Accelerat ing t he world's research.

The Cassie equation: How it is meant

to be used
Andrew J. B. Milne

Advances in Colloid and Interface Science

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

T heoret ical mass sensit ivit y of Love wave and layer guided acoust ic plat e mode sensors
Glen McHale

Nat ural and biomimet ic art ificial surfaces for superhydrophobicit y, self-cleaning, low adhesion, and dr…
Bharat Bhushan,

An Elect rowet t ing Model for Rough Surfaces Under Low Volt age
Wei Dai
 Advances in Colloid and Interface Science 170 (2012) 48–55

Contents lists available at SciVerse ScienceDirect

Advances in Colloid and Interface Science

journal homepage: www.elsevier.com/locate/cis

The Cassie equation: How it is meant to be used

A.J.B. Milne, A. Amirfazli ⁎
Department of Mechanical Engineering, University of Alberta, Edmonton, Alberta, Canada T6G 2G8

a r t i c l e i n f o a b s t r a c t

Available online 13 December 2011 A review of literature shows that the majority of papers cite a potentially incorrect form of the Cassie and
Cassie–Baxter equations to interpret or predict contact angle data. We show that for surfaces wet with a
Keywords: composite interface, the commonly used form of the Cassie–Baxter equation, cos θc = f1 cos θ − (1 − f), is
Cassie only correct for the case of flat topped pillar geometry without any penetration of the liquid. In general,
Cassie–Baxter the original form of the Cassie–Baxter equation, cos θc = f1 cos θ1 − f2, with f1 + f2 ≥ 1, should be used. The dif-
Contact angle
ferences between the two equations are discussed and the errors involved in using the incorrect equation are
Contact area
Contact line
estimated to be between ~ 3° and 13° for superhydrophobic surfaces. The discrepancies between the two
Rough surface equations are also discussed for the case of a liquid undergoing partial, but increasing, levels of penetration.
Finally, a general equation is presented for the transition/stability criterion between the Cassie–Baxter and
Wenzel modes of wetting.
© 2011 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2. Error in contact angle prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3. Effects of liquid penetration on contact angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4. Favorability of the Cassie–Baxter versus Wenzel mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5. Range of expected errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
6. Final remark on influence of liquid–vapor fraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Appendix A. Supplementary data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

1. Introduction techniques. Superhydrophobic surfaces are surfaces with extremely

A rough surface can be wet in one of two modes, the Cassie–Baxter
mode [1–4] (for which parts of the interface under the drop is liquid–
vapor due to vapor remaining beneath the drop in the troughs of the
rough surface), or the Wenzel mode [5] (for which the entire solid
area under the drop is wetted). The works of Cassie and Cassie/Baxter
[1–4] have been greatly referenced in recent years, since their equa-
tion can be used to help describe the phenomenon of superhydropho-
bicity (extreme water repellency) displayed both by natural surfaces
(e.g. such as lotus leaves and ducks feathers) and by surfaces
manufactured using lithography, micromachining, etching, or similar
⁎ Corresponding author at: 4-8 Mechanical Engineering Building, University of
Alberta, Edmonton, Alberta, Canada T6G 2G8. Tel.: + 1 780 492 6711; fax: + 1
780 492 2200.
E-mail address: a.amirfazli@ualberta.ca (A. Amirfazli).
high contact angles and low contact angle hysteresis, commonly un-
derstood to be due to the low adhesion properties of the air remain-
ing below the drop in the Cassie–Baxter wetting mode.
A combined Web of Science and Google Scholar search 1 returns
2686 papers referencing the works of Cassie and Cassie/Baxter
1
Baxter, S.; Cassie, A. B. D. Journal of the Textile Institute Transactions 1945, 36, 67–90
was not indexed in Web of Science, and so could not be searched there. A Google Schol-
ar search found only 35 references to this work, which were included. In general, Web
of Science returned more results than Google Scholar, and so its results were used for
the other three works of Cassie and Baxter [2–4]. Further, Cassie, A. B. D. Discussions of
the Faraday Society 1948, 3, 11 is sometimes incorrectly cited as Cassie, A. B. D. Trans-
actions of the Faraday Society. 1948, 3, 11. Thirty-six references to the later were found
with a Web of Science search, even though no such volume or issue were published in
Transactions in that year. The thirty-six references were included in the total count of
references to Cassie and Baxter's works. Finally, of the original 100 references random-
ly selected, 7 could not be retrieved/were in a foreign language. Seven additional refer-
ences were chosen at random and used in the sample.

0001-8686/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.cis.2011.12.001
 A.J.B. Milne, A. Amirfazli / Advances in Colloid and Interface Science 170 (2012) 48–55 49

[1–4]. A random sample of 100 of these papers [6–105] gives 53 that
present some form of the Cassie/Cassie–Baxter equations [53–105],
of which 40 use a form of the equation to interpret contact angle
data on rough surface [66–105] (instead of e.g. heterogeneous
smooth surfaces). Of these 40 references, the majority (~65%) use
Eq. (1) to interpret or explain their data [80–105]; this can be a
cause for alarm on three different fronts as explained below. Consider
Eq. (1):
cosθc ¼ f cosθ−ð1−f Þ
where θc is the predicted Cassie–Baxter contact angle, and f is thought
to be the fractional projected area of a material with smooth surface
contact angle, θ, and the (1 − f) term is thought to reflect the contri-
bution of air remaining under the drop. The use of Eq. (1) potentially
introduces errors in (at least) three aspects of wetting: 1) the predic-
tion and/or interpretation of contact angle data on surfaces, 2) the
effect on contact angle predictions as liquids penetrate into rough
surfaces, and 3) the stability or transition criterion of the Cassie–
Baxter and Wenzel wetting modes on rough surfaces.
The concerns with Eq. (1) stem from the definition of f. Eq. (1)
defines f as the projected area of the solid–liquid interface per unit
projected area under the drop. This means that Eq. (1) is not, in
general, the Cassie–Baxter equation. The original proper Cassie–Baxter
equation, as formulated by Cassie and Baxter [1–4], is Eq. (2):
cosθc ¼ f 1 cosθ1 −f 2
Cassie and Baxter defined f1 as the total area of solid under the
drop per unit projected area under the drop, with θ1 as the contact
angle on a smooth surface of material 1. Likewise, f2 is defined in an
analogous way, with material 2 as air (θ2=180°). Cassie and Baxter
thought that θ1 could be either of the advancing or receding smooth
surface contact angles, giving advancing and receding predictions of
the Cassie–Baxter contact angle, respectively. Johnson and Dettre
[106] later rederived Cassie and Baxter's equation using thermody-
namic principles, showing that θc and θ1 must each be equilibrium
contact angles. However, the form of the equation derived [106]
remained the same as Eq. (2), with the same definitions for f1 and
f2. Marmur and Bittoun have more recently shown that the Cassie–
Baxter, (and Cassie, and Wenzel) equations are approximations
which become valid as the drop size becomes sufficiently large
compared to the wavelength of the roughness/heterogeneity of the
surface [107]. However, when this requirement is satisfied, they find
the same form for the equations. In this review we also focus on
cases where drop size is much larger than the wavelength of the
roughness/heterogeneity of the surface.
ð1Þ Fig. 1(i) shows how the values of f1 and f (and by extension, Eqs. 1
and 2) can be identical in the limiting case of coplanar interfaces.
Fig. 1(ii) and (iii) shows how f1 and f2 are different from the values
of f and 1-f in the general case. As the first problem then, Eq. (1)
will not generally return correct predictions of contact angles on
rough surfaces, while Eq. (2) will. To put the difference between
Eqs. (1) and (2) in another way, using Eq. (1) effectively ignores
any roughness at the solid–liquid and liquid–vapor interface, and is
thus only valid for surfaces such as those in Fig. 1(i). Fig. 1(i) is obvi-
ously not a general rough surface, and thus Eq. (1) is not a general
correct form of the Cassie–Baxter equation. However, many studies
[83–105] (~58% of the randomly selected references which study
rough surfaces), and four works by the present authors [108–111]
examine general rough surfaces such as those in Fig. 1(ii) and (iii),
and use Eq. (1) nevertheless.
The second problem with the use of Eq. (1) stems from its mathe-
matical inconsistency as liquid penetrates the troughs of a rough
ð2Þ surface (i.e. as f → 1). As f → 1, Eq. (1) returns the behavior of the
Young equation. In their largely forgotten passage, however, Cassie
and Baxter noted, “When [f2 is zero equation [2]] reduces to Wenzel's
equation for the apparent contact angle of a rough surface with
the roughness factor f1.”. The Wenzel roughness factor, identical
to f1 in Cassie and Baxter's work if the surface is fully penetrated, is
commonly given the symbol, r. The correct form of the Cassie–Baxter
equation (Eq. 2) therefore must, and does, capture both the Wenzel
equation:
cosθc ð f 1 →r; f 2 →0Þ ¼ cosθw ¼ r cosθ1 ð3Þ

Coplanar (i.e. in the same line) liquid-vapor

and solid-liquid interfaces
a b
f1 = f = , f 2 = = 1 − f
c c

a+b c d
f1 = , f2 = , f =
e e e

f1 ≠ f , f 2 ≠ 1 − f

Fig. 1. Schematic close up of three rough 2-D surfaces. Solid is blue/gray, air is white, liquid is the cross-hatched area above the surface. Liquid–vapor and solid–liquid interfaces of
drop are denoted by the black line. (i) A smooth-topped rough surface, which (for zero penetration of liquid) has coplanar solid–liquid and liquid–vapor interface (i.e. interfaces are
in line with each other). This yields f1 = f and f2 = (1-f) for (i). (ii) An arbitrary rough surface showing the values of f1, f2 and f, with f1 ≠ f and f2 ≠ (1-f) in general. (iii) A dual-scale
rough surface showing (left) a curved liquid–vapor interface, (middle) contact of the drop with pillars of different heights, and (right) partial penetration into a dual scale structure,
all of which increase the effective area of the liquid–vapor interface under the drop. Neither f1 nor f2 are functions of f in all three case of (iii). Further, f1 and f2 can both be greater
than unity (depending on spacing, etc.).
50 A.J.B. Milne, A. Amirfazli / Advances in Colloid and Interface Science 170 (2012) 48–55
and the Young equation (as r approaches 1):
cosθc ð f 1 →1; f 2 →0Þ ¼ cosθy ¼ cosθ1
in one self consistent equation. The Wenzel equation (with θw being the
Wenzel contact angle) is returned as the liquid penetrates the troughs
of the roughness fully, causing f2 → 0 and f1 → r. The Young equation
(with θy being the Young contact angle) is returned as f2 → 0 and
f1 → 1 (i.e. as the surface becomes smooth and homogeneous).
Some researchers (~ 18% of the randomly selected references that
examine rough surface) have noticed the above issues in using
Eq. (1), and have proposed instead Eq. (5):
cosθc ¼ r loc f cosθ þ ð f −1Þ
where rloc is the total solid–liquid area under the drop divided by
the projected solid–liquid area under the drop. Whereas Eq. (5)
does account for the roughness of the solid–liquid interface, it still
carries with it three issues of varying importance. First, it uses a
definition of f which is different from the original ideas of Cassie
and Baxter. Second, one must always remember, and carefully ac-
count for, the fact that the value of rloc can change as liquid penetrates
into the troughs of a surface. Further, in general, rloc does not equal
the Wenzel roughness factor until the liquid fully penetrates the
troughs. Third, Eq. (5) assumes that the liquid–vapor interface is
smooth and lies flat in the macroscopic plane of the solid. This
seems at first to be a fair approximation since as Marmur and Bittoun
pointed out the radius of the drop must be much larger than the spac-
ing of the roughness for the Cassie–Baxter equation's assumptions
to be valid [107]. However, the liquid–vapor interface can contain ap-
preciable corrugations (i.e. roughness) of its own by at least three
means. First, it can be caused by means of excessive curvature of
the interface (due to, e.g., hydrostatic pressure, Fig. 1(iii) (left
side)). Hydrostatic pressure is an important consideration for the
application of superhydrophobic surfaces (SHS) to underwater struc-
tures or to drop impact on SHS. Two addition causes of liquid–vapor
interface corrugation/roughness are exposure of the drop to pillars
of different heights, or penetration into the larger scale of a dual
scale structure, as shown in Fig. 1(iii) (middle and right side). These
situations can again be caused by hydrostatic pressure, but can be
present even for the case of quiescent sessile drops since in these
situations the liquid–vapor interface can have small curvature, but a
high ratio of true to projected area. Regardless of cause, f2 ≠ (1 − f)
is a general statement. Therefore, use of equations in the form of
Eq. (5) is generally not recommended, and Eq. (2) should be used
instead.
As a third source of error in using Eq. (1), consider the common
practice of determining the favorability of either the Cassie–Baxter
or Wenzel mode of wetting by equating Eqs. (1) and (3) and solving
for θ1 = θcr, where θcr is the minimum intrinsic contact angle needed
for a stable Cassie–Baxter mode [112]. The potential error in Eq. (1)
is therefore propagated to a prediction of stability/transition. Consid-
ering this, a more general stability criterion is presented at the end of
this paper.
The use of Eq. (1) is by no means universal (e.g. see references
[66–72,113–118] which properly cite or otherwise acknowledge
Eq. (2) as the Cassie–Baxter equation). Indeed, J. R. Henderson gave
a general comment in a recent Faraday Discussion [119] that covered
many of the same points made in this paper. Henderson proposes the
use of Eq. (5), and notes that in the most general case, one would
need to account for liquid–vapor interface roughness as well. He
notes that, “some current workers refer to the Cassie equation as if
there were no Wenzel roughness factors present on the right side.
This is definitely not what Cassie had in mind” [119]. He also makes
reference to the effects of pillar wall angle as we shall later discuss
in this paper, and discusses the change in energy as liquids penetrate
a surface from the tops of the pillars. Henderson does not, however,
highlight the errors inherent in the use of Eq. (1), or present a general
ð4Þ form of the transition/stability criterion between the Wenzel and
Cassie–Baxter modes. Further, his use of Eq. (5) would still lead to
errors for surfaces similar to those in Fig. 1(iii).
It is worthy to note here the contribution by Pease [120] of an
alternate equation for contact angles based on force balances con-
ducted around the contact line. Also of note is the more resent discus-
sion of contact area versus contact line importance (e.g., see
references [121,122]), and how for uniformly rough surfaces there is
little difference (e.g., see references [58,107,123,124]). Regardless,
the same arguments put forth in this paper regarding the correct
form of the Cassie–Baxter equation (Eq. 2), hold if one uses Pease's
ð5Þ equation. In Pease's equation, the sum of the line fractions is greater
than or equal to one, just as how in the Cassie and Cassie–Baxter
equations the sum of the area fractions is greater than or equal to
one. This means that whether one attributes the predicted contact
angle on a rough surface to area or contact line fractions, the same
implications as outlined above will persist.
In the literature, the terms ‘Cassie–Baxter’ and ‘Cassie’ have largely
been used interchangeably. In this review, ‘Cassie–Baxter’ is used for
the case of a rough (heterogeneous or homogeneous) surface incom-
pletely wet by a liquid with vapor remaining below the drop. The
term ‘Cassie’ is reserved for the case of heterogeneous (smooth or
rough) solid surfaces wet completely by the liquid. The distinction
is somewhat arbitrary since the most full form of the Cassie equation:
 
cosθc ¼ ∑ f n cosθn ð6Þ
n
will capture all forms of wetting (e.g. Cassie–Baxter, Cassie, Wenzel,
and Young). As such, Eq. (6) can predict contact angle for an arbitrary
(rough or smooth, chemically homogeneous or heterogeneous, wet
with air remaining or completely wetted) surface. Eq. (6) is fully arbi-
trary as long as one allows for the total areas of each (solid–liquid or
liquid vapor) interface under the drop such that ∑ f n ≥1, and such
n
that each of the n interfaces under the drop can have a total area in
excess of its planar area. Put another way, a heterogeneous surface
can be rough, and further, can be wet by a liquid with or without
air remaining under the drop (the presence of air further adding to
the heterogeneous and/or rough interface under the drop). In this
paper, we will focus on the case of a homogeneous rough surface
wet with vapor remaining below the drop, for which Eq. (6) sim-
plifies to Eq. (2). However, all of the arguments presented in this
paper can be applied to Eq. (6).
In summary, while the incorrect use of Eq. (1) may not be univer-
sal, it is certainly persistent. We have therefore decided to document
comprehensively the distinctions between Eq. (2) and Eqs. (1) and
(5), as well as to present a general Cassie–Baxter to Wenzel transition
criterion.
2. Error in contact angle prediction
To illustrate the error inherent in the use of Eq. (1), consider
the surfaces shown in Fig. 2, which are wet to some uniform level of
penetration. The surfaces shown in Fig. 2 are meant to represent sim-
plified forms of common topographies of superhydrophobic surfaces
(SHS). Contact between the liquid and the surface must cause some
level of penetration of the drops on these surfaces (except for
Fig. 2a), since otherwise the drop would contact the surface at a series
of mathematical points, which would be non-physical. Assuming a
uniform level of penetration means that the liquid–vapor interface
under the drop can be approximated as planar (neglecting for now
the curvature of the interface), while the solid–liquid interface
under the drop is rough. Making this approximation allows one to
 A.J.B. Milne, A. Amirfazli / Advances in Colloid and Interface Science 170 (2012) 48–55
a b
Liquid Level
c d
Liquid Level
Fig. 2. Schematic close up showing surfaces decorated by (a) flat topped square pillars,
(b) equilateral pyramid topped square pillars, (c) conical pillars (half angle of 15°,
not to scale), and (d) discrete spherical particles. Each surface is assumed to support
a planar liquid–vapor interface at some arbitrary level denoted by the thick dashed
lines and label.
express f2 = 1 − f, regardless of the roughness of the solid–liquid in-
terface for a given surface (i.e. regardless of f1 as defined for
Eq. (2)). Contact angle predictions are therefore presented versus
f2 = 1 − f to allow comparison between Eqs. (1) and (2).
The predictions of Eqs. (1) and (2) for the surfaces in Fig. 2
are compared in Fig. 3, for θ1=100°. Contact angle is plotted against
liquid–vapor surface fraction f2, varied by changing the spacing be-
tween pillars. Predictions for the spherical particles (Fig. 2d) start at
f2 ≈ 0.0931, the closest packing possible for this case. Similarly, the
conical pillars have the same maximum packing density at their
base. The conical pillars and spherical particles are assumed to be
wet to 1/2 their height as shown in Fig. 2. Because of this, the conical
pillar contact angle predictions start at f2 ≈ 4 × 0.0931 ≈ 0.372.
Fig. 3 shows the prediction of Eq. (1) does not follow the predic-
tion of Eq. (2), and fails to return identical predictions at any value
of f2 = (1-f), expect for flat topped surfaces (see Fig. 2a). The error
introduced by using Eq. (1) is higher at low values of f2 = (1-f), and
decreases with f2. For example, at f2 ≈ 0.37, the error introduced by
using Eq. (2) can be as high as ~ 24° for the geometries considered,
while at f2 = 0.75 and f2 ≈ 0.95, the errors are ~ 14° and ~6°,
Low f1
Fig. 3. Contact angles predicted by Eqs. (1) and (2) on surfaces shown in Fig. 2
for θ1 = 100°. Inset shows predictions at high values of f2 (equal to (1-f) in the case
of planer liquid–vapor interface). Low values of f1, corresponding (roughly) to high
values of f2, are interesting for application to superhydrophobic surfaces. Contact
angle for f2 = 0 correspond to Wenzel mode (fully penetrated) predictions for surfaces
made up of only the pillar tops with no space between them (e.g. a smooth surface, and
a surface covered in square based equilateral pyramids, respectively).
51
respectively. The decrease in error with f2 = (1-f) is due to the de-
creasing importance of the f1 term as f2 increases.
In Fig. 3, whereas the value of f2 affects the magnitude of the error
in using Eq. (1), the error itself is due to the failure of Eq. (1) to ac-
count for the roughness of the pillar tops (i.e. the increased area of
the solid–liquid interface under the drop). At the assumed level of
penetration, the greatest increase in area is for the conical pillars (at
1
sinð15∘Þ ≈3.86 times the area for a half angle of 15°). The distributed
spheres (at 2 times the area) have the next greatest effect. The pyra-
midal topped pillars have the least effect of the structures considered
(~1.73 times the area), except for the smooth topped pillars (which,
by definition, have no increased area if there is no penetration). In-
creasing roughness on the pillars increases the contact angle, and
therefore increases the error in using Eq. (1). The amount of increased
area is obviously dependent on the level of penetration and the shape
of the topography (steepness of pyramid or cone angle, spacing of
roughness (captured by f2), etc.). The analysis assumes the roughness
size is sufficient to affect the drop (i.e. nanometer scale or larger). This
argument of a limiting scale is supported by literature [125,126],
though the other arguments presented [125,126] of a molecular
scale Cassie–Baxter like mode is questionable and will be left for
now. The effects of penetration depth will be explored after the ef-
fects of intrinsic contact angle are discussed.
Fig. 3 showed predictions for θ1 = 100°. For θ1 = 110° (not shown)
the prediction for conical pillars are 180° for all values of f2 = (1 − f). If
θ1 = 120°, similar predictions of 180° are seen for the spherical parti-
cles as well. These results are in the online supplement, along with re-
sults for two additional pillar shapes (hemisphere and hemicylinder
topped pillars). The combination of the intrinsic contact angle and
the roughness of the pillar tops leads to these nonsensically high con-
tact angles, and indicate that the drop would likely retreat further up
these rough toped pillars. Interestingly, as the liquid does so, the pre-
dicted contact angles are still 180° for the case of conical pillars with
half angle of 15° as shown in Fig. 2. This is similar to the findings of
Marmur [127]. Besides this finding, the errors in using Eq. (2) are in-
creased as θ1 increases, assuming the liquid–vapor interface remains
fixed under the drop. Since SHS are often made with intrinsic contact
angles of θ1 ≈ 120° (from, e.g. PTFE), the error in using Eq. (1) will be
even further increased from the amounts reported in Fig. 3.
As a note, SHS are widely defined as having contact angles greater
than 150°. The results above suggest therefore, that for a surface com-
posed of scattered spherical particles (or perhaps for hemisphere
topped pillars) a SHS may be possible with liquid–vapor area frac-
tions as low as f2 ≈ 0.6 and θ1 = 110° (this compares to requirements
of f2 ≈ 0.8 for flat topped pillars). Whether such a spherical/hemi-
sphere based surface would possess the low contact angle hysteresis
necessary for classification as a superhydrophobic surface cannot be
answered, since Eqs. (1) and (2) both deal with equilibrium contact
angle. Regardless, Eq. (2) explains why researchers have had success
forming SHS from spherical particles deposited on surfaces. Such an
explanation would be more difficult using Eq. (1).
3. Effects of liquid penetration on contact angle
As stated above, the predicted contact angle will depend on the
amount of penetration of liquid into the troughs of the rough surface.
A given level of penetration for liquid was assumed for the surfaces
in Fig. 2. Consider now Fig. 4, showing predictions (based on Eqs. 2
and 1) of contact angle on a surface decorated with spherical particles
(i.e. the surface shown in schematic form in Fig. 2d) for various
amounts of liquid penetration. The spheres are placed so that at 50%
penetration height, f1 = 0.4, f2 = 0.8. The intrinsic contact angle varies
from θ1 = 90° to θ1 = 120°.
Fig. 4 shows that for a surface decorated with spherical particles,
the contact angle decreases with penetration to a minimum value,
52 A.J.B. Milne, A. Amirfazli / Advances in Colloid and Interface Science 170 (2012) 48–55
Fig. 4. Correct predictions (from Eq. 2) and incorrect predictions (from Eq. 1) for
contact angle on a surface decorated with non-close-packed spherical particles as a
function of depth of penetration into the troughs of the surface. Intrinsic contact angles
varies from θ1 = 120° to θ1 = 90° and the spheres are placed so that at 50% penetration
depth, f1 = 0.4, f2 = 0.8.
and then increases with further penetration. Eq. (2) shows that the
fractional penetration for minimum contact angle is inversely depen-
dent on the intrinsic contact angle, θ1. Further, Eq. (2) correctly pre-
dicts a discontinuous jump to the Wenzel contact angles when full
penetration is reached. Eq. (1), on the other hand, incorrectly predicts
the minimum at 50% penetration for all intrinsic contact angles, and
only matches Eq. (2) for the case of θ1 = 90°, for which solid–liquid
roughness has no effect. Eq. (1) also incorrectly jumps to the intrinsic
contact angle at full penetration (not shown on the graph). It is worth
noting that not all levels of penetration may be stable on a given sur-
face, i.e. not all loci shown in the curves of Fig. 4 are necessarily at
equilibrium from a thermodynamic perspective, and the liquid may
therefore immediately jump to either the Wenzel mode or to another
more stable level of penetration in the Cassie–Baxter mode. Based on
the concepts discussed in Section 4, it is likely that the Wenzel mode
is favorable in all cases for Fig. 4, with metastable Cassie–Baxter
modes at the minima of Fig. 4. The concept of stability of different
wetting modes will be discussed more in Section 4.
The increase or decrease in contact angle with penetration de-
pends on the intrinsic contact angle and the local slope of the rough-
ness. The variation in roughness slope for the spherical particles gives
Fig. 4 its minima. Surfaces with constant roughness slope (such as
those in Fig. 5) still show a change in contact angle with penetration,
however, and reinforce the idea that Eq. (1) is only applicable for flat
topped pillars with no penetration of the liquid. Consider surfaces
with unit cells shown in schematic form in Fig. 5. The liquid–vapor in-
terface starts at the top of the pillars (shown as the top dotted line)
and the contact angle behavior is considered as the liquid penetrates
(shown as the progressively lower dotted lines).
Predictions of the contact angle behavior based on Eqs. (2) and (1)
are shown in Fig. 6 for both surfaces considered in Fig. 5, for θ1 = 100°.
In Fig. 6a it can be seen that Eq. (1) fails to properly predict the
contact angle as the liquid penetrates the straight walled surface. It
Fig. 5. Schematic close up of two 2-D structures wet by a drop at progressively greater
amounts of partial penetration (denoted by dotted lines). Surface ‘a’ is a surface of flat
topped pillars with f = 0.25 and an aspect ratio (pillar height to width) of 3. Surface ‘b’
is made up of close-packed equilateral triangular prisms.
shows no increase in contact angle as the liquid penetrates. Likewise,
at complete penetration, the prediction discontinuously jumps to the
intrinsic contact angle for the smooth surface. In contrast, the predic-
tion of Eq. (2) shows that contact angle increases as liquid penetrates
the surface. This is true for any intrinsic contact angle above 90° on
any vertical walled pillared surfaces such as this. The prediction of
Eq. (2) also shows a discontinuity at full penetration, this time
correctly jumping from the Cassie–Baxter contact angle (180°) to
the Wenzel contact angle (~116°), as full penetration occurs and the
liquid vapor interfaces under the drop are replaced by solid–liquid
interfaces. Using the concepts outlined in Section 4, it is likely that
the Wenzel mode is favorable.
Fig. 6b shows predictions of contact angle on a surface with angled
walls as shown in Fig. 5b. On such a surface, the liquid–vapor area
decreases as the liquid penetrates. For the wall angle and intrinsic
contact angle chosen, the energy of this decrease offsets the increase
in energy from the increase in solid–liquid area. As a result, contact
angle decreases as penetration progresses, with Eq. (2) correctly
returning the Wenzel contact angle at full penetration. In Fig. 6b
there is no discontinuity in contact angle at full penetration since
the liquid–vapor area approaches zero as penetration increases.
Eq. (1), on the other hand, still predicts incorrect contact angles for
all but the impossible case of zero solid–liquid contact (zero penetra-
tion), and again incorrectly returns the Young's contact angle on a
fully penetrated rough surface. It is worth noting here that other com-
binations of wall angles less than 90° can still show an increase in
contact angle with penetration. The intrinsic contact angle is the
determining factor of which wall angles will show such behaviors.
Similar to Fig. 4, not all loci on the curves of Fig. 6 are necessarily
energetically favorable, and the concepts in Section 4, suggest that
the Wenzel mode is favorable.
a
Two
correct
limits
Two incorrect
limits
b
Fig. 6. Predictions (based on Eqs. 1 and 2) of contact angle on the two surfaces (a and b)
shown schematically in Fig. 5. Surface ‘a’ is a 2-D surface of flat topped pillars with
f = 0.25 and an aspect ratio of 3. Surface ‘b’ is a 2-D surface made up close-packed
equilateral triangular prisms. For both plots, θ1 = 100°.
 A.J.B. Milne, A. Amirfazli / Advances in Colloid and Interface Science 170 (2012) 48–55 53

4. Favorability of the Cassie–Baxter versus Wenzel mode

In 2002, researchers [112] used differential energy balances to de-

fine a stability criterion to determine if the Cassie–Baxter or Wenzel
mode of wetting is favorable on a given surface. The equation devel-
oped was:

f −1
cosθcr ¼ ð7Þ
r−f
where θcr is the critical intrinsic contact angle. According to Eq. (7),
if θ1 > θcr, Cassie–Baxter mode (unpenetrated) wetting is favorable.
If θ1 b θcr, Wenzel mode (penetrated) wetting is instead favorable.
Unfortunately, while the concept of solid–liquid interface rough-
ness was discussed earlier in the original paper, and a previous
paper by the same authors (for the specific case of hemisphere
topped pillars) [113], it was not used in the determination of the sta- Fig. 7. Regions of favorability (stability) for the Cassie–Baxter and Wenzel wetting
bility criterion of Eq. (7), for which the authors used a simplified, flat modes for a surface with r = 2, composed of equilateral pyramid topped square pillars
topped pillar geometry. As such, the applicability of Eq. (7) is limited wet to the base of the pyramid (see Fig. 2b). The spacing of the pillars determines f2
to only flat topped pillars with either no penetration or full penetra- with f2 = (1-f). As solid–liquid interface roughness increases, the prediction of Eq. (7)
becomes less accurate (see arrow).
tion of the roughness. For this reason, its use for SHS prepared by
particle deposition, etching, electrospinning, etc. is questionable.
The original authors noted that using a flat topped pillared surface
penetration will be correctly treated by Eq. (7), while Eq. (8) will han-
was a simplification [113], but readers have not necessarily noted
dle this specific case, and all other cases.
this fact in applying Eq. (7).
Studying Eq. (8) (and knowing that, by definition, f1 ≤ r), it is
Yang et al. also presented a stability criterion using a form of the
seen that increasing either f1 or f2 increases the critical contact
Cassie–Baxter equation similar to Eq. (5) [126], finding that high
angle, making the Cassie–Baxter wetting mode less favorable by re-
roughness is required to stabilize the Cassie–Baxter mode. Further-
quiring higher intrinsic contact angles. This is shown graphically in
more, Marmur presented an excellent and advanced analysis of over-
Fig. 7 by observing the decreased range of intrinsic contact angles
all drop stability [127]; in it, Eq. (5) was used as the Cassie–Baxter
that favor the Cassie–Baxter mode (the white region on the graph)
equation to argue that there are additional criteria for the stability
at higher values of f2. Likewise, increased f1 corresponds to increased
of the Cassie–Baxter mode based on the positive second derivative
solid–vapor roughness, increasing the curvature of the dotted line
of roughness shape. Further, Marmur found that lower contact angle
corresponding to Eq. (8) in Fig. 7, and decreasing the range of favor-
mode indicated a more stable mode.
ability for the Cassie–Baxter mode. For intrinsically hydrophobic ma-
Neither Yang et al. [126] nor Marmur [127] presented a fully gen-
terials, increasing f1 or f2 both increase the Cassie–Baxter contact
eral form similar to Eq. (7) since both used Eq. (5) as their basis.
angle, so one can say that as the Cassie–Baxter contact angle in-
Eq. (5) is a more correct form of the Cassie–Baxter equation, but it
creases, the Cassie–Baxter wetting mode becomes less favorable
still neglects the possibility of liquid–vapor corrugations/roughness
(similar to the finding of Marmur [127]). Similar observations can
under the drop as explained in the discussion of Fig. 1iii. We now dis-
be made using Eq. (7), but since f is bounded between 0 and 1, and
cuss such a general form and will focus on Eq. (7) since it is widely
since it exists in both the numerator and denominator, it is impossible
used.
to describe situations where f1 or f2 vary independently with Eq. (7).
Following a differential energy balance similar to Bico et al. [112],
Further, Eq. (7) suggests that for θ1 = 120° (the maximum attain-
a general form of the stability criterion is given below, written using f1
able smooth surface contact angle with water) a finite value of r = 2 is
and f2 as:
sufficient to favor the Cassie–Baxter mode for any value of f. Eq. (8),
on the other hand, illustrates that as f2 increases, r must increase
−f 2 without bounds to maintain the favorability of the Cassie–Baxter
cosθcr ¼ ð8Þ
r−f 1 mode for θ1 = 120°. As a reminder, f2 can increase due either to curva-
ture of the liquid–vapor interface under the drop or due to partial
Eq. (8) can also be derived by equating the general form of the penetration into a dual scale structure such as that seen in Fig.
Cassie–Baxter equation (Eq. 2) with the Wenzel equation (Eq. 3). 1(iii). This suggests that fractal surfaces (with an effectively infinite
Fig. 7 shows the stability criterion as predicted by Eqs. (8) and (7), r value, but more slowly growing values of f1 and f2), may be the
for a surface (similar to that shown in Fig. 2b) made up of square best surfaces for ensuring Cassie–Baxter mode wetting, similar to
pillars topped with equilateral pyramids wet to the base of the pyra- the arguments put forth by Herminghaus [128].
mids. The increased area of the pillar tops is therefore ~1.73 times,
while the height of the pillars increases with spacing to maintain 5. Range of expected errors
r = 2 for complete penetration. The spacing of the pillars is varied
to change f2 with the assumption that the liquid–vapor interface is Since the (correct or incorrect) Cassie–Baxter equation is often
flat, allowing one to express f2 = (1-f) for comparison of Eqs. (6) used to predict contact angles on superhydrophobic surfaces (SHS),
and (7). Fig. 7 shows that Eq. (7) incorrectly suggests that the it is of interest to quantify the error in making predictions of such sys-
Cassie–Baxter mode is more stable than it actually is. While the tems using Eq. (1) instead of Eq. (2). This is important when readers
prediction is correct at f2 = 0 and f2 = 1, these extremes are not of study the results of the papers reviewed here, or any similar works.
interest (since f2 = 0 corresponds to the Wenzel mode in all cases) Since SHS tend to have a high amount of liquid–vapor interface
or attainable (since f2 = 1 corresponds to zero contact with the solid under the drop, analysis will focus on f2 ≥ 0.8. For illustrative pur-
in this case, which is non-physical). Similar curves can be produced poses, the surfaces already discussed in this paper will be considered.
for surfaces with different roughness geometries, giving different A given level of penetration was assumed for the surfaces in Fig. 2,
values of r, f1 and f2. Only the case of flat topped pillars with no and contact angle predictions were made for a range of penetration
54 A.J.B. Milne, A. Amirfazli / Advances in Colloid and Interface Science 170 (2012) 48–55
depths for a surface of the type in Fig. 2f. It is difficult to determine
what exact level of penetration will be stable for a given SHS (and
therefore what the accompanying error in using Eq. (1) would be).
However, as a first approximation one can take the minimum of
Eq. (2) on Fig. 4 (i.e. for f2 = 0.8). Doing this, the error in using
Eq. (1) is the difference between the curves in Fig. 4 for each intrinsic
contact angle at the minimum of Eq. (2) for that intrinsic contact
angle. This suggests that the error inherent in using Eq. (1) is ~ 4.1°
and ~ 3.4° for θ1 = 120° and θ1 = 105°, respectively, for this surface.
A similar analysis for the equilateral pyramid topped pillars in
Fig. 2b (with f2 = 0.8 at penetration to the base of the pyramids) re-
sults in the assumed most stable contact angle being for penetration
to the base of the pyramid. At this level of penetration, the error in
using Eq. (1) is estimated as ~ 12.5° and ~ 4.4° for θ1 = 120° and
θ1 = 105°, respectively. Better estimates of the error would require
more knowledge of the exact level of wetting that each surface expe-
riences, but the above estimates can be taken as approximations of
the bounds of error (i.e. ~ 3° to ~13°). Greater errors would be possi-
ble for surfaces with even greater solid–liquid interface roughness
(such as the conical pillared surface), but analysis of the most stable
contact angle for such a surface is complicated by its geometry and
will not be attempted. Conversely, smaller error is possible for
lower intrinsic contact angles or lower values of f1, but these criterion
run counter to the design requirements of a SHS. Higher values of f2
(attainable by increasing the spacing of roughness elements without
increasing their size) would also decrease the error, so extremely
sparsely pillared surfaces (e.g. f2 = 0.95) would suffer little error
(~1.3º) for pyramidal topped pillars.
6. Final remark on influence of liquid–vapor fraction
For the sake of simplicity, the assumption of a flat liquid–vapor in-
terface has been made throughout this paper whenever calculations
have been performed. As noted in the Introduction, this need not be
the case, and the correct form of the Cassie–Baxter equation can
capture the effects on contact angle of non-planar liquid–vapor inter-
faces. Any non planar portions of the liquid–vapor interface would
have a substantial effect on contact angle, since the effective contact
angle on liquid–vapor portions is 180°. One could purposely increase
the contact angle in this way by designing surfaces similar to those in
Fig. 1(iii), and Eq. (2) can aid in understanding the benefits of dual
scale roughness on repellency when considered in light of the two
surface types shown at the right of Fig. 1(iii). Alternatively, if one con-
siders Pease's equation [120], a similar effect of increase liquid–vapor
line fraction could be achieved by changing the spacing/heights of the
roughness elements in such a way as to increase the tortuosity of the
contact line. The importance of tortuous contact lines has been noted
before [129].
7. Conclusions
In conclusion, two commonly used variants of the Cassie–Baxter
equation, Eqs. (1) and (less often) (5), are only valid in limited
cases. Eq. (1) fails to properly predict contact angles on general sur-
faces displaying roughness of either the solid–liquid or liquid–vapor
interface under the drop. On superhydrophobic surfaces, the errors
are likely in the range of ~3°–13° or greater. Eq. (5) fails to properly
predict contact angles on general surfaces displaying roughness of
the liquid–vapor interface under the drop due, e.g., to curvature of
the liquid–vapor interface under the drop or penetration of the drop
into the larger structure of a dual scale roughness. In addition to
these limitations, neither Eq. (1) nor Eq. (5) can be used in general
to predict contact angles as liquids partially penetrate into a rough
surface. The original, correct, Cassie–Baxter equation (Eq. 2) does
not suffer from any of these difficulties. Eq. (2) does require knowl-
edge of the extent of penetration of the liquid into the troughs of a
rough surface, however, with this knowledge it can make correct pre-
dictions, while Eqs. (1) and (5), in general, cannot.
Eq. (7) (describing the stability of the Cassie–Baxter versus
Wenzel modes of wetting, and based on a similar premise as Eq. 1)
cannot, in general, be used to determine the favorability of a given
wetting mode and introduces errors in plots of stable regions of
wetting. An analogous, but general, equation for stability is given as
Eq. (8) for the first time.
Given the prevalent use of Eqs. (1) and (7) in situations where
they are not valid, the authors to stress the importance of using
Eqs. (2) and (8). Eqs. (1) and (7) may only be used for the case of
flat topped pillars with no penetration. Eq. (5) may be used to account
for roughness of the solid–liquid interface, but neglects roughness of
the liquid–vapor interface. In contrast, Eqs. (2) and (8) can be used in
all cases, and will simplify to Eqs. (1) (or 5) and (7) when allowable.
Although not discussed at length in this paper, the same argument
presented here can be applied to an arbitrary (rough or smooth,
chemically homogeneous or heterogeneous, wet with air remaining
or completely wetted) surface via the use of Eq. (6). Alternatively,
one can apply the arguments presented here to calculation of wetting
behavior by Pease's contact line approach.
Acknowledgments
This research was funded by a Natural Sciences and Engineering
Research Council of Canada (NSERC) Discovery grant. A. Amirfazli
holds a Canada Research Chair in Surface Engineering. A.J.B. Milne
was supported by the Izaak Walton Killam Memorial Scholarship,
an NSERC CGS-D, and an Alberta Innovates Technology Futures
Nanotechnology Graduate Research Scholarship.
Appendix A. Supplementary data
Supplementary data to this article can be found online at doi:10.
1016/j.cis.2011.12.001.
References
[1] Baxter S, Cassie ABD. J Text Inst Trans 1945;36:67.
[2] Cassie ABD. Discuss Faraday Soc 1948;3:11.
[3] Cassie ABD, Baxter S. Trans Faraday Soc 1944;40:546.
[4] Cassie ABD, Baxter S. Nature 1945;155:21.
[5] Wenzel RN. Ind Eng Chem 1936;28:988.
[6] Badre C, Pauporté T, Turmine M, Lincot D. Nanotechnology 2007;18:365705.
[7] Ball P. Nat Mater 2009;8:250.
[8] Barnette AL, Asay DB, Kim SH. Phys Chem Chem Phys 2008;10:4981.
[9] Bhushan B. Microelectron Eng 2007;84:387.
[10] Brandner BD, Hansson PM, Swerin A, Claesson PM, Wåhlander M, Schoelkopf J,
Gane Patrick AC. Soft Matter 2011;7:1045.
[11] Cai Y, Yun YH, Zhang Newby B. Colloids Surf B Biointerfaces 2010;75:115.
[12] Chang K-C, Chen H, Huang C-K, Huang S-I. J Appl Polym Sci 2007;104:1646.
[13] Darmanin T, Guittard F. Chem Commun 2009;16:2210.
[14] Du J, Michielsen S, Lee HJ. Langmuir 2010;26:16000.
[15] Etzler FM. In: Mittal KL, editor. Contact angle, wettability and adhesion, part II,
Vol. 3; 2003. VSP - An imprint of BRILL, Chapter 12.
[16] García N, Benito E, Tiemblo P, Hasan MMB, Synytska A, Stamm M. Soft Matter
2010;6:4768.
[17] Gunnells DW, Gardner DJ, Wolcott MP. Wood Fiber Sci 1994;26:447.
[18] Gupta TK. Thin Solid Films 1981;75:319.
[19] Hou W, Wang Q. Langmuir 2009;25:6875.
[20] Hsieh C-T, Yang S-Y, Lin J-Y. Thin Solid Films 2010;518:4884.
[21] Janssen D, De Palma R, Verlaak S, Heremans P, Dehaen W. Thin Solid Films
2006;515:1433.
[22] Kalinin YV, Berejnov V, Thorne RE. Langmuir 2009;25:5391.
[23] Kannan R, Sivakumar D. Exp Fluids 2007;44:927.
[24] Kwok DY, Neumann AW. Adv Colloid Interface Sci 1999;81:167.
[25] Lin T-S, Wu C-F, Hsieh C-T. J Vac Sci Technol, B : Microelectron Nanometer Struct
2006;24:855.
[26] Liu S, Zhang C. J Phys Conf Ser 2009;188:012026.
[27] Liu T, Chen S, Cheng S, Tian J, Chang X, Yin Y. Electrochim Acta 2007;52:8003.
[28] Manoudis PN, Karapanagiotis I, Tsakalof A, Zuburtikudis I, Panayiotou C.
Langmuir 2008;24:11225.
[29] Mao C, Liang C, Luo W, Bao J, Shen J, Hou X, et al. J Mater Chem 2009;19:9025.
[30] Minko S, Usov D, Goreshnik E, Stamm M. Macromol Rapid Commun 2001;22:
206.
 A.J.B. Milne, A. Amirfazli / Advances in Colloid and Interface Science 170 (2012) 48–55 55

[31] Narhe R, Beysens D. Phys Rev Lett 2004;93:076103. [80] Caputo G, Cortese B, Nobile C, Salerno M, Cingolani R, Gigli G, et al. Adv Funct
[32] Patankar NA. Langmuir 2010;26:8783. Mater 2009;19:1149.
[33] Patra S, Sarkar S, Bera SK, Ghosh R, Paul GK. J Phys D: Appl Phys 2009;42:075301. [81] Chou T-H, Hong S-J, Sheng Y-J, Tsao H-K. J Phys Chem B 2010;114:7509.
[34] Ralston J, Popescu M, Sedev R. Annu Rev Mater Res 2008;38:23. [82] Vrancken RJ, Kusumaatmaja H, Hermans K, Prenen AM, Pierre-Louis O, Bastiaansen
[35] Ramaratnam K, Iyer SK, Kinnan MK, Chumanov G, Brown PJ, Luzinov I. J Eng CWM, et al. Langmuir 2010;26:3335.
Fibers Fabr 2008;3:1. [83] Cao M, Song X, Zhai J, Wang J, Wang Y. J Phys Chem B 2006;110:13072.
[36] Rogacs A, Steinbrenner JE, Rowlette JA, Weisse JM, Zheng XL, Goodson KE. J [84] Chen M, Cao B-Y, Guo Z-Y. Proceedings of the sixth international ASME confer-
Colloid Interface Sci 2010;349:354. ence on nanochannels, microchannels and minichannels ICNMM2008, ASME
[37] Salerno MB, Logan BE, Velegol D. Langmuir 2004;20:10625. 2008 6th international conference on nanochannels, microchannels, and mini-
[38] Shaohua L, Xiang Z, Qihong L, Tianmin T, Zhiqi X. Lasers Mater Process Manuf channels (ICNMM2008), Darmstadt, Germany, 23–25 June 2008. ASME; 2008.
2002;4915:272. p. 1661.
[39] Shirwaiker RA, Wysk RA, Kariyawasam S, Carrion H, Voigt RC. Biofabrication [85] Cheng YH, Chou CK, Chen C, Cheng SY. Chem Phys Lett 2004;397:17.
2011;3:015003. [86] Cheng Y-T, Rodak DE. Appl Phys Lett 2005;86:144101.
[40] Takeda K, Nakajima A, Hashimoto K, Watanabe T. Surf Sci 2002;519:L589. [87] Cheng Z, Feng L, Jiang L. Adv Funct Mater 2008;18:3219.
[41] Tavana H, Neumann AW. Adv Colloid Interface Sci 2007;132:1. [88] Detry JG, Sindic M, Deroanne C. Crit Rev Food Sci Nutr 2010;50:583.
[42] Tsai P, Lammertink RGH, Wessling M, Lohse D. Phys Rev Lett 2010;104:116102. [89] Doshi DA, Shah PB, Singh S, Branson ED, Malanoski AP, Watkins EB, et al.
[43] Vafaei S, Wen D. Adv Colloid Interface Sci 2010;159:72. Langmuir 2005;21:7805.
[44] Waugh DG, Lawrence J. Opt Lasers Eng 2010;48:707. [90] Fresnais J, Chapel JP, Benyahia L, Poncin-Epaillard F. J Adhes Sci Technol 2009;23:
[45] Wu AHF, Cho KL, Liaw II, Moran G, Kirby N, Lamb RN. Faraday Discuss 2010;146: 447.
223. [91] Hiroyuki S, Ken-ichi T, Goro Y. Mater Sci Res Int 1997;3:185.
[46] Yanagisawa T, Nakajima A, Sakai M, Kameshima Y, Okada K. Mater Sci Eng: B [92] Hong YC, Shin DH, Uhm HS. Surf CoatTechnol 2007;201:5025.
2009;161:36. [93] Hou S, Li X, Li X, Feng X. Biofabrication 2009;1:035004.
[47] Yun Y, Shanov VN, Balaji S, Schulz MJ, Tu Y, Yarmolenko S, et al. Proceedings [94] Hsieh CT, Wu FL, Chen WY. Surf CoatTechnol 2009;203:3377.
of SPIE, Smart Structures and Materials 2006: Sensors and Smart Structures [95] Kim D, Kim J, Park HC, Lee KH, Hwang W. J Micromech Microeng 2008;18:
Technologies for Civil, Mechanical, and Aerospace Systems, San Diego, CA, USA, 015019.
27 February 2006; 2006. p. 61743Z. [96] Kim MJ, Park J-E, Song S, Lee HH. J Vac Sci Technol B: Microelectron Nanometer
[48] Zhang H, Lamb RN, Cookson DJ. Appl Phys Lett 2007;91:254106. Struct 2007;25:1412.
[49] Zhang J, Sheng X, Jiang L. Langmuir 2009;25:1371. [97] Lai KY, Lin Y-R, Wang H-P, He J-H. CrystEngComm 2011;13:1014.
[50] Zhang J, Mohammadi R, Kwok DY. ASME 3rd International Conference on Micro- [98] Li B, Zhou M, Yuan R, Cai L. J Mater Res 2011;23:2491.
channels and Minichannels, Part B cont'd, ASME 3rd International Conference on [99] Lim J-M, Yi G-R, Moon JH, Heo C-J, Yang S-M. Langmuir 2007;23:7981.
Microchannels and Minichannels (ICMM2005), Toronto, Ontario, 13–15 June [100] Nakae H, Inui R, Hirata Y, Saito H. Acta Mater 1998;46:2313.
2005; 2005. p. 523. [101] Nakajima A, Abe K, Hashimoto K, Watanabe T. Thin Solid Films 2000;376:140.
[51] Z-guo Zhou, Z-wen Liu. J Bionic Eng 2009;6:1. [102] Sun X, Tao R, Lin L, Li Z, Zhang Z, Feng J. Appl Surf Sci 2011;257:3861.
[52] Zou S, Ma Y, Hempenius MA, Schönherr H, Vancso GJ. Langmuir 2004;20:6278. [103] Wang C, Song Y, Zhao J, Xia X. Surf Sci 2006;600:38.
[53] Bormashenko E. Langmuir 2009;25:10451. [104] Xu J, Li M, Zhao Y, Lu Q. Colloids Surf A Physicochem Eng Asp 2007;302:136.
[54] Cubaud T, Fermigier M. J Colloid Interface Sci 2004;269:171. [105] Zhu M, Gao H, Li H, Xu J, Chen Y. J Solid State Chem 2010;183:595.
[55] Drelich J, Miller JD, Good RJ. J Colloid Interface Sci 1996;179:37. [106] Johnson RE, Dettre RH. In: Fowkes FM, editor. Contact angle, wettability, and
[56] Drelich J, Miller JD. Langmuir 1993;9:619. adhesion, vol. 43. American chemical society; 1964.
[57] Gotoh K, Yasukawa A, Taniguchi K. J Adhes Sci Technol 2011;25:307. [107] Marmur A, Bittoun E. Langmuir 2009;25:1277.
[58] Hey MJ, Kingston JG. Chem Phys Lett 2007;447:44. [108] Mohammadi R, Wassink J, Amirfazli A. Langmuir 2004;20:9657.
[59] Ito Y, Heydari M, Hashimoto A, Konno T, Hirasawa A, Hori S, et al. Langmuir [109] Antonini C, Innocenti M, Horn T, Marengo M, Amirfazli A. Cold Reg Sci Technol
2007;23:1845. 2011;67:58.
[60] Pignatelli F, Carzino R, Salerno M, Scotto M, Canale C, Distaso M, et al. Thin Solid [110] Tavana H, Amirfazli A, Neumann AW. Langmuir 2006;22:5556.
Films 2010;518:4425. [111] Milne AJB, (MSc Thesis) Clean Surfaces, Dirty Water: Topography and Chemistry
[61] Rinastiti M, Ozcan M, Siswomihardjo W, Busscher HJ. J Dent 2010;38:29. in the Wetting of Superhydrophobic Surfaces by Pure Liquids and Surfactant
[62] Shovsky A, Schönherr H. Langmuir 2005;21:4393. Solutions. University of Alberta; 2008.
[63] Silberzan P, Leger L, Ausserre D, Benattar JJ. Langmuir 1991;7:1647. [112] Bico J, Thiele U, Quéré D. Colloids Surf A Physicochem Eng Asp 2002;206:41.
[64] Sohn EH, Kim BG, Chung JS, Lee JC. J Colloid Interface Sci 2010;343:115. [113] Bico J, Marzolin C, Quéré D. Europhys Lett 1999;47:220.
[65] Zhang H-L, Chen M, Li H-L. J Phys Chem B 2000;104:28. [114] Daub CD, Wang J, Kudesia S, Bratko D, Luzar A. Faraday Discuss 2010;146:67.
[66] Elowson AM. The Auk 1984;101:371. [115] Wang J, Bratko D, Luzar A. Proc Natl Acad Sci 2011;108:6374.
[67] Holdgate MW. J Exp Biol 1955;32:591. [116] Blanco-Gomez G, Flendrig LM, Cooper JM. Langmuir 2010;26:7248.
[68] Kawase T, Fujii T, Minagawa M. Text Res J 1987;57:185. [117] Chen W, Fadeev AY, Hsieh MC, Öner D, Youngblood J, McCarthy TJ. Langmuir
[69] Kawase T, Misono K, Fuj T, Minagawa M. Text Res J 1990;60:345. 1999;15:3395.
[70] Marmur A. Adv Colloid Interface Sci 1994;50:121. [118] Spori DM, Drobek T, Zürcher S, Spencer ND. Langmuir 2010;26:9465.
[71] Morra M, Occhiello E, Garbassi F. Langmuir 1989;5:872. [119] Henderson JR. Faraday Discuss 2010;146:79.
[72] Raibeck L, Reap J, Bras B. Proceedings of the ASME 2008 international design [120] Pease DC. J Phys Chem 1945;49:107.
engineering technical conferences & computers and information in engineering [121] Extrand CW. Langmuir 2003;19:3793.
conference IDETC/CIE 2008, Brooklyn, New York, 3–6 August 2008; 2008. [122] Gao L, McCarthy TJ. Langmuir 2007;23:3762.
[73] Bhushan B, Chaejung Y. Ultramicroscopy 2007;107:1033. [123] McHale G. Langmuir 2007;23:8200.
[74] Cha T-G, Yi JW, Moon M-W, Lee K-R, Kim H-Y. Langmuir 2010;26:8319. [124] Nosonovsky M. Langmuir 2007;23:9919.
[75] Gostick JT, Fowler MW, Ioannidis MA, Pritzker MD, Volfkovich YM, Sakars A. J [125] Yang C, Tartaglino U, Persson B. Phys Rev Lett 2006;97.
Power Sources 2006;156:375. [126] Yang C, Tartaglino U, Persson BNJ. Eur Phys J E 2008;25:139.
[76] Im M, Im H, Lee J-H, Yoon J-B, Choi Y-K. Langmuir 2010;26:17389. [127] Marmur A. Langmuir 2003;19:8343.
[77] Lawrence J. Proc R Soc Lond A Math Phys Eng Sci 2004;460:1723. [128] Herminghaus S. Europhysics Letters 2000;52:165.
[78] Triantafyllidis D, Li L, Stott FH. Mater Sci Eng A 2005;390:271. [129] Öner D, McCarthy TJ. Langmuir 2000;16:7777.
[79] Zhu L, Xu J, Xiu Y, Sun Y, Hess DW, Wong C-P. J Phys Chem B 2006;110:15945.