Sie sind auf Seite 1von 8

Measurement of the tangential momentum accommodation coefficient of

H2 on stainless steel, extreme ultraviolet-resist, and polyimide


Johannes F. M. Velthuis and Laurens van Bokhoven

Citation: J. Vac. Sci. Technol. A 31, 061605 (2013); doi: 10.1116/1.4816941
View online: http://dx.doi.org/10.1116/1.4816941
View Table of Contents: http://avspublications.org/resource/1/JVTAD6/v31/i6
Published by the AVS: Science & Technology of Materials, Interfaces, and Processing

Related Articles
Plasma etching: Yesterday, today, and tomorrow
J. Vac. Sci. Technol. A 31, 050825 (2013)
Nucleation and growth of MgO atomic layer deposition: A real-time spectroscopic ellipsometry study
J. Vac. Sci. Technol. A 31, 06F101 (2013)
Enhanced response to molecular adsorption of structurally defective graphene
J. Vac. Sci. Technol. B 31, 030602 (2013)
Effects of Cs adsorption on the field emission characteristics of closed single-walled carbon nanotubes
J. Vac. Sci. Technol. B 31, 021802 (2013)
Copper deposition on TiO2 from copper(II)hexafluoroacetylacetonate
J. Vac. Sci. Technol. A 31, 01A121 (2013)

Additional information on J. Vac. Sci. Technol. A
Journal Homepage: http://avspublications.org/jvsta
Journal Information: http://avspublications.org/jvsta/about/about_the_journal
Top downloads: http://avspublications.org/jvsta/top_20_most_downloaded
Information for Authors: http://avspublications.org/jvsta/authors/information_for_contributors
Downloaded 18 Sep 2013 to 139.63.40.192. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvsta/about/rights_and_permissions
Measurement of the tangential momentum accommodation coefficient
of H
2
on stainless steel, extreme ultraviolet-resist, and polyimide
Johannes F. M. Velthuis
a)
TNO Science and Industry, P. O. Box 155, 2600AD Delft, The Netherlands
Laurens van Bokhoven
b)
ASML Netherlands B.V., De Run 6501, 5504 DR Veldhoven, The Netherlands
(Received 12 March 2013; accepted 12 July 2013; published 9 September 2013)
The tangential momentum accommodation coefcient (TMAC) of H
2
on electropolished stainless
steel 304, extreme ultraviolet-resist, and polyimide was determined by measuring the mass ow
through a macroscopic rectangular channel (100 cm 10 cm 1 cm) as a function of the pressure
drop in the Knudsen range Kn
ave
~ 00.6. The TMAC and Knudsen number determine the amount
of velocity slip taking place at the wall at rareed conditions. By comparing the measurements
with an analytical expression for the mass ow through the channel, including the 1st order
slip ow contribution to continuum, the only remaining unknown, that is, the TMAC, was extracted.
VC
2013 American Vacuum Society. [http://dx.doi.org/10.1116/1.4816941]
I. INTRODUCTION
The value of the tangential momentum accommodation coef-
cient (TMAC), also referred to as r, and the thermal accom-
modation coefcient (TAC), also referred to as a, is important
as boundary conditions for the accurate simulation of ow and
thermal effects in extreme ultraviolet (EUV)-wafer scanners.
These scanners operate in a hydrogen environment in the
slip ow regime (0.01 <Kn <0.1) and the early transitional
ow regime (Kn 0.4). Simulations models that are being
used are CFD (computational uid dynamics) and DSMC
(direct simulation Monte Carlo) amongst others. Continuum
ow (Kn <0.01) is modeled by applying zero velocity slip
at the wall. At rareed conditions (small pressures and/or
small channel dimensions), both velocity and temperature
slip take place at the wall. The velocity slip (1st order slip
ow expansion of continuum) is a function of Knudsen num-
ber Kn and TMAC. The Knudsen number is a measure of
rarefaction and hence of the amount of velocity slip. For
channels, the Knudsen number computed at the channel av-
erage pressure (of the channel inlet and outlet) is representa-
tive, denoted Kn
ave
. The TMAC is a measure of the
momentum exchange between the gas and the surface, which
is a combined property of the gas and the surface involved.
TMAC data of H
2
on surfaces relevant for EUV-wafer
scanners could not be found in the literature; therefore, it
was decided to measure these parameters. The 1st order slip
ow expansion of continuum is also known to be valid in the
early transitional ow regime (Kn
ave
0.4), the extent of
which is investigated by varying the Knudsen number over
the range Kn
ave
~ 00.6.
Several methods exist to measure TMAC, see, for exam-
ple, Agrawal and Prabhu.
1
A popular method involves meas-
uring the ow rate over microchannels manufactured with
MEMS techniques using the rate of pressure rise (ROR) tech-
nique, see Arkilic et al.
2
By comparing the measured mass
ow rate over the channel with the analytical expression for
the mass ow rate over the channel, including the 1st order
slip ow contribution to continuum, the remaining unknown,
that is, TMAC, can be determined.
We use a similar method but now applied to macroscopic
channels. Thus, the method is not limited to MEMS compati-
ble materials but can be applied to actual surfaces and condi-
tions, such as found in EUV-wafer scanners. Standard mass
ow sensors can be used, avoiding the difculties associated
with the ROR technique. First, the TMAC of H
2
in an elec-
tropolished stainless steel 304 (SS304) channel is measured.
Next, the channel is spray-coated to measure the TMAC on
EUV-resist (SPUR-V002). Finally, the channel is covered
with thin sheet material to measure the TMAC on polyimide
(Dupont Kapton type H-lm).
II. THEORY
A. Channel mass flow rate
Figure 1 shows the channel with rectangular cross section.
Starting from the NavierStokes equations, a theoretical
expression for the laminar mass ow rate over the channel
can be derived, which includes a 1st order slip ow contribu-
tion to continuum as proposed by Maxwell in 1879.
25
The
mass ow rate expression reads
_ m
p
2
i
p
2
o
= A B
2
p
o
p
i

; (1)
where
A =
H
3
W M
3
4
C
1
24 l L R T
; (2)
B = A
C
2
C
1

2 r
r
Kn
o
p
o
; (3)
C
1
=
4
3
1
192
p
5

H
W

n =1;3;5::
1
n
5
tanh
np
2

W
H
" #
;
(4)
a)
Electronic mail: Han.Velthuis@tno.nl
b)
Electronic mail: Laurens.van.Bokhoven@asml.com
061605-1 J. Vac. Sci. Technol. A 31(6), Nov/Dec 2013 0734-2101/2013/31(6)/061605/7/$30.00 VC
2013 American Vacuum Society 061605-1
Downloaded 18 Sep 2013 to 139.63.40.192. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvsta/about/rights_and_permissions
C
2
= 8 C
1

256
p
4
1
H
W

n=1;3;5::
1
n
4
tanh
2
np
2

W
H

:
(5)
The mass ow rate _ m (kg/s) is scaled with p
i
2
p
o
2
, where
p denotes pressure and subscript i and o refer to the
channel inlet and the channel outlet, respectively. The aver-
age pressure is dened as p
ave
=(p
i
p
o
)/2. For an isother-
mal channel of given dimension and given TMAC value r,
the terms A and B in Eq. (1) are constants, leading to a
linear relationship between the scaled mass ow rate and
inverse average pressure. The rst term A represents the
scaled mass ow rate in the continuum limit, whose theoreti-
cal value is given by Eq. (2). The second term B accounts
for slip ow effects, whose theoretical value is given by
Eq. (3). Slip has the effect of enhancing the mass ow. The
slip term in Eq. (1) disappears when the absolute pressures
rises, despite B having a nite value. The Knudsen number
Kn is a measure for noncontinuum effects and is dened as
Kn=k/H, where H is the channel height and k denotes
the mean free path of gas molecules between collisions
k =
l
q

p M
2 R T
r
: (6)
The density q of hydrogen is calculated with the ideal the
gas law, where R is the universal gas constant, T is the
temperature, and M is the molecular weight (2.016 g/mol).
The dynamic viscosity l of hydrogen is taken according to
Table I. For comparison, at isothermal conditions, the mean
free path of hydrogen is inversely proportional to pressure
and reads k ~ 12 mm at 1 Pa and 293.5 K. We differentiate
between the outlet Knudsen number Kn
o
(evaluated at the
channel outlet pressure p
o
) and the average Knudsen num-
ber Kn
ave
(evaluated at the average channel pressure p
ave
).
As in our measurements, we always enforce p
i
to be greater
than or equal to two times p
o
; this in practice means that
Kn
ave
is always less than 2/3Kn
o
. Note that for isothermal
conditions, the product (Knp), for example, appearing in
Eq. (3), is constant. The coefcients C
1
[Eq. (4)] and
C
2
[Eq. (5)] account for the channel aspect ratio (H/W).
For a plan-parallel channel with innitely small aspect ratio
(H/W=0), the coefcients read C
1
=4/3 and C
2
/C
1
=6.
For a channel with nite aspect ratio (H/W~ 0.1), as applies
for our experimental setup, the channel sides reduce the con-
tinuum mass ow rate by approximately 7%, when com-
pared to a channel with zero aspect ratio.
By linear curve-tting the measured data for the scaled
mass ow rate versus inverse average pressure according to
Eq. (1), an experimental value is obtained for the line inter-
cept A and slope B. From these and the previous given theo-
retical expression for the slope [Eq. (3)], the unknown
TMAC value r is extracted, according to
r =
2
1
B
A

1
Kn
o
P
o

C
1
C
2
: (7)
In this paper, we will determine the TMAC value, or r, in
the Knudsen range Kn
o
=01.2, equivalent to Kn
ave
~ 00.6.
B. Assumptions
In the derivation of the scaled mass ow rate expression
[Eq. (1)], a number of simplifying assumptions are used. The
experimental conditions must meet these simplifying
assumptions (see Sec. IV), otherwise the extracted TMAC
values become ill-dened. The simplifying conditions are as
follows:
(1) Laminar ow. For this, the hydraulic Reynolds numbers
should stay below the laminar-turbulent ow transition,
Re
2H
~ 2300.
(2) Negligible inertial effects. For this, the product of Mach
number at the channel inlet M
i
, Mach number at the
channel outlet M
o
, and heat capacity ratio c should be
smaller than unity (cM
o
M
i
1), see Karniadakis and
Beskok.
3
(3) Isothermal channel. For isothermal conditions, the varia-
tion in channel temperature in time and in place must be
limited, i.e., less than 3 K, which corresponds to 1% rela-
tive error in absolute ambient temperature.
(4) Slip ow (Kn
ave
<0.1). Notwithstanding this condition, it
is generally known that Eq. (1) with associated TMAC
(derived for the 1st order slip ow regime) also reprodu-
ces the mass ow rate in the early transitional ow regime
(Kn
ave
0.4) quite well. When lumping higher order
Knudsen effects to Eq. (1) by simply modifying the
TMAC value, this per denition will result in a Knudsen
dependency of the TMAC. Plotting these lumped TMAC
values is of limited practical signicance, other than as an
indicator for higher order Knudsen effects. As in our
experiments Kn
ave
does not exceed the value 0.6, we
expect that this Knudsen dependency is still limited.
(5) Developed ow. To say negligible entrance and exit effects
on the channel pressure drop. Duan and Muzychka
7
give
FIG. 1. Channel with rectangular cross section (L ~ 100 cm, W ~ 10 cm,
H ~ 1 cm).
TABLE I. Dynamic viscosity of H
2
[(reported (Ref. 6) accuracy 60.5%].
T (K) l (lPas)
290 8.766
300 8.969
310 9.170
061605-2 J. F. M. Velthuis and L. van Bokhoven: Measurement of the TMAC 061605-2
J. Vac. Sci. Technol. A, Vol. 31, No. 6, Nov/Dec 2013
Downloaded 18 Sep 2013 to 139.63.40.192. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvsta/about/rights_and_permissions
an expression for the apparent friction factor for ow in a
plane parallel channel including slip ow and entry
effects. Using this expression, it can be shown that the rel-
ative error in the scaled mass ow rate due to entry/exit
effects, relative to that of a channel with developed ow,
stays below 0.1% for Reynolds numbers Re
2H
<25,
where it is assumed that the exit effect is of the same
order of magnitude as the entrance effect.
III. EXPERIMENT
A. Vacuum system
The vacuum system consists of two vacuum vessels,
referred to as inlet vessel and outlet vessel, see Fig. 2.
Each vessel has a volume of ~1.4 m
3
. A stainless steel 304
channel connects the two vessels (see Sec. III B for details).
The inlet vessel is not pumped and the respective pump-
openings are closed. The outlet vessel is pumped by option-
ally four Turbo Molecular Pumps (TMP), type Shimadzu
TMP-H3153LMC. The pumps are water-cooled with a
chiller to lab temperature level (~293 K). The individual
pump speed is 2600 l/s H
2
at 24 000 rpm (=100%) for abso-
lute pressures in the normal operating range (<1 Pa). In the
experiments, the pumps are used outside their normal operat-
ing range, characterized by a considerable drop in pump
speed. The pump speed can be manually adapted by setting
the rotational speed between 25% and 100% in discrete steps
of 1%. H
2
gas can be injected into the inlet vessel and sepa-
rately into the outlet vessel, using mass ow controllers
(MFCs). At the inlet vessel, two MFC are present, covering
two ow ranges. Sensor details can be found in Table II. For
absolute outlet pressures p
o
_5 Pa, the MFC with the highest
ow range is selected, otherwise the MFC with the lowest
ow range is used. The absolute pressure of the inlet vessel
and outlet vessel is monitored using Baratron capacitance
sensors. For H
2
safety reasons, the absolute vessel pressure
is limited to ~120 Pa. The temperature of the SS304 channel
is monitored using 3 NTCs in the ow direction, glued to the
external channel surface. For the EUV-resist and the polyi-
mide measurement, only 1 NTC is used halfway along the
channel, as the channel was found to be isothermal. Prior to
the experiments, the vacuum system was tested for leaks/out-
gassing, which gave equivalent mass ow rates one order of
magnitude smaller than the resolution of the smallest MFC
used in the experiments.
The absolute pressure p
i
at the channel inlet and at the
channel outlet p
o
can be set independently of each other,
within a limited range, by playing with the number of pumps
in operation and their rotational speeds, in combination with
adjusting the H
2
ow to the inlet vessel. Optionally, H
2
can be
injected into the outlet vessel in the case that the minimum
pump speed is reached and the outlet pressure still needs to be
raised. After a change in set-point value for the ow rate or
the rotational speed, it takes approximately 5 min for the sys-
tem pressures to become steady. In the steady state, the mass
ow rate as recorded by the MFC present at the inlet vessel
now equals the mass ow rate through the channel.
B. Channel
The channel consists of two electropolished stainless steel
304 parts bolted together, see Fig. 3. The channel dimensions
with uncertainties are given in Table III. The variation in
channel height (bottom part) was less than 2% based on 24
equidistant measurement points in ow direction, that is,
over a length of the channel of 1 m. This does not include
any curvature of the channel lid (i.e., due to bolting stresses),
which may lead to a somewhat different effective channel
height. The roughness (arithmetic average of absolute val-
ues) after electropolishing varied between R
a
=0.2 and
0.6 lm (Mahr Perthometer M2P, pfk measure head). Kapton
self-adhesive tape was applied to the sides of the channel
combined with Teon washers to cancel any remaining side-
leakage. The channel was wrapped in four layers of aluminum
foil, serving as a radiation shield, in order to prevent the
channel collecting any heat radiation from the pumps.
FIG. 2. (Color online) Test setup which is sliced-open to show the channel
inside. In steady state operation, the channel ow rate equals the H
2
injec-
tion rate.
TABLE II. Sensors.
Sensor Manufacturer Item Value
MFC Horiba STEC
(Z512MGXC)
Full scale (FS) 300 and 2760 sccm
a
Accuracy
b
60.25% of FS
(ow < 25% FS)
61% of SP
(ow > 25% FS)
Resolution 0.02% FS
Absolute
pressure
Absolute
Capacitance
Manometer
MKS Baratron
VR
Type 624B
Range 100 Pa (FS)
Resolution 0.01 Pa
Accuracy 0.12% of reading
at 90%
c
condence
level
Temperature NTC, YSI
thermistor 44031
Accuracy
b
60.1

C
a
Standard conditions: 1013 hPa, 273.15 K.
b
99.8% condence level (3stdev/mean).
c
Source: private email MKS, equivalent to 0.22% of reading at 99.8% con-
dence level.
061605-3 J. F. M. Velthuis and L. van Bokhoven: Measurement of the TMAC 061605-3
JVSTA - Vacuum, Surfaces, and Films
Downloaded 18 Sep 2013 to 139.63.40.192. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvsta/about/rights_and_permissions
C. Surface preparation
1. Stainless steel
The cleaning procedure of the electropolished stainless
steel 304 channel parts consisted of degreasing, rinsing in
demineralized water, alkaline cleaning, and repeated rinsing
in demineralized water of increasing purity, followed by dry-
ing in air. After mounting in the test setup, the channel
resided in vacuum (1 Pa) for a few days and was allowed
to outgas before the experiments started. This outgassing
procedure was followed for all materials studied.
2. EUV-resist
For the TMAC measurement on EUV-resist, the internal
wetted surface of the stainless steel channel was spray-
coated with several layers of EUV-resist (SPUR-V002, see
Table IV), with drying in between. The resist itself is almost
colorless in appearance. After applying a few layers, the
coating showed colorization, indicating a layer thickness of
typically 0.25 lm (
1
=
4
wavelength of visible light). Between
5 and 10 layers were applied in total until the colorization
disappeared. Subsequently, the resist was baked at ~105

C,
using a lint heater wrapped around the channel.
3. Polyimide (Kapton)
For the TMAC measurement on polyimide, the internal
wetted surface of the stainless steel channel was coated with
a 25 lm thin polyimide sheet (Dupont Kapton type H lm),
with negligible effects on channel height and the variation in
channel height.
D. Measurement procedure
A typical measurement series consists of tuning the outlet
channel pressure p
o
to a constant value (within 1%) and
varying the inlet pressure to approximately multiples of the
outlet pressure (p
i
=np
o
, n _2). Next, the scaled mass ow
rate is plotted versus inverse average pressure according to
Eq. (1), including error bars in both coordinate directions
due to sensor accuracies or resolution, whichever is larger.
The error bars are determined using linear error propagation
techniques and assuming square-root weighted averaging of
the individual error contributions, assuming independent
measurement quantities. All uncertainties mentioned in this
paper are based on a 99.8% condence level (3stdev/
mean). Next a linear curve t of the plotted data is made
using the total least squares (TLS) technique.
8
The TLS tech-
nique takes into account the errors bars present in the data,
in both coordinate directions. The curve t procedure results
in a value for the line intercept A (scaled mass ow in the
continuum limit) and for the slope B (slip effect), together
with their associated uncertainties DA and DB. Next,
the TMAC value r is calculated according to Eq. (7). The
uncertainty Dr is determined again using linear error propa-
gation techniques, now with linear addition of the individual
error contributions. Thus, a certain measurement series
can be associated with a specic (constant) channel outlet
pressure p
o
, a corresponding Knudsen number Kn
o
, and a
TMAC value. By measuring multiple data series at varying
discrete channel outlet pressures, the TMAC values can be
plotted as function Kn
o
. As already stated in Sec. II B, the
relevance of plotting TMAC beyond the early transition
range (Kn
ave
0.4) is limited, other than as an indicator for
higher order Knudsen effects. Typical discrete outlet pres-
sures selected are in the range p
o
=201 Pa. The highest
occurring channel outlet Knudsen numbers is Kn
o
=1.2,
which is equivalent to a measurement series in which the
average Knudsen number is varied between Kn
ave
~ 0 and
0.6. Measurements at even higher Knudsen numbers (lower
pressures) were not conducted, as these where limited by the
resolution of the pressure sensors and accuracy of the mass
ow controllers.
To determine the TMAC valid for the slip ow regime
(and early transitional ow regime), all measurement series
were collected in one plot (scaled mass ow rate versus
inverse average pressure) and tted to Eq. (1) in the range
Kn
ave
0.4. The resulting scaled mass ow in the continuum
FIG. 3. (Color online) Stainless steel channel consisting of bottom part and lid.
TABLE III. Channel dimensions.
Item Description
a
Height H 10.16 60.17 mm
Width W 100.0 60.1 mm
Length L 1000.8 60.5 mm
C
1
1.248
b
C
2
7.588
b
a
99.8% condence level (3stdev/mean).
b
At nominal dimensions.
TABLE IV. EUV-resist (SPUR-V002). Source material safety data sheet.
Item Description
Product SPUR-V002
Supplier Shin-Etsu Chemical Co Ltd., Japan
Composition <30% PGMEA, >50% cyclohexanone,
<10% 1-methoxy-2-propanol, <10% Polymer
Appearance Transparent, colorless to pale yellow
061605-4 J. F. M. Velthuis and L. van Bokhoven: Measurement of the TMAC 061605-4
J. Vac. Sci. Technol. A, Vol. 31, No. 6, Nov/Dec 2013
Downloaded 18 Sep 2013 to 139.63.40.192. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvsta/about/rights_and_permissions
limit (intercept A) is compared with the theoretical value
Eq. (2), by inserting in this equation the channel dimensions,
temperature, and corresponding H
2
dynamic viscosity. The
uncertainty in the theoretical value is determined based on
square-root weighted averaging of the individual error
contributions.
IV. RESULTS
Over the course of the electropolished stainless steel
experiments, the channel temperatures stayed within
293.8 6 1.5 K, indicating that the channel can be considered
isothermal during the experiments (Sec. II B). As already
mentioned in Sec. III D, uncertainties given in this paper
refer to a 99.8% condence level (3stdev/mean). The tem-
perature of the channel during all the EUV-resist experi-
ments stayed within 293.5 6 0.5 K. The temperature of the
channel during all the polyimide experiments stayed within
293.2 60.3 K.
The Reynolds numbers (not shown) during all ow
experiments stayed below 10, well below the previously
determined criterion for the laminar to turbulent ow
transition of Re ~ 2300 (Sec. II B). Entrance and exit
effects on the channel pressure drop can therefore be safely
neglected, and the ow can be considered developed
throughout the channel. Inertia effects are also considered
negligible, as the product of inlet Mach number, outlet
Mach number, and heat capacity ratio c during all experi-
ments stayed below 0.05, far less than the previously deter-
mined criterion of unity (Sec. II B).
Figure 4 shows a typical measured data series in which
the channel outlet pressure p
o
is kept constant during
the experiments and the channel inlet pressure p
i
is varied.
For this particular example, p
o
=10 Pa (equivalent to
Kn
o
=0.12). The outlet pressure p
i
is varied between 20
and 100 Pa, which is equivalent to 1/p
ave
~ 0.020.07 or
Kn
ave
~ 0.020.08. The mass ow rate is scaled with the the-
oretical continuum mass ow rate, which is equal to a con-
stant times (p
i
2
p
o
2
). The inverse average pressure 1/p
ave
is
given along the x-axis. To convert the x-axis to Kn
ave
, a lin-
ear scaling factor of 1.2 must be used. The error bars in the
scaled mass ow rate become larger toward the right side of
the graph (lower pressure range), as the associated mass ow
rates fall into the lower, less accurate ow range of the
MFCs. The error bars in the x-direction are not visible on
this scale. As expected, the measured data fall on a straight
line conform Maxwells model. The associated TMAC for
this data series with Kn
o
=0.12 is given in Fig. 5, together
FIG. 4. (Color online) Scaled mass ow rate vs inverse average pressure. To
convert to Kn
ave
multiply the x-axis with 1.2. Shown is a typical data series
in which p
o
is kept constant and p
i
is varied. Straight lines are Maxwells
model [Eq. (1)] with TMAC as t parameter.
FIG. 5. (Color online) TMAC as function of Kn
o
obtained by tting a meas-
ured series of mass ow data to Maxwells model [Eq. (1)]. A specic value
of Kn
o
corresponds to a data series in which p
o
is kept constant and p
i
is var-
ied (Kn
ave
varies between roughly 0 and 2/3Kn
o
).
FIG. 6. (Color online) Scaled mass ow rate vs inverse average pressure for
SS304. All data series collected. To convert to Kn
ave
multiply the x-axis
with 1.2. The straight line is Maxwells model [Eq. (1)] with the TMAC t-
ted in the range Kn
ave
0.4.
FIG. 7. (Color online) Similar as Fig. 6 but now for EUV-resist.
061605-5 J. F. M. Velthuis and L. van Bokhoven: Measurement of the TMAC 061605-5
JVSTA - Vacuum, Surfaces, and Films
Downloaded 18 Sep 2013 to 139.63.40.192. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvsta/about/rights_and_permissions
with the results of the other data series. The highest outlet
Knudsen number reads Kn
o
=1.2, which corresponds to a
data series in which the average Knudsen number varies
between Kn
ave
=0 and 0.6. As remarked in Sec. II B 4 plot-
ting TMAC values as a function of Kn
o
based on measure-
ment data lying outside the early transition ow regime
(Kn
ave
0.4) is of limited practical signicance, other than
as an indicator of higher order Knudsen effects. The
Knudsen dependency of the TMAC shows a similar trend as
that observed in Refs. 2 and 3 for nitrogen and argon gas in
contact with a silicon microchannel in the range Kn
o
<0.4,
namely, a small decrease in lumped TMAC value up to a
Kn
o
~ 0.2, after which the TMAC becomes more or less
constant. This seems consistent with the nding that the 1st
order slip ow model still holds in the early transitional ow
regime (Kn
ave
0.4) as the measurements stay close to this
range. The TMAC of SS304 at Kn
o
=1.2 shows a deviation
and larger uncertainty than the EUV-resist and polyimide
results. This is most probably due to the limited variation of
the channel inlet pressure used in the SS304 experiments. As
the SS304 channel at that time was already coated with
resist, this particular SS304 experiment could not be repaired.
This was rectied in later experiments for EUV-resist and pol-
yimide, by adding more pressure data points.
Figures 68 show all measured data series collected in
one graph per surface material. The mass ow rate is scaled
with the theoretical continuum mass ow rate, which is
equal to a constant time (p
i
2
p
o
2
). To convert the inverse
average pressure along the x-axis to Kn
ave
, a linear scaling
factor of 1.2 must be used. The straight lines are Maxwells 1st
order slip ow model [Eq. (1)], where the associated TMAC is
tted on that subset of the data for which Kn
ave
0.4, that
is, lying in the early transitional ow regime. These TMAC
values are summarized in Table V. This table also compares
the intercept of the tted line with the theoretical value as
given by Eq. (2). These values are equal to each other within
the measurement uncertainty. The slope of the curve connect-
ing the measurement points increases only very gradually,
indicating the onset of higher order rarefaction effects at higher
Knudsen numbers. The data support the general nding that
the 1st order slip ow model can be used throughout the early
transitional ow regime (Kn
ave
0.4).
Another way of representing the data is plotting the mass
ow rate versus the average Knudsen number, similar as in
given Ref. 3 (Fig. 5.14). The mass ow rate is scaled with the
theoretical free molecular mass ow rate which is equal to a
constant times the pressure difference (p
i
p
o
) over the channel.
The resulting graph is shown in Fig. 9, ignoring the exact value
of the constant. Maxwells model [Eq. (1)] with TMAC=0.83
is given as comparison. As the measurement range does not
exceed Kn
ave
=0.6, it is still too early to see a local minimum
in the scaled mass ow rate which in general is observed
around Kn
ave
~ 1, the so-called Knudsen minimum.
3
TACs on the same prepared surfaces were separately
measured by Philips,
9
using the heated parallel disk method
and by tting the results with the ShermanLees
10
relation-
ship for heat transfer between parallel gaps in the continuum
FIG. 8. (Color online) Similar as Fig. 6 but now for polyimide.
TABLE V. TMAC and TAC (Refs. 9 and 10) of H
2
on SS304, EUV-resists, and polyimide (Kn
ave
< 0.4). Condence level 99.8% (3stdev/mean).
Surface
Theoretical
scaled mass
flow [Eq. (2)]
Measured scaled
mass flow in the
continuum limit [Eq. (1)]
Measured slope
[Eq. (1)]
Momentum accommodation
coefficient TMAC
Thermal accommodation
coefficient TAC (Refs. 9 and 10)
A
theory
A
meas
B
meas
r a
[kg/(s Pa
2
)] [kg/(s Pa
2
)] [kg/(s Pa)] [] []
SS304 3.8 60.2 10
10
3.64 6 0.02 10
10
3.71 60.02 10
9
0.83 6 0.01 0.32 60.04
EUV-resist 3.8 60.2 10
10
3.65 6 0.02 10
10
3.70 60.02 10
9
0.84 6 0.01 0.39 60.04
Polyimide 3.8 60.2 10
10
3.67 6 0.02 10
10
3.78 60.02 10
9
0.83 6 0.01 0.35 60.04
FIG. 9. (Color online) Mass ow rate channel vs Kn
ave
. The mass ow rate is
scaled with the theoretical free molecular mass ow rate, which is propor-
tional to (p
i
p
o
). The line is Maxwells model [Eq. (1)] with TMAC=0.83
(tted in the range Kn
ave
0.4).
061605-6 J. F. M. Velthuis and L. van Bokhoven: Measurement of the TMAC 061605-6
J. Vac. Sci. Technol. A, Vol. 31, No. 6, Nov/Dec 2013
Downloaded 18 Sep 2013 to 139.63.40.192. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvsta/about/rights_and_permissions
to transitional ow regime (Kn ~ 01.2). These TAC values
can be found in Table V.
V. SUMMARYAND CONCLUSIONS
The tangential momentum accommodation coefcient
(TMAC) of H
2
on various surfaces was determined, by meas-
uring the mass ow through a macroscopic rectangular chan-
nel (100 cm 10 cm 1 cm) as a function of pressure drop in
the Knudsen range Kn
ave
<0.6. This works main advantage is
the possibility of using macroscopic channels (as compared to
frequently used MEMS microchannels and MEMS materials).
Therefore, this method can be easily applied to a number of
actual surfaces, for example, relevant for EUV-wafer scanners.
The results support the general nding that the 1st order slip
ow expansion can be used throughout the early transitional
ow regime (Kn
ave
0.4). The TMAC value for H
2
on stain-
less steel 304 and polyimide (r =0.83 6 0.01) was found to
be almost identical to that on EUV-resist (r =0.84 60.01).
ACKNOWLEDGMENT
This project is funded by ASML Netherlands B.V.
1
A. Agrawal and S. V. Prabhu, J. Vac. Sci. Technol. A 26, 634 (2008).
2
E. B. Arkilic, K. S. Breuer, and M. A. Schmidt, J. Fluid Mech. 437, 29
(2001).
3
G. E. Karniadakis and A. Beskok, Micro Flows (Springer, New York, 2002).
4
J. Jang and S. T. Wereley, J. Micromech. Microeng. 16, 493 (2006).
5
F. M. White, Viscous Fluid Flow (McGraw-Hill, New York, 2006).
6
M. J. Assaei, S. Mixafendi, and W. A. Wakeham, J. Phys. Chem. Ref.
Data 15, 1315 (1986).
7
Z. Duan and Y. S. Muzychka, J. Fluids Eng. 132, 011201(2010).
8
W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery,
Numerical Recipes in Fortran (Cambridge University Press, Cambridge,
1992).
9
S. Box and G. Hannen, Measurement of the thermal accommodation
coefcient for ASML, Philips Research, Report Ref: ATP591-11-0405/
01, 2011.
10
D. J. Rader, W. M. Trott, J. R. Torczynski, J. N. Casta~neda, and T. W.
Grasser, Measurements of thermal accommodation coefcients, Sandia
Report SAND2005-6084, 2005.
061605-7 J. F. M. Velthuis and L. van Bokhoven: Measurement of the TMAC 061605-7
JVSTA - Vacuum, Surfaces, and Films
Downloaded 18 Sep 2013 to 139.63.40.192. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvsta/about/rights_and_permissions

Das könnte Ihnen auch gefallen