of the Vc2/3-t plot.15
Time Dependence of rb and h. For constant contact
angle evaporation mode, McHale et al.10 derived the time
dependence of contact radius, rb, so that

4D(cs - c) sin2

rb2 ) rbi2 -

FL(1 - cos )(2 + cos )

(19)

Similarly, when the Picknett and Bexon approach is used

in the derivation, one obtains

8D(cs - c)(1 + cos )f()

rb2 ) rbi2 -

FL(1 - cos )(2 + cos )

(20)

It is also possible to derive the time dependence of drop

height, hd, for constant contact angle evaporation mode.
When the chain rule is applied

( ) ( )( )
Vc
t

Vc
hd

hd
t

(21)

By using eqs 10-12, one obtains the derivative of drop

volume with respect to drop height

( )
Vc
hd

hd2(2 + cos )

(1 - cos )

(22)

By combining eqs 16, 17, 21, and 22 one obtains

( )
hd
t

4D(cs - c)f()
Fhd(2 + cos )

Results and Discussion

(23)

Equation 23 can be readily integrated between hdi (initial)

when t f 0 and hd when t f t
2

hd )

hdi2

8D(cs - c)f()
t
F(2 + cos )

circular windows were opened on two opposite sides of the cell

to enable photography and the light entrance.21 Activated carbon
was regenerated before each experiment by keeping the filled
bags in an oven at 110 C for 3 days. The thin tube evaporation
experiments were conducted at constant temperatures by placing
the cell in a constant temperature oven or refrigerator. The
temperature inside the cell was measured by both digital and
glass thermometers fitted on the cover of the cell. Double-distilled
Merck grade toluene was used in the experiments. Two thin
glass tubes having a height of 12 cm and cross-sectional areas
of 0.249 and 0.252 cm2 were placed vertically in a small tube
holder in the cell, and the decrease of the liquid level by time was
monitored manually by using a length scale placed on the tubes
which was carefully prepared by a compass in advance.21 Since
the rate of evaporation is very slow, i.e., 2.1-9.6 mm/day
depending on the cell temperature, the precision of the experiments was found to be sufficient.
For the sessile drop evaporation experiments, a flat, transparent, and smooth Teflon sheet produced by DuPont (FEP) was
used as a substrate. The surface of the Teflon sheet was cleaned
of impurities and oily material by wiping with a piece of cloth
containing dilute soap solution, by washing with distilled water,
acetone, ethyl alcohol, and again with distilled water thoroughly,
and finally drying at 50 C overnight and was kept in a
desiccator.21 A microliter syringe was used to deposit the toluene
drop from a small hole in the Plexiglas cell lid onto the Teflon
substrate, which was previously placed inside the cell, and the
upper hole was closed before taking photographs. Drops were
illuminated with a side lamp, and placing a large water cup
between the lamp and the cell prevented heat transfer by
convection. Since the surface area of the toluene sessile drop was
very small, the heat transfer from a 60-W lamp to the drop by
radiation was in the order of 10-6 W according to the StefanBoltzmann law and thus can be neglected. A magnifying
photography set containing a Nikon FM-10 body equipped with
AF Micro-Nikkor 60 mm f/2.8D lens, Nikon PB-6 and PB-6E
extension bellows joined with each other having a total length
of 45 cm, and AR-3 shutter cable release was used. The
evaporation of the drops was photographed within 15 or 20 s
intervals in the closed cell in still air at the temperature, which
was precisely measured with a thermometer placed near the
sessile drop in the cell. The polymer surface was sufficiently
reflective to enable an optical profile of a droplet and its reflection
to be obtained. For calibration of actual lengths, a metal wire
having a precisely determined diameter of 0.80183 mm was also
photographed after each experiment. The height, contact radius,
and the contact angle of the toluene sessile drops were measured
on the photographs by compass and transformed to their real
dimensions by means of the calibration wire.21

(24)

In theory, it is possible to calculate the diffusion coefficient

values from the slopes of rb2-t or hd2-t plots by using eqs
20 and 24; however it is more convenient to use the
linearity in these plots as an indication of the constant
contact angle evaporation mode and use only eq 18 and
Vc2/3-t plot for D calculation due to the possible errors in
one drop dimension measurement.
Experimental Section
A cube-shaped glass cell having dimensions 25 25 25 cm
with a Plexiglas cover was used in all the experiments. The edges
of the cell were isolated with silicone mastics. All the six faces
of the cell were covered with flat polyester fabric bags having 1.5
cm thickness, containing Merck grade activated carbon. Two

Temperature dependence of the liquid toluene density

in kg/m3 was given as16

F ) 882.86 - 0.85T(C)

(25)

and temperature dependence of the saturated toluene

vapor pressure in mmHg was given as16

ln Pvs ) 19.275 -

4748.4
T(K)

(26)

All the liquid densities at both the thin tube and sessile
drop experiment temperatures were calculated by using
eq 25 and vapor pressures by using eq 26 and are given
in Tables 1 and 2. Then, Stefans correction was applied
and the actual vapor pressures, Pv, were calculated by
using eq 6 and are given in Tables 1 and 2.
Thin Tube Evaporation Experiments. All the thin
tube experiments were performed by monitoring the
evaporation of toluene in two thin glass tubes simulta(21) Avc, Y. Msc. Thesis, Science Institute, Chemistry Department,
Kocaeli University, 2001.

Diffusion Coefficient of Toluene

Langmuir, Vol. 18, No. 13, 2002 5117

Table 1. Thin Tube Evaporation Experiment Results

liquid
sat. vapor
vapor
Tair density
pressure
pressure
3
(C) F (kg/m ) Pvs (mmHg) Pv (Pa)
1
7
16
26
33

882.0
876.9
869.3
860.8
854.8

7.06
10.23
17.34
30.02
43.17

946
1374
2339
4085
5925

KExp.

(m2/s)

2.65 10-10
4.15 10-10
7.50 10-10
1.425 10-9
2.20 10-9

diffusion
coefficient in
air, D (m2/s)
6.11 10-6
6.70 10-6
7.27 10-6
8.11 10-6
8.76 10-6

Table 2. Sessile Drop Evaporation Experiment Results

Tair
(C)

liquid
density
F (kg/m3)

sat. vapor
pressure
Pvs (mmHg)

vapor
pressure
Pv (Pa)

[K.f()]Exp.
(m2/s)

diffusion
coefficient in
air, D (m2/s)

17.5
22.5
26.5
32
33

868.0
863.7
860.3
855.7
854.8

18.88
24.89
30.84
41.03
43.17

2549
3373
4197
5623
5925

4.32 10-9
5.78 10-9
7.80 10-9
1.08 10-8
1.16 10-8

7.17 10-6
7.33 10-6
8.03 10-6
8.40 10-6
8.55 10-6

Figure 2. Time dependence of the square of the drop contact

radius, rb2, and the drop height, h2, at 33 C in the sessile drop
evaporation experiment.

Figure 1. Time dependence of the square of the air column

height, h2, at 16 C in the thin tube evaporation experiment.

neously in the same cell, and both data are plotted in the
same graph. An indicative h2-t plot for the thin tube
evaporation experiment at 16 C is given in Figure 1. The
regression coefficients obtained for each temperature are
generally high, R2 ) 0.9890-0.9994. K constants are
calculated from the slope of these linear plots for every
experiment temperature by using eq 9 and are given in
Table 1. Then toluene-in-air diffusion coefficients obtained
from thin tube experiments, D, are calculated by using eq
8 and are also given in Table 1. The change of the toluene
diffusion coefficient by the increase of the temperature
(in C) is given in Figure 4. The linear dependence relation
was found as

D(m2/s) ) (8.075 10-8)T(C) + 6.051 10-6

(27)

with a regression coefficient of R2 ) 0.9963. For various

diffusing liquid vapors, the linear dependence of D to T
is not surprising and has been reported in the literature
previously.17,22 At 0 C eq 27 results in D0 ) 6.05 10-6
m2/s. To determine the power dependence of diffusion
(22) OConnell, J. P.; Gillespie, M. D.; Krostek, W. D.; Prausnitz, J.
M. J. Phys. Chem. 1969, 73, 2000.

Figure 3. Time dependence of the two-thirds power of the

sessile drop volume, Vc2/3, at 33 C.

coefficient with temperature, an ln D vs ln T (in K) plot

is drawn and is given in Figure 5. The power dependence
relation in this temperature range was found as

ln D ) 3.174 ln T - 29.814

(28)

with a regression coefficient of R2 ) 0.9964, which

corresponds to a power factor of n ) 3.174 in the wellknown (D1/D2) ) (T1/T2)n equation where usually the power
factor, n )1.75-2.00 was used.16,17 At 0 C, eq 28 results

5118

Langmuir, Vol. 18, No. 13, 2002

Figure 4. The change of the diffusion coefficient by temperature for both thin tube and sessile drop experiments.

Erbil and Avci

D25 ) 9.08 10-6 and D45 ) 9.70 10-6 m2/s in 1960.26

Lugg27 reported a value of D25 ) 8.49 10-6 m2/s in 1968
and compared with theoretical estimates of D25 ) 7.68
10-6 of Chen and Othmer28 and D25 ) 7.83 10-6 of Wilke
and Lee.29 Langstroth, Diehl, and Winhold30 reported a
value of D20 ) 7.09 10-6 in 1950 in their work of drop
evaporation in still air. It seems that both the values
reported by Fairbanks and Wilke and Altshuller and
Cohen were unreasonably high, and the mean of the
remaining three experimental values is D25 ) 8.49 10-6
m2/s, which is 5.3% larger than our D25 ) 8.06 10-6 m2/s
value. However, the mean value of the theoretical
estimates of Chen and Othmer and Wilke and Lee is 7.75
10-6 m2/s, and when incorporated with the experimental
values, it gives an arithmetic mean of D25 ) 8.12 10-6
m2/s, which is only 0.7% larger than our experimental
value. On the other hand, the discussion of the temperature dependence of toluene diffusion coefficient is given
in very few publications.24-27,31 Fairbanks and Wilke25 and
Lugg27 directly used the temperature coefficient n ) 2 in
the general (D1/D2) ) (T1/T2)n equation. Gilliland24 reported
three D values for three different temperatures, which
resulted in n ) 1.82, and his values also give a straight
line when plotted as D vs T (in C). Altshuller and Cohen
examined the temperature dependence of diffusion coefficient in detail; unfortunately they only determined n )
1.95-2.00 for benzene and did not examine n for toluene.26
The closest n figure to our n ) 3.174 was reported as n
) 3 by Yuan and Cheng for a range of 37-70 C in 1967.31
In general, the temperature dependence of diffusion
coefficient was derived theoretically by Maxwell32 from
the gas kinetic theory so that

D)

1
1
+
M1 M2

AT3/2

1/2

(29)

122PT

where A is a constant, T is absolute temperature, M1 and

M2 are the mole masses of two gases, and 12 is the distance
between the centers of the two unlike molecules at collision.
Kinetic theory postulated that the molecular transport
by diffusion is entirely due to the molecular collisions,
which are controlled by interaction forces between the
molecules. Sutherland33 introduced a new C constant to
the Maxwell equation, depending on the nature of the
molecules, which can be obtained from pure gas viscosity
data
Figure 5. ln D versus ln T plot to determine the power
dependence of diffusion coefficient with temperature for both
thin tube and sessile drop experiments.

in D0 ) 6.10 10-6 m2/s. The arithmetic mean of both D0

values of toluene vapor in air at 0 C is 6.08 10-6 m2/s,
which is 20% lower than the reported value of 7.60 10-6
m2/s in Perrys Handbook. At 30 C, we found, D30 ) 8.48
10-6 m2/s by both methods, which is only 3.6% lower
than the Perry value of D30 ) 8.80 10-6 m2/s.16 D values
for toluene vapor in air reported in the literature were
found to scatter very much: Mack reported a value of D25
) 8.44 10-6 m2/s in 1925.23 Gilliand reported D25.9 ) 8.6
10-6, D39.4 ) 9.2 10-6, and D59.0 ) 10.4 10-6 m2/s
in 1934.24 Fairbanks and Wilke reported a value of D27.7
) 9.80 10-6 m2/s in 1950.25 Altshuller and Cohen reported
(23) Mack, E. J. Am. Chem. Soc. 1925, 47, 2468.
(24) Gilliand, E. R. Ind. Eng. Chem. 1934, 26, 681.

1
1
+
M1 M2
C
122PT 1 +
T

BT3/2
D)

1/2

(30)

Application of kinetic theory using the Lennard-Jones

potential for intermolecular forces by Chapman and
(25) Fairbanks, D. F.; Wilke, C. R. Ind. Eng. Chem. 1950, 42, 471.
(26) Altshuller, A. P.; Cohen, I. R. Anal. Chem. 1960, 32, 802.
(27) Lugg, G. A. Anal. Chem. 1968, 40, 1072.
(28) Chen, N. H.; Othmer, D. F. J. Chem. Eng. Data 1962, 7, 37.
(29) Wilke, C. R.; Lee, C. Y. Ind. Eng. Chem. 1955, 47, 1253.
(30) Langstroth, G. O.; Diehl, C. H. H.; Winhold, E. J. Can. J. Res.
1950, 28A, 580.
(31) Yuan, H. C.; Cheng, M. H. J. Chin. Chem. Soc. 1967, 14, 1 (Chem.
Abstr. 1968, 68, 62880z).
(32) Maxwell, J. C. Philos. Mag. 1868, 35, 185.
(33) Sutherland, W. Philos. Mag. 1894, 38, 1.

Diffusion Coefficient of Toluene

Langmuir, Vol. 18, No. 13, 2002 5119

Cowling34 and Hirschfelder, Curtiss, and Bird35 leads to

D T3/2/(D)

(31)

where (D) is a weakly temperature-dependent collision

integral so that the net dependence of the molecular
diffusion coefficient is in the range of n ) 1.6-2.0.
However, there is still considerable uncertainty about
what n will be for a particular system especially for
polyatomic molecules.17,19,26 Theoretical models were
derived primarily for spherical, symmetric molecules doing
elastic collisions, and many approximations were done to
apply these models to nonspherical, unsymmetrical, and
polar molecules doing inelastic collisions giving an interchange of energy of rotational and translational kinetic
energy during collision.19 In this respect, it seems best to
determine the temperature dependence of diffusion coefficient experimentally in the working temperature range
and to avoid long extrapolations.
There are some other related factors that should also
be considered: The magnitude of the liquid surface
temperature lowering due to the evaporative cooling is
related to the rate of evaporation. Since, the rate of toluene
evaporation is very slow in thin tube experiments, the
magnitude of the error from this source is neglected in
our analysis. In some diffusion tube experiments a stream
of a carrier gas was passed over the open end of the vertical
tube in order to maintain zero vapor concentration at the
tube outlet.17,26,27,36 As the flow rate of the carrier gas is
increased, the turbulent behavior in the upper chamber
extends eddies down into the diffusion tube and a
correction needs to be made.17,36 Since we used an activated
charcoal lined cell, we did not pass any carrier gas and
also the high linearity of the h2-t plots is an indication
of the absence of any error from this source.
Drop Evaporation Experiments. All the toluene
sessile drop evaporation on Teflon substrate experiments
was carried out in the same cell used in thin tube
experiments within the constant room temperature range
of 17.5-33 C. The presence of the large body of the
magnification photography set did not allow placing the
cell in an oven or refrigerator. The constant contact angle
of the sessile drop was found to be 43 ( 1 for all the
experiments and did not change for most of the evaporation
process. This value is very close to the 43.8 contact angle
value for the toluene drop on Teflon, which was determined
precisely by a digital camera and image analysis system.15
The time dependence of the square of the drop contact
radius, rb2, and the drop height, h2, at 33 C are given
Figure 2 as an indicative plot. The regression coefficients
(34) Chapman, S.; Cowling, T. G. The Mathematical Theory of NonUniform Gases, 3rd Ed.; Cambridge University Press: New York, 1970.
(35) Hirschfelder, J. O.; Curtiss, C. F.; Bird, R. B. Molecular Theory
of Gases and Liquids; Wiley: New York, 1954.
(36) Lee, C. Y.; Wilke, C. R. Ind. Eng. Chem. 1954, 46, 2381.

obtained for each constant temperature are not so high

for these parameters, R2 ) 0.9610-0.9906 for the rb2-t
plot and R2 ) 0.9696-0.9945 for the h2-t plot. As another
indicative plot, the time dependence of the two-thirds
power of the sessile drop volume, Vc2/3, at 33 C in the
sessile drop evaporation experiment is given Figure 3.
The regression coefficients obtained for Vc2/3-t plots are
generally high, R2 ) 0.9893-0.9991 being better than
both for rb2-t and h2-t plots for every measurement. [K
f()]Exp. constants are calculated from the slope of these
linear plots for every experiment temperature by using
eq 18 and are given in Table 2. Then, D of the evaporating
toluene vapor is calculated from [K f()]Exp. values by using
eq 17 and are also given in Table 2. The change of the
toluene diffusion coefficient obtained from the drop
evaporation experiments by the increase of the temperature (in C) is given in Figure 4. In addition, ln D versus
ln T (in K) plot is also given in Figure 5. As seen in both
Figures 4 and 5, the data points obtained from the drop
evaporation experiments are close to the data points and
the line obtained from thin tube evaporation experiments.
It should be noted that f() ) 0.2532 was calculated for
) 43, and this shows the importance of the f() factor,
which physically represents the restriction of evaporation
space by the presence of the substrate surface and thus
decreases the rate of mass loss in drop evaporation. There
should be a large error in D values for the drop evaporation
method if we neglect the f() factor.
Conclusion
The toluene-in-air diffusion coefficient was simultaneously measured in the same cell whose walls were lined
with activated carbon by following the evaporation of
toluene both in thin glass tube and as a sessile drop on
Teflon substrate, and the results are compared. Photography was applied to monitor the sessile drop evaporation, which occurred with constant contact angle mode. A
reasonably good fit of both data was obtained when the
restriction of evaporation space by the presence of the
substrate surface [f () factor] was considered in drop
evaporation on a substrate. It was also shown that the
decrease of the square of the height of the sessile drop
should vary linearly with time for this mode of drop
evaporation and the experimental results confirmed this
fact. On the other hand, a 1.75-2.00 power dependence
of the diffusion coefficient to the absolute temperature
was reported in the existing literature; however, a 3.17
power dependence was experimentally determined in the
working temperature range of 1-33 C.
Acknowledgment. The authors gratefully acknowledge the support of Kocaeli University Research Grant
No. 1999-28 for funding this work.
LA011557A