You are on page 1of 14

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/261767201

Water Condensation: A Multiscale Phenomenon

Article in Journal of Nanoscience and Nanotechnology · February 2014


DOI: 10.1166/jnn.2014.9108 · Source: PubMed

CITATIONS READS

11 6,360

4 authors, including:

Rasmus Lund Jensen Leonid Gurevich


Aalborg University Aalborg University
119 PUBLICATIONS 1,694 CITATIONS 95 PUBLICATIONS 2,049 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Early Stage Decision Support for Sustainable Building Renovation View project

Holistic BIM-Based Sustainable Building Design and Performance Assessment View project

All content following this page was uploaded by Leonid Gurevich on 30 November 2016.

The user has requested enhancement of the downloaded file.


Review
Journal of
Copyright © 2014 American Scientific Publishers
Nanoscience and Nanotechnology
All rights reserved Vol. 14, 1859–1871, 2014
Printed in the United States of America www.aspbs.com/jnn

Water Condensation: A Multiscale Phenomenon


Kasper Risgaard Jensen1 , Peter Fojan1 , Rasmus Lund Jensen2 , and Leonid Gurevich1 ∗
1
Department of Physics and Nanotechnology, Aalborg University, Skjernvej 4A, 9220 Aalborg East, Denmark
2
Department of Civil Engineering, Aalborg University, Sohngaardsholmsvej 57, 9000 Aalborg, Denmark

The condensation of water is a phenomenon occurring in multiple situations in everyday life, e.g.,
when fog is formed or when dew forms on the grass or on windows. This means that this phe-
nomenon plays an important role within the different fields of science including meteorology, build-
ing physics, and chemistry. In this review we address condensation models and simulations with
the main focus on heterogeneous condensation of water. The condensation process is, at first,
described from a thermodynamic viewpoint where the nucleation step is described by the classical
nucleation theory. Further, we address the shortcomings of the thermodynamic theory in describing
the nucleation and emphasize the importance of nanoscale effects. This leads to the description of
condensation from a molecular viewpoint. Also presented is how the nucleation can be simulated
by use of molecular models, and how the condensation process is simulated on the macroscale
using computational fluid dynamics. Finally, examples of hybrid models combining molecular and
macroscale models for the simulation of condensation on a surface are presented.
Keywords: Water Condensation, Heterogeneous Nucleation, Hydrophilicity/Hydrophobicity,
Thermodynamics of Water Condensation, Water Condensation Modelling.

CONTENTS on which the phenomenon is relevant. That is, within each


1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1859 branch is an attempt to model the condensation as simple
2. The Macroscopic Approach—Thermodynamics as possible without omitting important details. In reality
of Condensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1859
2.1. Effect of the Liquid-Gas Interface—Homogeneous
the droplet growth during condensation is a multiscale pro-
Nucleation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1860 cess starting with assembly of just a few molecules and
2.2. Condensation on a Substrate—Heterogeneous ending with drops of macroscale dimensions.1
Nucleation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1861 In this review we present how condensation is described
2.3. Droplet Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1863
3. Nanoscopic Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1863
and modelled at different length scales ranging from
3.1. Nanoscopic Effects on the molecular to macroscale. Section 2 contains the macro-
Classical Nucleation Theory . . . . . . . . . . . . . . . . . . . . . . . 1863 scopic description of condensation, whereas nano- and
3.2. Nanoscopic Effects on the Contact Angle . . . . . . . . . . . . . 1863 microscopic effects on condensation are presented in
3.3. Adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1865
Section 3. In Section 4 it is described how condensa-
4. The Molecular Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1865
5. Simulation of Condensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1867 tion appears from an atomic viewpoint. Following these
5.1. Drop Growth Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1867 sections it is discussed in Section 5 how condensation is
5.2. Nucleation Models—Molecular Modelling . . . . . . . . . . . . . 1867 simulated using different approaches.
5.3. Heat and Mass Transfer Models . . . . . . . . . . . . . . . . . . . . 1869
5.4. Simulation of the Solid–Liquid Interface . . . . . . . . . . . . . . 1869
6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1870 2. THE MACROSCOPIC
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1870 APPROACH—THERMODYNAMICS OF
References and Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1870
CONDENSATION
1. INTRODUCTION From a thermodynamics point of view, the phase transi-
Every science has developed different ways of describing tion at condensation is expected to occur when the chem-
condensation, depending on the context and length scale ical potential of the vapor phase exceeds the chemical
potential of the liquid phase. For water it occurs when

Author to whom correspondence should be addressed. the relative humidity exceeds 100%. This condition can

J. Nanosci. Nanotechnol. 2014, Vol. 14, No. 2 1533-4880/2014/14/1859/013 doi:10.1166/jnn.2014.9108 1859


Water Condensation: A Multiscale Phenomenon Jensen et al.

be reached either when temperature is lowered below the and substrate effects, as is described in the following
dew point temperature or when vapor pressure is increased sections.
above the saturation vapor pressure, as can be seen in the
schematic phase diagram shown in Figure 1. In general 2.1. Effect of the Liquid-Gas
a phase transition depends on the physical properties of Interface—Homogeneous Nucleation
the matter in question, such as enthalpy and volume of The transition from gas to liquid phase requires the for-
transition of vaporization.2 This relatively simple approach mation of an interface between the two phases. When
for defining the phase transition is often used when heat droplets are formed, as is the case in dew formation, two
and mass transfer models are used for the prediction of terms add to the change of Gibbs energy, according to the
condensation.3–9 However, this simplification does not take classical nucleation theory (CNT), and thereby determine
into account that the phase transition is altered by surface whether the droplet formation takes place. One term is the

Kasper Risgaard Jensen is a Ph.D. student at the Department of Physics and Nanotech-
nology, Aalborg University, Denmark. He graduated with a master’s degree in nanotech-
nology from Aalborg University in 2009. His doctoral research is mainly focussed on
experimental investigation of dropwise condensation.

Peter Fojan got his Ph.D. in Biotechnology at the University of Technology, Graz, Austria
in 1997. He initially worked on industrial genetics of eukaryotic organisms. During his
PostDoc time at Aalborg University at the Department of Biotechnology, he moved into
the area of protein physics and molecular modelling. With the startup of Nanotechnology
at AAU he moved to the Department of Physics and Nanotechnology where he became
an Associate Professor in 2009. His research interests are centered around biological and
small molecules and their interactions with cells and surfaces in general, for medical,
sensor applications and as antibacterial agents.

Rasmus Lund Jensen got his Ph.D. in Civil Engineering from Aalborg University,
Denmark in 2005. He has been involved in research in air distribution and ventilation for
more than 10 years as well as in development of building thermal simulation tools. He is
the leader of the laboratory for Energy and Indoor Environment at Aalborg University and
has as such been heavily involved in the development and quality assurance of measure-
ment techniques and setups to measure energy use and indoor environment in the building
environment.

Leonid Gurevich got his Ph.D. in Physics at the Institute of Solid State Physics
(Chernogolovka, Russia) in 1994. He initially worked on high-Tc superconductors but
during his postdoc stay at Delft University of Technology became excited about nanotech-
nology and the possibility of charge transport through a single molecule. Since 2005 he has
been an Associate Professor at Aalborg University. His research interests are focused on
molecular electronics, biosensors, functional surfaces and nanofabrication. Surface modi-
fication and functionalization is one of his most essential tools interfusing all these fields.

1860 J. Nanosci. Nanotechnol. 14, 1859–1871, 2014


Jensen et al. Water Condensation: A Multiscale Phenomenon

wherefore the condensation is said to be initiated by


nucleation.10 11 That is, the water may be kept in the kinet-
ically stabilized metastable gas phase as the convex cur-
vature of the liquid surface imposes an energy barrier,
Gcr , for the vapor-to-liquid phase transition. Therefore
the gas phase should be supersaturated in order to initiate
condensation.12
The surface influence is also manifested in the Kelvin
equation (Eq. (3)). It states that the curvature of the droplet
surface gives rise to a difference in vapor pressure com-
pared to a flat surface, as:
pc = p e2Vm /rRT (3)
where pc is the vapor pressure in the presence of the
curved surface, p is the vapor pressure without a curved
Figure 1. Schematic phase diagram of water. Condensation is the phase
transition from gas to liquid phase. surface,  is the liquid–vapor surface tension, Vm is the
molar volume of the liquid phase, r is the radius of cur-
negative contribution from the volumetric Gibbs energy, vature, R is the gas constant, and T is the temperature.2
Gv , which favors the transition, whereas the other term From Eq. (3) it is observed that the smaller the radius of a
is an unfavorable positive contribution originated from the droplet the higher is the increase in vapor pressure caused
surface Gibbs energy, which is also called interfacial free by the surface curvature. It also appears from the equation
energy or surface tension, . The change in Gibbs energy that condensation may actually occur above the dew point
is shown in Eq. (1), where V is the liquid volume and S is temperature/below saturation vapor pressure if a concave
the liquid surface area, R is the drop radius, and  is the liquid surface is formed e.g., in the cavities on a rough
surface tension of a planar surface. surface.13

4R3 2.2. Condensation on a Substrate—Heterogeneous


G = −VGV + S = − GV + 4R2  (1)
3 Nucleation
This expression has a maximum at G/R = 0 where: Nucleation on a substrate is said to be heterogeneous,
whereas the nucleation by spontaneously formed drops
3
16 above the critical size is called homogenous.11 The sub-
Gcr = (2)
3GV 2 strate affects the nucleation process by lowering the energy
barrier of heterogeneous nucleation compared to that of
and the drop radius equals the critical radius Rcr = −2 / homogeneous nucleation.2 10
GV , see Figure 2. This means that droplets smaller than According to the CNT the energy barrier of condensa-
the critical size are unstable and will evaporate whereas tion with heterogeneous nucleation depends on the wetting
larger droplets will grow. Hence, a nucleus with a size properties of the substrate, where the wetting properties are
above the critical size is needed to facilitate condensation, characterized by the contact angle, , between substrate
and liquid, see Figure 3.10 11 14–16 When the contact angle
is 0 the solid surface is fully wetted by the liquid, which
forms a film, whereas the liquid is completely repelled by
the surface when the contact angle is 180 and the liquid
forms a whole sphere on the surface. If it is assumed that
a droplet forms a spherical cap on a plane solid substrate
the energy barrier of nucleation becomes:17
3
16
Gcr het = f  (4)
3GV 2
where the geometric factor, f  , is:
1
f  = (5)
2 − 3 cos + cos3 /4
From Eqs. (4) and (5) it can be seen that the energy barrier
Figure 2. Schematic change in Gibbs free energy at droplet formation for heterogeneous nucleation is always lower than the bar-
as a function of droplet radius. rier for homogeneous nucleation, and in fact this can be

J. Nanosci. Nanotechnol. 14, 1859–1871, 2014 1861


Water Condensation: A Multiscale Phenomenon Jensen et al.

Figure 3. Schematics of gas–solid (GS ), liquid–solid (LS ), and gas–


liquid () surface tensions at the contact line of a drop on a solid sub-
strate. The contact angle of the drop is .

generalized to the cases where the substrate is not planar.16


The nucleation process is thermally activated, and there-
fore the nucleation rate, J , exhibits the Arrhenius behavior:
 
Gcr
J = J0 exp − (6)
kB T
where J0 is a kinetic constant and kB is the Boltzmann
constant.10 This shows that the reduction of the nucleation
energy barrier of heterogeneous nucleation also entails a
higher nucleation rate. When the contact angle is 180 the
energy barriers of homogeneous and heterogeneous con-
densation are equal. The lower energy barrier and thereby Figure 4. Environmental scanning electron microscope (ESEM) images
higher nucleation rate on hydrophilic surfaces, compared of water condensation on surface with hydrophilic and hydrophobic paths
to hydrophobic surfaces, has been examined experimen- with a width of 25 m. The water contact angle on the hydrophilic
tally by Varanasi et al.18 On a surface with alternating regions is ∼ 25 , whereas it is ∼ 110 on the hydrophobic regions. In the
first image (a) the surface is dry whereafter the humidity is increased
hydrophilic and hydrophobic paths they observe prefer-
(b)–(h) until droplets are observed. The drops are observed to form
ential nucleation on the hydrophilic regions, as can be preferably on the hydrophilic paths. Reprinted with permission from [18],
observed in Figure 4. Similarly, Na and Webb19 found K. K. Varanasi et al., Appl. Phys. Lett. 95, 094101 (2009). © 2009,
that the supersaturation required for nucleation depends on American Institute of Physics.
the contact angle, as the surfaces with low contact angle
require lower levels of supersaturation for nucleation to introduced a roughness factor, r, given by the actual sur-
occur, as compared to the surfaces with high contact angle. face area divided by the geometrical plane surface area,
The question is then, what determines the contact angle of into the interfacial tension balance in Eq. (7):22
the liquid drop on the solid substrate.
When condensation takes place on a substrate, two more cos rough = r cos flat (8)
interfaces are created in addition to the liquid–gas inter-
face; namely the solid–liquid and the solid-gas interfaces. Wenzel’s modification is based on the assumption that liq-
By balancing forces along the contact line, as presented uid penetrates into the depressions of rough surfaces. From
in the Young equation in Eq. (7) and in Figure 3, it fol- the modification it appears that a hydrophilic substrate
lows that at thermodynamic equilibrium the contact angle will become more hydrophilic when roughened, whereas
on a smooth surface is determined by the gas–solid (GS ), a hydrophobic substrate will become more hydrophobic.
liquid–solid (LS ), and gas–liquid () surface tensions.20 This is for instance illustrated by experiments performed
by Klemm et al.11 23 by roughening of glass. According to
GS − LS their experiments, increased roughness of the hydrophilic
cos = (7)
 glass results in water vapor condensation taking place at
higher temperatures.23
That is, the contact angle of dew on a substrate is deter-
To describe the situation where the liquid does not pen-
mined by the surface energy of the substrate, which is
etrate the depressions of a rough hydrophobic surface and
directly related to the chemical composition of the sur-
air is trapped in the depressions the following equation
face. It can be observed from Eq. (7) that when GS < LS
derived by Cassie and Baxter for calculating the contact
the contact angle is > 90 , and in this case the surface
angle on chemically heterogeneous surfaces, het , can be
is designated as hydrophobic. Contrarily, the surface is
used.24
called hydrophilic when GS < LS and the contact angle
cos het = f1 cos 1 + f2 cos 2 (9)
is < 90 . The highest water contact angle attainable on a
smooth surface is reported to be around 120 .21 However, Here, f1 and f2 are the fractional areas of surface type 1
the contact angle is also affected by the roughness of the and 2, respectively, and 1 and 2 are the local contact
surface, as was initially recognized by Wenzel. He has angles on surface type 1 and 2, respectively. When Eq. (9)

1862 J. Nanosci. Nanotechnol. 14, 1859–1871, 2014


Jensen et al. Water Condensation: A Multiscale Phenomenon

is used to describe the situation where air is trapped in area at saturation varies depending on the liquid contact
the depressions under a drop on a rough surface, hence angle on the substrate. On an inclined surface the largest

2 = 180 , the equation reduces to: drops will eventually slide off due to gravity, whereby sur-
face is also released for growth of new droplets.1 30 In this
cos rough = f1 cos 1 − f2 (10) case, the condensation process is cyclic.
with f1 being the fractional area of fluid contact and f2 the
fractional area of air-liquid contact underneath the droplet, 3. NANOSCOPIC EFFECTS
both normalized by the geometrical plane surface area. As stated above, the initial nucleation step of condensation
It is now generally accepted that heterogeneous conden- is the limiting step in a phase transition. Therefore it is also
sation is initiated by nucleation. However, another model notable that even though the condensation process may
where the condensation is initiated from a thin film on a be considered from a macroscopic perspective the initial
substrate does also exist.1 This will, though, not be further step occurs on the nanoscale where bulk properties and
considered in this review. According to Merikanto et al.25 assumptions are not necessarily valid. In general, surface
the nucleation step is the limiting step in a phase transition. effects become increasingly important compared to volume
effects at decreasing droplet sizes.
2.3. Droplet Growth
When describing droplet growth patterns on a substrate 3.1. Nanoscopic Effects on the
the process is normally divided into three stages of which Classical Nucleation Theory
nucleation is the first one. In the second stage of the pro- The CNT, which is commonly used for the description of
cess the droplets grow primarily by diffusion of vapor from the initial step of phase transition, is reported to be inac-
the surroundings, and in the third stage the drop growth is curate at the nanoscale.25 31 It is reported by Merikanto
mainly caused by coalescence.10 14 26 et al.25 that the modelling of small clusters as bulk liq-
The growth of an isolated droplet is governed by diffu- uid, in accordance with the CNT, introduces an error to
sion. This mechanism can encompass both diffusion of sin- the calculated energy barrier of nucleation at cluster sizes
gle molecules from the gas phase and diffusion of clusters below 8 to 50 molecules, depending on temperature and
and adsorbed molecules along the substrate surface.14 26 molecule species. Likewise, Prestipino et al.31 and Ghosh
Experimental studies by Leach et al.27 show that smaller and Ghosh32 have introduced corrections to the CNT by
droplet growth primarily occurs by diffusion of adsorbed assuming a nonsharp interface. Thereby they have obtained
water along the substrate surface, whereas larger droplets a more accurate description of the surface tension and
grow mainly by diffusion of vapor from the gas phase. The liquid density dependences on cluster size, whereby the
diffusion is driven by the concentration gradient around energy barrier of nucleation is reduced compared to the
the droplet caused by the fact that the droplet acts as a CNT. This has also been demonstrated in molecular dynam-
mass sink, and in fact this may lower the vapor pressure ics (MD) simulation by Toxvaerd,33 which shows that the
sufficiently to prevent nucleation within a certain perime- solid–liquid and liquid–vapor interfaces cannot be sepa-
ter around already existing drops. In the diffusion regime rated for nuclei consisting of ∼ 100 molecules. Therefore
the droplet size distribution is approximately monodisperse the surface tensions on nanoscale become smaller than the
and the increase of the mean drop radius follows a power bulk surface tensions. Furthermore, the simulations show
law with an exponent of 1/3.14 26 that the vapor density near the surface is enhanced, which
When droplets have grown so large by diffusion that should be taken into account in the CNT. The use of molec-
their perimeters come into contact with each other the ular simulations may be considered as a way of avoiding
droplets coalesce. The coalescence typically starts when inaccuracies of the CNT. However, this is not necessarily
the drops cover more than 30% of the substrate, which providing more accurate results relative to the CNT when
means that the transition from the diffusion to the coales- compared to experimental results.16
cence growth stage occurs faster at lower contact angles.28
In contrary to the growth by diffusion, the coalescence 3.2. Nanoscopic Effects on the Contact Angle
by itself does not increase the mass of condensate on The Wenzel and Cassie-Baxter equations presented above
the surface. However, it releases the substrate surface that describe the contact angle of a fluid on a rough sur-
becomes available for growth of a new generation of face in the states where liquid, respectively, penetrates
droplets by nucleation and diffusion, and with time the or does not penetrate the depressions of the surface.
droplet size distribution on the surface becomes highly It has been suggested to use the Wenzel equation in
polydisperse. In this growth regime the ratio between the hydrophilic region and the Cassie-Baxter equation in
dry and wet surface areas saturates at the wet area frac- the hydrophobic region.34 However, according to other
tion of approximately 55%, and the growth of the mean authors it is not unambiguously described by macroscopic
drop radius follows a power law with an exponent of parameters which of the two states are occupied under
unity.10 14 29 However, according to Leach et al.27 the wet which conditions.35 The energetically most favorable of the

J. Nanosci. Nanotechnol. 14, 1859–1871, 2014 1863


Water Condensation: A Multiscale Phenomenon Jensen et al.

Wenzel and Cassie-Baxter states is the one with the lowest A consequence of the influence of line tension is that
apparent contact angle. However, this does not mean that the contact angle depends on the drop size, as appears
a drop will attain this state as the state initially attained from Eq. (11).45 46 49–51 Measurements by Cao et al.52
depends on how the drop is formed on the surface, and show that the contact angle is drastically reduced for
an activation energy barrier is present between the two micrometer or submicrometer sized droplets on hydropho-
states.36 37 The presence of an activation barrier between bic surfaces, which correlates well with results presented
the Wenzel and Cassie-Baxter states renders it possible by Barberis and Capurro50 showing that at the nanoscale
for the states to be metastable, and actually the Cassie- the contact angle increases with increasing droplet size.
Baxter to Wenzel transition is irreversible according to However, through theoretical models it has been shown
Nosonovsky and Bhushan.38 For instance, experiments by that the contact angle can both increase or decrease with
Cheng and Rodak39 and Xiao et al.40 have shown that some increasing droplet size depending on the line tension,47
rough surfaces, which are normally designated as superhy- surface heterogeneity,53 and surface roughness.54 In respect
drophobic, only carry this characteristic when water drops to CNT, the inclusion of line tension can theoretically
of macroscopic size are deposited on the surfaces. When make homogeneous nucleation more thermodynamically
instead water condensed on the surfaces the contact angles favorable than heterogeneous nucleation.47
were well below 90 . This has been explained by the pres-
ence of condensed water in the depressions of the rough 3.2.3. Contact Angle Hysteresis
surfaces, whereby the fractional area of air-liquid contact According to Tadmor,55 the line energy is a function of
beneath the droplet has been reduced. surface irregularities, which gives rise to contact angle hys-
teresis. That is, when a drop advances on a surface the
3.2.1. The Three-Phase Contact Line contact angle is higher than when a drop recedes on the
It has been argued by Gao and McCarthy41 that only rough- same surface despite the fact that the macroscopic interfa-
ness at the three-phase contact line of a drop on a surface cial energies are the same.56 Microscale roughness is one
is relevant when determining the liquid contact angle, and of the irregularities contributing to contact angle hysteresis
therefore it is incorrect when the surface roughness of the as the roughness locally disturbs the shape of the spheri-
liquid–solid contact area is used in the Wenzel and Cassie- cal cap anticipated in the Young equation and CNT. Due
Baxter equations. This is, however, only relevant on sur- to pinning of the three-phase contact line the microscale
faces where the surface roughness is non-uniform or where roughness makes a range of contact angles energetically
the characteristic dimension of the surface roughness is favorable, in addition to those described by the Wenzel
comparable to the drop size.42 43 In cases where either of and Cassie-Baxter equations.35 38 55 Furthermore, adhesion
the two applies, the local surface roughness at the three- hysteresis38 and microscale chemical heterogeneity55 con-
phase contact line should be used in the Wenzel and Cassie- tributes to the contact angle hysteresis. However, contact
Baxter equations to obtain the local contact angle.38 44 It angle hysteresis is also observed on smooth homogeneous
is noteworthy that the small droplets formed in the initial solid substrates. According to Starov12 this is caused by
stages of dew formation actually have a size comparable to the fact that at any contact angle between the receding
the characteristic dimension of the surface roughness. and the advancing ones the droplet is in slow “micro-
scopic” motion and therefore it is not in equilibrium, as
3.2.2. Line Tension required for the Young equation to be valid. When the
One of the parameters neglected when describing the con- contact angle is below or above the receding or advancing
tact angle by the Young equation, and thereby also when contact angles, respectively, the droplet motion becomes
describing the nucleation rate by the CNT, is the line ten- “macroscopic.”
sion acting along the three phase contact line of a droplet,
analogous to surface tension acting at the surface between 3.2.4. Nucleation Sites
two phases.45–49 That is, the line tension is a consequence
The surface heterogeneity and roughness give rise to local
of an imbalance of intermolecular forces in the junction
variations on a surface. In this respect the surface het-
of the three phases. The line tension typically amounts to
erogeneity may be caused by both internal effects, such
10−10 to 10−6 N and it has high influence on the contact
as local variations in chemical composition and rough-
angle of nano- and submicron droplets, whereas the drops
ness, and by external effects, such as adsorption of water
with macroscopic dimensions are not significantly affected
vapor and contaminants.19 57 58 This creates nucleation
by the line tension.45 46 The effect of the line tension, ,
sites where nucleation is particularly favorable, as has
on the contact angle has been introduced through the mod-
been shown in experiments by Mu et al.59 According
ified Young equation:45 49
to their experiments the nucleation site density is higher
cos = cos − k (11) on surfaces with higher fractal dimensions. Apart from
 the surface heterogeneity and topography, the nucleation
where  is the contact angle described by Eq. (7) and site density also depends on the minimum droplet radius,
k is the geodesic curvature of the three-phase contact line. which, in turn, depends on the surface subcooling.60 Even

1864 J. Nanosci. Nanotechnol. 14, 1859–1871, 2014


Jensen et al. Water Condensation: A Multiscale Phenomenon

though the nucleation rate is expected to be higher on


hydrophilic than on hydrophobic surfaces according to
CNT, Leach et al.27 have not observed significant differ-
ences in the nucleation densities between surfaces with
the contact angles of 60 and 100 . However, their mea-
surements may not be entirely reliable since the effects
of surface topography and heterogeneity have not been
controlled.
Figure 5. Schematics of the transition from macroscale droplet to
3.3. Adsorption nanoscale film. Reprinted with permission from [12], V. M. Starov, Adv.
Characterization of a substrate from a macroscopic point Colloid Interface Sci. 161, 139 (2010). © 2010, Elsevier.
of view, as discussed in Section 2,2, requires knowledge
of the surface tensions. The latter may vary locally due to
necessarily directly related to the molecular adsorption
chemical and topographical heterogeneity of the substrate,
of water on surfaces covered by different self-assembled
as described above. Besides that, another effect to consider
monolayers. However, according to Furuta et al.62 the con-
in this respect is adsorption of water on the substrate. The
tact angle of water on a hydrophobic surface decreases
adsorption of water molecules on a surface is controlled
with decreasing temperature due to an increase in water
by surface forces; such as Van der Waals forces, hydrogen
adsorption which renders the surface more hydrophilic.
bonding, electrostatic forces, and steric forces.12 That is,
The ordered structure of adsorbed liquid on a solid sur-
the water molecules are bound to the surface and there-
face also influences the interfacial energy transport effi-
fore this water does not possess the characteristics of bulk
ciency at the solid–liquid interface. At the interface there
liquid.
is a thermal resistance, known as the Kapitza resistance,
At ambient conditions the gas–solid surface tension is
always affected by adsorption of water molecules from the due to the mismatch between the properties of the solid
air, as and the liquid. MD simulations have revealed that the
GS = GS
0
1 −  + GSa
(12) Kapitza resistance depends on the wetting properties of
the solid, as the stronger adsorption and thereby also more
with being the fraction of solid surface covered by ordered liquid on wetting interfaces reduces the Kaptiza
0
adsorbed water molecules, GS is the surface tension resistance, whereas it is increased on non-wetting inter-
a
of the bare solid surface, and GS the surface ten- faces where the adsorbed liquid is less ordered at the inter-
sion of the surface covered entirely by adsorbed water.12 face. The simulations have been made both for a general
Furthermore, the fraction of solid surface covered by case of a Lennard-Jones fluid on a Lennard-Jones solid63
adsorbed water molecules depends on the vapor pressure, and for a specific case of water on different self-assembled
which also makes GS a function of the vapor pressure. monolayers.64 The simulations by Xue et al.63 also showed
On hydrophilic surfaces the adsorption of water lowers that the thermal conductivity in the structured liquid was
the gas–solid surface tension which results in adsorption indistinguishable from that of bulk liquid.
of a thin water film on the solid. This results in the The effects described in Section 3, all show that a sub-
formation of a smooth transition zone between a liquid strate cannot be characterized solely by the macroscopic
droplet and the adsorbed liquid film on such surface, as contact angle determined from surface tension and sur-
illustrated in Figure 5. The profile in the transition zone face roughness. First of all, the surface tension and surface
is governed by surface forces, resulting in the so-called roughness vary locally on the surface, and their effects on
disjoining or Derjaguin’s pressure.12 On the contrary, the contact angle depend on drop size. Secondly, several
adsorption of water on hydrophobic surfaces increases the other parameters affect the contact angle. To summarize
surface tension. Adsorption still occurs on hydrophobic the discussion of condensation theory presented until now
surfaces as long as the contact angle is lower than 180 , in this review, the condensation process is very complex
though, in general, to a lower extent than on hydrophilic and it involves phenomena on several length scales rang-
surfaces.12 46 61 Measurements by Cao et al.52 have shown ing from the atomic scale at nucleation to the macro scale
that nanometer-sized droplets of water are adsorbed on where drops will slide off an inclined surface.
hydrophobic surfaces at ambient conditions where the rel-
ative humidity is much lower than 100%, indicating that
the physics at the nanometer scale cannot be ignored when 4. THE MOLECULAR APPROACH
considering condensation. According to their measure- As opposed to the macroscopic approach presented above,
ments, adsorption of water on hydrophobic surfaces pri- the condensation process can also be described from
marily occurs on surface defects, which act as nucleation a molecular viewpoint, i.e., based on the interactions
centers. This is consistent with the measurements by James between individual molecules. Two methods are com-
et al.61 showing that the macroscopic contact angle is not monly used for the description; namely, MD or Monte

J. Nanosci. Nanotechnol. 14, 1859–1871, 2014 1865


Water Condensation: A Multiscale Phenomenon Jensen et al.

Carlo (MC) methods.65 In the MD calculation Newton’s


equations are solved to obtain the trajectories of individ-
ual molecules in the system. This requires the descrip-
tion of intermolecular forces, such as Van der Waals
forces, electrostatic forces, hydrogen bonding, and steric
interactions etc. by an intermolecular potential. From a
classical approach the intermolecular potential of unpolar-
ized molecules, such as argon, is often described by the
Lennard-Jones potential, as follows:16
 6  12
A B
VLJ = − + (13)
r r
The first term describes the attractive Van der Waals
forces, whereas the second term describes the Pauli repul-
sion originating from the Pauli exclusion principle of
electrons.66 However, this description is not sufficient for
molecules with an inherent dipole moment, such as water.
In these cases, the dipole character of water needs to be
considered as a permanent positive charge on the oxy-
gen atom and permanent negative charges on the hydrogen
atoms.67 The interaction between charges is described by
the Coulomb potential between point charges, as follows:68
1 q1 q2
VCoulomb = (14)
4 0 r

where q1 and q2 are the interacting charges separated


by the distance r, and 0 is the vacuum permittivity.
According to Molinero and Moore,69 around hundred ver- Figure 6. (A) Nucleation onset pressure as a function of temperature
sions of the intermolecular potential of water exist, with for argon at a nucleation rate of 107 s−1 cm−3 calculated from differ-
the most popular ones being SPC,70 SPC/E,67 TIP3P,71 ent models (CNT, SC-CNT, RKK, MC simulation-based) and measured
TIP4P,71 and TIP5P.72 All these water models treat the experimentally, and (B) nucleation rates for water as a function of super-
saturation. Reprinted with permission from [25], J. Merikanto et al., Phys.
water molecules as rigid molecules. However, the popular
Rev. Lett. 98, 145702 (2007). © 2007, The American Physical Society.
potentials all include the electrostatic term, which is long-
ranged and therefore computationally heavy, to obtain a
correct description of the hydrogen bonds in water. There- to calculate the movement of the individual particles, and
fore, in order to improve computational efficiency other as the time steps need to be in the femtosecond range.65 73
simpler potentials have been developed. The monatomic For instance, Nagayama and Tsuruta74 75 has used MD
water (mW) model presented by Molinero and Moore69 is simulations in the study of condensation coefficients of
an example of such an improved system, where the electro- argon and water.
static term is replaced by a short-range term, which favors In comparison to MD, the MC approach for modelling
tetrahedral configurations of the water molecules. The of condensation at the molecular scale is computationally
description of intermolecular potentials may also be done less heavy, and the method is widely used for simulation
by applying a quantum mechanical approach, whereby the of condensation.25 57 76–78 Figure 6 shows an example of
description is actually at the electron level.16 results from MC simulations by Merikanto et al.25 com-
Additionally, in the case of condensation on a sub- pared to results obtained from CNT and two modified
strate, the intermolecular potentials between the substrate versions of CNT (SC-CNT79 and RKK80 ). The simulation
molecules and the forces between molecules of the sub- results are furthermore compared with results from nucle-
strate and water molecules must be known.57 ation pulse experiments.81 82
When knowing the intermolecular potentials in a system The first flow scheme for an MC method was presented
it is possible to calculate the movement of the individ- by Metropolis et al.65 From an initial configuration of
ual molecules within a certain time interval by MD when molecular positions a new configuration is generated by
the initial positions and velocities of the molecules are random displacement of a molecule in the initial configu-
known. Hereby, it is possible to calculate e.g., how the ration. Hereafter the potential energies of the old and new
condensation process progresses in time and the nanoscale configurations, calculated by intermolecular potentials, are
effects discussed in the previous sections are inherent. compared, and if the new configuration has a lower energy
However, it is computationally very heavy due to the need it is accepted whereafter it is used as basis for generation

1866 J. Nanosci. Nanotechnol. 14, 1859–1871, 2014


Jensen et al. Water Condensation: A Multiscale Phenomenon

of another configuration change in the iteration process. from thermodynamic considerations by Rose60 as pre-
If, on the contrary, the new configuration has a higher sented in Eq. (15).
energy than the old one it is compared with a random
2Tw
number to determine if it is accepted or rejected as basis rmin = (15)
for the next configuration change. The process is iterated hLV · L · TV − Tw 
until an equilibrium state is achieved. An ensemble of con- Here, Tw and TV are the wall and vapor temperatures,
figurations is obtained from the MC method whereby it is respectively, hLV is the latent heat of condensation, and
possible to calculate equilibrium properties by sampling. L is the liquid density. According to this description the
That is, the MC method is stochastic in nature as opposed minimum drop size is depending on the liquid properties
to the deterministic MD method.65 and wall subcooling. As stated in Section 3.2.4 and has
been pointed out by Sikarwar et al. the nucleation site
5. SIMULATION OF CONDENSATION density depends on the surface roughness59 and surface
The dynamic simulation of heterogeneous condensation is subcooling.60
not a trivial task, as illustrated by the condensation theory In the model by Sikarwar et al. the growth of individual
outlined above. In principle, it is required to combine sev- drops is described by:1
eral models in order to simulate the phase change occur- dr 4Tt 1−rmin /r 1−cos
ring in a multiscale condensation process. The core of the = · · (16)
dt L ·hLV 2/hi +r/kl 2−3cos +cos3
simulation is suggested to be the growth models describ-
ing the different regimes of drop growth in condensation. where Tt is the total temperature drop made up of tem-
That is, nucleation, diffusion, and coalescence, and addi- perature drops due to resistance of interfacial heat trans-
tionally drop slide-off on inclined surfaces should be mod- fer, resistance of thermal conduction, and resistance due to
elled. To simulate the nucleation correctly it is envisaged droplet curvature. The total temperature drop will balance
that the nucleation model should be coupled to a surface the subcooling. hi is the interfacial heat transfer coefficient
model describing the distribution of nucleation sites on and kl is the thermal conductivity of the liquid. Coales-
the substrate. That is, the surface model should describe cence is considered, as drops that make contact are antici-
the substrate heterogeneity. The driving force in condensa- pated to coalesce instantaneously, whereby the coalescing
tion is provided by temperature and humidity differences. drops form a new drop with volume conserved and located
Therefore coupling of a heat and mass transfer model to at the weighted center of mass. The critical drop size at
the droplet growth models is essential for the simulation slide-off from an inclined substrate surface is obtained in
of condensation. the model by setting up the balance of forces acting on
the drops with gravity favoring sliding and surface tension
5.1. Drop Growth Models preventing it. Likewise, a maximum drop size is given, as
In literature there has been focus on dropwise conden- drops larger than this size will fall off the surface due to
sation as the heat transfer coefficient of dropwise con- gravity when the surface is inclined more than 90 com-
densation is usually an order of magnitude higher than pared to horizontal. In the expressions for critical and max-
that for filmwise condensation, which is caused by the imum drop sizes the contact angle hysteresis is also taken
enhanced slide-off from hydrophobic surfaces compared into consideration.
to hydrophilic ones.1 58 To a great extent this interest has Figure 7 shows pictures from experiments compared to
driven the research in drop growth where the higher heat simulation results of Sikarwar et al.1 83
transfer coefficient may be utilized in the design of steam
condensers etc. One of the major challenges in the mod- 5.2. Nucleation Models—Molecular Modelling
elling of drop growth lies within the fact that the phe- The nucleation of a droplet during condensation involves
nomenon is multiscale, ranging from the atomic scale at relatively few molecules, and the nucleus size is in the
nucleation to the macroscale at drop slide-off from inclined nanoscale realm where continuum models do not apply.
surfaces.27 Numerous references describing the individual Therefore, molecular modelling is an obvious choice for
regimes of droplet growth can be found in the literature. simulation of the nucleation step of condensation.
However, only one model describing the entire growth For instance Mokshin and Galimzyanov84 have simu-
process from nucleation to slide-off has been found. The lated homogeneous nucleation of water by MD using the
model, developed by Sikarwar et al.1 83 consists of a loop mW model. From these simulations they have been able
for simulation of condensation by diffusion running in par- to calculate surface tension, nucleation rate, and critical
allel with a loop for simulation of coalescence. In the cluster size etc. The calculated surface tension matches
model it is assumed that the initial nucleation site density experimental data well, whereas the nucleation rate and
is 109 m−2 and that any free nucleation site is instanta- critical cluster size are only compared with other models,
neously occupied by a nucleus of smallest possible stable showing that the mW model gives higher nucleation rate
size. The radius of these nuclei, rmin , has been obtained (Fig. 8) and higher critical cluster size as compared to the

J. Nanosci. Nanotechnol. 14, 1859–1871, 2014 1867


Water Condensation: A Multiscale Phenomenon Jensen et al.

Figure 7. Dropwise condensation on a substrate. Left images show the steps of condensation observed experimentally, and right images show
simulation results of Sikarwar et al. (2012) for the steps. Reprinted with permission from [1], B. S. Sikarwar et al., Heat Transfer Eng. 33, 301 (2012).
© 2012, Taylor & Francis Ltd.

CNT. This illustrates one of the challenges in modelling have used the nucleation pulse technique for measuring the
nucleation; namely, the fact that the phenomenon is hard rates of homogeneous nucleation of water at different tem-
to examine experimentally, as e.g., the critical nucleus peratures. The measurements have shown that the CNT did
consists of less than 100 water molecules, wherefore not provide a good fit to the experimental data, though the
experimental data for comparison are scarce.85 However, technique has been later questioned by Winter et al.85 as
experimental measurements of the nucleation rate do exist, the nucleation rate measurements by the nucleation pulse
as for instance the measurements by Viisanen et al.86 who technique require that nuclei have grown to a detectable
size. An experimental study of heterogeneous nucleation
has been conducted by Lauri et al.87 using environmental
scanning electron microscopy (ESEM) showing that CNT
overestimates the supersaturation required for nucleation
onset. However, this technique also suffers from the fact
that nuclei of critical size are too small to be observed
and therefore the nuclei must grow before they become
detectable.
The failure of the CNT at very small cluster sizes has
also been demonstrated by Merikanto et al.25 who has used
molecular MC simulations to show that the CNT fails at
small cluster sizes in the range of 8–50 molecules depend-
ing on temperature and molecules in question. MC simu-
lations by Winter et al.76 85 show that inclusion of the line
tension in the Young equation increases the accuracy of the
CNT for describing the free energy of droplet formation
at heterogeneous nucleation. Examples of snapshots from
the simulations of Winter et al. are shown in Figure 9.
Figure 8. Nucleation rate as a function of temperature calculated from
MD simulations of a Lennard-Jones fluid by Inci and
MD simulations using different intermolecular potentials of water, and
from the CNT. Reprinted with permission from [84], A. V Mokshin Bowles88 have shown that the heterogeneous nucleation
and B. N. Galimzyanov, J. Phys. Chem. B 116, 11959 (2012). © 2012, on nanosized seed particles is well described by the CNT
American Chemical Society. with respect to the critical nucleus size, whereas the energy

1868 J. Nanosci. Nanotechnol. 14, 1859–1871, 2014


Jensen et al. Water Condensation: A Multiscale Phenomenon

the absolute humidity is anticipated not to vary in space


in both the models mentioned above. Firstly, steady state
methods are more inaccurate to predict the condensation
risk compared to transient methods.96 Secondly, dynami-
cal simulation of the development of condensation requires
a transient model, and mass transfer cannot be ignored,
meaning that temperature and humidity may vary both
in time and space.4 A transient model including conduc-
tive and convective heat transfer and diffusive and con-
vective mass transfer has been made by Liu et al.4 The
Figure 9. Snapshots of MC simulations by Winter et al. (2009) of het- model includes local variations in temperature and abso-
erogeneous condensation with contact angles of ∼ 58 (left) and 90 lute humidity and defines condensation rate from the local
(right). Reprinted with permission from [85], D. Winter et al., J. Phys.:
supersaturation of vapor and the moisture transfer coef-
Condens. Matter 21, 464118 (2009). © 2009, Institute of Physics.
ficient, which is again calculated from the heat transfer
coefficient. This is based on an assumption of heat and
barrier of nucleation, and thereby the nucleation rate, is mass transfer analogy, and eliminates the need to solve the
overestimated by CNT. For a comprehensive review on mass diffusion equation.91 However, according to Croce
molecular simulations and other models of nucleation the et al.91 the assumption is not always valid. The model of
reader can refer to the book by Vehkamäki.16 Liu et al. illustrates another problem of obtaining an accu-
rate model; namely the issue of modelling turbulent flow
5.3. Heat and Mass Transfer Models accurately. Though, the accuracy of turbulence models is
Simulation of heat and mass transfer is often done based outside the scope of this review, and it will therefore not
on numerical solving of the governing equations of fluid be further discussed.
dynamics; namely the mass, momentum, and energy equa- A transient heat and mass transfer model taking some
tions founded on the assumption of mass conservation, substrate effects into consideration has been produced
Newton’s second law, and the first law of thermodynamics, by Croce et al.91 92 97 98 In their model they distinguish
respectively.89 This method takes each phase into account between filmwise and dropwise condensation for the cal-
as a continuum described by their respective macroscopic culation of heat flux to the substrate and they find that
properties. Through the governing equations it is possi- calculation of the condensation rate from the mass diffu-
ble to describe the heat transfer, which can occur by the sion equation is more accurate than using the heat and
mechanisms of conduction, convection, and radiation,90 mass transfer analogy.91 Furthermore, the latent heat of
and analogously it is possible to describe the mass transfer condensation is included in the model and it is found to
by diffusion and convection. In the case of water con- be an important factor in the simulation of condensation
densation on a solid substrate three different domains are and evaporation with natural convection.
identified. These are the vapor flow domain, the water In all of the mentioned models it is assumed that con-
flow domain, and the solid domain.91 In this instance, the densation occurs when the temperature gets below the dew
problem of dew formation on a surface is essentially a point temperature, whereby the effect of the substrate, as
conjugate fluid-solid heat and mass transfer problem in is part of the nucleation theory, is neglected. Furthermore,
the presence of a phase changing fluid layer on the solid a detailed resolution of the liquid–gas interface cannot be
surface.92 obtained by CFD due to the macroscopic nature of this
The modelling of condensation by heat and mass trans- approach. This is caused by the fact that the continuum
fer models alone has been attempted by several authors, assumption breaks down at nanoscale.99 100
and it is a common assumption that condensation will
occur when the temperature gets below the dew point 5.4. Simulation of the Solid–Liquid Interface
temperature.4 5 7 8 93 94 In these models nucleation as such In the simulation by Gu et al.101 who have simulated film-
is not considered and it is implicitly anticipated that the wise condensation of argon on a cooled substrate, the
condensation forms a film on a completely wetted sub- interface is simulated at the nanoscale by MD and this is
strate. In the simplest models only steady state heat trans- coupled to a direct simulation MC approach for simulation
fer by conduction is modelled, whereas convective and of bulk gas flow tangential to the substrate surface in order
radiative heat transfer are considered by conversion of to capture the interfacial details. However, the MD part of
these to an equivalent conductive heat transfer.8 95 A more the simulation only includes the Lennard-Jones potential
advanced steady state model has been presented by Bong to calculate the interaction between molecules, which is an
et al.7 where both conductive and convective heat trans- oversimplification for a realistic water simulation. Further-
fer are considered with a k– turbulence model used for more, only filmwise condensation on a perfect crystalline
the convection. Though, radiative heat transfer and mass surface is modelled and nucleation sites are not taken into
transfer are not considered in this model, and furthermore consideration.

J. Nanosci. Nanotechnol. 14, 1859–1871, 2014 1869


Water Condensation: A Multiscale Phenomenon Jensen et al.

Another hybrid model for the condensation of argon on 13. R. W. Carpick, Micro- and Nanoscale Phenomena in Tribology,
a platinum substrate has been developed by Sun et al.99 edited by Y.-W. Chung, CRC Press, London (2011), pp. 51–76.
which combines MD simulation at the interface, where 14. D. Beysens, Atmos. Res. 39, 215 (1995).
15. F. Eslami and J. A. W. Elliott, J. Phys. Chem. B 115, 10646 (2011).
microscopic phenomena mainly exist, with finite volume 16. H. Vehkamäki, Classical Nucleation Theory in Multicomponent
methods for simulation of heat transfer and fluid flow Systems, Springer-Verlag, Berlin, Heidelberg (2006).
in the bulk gas phase, where the continuum assumption 17. D. Turnbull, J. Chem. Phys. 18, 198 (1950).
applies. Again, the MD simulation only uses a Lennard- 18. K. K. Varanasi, M. Hsu, N. Bhate, W. Yang, and T. Deng, Appl.
Phys. Lett. 95, 094101 (2009).
Jones potential to describe the interaction between argon
19. B. Na and R. L. Webb, Int. J. Heat Mass Transfer 46, 3797 (2003).
molecules and a perfectly crystalline surface. Due to the 20. T. Young, Phil. Trans. R. Soc. Lond. 95, 65 (1805).
lack of inherent dipole moments in this system the results 21. T. Nishino, M. Meguro, K. Nakamae, M. Matsushita, and Y. Ueda,
cannot be ported directly to water systems. Langmuir 15, 4321 (1999).
22. R. N. Wenzel, Ind. Eng. Chem. 28, 988 (1936).
23. A. J. Klemm, P. Klemm, K. Rozniakowski, and G. Galbraith, Build.
6. CONCLUSION Environ. 37, 1215 (2002).
The condensation process is a multiscale phenomenon, as 24. A. B. D. Cassie and S. Baxter, Trans. Faraday Soc. 40, 546 (1944).
25. J. Merikanto, E. Zapadinsky, A. Lauri, and H. Vehkamäki, Phys.
drop sizes span from molecular scale to typically mil-
Rev. Lett. 98, 145702 (2007).
limeter scale during their growth starting with nucleation 26. M. Sokuler, G. K. Auernhammer, C. J. Liu, E. Bonaccurso, and
and ending with slide-off on inclined surfaces. In litera- H. J. Butt, Epl 89, 36004 (2010).
ture, the condensation process has been thoroughly stud- 27. R. N. Leach, F. Stevens, S. C. Langford, and J. T. Dickinson,
ied, and modelled at both the molecular scale and the Langmuir 22, 8864 (2006).
28. H. Zhao and D. Beysens, Langmuir 11, 627 (1995).
macroscale. However, to the best of our knowledge a holis- 29. T. Rogers, K. Elder, and R. Desai, Phys. Rev. A 38, 5303 (1988).
tic model of condensation, spanning the entire process, 30. E. De Candido, G. Croce, and P. D’Agaro, Heat Transfer. Eng.
does not exist. This may be considered as a natural con- 33, 1130 (2012).
sequence of the fact that the molecular models used to 31. S. Prestipino, A. Laio, and E. Tosatti, Phys. Rev. Lett. 108, 1 (2012).
describe nucleation become too complex and computation- 32. S. Ghosh and S. K. Ghosh, J. Chem. Phys. 134, 024502 (2011).
33. S. Toxvaerd, J. Chem. Phys. 117, 10303 (2002).
ally heavy when larger drops are considered, and the con- 34. J. Bico, C. Marzolin, and D. Quéré, Europhys. Lett. 47, 220 (1999).
tinuum models used to describe larger drops are inaccurate 35. M. Nosonovsky and B. Bhushan, Nano Lett. 7, 2633 (2007).
if they are applied to the nanometer-sized droplets consid- 36. N. A. Patankar, Langmuir 19, 1249 (2003).
ered in the nucleation. The solution to this issue may be 37. A. Marmur, Langmuir 19, 8343 (2003).
the development of hybrid models, however the level of 38. M. Nosonovsky and B. Bhushan, Langmuir 24, 1525 (2008).
39. Y.-T. Cheng and D. E. Rodak, Appl. Phys. Lett. 86, 144101 (2005).
details necessary for describing the interface is a critical 40. X. Xiao, Y. T. Cheng, B. W. Sheldon, and J. Rankin, J. Mater. Res.
and highly debated issue. 23, 2174 (2011).
41. L. C. Gao and T. J. McCarthy, Langmuir 23, 3762 (2007).
Acknowledgments: The authors would like to thank 42. Y. T. Cheng, D. E. Rodak, A. Angelopoulos, and T. Gacek, Appl.
Phys. Lett. 87, 194112 (2005).
Eirik Bjørn for insightful discussions. This work was sup-
43. M. Nosonovsky, Langmuir 23, 9919 (2007).
ported by DOVISTA and The Danish Agency of Science 44. G. McHale, Langmuir 23, 8200 (2007).
Technology and Innovation. 45. A. Amirfazli and A. W. Neumann, Adv. Colloid Interface Sci.
110, 121 (2004).
46. L. Boinovich and A. Emelyanenko, Adv. Colloid Interface Sci.
References and Notes 165, 60 (2011).
1. B. S. Sikarwar, S. Khandekar, S. Agrawal, S. Kumar, and 47. G. Navascueìs and P. Tarazona, J. Chem. Phys. 75, 2441 (1981).
K. Muralidhar, Heat Transfer. Eng. 33, 301 (2012). 48. J. W. Gibbs, The Scientific Papers of J. Willard Gibbs, Longmans,
2. P. Atkins and J. de Paula, Atkins’ Physical Chemistry, 8th edn., Green and Co., New York (1906), p. 288.
Oxford University Press, Oxford (2006). 49. D. Aronov, G. Rosenman, and Z. Barkay, J. Appl. Phys.
3. N. Simoes, F. Branco, and A. Tadeu, Eng. Anal. Boundary Elem. 101, 084901 (2007).
26, 527 (2002). 50. F. Barberis and M. Capurro, J. Colloid Interface Sci. 326, 201
4. J. Liu, H. Aizawa, and H. Yoshino, Build. Environ. 39, 905 (2004). (2008).
5. R. Hohota, G. Rusaouen, M. Woloszyn, and L. Hedegaard, Inter- 51. J. Rafiee, X. Mi, H. Gullapalli, A. V. Thomas, F. Yavari, Y. Shi,
national Energy Agency, Energy Conservation in Buildings and P. M. Ajayan, and N. A. Koratkar, Nat. Mater. 11, 217 (2012).
Community Systems Programme, Zurich (2004), pp. 1–6. 52. P. Cao, K. Xu, J. O. Varghese, and J. R. Heath, Nano Lett. 11, 5581
6. H. J. Glaser, Glass Sci. Technol. 70, 146 (1997). (2011).
7. T. Y. Bong, H. Xue, and H. C. Liew, Build. Environ. 33, 143 (1998). 53. D. Li, F. Y.. Lin, and A. Neumann, J. Colloid Interface Sci.
8. S. Y. Song, J. H. Jo, M. S. Yeo, Y. D. Kim, and K. D. Song, Build. 142, 224 (1991).
Environ. 42, 940 (2007). 54. F. Lin, D. Li, and A. Neumann, J. Colloid Interface Sci. 159, 86
9. M. Davies, Build. Sci. 8, 97 (1973). (1993).
10. D. Beysens, C. R. Phys. 7, 1082 (2006). 55. R. Tadmor, Langmuir 20, 7659 (2004).
11. P. Klemm, A. J. Klemm, and I. Ibrahim, Energy Build. 38, 1468 56. L. C. Gao and T. J. McCarthy, Langmuir 22, 6234 (2006).
(2006). 57. S. V. Shevkunov, J. Exp. Theor. Phys. 107, 965 (2009).
12. V. M. Starov, Adv. Colloid Interface Sci. 161, 139 (2010). 58. N. A. Patankar, Soft Matter 6, 1613 (2010).

1870 J. Nanosci. Nanotechnol. 14, 1859–1871, 2014


Jensen et al. Water Condensation: A Multiscale Phenomenon

59. C. Mu, J. Pang, Q. Lu, and T. Liu, Chem. Eng. Sci. 63, 874 (2008). 81. A. Fladerer and R. Strey, J. Chem. Phys. 124, 164710 (2006).
60. J. W. Rose, Int. J. Heat Mass Transfer. 19, 1363 (1976). 82. J. Wölk and R. Strey, J. Phys. Chem. B 105, 11683 (2001).
61. M. James, T. A. Darwish, S. Ciampi, S. O. Sylvester, Z. Zhang, 83. B. S. Sikarwar, N. K. Battoo, S. Khandekar, and K. Muralidhar,
A. Ng, J. J. Gooding, and T. L. Hanley, Soft Matter 7, 5309 (2011). J. Heat Transfer. 133, 021501 (2011).
62. T. Furuta, M. Sakai, T. Isobe, and A. Nakajima, Langmuir 84. A. V. Mokshin and B. N. Galimzyanov, J. Phys. Chem. B
26, 13305 (2010). 116, 11959 (2012).
63. L. Xue, P. Keblinski, S. Phillpot, S. U.-S. Choi, and J. Eastman, 85. D. Winter, P. Virnau, and K. Binder, J. Phys.: Condens. Matter
Int. J. Heat Mass Transfer. 47, 4277 (2004). 21, 464118 (2009).
64. N. Shenogina, R. Godawat, P. Keblinski, and S. Garde, Phys. Rev. 86. Y. Viisanen, R. Strey, and H. Reiss, J. Chem. Phys. 99, 4680
Lett. 102, 156101 (2009). (1993).
65. F. J. Keil, Top. Curr. Chem. 307, 69 (2012). 87. A. Lauri, I. Riipinen, J. A. Ketoja, H. Vehkamäki, and M. Kulmala,
66. J. N. Israelachvili, Intermolecular and Surface Forces, 2nd edn., Langmuir 22, 10061 (2006).
Academic Press, London (1992). 88. L. Inci and R. K. Bowles, J. Chem. Phys. 134, 114505 (2011).
67. H. J. C. Berendsen, J. R. Grigera, and T. P. Straatsma, J. Phys. 89. H. K. Versteeg and W. Malalasekera, An Introduction to Com-
Chem. 91, 6269 (1987). putational Fluid Dynamics, The Finite Volume Method, 2nd edn.,
68. J. R. Reitz, F. J. Milford, and R. W. Christy, Foundations of Elec- Pearson, Harlow (2007).
tromagnetic Theory, 4th edn., Addision-Wesley, New York (1993). 90. R. A. Serway and R. J. Beichner, Physics for Scientists and Engi-
69. V. Molinero and E. B. Moore, J. Phys. Chem. B 113, 4008 (2009). neers with Modern Physics, 5th edn., Saunders College Publishing,
70. H. J. C. Berendsen, J. P. M. Postma, W. F. Gunsteren, and Fort Worth (2000), pp. 623–629.
J. Hermans, Intermolecular Forces, edited by B. Pullman, Reidel, 91. G. Croce, E. De Candido, and P. D’Agaro, Appl. Therm. Eng.
Dordrecht (1981), pp. 331–342. 29, 1366 (2009).
71. W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, and 92. G. Croce, P. D’Agaro, and F. D. Mora, Int. J. Comput. Fluid D
M. L. Klein, J. Chem. Phys. 79, 926 (1983). 19, 437 (2005).
72. M. W. Mahoney and W. L. Jorgensen, J. Chem. Phys. 112, 8910 93. A. Dehbi, F. Janasz, and B. Bell, Nucl. Eng. Des. 258, 199 (2013).
(2000). 94. J.-D. Li, Int. J. Heat Mass Transfer. 57, 708 (2013).
73. J. Choe and B. Kim, Bull. Korean Chem. Soc. 21, 419 (2000). 95. Dansk Standard, DS/EN ISO 10077-2, 1 (2004).
74. G. Nagayama and T. Tsuruta, J. Chem. Phys. 118, 1392 (2003). 96. D. Mumovic, I. Ridley, T. Oreszczyn, and M. Davies, Build. Serv.
75. T. Tsuruta and G. Nagayama, Microscale Thermophys. Eng. 6, 267 Eng. Res. Technol. 27, 219 (2006).
(2002). 97. P. D’Agaro, G. Croce, and G. Cortella, Appl. Therm. Eng. 26, 1927
76. D. Winter, P. Virnau, and K. Binder, Phys. Rev. Lett. 103, 225703 (2006).
(2009). 98. G. Croce, P. D’Agaro, A. DeAngelis, and F. Mattiello, P. I. Mech.
77. Q. Sun and I. D. Boyd, 38th AIAA Thermophysics Conference, Eng. D-J. Aut. 221, 1241 (2007).
American Institute of Aeronautics and Astronautics, Inc., Toronto 99. J. Sun, Y.-L. He, and W.-Q. Tao, Microfluid Nanofluid 7, 407
(2005), pp. 1–13. (2009).
78. R. P. Sear, J. Phys. Chem. B 110, 4985 (2006). 100. J. Sun, Y. L. He, W. Q. Tao, J. W. Rose, and H. S. Wang, Microfluid
79. S. L. Girshick and C.-P. Chiu, J. Chem. Phys. 93, 1273 (1990). Nanofluid 12, 991 (2012).
80. H. Reiss, W. K. Kegel, and J. L. Katz, Phys. Rev. Lett. 78, 4506 101. K. Gu, C. B. Watkins, and J. Koplik, Phys. Fluids 22, 112002
(1997). (2010).

Received: 30 July 2013. Accepted: 18 August 2013.

J. Nanosci. Nanotechnol. 14, 1859–1871, 2014 1871

View publication stats

You might also like