Professional Documents
Culture Documents
Evolution of Microchannel
Flow Passages
Thermohydraulic
Performance and
Fabrication Technology
SATISH G. KANDLIKAR
Mechanical Engineering Department, Rochester Institute of Technology, Rochester, NY, USA
WILLIAM J. GRANDE
Microelectronic Engineering Department, Rochester Institute of Technology, Rochester, NY, USA
This paper provides a roadmap of development in the thermal and fabrication aspects of
microchannels as applied in microelectronics and other high heat-flux cooling applications.
Microchannels are defined as flow passages that have hydraulic diameters in the range of 10 to
200 micrometers. The impetus for microchannel research was provided by the pioneering work of
Tuckerman and Pease [1] at Stanford University in the early eighties. Since that time, this
technology has received considerable attention in microelectronics and other major application
areas, such as fuel cell systems and advanced heat sink designs.
After reviewing the advancement in heat transfer technology from a historical perspective, the
advantages of using microchannels in high heat flux cooling applications is discussed, and research
done on various aspects of microchannel heat exchanger performance is reviewed. Single-phase
performance for liquids is still expected to be describable by conventional equations; however, the
gas flow may be influenced by rarefaction effects. Two-phase flow is another topic that is still under
active research.
The evolution of research in microchannel flow passages has paralleled the advancements made
in fabrication technology. The earliest microchannels were built in silicon wafers by anisotropic wet
chemical etching and sawing. While these methods have been exploited successfully, they impose a
number of significant restrictions on channel geometry. A variety of advanced micromachining
techniques have been developed since this early work. The current state of fabrication technology is
reviewed, taxonomically organized, and found to offer many new possibilities for building
microchannels. In particular anisotropic dry etching and other high aspect ratio techniques have
removed many of the process-induced constraints on microchannel design. Other technologies such
as surface micromachining, microstamping, hybridization, and system-on-chip integration will
enable increasingly complex, highly functional heat transfer devices for the foreseeable future. It is
also found that the formation of flow passages with hydraulic diameters below the microchannel
regime will be readily possible with current fabrication techniques.
Address correspondence to Satish G. Kandlikar, Mechanical Engineering Department, Rochester Institute of Technology, Rochester, NY 14618,
USA. E-mail: sgkeme@rit.edu
INTRODUCTION
Historical Advancements in Heat
Transfer Technologies
Heat, or thermal energy, represents the ultimate manifestation of all forms of energy. Transfer of heat from
one location to other, from one medium to another,
and meeting the challenges of accomplishing this transfer under a variety of constraints have been the objectives of heat transfer research ever since fire was
domesticated.
Heat transfer by convection provides a means of
transferring heat quickly away from heat exchange surfaces. Fluids employed in engineering systems and processes undergo the changes in their thermal state in heat
exchangers. The basic equation of heat transfer by convection is expressed as follows:
q = h A(Ts T f )
(1)
Nature provides us with some important clues regarding the heat and mass transfer processes. For example, the African elephants have larger ears than those
in Asiathe higher temperature in the desert environment in Africa requires a larger surface area for the
ears, which are the main heat dissipation devices for
elephants. Looking at the biological systems, such as
a human body, Chen and Helmes [2] found that the
blood vessels that are largely responsible for thermal exchange (known as thermally significant blood vessels)
have sizes on the order of hundreds of micrometers, with
175 m diameter being typical. The mass transfer processes, on the other hand, take place in much smaller
sized vessels, such as alveoliwhich are on the order
of a few micrometersand form the air sacs at extremities of the air passageways in the lungs. The arterioles
and venules, which are the smallest vessels for blood
transportation, are only 10 to 15 m in diameter. The
capillaries, where most of the mass transfer processes
occur, are only 4 m in diameter. The mass transfer effectiveness of these three unitsarterioles, venules, and
capillariesare over three orders of magnitude higher
than the larger vessels (Lightfoot and Duca [3]).
Channel Classification
The thermal scientists in the last two decades took
another look at Eq. (1) in their attempt to address the
challenges posed by the high heat flux devices. High
heat fluxes coupled with the small device sizes led to
smaller channel dimensions. The word micro was embraced enthusiastically with the opening of its newest
branch in microscale heat transfer. The classification of
small channel dimensions, as proposed by Mehendale
et al. [4], divides the range from 1 m to 100 m as microchannels, 100 m to 1 mm as meso-channels, 1 mm
to 6 mm as compact passages, and >6 mm as conventional passages. This classification is based simply on
the dimensions of the channels. The classification provided by Kandlikar [5, 6], further refined below, is based
on the flow considerations.
Conventional channels are identified as those with
a hydraulic diameter of 3 mm or larger. The channel
sizes below 3 mm are formed as narrow fin passages,
as in plate-fin heat exchangers. The regenerative heat
exchanger matrix and plate heat exchangers belong to
this category. The lower limit for manufacturing smaller
channels is really imposed by the major changes in
vol. 24 no. 1 2003
(2)
where is the mean free path for the gas calculated from
the following equation:
(3)
=
2RT
where Rgas constant, J/kgK; dynamic viscosity,
N/ms; density, kg/m3 ; and T is absolute temperature
in K.
Table 1 gives the values of mean free paths for different gases at 300 K. As an example, the mean free path
for air at 300 K is 0.068 m. The microchannel range,
covering 10 m to 200 m, is generally affected by
the rarefaction effects for many gases. The continuum
approach with no wall slip is modified in these channels; this approach being valid for 0.1 > Kn > 0.001, the
region identified as the slip region.
Below 10 m, depending on the gas and the pressure,
the transitional region is encountered, where rarefaction
effects are more severe and approach the molecular flow.
The range 10 > Kn 0.1 m is referred to as the transitional region. Tentatively, we may assign the channel
to a
dimensions from 10 m Dh > 0.1 m (1000 A)
newly defined region as Transitional Nanochannels.
On the basis of the above discussion, the following
classification is presented.
Conventional Channels:
Minichannels:
Microchannels:
Transitional Channels:
Transitional Microchannels:
Transitional Nanochannels:
Molecular Nanochannels:
Dh > 3 mm
3 mm Dh > 200 m
200 m Dh > 10 m
10 m Dh > 0.1 m
10 m Dh > 1 m
1 m Dh > 0.1 m
0.1 m Dh
Table 1
pressure
Gas
T, K
R, J/kg K
, kg/m3
, kg/m s
, m
Air
Helium
Hydrogen
Nitrogen
300
300
300
300
287.0
2077.03
4124.18
296.8
1.1614
0.1625
0.08078
1.1233
1.846 105
1.99 105
8.96 106
1.782 105
0.068
0.194
0.125
0.066
This article intends to provide a historical perspective to the developments in microchannels from different viewpoints. The task is very much complicated by
the fact that the microchannels represent a relatively
new technology in the heat transfer application, tracing
its roots with the pioneering work of Tuckerman and
Pease [1] just over twenty years ago. Active research on
vol. 24 no. 1 2003
2 f G2
pf
=
(5)
L
D
where p f /L is the frictional pressure gradient, f is
the Fanning friction factor, G is the mass flux, is the
density, and D is the hydraulic diameter. For fully developed laminar flow, we can write
f Re = C
THERMOHYDRAULIC PERFORMANCE
OF MICROCHANNELS
h = Nu
However, f varies inversely with Re, since the product f Re remains constant during fully developed laminar flow. The frictional pressure drop per unit length for
the flow of an incompressible fluid is given by
(4)
(6)
Figure 1a Variation of heat transfer coefficient with channel size for a square channel under a laminar flow, constant heat flux boundary
condition, assuming no rarefaction and compressibility effects.
Figure 1b Variation of pressure gradient for fully developed laminar flow in smooth circular tubes at 300 K for water at G = 200 kg/m2 s
(V = 0.20 m/s) and air at 5 kg/m2 s (V = 4.25 m/s), assuming no rarefaction and compressibility effects.
Range of
Knudsen numbers Type of flow
0.001 > Kn
0.1 > Kn > 0.001
10 > Kn > 0.1
Kn > 10
Gas
Continuum
flow
Slip flow
Air
Helium
Hydrogen
>67 m
>194 m
>123 m
0.6767 m
1.94194 m
1.23123 m
Transition
flow
0.00670.67 m
0.01941.94 m
0.01231.23 m
Free
molecular
flow
<0.0067 m
<0.0194 m
<0.0123 m
accounting for the wall effect during flow through parallel channels and rectangular microchannels.
In their formulation, Ebert and Sparrow [12] modeled
the wall slip condition using the first derivative of the
velocity profile at the wall. Aubert and Colin [14] point
out that the Taylor series form of the velocity profile
proposed by Ebert and Sparrow does not converge with
the second order boundary condition and instead used a
second order boundary condition at the wall. Although
their results provided a better fit with the experimental
data, they yield accommodation factors of greater than
1, suggesting that these be treated as empirical constants
at this time.
The wall effects also influence the heat transfer characteristics in microchannels. In the case of fluid flow, the
friction coefficient was reduced, yielding a higher mass
flow rate of gases as compared to the predictions from
conventional correlations. In the case of heat transfer,
a decrease in the heat transfer coefficient is expected
as channel dimensions become smaller or the Knudsen
number increases beyond 0.001.
The approach proposed by Li et al. [15] represents
an alternative model for analyzing the wall effect. The
temperature and velocity jumps at the wall proposed in
the slip flow models are replaced with a continuous variation of viscosity and thermal conductivity in the layer
close to the wall within a several mean free path distance.
Further evaluation of this model is needed before it can
be applied to other systems. This approach, however,
seems to be promising as the wall effects are modeled
through a continuously variable property rather than a
jump condition, which is not clearly understood.
Hadjiconstantinou and Simek [16] analyzed the fully
developed flow in the slip flow and transition regions for
the case of constant wall temperature. They included
the axial heat conduction term in the slip flow model.
Their results indicate that axial conduction increases the
Nusselt number by 10 percent.
Condensation in Microchannels
Surface tension plays an important role as channel dimensions become smaller. An exhaustive review
Figure 2a
Churn flow sequence observed during flow boiling of water in a 200 m square microchannel.
10
11
Figure 4
Figure 5
12
of microns of vertical height [49]. Final parts can be obtained in three distinct ways. First, the patterned resist
can be separated from the substrate and used as a precision machined polymer. Second, the molded deposit,
which is typically an electroformed metal such as nickel
or copper, can be separated from the substrate. Lastly,
the substrate and deposited metal can be used in combination as a high precision molding master. High aspect
ratio electrodeposited features were first obtained using
X-ray exposure by Romankiw and coworkers in the mid1970s [50]. In 1982, Ehrfeld and coworkers recognized
the potential of the process to build molding masters
that could be used to inexpensively create ultraprecise
high aspect ratio parts [51].
None of the technologies described above can individually produce a complete microsystem. Hybridization is the process of combining all the necessary
disparate substrates, structures, components, and subassemblies into a final product. An extremely versatile
variant of hybridization is wafer bonding, where two
flat substrates of nearly arbitrary composition can be
permanently attached. Direct wafer bonding is a collection of processes of which the exact details vary with
material but the technique can generally be tailored to
obtain a wide range of adhesive bond strengths. Plol
and Krauter have extensively reviewed wafer bonding
and its application to microsystems construction [65].
Three bonding techniques of particular interest are
fusion bonding, anodic bonding, and adhesive bonding.
In fusion bonding, two wafers of which the surfaces
are silicon or silicon compounds, such as oxide and nitride, can be covalently bonded through a combination
of chemical surface treatments, pressure, and annealing at elevated temperature. When properly performed,
the bond strength is at least as great as the bulk wafer
strength. Wafer stacks of greater than two wafers can be
bonded in either a serial or parallel fashion. Prior to fusion bonding, the wafers can be extensively machined.
A variety of fusion bonded microsystems with complex internal cavities and moving parts have been realized, including accelerometers [66], microfluidic valves
[67], and micro turbine engines [68]. In anodic bonding, silicon and ionic glass surfaces are joined through a
combination of pressure, temperature, and electric field.
While both fusion and anodic bonding can produce interfaces with great strength, they are quite material specific. For generic hetero-bonding, adhesive techniques
are the most general. Figure 8 shows how microchannel
devices can be formed by wafer bonding.
13
At these dimensions, many of the fabrication technologies described in the preceding section can still be
practiced with little modification, while others must be
modified or even abandoned at very small dimensions.
However, the exploding interest in nanotechnology is
beginning to offer unique fabrication tools for the
nanometer regime.
The dimensional range from 101 m represents
a transition between MEMS fabrication and standard
semiconductor fabrication. In general, both additive
(i.e., deposition) and subtractive (i.e., etching) fabrication technologies become mainstream and widely available at dimensions of a few microns down to approximately 0.1 m. Thus, products based on transitional
channels can readily leverage the global microelectronics manufacturing infrastructure. At ultrasmall dimensions, the salient question becomes: what is technologically possible and what is commercially viable?
Leading edge lithography today is practiced down to
0.13 m and is projected to reach 0.022 m by 2015
[69]. The ultimate feat of patterning was Eigler and
Schweizers spelling out of IBM in xenon atoms using
a scanning tunneling microscope [70]. For deposition
and etching techniques at ultrasmall dimensions, the
main issue is control and accuracy. Advanced processes
such as atomic layer epitaxy and digital etching represent the ultimate in dimensional control and are able to
add or remove monatomic layers. Thus, the practical as
well as physical limit to building molecular nanochannels is the atomic structure of the channel materials
themselves.
CONCLUDING REMARKS
Reducing channel dimensions yields larger surface
area per unit flow volume and a larger heat transfer co14
range has become a focus of interest for thermal transfer research over the last two decades since the work of
Tuckerman and Pease. Over the same time period, the
microfabrication community has also focused on this
dimensional range. Today, a wide variety of specialized
processes and fabrication toolssome derived from traditional machining, some derived from the semiconductor industry, and some clever adaptations of other
technologiesare optimized for the construction of microchannels. Thus, from a fabrication standpoint, there
are virtually no dimensional limits on building channels
for thermal transfer applications.
NOMENCLATURE
A
Dh
f
G
h
Kn
L
Nu
P
R
q
Re
T
Greek
Subscripts
f
s
fluid
surface
REFERENCES
[1] Tuckerman, D. B., and Pease, R. F., High Performance Heat
Sinking for VLSI, IEEE Electronic Device Letters, EDL-2,
pp. 126129, 1981.
[2] Chen, M. M., and Holmes, K. R., Microvascular Contributions in Tissue Heat Transfer, Annual NY Academy of Science,
vol. 325, pp. 137150, 1980.
[3] Lightfoot, E. N., and Duca, K. A., The Roles of Mass Transfer
in Tissue Function, The Biomedical Engineering Handbook,
2d ed., CRC Press and IEEE Press, Section 115, Table 115.8,
2000.
[4] Mehendale, S. S., Jacobi, A. M., and Shah, R. K., Fluid Flow
and Heat Transfer at Micro- and Meso-Scales with Applications to Heat Exchanger Design, Applied Mechanics Review,
vol. 53, pp. 175193, 2000.
[5] Kandlikar, S. G., Heat Transfer, Pressure Drop and Flow Patterns during Flow Boiling in Parallel Channel Compact Heat
Exchangers of Small Hydraulic Diameters, Heat Transfer Engineering, vol. 23, no. 5, pp. 523, 2001.
[6] Kandlikar, S. G., Fundamental Issues Related to Flow Boiling in Minichannels and Microchannels, Experimental Heat
Transfer, Fluid Mechanics and Thermodynamics, Proc. Fifth
World Conference on Experimental Heat Transfer, Fluid Mechanics, and Thermodynamics, Thessaloniki, Greece, vol. 1,
pp. 129146, Sept. 2001.
[7] Richter, M., Woias, P., and Wei, D., Microchannels for Applications in Liquid Dosing and Flow-Rate Measurement, Sensors
and Actuators, vol. A62, pp. 480483, 1997.
[8] Pfahler, J., Harley, J., and Bau, H., Liquid Transport in Micron
and Submicron Channels, Sensors and Actuators, vol. A21
A23, pp. 431434, 1990.
[9] Phillips, R. J., Glicksman, L., and Larson, R., ForcedConvection, Liquid-Cooled, Microchannel Heat Sinks for
High-Power Density Microelectronics, Proc. Intl. Symp. Cooling Technology for Electronic Equipment, Honolulu, Hawaii,
pp. 227248, 1987.
[10] Kandlikar, S. G., Steinke, M. S., and Balasubramanian,
P., Single-Phase Flow Characteristics and Effect of Dissolved Gases on Heat Transfer Near Saturation Conditions
in Microchannels, Paper submitted for presentation at the
IMECE02 conference, New Orleans, Nov. 1622, 2002.
[11] Harley, J. C., Compressible Gas Flows in Microchannels and
Microjets, Ph.D. Thesis, University of Pennsylvania, Philadelphia, PA, 1993.
[12] Ebert, W. A., and Sparrow, E. M., Slip Flow in Rectangular and
Annular Ducts, Journal of Basic Engineering, Transactions of
the ASME, Series D, pp. 10181024, 1965.
[13] Harley, J. C., Huang, Y., Bau, H. H., and Zemel, J. N., Gas
Flows in Microchannels, J. Fluid. Mech., vol. 284, pp. 257
274, 1995.
[14] Aubert, C., and Colin, S., High-Order Boundary Conditions for
Gaseous Flows in Rectangular Microducts, Microscale Thermophysical Engineering, vol. 5, pp. 4154, 2001.
[15] Li, J.-M., Wang, B.-X., and Peng, X.-F., Wall-adjacent Layer
Analysis for Developed-Flow Laminar Heat Transfer of Gases
in Microchannels, International Journal of Heat and Mass
Transfer, vol. 43, pp. 839847, 2000.
[16] Hadjiconstantinou, N. G., and Simek, O., Nusselt Number in
Micro and Nano-Channels, Journal of Heat Transfer, vol. 124,
pp. 356364, 2002.
[17] Kandlikar, S. G., A General Correlation for Saturated TwoPhase Flow Boiling Heat Transfer Inside Horizontal and
Vertical Tubes, ASME Journal of Heat Transfer, vol. 112,
pp. 219228, 1990.
[18] Kandlikar, S. G., and Steinke, M. E., Flow Boiling Heat Transfer Coefficient in MinichannelsCorrelation and Trends,
Proc. 12th International Heat Transfer Conference, Grenoble,
France, Paper #1178, August 2002.
[19] Friedrich, C. R., and Vasile, M. J., Development of the
Micromilling Process for High-Aspect-Ratio Microstructures,
J MEMS, vol. 5, no. 1, pp. 3338, 1996.
[20] See, for example, Kulicke & Soffas website at www.kns.com,
Disco Corps website at www.disco.co.jp, and American Dicing Inc.s website at www.americandicing.com.
15
[21] Masaki, T., Kawata, K., and Masuzawa, T., Micro ElectroDischarge Machining and its Applications, IEEE Micro Electro Mechanical Systems Workshop, Napa Valley, California,
pp. 2126, 1990.
[22] Kuo, C.-L., Masuzawa, T., and Fujino, M., A Micropipe Fabrication Process, IEEE Micro Electro Mechanical Systems Workshop, Nara, Japan, pp. 8085, 1991.
[23] Furutani, K., Enami, T., and Mohri, N., Dot-Matrix Electrical
Discharge Machining for Shaping Fine Structure, Proc. IEEE
10th Annual International Workshop on Micro Electro Mechanical Systems, Nagoya, Japan, pp. 180185, January 26
30, 1997.
[24] Petersen, K. E., Silicon as a Mechanical Material, Proc. IEEE,
vol. 70, no. 5, pp. 420457, May 1982.
[25] Muller, R. S., Howe, R. T., Senturia, S. D., Smith, R. L., and
White, R. M., eds., Microsensors, IEEE Press, New York, NY,
1991.
[26] Ristic, L., ed., Sensor Technology and Devices, Artech House,
London, 1994.
[27] Lang, W., Silicon Microstructuring Technology, Materials Science and Engineering, R17, pp. 155, 1996.
[28] Madou, M, Fundamentals of Microfabrication, CRC Press,
Inc., Boca Raton, FL, 1997.
[29] Trimmer, W. S., ed., Micromechanics and MEMS: Classic
and Seminal Papers to 1990, IEEE Press, New York, NY,
1997.
[30] Kovacs, G. T. A., Micromachined Transducers Sourcebook,
WCB/McGraw-Hill, New York, NY, 1998.
[31] Gad-el-Hak, M. The MEMS Handbook, CRC Press, Inc., Boca
Raton, FL, 2002.
[32] Bassous, E., Fabrication of Novel Three-Dimensional Microstructures by the Anisotropic Etching of (100) and (110)
Silicon, IEEE Trans. Electron Dev., vol. ED-25, no. 10,
pp. 11781185, 1978.
[33] Bean, K. E., Anisotropic Etching of Silicon, IEEE Trans.
Electron Dev., vol. ED-25, no. 10, pp. 11851193,
1978.
[34] Vangbo, M., and Baecklund, Y., Precise Mask Alignment to
the Crystallographic Orientation of Silicon Wafers Using Wet
Anisotropic Etching, J. Micromech. Microeng., vol. 6, pp. 279
284, 1996.
[35] Puers, B., and Sansen, W., Compensation Structures for Convex Corner Micromachining in Silicon, Sensors and Actuators
A, A23, pp. 10361041, 1990.
[36] Klaassen, E. H., et al., Silicon Fusion Bonding and Deep Reactive Ion Etching: A New Technology for Microstructures,
Digest of Technical Papers from Transducers 95/Eurosensors IX, Stockholm, Sweden, vol. 1, pp. 556559, June 2529,
1995.
[37] Fleming, J. G., and Barron, C. C., Characterization and Application of Deep Si Trench Etching, SPIE Conference on Micromachining and Microfabrication Process Technology II, SPIE
vol. 2879, pp. 7379, 1996.
[38] Pandhumsoporn, T., et al., High Etch Rate, Anisotropic Deep
Silicon Plasma Etching for the Fabrication of Microsensors, SPIE Conference on Micromachining and Microfabrication Process Technology II, SPIE vol. 2879, pp. 94102,
1996.
[39] Bhardwaj, J., Ashraf, H., and McGuarrie, A., Dry Silicon
Etching for MEMS, Proc. 1997 Meeting of the Electrochemical Society, Microstructures and Microfabricated Systems III
Symposium, Montreal, PQ, vol. 975, pp. 118130, May 49,
1997.
16
[70] Eigler, D.M., and Schweizer, E.K., Nature 344, April 5, 1990,
pp. 524526.
Satish Kandlikar has been a professor in the Mechanical Engineering Department at RIT for the
last twenty-two years. He received his Ph.D. from
the Indian Institute of Technology in Bombay in
1975, and was a faculty member there before coming to RIT in 1980. His research is mainly focused
in the area of flow boiling. After investigating
the flow boiling phenomenon from an empirical
standpoint, which resulted in widely accepted correlations for different geometries, he started to look at the problem from
a fundamental perspective. Using high speed photography techniques, he
demonstrated that small bubbles are released at a high frequency under flow
conditions. He is also working in the areas of binary flow boiling and bubble
formation in inkjet printing application. He has given a number of invited
and keynote talks nationally and internationally.
William Grande has worked in the field of microfabrication for over two decades. His research
has centered on the fabrication tools, processes,
and devices used in the areas of III-V semiconductor lasers, optoelectronics, microsystems, and
MEMS. He holds the BSEE and BSChE from the
New Jersey Institute of Technology and an M.S.
and Ph.D. in Applied and Engineering Physics
from Cornell University. After ten years of industrial research experience at the IIT Research Institute and Kodak Research
Labs, he joined the faculty of the Department of Microelectronic Engineering at RIT, where his activities are focused on expanding the curriculum and
research effort in Microsystems. Dr. Grande holds 16 patents and founded
Tiger Microsystems, Inc., which specializes in the commercialization of
microsystem-based products.
17