Heat Transfer Engineering, 27(4):41–52, 2006
C Taylor and Francis Group, LLC
Copyright ISSN: 0145-7632 print / 1521-0537 online
DOI: 10.1080/01457630500523774
Development of an Experimental
Facility for Investigating Single-Phase
Liquid Flow in Microchannels
MARK E. STEINKE and SATISH G. KANDLIKAR
Thermal Analysis and Microfluidics Laboratory, Mechanical Engineering Department, Rochester Institute of Technology,
Rochester, New York, USA
J. H. MAGERLEIN and EVAN G. COLGAN
Thomas J. Watson Research Center, IBM Corporation, Yorktown Heights, New York, USA
ALAN D. RAISANEN
Semiconductor & Microelectronics Fabrication Laboratory, Rochester Institute of Technology, Rochester, New York, USA
An experimental facility has been developed to investigate single-phase liquid heat transfer and pressure drop in a variety of
microchannel geometries. The facility is capable of accurately measuring the fluid temperatures, heater surface temperatures,
heat transfer rates, and differential pressure in a test section. A microchannel test section with a silicon substrate is used
to demonstrate the capability of the experimental facility. A copper resistor is fabricated on the backside of the silicon
to provide heat input. Several other small copper resistors are used with a four-point measurement technique to acquire
the heater temperature and calculate surface temperatures. A transparent pyrex cover is bonded to the chip to form the
microchannel flow passages. The details of the experimental facility are presented here. The experimental facility is intended
to support the collection of fundamental data in microchannel flows. It has the capability of optical visualization using
a traditional microscope to see dyes and particles. It is also capable of performing micro-particle image velocimetry in
the microchannels to detect the flow field occurring in the microchannel geometries. The experimental uncertainties have
been carefully evaluated in selecting the equipment used in the experimental facility. The thermohydraulic performance of
microchannels will be studied as a function of channel geometry, heat flux, and liquid flow rate. Some preliminary results for
a test section with a channel width of 100 micrometers, a depth of 200 micrometers, and a fin thickness of 40 micrometers
are presented.
INTRODUCTION
fluxes exceed the present limit of air cooling (approximately
800 kw/m2 ). The majority of available literature and ongoing
research programs are focusing on two-phase flow systems to
perform the high heat flux removal. However, the single-phase
cooling option offers considerable advantages over a two-phase
flow system.
Kandlikar and Grande [1, 2] describe the fabrication techniques and possible trends in microchannels. They remark that
the fluid flow in microchannels is further reaching than just efficient heat transfer, and it could open up completely new fields
not possible only a few years ago.
The upper limit of single-phase liquid-cooled systems is not
clearly established. It is therefore necessary to conduct experiments in an effort to arrive at an acceptable cooling system
The cooling of high heat flux microprocessor chips using
single-phase liquid flow presents an attractive option, as the heat
The fabrication of the test devices was performed at IBM T. J. Watson
Research Center. The device bonding and experimental investigation were performed at Rochester Institute of Technology in the Thermal Analysis and Microfluidics Laboratory. The second author acknowledges the IBM Faculty Award
in support of the work done in this paper. In addition, the support from the Semiconductor and Microsystems Fabrication Laboratory (SMFL) for device bonding
is appreciated.
Address correspondence to Dr. Mark E. Steinke, IBM Corporation, Systems
& Technology Group, Dept. 6T6A / Bldg. 060, 3039 Cornwallis Rd., Research
Triangle Park, NC 27709. E-mail: [email protected]
41
42
M. E. STEINKE ET AL.
performance and predict that performance. Establishing this
boundary will also help find ways to extend this limit with other
advanced techniques, such as incorporating enhanced structures
within microchannel flow passages.
The single-phase cooling system will have less system complexity and variability resulting from flow boiling and smaller
pressure drops compared to a two-phase flow system. In addition, the heat rejection side of the system will also have less complexity because the need for a condensation process is absent.
Steinke and Kandlikar [3] identified single-phase heat transfer enhancement techniques for use in microchannels and minichannels. They speculate that this increase in heat transfer performance from these techniques could place a single-phase
liquid system in competition with a two-phase system, thus simplifying the overall complexity and reliability. However, they
point out that the added pressure drop resulting from the techniques should be carefully evaluated.
An experimental test facility that allows for the determination
of important parameters influencing heat transfer and fluid flow
in microchannels and determining the associated experimental
uncertainties is desired. The facility should also have the ability to incorporate a variety of visualization techniques that are
helpful in microfluidics.
using an inert gas, usually nitrogen. The other method is to use a
pump. The style of pump used in the test loops varies; however,
the most desirable feature for all of them is a steady flow rate.
The most common pump used for steady flow is a gear pump.
The gear pump is a positive displacement pump that delivers a
constant volume of fluid at a steady rate, despite changes in upstream pressure. By contrast, a centrifugal pump’s flow rate depends upon the upstream pressure and therefore is less desirable.
A wide range of Reynolds numbers and heat fluxes are encountered in single-phase microchannel flow. A test facility that
produces a wide range of flow rates would be desirable. In addition, an appropriate power supply for delivering high current
would be required to generate large heat fluxes.
The objective of the present work is to develop an experimental facility capable of providing accurate data for single-phase
liquid heat transfer and pressure drop in microchannels. The
experimental system should provide accurate measurements to
generate microchannel heat transfer and pressure drop data. Additionally, the test section should provide the ability to perform
microchannel flow visualization. The development of the test
section for a variety of fluid flow, heat transfer, and geometry
arrangements will be performed.
EXPERIMENTAL TEST FACILITY
LITERATURE REVIEW
There are over 150 papers that address the single-phase flow
of liquids in microchannels. Bailey et al. [4] provided a review
of heat transfer and pressure drop in microgeometries, with a
main focus on pressure drop occurring in the microgeometry.
Palm [5] presented a review of single-phase and two-phase flow
in microchannels and identified some of the important parameters that govern the behavior. Palm concluded that more work
is needed to advance the understanding of fluid flow and heat
transfer in microchannels. Sobhan and Garimella [6] performed
a comparative analysis of some of the existing works on microchannels. They cite the need for further study on all fronts
to allow for the explanation of the discrepancies reported in the
literature. Recently, Morini [7] performed a literature review
and focused on the pressure drop and friction factor correlations
available. They also point to the need for more experimental
work to develop a greater fundamental understanding.
Table 1 shows selected works that have acquired data for
heat transfer and pressure drop in microchannels. The most wellknown work, by Tuckerman and Pease [8], is often considered to
be the pioneering study in microchannels. They used a pressure
vessel to drive the flow through the microchannels. There are several parameters for each work reported in the table, including the
hydraulic diameter, Reynolds number, mass flux, and heat flux.
There are many methods for driving the flow through a microchannel. The two most common methods used are a pressurized vessel and a pump. The pressure vessels are typically
containers that hold the working fluid and can be pressurized
heat transfer engineering
The experimental test facility contains all of the supporting equipment that supplies a metered working fluid to the
test section and the measurement of heat transfer and pressure
drop. Figure 1 shows the design schematic for the experimental
system.
General Description
The flow loop can be operated in an open-system or a closedsystem mode. (The type of experiment conducted determines
which is more appropriate.) The system represented in Figure 1
is an open flow loop that begins with a polycarbonate supply
tank.
A positive displacement, micro-gear pump drives the flow
through the loop. The gear pump is a Micropump brand pump
that delivers 1 to 300 mL/min of flow at a maximum pressure of
4.8 bars. This style of pump is chosen to give a constant flow rate
at the test section regardless of the pressure drop occurring in it.
This is because the pressure drop can significantly change due to
different microchannel geometries being tested. The micropump
is chosen because of its small footprint, accurate flow rates, and
ability to dispense small volumes.
A membrane filter follows the pump. The filter is housed
in a stainless steel casing that can maintain high pressures and
temperatures. The membrane used in the filter houses a standard
47 mm-diameter, replaceable nylon filter. There is a variety of
pore sizes readily available, ranging from 0.2 µm to 25 µm in
diameter. Typically, a 1.0 µm pore size is used in the filter.
vol. 27 no. 4 2006
M. E. STEINKE ET AL.
43
Table 1 Selected literature for single-phase liquid flow in microchannel passages
Author
Year
Fluid
Shape
Dh (µm)
Re
G (kg/m2 s)
q (W/cm2 )
Method for driving flow
Tuckerman & Pease [8]
Missaggia et al. [9]
Riddle et al. [10]
Gui & Scaringe [11]
Peng & Peterson [12]
Vidmar & Barker [13]
Ravigururajan &
Drost [14]
Lee et al. [15]
Qu & Mudawar [16]
1981
1989
1991
1995
1996
1998
1999
Water
Water
Water
Water
Water
Water
R124
Rectangular
Rectangular
Rectangular
Trapezoid
Rectangular
Circular
Rectangular
92–96
160
86–96
338–388
133–200
131
425
291–638
2350
96–982
834–9955
136–794
2452–7194
135–1279
2615–5678
12463
950–9545
2110–21937
579–4448
16005–46954
3.2–30.6
187–790
100
100–2500
12–112
5–45
506–2737
0.8–13
Pressure
Pressure
Pressure
Pressure
Pump
Pump
Pump
2002
2002
Water
Water
Rectangular
Rectangular
85
349
119–989
137–1670
1196–9944
335–4093
35
100–200
Pump
Pump
The next item in the flow loop is a flow meter bank that
contains flow meters in a parallel arrangement with different flow
rate ranges. Each flow meter has a different range to improve
the accuracy of the flow measurement. The advertised accuracy
of the rotameter is 2% of full scale. However, the measured
accuracy is actually 1.25% for the bottom range and less than
0.5% for the mid- and upper-flow ranges. The flow meters are
a rotameter type. For the flow range of the pump, the rotameter
gives reasonable accuracy for a moderate cost. For even lower
flow rates, a coriolis flow meter is utilized. The flow meters
have an overall flow range of 0.01 mL/min to 4,000 mL/min.
The individual flow meter ranges usually increase by an order
of magnitude. In other words, one flow meter would have a 3 to
75 mL/min flow range, and the next available flow meter would
have a 30 to 300 mL/min flow range.
A miniature shell and tube heat exchanger and a recirculation bath are used to control the fluid temperature entering the
test section. The tube diameters are 3 mm, and the entire heat
exchanger is constructed from stainless steel. The inlet heater is
designed to heat the working fluid up to 90◦ C. A typical load for
the heat exchanger is 100 W.
The experimental system is instrumental to measure temperatures, pressures, and flow rates. Integrated resistors on the test
device are used to measure local temperatures. They are formed
by depositing copper in a very thin line (more details given in a
later section). The remaining system temperatures are read using
thermocouples due to the inexpensive yet accurate measurement
capabilities. E type thermocouples are chosen for their accuracy
and their temperature range.
Pressure is measured using a differential pressure transducer
from Omega. The pressure transducer range of 0.02 to 6.8 bars
is selected to give the highest level of accuracy.
Flow Visualization
Studying the fluid flow in microchannels can be achieved
in part by visualizing the flow field. There are two main
Figure 1 Experimental flow circuit. Open system mode shown.
heat transfer engineering
vol. 27 no. 4 2006
44
M. E. STEINKE ET AL.
visualization techniques that can be employed with the experimental test facility by utilizing an optically clear microchannel
wall.
A standard optical microscope has been included in the experimental facility. Long distance objectives have been incorporated
to increase the working distance. A high-speed CCD camera is
attached to the microscope to capture images of the flow.
A micro-particle image velocimetry (µPIV) system is utilized
with a visualization test section to investigate the velocity field
in the microchannel. A high frame rate camera is utilized with
this system to visualize higher Reynolds number flows. The velocity vectors in the microchannel can be determined using this
technique. The flow distribution in the microchannels, the flow
in the header region, and the entrance length in the microchannel
are some of the aspects that can be studied with this experimental
facility.
TEST SECTION DESIGN
The design of the test section used to conduct the experiments
in the present work is discussed in this section. The test section
is located in the test flow circuit and provides support for the test
device. The test fixture is part of the test section and provides a
platform to contain the electrical and fluidic connections.
Test Device
The test device contains the microchannels, the heater, and
the temperature sensors. The geometries of the microchannels
that have been fabricated are presented in Table 2. Figure 2
shows the location of the geometry variables used in Table 2.
The parameters are the microchannel width (a), fin thickness
separating the microchannels (s), depth of the microchannel (b),
the total flow length of the microchannel (L), and the number of
microchannels (n).
There are twelve different microchannel configurations selected for the study. The channels are 10 mm in total length
and 8 mm wide. The number of channels varies to completely
fill the 8 mm width. The resulting number of channels varies
Table 2 Channel geometries of the test device
Number
a (µm)
s (µm)
Pitch (µm)
n
L (µm)
G1-001
G1-002
G1-003
G1-004
G1-005
G1-006
G1-007
G1-008
G1-009
G1-010
G1-011
G1-012
40
40
70
70
100
100
200
200
200
250
250
250
40
100
40
100
40
100
40
100
200
40
100
200
80
140
110
170
140
200
240
300
400
290
350
450
100
57
73
47
57
40
33
27
20
28
23
18
10000
10000
10000
10000
10000
10000
10000
10000
10000
10000
10000
10000
heat transfer engineering
Figure 2 Geometric parameters for the microchannels.
between 18 and 100. The test section is fabricated using two
different layers that are bonded with an adhesive. The first layer
is a silicon substrate that contains the microchannels, the heater,
and the temperature sensors. The second layer is made of Pyrex
7740 and contains the fluidic connections.
The microchannels are fabricated using deep reactive ion
etching (DRIE). A very high aspect ratio with straight sidewalls
can be achieved using this method. Figure 3a shows an example
of the microchannels etched into the silicon. There are also inlet
and outlet headers formed in the silicon. The back side of the
header is rounded to help direct the flow and minimize stagnation regions. Finally, the entrance to each channel is rounded to
reduce the pressure drop due to the entrance. The radius is one
half the fin thickness, s.
An insulating layer of silicon nitride is deposited on the backside of the device. Then, a layer of copper is deposited and patterned on the backside to form the heater and sense resistors.
Finally, a nitride cap is deposited to help minimize the corrosion
of the copper. The copper is patterned into a long serpentine resistor for the heater. The total length is 82 mm with a resistance
of 9.7 ohms. The current is controlled to allow for the introduction of a targeted heat flux of 200 W/cm2 . The current supplied
to the heater can be as large as five amps.
The six-temperature sensor resistors are kept very small to
increase the spatial resolution. A constant current is supplied to
each sense resistor. The resistance of the sensor is a function of
temperature allowing for the temperature to be determined. Each
sense resistor requires four connections: two for supply current,
and two for potential reading. The sensor will return a lineaveraged temperature along the length of the resistor. Figure 3b
shows the backside of the device with the heater, sensors, and
bonding pads.
To discuss the location of features on the test device, let the
origin be located at the center of the device. There are six temperature sensors located on the backside of the device. The locations of the sensors are at −4.750 mm, −3.375 mm, −1.875 mm,
0.375 mm, 3.375 mm, and 4.750 mm. The first and last sensors
vol. 27 no. 4 2006
M. E. STEINKE ET AL.
45
Figure 3 Microchannels in silicon substrate: (a) microchannels etched into front side of silicon; (b) electrical layout of copper back side.
are placed beneath the inlet and outlet plenums. The remaining
four sensors are interlaced within the serpentine of the heater.
The silicon substrate contains the microchannels with three
walls. A pyrex cover piece is placed on top of the silicon substrate to form the complete microchannel test device. The cover
piece has the same overall dimensions of the silicon substrate.
The thickness of the cover plate is 750 µm. The cover piece is
laser drilled with two holes to match with the location of the inlet
and outlet headers. The diameter of the plenum hole is 1.5 mm.
Adhesive Bonding
Adhesive bonding is the method chosen to attach the pyrex
cover to the silicon substrate. This method provides wide latitude with the surface preparation as well as surface roughness.
However, both parameters are still important in the success of
the bond interface. The adhesive used is M-Bond 610. It is a
two-component, solvent-thinned, epoxy-phenolic adhesive. The
operational temperature range is from −269◦ C to 370◦ C for a
short period, and the long term temperature range is −269◦ C to
260◦ C. The approximate composition is 60% tetrahydrofuran,
30% bisphenol F epoxy resin, and 10% methyl ethyl ketone.
device retaining plate, and the mounting plate. The pogo probes
are inserted into the pogo probe block. The pogo probes are a
high current version that can handle up to ten amps. Figure 4
shows the pogo probes inserted into their respective block; their
pattern there matches the contact pad pattern in Figure 3b. The
pogo probe block is inserted into the main block. Figure 5 shows
the test section assembly.
A critical part in the test fixture assembly is the deviceretaining plate shown in Figure 6. This part provides a clamping
force on the microchannel test section and the fluidic connections, and is made of an optically clear polycarbonate material. The fluid enters the piece and moves into an inlet plenum.
The fluid then enters the test device and leaves through the exit
plenum. The device is sealed using two double sealing O-rings.
Two miniature thermocouples are fabricated and placed in the
plenums. The thermocouple bead is 0.5 mm in order to minimize
the effect on the flow. They are also coated with an epoxy to seal
them and prevent corrosion.
Pressure ports are located in each of the plenums to get the
pressure drop across the test device. The plenums themselves
are carefully designed to allow for sufficient volume to reduce
Test Fixture
The test fixture is the part of the test section that provides the
support structure for the microchannel test section. The fixture
houses the electrical and fluidic connections. The material used
for the test fixture is black delrin. Delrin is an acetal resin from
DuPont—it is a lightweight plastic that is very stable and does
not experience significant creep.
The test fixture is comprised of the following major parts: the
pogo probe block, the main block, the bottom retaining plate, the
heat transfer engineering
Figure 4 Pogo probe block with probes inserted.
vol. 27 no. 4 2006
46
M. E. STEINKE ET AL.
Figure 5 Test fixture assembly.
the fluid flow velocity. This ensures that the measured pressure
drop is indeed measuring the change in static pressure and has no
dynamic pressure component included with the measurement.
Experiments are conducted to ensure that true static pressure is
measured.
EXPERIMENTAL UNCERTAINTY
The experimental uncertainties can become quite large for a
microchannel heat exchanger because the magnitudes of some
of the measurements are very small. In addition, the propagation
of errors in the system can become troublesome.
The physical size of the system being measured presents a
challenge. It is very difficult to fabricate a thermocouple small
enough to have any significant size resolution and attach it to the
chip at a desired location. A better approach is to use a resistance
element that has a known temperature dependence.
The magnitudes of the measurements represent a problem as
well. The heat transfer in microchannels is very efficient; therefore, the changes in temperature or the temperature difference
can be very small. The T can be on the order of only a few degrees. A typical experimental uncertainty value for temperature
is ±0.5◦ C on a two point calibration. An experimental uncertainty of ±0.05◦ C can be achieved by using the data collection
procedure presented later.
Fortunately, several of the standards for experimental uncertainties still apply. The two best standards for determining
experimental uncertainties are ASME PTC 19.1 [17] and NIST
Technical Note 1297 [18]. There are many similarities between
these standards and others. The primary standard discussed here
will be the ASME standard. In general, the total uncertainty is
comprised of two parts, bias error and precision error, which are
given by:
B 2
σ 2
U=2
+ √
(1)
2
N
where U is the total uncertainty, B is the bias error, σ is the
standard deviation, and N is the number of samples. The bias
error is a measure of the systematic error, and the precision
error is a measure of the random errors in the system.
Figure 6 Device retaining plate.
heat transfer engineering
vol. 27 no. 4 2006
M. E. STEINKE ET AL.
Table 3 Uncertainties for mean of experimental data
Parameter
Mean
Uncertainty (%)
Q (mL/min)
G (kg/m2 s)
Re
q (W/cm2 )
T (◦ C)
θ (◦ C cm2 /W)
p (kPa)
f
40
741
212
29
40
0.4
2.9
0.4
0.5
5.1
6.1
3.3
0.2
6.0
3.0
6.5
47
height. Therefore, even with the most careful pressure drop measurements, there is still going to be a large uncertainty due to
the measurement errors of the microchannel dimensions and the
flow rate.
SYSTEM CALIBRATION AND EXPERIMENTAL
PROCEDURE
There are several rules to follow when propagating errors
from a measured variable to calculated values. In general, Eq. (2)
gives the uncertainty of a calculated parameter.
n 2
∂ p
Up =
u σi
(2)
∂σi
i=1
where p is the calculated parameter. The uncertainty in any parameter is the sum of the uncertainties of the components used
to calculate that parameter.
A measurement technique is utilized to reduce the experimental uncertainties. In a data burst DAQ (data acquisition) mode,
several hundred samples of the same quantity, such as temperature, are measured at a high frequency. The mean and standard
deviation of the sample set is determined and used to report the
parameter being measured. The systematic errors are ones that
result from biases in the experimental setup and can be minimized by using calibrations. These errors are typically repeatable
and can be eliminated. The uncertainties for the major parameters as well as the parameter mean values are presented in Table 3.
Steinke and Kandlikar [19, 20] present a more detailed discussion of the experimental uncertainties occurring in microchannel
pressure drop and heat transfer. The uncertainty for calculated
parameters is derived in terms of the measured parameters. This
allows for the contribution from each measured parameter to be
determined.
The following equation is recommended for use in microchannel fluid flow. The uncertainty in the f Re product is shown
in Eq. (3) [19]:
1/2
 2 2 Up 2
Uµ
Uρ
+
+

2 ·
µ

 ρ
2 p 2


2
U f Re
UL
UQ
Ua


= +

+3·
+5·


f Re
L
Q
a

2
2
2


Ub
Ua
Ub
+5 ·
+2·
+2·
b
a+b
a+b
(3)
where Q is the volumetric flow rate, ρ is the density, µ is the
viscosity, p is the pressure drop, a is the microchannel width,
b is the microchannel depth, and L is the microchannel length.
It can be seen that the most dominant terms in the fRe uncertainty are the measurements of the microchannel width and
heat transfer engineering
The approach used for the system calibration and the experimental procedure are described in this section. The three main
topics include the system calibration, the experimental procedure, and the preparation of the working fluid.
Experimental Calibration
The experimental system requires calibration and some specific data acquisition techniques to generate the most accurate
data possible for a given system configuration. The thermocouples, temperature sensors, pressure transducers, flow meters, and
pump all require calibration and will be discussed in detail.
The thermocouples are calibrated using a heated block and an
ice bath. The ice bath gives a reference temperature of 0◦ C, and a
calibrated hot block from Omega is used to give the other known
temperature points. The range of operation for the thermocouple calibrator is 40◦ C to 480◦ C. The operating range for the
experimental system is from 0◦ C to 200◦ C. The thermocouples
are calibrated using twenty points chosen within the operational
range of the system. A linear curve fit is assumed and applied
to the collected points to determine the calibration equation. Finally, the calibration process is repeated using the equation that
was determined. The temperature measurements are verified in
this manner, and excellent agreement is found using this method.
All of the temperatures are measured to within ±0.03◦ C.
The temperature sensors on the test device are calibrated by
applying a known temperature to the device. Recall that these
sensors are actually copper resistors and the resistance is a function of temperature. The efficient heat transfer that occurs in
the microchannels will be used to our advantage. Temperaturecontrolled water is passed through the test device at a moderate
flow rate. A sufficient amount of time is allowed to let the test
device reach the fluid temperature, within ±0.05◦ C. Several external thermocouples are then used to ensure that an accurate
temperature and steady-state temperature are achieved. Finally,
the potential is measured for that applied temperature. A linear
curve fit is assumed and applied to the data. Once again, the
calibration is repeated with the calibration equation to verify the
accuracy of the fit. The test devices are calibrated prior to and
just after experimentation to ensure good linearity.
The pressure transducers are calibrated using known pressures and the measured response of the transducer. A pressure calibrator from Omega is used to apply a known value
of pressure. The range of the pressure calibrator is −100 kPa to
200 kPa. The high side pressure port on the differential pressure
vol. 27 no. 4 2006
48
M. E. STEINKE ET AL.
transducer is exposed to the known values of pressure. Over
twenty points are taken within the range of the specific pressure
transducer. A linear curve fit is assumed and used to generate
the calibration equation.
The flow meters and pump are calibrated at the same time.
The pump is set for a specific flow rate, and the fluid is collected
in a flask and timed. Then, the mass of the fluid is measured on
a calibrated balance for that period of time. Several points are
taken over the range of operation of the pump and for each of
the flow meter ranges. A polynomial curve fit is applied to get
the calibration equation for the pump and flow meters.
Experimental Procedure
A DAQ system monitors several thermocouples, pressure
transducers, and sensing resistors. In addition, the data acquisition system controls the pump used to drive flow in the system.
The DAQ is based upon the signal conditioning SCXI system
from National Instruments. This system conditions all incoming
signals and enables the possibility to have high channel counts.
The DAQ system used in the experimental system is capable
of sampling at a rate of 100,000 samples per second. The present
work is focused on steady-state performance; however, the fast
sampling will be used to increase the accuracy. The procedure for
data collection begins with the sampling rate. The channels are
sampled at a rate of 1.0 kHz, and the data collected is in the form
of a waveform. The channel information contains both amplitude
and frequency information. A thousand samples for each channel
are collected. The data are analyzed to determine the mean and
the standard deviation, σ, of the data set. Then, the mean and
twice the standard deviation, 2σ, are recorded. The calibration
for each measurement is applied offline of the data collection.
A test device is loaded in the test fixture. First, the top retaining plate is used to create the fluidic connections. Then, the
electrical connections are made via the pogo probes. The flow
loop is brought to the desired inlet temperature by using a test
section bypass. Next, the flow enters the test device and the input power is applied. After all of the data have been collected
at the specified heat flux, the input power is changed while the
flow rate remains constant. Therefore, the data are collected by
varying the heat flux for a fixed mass flow rate.
The system reaches steady state before the data collection
begins. The measured values are considered to be at steady state
if the values do not fluctuate more than the uncertainty values.
Typically, the data collection begins five minutes after the system
reaches steady state. The data are a time-averaged mean of over
5,000 points.
Working Fluid Preparation
The working fluid for the present work is distilled, de-ionized,
and degassed water. The present work is focused on generating
fundamental single-phase flow data for a variety of microchannel geometries. It is therefore important to eliminate all of the
heat transfer engineering
interfering variables. The amount of dissolved gas in the water
will affect the heat transfer performance at very large Ts. The
non-condensables can outgas during the experiments and cause
heat transfer enhancement.
The amount of dissolved gas in the water needs to be precisely
controlled to eliminate the heat transfer changes resulting from
the out-gassing of dissolved gases. Steinke and Kandlikar [21]
conducted an experimental investigation concerning the control
of dissolved gases. They demonstrated that the effects due to the
out-gassing of dissolved gases can be eliminated if the water is
treated to reduce the dissolved oxygen content to 5.4 parts per
million (ppm) at 25◦ C. To be conservative, the dissolved oxygen level used in these experiments is maintained at 3.2 ppm.
The water is checked randomly throughout the course of the
experiments.
A brief outline of the degassing procedure is given. A pressure vessel is filled with de-ionized, distilled water. The vessel is
heated to generate steam. A deadweight corresponding to 1atm
of pressure is applied. When the vessel reaches the desired pressure, the deadweight is removed. A vigorous boil results from
the sudden change in pressure. As a result, the dissolved gases
are released from the water.
EXPERIMENTAL RESULTS
Some experimental results obtained from the single-phase
experimental facility will be presented to demonstrate the parameters that can be investigated with the system. The results
shown are for a microchannel with a width of 250 µm, a depth of
200 µm, and a fin thickness of 200 µm. The heat transfer performance will be characterized in terms of the heat flux, temperature
difference, and unit thermal resistance of the microchannel test
section. The pressure drop performance will be characterized
in terms of the total pressure drop and apparent friction factor.
Table 4 presents the range of parameters investigated for this
microchannel geometry.
The following results do not reflect the maximum performance for a silicon microchannel. The smaller pitches and microchannel widths will provide a much higher heat transfer
performance. There are many papers that present better performance. For example, Colgan et al. [22] investigated the performance of silicon microchannels with some enhancement features and dissipated more than 2,750 kW/m2 . The test section
Table 4 Range of experimental data
Parameter
Q (mL/min)
G ( kg/m2 s)
Re
q (kW/m2 )
T (◦ C)
θ (◦ C m2 /W)
p (kPa {psi})
f
vol. 27 no. 4 2006
Min
Max
12
219
61
34.9
24.8
4100
0.7 {0.10}
0.11
68
1262
364
548.9
54.6
16900
4.8 {0.70}
0.8
M. E. STEINKE ET AL.
49
The slope of the line in Figure 7 is the unit thermal resistance
of the test section. Figure 8 shows the unit thermal resistance for
the different mass fluxes. The unit thermal resistance is a common parameter in the electronics cooling industry. It is meant
to give a direct comparison of performance for different cooling techniques, and a lower value is desired. The unit thermal
resistance can be calculated using Eq. (4):
θ=
T j
(T j − Ti )
=
q q (4)
used has the coarsest pitch and the widest microchannels. The
intention is to demonstrate some of the parameters that can
be determined from the experimental facility and associated
uncertainty.
Figure 7 shows the difference between the average temperature of the resistive sensor on the back side of the test section
and the inlet water temperature versus heat flux. The data are
represented by points and the uncertainty is shown as error bars.
As expected, the temperature difference increases linearly with
heat flux and decreases with an increasing mass flux.
For a mass flux of 1262 kg/m2 s and a Tavg of 23.5◦ C, the
unit thermal resistance is 0.40◦ C cm2 /W and the thermal resistance would be 0.15◦ C/W. A typical air-cooled heat sink thermal
resistance alone would be approximately 1◦ C/W. This does not
take into account the thermal resistances of the silicon, thermal
interface material, contact resistances, and heat spreader; therefore, this technique is already an improvement over air-cooled
heat sinks. A typical two-phase system would have a thermal
resistance that would be approximately 0.10◦ C/W. Once again,
the present work does not attempt to get maximum performance
from the microchannels. However, the thermal resistances of the
single-phase and two-phase systems can be on the same order
of magnitude.
The heat flux versus Reynolds number is shown in Figure 9.
The junction temperatures were limited to 55◦ C to minimize
corrosion of the copper resistors. The heat flux was adjusted for
each mass flux to obtain the data as long as all of the junction
temperatures remained less than 50◦ C. As a result, the number
of data points collected for each mass flux is not the same. As
expected, the higher mass fluxes can support higher heat fluxes.
The largest heat flux dissipated is 54.9 W/cm2 . An increase in the
maximum heat flux is demonstrated with an increase in Reynolds
number.
Figure 8 Unit thermal resistance vs. heat flux, G01–012: a = 250 µm; b = 200
µm; G = 219, 342, 513, 726, 977, 1262 kg/m2 s; Re = 61, 91, 132, 187, 250, 324.
Figure 9 Heat flux vs. Reynolds number, G01–012: a = 250 µm; b = 200 µm;
G = 219, 342, 513, 726, 977, 1262 kg/m2 s; Re = 61, 91, 132, 187, 250, 324.
Figure 7 Temperature difference vs. heat flux, G01–012: a = 250 µm;
b = 200 µm; G = 219, 342, 513, 726, 977, 1262 kg/m2 s; Re = 61, 91, 132,
187, 250, 324.
heat transfer engineering
vol. 27 no. 4 2006
50
M. E. STEINKE ET AL.
Figure 10 Pressure drop vs. Reynolds number, G01–012: a = 250 µm;
b = 200 µm; G = 219, 342, 513, 726, 977, 1262 kg/m2 s; Re = 61, 91, 132,
187, 250, 324.
The pressure drop in the test section is measured in the inlet
and outlet plenums. The entrance and exit losses are determined
and subtracted from the measurement to determine only the pressure drop occurring in the microchannel. Figure 10 shows the
resulting pressure drop in the microchannels. As expected, the
pressure drop increases with the Reynolds number. The small
variation seen in the Reynolds number is due to the water properties being calculated using the mean water temperature. The
mean temperature will increase as the outlet temperature rises
with an increase in heat flux.
The apparent friction factor can now be calculated from the
pressure drop. The inlet and exit losses are removed from the
measured pressure drop. The conventional expansion and constriction area correlations were used. The apparent friction factor
is calculated using Eq. (5) [23]:
p =
2( f app Re)µV̄ L
Dh2
(5)
where f app is the apparent friction factor, Re is the Reynolds
number, µ is the viscosity, V̄ is the mean velocity, L is the microchannel length, and Dh is the hydraulic diameter. The theoretical value for f Re can be used to normalize the friction factor data
to give the non-dimensional f Re ratio, C ∗ . The non-dimensional
f Re ratio versus Reynolds number is shown in Figure 11. This
is a more appropriate method to compare the theoretical prediction of the apparent friction factor [19]. A C ∗ value of 1.0 would
be a perfect match between experimental data and theoretical
value.
There seems to be good general agreement with the friction
∗
factor data. The Capp
ratio takes into account the developing
flows in the microchannels. The experimental uncertainties are
shown as error bars in the figure. A more detailed discussion
on the friction factors and the procedure for correcting for the
heat transfer engineering
Figure 11 C∗app vs. Reynolds number, G01–012: a = 250 µm; b = 200 µm;
G = 219, 342, 513, 726, 977, 1262 kg/m2 s; Re = 61, 91, 132, 187, 250, 324.
entrance and exit losses and the developing flow region are presented by Steinke and Kandlikar [19].
CONCLUDING REMARKS
A single-phase experimental test facility has been designed
and fabricated. The system is capable of delivering a working
fluid to a test section with a variety of inlet temperatures and
flow rates. A wide range of mass fluxes and heat fluxes can be
applied. The equipment has been carefully selected to give the
maximum amount of accuracy in the flow ranges.
The experimental facility also has the ability to perform flow
visualization using a traditional optical microscope and a microparticle image velocity system.
The system can deliver fluid from a flow range of 1 mL/min
to 4,000 mL/min. It can also provide an input power to the test
section of 1.5 kW.
The experimental uncertainties have been carefully controlled
to produce experimental data with very low uncertainty. Steinke
and Kandlikar [19, 20] present a more detailed discussion of the
experimental uncertainties occurring in the microchannel pressure drop and heat transfer.
The experimental results shown represent some of the parameters that can be determined using the single-phase test facility.
The facility can now be used to determine the performance of
single-phase microchannels.
The unit thermal resistance shown in Figure 7 seems to be
dependent upon the Reynolds number for low mass fluxes and
independent of the heat flux at the higher Reynolds numbers.
The diabatic friction factors are in agreement with conventional laminar Poiseuille flow theory after accounting for the
entrance and exit losses and developing region effects.
vol. 27 no. 4 2006
M. E. STEINKE ET AL.
NOMENCLATURE
a
b
B
Dh
f
G
L
n
N
p
P
Q
q
q
R
Re
s
T
ui
U
V̄
channel width, m
channel height, m
systematic error
hydraulic diameter, m
Fanning friction factor
mass flux, kg m−2 s−1
channel length, m
number of channels
number of measurements in sample
pressure, Pa
power, W
volumetric flow rate, L s−1
heat transfer rate, W
heat flux, W m−2
thermal resistance, ◦ C W−1
Reynolds number
fin thickness, m
temperature, ◦ C
uncertainty of parameter i
uncertainty
mean velocity, m s−1
Greek Symbols
η
µ
ρ
σ
θ
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
difference
fin efficiency
viscosity, N s m−2
density, kg m−3
standard deviation
unit thermal resistance ◦ C m2 W−1
Subscripts
[11]
[12]
[13]
app apparent
avg average
f
fluid
h
heater
i
inlet
o
outlet
s
surface
[14]
[15]
REFERENCES
[16]
[1] Kandlikar, S. G., and Grande, W. J., Evolution of Microchannel
Flow Passages—Thermohydraulic Performance and Fabrication
Technology, Heat Transfer Engineering, vol. 24, no. 1, pp. 3–17,
2003.
[2] Kandlikar, S. G., and Grande, W. J., Evaluation of Single Phase
Flow in Microchannels for High Heat Flux Chip Cooling—
Thermohydraulic Performance Enhancement and Fabrication
heat transfer engineering
[17]
[18]
[19]
51
Technology, Heat Transfer Engineering, vol. 24, no. 1, pp. 5–16,
2004.
Steinke, M. E., and Kandlikar, S. G., Review of Single-Phase Heat
Transfer Enhancement Techniques for Application in Microchannels, Minichannels and Microdevices, International Journal of
Heat and Technology, vol. 22, no. 2, pp. 3–11, 2004.
Bailey, D. K., Ameel, T. A., Warrington, R. O., and Savoie,
T. I., Single Phase Forced Convection Heat Transfer in
Microgeometries—A Review, Proc. Intersociety Energy Conversion Engineering Conference, Vol. 2: Environmental Impact,
pp. 301–310, 1995.
Palm, B., Heat Transfer in Microchannels, Microscale Thermophysical Engineering, vol. 5, no. 3, pp. 155–175, 2001.
Sobhan, C. B., and Garimella, S. V., A Comparative Analysis
of Studies on Heat Transfer and Fluid Flow in Microchannels,
Microscale Thermophysical Engineering, vol. 5, no. 4, pp. 293–
311, 2001.
Morini, G. L., Single-Phase Convective Heat Transfer in Microchannels: A Review of Experimental Results, International
Journal of Thermal Sciences, vol. 43, no. 7, pp. 631–651, 2004.
Tuckerman, D. B., and Pease, R. F. W., High-performance Heat
Sinking for VSLI, IEEE Electron Device Letters, vol. EDL-2, pp.
126–129, 1981.
Missaggia, L. J., Walpole, J. N., Liau, Z. L., and Phillips, R. J., Microchannel Heat Sinks for Two-Dimensional High-Power-Density
Diode Laser Arrays, IEEE Journal of Quantum Electronics, vol.
25, no. 9, pp. 1988–1992, 1989.
Riddle, R. A., Contolini, R. J., and Bernhardt, A. F., Design Calculations for the Microchannel Heatsink, Proc. Technical Program—
National Electronic Packaging and Production Conference, vol. 1,
pp. 161–171, 1991.
Gui, F., and Scaringe, R. P., Enhanced Heat Transfer in the Entrance Region of Microchannels, Proc. Intersociety Energy Conversion Engineering Conference, pp. 289–294, 1995.
Peng, X. F., Wang, B. X., Peterson, G. P., and Ma, H. B., Experimental Investigation of Heat Transfer in Flat Plates with Rectangular Microchannels, International Journal of Heat and Mass
Transfer, vol. 38, no. 1, pp. 127–137, 1995.
Vidmar, R. J., and Barker, R. J., Microchannel Cooling for a
High-Energy Particle Transmission Window, an RF Transmission Window, and VLSI Heat Dissipation, IEEE Transactions on
Plasma Science, vol. 26, no. 3, pp. 1031–1043, 1998.
Ravigururajan, T. S., and Drost, M. K., Single-Phase Flow Thermal Performance Characteristics of a Parallel Microchannel Heat
Exchanger, Journal of Enhanced Heat Transfer, vol. 6, no. 5,
pp. 383–393, 1999.
Lee, P. S., Ho, J. C., and Xue, H., Experimental Study on Laminar
Heat Transfer in Microchannel Heat Sink, Proc. Eighth Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems, 30 May–1 June 2002, ITHERM 2002,
pp. 379–386, June 2002.
Qu, W., and Mudawar, I., Experimental and Numerical Study of
Pressure Drop and Heat Transfer in a Single-Phase Micro-Channel
Heat Sink, International Journal of Heat and Mass Transfer,
vol. 45, no. 12, pp. 2549–2565, 2002.
ASME PTC 19.1, 1998.
NIST Technical Note 1297, 1994.
Steinke, M. E., and Kandlikar, S. G., Single-Phase, Liquid
Friction Factors in Microchannels, International Conference on
vol. 27 no. 4 2006
52
[20]
[21]
[22]
[23]
M. E. STEINKE ET AL.
Microchannels and Minichannels, Paper ICMM05-75112,
Toronto, Canada, June 13–15, 2005.
Steinke, M. E., and Kandlikar, S. G., Single-Phase Liquid Heat
Transfer in Microchannels, International Conference on Microchannels and Minichannels, Paper # ICMM05-75114, Toronto,
Canada, 2005.
Steinke, M. E., and Kandlikar, S. G., Control and Effect of Dissolved Air in Water during Flow Boiling in Microchannels, International Journal of Heat and Mass Transfer, vol. 47, pp. 1925–
1935, 2004.
Colgan, E. G., Furman, B., Gaynes, M., Graham, W., LaBianca,
N., Magerlein, J. H., Polastre, R. J., Rothwell, M. B., Bezama,
R. J., Toy, H., Wakil, J., Zitz, J., and Schmidt, R., A Practical Implementation of Silicon Microchannel Coolers for High Power
Chips, Proc. Semiconductor Thermal Measurement, Modeling,
and Management Symposium, March 15–17, 2005. San Jose, CA,
pp. 1–7.
Kakaç, S., Shah, R. K., and Aung, W., Handbook of Single-Phase
Convective Heat Transfer, John Wiley & Sons, New York, 1987.
Mark E. Steinke is a thermal engineer in the Systems & Technology Group eServers xSeries and
BladeCenter Servers at the IBM Corporation in
Research Triangle Park, NC. His research is focused on the system and CPU level cooling of
high-performance servers. He received his Ph.D.
in microsystems engineering from Rochester Institute of Technology in 2005. His dissertation
topic was on the development of cooling techniques for high heat flux microprocessors using
enhanced microchannels with single-phase liquid. Other areas of his research
include two-phase flow boiling in microchannels, critical heat flux (CHF) in microchannels, and liquid-vapor interfacial dynamics during CHF. He is a member
of the ASME and IEEE.
Satish Kandlikar is a Gleason professor of the
mechanical engineering department at RIT, where
he has been working for the last twenty-five years.
He received his Ph.D. from the Indian Institute of
Technology in Bombay in 1975 and has been a
faculty member there before coming to RIT in
1980. His research is mainly focused in the area
of flow boiling, high heat flux cooling, electronic
cooling, air-water flows in fuel cells, evaporating
meniscus studies, and other novel applications.
After investigating the flow boiling phenomena 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 highspeed photography techniques, he demonstrated that small bubbles are released
at a high frequency under flow conditions. His current work involves stabilizing
heat transfer engineering
flow boiling in microchannels, interface mechanics during rapid evaporation,
and advanced chip cooling with single-phase liquid flow. He has published over
130 journal and conference papers. He is a fellow member of ASME and has
been the organizer of the three international conferences on microchannels and
minichannels sponsored by ASME. Visit www.rit.edu/∼taleme for further information and publications.
John Magerlein is a research staff member and
manager of chip cooling and rf passives at the IBM
Thomas J. Watson Research Center in Yorktown
Heights, NY. He received his Ph.D. in physics
from the University of Michigan in 1975 and
worked at Bell Laboratories prior to joining IBM
in 1977. There, he has carried out research on
experimental Josephson junction circuits, GaAs
MESFET processing and characterization, and
electromagnetic modeling of high-performance
interconnects prior to assuming his current position. His current research interests include rf MEMS devices, cooling of very high power chips, and advanced packaging for high-performance computer systems. He is a member of
the American Physical Society and the IEEE.
Evan G. Colgan is a research staff member at
the IBM Thomas J. Watson Research Center and
is currently working on high-performance water cooling. He has worked on optical interconnects, high-resolution flat panel displays, liquid
crystal on silicon projection displays, the integration of both Cu- and Al-based metallizations
for integrated circuits, diffusion barriers, and silicide formation. He received his B.S. in applied
physics from the California Institute of Technology in 1982, and his Ph.D. in materials science and engineering from Cornell
University in 1987. He has published more than ninety papers and been issued
more than fifty patents.
Alan D. Raisanen received his B.A. in physics
from Drake University in 1985, and his Ph.D. in
materials science and engineering from the University of Minnesota in 1991. He spent ten years at
Xerox Corporation studying III-V and II-VI semiconductors and interface properties and working
as a microelectronics engineer in the thermal
inkjet program. Currently, he serves as associate
director in the Semiconductor and Microelectronics Fabrication Laboratory (SMFL) at RIT, where
he concentrates on developing facilities to support research into MEMS devices and integration with CMOS microelectronics. Research interests include
MEMS materials processing and characterization, micro-optical mechanical devices, and microfluidic devices.
vol. 27 no. 4 2006