Hydrodynamics and Heat Transfer in Gaseous Microflows: from Theory to Experiment Ecole thématique 2013 GDR CHANT Prapoutel les Sept Laux – January 10, 2013 S. Colin
Hydrodynamics and Heat Transfer in Gaseous Microflows:
from Theory to Experiment
Ecole thématique 2013 GDR CHANTPrapoutel les Sept Laux – January 10, 2013
S. Colin
2
Outline
1. Introduction : gas microflows applications
2. Gas microflows and rarefaction
3. Different regimes function of Kn
4. Importance of slip flow regime
5. Modelling of gaseous microflows
6. Examples of specific behaviours
7. Need for experimental data on gas microflows
8. Conclusions
3
Gas Microflows Applications
� Micro heat exchangers for cooling of electronic components or for chemical applications
� Micro nozzle for control of nano satellite position
� Micro gas analyzers – micro-chromatography
� Fluidic micro actuators for active control of aerodynamic flows
� Vacuum generators for extracting biological samples – Knudsen micropumps
� Mass flow and temperature micro-sensors
� Pressure gauges
� …Artificial lung
Micro-nozzle for
mass spectrometry
1. Introduction
4
Gas Microflows –Scales to Take Into Account
L
δλ
d
d : mean molecular diameter
δ : mean molecular spacing
λ : mean free path
L : characteristic length of the flow
2. Rarefaction
5
Limits of UsualContinuum Approach (N-S, classic BC…)
• sampling volume (SV) contains 10 000 molecules
� 1% of statistical fluctuations on macroscopic properties.
The mean volume occupied by a molecule is δ
4 310 22SV
l δ > ≈ 4 310L
δ≫
Negligible
Statistical Fluctuations
1L
λ≪ Thermodynamic Equilibrium
• local thermodynamic equilibrium: enough collisions inside the SV
� λ small compared with lSV
1SV
lλ ≪
• binary collisions
� enough molecular spacing
1d
δ≪ Dilute Gas
d : mean molecular diameter
δ : mean molecular spacing
λ : mean free path
L : characteristic length of the flow
2. Rarefaction
6
Dilute gas
Thermodynamic equilibrium
Negligible statistical fluctuations
1d δ ≪
1Lλ ≪
4 / 310L δ ≫
Conventional Limits
δδδδ / d
101 100102
100
101
102
103
104
105
106
107
L / d
δδ δδ /
d == ==
7
L / δδδδ ==== 100
λλλλ / L
==== 0,1
dense gasdilute gas
thermodynamic
disequilibrium
significant
statistical
fluctuations
negligible
statistical
fluctuations
air, standard conditions
thermodynamic
equilibrium C
S
T
M
d : mean molecular diameter
δ : mean molecular spacing
λ : mean free path
L : characteristic length
of the flow
BIRD G.A., Molecular gas dynamics and the
direct simulation of gas flows, Oxford,
Clarendon Press, 1998.
( ) ( )( )31 2 KnL d d Lλ πδ= =
( )3 22Hard Sphere
dλ δ π=
2. Rarefaction
7
( )2k RTλ µ ρ=
Reu Lρ
µ= Ma
u u
a RTγ= =
KnL
λ=
Knudsen number
2
MaKn
Rek γ=
δ / d101 100102
100
101
102
103
104
105
106
107
L d /K
n 10
=
-3
Standard conditions for He air SO2
Kn
10=
-1
Kn
10=
+1
(C)
(S)
(T)
(M)
δ=
/
7
d
Regimes as a function of Kn
3. Regimes
8
Kn
H2
H1
0,001 0,01 0,1 1 10
écoulement classique
Non raréfié
apparitionde glissementet d'un saut
de températureà la paroi
Légèrement raréfié
disparitionde la notion
de milieu continu
Modérèment raréfié
les chocs entre molécules
deviennent extrêmement
rares
Hautement raréfié
100 10 1 0,1 0,01
h(µm)
h
h
Regimes as a Functionof Hydraulic Diameter
Rarefaction increase
Continuum flow Slip flow Transition flow Free molecular flow
Classic BC
Continuum medium
classic flow negligible
intermolecular
collisions
continuum
hypothesis no
longer valid
slip velocity and
temperature jump
at the walls
Air in standard conditions
3. Regimes
9
Knudsen Layer and Slip at the Wall
� Thickness:
1 ou 2 λ
� Local
thermodynamic
disequilibrium
O (
)λ
Knu
dsen
Lay
er
s w allu u-
w allu
gasu
slipu
W all
G a s
n
s
4. Slip
10
10
2
sslip slip
uu u
nλ
∂ = + + ∂
sslip
uu
nλ
∂⇒ =
∂
• Basic explanation
~ λ
s
n
us(λ)
Slip at the Wall – Physical Phenomenon
4. Slip
11
Accommodation Coefficients
Specular reflection Diffuse reflexion
n
s
(a) (b)
1 − σu σu
Same behaviour for the temperature: σT
4. Slip
2u t
s wall
u
uu u
n
σλ
σ Γ
− ∂− =
∂
12
Slip Flow Regime: Some Examplesof Various Velocity Slip Boundary Conditions
� Initial form– Maxwell, J.C. (1879) Philosophical
Transactions of the Royal Society,170.
2 3
4
u t nt wall
u
u u R Tu u
n t p t
σ µλ
σ ΓΓ
− ∂ ∂ ∂ − = + +
∂ ∂ ∂
22 3 3
2 4
u tt wall
u
u T Tu u
n T s t T t
σ µ µλ
σ ρ ρΓΓ
− ∂ ∂ ∂− = − +
∂ ∂ ∂ ∂
2
1 *
u tt wall
u
Kn uu u
b Kn n
σ
σ Γ
− ∂− =
− ∂
( )2
tanh*
u tt wall
u
uu u Kn
n
σ
σ Γ
− ∂− =
∂
2 * ** *
* 2 *
u tt wall
u
u Kn pu u Kn Re
n t
σ
σΓ
− ∂ ∂ − = +
∂ ∂
2 2 22
2 2 2
2 92
16
u t t t tt wall
u
u u u uu u
n n s t
σ ∂ ∂ ∂ ∂λ λ
σ ∂ ∂ ∂ ∂Γ Γ
−− = − + +
� Curvature effects– Barber, R.W., et al. (2004) Vacuum, 76.
� Higher-order forms– Deissler, R.G. (1964) Int. Journal
of Heat and Mass Transfer, 7.
� Other hybrid dimensionless forms– Karniadakis, G.E. & Beskok, A. (2002) Microflows:
Fundamentals and Simulation, Springer-Verlag.
– Xue, H. & Fan, Q. (2000) Microscale ThermophysicalEngineering, 4.
– Jie, D., et al. (2000) Journal of Micromechanics and Microengineering, 10.
4. Slip
13
Slip Flow Regime: Examples of Temperature Jump Boundary Conditions
� Initial form– Smoluchowski, M. (1898) Annalen
der Physik und Chemie,64, 101-130.
� Additional term– Sparrow, E. M. & Lin, S. H.
(1962) Journal of Heat Transfer, 84, 363-369.
� Higher-order forms– Deissler, R. G. (1964) International
Journal of Heat and Mass Transfer,7, 681-694.
� Langmuir boundary condition– Myong, R. S., et al. (2006) International
Journal of Heat and Mass Transfer,49, 2502-2513.
2 2
1
Twall
T
TT T
Pr n
σ γ λ
σ γ Γ
− ∂− =
+ ∂
( )2 2 2 2
2 2 2
2 2
1
177 145 92
1 256
Twall
T
TT T
Pr n
T T T
n t s
σ γ λ
σ γ
γ λ
γ
Γ
Γ
− ∂− =
+ ∂
− ∂ ∂ ∂− + +
+ ∂ ∂ ∂
( ) ( )1wall nT aT a T λ=
= + −
( )2
2 2 4 1
1 1 2
t wallT T uwall
T T u p
u uTT T
Pr n c
σ γ λ γ σ σ
σ γ γ σ σΓ
−− ∂ −− = +
+ ∂ + −
4. Slip
14
Modelling Gas Microflows
Navier-Stokes equationswith no-slip B.C.
Navier-Stokes equations
with 1 -order slip B.C.st
QGD or QHD equations
with 1 -order slip B.C.
or
st
Navier-Stokes equations
with 2 -order slip B.C.nd
Burnett equationswith slip B.C.
DSMC or Lattice Boltzmann
Eu
ler
equ
atio
ns
Kn
1010-1
10-3
0
5. Modelling
15
Some Particularities of Gas Microflows:Thermal Creep
� Principle of Knudsen pump
2 3
4
u ss wall
u
u T
Tu u
n t
σλ
σ
µ
ρΓΓ
− ∂ − = +
∂
∂
∂
LT
LT
LT H
TH
TH
TL
T
stage 2 stage Nstage1flow flow
LT
continuum regime
slip or transition
regime
6. Specific Behaviours
16
Some Particularities of Gas Microflows:Thermal Creep
� Knudsen pump with curved parts
HT
HTHT
LT LT
HTLT HT
stage1
stage 2 stage N
HTLT LT
LT LT LT
LT LT
flow
flow
6. Specific Behaviours
17
Thermal Creep
� Simulation using Fluent
+ specific UDFs
1.3
1.5
1.7
1.9
2.1
2.3
2.5
0 5 10 15 20 25
P1*
s*
N = 2N = 4 N = 8
N = 25
2 3
4
v
fluid wall
wallv wall
u v R Tu u
n s P s
σ µα λ
σ
− ∂ ∂ ∂ − = + +
∂ ∂ ∂
2 2
1
Tfluid wall
wallT
TT T
Pr n
σ γ λβ
σ γ
− ∂− =
+ ∂
( )1* av LP P RTρ=
HT
HTHT
LT LT
HTLT HT
stage1
stage 2 stage N
HTLT LT
LT LT LT
LT LT
flow
6. Specific Behaviours
18
Heat Transfer in Slip Flow Regime
� Example
z
y
h−
h+
hϕ
Insulated wall
Uniform heat flux
( )
12
2
* 26 147 * 210 *
4 140 1 3 *Nu
ς ξ ξ
ξ
− + +
= + +
*wall
uu u
nξ
Γ
∂− =
∂
*w
TT T
nς
Γ
∂− =
∂Colin, S., Gas microflows in the slip flow regime: a critical review on
convective heat transfer, Journal of Heat Transfer, 134 (2), 2012.
6. Specific Behaviours
19
Motivation for Experimental Analysis of Gas Microflows
� Wide literature on modelling and numerical simulation of gas microflows, in different rarefaction regimes
� However, few available experimental data
� Crucial need of smart experimental data, for example to:
– Help identifying the best BC to be used in slip flow regime and the limit of applicability of the associated analytical models
– Analyse the influence of surface, which may vary with � different materials – silicon, metals, polymers, glass and fused silica…
� different kind of manufacturing – wet chemical etching, reactive ion etching, laser etching, moulding, embossing, drilling, micromilling…
7. Experimentation
20
Experimental Analysis of Gas Microflows
� Example: gas flow in a microchannel
Main quantities of interest
Mass flowrate
GLOBAL data
Dijkstra, M., et al. (2008) Sensors and Actuators A: Physical,
143, 1-6.
Pressure
Temperature
Velocity
LOCAL data
Concentration
7. Experimentation
21
Flowrate: Accurate Measurement
p Vm
R T=
1 dV dp pV dTm p V
R T dt dt T dt
= + −
ɺ
Tank
(m, p, T, V)
mɺ
µsystem
Constant Pressure Method Constant Volume Method
Droplet Tracking Method
� For low flowrates: need of specific setups
� Basic principle based on the equation of state – suitable for dilute gases
with good
thermal insulation
7. Experimentation
22
Réservoir
amont
Réservoir
aval
MicrosystèmeUpstream
tank
Downstream
tank
Microsystem
( )P t
( )T t
Example of Constant Volume Method
7. Experimentation
23
ICA Mass Flowrate Setup
Pitakarnnop, J., et al. (2010) Microfluidics and Nanofluidics, 8, 57-72.
Reservoir A
Reservoir B
7. Experimentation
24
Kn010-210-1100101
Example of Results: Flowrates of He or Ar
Pitakarnnop, J., et al. (2010) Microfluidics and Nanofluidics, 8, 57-72.
Issue: difficult to separate
the roles of
• the model of BC
• the accommodation coefficient
Interest to access a direct
measurement of the velocity profile
7. Experimentation
25
Main Flowrates Data on Gas Microflows
� Droplet tracking method– Pong, K.-C., et al. (1994) FED-197, ASME, New York, pp. 51-56.
– Harley, J. C., et al. (1995) Journal of Fluid Mechanics, 284, 257-274.
– Zohar, Y., et al. (2002) Journal of Fluid Mechanics, 472, 125-151.
– Maurer, J., et al. (2003) Physics of Fluids, 15, 2613-2621.
– Colin, S., et al. (2004) Heat Transfer Engineering, 25, 23-30.
– Ewart, T., et al. (2006) Experiments in Fluids, 41, 487-498.
� Constant pressure method– Jousten, K., et al. (2002) Metrologia, 39, 519-529.
� Constant volume method– Arkilic, E. B., et al. (1998) Experiments in Fluids, 25, 37-41
– Arkilic, E. B., et aL. (2001) Journal of Fluid Mechanics, 437, 29-43.
– Ewart, T., et al. (2007) Journal of Fluid Mechanics, 584, 337-356.
– Pitakarnnop, J., et al. (2010) Microfluidics and Nanofluidics, 8, 57-72.
– Szalmás, L., et al. (2010) Microfluidics and Nanofluidics, 9, 1103-1114.(mixtures of gases)
7. Experimentation
26
Pressure Measurements – Discrete Data
� Shih, J. C et al. (1996), ASME DSC-59, pp. 197-203.– First local data for gas flows in microchannels
– Channel 4,000×40×1.2 µm3
– He & N2
� Zohar, Y. et al. (2002) Journal of Fluid Mechanics, 472, 125-151.– Detailed measurements
– Channels 4,000×40×(0.53 & 0.97) µm3
– He, Ar & N2
� Jang, J. & Wereley, S. T. (2004) Microfluidics and Nanofluidics, 1, 41-51. – Rectangular channels with higher aspect ratio (0.36)
– Channels ?×105×39 µm3
– Air
� Turner, S. E. et al. (2004) Journal of Heat Transfer, 126, 753-763.– Entrance effects analysed; influence of roughness 0.4 % to 6 %: insignificant
– Channel 30,000×1,000×(2.3 to 50) µm3
– Air
2(He) (N )0.16 or 0.055
oKn =
2(Ar) (N )0.20 or 0.067oKn =
0.0018o
Kn =
0.15oKn =
7. Experimentation
27
Pressure Measurement
� Integrated pressure sensors
Zohar, Y. et al. (2002) Journal of Fluid Mechanics, 472, 125-151.
• Microchannel• length 4000 µm
• width 40 µm
• depth 0.5 or 1 µm
• Capillary connection• width 4 µm
• depth 0.2 µm
7. Experimentation
28
Pressure Measurement
Zohar, Y. et al. (2002) Journal of Fluid Mechanics, 472, 125-151.
Nitrogen flow• channel depth
0.97 µm
• outlet pressure
100 kPa
• accuracy 5 %
Model
• first order slip flow
• plane flow
• diffuse
accommodation
0.067o
Kn =
7. Experimentation
29
Pressure Measurement
Zohar, Y. et al. (2002) Journal of Fluid Mechanics, 472, 125-151.
Shih, J. C et al. (1996), ASME DSC-59, pp. 197-203.
Argon flow• channel depth
0.53 µm
• outlet pressure
100 kPa
• accuracy 5 %
Model
• first order slip flow
• plane flow
• diffuse
accommodation
0.20o
Kn =
Helium flow, channel depth 1.2 µm
7. Experimentation
30
Measurements of Pressure Fieldsat the Wall
� Pressure-sensitive paints (PSP)– Luminescent molecules coated at the wall; once excited, emit at a longer
wavelength
– Luminescent intensity depends on O2 concentration (oxygen quenching phenomenon), related to pressure
– Non intrusive technique - High spatial resolution
– Need calibration and transparent side
– Cannot be used for oxygen free gases
– Too thick for use at microscale� Huang, C. et al. (2007) Journal of Microelectromechanical Systems, 16, 777-785.
� Pressure-sensitive molecular films (PSMF)– Technique developed at Nagoya University
� Mori, H. et al. (2005) Physics of Fluids, 17, 100610.
� Matsuda, Y. et al. (2007) Experiments in Fluids, 42, 543-550.
� Matsuda, Y., et al. (2009) Experiments in Fluids, 47, 1025-1032.
� Matsuda, Y., et al. (2011) Microfluidics and Nanofluidics, 10, 165-171.
7. Experimentation
31
Pressure-Sensitive Molecular Films (PSMF)
� Example of luminophore:
Pt(II) Mesoporphyrin IX
� Langmuir–Blodgett (LB)
deposition method
Matsuda, Y. et al. (2009) Experiments in Fluids, 47, 1025-1032.
7. Experimentation
32
From PSP to PSFM
� Relative luminescent intensity fields - 160 160 µm2 surface
Matsuda, Y. et al. (2011) Microfluidics and Nanofluidics, 10, 165-171.
PSP
Standard deviation 0.23
PSFM
Standard deviation 0.016
7. Experimentation
33
Experimental Results
� Micro-nozzle flow
– 2 configurations
– Pi = 10 kPa
– Po = 1 kPa
Matsuda, Y. et al. (2011) Microfluidics and Nanofluidics, 10, 165-171.
7. Experimentation
34
PSMF – Comparison with DSMC
Pressure distribution Pressure along the centerline
Matsuda, Y. et al. (2011) Microfluidics and Nanofluidics, 10, 165-171.
7. Experimentation
35
PSMF – Comparison with DSMC
Pressure distribution Pressure along the centerline
Matsuda, Y. et al. (2011) Microfluidics and Nanofluidics, 10, 165-171.
7. Experimentation
36
Temperature Measurement
� Various available techniques for measurement at the wall
– Thin film Resistance Thermo Detectors (RTD)
– Thin Film ThermoCouples (TFTC)
� 25x25 µm2 to 80x80 µm2 embedded junction in a 100-150 nm thick film.
20 °C – 900 °C
Zhang, X., et al. (2006) Journal of Micromechanics and Microengineering,
16, 900.
– Semiconducting Sensors (SC)
– Temperature sensitive paint (TSP)
� Promising new techniques for measurement within the flow
– Molecular Tagging Thermometry (MTT)
� Hu, H., et al. (2010) Measurement Science and Technology, 21,085401:1-14
7. Experimentation
37
Velocity Measurement
� Two recent techniques (for gas) currently under
investigation
– Micro Particule Image Velocimetry (µPIV)
– Micro Molecular Tagging Velocimetry (µMTV)
7. Experimentation
38
Principle of Classic PIV
http://www.dantecdynamics.com
7. Experimentation
39
Difference Between PIV and µPIV
http://www.dantecdynamics.com
Meinhart, C. D. et al. (2000)
Measurement Science and Technology, 11, 809-814.
7. Experimentation
40
µPIV for Air Flow in Square Microchannels
Yoon, S. Y. R. et al. (2006) Journal of Power Sources, 160, 1017-1025.
� Square sections
– 1 1 mm2
� Tracers
– smoke particles
– water droplets
� Re = 50 - 820
� spatial resolution
– 40 - 60 µm
� Not in rarefied regimes
7. Experimentation
41
µPIV for Nitrogenin Rectangular Channels
� Rectangular channels
– 1 mm 0.5 mm
� Tracers
– fluorescent oil droplets
– diameter 0.5 to 2 µm
� Re = 26 – 130
� Not in rarefied regimes
Sugii, Y. and Okamoto, K. (2006) In Proceedings of ICNMM2006, Limerick, pp. ICNMM2006-96216:1-6.
7. Experimentation
42
µPIV
• Particles seeding
• The whole test section is illuminated
• Resolution: defined by the optical depth
of field
• Limitations:
• Need of small particles (100- 300 nm)
=> sensitivity to Brownian motion
=> difficult particles pattern detection
• Background noise due to out-off focus
particles
µMTV
• Molecular tracers
• A line of molecules is tagged
• Resolution: defined by the laser beam
diameter
• Non intrusive technique
• Limitations to be clarified
Maynes and Webb, Exp. Fluids, 32(1), 2002
Microtube Ø 705 µm
Velocimetry at Small Scale
t0 t1 u
7. Experimentation
43
Molecular Tagging Velocimetry (MTV) Principle
� Direct UV tagging of specific molecules
– Once excited: immediate fluorescence
– After a delay: phosphorescence
� Efficient with liquids
– Supramolecules
� Only tested with gases in
unconfined flows
– Acetone & Biacetyl
7. Experimentation
44
Micro Molecular Tagging Velocimetry & Thermometry (µMTV-µMTT) in Liquids
� Current data on microflows
only for liquids. Examples:
– µMTV in a Hagen-Poiseuille flow in a fused silica microtube
– µMTV and µMTT in electro osmotic flow
Maynes, D. and Webb, A. R. (2002)
Experiments in Fluids, 32, 3-15.
∆t = 200 µs
Hu, H., et al. (2010) Measurement Science and Technology, 21, 085401:1-14.
∆t = 4.5 ms
7. Experimentation
45
µMTV with Gas: Delicate Choice of Material for the Walls
� Fluorescence images obtained in TSC3 (a) and Suprasil (b)
channels
Samouda, F., et al. (2011) Proceedings of GASMEMS11, Bertinoro, Italy.
(a) (b)
7. Experimentation
46
PEEK® channel
multilayer rectangular channel
1 x 5 mm2 cross-section, 20 cm length
Microchannel Fabrication
Samouda F. (2012) PhD Thesis University of Toulouse
7. Experimentation
47
PEEK® channel
Suprasil® lenses
Microchannel Fabrication
7. Experimentation
48
PEEK® channel
Suprasil® lenses
Borofloat® window
Microchannel Fabrication
7. Experimentation
49
PEEK® channel
Suprasil® lenses
Borofloat® window
O-ring sealing
Microchannel Fabrication
7. Experimentation
50
PEEK® channel
Suprasil® lenses
Borofloat® window
Thermocouple
O-ring sealing
Microchannel Fabrication
7. Experimentation
51
Suprasil® lenses
PEEK® channel
Borofloat® window
PEEK® plate
Thermocouple
O-ring sealing
Microchannel Fabrication
7. Experimentation
52
PEEK® channel
Suprasil® lenses
Borofloat® window
O-ring sealing
Thermocouple
PEEK® plate
Swagelok® connectors
Microchannel Fabrication
7. Experimentation
53
Camera access
Laser access
Microchannel Fabrication
7. Experimentation
54
Inlet reservoir
Thermocouple
Outlet capacitive gauge
Visualization access
Suprasil® lens
Laser access
Flow
Suprasil® window
Thermocouples
Channel inlet Channel outlet
7. Experimentation
55
MTV SYSTEM
7. Experimentation
56
Laser guiding arm
CCD Camera
Intensified Relay Optics (IRO)
Argon
Seeding circuit
IRO Controller
Programmable Timing Unit
Computer
Channel
Laser
• Flow seeding
• Tagging
• Detecting
• Data processing
Velocity profiles measurement
7. Experimentation
57
m=200, n= 50 m=600, n= 50 m=1000, n= 50 m=60, n= 50
Post processing
time
12 s8 s2 s1 s
G
t0
Camera
Laser pulse
Image
acquisition
1 n
Final averaged
image
Average
G
t0
2 1 2 n
1 m
7. Experimentation
58
Data Processing
7. Experimentation
59
Standard averaging technique• t0 = 1 µs (35000 images) • t1 = 80 µs (50000 images)
Good agreement between theoretical and experimental data for the maximum
velocity
Experimental data
Theoretical
profile
Experimental flow conditions • Pmoy = Patm
• dp /dx = 1kPa/m• T= 25 °C• Re = 423
Comparison withTheory
7. Experimentation
60
MTV – First Results & Next Steps
� Preliminary steps in velocity field analysis– µMTV for confined flows: validated for continuum flows
– Towards rarefied flows: exploitable signal at lower pressure
� Reaching the slip flow regime: challenging for measurement
of slip at the wall
� Extensions of the technique to µMTT (thermometry)
– Lifetime and intensity of phosphorescence are a function of temperature
� Extensions of the technique to µMTM (manometry)
– Phosphorescence is quenched by O2
7. Experimentation
61
Conclusions – Future Needs
� A lot of theoretical and numerical investigations on gas
microflows and heat transfer
� No definitive consensus on some models, especially
concerning slip boundary conditions
� Need of smart local experimental data
– Velocity fields
– Temperature fields
– Data on heat transfer
62
Acknowledgements
� Funding from European
Community's 7th
Programme FP7/2007-
2013 under grant
agreement ITN
GASMEMS n°215504
� More information on
http://www.gasmems.eu