Article

pubs.acs.org/JPCA

A Unifying Model for Adsorption and Nucleation of Vapors on Solid

Surfaces
Ari Laaksonen*,†,‡
†
Finnish Meteorological Institute, P.O. Box 503, 00101 Helsinki, Finland
‡
University of Eastern, Finland, Department of Applied Physics, P.O. Box 1627, 70211 Kuopio, Finland
*
S Supporting Information
Downloaded via PONTIFICIA UNIV CATOLICA DE CHILE on September 26, 2022 at 13:05:47 (UTC).

ABSTRACT: Vapor interaction with solid surfaces is traditionally described with

See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

adsorption isotherms in the undersaturated regime and with heterogeneous

nucleation theory in the supersaturated regime. A class of adsorption isotherms is
based on the idea of vapor molecule clustering around so-called active sites.
However, as the isotherms do not account for the surface curvature effects of the
clusters, they predict an infinitely thick adsorption layer at saturation and do not
recognize the existence of the supersaturated regime. The classical heterogeneous
nucleation theory also builds on the idea of cluster formation, but describes the
interactions between the surface and the cluster with a single parameter, the
contact angle, which provides limited information compared with adsorption
isotherms. Here, a new model of vapor adsorption on nonporous solid surfaces is
derived. The basic assumption is that adsorption proceeds via formation of
molecular clusters, modeled as liquid caps. The equilibrium of the individual
clusters with the vapor phase is described with the Frenkel−Halsey−Hill (FHH) adsorption theory modified with the Kelvin
equation that corrects for the curvature effect on vapor pressure. The new model extends the FHH adsorption isotherm to be
applicable both at submonolayer surface coverages and at supersaturated conditions. It shows good agreement with experimental
adsorption data from 12 different adsorbent−adsorbate systems. The model predictions are also compared against heterogeneous
nucleation data, and they show much better agreement than predictions of the classical heterogeneous nucleation theory.

1. INTRODUCTION
Adsorption and heterogeneous nucleation of vapors on solid
surfaces are phenomena encountered in many industrial and
natural processes including atmospheric aerosol and cloud
formation.1−4 Vapor−liquid nucleation is an ill-understood
phenomenon that starts the condensation of a supersaturated
vapor. The difficulty in the theoretical treatment of nucleation
stems from the fact that the critical clusters of the nucleating
vaporsi.e. clusters just stable enough to be able to have a
greater probability to grow further rather than evaporateare
in most cases too large to be described using first principles, yet
too small to be well characterized using macroscopic concepts.
Nevertheless, the classical nucleation theory, based on a
macroscopic liquid drop model, has been surprisingly successful
in predicting critical cluster sizes in homogeneous nucleation,5,6
although it in most cases predicts the temperature dependence
of the nucleation rate incorrectly.3,5,7 In heterogeneous
nucleation, whereby the critical clusters appear on solid or
liquid surfaces, the situation is worse. For example, the classical
heterogeneous nucleation theory8 (CHNT) overpredicts the
onset supersaturation of water vapor nucleation on flat
surfaces9,10 and on nanoparticles11 drastically.
In CHNT, the interaction between the underlying substrate
and the critical cluster is described with just one parameter, the
contact angle Θ. According to Young’s equation,1 the contact
angle is determined by the interfacial tensions between the
substrate and the vapor (γs), between the critical cluster
(pictured as a liquid droplet) and the vapor (γl), and between
the droplet and the substrate (γsl). However, the contact angle
does not provide as complete information about the molecular
interactions between the substrate and the liquid as e.g.
adsorption isotherms do.
Whereas the CHNT is based on an incomplete description of
the substrate−liquid interactions, traditional adsorption theo-
ries do not include information about the surface tension effects
of the adsorption layer.12 For this reason, multilayer adsorption
isotherms are formulated so that the thickness of the adsorbed
layer tends to infinity as the saturation ratio of the adsorbing
vapor approaches unity; i.e., they do not recognize the fact that
a certain supersaturation level may have to be exceeded before
unlimited condensation of the vapor can occur. The multilayer
adsorption isotherms have been classified into a number of
subtypes,1 some of which are based on the idea that adsorption
proceeds via clustering of the vapor molecules around so-called
active sites, which can be chemical or physical heterogeneities
of the surface. As pointed out by Ruckenstein and Bhakta,13 it is
unlikely that even in the case of completely homogeneous
surfaces the adsorbed molecules are distributed uniformly, but
Received: January 12, 2015
Revised: March 31, 2015
Published: April 1, 2015

© 2015 American Chemical Society 3736 DOI: 10.1021/acs.jpca.5b00325

J. Phys. Chem. A 2015, 119, 3736−3745
The Journal of Physical Chemistry A
that the competition between entropy and attractive
interactions between the molecules leads to clustering. It thus
seems clear that adsorption and heterogeneous nucleation, both
of which occur as a result of vapor molecule clustering on the
surface, can be described within a unified theoretical frame-
work. The goal of this paper is to present such a theory for
nonporous surfaces and to show numerically that the new
theory can be used both to describe the adsorption of vapors at
subsaturated conditions and to predict the onset super-
saturation of heterogeneous nucleation on flat substrates. An
incomplete version of the theory has been presented earlier in a
conference proceedings.14
2. THEORETICAL FORMALISM
2.1. Adsorption. The adsorption model builds on the
Frenkel−Halsey−Hill (FHH) isotherm15−17 relating the
saturation ratio of the vapor S and the number adsorbed
monolayers (or surface coverage) N = V/Vm (where V denotes
the volume of the adsorption layer and Vm the volume of a
monolayer) via the equation
−A
ln(S) = B
N (1)
where the parameter A describes the interaction between the
surface and the attached vapor molecules, and the parameter B
determines how the strength of the interaction decays with
distance. In the original FHH theory, which applies for van der
Waals fluid, B = 3. However, for more complex systems, A and
B must be seen as adjustable parameters.
In order to describe the equilibrium between the vapor and a
single cluster adsorbed on a solid substrate, the cluster is
modeled as a spherical liquid cap with a contact angle Θ. The
orthogonal projection area of the droplet on the adsorbent
surface is denoted by Ω, and the average surface coverage over
Ω by Nd = δ/(v/σ), where δ is the average distance between the
substrate and the surface of the droplet (or the surface of the
upper hemisphere of the droplet, in case Θ > 90°). σ is the
cross-sectional area occupied by an adsorbed molecule having a
volume v, and thus v/σ is the monolayer thickness. The
curvature of the droplet causes an increase in its vapor pressure
that can be calculated via the Kelvin equation. By combining
the FHH and Kelvin terms, an equilibrium condition for the
cluster can be written as
−A 2γv
ln(S) = B +
Nd kTR (2)
where γ is surface tension, k is the Boltzmann constant, T is
temperature, and R is the radius of the spherical liquid cap. The
relation between R and Nd is given by (see Appendix for
derivation)
3vNd
R= with
σf (Θ) (3)
(1 − cos Θ)2 (2 + cos Θ)
f (Θ) = , Θ ≤ 90°;
sin 2 Θ
f (Θ) = 2 − 3 cos(Θ), Θ > 90°
Note that although the relation Nd = Vd/Vmd (where Vd is
droplet volume and Vmd is the volume of a monolayer filling the
area Ω) holds when Θ ≤ 90°, it does not hold when Θ > 90°.
The reason is that when the droplet’s contact area with the
3737
Article
substrate is smaller than its projection area, δ is not given by
Vd/Ω. For example, when Θ = 180°, Vd/Ω = (4/3)R, whereas
the average distance between the substrate and the upper half
of the droplet surface is (5/3)R.
If the average distance between adsorbed droplets is s, the
macroscopically observable surface coverage N can be related to
that of the droplets, Nd, via (see Appendix for derivation)
⎡ β ⎤2
N = π ⎢ ⎥ Nd 3 , Θ ≤ 90°
⎣ε⎦ (4)
⎡ πv 3g (Θ) ⎤
N=⎢ ⎥Nd 3 , Θ > 90°
⎣ 3ε 3s 2 ⎦ (5)
where β = 3v sin(Θ)/s, ε = σf(Θ), and g(Θ) = 4 − (1 + cos
Θ)2(2 − cos Θ).
The adsorption isotherm for the complete surface can now
be written as
⎡ πβ 2 ⎤ B /3 2γ ⎡ πεβ 2 ⎤
1/3
ln(S) = −A⎢ 2 ⎥ + ⎢ ⎥ , Θ ≤ 90°
⎣ε N ⎦ 3kT ⎣ N ⎦ (6)
⎛ ε ⎞ B⎡ πg (Θ) ⎤
B /3
ln(S) = −A⎜ ⎟ ⎢ 2 ⎥
⎝ b ⎠ ⎣ 3s N ⎦
2γv ⎡ πg (Θ) ⎤
1/3
+ ⎢ ⎥ , Θ > 90°
3kT ⎣ 3s 2N ⎦ (7)
Many surfaces have physical (e.g., kinks) or chemical
heterogeneities which may act as preferential adsorption sites,
determining the value of s. From now on s will be referred to as
the distance between active sites.
As the saturation ratio S increases, the droplets grow in size,
and depending on the distance between active sites and the
other parameters of the model, they may become large enough
to fill the surface completely. At that point the clusters coalesce
to form a uniform liquid film on the surface. At higher
saturation ratios the adsorption equilibrium is then described
using the conventional FHH isotherm. In a so-called FHH plot
(see Figure 1) showing the logarithm of −ln S vs the logarithm
of N, the isotherm is a straight line with a slope of −1/B in the
uniform liquid film regime. However, at the limit of zero
contact angle, the slope changes to −3/B in the liquid drop
regime as is the case with the red curve of Figure 1. The change
of the slope by a factor of 3 can be understood by noting that a
liquid film can only grow in one dimension, whereas clusters
grow in three dimensions. As the contact angle is increased
sufficiently from zero, the Kelvin effect becomes stronger, and
the FHH plot shows a downward bending curve in the droplet
regime instead of a straight line, as is the case with the blue
curve of Figure 1.
At nonzero contact angles, the merging of the clusters to a
uniform film also causes a stepwise increase in N as the
curvature of the liquid surface vanishes, causing a sudden drop
in vapor pressure that is balanced by the increased film
thickness (see blue curve in Figure 1). Stepwise adsorption has
been observed in many systems, but it is usually related to
phase changes in the adsorbed film.1 Whether the stepping due
to coalescence can be actually detected experimentally is
unclear as nonuniform distribution of active sites may lead to
coalescence taking place over a range of saturation ratios,
possibly smearing out the step feature. In any case, the size of
DOI: 10.1021/acs.jpca.5b00325
J. Phys. Chem. A 2015, 119, 3736−3745
The Journal of Physical Chemistry A
Figure 1. FHH plots for two systems having water properties, and
adsorption model parameters shown in the inset of the plot. The
distance between the active sites was taken to be 277 nm for the red
curve and 1.9 nm for the blue curve. The parameters for the blue curve
were chosen so as to show clearly both the step feature and the droplet
regime curvature, and they do not necessarily correspond to any real
system.
the step increases with increasing contact angle. An improved
model could possibly be developed that accounts for clusters
merging to larger clusters prior to the liquid film formation, but
this is out of the scope of the present paper.
In contrast to eq 1 (and other traditional multilayer
isotherms) that predicts infinite adsorption layer thickness at
saturation, eqs 6−7 predict finite N at saturation (provided that
the coalescence transition does not take place at S < 1). The S
vs N curve (see Figure 2) passes to the supersaturated region
and exhibits a maximum value before approaching S = 1 as the
adsorption layer thickness tends to infinity. As discussed below,
the maximum is related to the onset of heterogeneous
nucleation. The height of the maximum increases with the
contact angle. If the contact angle approaches 180°, the
maximum should get close to the onset supersaturation of
Figure 2. A schematic showing a traditional FHH isotherm and one
calculated using the new model.
3738
Article
homogeneous nucleation, which can vary from below 2 to close
to 100,18 depending on the substance and temperature.
2.2. Heterogeneous Nucleation. Heterogeneous nuclea-
tion occurs when the saturation ratio of the vapor is increased
past unity sufficiently so that the liquid caps start growing
spontaneously due to condensation. As with the Köhler theory
of cloud drop formation,19 this point of activation is denoted by
a maximum in a curve showing the saturation ratio vs droplet
size from eq 2 (see Figure 2). In order to be consistent with the
usual convention, the following equations were written in terms
of droplet radius rather than Nd. The maximum, denoted by *,
is given by
⎛ 2γvf (Θ)B ⎞1/(1 − B)
R =⎜
∗
B
⎟
⎝ kTα B ⎠ (8)
and
∗
⎛ α ⎞ B ⎛ 2γv ⎞
S = exp −⎜ ∗ ⎟ exp⎝
⎜ ⎟
⎝ f (Θ)R ⎠ kTR∗ ⎠ (9)
where the parameter α = 3A v/σ. Note that R* tends to
1/B
infinity if the value of B approaches unity, making B = 1 a
forbidden value. Furthermore, it can be shown that the widely
used BET-isotherm20 approaches the FHH-isotherm when A =
1, B = 1, and S → 1. Thus, the BET isotherm cannot be used in
place of the FHH isotherm in modeling heterogeneous
nucleation, as there will be no maximum in the S vs R curve.
When heterogeneous nucleation takes place on nanoparticles
with highly curved surfaces, the above equations need to be
modified somewhat. The nanoparticle curvature has to be taken
into account when calculating Vd using the equations given by
Fletcher,8 and the maximum of eq 2 needs to be located
numerically in order to find the size of the critical cluster and
the critical supersaturation. In the special case of zero contact
angle, eq 2 needs to be replaced by eq 4 of Sorjamaa and
Laaksonen4 describing uniform adsorption on a nanoparticle.
Note that another route to nucleation in the present model is
the droplet coalescence discussed above. If this occurs at
undersaturation, there will be no nucleation barrier to
condensational growth of the adsorbed film as saturation is
reached. It is also possible that coalescence triggers the growth
at a saturation ratio between unity and S*.
3. TESTING THE MODEL AGAINST EXPERIMENTAL
DATA
3.1. Comparison of Model Isotherms and Adsorption
Data. Figure 3 shows an FHH plot and two linear isotherm
plots of nitrogen adsorbing on nonporous alumina. The black
dots are experimental results,21 and the red curves were
produced with the new model. The procedure for obtaining the
model parameters was as follows. From the FHH plot of the
experimental data, a subset was selected that is clearly in the
multilayer regime, with linearly aligned data points. The FHH
parameters A and B were then obtained by linear least-squares
fitting to the subset. After that, the contact angle and the active
site distance were manually varied until a satisfactory fit to the
complete data set was obtained. Note that there usually is a
range of Θ- and s-values that produce such a satisfactory fit; as
the purpose of this paper is to demonstrate the new model,
only one pair of values was selected for each system. Note that
the numbers for Θ and s shown in Table 1 are not unique and
should therefore not be referred to as experimental values.
DOI: 10.1021/acs.jpca.5b00325
J. Phys. Chem. A 2015, 119, 3736−3745
The Journal of Physical Chemistry A Article

Figure 3. Nitrogen adsorbing on alumina at 77 K. The model parameters are listed in Table 1. Panels (a) and (b) show the FHH plot and linear
isotherm with model curves calculated using constant Θ and s, whereas in panel (c) the values of Θ and s were linearly increased and decreased,
respectively, between S = 0 and S = 0.5, as described in the main text.

Table 1. Systems Shown in Figures 2−5 and in the Supporting Information, and the Parameter Values Used in Generating the
Model Curves
system refa T (K) A B Θ (deg) s (nm) σ (Å2)b γ (N/m)
N2−Al2O3 21 77 3.32 2.88 5 38.0 16.7 0.009
H2O−SiO2 22 303 1.88 1.54 5 7.56 31.6c 0.071
H2O−glass 23 283 17.0 4.51 5 47.2 10.8 0.074
H2O−Al2O3 22 293 2.55 2.29 5 17.5 16.1 0.073
H2O−α-Al2O3 24 296 13.8 2.25 5 62.9 10.0 0.072
(CH2Br)2−glass 23 288 19.4 3.82 5 96.4 29.6 0.039
C2H5OH−glass 25 273 3.12 3.03 60 3.3 20.0 0.024
C6H6−MoS2 26 299 1.27 1.35 1 86.0 43.6d 0.028
C6H6−quartz 27 298 1.41 1.98 1 115.0 43.6d 0.028
C7H8−SiO2 28 298 3.33 2.44 10 15.0 55.2d 0.028
C8H18−SiO2 28 298 0.93 1.71 10 13.5 64.6d 0.021
H2O−PMMA 29 298 9.3 2.2 76 5.2 12.5d 0.072
a
ref = reference to experimental adsorption data. bThe values of the specific surface areas, were obtained from the experimental reference unless
indicated otherwise. cReference 30. dReference 31.

As can be seen from Figures 3a and b, the new model can be FHH plot 3a reveals that the data does not behave quite as
fitted very well to the whole nitrogen data set. However, the linearly in the liquid drop regime as does the model, and this
3739 DOI: 10.1021/acs.jpca.5b00325
J. Phys. Chem. A 2015, 119, 3736−3745
The Journal of Physical Chemistry A
difference is also reflected in the linear plot 3b. A possible
reason for the difference is that in reality the active sites have a
range of contact angles rather than just one fixed value.
Furthermore, it is plausible that the sites are filled progressively
starting from those with the lowest contact angles as the
saturation ratio is increased from zero. To test this possibility,
the model calculation was repeated allowing Θ to grow from 3°
to 9° and s to decrease from 53.5 to 13.8 nm as S was increased
from 0 to 0.5. The resulting linear isotherm is shown in Figure
3c; clearly, the match between the experiments and the model
is improved. However, for the sake of simplicity the single
contact angle model is utilized for the remainder of this paper.
Table 1 lists altogether 12 adsorption systems considered in
the present work. Besides the nitrogen adsorption on alumina,
three of the systems (water on glass, ethanol on glass, water on
plexiglass) are shown in Figures 4−6; FHH plots comparing
the data and model results for the remaining systems can be
found in the Supporting Information.
Figure 4. FHH plot of water adsorption on glass. Black dots:
experiments; red curve: model. The experimental details and model
parameters are listed in Table 1
Figure 4 shows experimental data and model results for water
adsorbing on glass. The model parameters shown in Table 1
were determined with the same procedure as described above.
As with many other systems shown in Table 1, a contact angle
value of Θ = 5° was sufficient to fit the model well. With water
on glass, the resulting active site distance is 47.2 nm, but that is
not necessarily a correct value, as the model could as well be
fitted using Θ = 10° and s = 30 nm. The model fit deteriorates
at contact angles larger than 10°, although literature values of
the macroscopic contact angle between glass and water vary
between about 10° and 40°.32−34 This may reflect the influence
of the microscopic contact angle (which refers to a change of
the slope of the liquid surface near to the contact line that can
only be seen at very small scales). The microscopic contact
angle is always smaller than the macroscopic one35 when Θ <
90°.
It should be noted that the value of the FHH parameter A
depends on the value of σ used when the experimental layer
thicknesses are converted from the primary measured quantity
to N. In some cases shown in Table 1, the conversion has been
done by the authors of the original publication using a σ-value
determined one way or another, and in other cases I have done
3740
Article
the conversion using a literature value of σ. For example, the A-
values determined for water adsorption on alumina differ by
quite much although the B-values are very close. It should be
noted the adsorbents differ somewhat, as one is polycrystalline
Al2O3 and the other is single-crystal α-Al2O3(0001), and that
the samples were treated differently before the measurements
were carried out. However, it is plausible that at least some
degree of the difference in the A’s stems from the different
approaches used in determining the water density at the
monolayer coverage (and thereby the σ).
Figure 5 shows the experimental and model results of
adsorption of ethanol on glass. In this case, a relatively large
Figure 5. FHH plot of ethanol vapor adsorbing on glass. Black dots:
experiments; red curve: model. The experimental details and model
parameters are listed in Table 1.
contact angle value of 60° improved the model fit clearly
compared to values close to zero. Interestingly, the data appear
to follow the step feature of the model curve; however, this may
be a coincidence created by somewhat noisy data.
In all of the systems shown in Figures 3−5 and in the
Supporting Information, the FHH curves created using the new
model are linear also in the liquid drop regime (with the
exception of the ethanol curve of Figure 5, which bends
slightly). Figure 6 shows the data for water adsorption on
plexiglass (polymethylmethacrylate) measured by Busscher and
co-workers.29 In order to fit the model to the data, the contact
angle value measured by Busscher et al., 76°, was used. The
other parameters (see Table 1) were then manually adjusted,
until a good fit was found. It appears that although the surface
coverage N reaches about 1.3 (assuming a molecular area of
12.5 Å2) at the highest saturation ratio of 0.97, the data are fully
in the liquid drop regime, i.e. below the coalescence transition.
Furthermore, because of the strong Kelvin effect, the model fit
is quite strongly curved.
3.2. Comparison of Model Predictions with Hetero-
geneous Nucleation Data. Figure 7 shows the data for water
vapor nucleation on different types of surfaces together with the
prediction of the classical heterogeneous nucleation theory8
and the new adsorption model (eq 9). The surfaces used by
Mahata and Alofs9 were differently prepared glasses and
plexiglasses, whereas Koutsky et al.36 used Teflon and various
crystalline surfaces. As has been found before, CHNT only
matches the data of Mahata and Alofs at the lowest contact
DOI: 10.1021/acs.jpca.5b00325
J. Phys. Chem. A 2015, 119, 3736−3745
The Journal of Physical Chemistry A
Figure 6. FHH plot of water adsorption on PMMA. Black dots:
experiments; red curve: model. The experimental details and model
parameters are listed in Table 1.
Figure 7. Experimental versus theoretical critical supersaturations as a
function of contact angle. See text for details.
angle values. With the data of Koutsky et al., there is somewhat
better agreement, but the data show a high degree of scatter.
Notably, the two data sets appear to be in complete
disagreement with each other, which was suspected by Mahata
and Alofs to result from flaws in the experimental design of
Koutsky et al. However, as shown in Figure 2, the new
adsorption model is able to at least partially explain the
experimental differences as being caused by varying adsorption
properties of the different surfaces used in the experiments. The
three model curves shown cover the range of the data points
between contact angle values of about 35° and 105° quite well,
with the exception of one data point measured by Koutsky et al.
The model predictions for the sizes of the critical clusters
corresponding to the data points of Mahata and Alofs between
Θ = 37.5° and 102.5° are shown in Figure 8 (squares). The
critical clusters are quite large with about 400 to 2000
molecules. In homogeneous nucleation experiments, critical
clusters are usually smaller than 100 molecules;5,7 however, the
critical supersturations are also much higher, so that the size
difference is not very surprising.
3741
Article
Figure 8. Minimum active size distances for nucleation to occur by
coalescence transition (dashed lines) and by critical cluster activation
(full lines), and numbers of molecules in critical clusters (squares) for
the same model systems as in Figure 7. See main text for details.
Figure 8 also indicates the active site distances at which the
new model predicts nucleation to occur via the droplet
coalescence transition between vapor saturation and the critical
supersaturation. For example, the model with α = 1.77 nm, B =
2 predicts that at a contact angle value of 80°, a multilayer
liquid film forms at subsaturation if s < 2.64 nm, nucleation
takes place via coalescence if 2.64 nm < s < 5.28 nm, and
nucleation takes place via activation of critical clusters at larger
values of s.
In order to examine whether the new model has actual
predictive power with regard to heterogeneous nucleation, the
adsorption parameters α and B should be determined
experimentally for the same vapor−substrate system as is
used in the nucleation experiment. The substrates used in the
nucleation experiments of Mahata and Alofs9 were uncoated
and coated (with oils and waxes) glasses and plexiglasses.
Literature data related to water vapor adsorption on uncoated
glass and plexiglass surfaces are available, and thus it is possible
to examine more closely whether the new model is able to
predict the critical supersaturations at Θ = 5° to 20° (glass) and
at Θ = 80° (plexiglass) shown in Figure 7.
Figure 9 shows the nucleation data together with model
predictions for glass. Mahata and Alofs9 used just one type of
glass but with different treatments to obtain the range of
contact angles. They heated the glass to 475 °C and observed
the lowest contact angle shortly after heating, the intermediate
angle after 9 h of exposure to room air, and the largest angle
after a prolonged exposure. As there are different types of
glasses, and as Mahata and Alofs did not provide details about
the type of glass they employed, I used adsorption parameters
for both pyrex and 3M glass obtained by fitting to the
experimental data sets of Jurinak et al.23 (see Figure 4 for the
3M glass data). As seen in Figure 9, using both the pyrex and
the 3M adsorption parameters in eq 9 results in predicted
critical supersaturations that are too low. Furthermore, using
adsorption parameters for water adsorption on glass obtained
from Busscher et al.29 results in even lower supersaturations,
very close to unity (not shown).
Figure 10 shows the experimental S* and model prediction
for plexiglass. The adsorption parameters for plexiglass were
obtained from Busscher et al.;29 see Figure 6. Mahata and
DOI: 10.1021/acs.jpca.5b00325
J. Phys. Chem. A 2015, 119, 3736−3745
The Journal of Physical Chemistry A
Figure 9. Critical supersaturation vs contact angle for heterogeneous
nucleation on glass substrates. Data are from Mahata and Alofs,9 and
the model parameters were obtained by fitting the FHH model to data
of water adsorption on pyrex (red squares) and 3M-glass (blue
squares) measured by Jurinak et al.23 The calculated critical droplet
sizes are between 27 000−1.7 million molecules for pyrex and 75 000−
13 million molecules for 3M-glass.
Figure 10. Critical supersaturation vs contact angle for heterogeneous
nucleation on a plexiglass substrate. Data are from Mahata and Alofs,9
the horizontal bar extending from the measured receding contact angle
(70°) to the advancing contact angle (90°). The model parameters
were obtained by fitting the FHH model to data of water adsorption
on plexiglass measured by Busscher et al.29 The calculated critical
droplet sizes are between 460−4900 molecules.
Alofs9 determined the advancing contact angle to be 90° and
the receding contact angle to be 70° (thus the data point is at
80° with the horizontal error bar between 70° and 90°),
whereas Busscher et al. measured a contact angle value of 76°.
Considering this uncertainty in the contact angle, the model
prediction is in excellent agreement with the experimental
critical supersaturation.
4. DISCUSSION
Already Rowley and Innes37 discussed in 1942 a “hypothetical
adsorption isotherm, not realizable experimentally” for a case
with nonzero contact angle that does not predict infinite
3742
Article
adsorption as vapor saturation is approached (similar to the red
curve in Figure 2). To my knowledge, two such isotherm
models have been presented since then. The first one is
Adamson’s potential distortion model,38,39 the purpose of
which was to deduce contact angles and spreading coefficients
from adsorption data. Curiously, Adamson and co-workers do
not mention heterogeneous nucleation in their papers. The
other model is the zeta-adsorption isotherm of Ward and co-
workers28,30,40−42 a four-parameter model which assumes
formation of molecular clusters, but does not explicitly invoke
the Kelvin term. In the paper by Zandavi and Ward,41 the
possibility of heterogeneous droplet formation in the super-
saturated regime is mentioned, but not considered further.
The application of the Kelvin equation in adsorption studies
is very common. However, it is almost always used for
calculating the depression of vapor pressure inside pores. An
exception is the work of Ruckenstein and Bhakta,13 in which
the increase of vapor pressure due to molecular clustering was
taken into account in the formulation of free-energy
expressions used to derive the adsorption isotherm. They
predicted the occurrence of a “jump” somewhat similar to the
stepping feature in Figure 1, with the exception that the jump
takes place at low occupation and leads to multilayer coverage.
As has been noted by Rowley and Innes37 and Adamson,38
the contact angle can be related to an integral of the adsorption
isotherm in the supersaturated regime via the spreading
pressure ΔI = γ(cos(Θ) − 1) as follows:
∞
ΔI = kt ∫PS
Γ d ln P
(10)
where the surface excess concentration Γ = N/σ and the ∞
denotes taking the integral to the limit where N approaches
infinity. It would of course be advantageous if eq 10 could be
used to reduce the number of free parameters of the adsorption
model; elimination of the contact angle would permit
unambiguous determination of the active site distance from
experimental isotherms. Unfortunately this is not possible, as
the Kelvin term makes the integral infinite.
As was noted above, the new isotherm, eqs 6−7 could be
improved especially at the low coverage regime by employing a
range of contact angles instead of a single value. The cost would
be an increased number of free parameters. Other possible
routes for improvement could be inclusion of the microscopic
contact angle and line tension effects. At least the latter could
have some effect on the nucleation predictions of the model
similarly to the case with CHNT.43
An interesting question relates to the quality of the
experimental nucleation data9 at low contact angles where the
model predicts clearly lower critical supersaturations. As seen
from Figure 4, a uniform film formed in the adsorption
experiments of Jurinak et al.23 at saturation ratios of around 0.5,
and in such a case there should exist no nucleation barrier at all
for unlimited condensation at saturation. Whatever the exact
value of the contact angle, a much lower active site density
would be needed for the coalescence transition to occur above
S = 1, let alone above S*. The nucleation experiments9 were
carried out using a diffusion cloud chamber. These devises have
been designed either to work at very low supersaturations,
relevant for cloud drop nucleation, or to work at very high
supersaturations, relevant for homogeneous nucleation. Prob-
lems have been detected with both types of instruments.44,45
For example, Snider et al.44 found, using well characterized test
aerosols, that supersaturations in a cloud condensation nucleus
DOI: 10.1021/acs.jpca.5b00325
J. Phys. Chem. A 2015, 119, 3736−3745
The Journal of Physical Chemistry A Article

chamber were 40% lower than values determined in a nucleation on plexiglass at a contact angle of ∼80° is in
conventional way. For these reasons, I am not completely excellent agreement with the experiment.
convinced about the reliability of the lowest critical super-
saturation(s) measured by Mahata and Alofs.
There is plenty of adsorption data available for many kinds of
■ APPENDIX
To derive eq 3, consider first Figure A1.
adsorbate−adsorbent systems that can be used for testing the
new model. One should be careful, however, to avoid testing
the model against systems with porous adsorbents (or at least,
to take the porosity into account when interpreting the results).
In contrast, reliable heterogeneous nucleation data is extremely
scarce for systems for which the adsorption parameters and
contact angles are available. At the moment, the best possibility
for further testing of the theory appears to be offered by
nucleation and cloud drop activation experiments on small
particles. Our FHH-activation theory4 of cloud droplets on
insoluble particles is a special case of the present model when
Fletcher’s curvature correction8 is taken into account, and the Figure A1. A spherical liquid cap with contact angle Θ ≤ 90° on a
contact angle is assumed to be zero. The FHH-activation substrate S.
theory gives a different functional dependence of critical
supersaturation on the dry particle size than the traditional The volume of the cap above the substrate S is
Köhler theory19 and can therefore be tested even if the
adsorption parameters were not independently available. πR3
V= (1 − cosΘ)2 (2 + cos Θ)
Indeed, Nenes and co-workers46−49 have found the FHH- 3
activation to be more consistent with experiments than the ⎛v⎞
traditional Köhler theory in the case of insoluble cloud = πa 2⎜ ⎟Nd
⎝σ ⎠
condensation nuclei. One further possible test case is
heterogeneous nucleation on nanoparticles,11,50 which will be ⎛v⎞
= πR2 sin 2 Θ⎜ ⎟Nd
a topic of future work. ⎝σ ⎠

5. CONCLUSIONS where the average distance between the substrate and the liquid
cap surface is (v/σ)Nd. Equation 3 for the case Θ ≤ 90° follows
It has been shown in this paper that combining the FHH directly. When Θ > 90°, the situation is as shown in Figure A2.
multilayer adsorption isotherm with the Kelvin equation yields
a new adsorption model that explicitly accounts for molecular
clustering and effectively extends the FHH theory to the
submonolayer regime. The new model predicts that the
coalescence transition, whereby the surface becomes filled
with clusters that merge to form a uniform liquid film, can be
detected from experimental adsorption data in an FHH-plot
showing the log of surface coverage vs loglog of vapor
saturation ratio. Specifically, the coalescence transition causes
the slope of the FHH curve to decrease by a factor of 3 when
the uniform liquid film forms (provided that the Kelvin term is
sufficiently small so that the submonolayer adsorption data Figure A2. A spherical liquid cap with contact angle Θ > 90° on a
behaves linearly). For systems with a strong Kelvin term, the substrate S.
new model predicts a step at the coalescence transition. The
model was shown to give an excellent description of
experimental adsorption isotherms for 12 different systems.
In case the coalescence transition does not take place at In this case we need to calculate the average distance
subsaturated conditions, the adsorption isotherm predicted by between the substrate and the surface of the upper hemisphere
the new model extends into the supersaturated regime. The of the droplet. The average height of the surface of the
model predicts an S vs R curve that shows a maximum at the hemisphere is given by (2/3)πR3/πR2 = (2/3)R, and the
so-called critical supersaturation where the clusters become
unstable, heterogeneous nucleation takes place, and the clusters hemisphere is lifted to height h = −R cos Θ above the substrate,
start growing by condensation. The growth can alternatively be so that (v/σ)Nd = R(2/3 − cos Θ), which gives eq 3 for Θ >
triggered by the coalescence transition, in case it takes place at a 90°.
supersaturation below the critical supersaturation. Comparisons To derive eqs 4 and 5 note that the total liquid volume must
between the model predictions and experimental nucleation remain constant when a droplet is spread over the whole
data show that the model is able to account for most of the surface. When Θ ≤ 90°, Ns2 = Ndπa2 = NdπR2 sin2 Θ, and eq 4
experimental features and explains apparent disagreements
between different nucleation experiments on flat surfaces. follows by substituting for R from eq 3.
Whereas the theoretical critical supersaturations for water When Θ > 90°, N(v/σ)s2 = (4/3)πR3 − (1/3)πR3[1 − cos(π
nucleating on low contact angle glass surfaces are somewhat − Θ)]2[2 + cos(π − Θ)], and eq 5 is obtained by again
lower than the experimental values, the prediction for water substituting for R from eq 3.
3743 DOI: 10.1021/acs.jpca.5b00325
J. Phys. Chem. A 2015, 119, 3736−3745
The Journal of Physical Chemistry A
■
*
ASSOCIATED CONTENT
S Supporting Information
Figures showing FHH plots of experimental data and model
results for adsorption in eight different systems that are listed in
Table 1. This material is available free of charge via the Internet
at http://pubs.acs.org.
■ AUTHOR INFORMATION
Corresponding Author
*E-mail: ari.laaksonen@fmi.fi.
Notes
The authors declare no competing financial interest.
■ ACKNOWLEDGMENTS
This work was supported by the Academy of Finland Center of
Excellence program (Grant No. 272041).
■
REFERENCES
(1) Adamson, A. W.; Gast, A. P. Physical Chemistry of Surfaces, 6th
ed.; John Wiley & Sons Inc.: 1997.
(2) Thomas, W. J.; Crittenden, B. J. Adsorption Technology and
Design; Butterworth-Heinemann: Oxford, 1998.
(3) Laaksonen, A.; Talanquer, V.; Oxtoby, D. W. Nucleation:
Measurements, Theory, and Atmospheric Applications. Annu. Rev.
Phys. Chem. 1995, 46, 489−524.
(4) Sorjamaa, R.; Laaksonen, A. The effect of H2O Adsorption on
Cloud-Drop Activation. Atmos. Chem. Phys. 2007, 7, 6175−6180.
(5) Strey, R.; Wagner, P. E.; Viisanen, Y. The Problem of Measuring
Homogeneous Nucleation Rate and the Molecular Contents of Nuclei:
Progress in the Form of Nucleation Pulse Measurements. J. Phys.
Chem. 1994, 98, 7748−7758.
(6) McGraw, R.; Laaksonen, A. Scaling Properties of the Critical
Nucleus in Classical and Molecular-Based Theories of Vapor−Liquid
Nucleation. Phys. Rev. Lett. 1996, 76, 2754−2757.
(7) Iland, K.; Wedekind, J.; Wölk, J.; Wagner, P. E.; Strey, R.
Homongeneous Nucleation Rates of 1-Pentanol. J. Chem. Phys. 2004,
121, 12259−12264.
(8) Fletcher, N. H. Size Effect in Heterogeneous Nucleation. J. Chem.
Phys. 1958, 29, 572−576.
(9) Mahata, P. C.; Alofs, D. J. Insoluble Condensation Nuclei: The
Effect of Contact Angle, Surface Roughness and Adsorption. J. Atmos.
Sci. 1975, 32, 116−122.
(10) Lee, T. D. Surface Characterization by Heterogeneous Nucleation
from the Vapor. 1998. Ph.D. thesis, Harvard University, 161 pp.
(11) Porstendörfer, J.; Scheibel, H. G.; Pohl, F. G.; Preining, O.;
Reischl, G.; Wagner, P. E. Heterogeneous Nucleation of Water Vapor
on Monodispersed Ag and NaCl Particles with Diameters Between 6
and 18 nm. Aerosol Sci. Technol. 1985, 4, 65−79.
(12) McMillan, W. G.; Teller, E. The Assumptions of the BET
Theory. J. Phys. Chem. 1951, 55, 17−20.
(13) Ruckenstein, E.; Bhakta, A. Clustering and Its Effects on
Adsorption. Langmuir 1994, 10, 2694−2698.
(14) Laaksonen, A. A Combined Theory of Heterogeneous
Nucleation and Adsorption of Vapors on Solid Surfaces. AIP Conf.
Proc. 2013, 1527, 270−273.
(15) Frenkel, J. Kinetic Theory of Liquids; Oxford University Press:
1946.
(16) Halsey, G. Physical Adsorption on Non-Uniform Surfaces. J.
Chem. Phys. 1948, 16, 931−945.
(17) Hill, T. L. Physical Adsorption and the Free Volume Model for
Liquids. J. Chem. Phys. 1949, 17, 590.
(18) Laaksonen, A.; Ford, I. J.; Kulmala, M. Revised Parametrization
of the Dillmann−Meier Theory of Homogeneous Nucleation. Phys.
Rev. E 1994, 49, 5517−5524.
(19) Köhler, H. The Nucleus in and the Growth of Hygroscopic
Droplets. Trans. Faraday Soc. 1936, 32, 1152−1161.
3744
Article
(20) Brunauer, S.; Emmett, P. H.; Teller, E. Adsorption of Gases in
Multimolecular Layers. J. Am. Chem. Soc. 1938, 60, 309−319.
(21) Č ejka, J.; Ž ilková, N.; Rathouský, J.; Zukal, A. Nitrogen
Adsorption Study of Organised Mesoporous Alumina. Phys. Chem.
Chem. Phys. 2001, 3, 5076−5081.
(22) Naono, H.; Hakuman, M. Analysis of Adsorption Isotherms of
Water Vapor for Nonporous and Porous Adsorbents. J. Colloid
Interface Sci. 1991, 145, 405−412.
(23) Jurinak, J. J.; Waldron, L. J.; Vomocil, J. A. Evidence of
Polymolecular Film Formation During Adsorption of Water and
Ethylene Dibromide on Glass Spheres. Soil Sci. Soc. Am. Proc. 1962, 26,
433−436.
(24) Al-Abadleh, H. A.; Grassian, V. H. FT-IR Study of Water
Adsorption on Alumina Surfaces. Langmuir 2003, 19, 341−347.
(25) Shereshefsky, J. L.; Russell, E. R. The Adsorption of Ethyl
Alcohol Vapor on Glass Spheres in Systems of Different Porosities. J.
Phys. Chem. 1953, 57, 660−665.
(26) Ballou, E. V.; Ross, S. The Adsorption of Benzene and Water
Vapors by Molybdenum Disulfide. J. Phys. Chem. 1953, 57, 653−657.
(27) Naono, H.; Hakuman, M.; Nakai, K. Determination of Pore Size
Distribution of Mesoporous and Macroporous Silicas by Means of
Benzene-Desorption Isotherms. J. Colloid Interface Sci. 1994, 165,
532−535.
(28) Zandavi, S. H.; Ward, C. A. Contact Angles and Surface
Properties of Nanoporous Materials. J. Colloid Interface Sci. 2013, 407,
255−264.
(29) Busscher, H. J.; Kip, G. A. M.; van Silfhout, A.; Arends, J.
Spreading Pressures of Water and n-Propanol on Polymer Surfaces. J.
Colloid Interface Sci. 1986, 114, 307−313.
(30) Ward, C. A.; Wu, J. Effect of Adsorption on the Surface
Tensions of Solid−Fluid Interfaces. J. Phys. Chem. B 2007, 111, 3685−
3694.
(31) McClellan, A. L.; Harnsberger, H. F. Cross-Sectional Areas of
Molecules Adsorbed on Solid Surfaces. J. Colloid Interface Sci. 1967, 23,
577−599.
(32) Eijkel, J. C. T.; Reemeijer, H. W.; Hermes, D. C.; Bomer, J. G.;
van der Berg, A. Strongly Accelerated and Humidity-Independent
Drying of Nanochannels Induced by Sharp Corners. Phys. Rev. Lett.
2005, 95, 256107.
(33) Birdi, K. S.; Vu, D. T.; Winter, A. A Study of the Evaporation
Rates of Small Water Drops Placed on a Solid Surface. J. Phys. Chem.
1989, 93, 3702−3703.
(34) Sumner, A. L.; Menke, E. J.; Dubowski, J.; Newberg, J. T.;
Penner, R. M.; Hemminger, J. C.; Wingen, L. M.; Brauers, T.;
Finlayson-Pitts, B. J. The Nature of Water on Surfaces of Laboratory
Systems and Implications for Heterogeneous Chemistry in the
Troposphere. Phys. Chem. Chem. Phys. 2004, 6, 604−613.
(35) Snoeijer, H. J.; Andreotti, B. A microscopic view on contact
angle selection. Phys. Fluids 2008, 20, 057101.
(36) Koutsky, J. A.; Walton, A. G.; Baer, E. Heterogeneous
Nucleation of Water Vapor on High and Low Energy Surfaces. Surf.
Sci. 1965, 3, 165−174.
(37) Rowley, H. H.; Innes, W. B. Relationships between the
Spreading Pressure, Adsorption, And Wetting. J. Phys. Chem. 1942, 46,
694−705.
(38) Adamson, A. W. an Adsorption Model for Contact Angle and
Spreading. J. Colloid Interface Sci. 1968, 27, 180−187.
(39) Tadros, M. E.; Hu, P.; Adamson, A. W. Adsorption and Contact
Angle Studies: I. Water on Smooth Carbon, Linear Polyethylene, And
Stearic Acid-Coated Copper. J. Colloid Interface Sci. 1974, 49, 184−
195.
(40) Wu, J.; Farouk, T.; Ward, C. A. Pressure Dependence of the
Contact Angle. J. Phys. Chem. B 2007, 111, 6189−6197.
(41) Ghasemi, H.; Ward, C. A. Surface Tension of Solids in the
Absence of Adsorption. J. Phys. Chem. B 2009, 113, 12632−12634.
(42) Zandavi, S. H.; Ward, C. A. Clusters in the Adsorbates of
Vapours and Gases: Zeta Isotherm Approach. Phys. Chem. Chem. Phys.
2014, 16, 10979−10989.
DOI: 10.1021/acs.jpca.5b00325
J. Phys. Chem. A 2015, 119, 3736−3745
The Journal of Physical Chemistry A Article

(43) Lazaridis, M. The Effects of Surface Diffusion and Line Tension

on the Mechanism of Heterogeneous Nucleation. J. Colloid Interface
Sci. 1993, 155, 386−391.
(44) Snider, J. R.; Petters, M.; Wechsler, P.; Liu, P. S. K.
Supersaturation in the Wyoming CCN Instrument. J. Atmos. Oceanic
Technol. 2006, 23, 1323−1339.
(45) Bertelsmann, A.; Heist, R. C. Two Dimensional Transport and
Wall Effects in the Thermal Diffusion Cloud Chamber. Part I: Analysis
and Operations Criteria. J. Chem. Phys. 1997, 106, 610−623.
(46) Kumar, P.; Nenes, A.; Sokolik, I. N. The Importance of
Adsorption for CCN Activity and Hygroscopic Properties of Mineral
Dust Aerosol. Geophys. Res. Lett. 2009, 36, DOI: 10.1029/
2009GL040827.
(47) Kumar, P.; Sokolik, I. N.; Nenes, A. Measurements of Cloud
Condensation Nuclei Activity and Droplet Activation Kinetics of Fresh
Unprocessed Regional Dust Samples and Minerals. Atmos. Chem. Phys.
2011, 11, 3527−3541.
(48) Kumar, P.; Sokolik, I. N.; Nenes, A. Measurements of Cloud
Condensation Nuclei Activity and Droplet Activation Kinetics of Wet
Processed Regional Dust Samples and Minerals. Atmos. Chem. Phys.
2011, 11, 8661−8676.
(49) Lathem, T. S.; Kumar, P.; Nenes, A.; Dufek, J.; Sokolik, I. N.;
Trail, M. T.; Russell, A. Hygroscopic Properties of Volcanic Ash.
Geophys. Res. Lett. 2011, 38, DOI: 10.1029/2011GL047298.
(50) Winkler, P.; Steiner, G.; Vrtala, A.; Vehkamäki, H.; Noppel, M.;
Lehtinen, K. E. J.; Reischl, G. P.; Wagner, P. E.; Kulmala, M.
Heterogeneous Nucleation Experiments Bridging the Scale from
Molecular Ion Clusters to Nanoparticles. Science 2008, 319, 1374−
1377.

3745 DOI: 10.1021/acs.jpca.5b00325

J. Phys. Chem. A 2015, 119, 3736−3745