Chemical Engineering Journal 227 (2013) 2–12

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Journal

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

Surface roughness of machined microchannels and its effect on

multiphase boundary conditions
Kristin Hecht a,⇑, Florian Messerschmidt a, Peter Pfeifer a, Roland Dittmeyer a,
Bettina Kraushaar-Czarnetzki b, Scott Hecht c
a
Institute for Micro Process Engineering (IMVT), Karlsruhe Institute of Technology (KIT), Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany
b
Institute of Chemical Process Engineering (CVT), Karlsruhe Institute of Technology (KIT), Fritz-Haber-Weg 2, 76131 Karlsruhe, Germany
c
Western Engineering, Inc., 1110 Road 9, Powell, WY 82435, USA

h i g h l i g h t s

" Machined metal surfaces are smoother than plastic surfaces.

" Machined surfaces are as good as polished based on the contact angle.
" The contact angle for any material can predicted with the suggested correlation.

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

Article history: The surface roughness is an important boundary condition needed for a detailed understanding and mod-
Available online 7 December 2012 eling of multiphase flow in microchannels. In this work the surface roughness and contact angles for the
conditions present in machined microchannels were measured for various materials. Surface roughness
Keywords: values are reported. Although the machined microchannels have a higher surface roughness than pol-
Multiphase flow ished surfaces, the resulting contact angles are very similar. A correlation has been developed that allows
Contact angle for the prediction of the contact angle from the liquid surface tension, volume of the liquid drop, and sur-
Surface roughness
face roughness.
Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction
Surface roughness impacts the surface area and friction occur-
ring at the interface of the wall and gas or liquid phase. This
parameter is of general interest for its effect on flow patterns and
pressure drop [1–3]. Microstructured elements can be fabricated
through numerous methods and in a wide variety of materials
[4,5]. The effect of the wall surface on flow is especially important
for devices with high surface area to volume ratios such as
microreactors.
The phase interface is especially critical to the performance of
microstructured devices for multiphase reactions such as gas/li-
quid microreactors. Experimental observations as well as simula-
tions have demonstrated that the three-phase boundary
dramatically influences the pressure drop and flow patterns of
gas/liquid microreactors. This interaction is typically quantified
according to the material contact angle, which combines a number
of factors such as surface roughness and material wettability into a
single boundary condition [6–11].
⇑ Corresponding author.
E-mail address: kristin.hecht@gmail.com (K. Hecht).
The Institute for Micro Process Engineering (IMVT) at the Kar-
lsruhe Institute of Technology (KIT) is active in investigating
microstructured devices for industrial use. Devices are designed
and fabricated in house, primarily through machining. To deter-
mine the influence of solid surface properties on gas/liquid flows,
samples of different metal and polymer materials were structured
through different processes. SEM (scanning electron microscopy)
images of the samples were collected. The surface roughness of
the samples was investigated with AFM (atomic force microscopy),
and the contact angle was measured on the prepared samples for a
range of solid/liquid/gas systems.
The measurements of surface roughness provide absolute
values for the surface roughnesses resulting from the polishing,
milling, and grinding of various plastics and metals. Some mea-
surements of surface roughnesses, for example of extruded pipes,
can be found in literature [12,13]. Microstructured reactors have
been rapidly gaining in significance, so estimates of the surface
roughnesses encountered in typical microstructured reactors are
needed. Contact angle measurements are frequently carried out
on ideal surfaces such as hydrophobic fluorinated polymers that
lend themselves well to such measurements [14,15]. From an engi-
neering standpoint, such ideal surfaces are rarely encountered.

1385-8947/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cej.2012.11.127
 K. Hecht et al. / Chemical Engineering Journal 227 (2013) 2–12
Nomenclature
Acronyms
PEEK polyether ether ketone
PMMA poly(methyl methacrylate)
PTFE polytetrafluoroethylene
UHMW PE ultra high molecular weight polyethylene
Symbols
f fraction of a surface consisting of a given component
g gravitational acceleration (m2/s)
h height (m)
P pressure (Pa)
r radius (m)
rw roughness factor used in Wenzel equation
Ra average roughness (nm)
Rmax maximum height variation between lowest and highest
point (nm)
Practical surfaces refer to surfaces as they are practically found, for
example metals typically absorb water molecules from the atmo-
sphere onto their surfaces [16]. Measurements on practical sur-
faces have proven challenging due to the wide variation
encountered in the results. In this work the effect of surface ten-
sion, pressure, surface roughness, and drop size have been investi-
gated for their effect on the contact angle. Statistical regression
analysis has then been applied to develop a rough correlation
which can be used to determine the contact angle when a direct
measurement thereof is not available or to anticipate the effect
of changes in surface tension, drop size, or surface roughness on
the resulting contact angle.
2. Materials and methods
2.1. Preparation of solid surfaces
The effect of solid surface characteristics on the contact angle
was investigated for the following materials:
 Glass (Menzel-Gläser microscope slides; Germany).
 Silicon wafer (n-type; <1-0-0>; Silicon Quest Intl.; USA).
 Nylon (Nylatron GS rod; Quadrant EPP USA, Inc.; Pennsylvania,
USA).
 PEEK (polyether ether ketone; Ensinger Tecapeek; Germany).
 PMMA (acrylic/poly(methyl methacrylate); Alro Plastics; USA).
 Polystyrene (Goodfellow ST313200; Germany).
 PTFE (polytetrafluoroethylene; Interstate Plastics; California,
USA).
 UHMW PE (ultra high molecular weight polyethylene; Tivar;
Quadrant EPP USA, Inc.; Pennsylvania, USA).
 8620 Alloy steel (nickel–chromium–molybdenum alloy; ASTM
A322 ASTM A304 steel; Magellan; China).
 Aluminum (Alloy 6081; Alaskan Copper & Brass Company;
USA).
 Bronze (660 bearing bronze; K.P. Bronze; Canada).
 Cast iron (G2 gray iron; Dura-bar Metal Services; Illinois, USA).
 Stainless steel (T-304/304-L; Viraj Impoexpo; India) (polished
and machined samples).
Two additional stainless steel samples were also investigated.
The manufacturer and exact type of stainless steel were not
known. The polished stainless steel sample from [6] was used in
the contact angle measurements, and the ground stainless steel
sample used for both the roughness and contact angle investiga-
3
Rq root mean square roughness (nm)
xCO2 mole fraction of CO2
Zi height of a given point (nm)
Greek Symbols
a fitting parameter (1/mm)
h contact angle (°)
h1 contact angle of an infinitely large drop (°)
h⁄ apparent contact angle (°)
q density (kg/m3)
r interfacial tension (mN/m)
rgl gas/liquid interfacial tension (mN/m)
rsg solid/gas interfacial tension (mN/m)
rsl solid/liquid interfacial tension (mN/m)
tions was prepared using steel from an unknown source. Both of
these samples are nickel–chromium steels; their compositions
have been determined with energy-dispersive X-ray spectroscopy
(Jeol JXA-8530F field emission electron probe). The results of the
elemental analysis are given in Table 1.
Samples from the various materials were prepared in three dif-
ferent ways to obtain different roughness: grinding, polishing, and
milling in the same manner as microchannels are machined at the
IMVT. Ground samples were prepared by grinding the materials
with 320 grit silicon carbide paper at 120 rpm on an ATM Saphir
550. To obtain polished samples, the ground samples were consec-
utively ground with 320, 600 and 1200 grit grinding papers and
then polished with a 3 micrometer diamond suspension (ATM
92002593) and a synthetic cloth at 100 rpm. The machined sam-
ples, except the PEEK sample, were milled (EMCO FB-3, dry, with-
out cooling) with an HSS shell end mill (35 mm diameter, 160/min,
150 mm/min). The PEEK machined sample was milled at a slightly
higher speed (245/min). The polishing process for the stainless
steel sample from [6] is unknown, but its surface roughness was
approximately comparable to the stainless steel 304 sample pre-
pared as specified (Rq = 18 nm; Ra = 14 nm; Rmax = 126 nm, surface
area difference = 0.099%).
2.2. Gases and liquids
Nitrogen (Air Liquide N50, >99.999%), carbon dioxide (Basi 4.5,
>99.995%), and 10% carbon dioxide in argon (Messer Griesheim
GmbH CO2 4.5/N2 5.0, 10.2% CO2) gases as well as the ambient
air atmosphere were used for measurements. The gases investi-
gated in this work were chosen because they relate to gas/liquid
systems that have been used to investigate microstructured de-
vices elsewhere [17,18]. Deionized water, glycerine (VWR
24397.467, >97%), ethanol (VWR 20821.296, >99.9%), isopropanol
(Merck 1.09634.1011, >99.8%), and mixtures of water and ethanol
were measured. All measurements were completed at ambient
temperature. The liquids may contain water absorbed from the
atmosphere. Water, ethanol, or isopropanol vapor may also be
present in the gas atmosphere when these liquids are measured.
2.3. SEM photographs
Secondary electron (SE) images of contact angle surface samples
were collected using a scanning electron microscope (SEM) (Jeol
JSM-6300). Nonconductive surfaces were sputtered with 10 nm
thick layers of platinum prior to measurement.
4 K. Hecht et al. / Chemical Engineering Journal 227 (2013) 2–12

Table 1 Contact angles were measured using the sessile drop method.
Composition of stainless steel samples from unknown sources. The apparatus consists of an optical cell (SITEC AG, Maur/Zürich,
Element Polished sample Ground sample 740.2086) with sapphire windows for observation of the droplet,
from [6] (mass%) (mass%) gas introduction with pressure regulation (Rotarex/SMT, Genlis,
Carbon 0.63 ± 0.13 0.55 ± 0.15 France, SL225-16), and liquid dosing through means of a hand-
Silicon 0.36 ± 0.05 0.26 ± 0.05 operated pump (SITEC AG, Maur/Zürich, 750.1400). An Imaging
Phosphorus 0.08 ± 0.03 0.09 ± 0.05 Source C3516-M(KP) (35 mm focal length) lens with 5 mm and
Chromium 18.6 ± 0.4 18.8 ± 0.4
Manganese 2.0 ± 0.2 1.9 ± 0.2
10 mm extension rings was used with a SONY AVC-D5CE CCD cam-
Iron 70.3 ± 0.5 70.2 ± 0.1 era. Light (HLV-24SW-NR-3W, CCS Inc., Japan) was passed through
Nickel 8.0 ± 0.6 8.2 ± 0.2 a series of lenses to create a parallel beam. Video frames collected
Sulfur/molybdenum Not detected 0.03 ± 0.01 from the apparatus were analyzed with the DropSnake plugin for
Image J [19] and Dataphysics SCA20.
The liquid sample was loaded into the pump and pumped to fill
the tubing. Solid samples were cleaned with water and isopropanol
2.4. Surface roughness measurements
and dried with compressed air and placed in the measurement
chamber. The measurement chamber was purged for several min-
AFM measurements of the surfaces of the samples prepared for
utes with the gas atmosphere to be measured before finally adjust-
contact angle measurements were performed with a Veeca AFM.
ing the final pressure. After adjusting the pressure to be measured,
The collected data was analyzed with Veeco NanoScope Analysis
a liquid drop was gently dropped onto the solid surface. A video of
1.10. PEEK and PTFE surfaces were sputtered with a 10 nm thick
the placement of first drop was recorded for one minute and
layer of platinum to prevent the surface from collecting a static
the contact angle over time was analyzed. The contact angle
charge that would interfere with the measurement. A cantilever
reached a stable value as soon as the drop settled onto the surface.
(Nanosensors Pointprobe Plus PPP-LFMR; Budget Sensors AIO)
In measurements contaminated with oil, the contact angle was
was used to scan the surface in contact mode or tapping mode.
observed to change over time; if contamination was observed,
Scan areas varied and could be square (37%), a rectangle with an
the experimental apparatus and sample were cleaned prior to
aspect ratio of 10:1 (52%), or be thin with a width of 1 lm (11%).
remeasurement.
For polished samples the longest dimension was at least 10 lm.
Surface tension was measured using the pendant drop method
For ground and milled samples the longest measurement dimen-
with the same apparatus as the contact angle measurements.
sion was at least 50 lm, and all measurements on these surfaces
Drops large enough to be deformed by gravity were freely sus-
were carried out such that the AFM cantilever traversed the burrs
pended from the syringe tip. The liquid surface tension was calcu-
left from the tool to give the maximum roughness value. If only a
lated from axisymmetric drop analysis using Dataphysics SCA20.
1 lm slice was scanned, the measurement was repeated at least
three times at different locations on the sample.
The collected profiles were adjusted for sample tilt through lin-
3. Results and discussion
ear regression assuming that the roughness manifests itself as scat-
ter around a line representing the position of the surface. This line
3.1. Effect of milling, polishing, and grinding on the surface roughness
represents a height of 0, and the height of each measured point (Zi)
of various materials
can have either a positive or negative value with respect to this
line. The roughness is reported in several forms: the root mean
SEM photos of stainless steel and PTFE samples in the milled
square roughness (Rq) (Eq. (1)); the average roughness (Ra) (Eq.
and polished states are shown in Fig. 1. Despite the materials being
(2)); the maximum height variation between the lowest point
prepared in the same manner, it can be seen that the resulting sur-
and the highest point (Rmax); the surface area difference, which is
faces differ. The machined grooves on the PTFE surface appear
the percentage increase in the surface area due to the actual sur-
wider than those of the milled stainless steel sample. Such pictures
face topography. The surface roughness values Rq and Ra were de-
provide an impression of the nature of the surfaces studied.
fined according to the number of measurement points (N) and the
To obtain detailed values for the roughness of the surface, a pro-
height measured at each point (Zi).
file or surface contour map was obtained from AFM measurements
sffiffiffiffiffiffiffiffiffiffiffi
P 2 of surfaces. AFM images were flattened to adjust for tilt in the sam-
Zi
Rq ¼ ð1Þ ple. The roughness values determined for all measured surfaces are
N presented in Table 2. It should be noted that the milled surfaces
were produced under suboptimal conditions (lack of cooling, high
1X N
speed) and that the microchannels produced at the IMVT under
Ra ¼ jZ i j ð2Þ
N i¼1 optimal conditions would have a lower surface roughness.
The photographs from the AFM have a lower resolution than the
SEM photos, but the features are the same. The AFM scans covered
2.5. Contact angle and surface tension measurements as large of a scan area as possible, typically slightly less than
100 lm. The differences between the results of the preparation
The contact angle is the angle formed between the solid–liquid methods on the different materials can be examined in Fig. 2,
and liquid–gas interfaces of a drop of liquid resting on a solid sur- which shows examples of three-dimensional renderings of the sur-
face. The contact angle is described according to Young’s equation: faces from the AFM for the examples of stainless steel and PTFE.
rsg ¼ rsl þ rgl  cosðhÞ ð3Þ In general the metal samples are much smoother than the poly-
mer materials prepared in the same way. Additionally, the general
where rsg is the solid/gas interfacial tension; the term rsl is the nature of the surface also differs. The machined metal and polymer
interfacial tension of the solid–liquid interface; the term rgl is the samples look similar to the eye. Both types of samples contain a
gas/liquid interfacial tension; and h is the contact angle. The term number of parallel grooves, but under the AFM or microscope,
interfacial tension is used interchangeably with the term surface the small ridges resulting from machining are much sharper for
energy and has units of mN/m. metals and more round for polymers. Harder materials are able
 K. Hecht et al. / Chemical Engineering Journal 227 (2013) 2–12
(a)
(b)
Fig. 1. SEM photos of a milled (a) and polished (b) stainless steel surface and milled (c) and polished (d) PTFE surface.
to hold an edge better than softer materials. Harder materials are
likewise, as seen here, able to be polished to higher degree than
soft materials.
The average roughness (Ra) for the polished, milled, and ground
surfaces is plotted against the Brinell material hardness in Fig. 3
and material density in Fig. 4. Generally, the resulting roughness
shows an inversely proportional trend to both material properties;
however, there is a large variation in the resulting roughness for
materials with low Brinell hardness and density values. The trend
compared to material tensile strength is similar to that of hardness
and density. For every material the polished surface roughness is
substantially less than the roughness resulting from milling or
grinding. Comparing the roughnesses achieved for a particular
material through grinding and milling, neither of the methods con-
sistently produced a smoother surface than the other. For hard
metals, ground surfaces were smoother than milled surfaces; in
some cases, the milled surfaces of hard metals were too rough to
be measured with AFM. However, this trend cannot be generalized
for softer materials such as plastics.
Rq, Ra, and Rmax are all nearly equivalent measures of roughness
irrespective of the preparation method of the samples. The surface
area difference percentage differs from these other values because
it depends on the shape of the surface roughness and not only on
the height difference between peaks and valleys.
No general correlation between surface roughness and material
properties was found. Polished surfaces were much smoother for
5
(c)
(d)
metals than for plastics. Grinding and machining produced sur-
faces with similar surface roughness values. The surface roughness
of plastics were higher and more spread out, and metal surfaces
were in general smoother. A summary of the average roughness
values categorized according to plastic or metal and surface treat-
ment is given in Table 3.
3.2. Gas/liquid/solid contact angles
3.2.1. Effect of pressure on the contact angle
Wesch et al. [20] found that the contact angle of water on var-
ious surfaces increased with pressure for pressures of up to
30 MPa. Fig. 5 shows the effect of pressure on drops of various sizes
placed into environments with a fixed pressure. These investiga-
tions revealed no substantial contribution of pressure to the con-
tact angles of drops formed at a fixed pressure. For pressures up
to 1000 kPa, the contact angle can be regarded as being indepen-
dent of the formation pressure, i.e. the results did not show an ef-
fect of pressure on the contact angle but seem to indicate a strong
influence of drop size on the measured angle.
3.2.2. Effect of drop size on the contact angle
Ponter and Yekta-Fard [21] have provided an extensive review
of literature and their own work where the drop size exhibits a
strong influence on the contact angle. In some systems (water/
polyethylene/water-saturated air, water/PMMA/water-saturated
6 K. Hecht et al. / Chemical Engineering Journal 227 (2013) 2–12

Table 2
Surface roughness determined with AFM.

Material Polished Ground Milled

Rq Ra Rmax Surface area Rq Ra Rmax Surface area Rq Ra Rmax Surface area
(nm) (nm) (nm) difference (%) (nm) (nm) (nm) difference (%) (nm) (nm) (nm) difference (%)
Metal
8620 11 8 106 0.0647 142 109 926 0.95 Too rough to measure
Aluminum 14 86 853 0.997 Not available 212 180 1076 1.79
Bronze 14 11 97 0.139 314 258 1608 2.38 162 135 697 0.693
Cast iron 20 12 623 0.188 215 175 1070 1.92 Too rough to measure
L80 4 3 29 0.118 141 111 854 1.07 Too rough to measure
Stainless 13 9 234 0.0184 111 85.7 666 0.688 230 200 920 0.583
steel

Plastic
Nylon 86 64 989 0.805 Not available 655 519 2877 1.52
PEEK 25 16 459 0.0906 343 262 1788 4.47 776 642 3486 1.43
PMMA 44 34 452 0.724 685 548 4094 15 703 455 4921 8.81
Polystyrene 20 16 129 0.0191 1353 979 8211 20.9 731 625 3135 7.53
PTFE 102 79 747 1.01 887 686 6043 8.47 1054 829 7406 4.83
UHMW PE 284 229 1453 2.18 559 447 3125 5.24 1076 867 9807 14.4

Glass slide 1.3 0.79 50 0.04

Silicon 0.97 0.40 33 0.000846
wafer

nitrogen, water/PMMA/water-saturated air, water/stainless steel/
water-saturated air, water/PTFE/water-saturated nitrogen, water/
FEP/water-saturated air, ethylene glycol/FEP/water-saturated air)
the contact angle increased with increasing drop radius. Other sys-
tems (boiling point benzene/PTFE/benzene vapor, boiling point
water/PTFE/water vapor, boiling point water/copper/water vapor)
show the reverse trend. Most of the curves leveled-off at a drop
diameter of around 0.5–1 cm.
The results shown in Fig. 5 show only an increase in contact an-
gle with increasing drop size with an eventual leveling-off occur-
ring when the drops reach a certain size, typically a drop base
radius of around 2 mm.
Gravity can deform the shape of a drop and influence the con-
tact angle when the drop is more than several millimeters large.
The Laplace pressure describes the pressure difference (DP) at a
meniscus or phase interface:
2r
DP ¼ Pinside  P outside ¼ ð4Þ
r Results are shown in Fig. 6. The slope of this relationship is
Here, Pinside refers to the pressure on the inside of the drop, Poutside is
the pressure on the outside of the drop, r is the interfacial tension,
and r is the drop radius.
When the drop and interface are additionally influenced by
gravity, the Laplace–Young equation takes the typical form:
 
1 1
DP ¼ qgh  r þ ð5Þ
r1 r2
where q is the density, g the gravitational acceleration, h the height
of the drop, and r1 and r2 are the radii of curvature describing the
drop shape flattened by gravity.
The application of the Laplace–Young equation requires the
drop to be deformed by gravity. The Dataphysics SCA20 software
can measure the contact angle either according to the Laplace–
Young equation or by fitting it to a spherical profile. Small drops
were typically better fit by a spherical profile, whereas drops de-
formed by gravity were better fit by the profile calculated accord-
ing to the Laplace–Young equation. The contact angle needed for
the Young equation is a balance of the various interfacial tensions
as the Laplace–Young equation is a balance of the interfacial ten-
sion and gravity. The contact angle needed for the Young equation
is therefore the angle that is constant for larger drop sizes.
The relationship between contact angle and the radius of the
drop base was fitted using a Langmuir-type linear regression ap-
proach (Eq. (6)). This type of nonlinear regression is able to describe
the curves seen in Fig. 5, where the contact angle approaches a con-
stant value as the drop size increases. The equation uses only two
parameters: ‘h1’ to describe the constant value of the contact angle
for an infinitely large drop size and ‘a’ to describe the slope or rate
at which the values approach the constant value.
ar
h¼ h1 ð6Þ
1þar
Eq. (6) can be linearized in order to determine the parameters ‘a’
and ‘h1’ from the measured data through linear regression (Eq.
(7)). The parameter h1 represents the constant value of the contact
angle reached for an infinitely large drop; a is a fitting parameter
r 1 1
¼ rþ ð7Þ
h h1 a  h1
equal to the inverse of the finally constant contact angle. The con-
tact angle values determined from the slope are identical for water
on stainless steel in both nitrogen and carbon dioxide atmospheres
with a h1 value of 91.7°. This is several degrees higher than the
average value of drops with bases of radius 2 mm or more shown
in Fig. 5.
In summary, the drop volume influences the shape and contact
angle of the single drops. Small drops assume spherical profiles.
The contact angle for the Young equation, Eq. (3), can be obtained
by analyzing drops of different sizes with a Langmuir linear regres-
sion or from single measurements as long as the drop profiles are
large enough to be deformed by gravity and fitted by the Laplace–
Young relationship given in Eq. (5).
3.2.3. Effect of surface tension on the contact angle
Mixtures of ethanol and deionized water were used for measur-
ing the contact angle for a range of surface tensions. Isopropanol,
glycerine, and pure water were also measured in some cases. The
surface tension was measured for mixtures of 51.3 ± 0.2 wt.%,
20.4 ± 0.2 wt.%, and 6.0 ± 0.2 wt.% ethanol. The measurements of
the surface tension for these mixtures in air, nitrogen, carbon diox-
ide, and 10% carbon dioxide in argon are shown in Fig. 7. The gas
atmosphere did not affect the measured interfacial tension to any
 K. Hecht et al. / Chemical Engineering Journal 227 (2013) 2–12 7

Fig. 2. Renderings of polished, machined, and ground PTFE and stainless steel surfaces from AFM data.

Fig. 3. Surface roughness dependence on material hardness. Fig. 4. Surface roughness dependence on material density.

perceptible extent. The measurements of the mixtures fit well the
experimental observation of Belda et al. [22]. The measured values
of pure ethanol may be higher than expected due to limitations of
the measurement method for liquids with low surface tension
where the formation of drops sufficiently large to be deformed
by gravity is not always possible. Surface tension values for isopro-
panol (21.7 mN/m) and glycerine (63 mN/m) were taken from lit-
erature [23].
According to the Young equation (Eq. (3)), the cosine of the con-
tact angle should be inversely proportional to the liquid/gas sur-
face tension. Fox and Zisman reported in a 1950 publication that
the cosine of the contact angle is also linearly proportional to the
surface tension [14]. Plots of the cosine of the contact angle against
the surface tension (Zisman plot) are shown for selected polished
surfaces in Fig. 8. Plots of the cosine of the contact angle against
the inverse of the surface tension (Young equation) are shown in
Fig. 9. Both of these representations appear to be approximately
linear, but the relationship according to the Young equation does
seem to introduce some curvature deviation from the linear pre-
dictions as the small contact angles lead to cosine values near
unity. The variability in the measurements is quite large. The mea-
surements with water (rgl = 72 mN/m) show a particularly large
8 K. Hecht et al. / Chemical Engineering Journal 227 (2013) 2–12
Table 3
Average surface roughness values for metals and plastics.
Material Surface treatment Ra (nm)
Metal Polished 22 ± 32
Machined or ground 157 ± 57
Plastic Polished 73 ± 81
Machined or ground 606 ± 226
Fig. 5. Contact angle variation with drop size at different formation pressures for water in N2 and CO2 on polished stainless steel.
variation. Aside from the statistical variation of repeated measure-
ments, there are a very large number of confounding variables that
affect contact angle measurements. For example, this variability
could be a result of drop size variation; drops smaller than 8.3 lL
have been excluded from these graphs, but the drops still vary in
size. The variability of the water measurements is not as large on
the PTFE and PEEK surface, but it is very large on stainless steel
and glass surfaces. Stainless steel and glass readily absorb water,
which could change the state of the surface being measured [24].
Most rigorous studies of wettability focus on hydrophobic surfaces
because their measurements are much more reproducible.
Although the surfaces here are not ideal, they represent real condi-
tions that might be encountered in a microreactor.
3.2.4. Effect of gas composition on the contact angle
The measurements shown in Fig. 8 separate the measurements
according to the gas phase in which the liquid was measured.
According to the Young equation, all three phases, gas, liquid,
and solid, are important in the determination of the contact angle.
It appears that the composition of the gas phase has a negligible
impact on the contact angle.
3.2.5. Effect of surface roughness on the contact angle
The cosine of the contact angle for different roughnesses of the
materials is plotted against the inverse of the surface tension
according to their relationship in Young’s equation in Fig. 9. De-
spite the scatter in the measurements, these graphs show an im-
pact of surface roughness on the contact angle that becomes
more significant for larger angles. The effect is insignificant for
small angles (cos(h) ? 1).
For PEEK, PTFE, and stainless steel, the contact angles measured
on the polished and milled samples are nearly identical. As the sur-
face roughness investigations revealed, it is not possible to simply
compare machined surfaces to each other since the processing
technique impacts each material differently. For example, Rq, Ra,
and Rmax for PEEK indicate that the milled PEEK sample is rougher
than the ground PEEK sample by nearly a factor of two, but the sur-
face area difference indicates just the opposite; the surface area in-
crease for the milled PEEK sample is 1.43% and for the ground PEEK
sample 4.47%. For PTFE the milled sample likewise has higher
roughness values in terms of Rq, Ra, and Rmax compared to the
Rq (nm) Rmax (nm) Surface area difference (%)
34 ± 54 324 ± 336 0.3 ± 0.4
191 ± 65 977 ± 296 1.3 ± 0.7
93.4 ± 99 705 ± 469 0.8 ± 0.8
774 ± 292 4751 ± 2457 8.7 ± 6.6
ground sample, but the surface area difference value is half
(4.83%) that of the ground sample (8.47%). For stainless steel the
milled sample has higher values of Rq, Ra, and Rmax, but the surface
area difference values are similar (0.688% ground; 0.583% milled).
The effect of surface roughness on the contact angle is typically
described by either the Wenzel equation [25] or the Cassie–Baxter
model [26]. The Wenzel equation describes the effect of roughness
on homogeneous systems.
cosðh Þ ¼ r W cosðhÞ ð8Þ
where h⁄ is the apparent angle; rW is the ratio of the actual surface
area to the geometric area, comparable to the surface area increase
determined from the AFM measurements; h is the intrinsic contact
angle of the material.
The Cassie–Baxter model describes systems composed of multi-
ple materials.
cosðh Þ ¼ f1 cosðh1 Þ þ f2 cosðh2 Þ ð9Þ
where f1 and f2 are the fractions of the surface covered by each
material, and h1 and h2 are the contact angles of the respective
materials. The Cassie–Baxter model is appropriate for use with sys-
tems where air bubbles may be trapped between solid structures.
If air is considered to have a contact angle of 180°, the Cassie–
Baxter model becomes:
cosðh Þ ¼ ð1  fair Þ cosðh1 Þ  fair ð10Þ
where fair is the fraction of the surface consisting of air. Fig. 10 dem-
onstrates the predications of the Wenzel equation and the Cassie–
Baxter model for the effect of roughness on the observed contact
angle. For the Wenzel equation the roughness ratio rW has been
shown for a 10% (rW = 1.1) and 50% (rW = 1.5) increase in the surface
area. For the Cassie–Baxter model the apparent contact angles have
been calculated for a surface consisting of 10% (fair = 0.1) or 50% air
(fair = 0.5). The Wenzel equation predicts that the roughness in-
creases cos(h) for h less than 90° and decreases cos(h) for h greater
than 90°. The effect is nothing at 90° and increases as the angle be-
comes either larger or smaller. The Cassie–Baxter model predicts
that cos(h) is decreased, and the effect is greater for smaller angles
(cos(h) ? 1).
Assuming that the polished sample represents the contact angle
of the intrinsic material, the explanations of these models can be
 K. Hecht et al. / Chemical Engineering Journal 227 (2013) 2–12 9

3.2.6. Effect of solid material on the contact angle

Table 4 summarizes the average contact angle for a selection of
liquid/solid systems examined in this work. The summary includes
measurements conducted on both polished and machined surfaces
and in all gas atmospheres and pressures. Drops were at least
8.5 lL in volume or 3.2 mm in base diameter. Table 4 also contains
the surface tension of the investigated liquids. In general, the con-
tact angles presented in this table follow the trend of contact angle
increasing with increasing liquid surface tension. There are, how-
ever, exceptions; for instance the contact angle of water on glass
is noticeably lower. The contact angles measured on glass and
stainless steel also show greater variance than contact angles mea-
sured on other materials. Glass and stainless steel surfaces are typ-
ically hydrated by atmospheric water; therefore, the contact angles
measured on these surfaces can depend on the atmospheric
humidity and previous handling of the solid surface [24]. The solid
surfaces in the table are generally organized from those with the
lowest contact angles to those with the highest. PTFE consistently
has the highest contact angle with all materials. Contact angles
measured on polystyrene are typically higher than those on the
other three above materials. It is more difficult to organize the
other materials since their interactions with the liquids produce
different responses. Polar, dispersive, and acid and base interac-
tions between the solid and liquid materials must be considered

Fig. 6. Langmuir linear regression on the relationship between contact angle and
drop size (the radius of the drop base) for (a) water/N2/stainless steel and (b) water/
CO2/stainless steel.

compared to the measurements presented in Fig. 9. The Wenzel

model predicted that cos(h) should be greater than that of the
intrinsic material when cos(h) is greater than 0 (h > 90°) and less
when cos(h) is less than 0 (h < 90°). The measurements on PTFE
shown in Fig. 9 could support this trend. The PEEK measurements
do not substantially differ as cos(h) ? 1 but are less than cos(h) for
the intrinsic material even before this value reaches 0, which is
somewhat better described by the Cassie–Baxter model.
Neither model provides a satisfactory explanation of the mea-
sured data. The machined sample often approximates the polished
sample, despite the difference in roughness between the polished
and milled samples being greater than that between the milled
and ground samples. The models have been developed for ideal
surfaces and geometries, typically constructed in hydrophobic
materials. The influence of the roughness on the contact angles
measured on the practical surfaces examined in this work is not
easy to quantify and could be influenced by aspects of surface
geometry that roughness values fail to quantify.

Fig. 7. Values of gas/liquid surface tension (rgl) measured for mixtures of ethanol Fig. 8. Relationship between contact angle and surface tension for polished (a)
and water compared to data from Belda et al. [22]. PEEK, (b) stainless steel and (c) PTFE surfaces.
10 K. Hecht et al. / Chemical Engineering Journal 227 (2013) 2–12

Table 4
Contact angles measured for some liquid/solid systems.

Liquid Isopropanol 50 wt.% ethanol in Glycerine Water

water
rgl (mN/m) 21.7 [23] 31 ± 1 57 ± 6 72 ± 2
Contact angle (°)
Glass 0 20 ± 2 80 ± 6 55 ± 16
Stainless 0 26 ± 5 72 ± 7 79 ± 11
steel
PEEK 0 27 ± 5 77 ± 3
Polystyrene 0 28 ± 10 88 ± 9 79 ± 1
PTFE 9±4 53 ± 6 95 ± 6 97 ± 6

Table 5
Significance of investigated parameters on contact angle determined from a multiple
linear regression analysis.

Factor Coefficient p-value

rgl (mN/m) 1.23 5.8  10131
Ra (nm) 0.0223 1.9  1031
P  102 (kPa) 0.0123 0.97
V (lL) 0.0661 8.8  1016
xCO2 2.25 0.044

in the development of the solid–liquid energy term [15]. The solid

and liquid are clearly both important, but their interaction is diffi-
cult to characterize due to its complexity.

3.2.7. Statistical significance of the investigated parameters

Fig. 9. Effect of surface roughness for contact angles measured on (a) PEEK, (b) PTFE
and (c) stainless steel.
Multiple linear regression analysis was performed for the com-
plete set of contact angle measurements using the regression func-
tion of Microsoft Excel to determine the significance of gas/liquid
surface tension, surface roughness, pressure, drop volume, and
the fraction of CO2 in the gas. This practical approach does not offer
a physical explanation of the mechanism by which each of these
variables effect the observed values of contact angle, but this anal-
ysis can be used to examine the relative influence of each param-
eter. The results are summarized in Table 5. For a typical level of
significance of 5% or greater, i.e. p-value < 0.05, the pressure was
the only factor not found to be significant (p-value > 0.05). Accord-
ing to the analysis, the gas/liquid surface tension, the surface
roughness, and the drop volume are very clearly significant.
The fraction of carbon dioxide in the gas phase also has a low
p-value but not on the same order of magnitude as the other
variables. The effect of CO2 in the gas phase is not actually an effect
of the CO2 gas directly on the gas/solid energy (Eq. (3)) or on the
gas/liquid energy (gas/liquid surface tensions were measured di-
rectly) but likely occurs when CO2 dissolves in the liquid thereby
lowering the pH. The vast majority of the measurements in this pa-
per were completed with water or water/ethanol mixtures. Since
measurements with gases under pressure require several minutes
of purging the gas chamber and then the adjustment of the pres-
sure, all of the liquids measured can be reasonably assumed to
be saturated with the gases according to the solubility of the
respective gases and the measured pressure.
Considering only the parameters of gas/liquid surface tension,
surface roughness, and drop volume, a simple, empirical correla-
tion for the contact angle was calculated:

h ¼ 11:8 þ 1:22rgl þ 0:0232Ra þ 0:0674V ð11Þ

where rgl is the liquid surface tension in mN/m, Ra is the average

Fig. 10. Effect of surface roughness on the observed contact angle according to the
relations of Wenzel (red) and Cassie–Baxter (blue). (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of
this article.)
roughness of the surface in nm, and V is the liquid drop volume
in lL.
A parity plot showing the experimental values versus the values
predicted from this regression model is shown in Fig. 11. There is a
 K. Hecht et al. / Chemical Engineering Journal 227 (2013) 2–12
Fig. 11. Experimentally measured values of contact angle shown against values predicted by model.
large amount of scattering, but the data seem to spread symmetri-
cally around the bisecting line indicating that the model repro-
duces the general trend. However, it should be noted that the
solid material is nowhere accounted for in this analysis. Next to
the gas/liquid surface tension, this factor is probably one of the
most important as indicated in the Young equation. This model
may be useful to estimate the contact angle when contact angle
measurements are not available or to estimate the effect of surface
roughness and drop size on the contact angle when applying mea-
surements on polished surfaces to real systems. If a more accurate
value of the contact angle is needed, the contact angle must be
measured directly. Measurements of solid/liquid surface energy
are complicated rare but offer the ultimate solution to calculating
the contact angle directly according to Eq. (3).
4. Conclusions
In this work the effect of surface roughness in machined micro-
channels has been investigated with respect to its impact on the
contact angle, a parameter that characterizes the boundary condi-
tion in gas/liquid flow.
Plastic surfaces were generally rougher than metal surfaces for
all of the investigated surface preparation procedures (milling,
grinding and polishing).
Machined surfaces were found to have a surface roughness sim-
ilar to that of ground surfaces. Theory predicts that the contact an-
gle of the surface should depend on the surface roughness;
however, the contact angles measured on the milled surfaces were
closer to the angles measured on the polished surfaces than on the
ground surfaces. Generally, it appears that the contact angles on
milled surfaces are nearly the same as those of the polished sur-
faces. This would be even more true for flows in microchannels
as the microchannels are usually milled with parameters (lower
speeds, with cooling, etc.) that produce a more smooth surface.
A correlation based on linear regression has been developed
that relates the contact angle to the liquid surface tension, surface
roughness, and drop volume. Although the model can at best pro-
vide a very rough estimate of the contact angle since it neglects the
impact of the chemical interaction of the solid and liquid, it pro-
vides estimates for the contact angle from values that are obtain-
able from literature or estimation and represents the effect of the
drop size. Especially for organic liquids with low surface tensions
(ethanol, methanol and isopropanol), the model can quickly be
used to determine that these liquids have a low contact angle/good
wettability, in which case further contact angle measurements are
not necessary.
11
Acknowledgments
This work was carried out with the support of the Karlsruhe
Nano Micro Facility (KNMF, www.kit.edu/knmf), a Helmholtz Re-
search Infrastructure at Karlsruhe Institute of Technology (KIT,
www.kit.edu). Special thanks to Hendrik Hölscher and Richard
Thelen for support with AFM measurements.
References
[1] C.Y. Lee, S.Y. Lee, Influence of surface wettability on transition of two-phase
flow pattern in round mini-channels, Int. J. Multiphase Flow 34 (2008) 706–
711.
[2] C.Y. Lee, S.Y. Lee, Pressure drop of two-phase plug flow in round mini-
channels: influence of surface wettability, Exp. Therm. Fluid Sci. 32 (2008)
1716–1722.
[3] C.Y. Lee, S.Y. Lee, Pressure drop of two-phase dry-plug flow in round mini-
channels: effect of moving contact line, Exp. Therm. Fluid Sci. 34 (2010) 1–9.
[4] M.J. Madou, Fundamentals of Microfabrication: The Science of Miniaturization,
CRC Press, 2002.
[5] J.J. Brandner, T. Gietzelt, T. Henning, K. Manfred, H. Mortiz, W. Pfleging,
Microfabrication in metals and polymers, in: N. Kockmann (Ed.), Micro Process
Engineering: Fundamentals, Devices, Fabrication, and Applications, Wiley-
VCH, Weinheim, Germany, 2008.
[6] F. Özkan, K. Hecht, P. Pfeifer, K. Schubert, B. Kraushaar-Czarnetzki, Influence of
the contact angle on two-phase flow in microreactors for nitrobenzene–
hydrogen–stainless steel/carbon, Surf. Interface Anal. 42 (2010) 1122–1127.
[7] R.M. Santos, M. Kawaji, Numerical modeling and experimental investigation of
gas–liquid slug formation in a microchannel T-junction, Int. J. Multiphase Flow
36 (2010) 314–323.
[8] A.M. Barajas, R.L. Panton, The effects of contact angle on two-phase flow in
capillary tubes, Int. J. Multiphase Flow 19 (1993) 337–346.
[9] A. Günther, K.F. Jensen, Multiphase microfluidics: from flow characteristics to
chemical and materials synthesis, Lab Chip 6 (2006) 1487–1503.
[10] V. Haverkamp, V. Hessel, H. Löwe, G. Menges, M.J.F. Warnier, E.V. Rebrov,
M.H.J.M. de Croon, J.C. Schouten, M.A. Liauw, Hydrodynamics and mixer-
induced bubble formation in micro bubble columns with single and multiple
channels, Chem. Eng. Technol. 29 (2006) 1015–1026.
[11] G. Rosengarten, D.J.E. Harvie, J. Cooper-White, Contact angle effects on
microdroplet deformation using CFD, Appl. Math. Model. 30 (2006) 1033–
1042.
[12] F. Farshad, H. Rieke, J. Garber, New developments in surface roughness
measurements, characterization, and modeling fluid flow in pipe, J. Petrol. Sci.
Eng. 29 (2001) 139–150.
[13] L.F. Moody, Friction factors for pipe flow, Trans. ASME 66 (1944) 671–684.
[14] H.W. Fox, W.A. Zisman, The spreading of liquids on low energy surfaces. I.
Polytetrafluoroethylene, J. Colloid Interface Sci. 5 (1950) 514–531.
[15] C.J. van Oss, Interfacial Forces in Aqueous Media, CRC Press, Boca Raton, USA,
2006.
[16] B.B. Dayton, Outgassing rate of contaminated metal surfaces, in: 1961
Transactions of the Eighth National Vacuum Symposium (American Vacuum
Society) Combined with the Second International Congress on Vacuum Science
and Technology (International Organization for Vacuum Science and
Technology), Symposium Publications Division, Pergamon Press,
Washington, DC, 1962, pp. 42–57.
[17] K. Hecht, M. Kraut, Thermographic investigations of a microstructured thin
film reactor for gas/liquid contacting, Ind. Eng. Chem. Res. 49 (2010) 10889–
10896.
12 K. Hecht et al. / Chemical Engineering Journal 227 (2013) 2–12
[18] N. Schüler, K. Hecht, M. Kraut, R. Dittmeyer, On the solubility of carbon dioxide
in binary water–methanol mixtures, J. Chem. Eng. Data 57 (2012) 2304–2308.
[19] A.F. Stalder, G. Kulik, D. Sage, L. Barbieri, P. Hoffmann, A snake-based approach
to accurate determination of both contact points and contact angles, Colloids
Surf., A 286 (2006) 92–103.
[20] A. Wesch, N. Dahmen, K. Ebert, J. Schön, Grenzflächenspannungen,
Tropfengrößen und Kontaktwinkel im Zweiphasensystem H2O/CO2 bei
Temperaturen von 298 bis 333 K und Drücken bis 30 MPa, Chem. Eng.
Technol. 69 (1997) 942–946.
[21] A.B. Ponter, M. Yekta-Fard, The influence of environment on the drop size –
contact angle relationship, Colloid Polym. Sci. 263 (1985) 673–681.
[22] R. Belda, J.V. Herraez, O. Diez, A Study of the refractive index and surface
tension synergy of the binary water/ethanol: influence of concentration, Phys.
Chem. Liq. 43 (2005) 91–101.
[23] Lange’s Handbook of Chemistry, 13th ed., McGraw-Hill, Inc., New York, 1985.
[24] T. Engländer, D. Wiegel, L. Naji, K. Arnold, Dehydration of glass surfaces
studied by contact angle measurements, J. Colloid Interface Sci. 179 (1996)
635–636.
[25] R.N. Wenzel, Resistance of solid surfaces to wetting by water, Ind. Eng. Chem.
28 (1936) 988–994.
[26] A.B.D. Cassie, S. Baxter, Wettability of porous surfaces, Trans. Faraday Soc. 40
(1944) 546–551.