A single phase hybrid micro heat sink using impinging micro-jet arrays and microchannels

Accepted Manuscript A Single Phase Hybrid Micro Heat Sink Using Impinging Micro-Jet Arrays and Microchannels A.J. Robinson, R. Kempers, J. Colenbrander, N. Bushnell, R. Chen PII: DOI: Reference:

S1359-4311(17)34995-5 https://doi.org/10.1016/j.applthermaleng.2018.02.058 ATE 11840

To appear in:

Applied Thermal Engineering

Received Date: Revised Date: Accepted Date:

4 August 2017 4 February 2018 15 February 2018

Please cite this article as: A.J. Robinson, R. Kempers, J. Colenbrander, N. Bushnell, R. Chen, A Single Phase Hybrid Micro Heat Sink Using Impinging Micro-Jet Arrays and Microchannels, Applied Thermal Engineering (2018), doi: https://doi.org/10.1016/j.applthermaleng.2018.02.058

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A Single Phase Hybrid Micro Heat Sink Using Impinging Micro-Jet Arrays and Microchannels A.J. Robinson 1,*, R. Kempers1,2, J. Colenbrander 1, N. Bushnell3, R. Chen4 1

Confluent Research Ltd., 6 The Avenue, Inse Bay, Laytown, Ireland 2 York University, Toronto, Canada 3 SimuTech Group, 11611 Airport Rd. Suite 201, Seattle, Washington, U.S.A. 4 Microfabrica, 7911 Haskell Ave, Van Nuys, California, U.S.A. * Corresponding Author: [email protected] Abstract This work describes the development of a single phase water-cooled microfluidic heat exchanger for cooling very high heat flux electronics. The heat sink was designed so that it can be manufactured using the MICA Freeform process, which is an ultra-precision additive manufacturing method for millimeter-scale metallic parts with micron-scale features. The heat sink targets a heat flux of 1000 W/cm2 with an average base temperature constraint of 65oC for a 4 mm x 3 mm heat source. The hydraulic constraints that were imposed on the design were that the pressure drop and flow rates could not exceed 100 kPa and 0.5 L/min respectively. The design was undertaken using Simulation-Driven Design whereby the commercial Computational Fluid Dynamics software ANSYS Fluent was utilized. The final embodiment of the design is a hybrid Microjet-Microchannel heat sink. The thermal performance is quite extraordinary, with a predicted effective thermal conductance of 400 kW/m2K for a flow rate of 0.5 L/min. With such an exceptionally high thermal conductance, the heat sink is predicted to maintain an average base temperature of under 58oC with a maximum variation about the mean of ±3oC for an imposed heat flux of 1000 W/cm2.

Keywords: liquid cooling, additive manufacturing, microjets, microchannels, MEMs heat exchanger, high heat flux, microfluidics Nomenclature A

area, m2

COP

Coefficient of Performance, -

P

pressure, kPa

h

thermal conductance, W/m2K

k

thermal conductivity, W/m-K

q//

heat flux, W/m2

Q

power, W

T

temperature, oC

u

velocity, m/s

V

volumetric flow rate, m3/s

Greek symbols ρ

mass density, kg/m3

µ

dynamic viscosity, N-s/m2

Subscripts l

liquid

s

surface

w

water

1. Introduction Reduced size, increased power levels and increased density of circuit elements have escalated to the extent that many current and next generation high performance electronic devices are reaching heat flux levels so high that new liquid-based heat exchange technologies are required. Without adequate cooling, components and packages can exceed temperatures beyond acceptable design limitations which can reduce performance and severely limit long term reliability [1]. In this context, thermal management of electronics is one of the most crucial technological barriers inhibiting the advancement of electronic components and packages. The breadth of the problem spans personal portable electronics all the way through to high-end military radar and avionics technology, affecting almost all electronic systems in between. When heat flux levels exceed about 100 W/cm 2, conventional air and liquid cooling strategies generally become inadequate [2]. This was recognized over thirty years ago with the pioneering work of Tuckermann and Pease [3], who are credited with being the first to realize heat flux removal rates well above 100 W/cm 2 using microfluidic heat exchangers. What Tuckermann and Pease [3] correctly realized was that, based on classical heat transfer theories, the smaller the channel size of a heat exchanger, the higher the effective thermal conductance of the device. For a given applied heat load, the higher conductance will result in cooler device operation. Thus, reducing the scale of heat exchanger fluidic channels from centimeter scales to ~100 µm or less will result in exceptional improvement in the heat transfer. However, the penalty was that the required hydraulic power also increased with reduced channel size, and this would have to be carefully considered when engineering micron-sized heat exchanger technology. After a notably slow start [4], renewed interest in microfluidic heat exchange technology began early this century [5]. The intensifying thermal demand of electronics developed in parallel with high speed computing capabilities and the associated maturing of codes that could solve complex heat transfer and fluid mechanics problems. After about a decade of confusion and controversy regarding the existence of special micro-effects at these scales, the advent of properly posed and simulated flow and heat transfer, in predominantly microchannel geometries, resolved the on-going debates. The end result was that researchers could in fact apply conventional continuum mechanics to liquid-based microfluidic heat exchangers. A significant escalation in microchannel and microfluidic research ensued, driven by the demand of industry and the newly justified capacity to use the simulation environment in an engineering design capacity, as opposed to a scientific one. The next decade saw the conceptual development of a vast number of different microfluidic heat exchangers, using Computational Fluid Dynamics (CFD) as the main simulation design tool. The primary goal of the various designs was to invent and verify techniques of enhancing the heat transfer without causing an excessive hydraulic penalty. The concepts ranged from simple consideration of the channel cross-sectional geometry [6 7], flow obstructions [8-12], wavy and zig-zag channels [13-15] and periodic pinching of the channels and channel roughness [16-18]; all ostensibly aimed to increase mixing and/or surface area within the channels thus improving the effectiveness of heat transfer. More complex three dimensional conceptual configurations [19-21] also emerged, partially aimed at the problem associated with such high imposed heat fluxes causing large surface temperature gradients, which are also deleterious to electronics performance and lifespan. One method of achieving exceptionally heat transfer rates, along with offering good surface temperature uniformity, is the use of microjets. As with microchannel-type flows, the effective thermal conductance increases as the size of the jet

decreases [22]. Generally speaking, microjet heat transfer coefficients are higher than that of equivalently posed microchannels, and this can be up to an order of magnitude which is significant [23-26]. Opposed to microchannels, where fluid is heated continually from inlet to outlet, jet arrays can disperse cold liquid to the entire heated surface thus having the potential to significantly reduce undesired surface temperature gradients. However, one drawback of microjets arrays is that there exists comparatively few experimental works in the open literature, with far fewer theoretical ones, making engineering design calculations problematic and risky. Another issue with microjet-based heat sinks is that of channeling the fluid into and out of the heat exchanger system, in particular if very low board-level form factors are required; which of course is the case for high component density electronic packages. Some hybrid microchannel-microjet systems have been conceived in order to take advantage of the cooling merits of each technology [27]. The general concept is that if the liquid is gradually introduced through microjets into a microchannel, the high cooling capacity of each is taken advantage of whilst also mitigating high temperature gradients and hydraulic penalties [27-30]. Albeit promising technologies, the hybrid microchannel-microjet heat sinks are still of fairly simple geometries since they rely on conventional micro-fabrication techniques from the semiconductor industry. As a result they offer limited improved performance over other microfluidic heat exchanger systems. Although the state of the art of single phase liquid-cooled microfluidic heat exchangers has advanced considerably over the past decade, the performance is still largely limited by their fairly simple internal geometries and this is likely due to constraints posed by conventional micro-fabrication techniques and materials. In order to realize their highest potential, a combination of extremely high liquid heat transfer coefficients must be combined with as high as feasible surface area per unit volume in the vicinity of the heated surface. These, combined with the fact that fluid must be channeled to, from and within the heat sink, requires complex internal structures that require an equally complex micro-manufacturing technology to realize them in reality.

1.2 The MICA Freeform Process As with other 3D printing processes, parts fabricated using the MICA Freeform process are built with layers. The MICA Freeform process involves three primary steps per layer. First, a fully-dense structural metal is electrodeposited onto a substrate in selected regions corresponding to the desired cross section of the part to be fabricated. Deposition occurs through apertures in a photoresist pattern, in a method similar to that used in semiconductor fabrication facilities. After removal of the photoresist, a sacrificial metal is blanket-electrodeposited over the structural metal. Finally, both metals are planarized to yield a layer that is flat, planar, and of precisely-controlled thickness. The three steps are then repeated for all layers required, after which a chemical etchant is used to dissolve the sacrificial metal, releasing the parts. It is within the context of this fabrication process that the proposed microfluidic heat exchanger was designed.

1.3 Problem Statement The preponderance of microchannel concepts proposed in the recent literature target heat fluxes in the region of 100 W/cm 2 with maximum surface temperature constraints commensurate with those of Si-based electronics (~75oC to 85oC). For the current design, small sized electronics (~5 mm length scale) with extremely high heat flux levels (~1000 W/cm2) are targeted. This would be commensurate with high power amplifiers for RF technologies, for example. For this design, a target maximum average surface temperature was selected to be 65oC for a coolant inlet temperature of about 27oC (300K). The maximum surface temperature variation of below 10oC was also a thermal design constraint. From a packaging perspective, where board

level real-estate is limited and compactness is key, the heat exchanger was designed to be very compact, with a maximum height restriction of 2 mm. With water as the working fluid, the hydraulic constraints for maximum allowable volumetric flow rate and pressure drop were 0.5L/min and 100 kPa respectively.

2. The Hybrid Microjet-Microchannel Heat Sink Design The hybrid water-cooled microjet-microchannel heat sink design, heretofore referred to as FINJET, is depicted in Fig. 1. The main housing of the heat sink is Valloy 120 (80% nickel, 20% cobalt, k=91 W/mK), however the MICA Freeform process allows for the embedding of copper, and this has been taken advantage of in the region close to the base. In this way, the copper layer (k=385 W/mK) acts as a heat spreader since it is much more conductive than Valloy 120. As shown in Fig. 2 the liquid enters the heat sink via a 0.95mm high chamber. This upper plenum is designed specifically to expand the inlet flow into a large enough volume that it stagnates so that a nearly uniform pressure exists above the inlet of all of the jets in the array. This is an important aspect of the design because; 

It forces the flow evenly into the individual jet nozzles resulting in well-conditioned downward-flowing jets when impinging upon the heat transfer surface, thus increasing the convective heat transfer coefficient.



It facilitates even distribution of flow across all of the nozzles in the array, not only increasing each of their individual heat transfer potential, but also causing each jet to have nearly the same thermal-hydraulic behavior. This reduces the generation of severe temperature gradients at the lower heated surface. The jet orifices themselves are 200 µm long by 70 µm wide diamond-shaped slot jets.

Figure 1: Isometric view of full heat sink less the header (Top) and cross section of the FINJET heat sink showing the inlet port and upper plenum chamber (bottom).

From the jet inlet through the various interior channels to the exit port of the heat sink involves several unique design features that are made possible only by precision 3D micro-fabrication processes, such as MICA Freeform. One in particular is the use of a cellular structure, whereby the individual nozzles issue jets into partially enclosed chambers, as depicted in Figs. 1 and 2. In contrast to other hybrid jet-channel systems, where the jets collect in linear channels and are thus exposed to escalating cross flow and heat transfer deterioration [27-30], this design isolates the pre-conditioned jets into individual cells allowing them to impinge on the target surface without obstruction or any other deleterious effects associated with the neighboring jets. In this way, the very high potential of each impinging microjet is realized resulting in exceptionally high heat transfer coefficients in the impingement zone. Another novel design feature is the upturning of the spent flow from the jet nozzle impingement zone into what can be considered as short vertical microchannels. As depicted in Fig. 2, subsequent to impinging on the lower heated surface, the flow is directed upward through channels with a nominal size of 35µm. This is particularly advantageous because the coolant wets significant surface area before being forced out of the exhaust port of the cell. Once again, this highlights the novelty of using the micro-3D printing fabrication process as it allows for complex micron-sized features and surface extensions to be packed very tightly within the individual jet-cell structures. This simultaneously enhances the heat transfer and extends the surface area, both of which being hugely advantageous to the local heat transfer. In order to improve fin efficiency and lateral heat spreading, copper is inlayed within the base structure and the channel walls, which act as fins.

Figure 2: (Top) Close up of jet-cell structure showing an individual cell and cell array, and (Bottom) an single cell and cutaway across the base of the internal microchannel: Red: copper, Aqua: Valloy, Green: Cut section.

The up-washing coolant exits the jet cell through an internal conduit located at the upper region of the cell. The spent flow from the cell, albeit heated to some extent, is still relatively cool owing to the high relative flow rates compared with the channel

jet-cell dimensions. The FINJET heat sink design takes advantage of this by collecting the spent flow from each cell into an integrated internal microchannel with nominal size of 100µm wide and 700µm tall, shown in Fig. 3. Within the microchannel the flow remains well mixed, in part due to the nature in which the flow from each jet-cell enters the channel and in part due to meandering path of the channel and internal flow obstructions. This is of course ideal from a heat transfer standpoint where there now exists a well-mixed flow of liquid coolant in a microchannel with substantial exposed surface area. Each jet nozzle is mechanically connected to the cell walls (see Fig. 2) in such a way that they are in thermal communication with the lower heated section. In this way, the internal channel carrying the spent flow from the jet impingement cells is an offset strip fin microchannel. The diamond shape of the nozzle cross section, highlighted as green in Fig. 2, was chosen to streamline the flow in the microchannel section whilst providing significant surface area for heat transfer. At the exit of the FINJET heat sink design is a side-attached exit chamber where the fluid is collected and routed to an exit port, as seen in Figs. 1 & 3. Compared with the state of the art, the design possibilities of micro-3D printing have allowed the design of a truly integrated system with millimeter scale form factor and micron-sized features.

Figure 3: Section showing internally integrated microchannels and relative position of jet-cells.

3 Numerical Modelling and Data Reduction 3.1 Numerical Modelling The design was undertaken using the Simulation-Driven Design platform whereby the commercial Computational Fluid Dynamics (CFD) software ANSYS Fluent Version 16.2 was utilized. For the simulations, a three-dimensional section of the heat sink was considered for the computational domain owing to the symmetry plains that exist and the laminar nature of the flow.

This of course was done to reduce computational resources. The computational domain together with the mesh is depicted in Figs. 4a and 4b respectively. A combination of relatively low fluid velocity and small characteristic length scales results in low Reynolds number flow through this device and a laminar flor model is justified. The governing equations for steady laminar flow were solved in conjunction with energy equations in both the liquid and solid phases. These are given as,

Liquid

   ( u )  0

(1)

   (u  )(  u )  p    (u )

(2)

(3) Solids

k2Ts  0

(4)

The boundary conditions imposed on the liquid are a constant inlet pressure, up to P = 1 atm at the inlet port (depending on desired flow rate), a constant pressure of P= 0 atm at the outlet, symmetry at the two side planes and the no slip condition on all wetted surfaces. The fluid inlet temperature was 300K and a balance of the conduction heat transfer was imposed as the solidliquid and solid-solid interfaces. At the bottom surface, a uniform heat flux of 1000 W/cm2 was applied and an insulation boundary condition was applied to the upper-most surface. The domain consisted of 13,505,082 mesh elements which was determined to give grid independent solutions for all of the flow rates tested. The simulations were carried out for steady flow scenario, though a transient model was also solved in order to check this assumption. The transient model showed some minor unsteadiness at the outlet though this did not have a notable effect on the average flow and thermal fields.

(a)

(b) Figure 4a): Wireframe diagram of computational domain, b) Illustration of the meshed domain.

3.2 Data Reduction The primary performance indicator for the FINJET heat sink is the effective thermal conductance, here defined as h owing to its close relation to the convective heat transfer coefficient,

h

q s'' T LM

(5)

where q//s is the applied heat flux at the bottom surface and ΔTLM is the log-mean temperature difference based on a constant average surface temperature,

(6)

In this way, h reflects the overall thermal conductance of the heat sink including conjugate heat transfer in the solid and convective heat transfer between the solid and fluid. The hydraulic performance of the heat exchanger is related to the pressure drop between the inlet and outlet as well as the hydraulic power requirement, respectively. These are given as,

P  Pin  Pout

(7)

Qpump  PV

(8)

and

where V is the volumetric flow rate. A final performance indicator considered here is the Coefficient of Performance which is the ratio of the thermal power transported through the heat exchanger and the hydraulic power required to force the liquid through it,

COP 

QHeat QPump

(9)

where Qheat=q//sAs and As is the surface area of the base of the heat sink. For illustration purposes the base area has been selected as As=3x4 mm2, which is commensurate with small high power amplifiers.

4 Results and Discussion 4.1 Overall Thermal Hydraulic Performance The thermal performance of microfluidic heat exchangers is posed in terms of both the thermal and hydraulic behavior using the performance metrics described above. Fig. 5 summarizes the thermal behavior of the FINJET heat sink design. As is typical with forced convection heat sinks, the overall thermal conductance increases quickly with initial increase in flow rate (~0-0.5 L/min) after which it begins to plateau. It is important to note that the overall thermal conductance of the heat sink reaches 400 kW/m2K at a flow rate of 0.5 L/min. This is an exceptionally high thermal conductance for a single phase heat exchanger and rivals the higher performance two-phase concepts, without issues related to CHF at such high heat fluxes. The FINJET heat sink is able to maintain a base temperature of under 58oC with an imposed heat flux of 1000 W/cm2, which is comfortably below the typical allowable junction

600

90 80

500

70

400

60 50

300

40

200

30

100

20

Tin=27oC (300K) q//=1000 W/cm2

0 0

0.1

0.2

0.3

0.4 0.5 0.6 0.7 Flow rate (L/min)

0.8

0.9

10

Max Base Temperature (oC)

Thermal Conductance (kW/m2 K)

temperature of Si semiconductor components and leaves some head-room in the thermal budget for a thermal interface material.

0 1

Figure 5: Effective thermal conductance and maximum base temperature of the simulated FINJET heat sink with varying flow

0.045

0.4

0.04

0.35

0.035

0.3

0.03

0.25

0.025

0.2

0.02

0.15

0.015

0.1

0.01 Tin=27oC (300K) q//=1000 W/cm2

0.005

0 0

0.1

0.2

0.3

0.4 0.5 0.6 0.7 Flow rate (L/min)

0.8

0.9

0.05

Thermal Resistance (K/W)

R-Value (cm2 K/W)

rate.

0 1

Figure 6: R-Value and thermal resistance of the simulated FINJET heat sink with varying flow rate.

From a thermal network modelling perspective it is more convenient to consider the heat sink in terms of its effective resistance to heat flow. Fig. 6 shows the R-Value, which is simply the inverse of the thermal conductance, h, and the thermal resistance, 1/(hAs). It is worth noting that the thermal resistance is still reasonably low, even for such a concentrated heat source. Fig. 7 shows the relevant hydraulic characteristics of the FINJET heat sink. Unlike the heat transfer, which tends to taper-off at higher flow rates, the rate of increase of the pressure drop and associated pumping power escalate with increasing flow rate. This illustrates the challenge in microfluidic heat sink design, where it is crucial that the design flow rate is high enough to achieve the desired heat transfer whilst also being low enough that the hydraulic penalties are realistic. Here, the FINJET heat sink achieves the effective conductance of 400 kW/m2K at a flow rate of 0.5 L/min, with a pressure drop of 1 bar (~100 kPa) and a hydraulic pumping requirement of less than 1.0 W. Regarding this last point, it is imperative that the required hydraulic power of the heat sink is significantly less than the thermal power being transported. This is quantified as the Coefficient of Performance (COP). As Fig. 8 shows, the COP of the FINJET heat sink is very high; being ~200 for the nominal flow rate of 0.5 L/min i.e. 200 times more thermal energy is transported through the heat sink than is required to force the liquid through it.

300

4

Pressure Drop (kPa)

3

200

2.5

150

2 1.5

100

1 50

Tin=27oC (300K) q//=1000 W/cm2

0 0

0.1

0.2

0.3

0.4 0.5 0.6 0.7 Flow rate (L/min)

0.8

0.9

Pumping Power (W)

3.5

250

0.5 0 1

Figure 7: Pressure drop and pumping power of the simulated FINJET heat sink with varying flow rate.

10000

COP (-)

1000

100

10

Tin=27oC (300K) q//=1000 W/cm2

1

0

0.1

0.2

0.3

0.4 0.5 0.6 0.7 Flow rate (L/min)

0.8

0.9

1

Figure 8: Coefficient of performance of the simulated FINJET heat sink with varying flow rate.

4.2 Thermal and Flow Analysis

The thermal hydraulic behavior of the FINJET heat sink will be discussed for the nominal flow rate of 0.5 L/min, applied heat flux of 1000 W/cm2 and inlet water temperature of 300 K, where it is understood that the same general flow and heat transfer behavior is observed over the range of flow rates, albeit for different magnitudes of the primary variables (flow velocity, temperature etc.) which have already been summarized above. Fig. 9 shows detail of the fluid flow through the heat sink by considering the flow velocity of a cross-section through the centerline of the jets. As it is shown, the liquid enters the upper chamber at a low velocity due to expansion of the flow as discussed. It is clear that the velocity is low and uniformly distributed over the jet nozzles. This forms an even distribution of high pressure liquid above the jets that forces it through the nozzles. As Fig. 9 shows, the flow is well distributed across the jets (from front to back), though the velocity increases somewhat moving upstream from the entrance. This is not necessarily a negative result since, on the exit port-side of the heat exchange system, the fluid is continually getting warmer which tends to cause increases in the base temperature in microchannels. The monotonically increasing jet velocities will have associated higher effective heat transfer coefficients thus helping to mitigate temperature gradients along the base. In fact, as the temperature map of Fig. 10 shows, the base temperature tends to decrease marginally from the front to the back of the heat sink, which is opposite to conventional microfluidic heat exchangers. Fig. 10 highlights that the uniform distribution of the flow results in good consistency in the heat transfer performance between the inlet and outlet, with only small variations in the temperature distribution profile from front to back. From an electronic component reliability point of view the uniformity of the base temperature would be of primary interest. Of course any local or global variations in the effective thermal conductance will cause variations in the base temperature and these will escalate with increased heat flux. Fig. 11 show the base temperature distribution for the 1000 W/cm2 test condition at 0.5 L/min. The result shows a difference between the maximum and minimum temperature of only 6oC (about ±3oC from mean) which, to the best of knowledge, is uncommonly low for this high an imposed heat flux for a single phase heat exchanger.

Figure 9: Velocity magnitude at the mid-plane of the FINJET heat sink (0.5 L/min, 1000 W/cm2).

Figure 10: Temperature at the mid-plane of the FINJET heat sink (0.5 L/min, 1000 W/cm2).

Figure 11: Two dimensional temperature distribution along the base (0.5 L/min, 1000 W/cm2). Fig. 12 and 13 show the flow and temperature characteristics for the 2nd and 3rd internal centerline jets. Fig. 12 shows the high velocity core of the flow within the nozzle which is of uniform temperature (Fig. 13). When the flow exits the nozzle it impinges on the surface forming a stagnation region directly beneath it. From the literature, the stagnation region is known to be the region of highest heat transfer and it is important to note here that the stagnation region is quite broad, in the sense that it is better than half the size of the jet nozzle, due to the slot jet-type of nozzle used here. If not for the heat spreading of the copper core within the base of the heat sink, one would observe a much more severe depression in the base temperature directly beneath the stagnation region. Once the jet has impinged vertically upon the surface it is redirected horizontally outward from the stagnation region. For conventional impinging jets, a wall jet is formed which has associated with it a severe drop in the local convective heat transfer coefficient [33]. This is due to jet deceleration as a result of expansion along the wall and entrainment as the wall thermal and hydraulic boundary layers develop [33]. To mitigate this deleterious effect, the chosen cellular jet structure confines the outward flowing liquid between the nozzle and the upward fin protrusion which forces the flow upward within a vertically orientated confined channel. Importantly, the pinching of the flow between the nozzle and the wall causes considerable acceleration of the flow and thus very effective heat transfer moving outward along the wall from the stagnation region.

Although there is some inevitable recirculation in the region where the flow is then forced upward, this is more than offset by the effective expansion of the flow into the vertical channel, noting that the vertical channel wall is an extended surface with a copper core, as depicted in Fig. 13. The combined effect of the highly mixed and confined flow and the heated surface extensions facilitates exceptional local heat transfer.

Figure 12: Velocity magnitude at the mid-plane of the FINJET heat sink for 2nd interior jet (0.5 L/min, 1000 W/cm2).

Figure 13: Temperature at the mid-plane of the FINJET heat sink for 2nd interior jet (0.5 L/min, 1000 W/cm2).

At the exit of the cell the fluid is exhausted into the microchannel section of the heat sink (Figs. 2 & 3). Here, the flow accumulates in an upper raised conduit section, flowing towards the exit of the heat exchanger. This is depicted in Fig. 14 which shows the flow velocity magnitude 0.1 mm above the floor of the microchannel section. Fig. 14 shows the associated temperature of the floor of the microchannel section.

Figure 14: Flow velocity magnitude 0.1 mm above the floor of the microchannel section (0.5 L/min, 1000 W/cm2).

Figure 15: Temperature of solid floor section of microchannel section (0.5 L/min, 1000 W/cm2).

Fig. 15 clearly shows that the fluid exiting the cells is cool relative to the temperature of the solid floor. In this way, this raised exhaust cannel can be considered a second-stage imbedded heat exchanger. An interesting feature of this type of design is that, since the local flow rates and associated velocities must increase from the entrance to exit (left to right) due to the addition of liquid from each subsequent jet cell, it causes the flow to accelerate, as depicted in Fig. 14. Again, this differentiates this technology from more conventional microchannel systems, where the unchanging flow rate combined with the escalating liquid temperature causes high temperature gradients when high heat fluxes are imposed. This accelerating flow effect in the 2 nd stage heat exchanger channel combined with the higher jet velocities previously discussed result in very small temperature gradients for this level of applied heat flux. Together they would also explain why there tends to be a reverse temperature gradient in the base of the heat sink i.e. in the opposite direction to the flow.

4.3. Performance Comparison Microfluidic heat exchangers have been conceptualized for a broad range of applications. As a result it is difficult to make accurate like-for-like performance comparisons of the current concept with other liquid cooled micro-heat exchangers. Here, an attempt is made to position the performance of the FINJET heat sink technology in the context of the state of the art of what can be considered compact and high performance heat exchangers. ‘Very compact’ has been rather arbitrarily chosen here as base areas ≤1 cm2 and ‘high performance’ relates to effective thermal conductance levels greater than 100,000 W/m2K and heat fluxes in the region of 1000 W/cm2. Performance penalties would be related to hydraulic considerations i.e. excessive flow

rate, pressure drop and/or pumping power. In order to attempt a like-for-like comparison with the literature, the heat transfer coefficient for the FINJET concept is evaluated based on the wall to liquid inlet temperature difference, opposed to the log-mean temperature difference, because insufficient information is given for the other chosen concepts to calculate the LMTD. This results in a moderate decrease of the approximated effective thermal conductance of about 7.5%. A range of micro-heat exchanger concepts are selected and they have been evaluated using a mixture of theoretical and/or experimental approaches. This includes the multi-objective-optimized predictions of Ndao et al. [31] for both straight channels and offset strip fins. Also included is the strip-fin device fabricated and tested by Colgan et al. [32]. Ditri et al. [34] proposed a compact embedded liquid microjet cooling concept specifically targeting high Power Amplifiers (HPAs), whereas Han et al. [35] and Escher et al. [36] both developed hybrid microchannel-microjet concepts. It should be noted that this is not an exhaustive list of compact and high performance liquid cooled micro heat exchangers, though can be considered a representative cross-section of existing concept technologies. It should also be noted that the heat flux capacity is calculated for each case assuming that the thermal conductance is independent of heat flux and it is increased until a 45 K surface temperature rise over the inlet coolant temperature occurs. Table 1 illustrates that the FINJET heat sink can be considered very compact and is predicted to achieve comparatively high thermal performance, with an effective thermal conductance and heat flux that is almost double that of the highest performing microfluidic heat exchanger concept of the chosen studies from the literature. The hydraulic penalties are reasonable and in-line with the other technology concepts. Also, a thermal performance metric that is difficult to quantify from existing studies is the base temperature homogeneity. As discussed, the FINJET concept is predicted to have very good base temperature uniformity for a single phase heat exchanger considering the level of heat flux. To put this into another context, the thermal conductance is above what is generally achieved with aggressive convective boiling, and the heat flux level is far above the CHF of advanced convective boiling concepts, see Refs [37-39].

Table 1. Thermal-hydraulic performance comparison

$

Thermal conductance of entire heat sink including base Based on a 45 K temperature rise of base over inlet water temperature

$$

5. Conclusions

A microfluidic heat exchanger has been designed with geometrical features and composite material structures that are possible to manufacture using the MICA Freeform manufacturing process. The underlying aim of the work was to take advantage of the unique manufacturing capabilities of this micro-3D metallic printing technology in order to design a very compact heat exchanger (<1 cm2) and very high heat flux (~1000 W/cm2) applications. To achieve this objective, a SimulationDriven Design process was followed whereby multi-physics simulations were carried out to solve the flow and conjugate heat transfer problem. The design concept that emerged was a hybrid microjet-microchannel (FINJET) heat sink, where the impinging microjets are confined within individual cells. The novelty of the work includes; 

The application of a novel metal composite additive manufacturing technology to the fabrication of microfluidic heat exchangers and the design of such a heat exchanger geometry based on the hybridization of two popular micro-heat exchange designs, namely microjets and microchannels. The fabrication of such a complex geometries and composite metallic structures is made possible with the MICA Freeform 3D additive manufacturing process considered here.



Although there is a large body of work for impinging jet arrays, including some that use hybrid microjets and microchannels, the concept of containing the jets in individual cells is novel and tackles the issue of wall jet formation and cross flow deteriorating jet thermal performance. Thus, high convective heat transfer coefficient levels of the stagnation region are maintained within the cells. This coupled with the dual purposing of the cell walls, which facilitate the high convective heat transfer coefficients, acting as fins, (including internal copper vias) is a new concept and is what facilitates such high overall thermal conductance.



The cellular concept ensures exceptional temperature uniformity, keeping in mind that this is absolutely crucial for microelectronics cooling applications, especially at this target level of heat flux of 1000 W/cm2.



The concept used all available surface area for heat transfer. The internal microchannel serves the hydraulic purpose of collecting and ejecting the spent flow from the jet cells. However, the design thermally links the channel floor as well as the protruding jet nozzles, which are arranged as offset strip fins, with the heat source. In this way the microchannel and jet nozzles also contributes to the heat transfer and the overall low thermal conductance of the heat sink.



The results of this study have yielded a design for a manufacturable single-phase micro-scale heat exchanger with unprecedented thermal conductance at acceptably low flow rates and pressure drops compared to previous studies found in the literature.

Figure 16 shows a photograph of the first manufactured FINJET prototype using the MICA Freeform process. Future work will involve experimental characterization of this new liquid cooling concept.

Figure 16: Photograph of first manufactured prototype FINJET heat exchanger. Acknowledgement Microfabrica Inc. would like to thank all of the consultants at Confluent Research Ltd. and at the SimuTech Group who contributed to this project.

References [1] W. Nakayama, ‘Heat in Computers: Applied Heat Transfer in Information Technology,’ ASME J. Heat Transfer, 136, 2014, pp. 013001-1 [2] M. Asadia, G. Xie, B. Sunden, A review of heat transfer and pressure drop characteristics of single and two-phase microchannels, International Journal of Heat and Mass Transfer, 79 (2014) 34–53 [3] D.B. Tuckerman and R.F.W. Pease, ‘High performance heat sinking for VLSI,’ IEEE Elec. Dev. Lett., 2, 1981, pp. 126129. [4] S.G. Kandlikar, History, Advances, and Challenges in Liquid Flow and Flow Boiling Heat Transfer in Microchannels: A Critical Review, Journal of Heat Transfer, 134 (2012) 034001-1 [5] S.G. Kandlikar, S. Colin, Y. Peles, S. Garimella, R.F. Pease, J.J. Bradner, D.B. Tuckerman. Heat Transfer in Microchannels—2012 Status and Research Needs, Journal of Heat Transfer, 135 (2012) 091001-1 [6] A.A. Alfaryjat, H.A.Mohammedb, Nor Mariah Adam, M.K.A. Ariffin, M.I. Najafabadi, Influence of geometrical parameters of hexagonal, circular, and rhombus microchannel heat sinks on the thermohydraulic characteristics, International Communications in Heat and Mass Transfer 52 (2014) 121–131 [7] J. Zhang, Y.H. Diao, Y.H. Zhao, Y.N. Zhang, An experimental study of the characteristics of fluid flow and heat transfer in the multiport microchannel flat tube, Applied Thermal Engineering, 65 (2014), pp. 209-218 [8] G. Xie, F. Zhang, B. Sundén, W. Zhang, Constructal design and thermal analysis of microchannel heat sinks with multistage bifurcations in single-phase liquid flow, Applied Thermal Engineering, 62 (2014) 791-802 [9] Yong Jiun Lee, Pawan K. Singh, Poh Seng Lee, Fluid flow and heat transfer investigations on enhanced microchannel heat sink using oblique fins with parametric study, International Journal of Heat and Mass Transfer, 81 (2015) 325–336. [10] Y. Li, F. Zhang, B. Sunden, G. Xie, Laminar thermal performance of microchannel heat sinks with constructal vertical Yshaped bifurcation plates, Applied Thermal Engineering 73 (2014) 185-195 [11] A. Ebrahimi, E. Roohi, Saeid Kheradmand, Numerical study of liquid flow and heat transfer in rectangular microchannel with longitudinal vortex generators, Applied Thermal Engineering 78 (2015) 576-583 [12] J. Zhao, Y. Wang, G. Ding, Y. Sun, G. Wang, Design, fabrication and measurement of a microchannel heat sink with a pinfin array and optimal inlet position for alleviating the hot spot effect, Journal of Micromechanics and Microengineering, 24 (2014) 115013 (9pp). [13] G. Xie, J. Liu, Y. Liu, B. Sunden, W. Zhang,Comparative Study of Thermal Performance of Longitudinal and TransversalWavy Microchannel Heat Sinks for Electronic Cooling, Journal of Electronic Packaging, 135 (2013) 021008-1 [14] J. Rostami, A. Abbassi, M. Saffar-Avval, Optimization of conjugate heat transfer in wavy walls microchannels, Applied Thermal Engineering 82 (2015) 318-328. [15] Z. Zheng, D.F.Fletcher, B.S. Haynes, Laminar heat transfer simulations for periodic zigzag semicircular channels: Chaotic advection and geometric effects, International Journal of Heat and Mass Transfer 62(2013)391–401. [16] L.Chai, G. Xia, L. Wang, M. Zhou, Z. Cui, Heat transfer enhancement in microchannel heat sinks with periodic expansion– constriction cross-sections, International Journal of Heat and Mass Transfer 62 (2013)741–751. [17] Y.L. Zhai, G.D. Xia, X.F. Liu, Y.F. Li, Heat transfer in the microchannels with fan-shaped reentrant cavities and different ribs based on field synergy principle and entropy generation analysis, International Journal of Heat and Mass Transfer, 68 (2014) 224–23. [18] J.Q. Yong, C.J. Teo, Mixing and Heat Transfer Enhancement in Microchannels Containing Converging-Diverging Passages, Journal of Heat Transfer, 136 (2014) 041704-1. [19] C. Leng, X.D. Wang, T.H. Wang, An improved design of double-layered microchannel heat sink withtruncated top channels, Applied Thermal Engineering, 79 (2015) 54-62 [20] B. Lu, W.J. Meng, F. Mei, Experimental investigation of Cu-based, double-layered, microchannel heat exchangers, Journal of Micromechics and Microengineering, 23 (2013) 035017 [21] A. Abdoli, G.S. Dulikravich, Multi-objective Design Optimization of Branching, Multifloor, Counterflow Microheat Exchangers, Journal of Heat Transfer, 136 (2014) 101801-1 [22] A.J. Robinson, E, Schnitzler, An Experimental Investigation of Free and Submerged Liquid Miniature Jet Array Impingement Heat Transfer, Experimental Thermal & Fluid Science, 32 (2007) 1-13. [23] C. J. Chang, H. T. Chen, C. Gau, Flow and Heat Transfer of a Microjet Impinging on a Heated Chip: Part II — Heat Transfer, Heat Transfer, Nanoscale and Microscale Thermophysical Engineering, 17(2) 92-111

[24] G.J. Michna, E.A. Brown, Y. Peles, M.K. Jensen, The effect of area ratio on microjet array heat transfer, International Journal of Heat and Mass Transfer, 54 (2011) 1782–1790 [25] E.A. Browne, G.J. Michna, M.K. Jensen, Y. Peles, Experimental Investigation of Single-Phase Microjet Array Heat Transfer, Journal of Heat Transfer, 132 (2010) 041013-1 [26] G.J. Michna, E.A. Browne, Y.Peles, M.K. Jensen, Single-Phase Microscale Jet Stagnation Point Heat Transfer, , Journal of Heat Transfer, 131 (2009) 111402-1 [27] M.K. Sung, I. Mudawar, Single-Phase and Two-Phase Hybrid Cooling Schemes for High-Heat-Flux Thermal Management of Defence Electronics, Journal of Electronic Packaging, 131 (2009) 021013-1 [28] Y. Han, B. L. Lau, X. Zhang, Package-Level Microjet-Based Hotspot Cooling Solution for Microelectronic Devices, IEEE Electronic Device Letters, 36(5) (2015) 502-504 [29] Y. Han, B.L. Lau, X. Zhang, Y.C.Leong, K.F. Choo, Thermal Management of Hotspots With a Microjet-Based Hybrid Heat Sink for GaN-on-Si Devices, IEEE Transactions on Components, Packaging and Manufacturing Technology, 4(9) (2014) 1441-1450. [30] Y. Han, B.L. Lau, H. Zhang, X. Zhang, Package-level Si-Based Micro-jet Impingement Cooling Solution With Multiple Drainage Micro-Trenches IEEE 16th Electronics Packaging Technology Conference, 2014. [31] S. Ndao, Y. Peles, M.K. Jensen, Multi-objective thermal design optimization and comparative analysis of electronics cooling technologies, International Journal of Heat and Mass Transfer 52 (2009) 4317–4326. [32] E.G. Colgan, B. Furman, M. Gaynes, W.S. Graham, N. LaBianca, J.H. Magerlein, R.J. Polastre, M.B. Rothwell, R.J. Bezama, R. Choudhary, K.C. Marston, H. Toy, J. Wakil, J.A. Zitz, R.R. Schmidt, A practical Implementation of Silicon Microchannel Coolers for High Power Chips, IEEE Transactions on Components and Packaging Technologies, 30(2) (2007) 218-225. [33] T.S. O'Donovan, D.B. Murray, Jet impingement heat transfer - Part I: Mean and root-mean-square heat transfer and velocity distributions, International Journal of Heat and Mass Transfer, 50 (17-18) (2007) 3291-3301. [34] J. Ditre, J. Hahn, R. Cadotte, M. McNulty, D. Luppa, Embedded Cooling of High Heat Flux Electronics Utilizing Distributed Microfluidic Impingement Jets, Proceedings of the ASME 2015 International Technical Conference and Exhibition on Packaging and Integration of Electronic and Photonic Microsystems InterPACK2015, nJuly 6-9, 2015, San Francisco, California, USA. [35] Y. Han, B. Long Lau, X. Zhang, Y.C. Leong, K.F. Choo, Thermal Management of Hotspots With a Microjet-Based Hybrid Heat Sink for GaN-on-Si Devices, IEEE Transactions on Components, Packaging and Manufacturing Technology, 4(9) (2014), 1441-1450 [36] W. Escher, T. Brunschwiler, B. Michel, D. Poulikakos, Experimental Investigation of an Ultrathin Manifold Microchannel Heat Sink for Liquid-Cooled Chips, J. Heat Transfer 132(8), 081402. [37] R. Jenkins, C. de Brún, T.L. Lupton, R. Lupoi, R. Kempers, A.J. Robinson, (2017) “Convective Boiling of Confined Impinging Jet Arrays,” Experimental Thermal Fluid Sciences, 86 (2017) 224-234 [38] R. Jenkins, R. Lupoi, R. Kempers, A.J. Robinson, (2017) “Heat Transfer Performance of Boiling Jet Array Impingement on Micro-Grooved Surfaces,” Experimental Thermal Fluid Sciences, 80 (2017) 293–304. [39] S.G. Kandlikar, T. Widger, A. Kalani, V. Mejia, “Enhanced flow boiling over open microchannels with uniform and tapered gap manifolds,” Journal of Heat Transfer, 135 (6) (2013) 061401.

    

CFD simulated performance of a MEMS single phase liquid water cooled heat sink. A hybrid micro-jet and microchannel concept is proposed. Design in amenable to MICA Freeform additive manufacturing capabilities. Simulations were performed for a 3 mm x 4 mm chip with heat flux of 1000 W/cm2. An effective thermal conductance of 400 W/cm2K has been predicted.