Metal-based microchannel heat exchangers (MHEs) are of current interest due to the combination of high heat transfer performance and improved mechanical integrity. Efficient methods for fabrication and assembly of functional metal-based MHEs are essential to ensure the economic viability of such devices. Al- and Cu-based high-aspect-ratio microscale structures (HARMS) have been fabricated through molding replication using metallic mold inserts. Such metallic HARMS were assembled through eutectic bonding to form Al- and Cu-based MHEs, on which heat transfer tests were conducted to determine the overall cooling rate and time constants. Electrically heated Cu blocks were placed outside the MHEs and provided a constant flux, and water flowing within the microchannels acted as the coolant. Experimental results show a great influence of the type of metal, flow rate, and the surrounding conditions on the overall cooling performance of the MHEs.
- A 1 constant
- A 2 constant
- D h microchannel hydraulic diameter (μ m)
- T temperature (°C)
- t time (s)
- Greek Symbols
- τ 1 cooling time constant corresponding to the heat flux from the heater (s − 1)
- τ 2 cooling time constant corresponding to the heat flux from the surroundings (s − 1)
Science and technology related to new and more environmentally friendly ways of producing and distributing energy are being intensely investigated. New technologies capable of increasing the energy efficiency of transportation, housing, and appliance systems, if successfully commercialized, will reduce energy consumption and positively impact the global environment [[
In 1981, Tuckerman and Pease pointed out that decreasing liquid cooling channel dimensions to the micrometer scale will lead to increases in heat transfer rates, and demonstrated a 40-fold improvement in heat-sinking capability experimentally in Si-based microchannels anodically bonded to Pyrex cover plates [[
Various techniques for fabricating metal-based HARMS have been studied. Serial subtractive techniques, include micromilling [[
Molding replication offers an important alternative toward efficient fabrication of metal-based HARMS. The process of replication involves the use of primary HARMS, produced by a combination of lithography, etching, deposition, and other techniques, as a mold insert to create the negative of the insert pattern in metals by molding. One important example of molding replication is contained within the LiGA protocol for HARMS fabrication, combining deep x-ray/ultraviolet (UV) lithography (Lithographie) on polymeric resists, metal electrodeposition (Galvanoformung) into developed resist recesses, and molding replication (Abformung) of secondary HARMS [[
Since 2003, successful HARMS replication by direct molding has been demonstrated in Pb and Zn [[
Using these surface-engineered refractory mold-inserts, microchannels with complex geometries have been replicated successfully in metals such as Al and Cu by compression molding. Al-based microchannels were successfully bonded to other Al microchannels and plates by eutectic bonding with vapor-phase co-deposited Al–Ge [[
Heat transfer characteristics of Cu-based MHEs have been measured previously by Lee et al. in Cu microchannels fabricated by mechanical cutting and sealed with a polymeric top plate by gluing [[
Heat transfer in microchannels depends significantly on channel geometry. Extensive studies of microchannel flow and heat transfer have been carried out in the past two decades. The focus of these studies has been on characterizing the flow and thermal transfer at a fundamental level, while application development has been a more recent focus.
The first experimental results of fluid behavior in microchannels were obtained for trapezoidal shaped microchannels with hydraulic diameters 50 μ m [[
Recently there have been several studies that have focused on various aspects of microchannel geometry to enhance heat transfer. Liu et al. [[
In the present work, fabrication of metallic MHE by molding replication is discussed. Additionally, heat transfer tests have been performed to investigate the overall cooling performance of such devices using infrared thermography. Cooling rates are evaluated for both Al- and Cu-based MHE devices.
Microscale mold inserts were fabricated from nickel-based superalloy Inconel X750 plates. As-received Inconel plates were machined to square insert blanks, with an active area of ∼ 15,000 μ m × ∼ 15,000 μ m, ∼ 3200 μ m in height. The top surface of the blank was mechanically polished with SiC abrasive papers down to 1200 grit size. Insert fabrication from the blanks involved three main steps: μ EDM of the active area, electrochemical polishing (ECP) of as-machined microscale Inconel features, and deposition of a conformal amorphous silicon nitride (a-Si:N) coating over electrochemically polished microscale features. A SARIX high-precision micro-erosion machine (model SR-HPM-B) was used for insert μ EDM. Flat molybdenum (Mo) sheets with a thickness of 500 μ m were used as blade electrodes. A series of parallel cuts was performed with the Mo electrodes, resulting in an array of parallel rectangular microprotrusions on the insert active area. As-machined Inconel blanks were electrochemically polished for a total of 10 min with current control in a mixed acid solution of HClO
Graph: Figure 1 SEM image of a Si:N coated Inconel insert.
Al 6061 (90 at.% Al, 4 at.% Si, 2 at.% Mg) and Cu 110 (99.9+ at.% Cu) coupons, with the same geometry of 35.5 mm × 35.5 mm square and 6.4 mm in thickness, were molded at high temperatures with one single Inconel insert coated with an a-Si:N coating. The molding process replicates the array of parallel rectangular microprotrusions on the insert into an array of parallel rectangular microchannels on the molded Al and Cu coupons. Before molding, the top surfaces of Al and Cu coupons were mechanically polished to less than 1 μ m in surface roughness. Compression molding was carried out in an MTS858 single-axis testing system interfaced to a custom-built high-vacuum, high-temperature, instrumented molding apparatus. A linear actuator could be programmed to move according to prescribed load forces in the force-controlled mode or actuator displacements in the displacement-controlled mode. Total axial force and axial displacement were measured continuously during the entire molding and demolding process. All molding runs included five segments: a fast approach followed by a slow approach in displacement control, a constant-loading-rate molding followed by a 15-s constant-force hold in force control, and a constant-rate demolding in displacement control. Further details on the molding apparatus and the molding protocol are given elsewhere [[
One fluid supply channel and one fluid drain plenum were cut into replicated arrays of parallel Al and Cu HARMS replicated by molding, which were further fabricated using μ EDM to cut inlet and outlet channels and using a drilling machine to make connection holes. Then these holes were typed to connect plastic adaptor. Surface morphologies of molded HARMS were examined on a Hitachi S3600N scanning electron microscope (SEM). Figure 2a and Figure 3a show the fabricated Cu and Al HARMS, respectively. Two inlets are connected by a μ EDM-ed channel, which also connected to the microchannels. On the other end, a wide channel is μ EDM-ed as a pool to connect three outlets, one in the center of the active area and the other two on its sides. Figure 2b and Figure 3b show SEM views of one of the inlets. The μ EDM-ed inlet channel of each structure has sharp sidewalls and typical μ EDM-ed surface morphologies, and well-connected microchannels and inlet holes. For Al structure, some dark spots can be found in the μ EDM-ed channel. These spots were produced during the EDM process due to the relatively lower melting temperature. Figure 2c and Figure 3c show high-magnification views of molded Cu/Al microchannels. The microchannel has vertical sidewalls and sharp sidewall-to-bottom transitions. It is clear to see the typical sidewall and bottom surface roughness due in part to the sidewall roughness of the rectangular microprotrusion on the Inconel insert. Because of the same insert, both of them show similar surface morphologies and the same surface roughness in size.
Graph: Figure 2 Close views of the copper micro heat exchanger base [[
Graph: Figure 3 Close views of the aluminum micro heat exchanger base.
Al/Cu HARMS and flat coupons were mechanically polished again to get a smooth surface for better bonding quality by using Al–Ge intermediate layers, which were deposited in the ICP assisted hybrid system. Al–Ge composite films were co-deposited in pure Ar (99.999%), with a total pressure of ∼ 1.3 mtorr, from two separate sputter sources, one for pure Al (99.99%) and the other for pure Ge (99.99%). The polished substrates, Al/Cu HARMS and flat coupons, were ultrasonically cleaned in acetone and methanol before being mounted on a rotatable holder situated in the middle of the deposition zone. The deposition sequence consisted of a substrate surface etch followed by co-deposition of Al and Ge. Substrate etching occurred in a pure Ar ICP with a total rf input power of 1000 W, a substrate bias of –100 V, and an etch duration of 20 min. Sputtering of Al and Ge cathodes commenced immediately after substrate surface etch. Substrates were rotated continuously at about 12 rpm during both etching and deposition. All Al–Ge depositions were carried out with a fixed Al cathode current of 1.0 A and a fixed Ge cathode current of 0.45 A, respectively. This deposition composition resulted in a Ge to Al composition ratio close to that corresponding to the Al–Ge eutectic composition of Al
Bonding experiments were carried out using the MTS858 single-axis testing system interfaced to a custom-built high-vacuum chamber. One heating station was mechanically attached to the bottom of the vacuum chamber. A second heating station was mechanically attached to the top linear actuator of the MTS858 testing system through a bellow-sealed motion feedthrough. The two heating stations were heated separately by resistive heating cartridges and the temperatures were measured by two separate K-type thermocouples. Further details on this high-vacuum, high-temperature bonding system have been given in our previous paper.
Flat Al/Cu coupons and HARMS with Al–Ge composite films deposited on the bonding surfaces were placed face-to-face on the lower heating station. The chamber was evacuated, and both heating stations were heated. Al HARMS/coupon bonding was carried out at the temperature of 500°C with an applied pressure of ∼ 1.5 MPa, while Cu HARMS/coupon was bonded at the temperature of 540°C with an applied pressure of ∼ 3 MPa. Figure 4 shows an optical overview of bonded Al and Cu devices connecting with plastic adaptors.
Graph: Figure 4 Assembled view of the copper and aluminum MHE devices [[
To get exact dimensions of bonded Al and Cu devices, we cut these two devices after we finished the heat transfer experiments. Figure 5a and Figure 6a show the cross-section views of portions of the bonded microchannel structures, obtained by mechanical cutting perpendicular to the microchannel arrays. The entire structure contains 19 microchannels in each device. However, there is one channel blocked in the Al device, meaning the Al device has 18 through microchannels. Figure 5b and Figure 6b show higher magnification views of two typical bonded microchannels. The bonding interface is not discernable in this higher magnification view, indicative of the quality of bonding achieved. From the SEM picture, the average dimension of each microchannel can be obtained, as shown in Table 1.
Graph: Figure 5 Cut away view of the Cu microchannel after top plate bonding [[
Graph: Figure 6 Cut away view of the Al microchannel after top plate bonding.
Table 1 Dimensions of Cu and Al devices
Average Average Micro-channel Inlet channel Inlet channel Inlet channel Number of Dh height (μ m) width (μ m) length (μ m) height (μ m) width (μ m) length (μ m) channels (μ m) Cu device 319.46 149.80 12,650 620.50 1260 16,200 19 203.96 Al device 290.82 114.56 13,050 686.98 1242 17,500 18 164.37
Figure 7 shows a schematic of the testing apparatus designed and built to study the thermal behavior of the micro heat exchanger fabricated using the already discussed procedure. The apparatus consisted of three sections: pressuring section, test section, and data acquisition section. For the experiments, the micro heat exchanger was used as a cooling device with water as the coolant. The first section consisted of a water holding tank connected to a compressed air bottle and was used to obtain sufficient pressure to push the coolant through the micro channels. This approach was preferred to the use of pump and provided a smooth and stable flow. A needle valve placed at the bottle neck helped release the required amount of air from the bottle and that placed downstream of the tank exit was used to make fine adjustments to the flow rate and attain smooth and stable flow through the channels. A pressure relief valve was also connected to the tank for safety measures.
Graph: Figure 7 Schematic of the experimental setup [[
The test section consisted of either an Al- or a Cu-based specimen enclosed in a rigid polyvinyl chloride (PVC) foam insulation with the heater in contact with the top face of the specimen. The Al specimen consisted of 18 parallel channels with an average cross-sectional dimension of 291 μ m × 114 μ m, while the Cu specimen had 19 parallel channels with an average cross-sectional dimension of 319 μ m × 150 μ m. Figure 8 shows a detailed view of the test section. Two rectangular heaters were used. Each rectangular heater was fabricated by inserting a super watt cartridge heater (200 W) into a single machined copper block (5/8 × 5/8 inch) of same length(1/4 inches). The two copper blocks were bonded together, then attached to the specimen with the help of highly conducting arctic silver glue and then the cartridge heaters were inserted. Before attaching, the surface of the heater was polished and smoothed. Also, the arctic silver glue provided a uniform contact and eliminated the air gaps between the heater and specimen surfaces. Holes were drilled into the insulation, near the bottom surface of the specimen, for the inlet and outlet tube connections. In order to measure the inlet and the outlet temperatures, 36 gauge K-type thermocouples were inserted into the inlet and the outlet plenum through the inlet and the outlet tubes (Figure 9). Other K-type thermocouples were placed at the top surface, bottom surface, and heater surface. An Instrunet data acquisition system connected to a PC (personal computer) was used to collect the thermocouple read-outs. The differential pressure across the specimen was measured using a Dwyer digital manometer with a least count of 0.1 psi.
Graph: Figure 8 Detailed views of the test section (not to scale): (a) bottom view, (b) top view, and (c) cross-sectional view.
Graph: Figure 9 Schematic showing the placing of the inlet thermocouple.
Constrained by the sample dimensions, a maximum of two super watt heaters were used with a maximum input flux of 37 W/cm
The two specimens were tested under similar conditions of temperature and pressure and the same experimental procedure was followed. First thermocouples were inserted into each inlet and tube with the help of a T-fitting and then sealed with epoxy (Figure 9). In the same fashion, thermocouples were inserted into each outlet tube but the epoxy sealing was not required. The tubes along with the thermocouples were then connected to the inlet and outlet of the device. Another thermocouple was placed at the bottom surface of the device, which was then placed in the insulation. Four other thermocouples and the heater were placed on the top surface of the device. Further, connections to measure the differential pressure and flow rate were made. These settings were maintained the same for both the specimens.
Infrared thermography was used to determine the surface temperature of the MHE device to determine the overall cooling rate and time constant for the device. The surface was coated black to improve emissivity up to 0.96. The test device was initially heated to a known temperature and cooler water was suddenly introduced into the test channels. The surface temperature response was captured at 60 Hz capture rate with the FLIR SC500 camera. The embedded thermocouples were used to calibrate the infrared (IR) response in situ. In each IR image a rectangular area was marked and the average temperature of that area was used to plot the transient temperature profile.
Figure 10 shows the surface temperature drop as coolant is suddenly allowed into the MHE. Initially, the heater is turned on and the MHE is allowed to achieve a steady temperature around 100°C. The coolant is suddenly introduced into the MHE and the surface temperature is monitored. It is clear that there is a significant temperature drop in the first 2–3 s from 100 to 40°C. Then there is a gradual decrease from 40 to 30°C in the next 60 s. The first drop in temperature occurs when the coolant draws the heat away from the copper or aluminum. The second gradual drop occurs when the heat is drawn from the surrounding insulation. Since the insulation has high heat capacity, this results in a slower gradual decrease in temperature. However, the entire device cools down to acceptable temperatures within 60 s of coolant introduction.
Graph: Figure 10 Surface temperature response of the MHE in a transient cooling mode [[
Based on the preceding hypothesis, the transient temperature profile for all the test cases were normalized and then approximated with the following decaying formula:
Graph
where T(t) is the normalized transient temperature, τ
Figure 11 shows the normalized cooling rate curve that has been fitted using Eq. (
Graph: Figure 11 Fitting of the normalized transient cooling curve [[
Table 2 Constants for Al-based MHE
Pressure drop (psi) A1 τ1(s) A2 τ2(s) 25.2 0.63319 2.18872 0.36696 15.50716 25.1 0.65515 2.3011 0.33922 18.14155 25 0.67993 2.17843 0.34203 17.7377 40.4 0.72496 1.88696 0.29739 15.57753 40.5 0.80313 2.03842 0.29047 15.18432 40.4 0.70437 1.82759 0.30823 13.26754
Table 3 Constants for Cu-based MHE
Pressure drop (psi) A1 τ1(s) A2 τ2(s) 10.2 0.35755 1.65058 0.64535 11.47543 10 0.40668 1.81968 0.59276 12.55087 10.5 0.48393 1.97372 0.57198 13.20602 20.5 0.59791 1.66263 0.50274 10.27134 20.3 0.50407 1.39206 0.50344 9.90178 20.2 0.47152 1.46668 0.54768 7.81155 40.2 0.51694 1.03597 0.44524 8.34396 40 0.55367 1.11859 0.44608 7.98393 39.9 0.674 1.15714 0.36364 7.97655
Figure 12a and Figure 12b present the time constant based on the transient cooling mode behavior of the MHE. The results show that the time constant for the device decreases with increasing cooling flow rate, as expected. Time constants corresponding to the heat flux from the heater are in the range of 1 to 2 s for the copper device and 2.5 to 2 s for the aluminum device. Those corresponding to the heat flux from the surrounding insulation show a similar trend but are one order higher in magnitude. Such reduced response is due to a large difference in the thermal conductivities of the insulation and the metallic device. Moreover, the aluminum device response is slower than the copper device due to the basic difference in thermal conductivity of the metals. Copper has higher thermal conductivity and thus results in lower time constants.
Graph: Figure 12 MHE time constant based on transient cooling mode results: (a) τ1 [[
Figure 13 shows the transient cooling mode behavior of metallic MHE from the initial state to the steady state. At instant 1, the heaters were switched on to provide an input power of 370 W. The surface temperature was constantly monitored with help of thermocouples placed on the surface was allowed to rise. As soon as the surface temperature reached 100°C, the valve was opened to push the water through the microchannels (instant 2) and within a minute the system reached steady state (instant 4). Infrared images were taken for the whole duration at 60 Hz capture rate and processed to obtain the cooling curve (Figure 13). Figure 14 shows the thermographs corresponding to each instant. The thermographs at instants 2 and 3 give an idea of the cooling effectiveness of the metal-based microchannel device. The time difference between instant 2 and instant 3 is just about a second. The fourth thermograph indicates the rise in coolant temperature as well as even heating of the coolant as it passes through the microchannels.
Graph: Figure 13 Temperature profile from initial state to steady state.
Graph: Figure 14 Thermographs at different instants: (a) t = 5 s, (b) t = 33.091 s, (c) t = 34.042 s, and (d) t = 60 s.
Fabrication of Al- and Cu-based high-aspect-ratio microscale structures (HARMS) through molding replication using metallic mold inserts has been demonstrated. Such metallic HARMS were assembled through eutectic bonding to form Al- and Cu-based MHEs. Heat transfer tests were conducted to determine the overall cooling rate and time constants. The time constant for the MHE device was found out to be lower for Cu channels with times around 1–2 s response. Aluminum is only slightly lower due to the lower thermal conductivity. Surrounding conditions were observed to play an important role by considerably increasing the cooling time.
The demonstration of fabrication and thermal performance for metal-based MHE devices has enormous potential in many heat transfer problems, primarily in avionics and electronic cooling. It is clear that the use of metal-based micro heat exchangers may result in significant improvement in cooling efficiency for microelectronic systems.
Graph
Pritish R. Parida received his B.Tech. degree from Indian Institute of Technology Guwahati (IITG), Guwahati, India, in 2006 and his M.S. degree from Louisiana State University, Baton Rouge, in 2007, both in mechanical engineering. He is currently a Ph.D. candidate in the Department of Mechanical Engineering, Virginia Tech. His research interests include thermal packaging for power electronics, cooling systems, investigation of enhanced heat transfer designs, and modeling and simulation of micro-scale conjugate heat transfer problems using the lattice Boltzmann method.
Graph
Fanghua Mei received the B.S. and M.S. degrees in materials science and engineering from Shanghai Jiao Tong University, Shanghai, China, in 2002 and 2005, respectively. He is currently a Ph.D. candidate in the Department of Mechanical Engineering, Louisiana State University, Baton Rouge. His research interests include thin film deposition, fabrication of metal-based MEMS, and testing of assembled MEMS devices.
Graph
Jing Jiang received the B.S. and M.S. degrees in the Department of Mechanics and Mechanical Engineering from the University of Science and Technology of China, Hefei, Anhui, China, in 1999 and 2002, respectively. He is currently a Ph.D. candidate in the Department of Mechanical Engineering, Louisiana State University, Baton Rouge. His research interests include fabrication, modeling, and simulation of metal-based MEMS.
Graph
Wen Jin Meng received the B.S. degree in physics with honors from California Institute of Technology (Caltech), Pasadena, in 1982 and the Ph.D. degree in applied physics from Caltech in 1988. He served as a postdoctoral research fellow in the Materials Science Division of Argonne National Laboratory from 1988 to 1989 and as a staff research scientist in the General Motors/Delphi Automotive Systems R&D Center from 1989 to 1999. Since 1999, he has been with Louisiana State University, Baton Rouge, where he is currently the Gerald Cire & Lena Grand Williams Professor in the Department of Mechanical Engineering. His research spanned topics concerning solid-state phase transformations, heteroepitaxial growth of metal and ceramic thin films, plasma-assisted vapor-phase deposition processes, nanostructured ceramic coatings, and surface engineering. His recent research focuses on the materials aspects of microscale fabrication and assembly, especially on metal-based microsystems.
Graph
Srinath V. Ekkad received his B.Tech. degree from JNTU in Hyderabad, India, in 1989 and then his M.S. from Arizona State University in 1991 and Ph.D. from Texas A&M University in 1995, all in mechanical engineering. He was a research associate at Texas A&M University and a senior project engineer at Rolls-Royce, Indianapolis, before he joined Louisiana State University as an assistant professor in 1998. He moved to Virginia Tech as an associate professor of mechanical engineering in fall 2007. His research is primarily in the area of heat transfer and fluid mechanics with applications to heat exchangers, gas turbines, and electronic cooling. He has written more than 100 articles in various journals and proceedings and one book on gas turbine cooling. His research focuses on enhanced heat transfer designs, with a variety of applications.
By PritishR. Parida; Fanghua Mei; Jing Jiang; WenJin Meng and SrinathV. Ekkad
Reported by Author; Author; Author; Author; Author