IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 5, MAY 2012 825 Numerical Analysis of Novel Micro Pin Fin Heat Sink With Variable Fin Density Carlos A. Rubio-Jimenez, Satish G. Kandlikar, and Abel Hernandez-Guerrero Abstract— Numerical analyses to characterize and design micro pin fin heat sinks for cooling the 2016s IC chip heat generation are carried out in this paper. A novel design with variable fin density is proposed to generate a more uniform temperature of the IC chip junctions. The variable-density feature allows the gradual increase of the heat transfer area as coolant passes through the system. Single-phase water in the laminar regime is employed. Four different fin shapes (circle, square, elliptical, and flat with two redounded sides) are analyzed. The junction temperature and pressure drop variations in the heat sink generated by these shapes are presented. The effects of varying the fin length and height are also studied. The best heat sink configuration has a thermal resistance ranging from 0.14 to 0.25 K/W with a pressure drop lower than 90 kPa and a junction temperature ∼314 K under the conditions studied. The temperature gradient at the bottom wall of the heat sink is considered as a parameter for comparing various heat sink designs. The novel cooling device has an overall temperature gradient lower than 2 °C/mm, which is significantly lower than the temperature gradients in other schemes reported in literature. Index Terms— Micro pin fin heat sink, single-phase flow, uniform junction temperature, variable fin density. I. I NTRODUCTION C URRENTLY, electronic devices have revolutionalized industrial processes and the entertainment technology sector. IC chips are the key components in these devices. The new generation of electronic devices has a higher capacity to process information and is smaller in size than its predecessor. This development has been possible because every component is now highly reliable, smaller, lighter, and cheaper with short design cycles. Several technological challenges have helped to achieve this status, such as new materials and manufacturing techniques, novel electronic designs, better handling of high heat fluxes, and so on. In the heat dissipation field, according to The International Technology Roadmap for Semiconductors, the challenge is transcendental. The 2016s 22-nm complementary metal–oxide–semiconductor devices are expected to Manuscript received September 1, 2011; accepted February 17, 2012. Date of publication April 10, 2012; date of current version May 3, 2012. This work was supported in part by CONACYT-Mexico. Recommended for publication by Associate Editor D. Agonafer upon evaluation of reviewers’ comments. C. A. Rubio-Jimenez was with the University of Guanajuato, Salamanca 36885, Mexico. He is now with the Department of Mechanical Engineering, Rochester Institute of Technology, Rochester, NY 14623 USA (e-mail: caalruji@yahoo.com.mx). S. G. Kandlikar is with the Department of Mechanical Engineering, Rochester Institute of Technology, Rochester, NY 14623 USA (e-mail: sgkeme@rit.edu). A. Hernandez-Guerrero is with the Department of Mechanical Engineering, University of Guanajuato, Salamanca 36885, Mexico (e-mail: abel@ugto.mx). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TCPMT.2012.2189925 generate up to 288 W of heat, while limiting the maximum design temperature to <85 °C [1]. Therefore, the new cooling systems to be used in these electronic devices should have a thermal resistance ranging from 0.14 to 0.25 °C/W, with higher reliability and smaller size than the current cooling systems (air-cooling technologies) which are reaching their maximum performance limit. One alternative for cooling electronic devices emerged with the pioneering work of Tuckerman and Pease [2] three decades ago. Their experiments, based on internal flow, demonstrated that the heat transfer coefficient increases as the hydraulic diameter is reduced to microscale. Their results showed that a microchannel heat sink can achieve a thermal resistance of 0.09 K/W and is capable of dissipating up to 790 W/cm2 with a temperature rise of 71 °C. Although these results indicate that this cooling system dissipates the heat fluxes that are expected in the near future, there are some important technical drawbacks: 1) the fluid pressure drop is very large (>200 kPa); and 2) the junction temperature of the heat sink and the chip is highly nonuniform along the flow direction (∼70 °C of temperature difference in only 10 mm). These drawbacks make the cooling system unreliable and seriously affect the performance and lifetime of the IC chips [3], [4]. Severe damage, such as cumulative failure, fatigue per cycle in some materials, acceleration of chemical reactions, metal and electronic migration, and so on, may result on chips when the heat removal is handled inappropriately. Moreover, the temperature variation causes clock skew [5]. In the last few years, numerous studies have been carried out considering microchannels in order to understand the microscale phenomena [6]–[8] and generate reliable cooling systems [9]–[11]. In a similar way, nonconventional microcooling systems based on different techniques have been proposed every year [12]–[14]. However, only a few works have shown good performance. Peles et al. [15] proposed a micro pin fin heat sink as an alternative for cooling IC chips. Their analytical–experimental study was based on correlations developed for coolant flow through a bank of tubes at macroscale. Their results showed that this arrangement dissipates up to 790 W/cm2 with a temperature rise of 30.7 °C; this means that the system achieves a thermal resistance of 0.039 K/W. Unfortunately, the pressure drop was larger than 200 kPa. Colgan et al. [16], [17] proposed a nonconventional heat sink formed by two layers using water and a fluorinated hydrocarbon fluid as coolants. Several staggered fins and multiple inlet/outlet sections were manufactured on the first and second layers, respectively. Their results showed that this system, using water, dissipates up to 500 W/cm2 and 2156–3950/$31.00 © 2012 IEEE 826 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 5, MAY 2012 achieves a thermal resistance as low as 12 mm2 °C/W with a maximum pressure drop of 65 kPa. Stenike and Kandlikar [18] experimentally determined the heat transfer coefficient with the silicon offset strip fin arrangement to be in excess of 500 000 W/m2 °C using single-phase flow of water. Recently, Wälchli et al. [19] proposed a heat sink configuration formed by three interconnected layers. Water flowing through the system experiences several direction changes due to the horizontal and vertical channel interconnections. This results in an increase in the heat transfer coefficient and, thus, a reduction in the thermal resistance. Their experimental results showed that this system yields a thermal resistance of around 0.08 K/W with a pressure drop of 40 kPa. Following the same research theme, Escher et al. [20] proposed a novel heat sink configuration based on manifold channels. Their numerical results showed that this heat sink dissipates up to 750 W/cm2 with a temperature difference between the fluid inlet and the chip of 65 K and a pressure drop lower than 10 kPa. A thermal resistance of 0.087 cm2 K/W was reported. The heat sink by Escher et al. [20] is able to dissipate the same heat flux as reported by Tuckerman and Pease [2], but with a pressure drop that is 10 times lower. Clearly, these recent studies provide a promising path for meeting the heat dissipation requirements for the 2016s IC chips. Moreover, the technical literature shows that the uniformity of the chip temperature has not been considered as a design parameter in cooling devices employing single-phase flow in laminar region. Microchannel heat sinks using two-phase flow or boiling have been studied as an alternative for providing uniform chip temperature [21]; however, they require the usage of water at subatmospheric pressures due to the IC chip temperature limit of around 80 °C. The usage of a refrigerant adversely affects the performance and increases the cost of the system. In this paper, a novel micro pin fin heat sink configuration with variable fin density is proposed and analyzed. This configuration offers a reliable cooling system capable of dissipating high heat fluxes at a low pressure drop while maintaining uniform junction temperature. II. U NIFORM T EMPERATURE D ISTRIBUTION C ONCEPT A large part of the liquid-cooling technologies is based on forced convection mechanism in internal flow. According to the theory, the temperature difference between the coolant and channel wall remains constant along the flow length when fully developed condition and constant properties are assumed under a constant heat flux boundary condition. The coolant temperature rises along the flow length [Tfluid = Tfluid (z)] and, therefore, the channel wall temperature increases accordingly. Assuming that the temperature difference between the heat sink and the junction remains constant, the junction temperature will then vary in a similar manner as the fluid temperature along the flow direction [TIC = TIC (z)]. Based on the conservation of thermal energy, the temperature difference between coolant and channel wall is determined by the heat flux and the convective thermal resistance (Fig. 1). Thus, if the thermal resistance is constant, it would result in a temperature Fig. 1. Sketch of temperature variation in internal flow systems subject to constant heat flux and thermal resistance. Fig. 2. Sketch of temperature variation in an internal flow system subject to constant heat flux, uniform wall temperature, and variable total thermal resistance. variation as shown in Fig. 1 and described by Eq. (1) Tbottom (z) − Tfluid (z) = q Rtotal . (1) According to (1), and considering a constant heat flux, a constant junction temperature is achieved when the thermal resistance reduces along the flow length as indicated in Fig. 2. Since the thermal resistance varies inversely with the product h A, this product should increase linearly along the flow length to achieve a uniform surface temperature. Microchannels technical literature indicates that the heat transfer coefficient increases when: 1) the channel hydraulic diameter decreases; or 2) the channel aspect ratio decreases. Although these options improve the heat dissipation in conventional microchannel heat sinks, the pressure drop goes up in an inverse cubic proportion to the hydraulic diameter. Moreover, a single microchannel should be divided into a number of sections along the flow length in order to gradually increase the heat transfer coefficient and generate uniform junction temperature. This clearly affects the complexity and cost of the cooling device. On the other hand, using variable density pin fins provides a simple way to enhance heat transfer rate while maintaining a uniform surface temperature. Pin configuration has been analyzed by a few investigators [15], [22], but with constant fin density. Section III presents the design of the fin arrangements with variable fin density for the heat sinks proposed in this paper. RUBIO-JIMENEZ et al.: NUMERICAL ANALYSIS OF NOVEL MICRO PIN FIN HEAT SINK SL,I SL,III ST,I Flow direction ST,III ST,II SL,II 827 ... ... SI Fig. 3. SII SN SIII Sketch of sections marked on the square heat sink. The transversal and longitudinal rows, ST and S L dimensions, and fin places are shown. TABLE I H EAT S INK PARAMETERS FOR THE T HREE C ONFIGURATIONS 50 µ m Parameter Flow direction 25 µm Fin length Fig. 4. Fin length Pin fin shapes and dimensions considered in the analyses. III. M ODEL D EFINITION A 10-mm × 10-mm IC chip surface is considered as the base for designing and building the novel micro pin fin heat sinks with variable fin density. The cooling system is formed by fins placed on a 200-μm-thick silicon substrate. In order to generate the variable fin density, sections (SI, SII, … SN) are marked parallel to the flow direction on this surface (Fig. 3). A number of transversal and longitudinal rows separated by ST and SL , respectively, are marked at each section. One pin fin is placed at each intersection. Three overall configurations are developed. 1) Configuration I: micro pin fin heat sinks formed by 990 fins placed on four sections. The fin height is 200 μm. Four fin shapes are considered (circle, square, ellipse, and flat with two rounded sides as shown in Fig. 4). The fin length is varied (75, 100, and 150 μm). 2) Configuration II: micro pin fin heat sinks formed by 3696 fins placed on four sections. The fin height is varied (100, 200, and 300 μm). Flat-shaped fins are used. 3) Configuration III: micro pin fin heat sinks formed by 4748 fins placed on three sections. The fin height is 200 μm. Flat-shaped fins are considered. Table I shows the geometrical parameters for these configurations. Additionally, the following two configurations are also analyzed for comparison. 1) Microchannel heat sink formed by 33 rectangular channels with 200-μm height, channel aspect ratio of 1.0 and space between the channels of 100 μm. Length (mm) No. of pin fins No. of longitudinal rows No. of transversal rows S L (μm) ST (μm) ρ pin fin (no. of pin fins/mm3 ) Section I Section II Section III Section IV Configuration I 1.4 2.8 4.2 1.6 0 231 462 297 0 7 7 3 0 33 66 99 0 0 400 300 600 150 600 100 0 41.66 55.55 93.75 4.2 1848 1.8 1188 Length (mm) No. of pin fins No. of longitudinal rows No. of transversal rows S L (μm) ST (μm) ρ pin fin (no. of pin fins/mm3 ) 1.0 0 Length (mm) No. of pin fins No. of longitudinal rows No. of transversal rows S L (μm) ST (μm) ρ pin fin (no. of pin fins/mm3 ) 1.5 330 Configuration II 3.0 660 0 20 28 12 0 33 66 99 0 0 150 300 150 150 150 100 0 111.11 222.22 333.33 3.3 2178 – – Configuration III 5.2 2240 10 34 22 – 33 66 99 – 150 300 150 150 150 100 – – 111.11 217.94 333.33 – 2) Heat sink formed by a single large channel of 200-μm height and 9.9-mm width. This model is called non-pin fin heat sink. The thickness of the silicon substrate under the channels for all models is 200 μm. Due to the symmetry that the heat sinks present, only 1/66 part of the heat sink was modeled. The effects of the external walls in the large channel are considered negligible. 828 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 5, MAY 2012 Symmetries Flat-shaped fins the assumptions and boundary conditions mentioned previously. The finite-volume vertex-centered code was used for getting the initial approach. A second-order upwind scheme was used for discretizing the momentum equation, whereas a SIMPLE algorithm was used for the pressure–velocity coupling [25]. Convergence criterion was set to 1 × 10−6 for residuals ∇U = 0 ρ (U · ∇U ) + P − = 0 U · ∇T = 0 ρ c p (U · ∇T ) − k f ∇ 2 T = 0 Fluid flow Silicon substrate Fig. 5. Mesh generated in the MF-50 × 100 ×200-66 model formed by >500 000 hexahedral elements. IV. N UMERICAL A NALYSIS A. Assumptions The following assumptions are made in the numerical modeling: 1) steady state; 2) water enters the heat transfer area with uniform velocity at ambient temperature (293 K); 3) constant fluid and solid thermal properties (except fluid viscosity, given by (2) [23], [24]); 4) constant and uniform heat flux from the substrate; 5) negligible radiation effects (2) μ = 2.414 × 10−5 10247.8/(T −140) . B. Mesh Generation and Boundary Conditions The mesh generated for each model is formed by approximately half million hexahedral elements distributed uniformly, except in the section near fin walls where a cell ratio of 1.025 is set. This mesh is adapted to the fin shape. Fig. 5 shows a part of the solid domain and its mesh. A mesh sensitivity analysis was performed to reach the appropriate number of elements for achieving convergence. The boundary conditions for the models are adjusted according to the fluid-surrounding interactions. A constant mass flow rate is considered at the fluid inlet section of the model. A constant heat flux is set at the bottom wall of the solid domain. A mass conservation condition was adjusted at the fluid outlet section. Symmetry conditions for both domains are considered at the symmetry walls. The walls which interact between both domains are set as interface conditions. The upper wall of the channel and the fins are considered adiabatic. C. Numerical Model and Validation Commercial CFD software was used for numerically solving the governing equations (3)–(6) for both domains considering (3) (4) for solid (5) for fluid. (6) Results are generated for the flow and temperature fields, and pressure domain in both domains. In order to validate the numerical results, the fluid temperature difference in a micro pin fin heat sink model and the pressure drop in the microchannel heat sink model were determined and compared with the analytical results. The average temperature and pressure values at the fluid inlet and outlet sections are calculated using numerical surface integration using (7) and (8), respectively. Equations (9) and (10) were used to determine analytically the fluid temperature and pressure difference. For the pin fin heat sink geometry with 1 mL/s of water and a heat flux of 100 W/cm2 , the numerical and analytical fluid temperature differences are 24.07 and 23.91 K, respectively. The error is lower than 0.7%. For the microchannel heat sink geometry with 1 mL/s of water and the same heat flux, the numerical and analytical pressure drop are 1.65 and 1.70 kPa, respectively. Overall, the agreement with the numerical results and theoretical predictions is seen to be reasonable 1 Tx,y d Ac out − Tx,y d Ac in (7) T = A c Ac A c 1 P = (8) Px,y d Ac out − Px,y d Ac in A c Ac Ac q As T = (9) ṁcp 2Poμu L ρu 2 P = . (10) + K Dh 2 2 V. R ESULTS A. Effects of Pin Fin Shapes Fig. 6 shows the junction temperature variation of the micro pin fin heat sink based on Configuration I with different fin shapes along the dimensionless flow direction (z ∗ = z/L). The operating conditions are: flow rate of 1 mL/s and heat flux of 100 W/cm2 . For simplification, the heat sinks are called as MA-50 × LLL × HHH-NN, where: 1) A indicates the fin shape (C-circle, S-square, E-ellipse, and F-flat); 2) LLL indicates the fin length (μm); 3) HHH indicates the fin height (μm); 4) NN indicates the number of longitudinal rows that forms the arrangement. The temperature variations for microchannel and non-pin fin heat sink are studied as well. Although these last two models RUBIO-JIMENEZ et al.: NUMERICAL ANALYSIS OF NOVEL MICRO PIN FIN HEAT SINK Fig. 6. Variation of the junction temperature along the dimensionless flow length for the (a) microchannel heat sink, (b) non-pin fin heat sink, and (c)–(f) micro pin fin heat sinks based on Configuration I with different fin shapes (Table II). q = 100 W/cm2 and flow rate = 1 mL/s. TABLE II OVERALL P ERFORMANCE OF M ICROCHANNEL , N ON -P IN F IN , AND M ICRO P IN F IN H EAT S INK B ASED ON C ONFIGURATION I W ITH D IFFERENT F IN S HAPES Heat sink a) Microchannel b) Non-pin fins c) MC50 × 50 × 200-17 d) MS-50 × 50 × 200-17 e) ME-50 × 75 × 200-17 f) MF-50 × 75 × 200-17 342.75 389.16 349.06 41.31 68.28 14.72 5.80 1.61 4.63 Pumping γ R Power (K/W) (mm2 /mm3 ) (W) 0.498 0.194 15.00 0.962 0.053 5.20 0.561 0.152 6.91 344.02 12.06 5.76 0.510 0.190 7.40 344.30 10.06 5.01 0.513 0.166 7.42 342.46 8.47 5.61 0.495 0.185 7.47 Tbottom,avg Tbottom (K) (K) P (kPa) have different temperature magnitudes caused mainly by the increase in the heat transfer coefficient when microchannels are used, the paths are similar. The lowest temperature is found at the fluid inlet section as expected and increases almost linearly. The largest temperature is found at z ∗ = 1.0. The temperature differences between the bottom wall and the fluid at the outlet are ∼41 and ∼68 K for microchannel and non-pin fin heat sinks, respectively. Moreover, the temperature profiles generated by the proposed micro pin fin heat sinks at this wall are more uniform because of the fin density variation. These temperature curves show four slight peaks/valleys near the transition zones among fin sections (Fig. 3). These variations are caused by three factors: 1) increase in the heat transfer area from one section to the next; 2) increase in the heat transfer coefficient with the increase in fin density; and 3) the entrance region effect. In the temperature curves for the variable fins (C-E), it is observed that the circular pin fin (C) presents the highest junction temperature as well as the largest temperature difference between maximal and minimal points. The flatshaped fin improves the heat dissipation since it has the lowest junction temperature as well as the lowest temperature difference. The overall performance of these configurations is shown in Table II. The heat transfer area to fluid volume 829 Fig. 7. Variation of the pressure drop along the dimensionless flow length for the (a) microchannel heat sink, (b) non-pin fin heat sink, and (c)–(f) micro pin fin heat sinks based on Configuration I with different fin shapes (Table II). q = 100 W/cm2 and flow rate of 1 mL/s. ratio γ is considered as a parameter for comparison. The results show that the heat dissipation is improved when γ is slightly increased (e.g., the increase in γ between circleand flat-shaped fin configurations is ∼1.1%) and the overall temperature difference is reduced by 6.25 K. The square (D) and the elliptical (E) fin configurations present similar temperature results with minimal differences at the inlet section. One important observation is that the fin shape does not affect the heat dissipation since it is possible to achieve a similar temperature profile with square- or ellipse-shaped fin configurations because γ is almost the same in both cases. Fig. 7 shows the pressure drop variation of micro pin fin heat sink configurations with 17 rows and different fin shapes. Microchannel and non-pin fin heat sinks pressure drop variations along the flow direction are shown as well. For these last two cases, the pressure drop variation is linear with slight curvature at the fluid inlet section due to the hydrodynamic development effects. There is a difference of ∼4 kPa between the two heat sinks due to the decrease of the channel hydraulic diameter when microchannels are used. Moreover, the pressure curves of pin fin heat sink models show several waves along the flow length. The increase/decrease of the fluid velocity across the fin arrangement and the stagnation points near the fin walls cause these paths. Overall, the results show that the pressure drop increases when the transversal pitch is reduced, for a given longitudinal pitch in the section. Also, it is observed that the fin shape plays a very important role in the energy lost by friction effects. Analyzing the results for these four configurations, it is observed that the lowest pressure drop is generated when circular fins are used (4.6 kPa). Moreover, the largest pressure drop is found in square- and flat-shaped fin configurations (∼5.7 kPa), although there is a 50% difference of fin length between the two geometries. Thus, it is possible to highlight that the frictional losses are greatly reduced when the fin walls perpendicular to the flow direction are redounded, and rise slightly when the fin length increases. These effects are more complex when ellipse- and flat-shaped fins are compared. Although they have the same fin length, the pressure drop is lower in the former (∼12% lower), because it has a larger wall curvature than the latter. According to these results, the usage of ellipse- and flat-shaped fin arrangements formed by 17 rows 830 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 5, MAY 2012 Fig. 8. Overall junction temperature and pressure drop for pin fin heat sinks with ellipse- and flat-shaped fins based on Configuration I and different fin lengths. q = 100 W/cm2 and flow rate = 1 mL/s. Fig. 9. Variation of the junction temperature along the flow length in the heat sinks based on Configuration II with different fin heights and flow rates. q = 100 W/cm2 . TABLE III OVERALL P ERFORMANCE OF M ICROCHANNEL , N ON -P IN F IN , AND M ICRO P IN F IN H EAT S INK BASED ON C ONFIGURATION I W ITH E LLIPSE - AND F LAT-S HAPED F INS S UBJECT TO D IFFERENT F IN L ENGTHS Heat sink g) ME-50 × 75 × 200-17 h) ME-50 × 100 × 200-17 i) ME-50 × 150 × 200-17 j) MF-50 × 75 × 200-17 k) MF-50 × 100 × 200-17 l) MF-50 × 150 × 200-17 Tbottom,avg Tbottom (K) (K) γ 344.30 10.06 P (kPa) 5.01 340.73 8.50 5.39 0.477 8.00 335.83 4.31 5.74 0.428 9.26 342.46 8.47 5.61 0.495 7.47 339.13 4.19 6.17 0.461 8.14 335.33 5.15 7.46 0.423 9.96 R (K/W) 0.513 (mm2 /mm3 ) 7.42 produces better performing heat sinks. However, the thermal resistance is larger than the value required by 2016s IC chips. B. Effects of Fin Length Fig. 8 shows the junction temperature and pressure drop of the heat sink with ellipse- and flat-shaped fins for three different fin lengths (75, 100, and 150 μm). The heat dissipation is improved when the fin length is increased because of the increase of γ (Table III). The average bottom temperature is reduced 2.5 and 2.1% for ellipse- and flat-shaped fins, respectively, when the fin length is varied from 75 to 150 μm. Moreover, the pressure drop is larger when flat-shaped fins are used due to the increased area of the surface walls parallel to the flow direction. The results indicate that, although the ellipse-shaped fin heat sink has the largest improvement in the heat dissipation with the increase in the fin length, the lowest IC chip temperature is achieved using flat fins. Thus, it is highly recommended to use the latter fin shape in this kind of micro heat sink. C. Effects of Fin Height and Flow Rate Fig. 9 shows the junction temperature variation along the flow length for micro heat sinks based on Configuration II with different fin heights and flow rates. Clearly, the heat dissipation Fig. 10. Thermal resistance variation for different flow rates and fin heights (h f ) based on Configuration II. q = 100 W/cm2 . is improved when the amount of coolant increases; however, the junction temperature variation loses uniformity. Two different effects in the two adjacent sections cause this uniformity loss. First, the heat sink works as a single channel in the first section (SI) whereby the heat dissipation is improved when the fin height (channel height) decreases and vice versa (e.g., the bottom temperature at z ∗ = 0.1 is 321.5 and 331.46 K when the fin height is 100 and 300 μm, respectively). Second, the increase in the fin height improves the heat dissipation at the subsequent sections (SII–SIV) because the heat transfer area increases (e.g., the bottom temperature at z ∗ = 0.8 is 328.06 and 324.2 K when the fin height is 100 and 300 μm, respectively). The variation in the fin height positively affects the heat dissipation of the system at one section, and negatively affects at another section at the same time, thus seriously affecting the temperature uniformity. Therefore, it is recommended to avoid sections without fins in this kind of micro heat sink. Fig. 10 presents the overall thermal resistance variation in the heat sinks based on Configuration II for different flow rates and fin heights. The thermal resistance is almost the same for the three fin height cases when the flow rate is below 1.5 mL/s. For flow rates higher than 1.5 mL/s, the effects of increasing the heat transfer area becomes more important for the heat dissipation. These results show that the 60-rows heat sink configuration requires a higher flow rate of water for getting thermal resistance values around 0.1 K/W since this curve decreases slightly after 2 mL/s. RUBIO-JIMENEZ et al.: NUMERICAL ANALYSIS OF NOVEL MICRO PIN FIN HEAT SINK Fig. 11. Pressure drop variation for different flow rates and fin heights based on Configuration II. q = 100 W/cm2 . TABLE IV 831 Fig. 12. Variation of fluid temperature and junction temperature along the flow direction for (a) conventional microchannel heat sink, (b) non-pin fin heat sink, and (c) conventional pin fin heat sink with constant fin density based on Configuration III. OVERALL P ERFORMANCE OF M ICRO P IN F IN H EAT S INK BASED ON C ONFIGURATION II W ITH D IFFERENT F IN H EIGHTS AND F LOW R ATES Tbottom (K) R (K/W) Mass flow rate (mL/s) Tbottom,avg (K) 1 326.09 10.68 44.13 0.339 2 316.15 6.32 98.80 0.232 3 311.88 7.72 163.04 0.189 P(kPa) m) MF-50100 × 100-60 n) MF-50 × 100 × 200-60 1 326.47 10.30 13.87 0.335 2 3 316.92 312.78 10.50 9.01 30.04 49.50 0.239 0.198 1 326.55 11.90 7.99 0.336 2 317.24 12.32 17.30 0.242 3 313.41 11.57 27.68 0.204 o) MF-50 × 100 × 300-60 Fig. 11 shows the pressure drop variation when the fin height is varied for different flow rates. The pressure rises when the flow rate increases; however, the friction losses are larger for fin heights lower than 200 μm. The effects of the single channel at SI and the decrease of the gap between fins (“apparent channels”) in the subsequent sections cause this large pressure drop. The fin height effects are minimal after this point for fin heights larger than 250 μm. Table IV presents the overall performance of Configuration I subject to different fin heights and flow rates. These results show that the usage of fins with 200-μm height generates good results. The heat sink could achieve thermal resistance values around 0.2 K/W. D. Performance of MF-50×100×200-66 Heat Sink Fig. 12 shows the junction temperature profile of the micro pin fin heat sink based on Configuration III. The temperature profiles generated with microchannel heat sink and non-pin fin heat sink models are added for comparison. Fig. 13 shows that this heat sink configuration reaches a comparably lowaverage bottom temperature with a uniform temperature profile (Tbot,ave = 39.9 °C, T = 5.7 °C) as compared to the Fig. 13. Thermal resistance and pressure drop variation for different flow rates in MF-50×100×200-66 heat sink. microchannel heat sink (Tbot,ave = 57 °C, T = 38 °C) and the non-pin fin heat sink (Tbot,ave > 100 °C, T = 68 °C). In this paper, the temperature gradient along the flow length is used as a parameter for comparison. Thus, according to the temperature differences shown above, the microchannel heat sink, the non-pin fin heat sink, and the MF-50×100×200-66 heat sink have temperature gradients of 6.83 °C/mm, 3.75 °C/mm, and only 1.63 °C/mm, respectively. Since the lower temperature gradients reduce thermal stresses and other adverse effects, the proposed heat sink is expected to increase the lifetime of the IC chips. Revisiting Fig. 12, there are two drastic zones found in the temperature curves of the MF-50 × 100 × 200-66 heat sink at the positions where changes of section take place. Considering the maximal and minimal points at the vicinity of these zones, the temperature gradient is ∼2.7 °C/mm. Although this value is larger than the overall temperature gradient of the device, it remains below the gradient values found in the microchannel heat sink and non-pin fin heat sink. It is possible to redesign the local fin configuration in this zone to reduce this effect. Fig. 13 shows the thermal resistance and pressure drop variation of the MF-50 × 100 × 200-66 heat sink for different flow rates. These results show that this proposed heat sink is attractive since it is capable of achieving the thermal resistance 832 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 5, MAY 2012 TABLE V P ERFORMANCE OF S EVERAL H EAT S INKS I NTO THE L IQUID -C OOLING T ECHNOLOGIES AND THE MF-50 × 100 × 200-66 H EAT S INK Author Description Tuckerman and Pease (1981) [2] Knight et al. (1992) [9] Rectangular microchannel heat sink Rectangular microchannel heat sink using turbulent flow Rectangular microchannel heat sink for multichip modules Micro heat sink with circular staggered pin fins Micro heat sink with semielliptical staggered pin fins Optimized microchannel heat sink Radial heat sink with boiling fluid Gillot et al. (2000) [10] Peles et al. (2005) [15] Colgan et al. (2007) [17] Husain and Kim (2008) [11] DaguenetFrick et al. (2010) [26] Escher et al. (2010) [20] This paper (2011) Heat sink formed by manifold channels MF-50 × 100 × 200-66 P (kPa) R (K/W) 207 0.090 Pump q ∗ Power max 2 (W/cm ) (W) 2.3 >650 207 0.056 >10.0 >1000 180 0.092 ∼47.0 >650 203 0.039 – >1500 <35 0.105 <0.9 >500 – 0.081 – >700 – 0.080 0.05 750 <10 0.087 ∼0.15 >680 29.5 0.207 0.04 230 ∗ Considering 1 × 1-cm IC chip with maximum design temperature of 85 °C and ambient temperature of 25 °C. needed by 2016s IC chips [1] (0.14–0.25 °C/W) with a reasonable pressure drop penalty (28–90 kPa). Also, the pumping power required in the systems is ranging from 0.02 to 0.34 W which is at least 10 times lower than microchannel heat sinks. Table V presents the overall performance of different micro heat sink systems studied in the literature. The MF-50× 100×200-66 performance is also shown for comparison. The results show that the proposed heat sink has a pressure drop considerably lower than other microchannel heat sinks [2], [9], [10] and the pin fin heat sinks proposed by Peles et al. [15]. The cooling device proposed by Escher et al. [20] is seen to have a lower resistance value due to multiple manifolds. However, the low flow rate required by MF-50×100×200-66 makes it more reliable than Escher et al.’s device [20] and microchannel heat sinks [2] because the pumping power is only 0.04 W (∼4 times lower than the Escher et al.’s heat sink and ∼50 times lower than Tuckerman and Pease’s heat sink). Only the boiling system proposed by Daguenet-Frick et al. [26] presents similar pumping power value. Although this proposed cooling system has the largest thermal resistance of all devices, it could effectively dissipate the maximum heat generation rate of 288 W expected in 2016 IC chips while maintaining the surface temperature within allowable limit and presenting the lowest temperature gradients among the systems analyzed in this paper. VI. F UTURE W ORK The analysis presented in this paper clearly identifies the benefits of using variable-density fins in 3-D IC packaging for achieving uniform junction temperature along the flow length. Efforts are being continued at Rochester Institute of Technology to experimentally validate these findings. VII. C ONCLUSION Micro pin fin heat sinks with variable fin density were analyzed numerically in order to dissipate the high heat fluxes expected from the 2016s IC chips with low pressure drops and uniform junction temperatures. The results showed that the fin shape plays an important role in the pressure drop rather than the heat dissipation which was mainly affected by the heat transfer area to fluid volume ratio γ . The best performance was obtained when flat-shaped fins were used. The 100-μm fin length made the system more desirable (good heat dissipation with a reasonable pressure drop). Moreover, the results generated with the fin height variation were interesting since two different effects were found. According to these observations, it was recommended to avoid the usage of sections without fins in this kind of micro heat sink systems. The MF-50 × 100 ×200-66 heat sink presents better performance compared to previous microcooling devices reported in literature. This proposed device can dissipate the heat flux expected by the 2016s IC chips with a pressure drop around 20 kPa and a pumping power as low as 0.04 W. Further, the system keeps the junction temperature uniform within 6 °C. Considering the temperature gradient as a parameter for comparison, this cooling system has an overall value of 1.63 °C/mm which was four times lower than the temperature gradient generated in rectangular microchannel heat sinks subject to similar sizes and operating conditions. ACKNOWLEDGMENT This paper was conducted in the Thermal Analysis, Microfluidics, and Fuel Cell Laboratory at Rochester Institute of Technology (RIT), Rochester, NY. The authors would like to thank D. Kudithipudi in the Department of Computer Engineering, RIT, for his support and insightful discussions. R EFERENCES [1] S. P. Gurrum, S. K. Suman, Y. K. Joshi, and A. G. Federov, “Thermal issues in next-generation integrated circuits,” IEEE Trans. Dev. Mater. Reliabil., vol. 4, no. 4, pp. 709–714, Dec. 2004. [2] D. B. Tuckerman and R. F. W. Pease, “High-performance heat sinking for VLSI,” IEEE Trans. Electron Dev., vol. 2, no. 5, pp. 126–129, May 1981. [3] M. T. Bohr, “Interconnecting scaling the real limiter to high performance ULSI,” in Proc. IEEE Int. Electron Dev. Meeting, Dec. 1995, pp. 241– 244. [4] P. Lall, M. G. Pecht, and E. B. Hakin, Influence of Temperature on Microelectronic and Reliability. Boca Raton, FL: CRC Press, 1997. [5] G. O. Workman, J. G. Fossum, S. Krishnan, and M. M. Pelella, “Physical modeling of temperature dependences of SOI CMOS devices and circuits including self-heating,” IEEE Trans. Electron Dev., vol. 45, no. 4, pp. 125–133, Jan. 1998. [6] S. G. Kandlikar and W. J. Grande, “Evaluation of single phase flow in microchannels for high flux chip cooling: Thermohydraulic performance enhancement and fabrication technology,” in Proc. ASME 2nd Int. Conf. Microchann. Minichann., Jun. 2004, pp. 5–16. RUBIO-JIMENEZ et al.: NUMERICAL ANALYSIS OF NOVEL MICRO PIN FIN HEAT SINK [7] M. Renksizbulut and H. Niazmand, “Laminar flow and heat transfer in the entrance region of trapezoidal channels with constant wall temperature,” J. Heat Trans., vol. 128, no. 1, pp. 63–74, 2006. [8] J. P. McHale and S. V. Garimella, “Heat transfer in trapezoidal microchannels of various aspect ratios,” Int. J. Heat Mass Trans., vol. 53, nos. 1–3, pp. 365–375, 2010. [9] R. W. Knight, D. J. Hall, J. S. Goodling, and R. C. Jaeger, “Heat sink optimization with application to microchannels,” IEEE Trans. Compon. Hybrids Manuf. Technol., vol. 15, no. 5, pp. 832–842, Oct. 1992. [10] C. Gillot, C. Schacffer, and A. Bricard, “Integrated micro heat sink for power multichip module,” IEEE Trans. Ind. Appl., vol. 36, no. 1, pp. 217–221, Jan.–Feb. 2000. [11] A. Husain and K. Y. Kim, “Optimization of a microchannel heat sink with temperature dependent fluid properties,” Appl. Therm. Eng., vol. 28, nos. 8–9, pp. 1101–1107, Jun. 2008. [12] D. V. Pence, “Reduced pumping power and wall temperature in microchannel heat sinks with fractal-like branching channel networks,” Microscale Thermophys. Eng., vol. 6, no. 4, pp. 319–330, 2002. [13] C. Biserni, L. A. O. Rocha, G. Stanescu, and E. Lorenzini, “Constructal H-shaped cavities according to Bejan’s theory,” Int. J. Heat Mass Trans., vol. 50, nos. 11–12, pp. 2132–2138, 2007. [14] X.-Q. Wang, P. Xu, A. S. Mujumdar, and C. Yap, “Flow and thermal characteristics of offset branching networks,” Int. J. Thermal Sci., vol. 49, no. 2, pp. 272–280, 2010. [15] Y. Peles, A. Koşar, C. Mishra, C. J. Kuo, and B. Schneider, “Forced convective heat transfer across a pin fin micro heat sink,” Int. J. Heat Mass Trans., vol. 48, no. 17, pp. 3615–3627, 2005. [16] E. G. Colgan, B. Furman, M. Gaynes, N. LaBianca, J. H. Magerlein, R. Polastre, R. Bezama, K. Marston, and R. Schmidt, “High performance and subambient silicon microchannel cooling,” J. Heat Mass Trans., vol. 129, no. 8, pp. 1046–1051, 2007. [17] E. G. Colgan, B. Furman, M. Gaynes, W. S. Graham, N. LaBianca, J. H. Magerlein, R. Polastre, M. B. Rothwell, R. J. Benzema, R. Choudhary, K. C. Marston, H. Toy, J. Walkil, J. A. Zitz, and R. R. Schmidt, “A practical implementation of silicon microchannel coolers for high power chips,” IEEE Compon. Packag. Technol., vol. 30, no. 2, pp. 218–225, Jun. 2007. [18] M. E. Steinke and S. G. Kandlikar, “Single phase liquid heat transfer with plain and enhanced microchannels,” in Proc. 4th ASME Int. Conf. Nanochann. Microchann. Minichann., Limerick, Ireland, 2006, pp. 943– 951. [19] R. Wälchli, T. Brunschwiler, B. Michel, and D. Poulikakos, “Combined local microchannel-scale CFD modeling and global chip scale network modeling for electronic cooling designs,” Int. J. Heat Mass Trans., vol. 53, pp. 1004–1014, Dec. 2009. [20] W. Escher, B. Michel, and D. Poulikakos, “A novel high performance, ultrathin, heat sink for electronics,” Int. J. Heat Fluid Flow, vol. 31, no. 4, pp. 586–598, 2010. [21] G. Hetsroni, A. Mosyak, Z. Segal, and G. Ziskind, “A uniform temperature heat sink for cooling of electronic devices,” Int. J. Heat Mass Trans., vol. 45, no. 16, pp. 3275–3286, 2002. [22] M. E. Steinke, S. G. Kandlikar, J. H. Magerlein, E. G. Colgan, and A. D. Raisanen, “Development of an experimental facility for investigating single-phase liquid flow in microchannels,” Heat Trans. Eng., vol. 27, no. 4, pp. 41–52, 2006. [23] K. C. Toh, X. Y. Chen, and J. C. Chai, “Numerical computation of fluid flow and heat transfer in microchannels,” Appl. Thermal Eng., vol. 25, no. 26, pp. 1472–1487, 2005. [24] Z. Li, X. Huai, Y. Tao, and H. Chen, “Effects of thermal property variations on the liquid flow and heat transfer in microchannel heat sinks,” Appl. Thermal Eng., vol. 27, nos. 17–18, pp. 2803–2814, 2007. [25] Fluent 6.2 User’s Guide, Fluent, Inc., Lebanon, NH, 2005. [26] X. Daguenet-Frick, J. Bonjour, and R. Revellin, “Constructal microchannel network for flow boiling in a disc-shaped body,” IEEE Trans. Compon. Packag. Technol., vol. 33, no. 1, pp. 115–126, Mar. 2010. 833 Carlos A. Rubio-Jimenez received the B.S. and M.S. degrees in mechanical engineering from the University of Guanajuato, Salamanca, Mexico, in 2007 and 2008, respectively, where he is currently pursuing the Ph.D. degree in mechanical engineering. He is involved in designing reliable micro heat sinks for cooling high-tech electronic devices. He has published a few conference and journal papers. He is a CONACYT Scholar and received a Fulbright Garcia-Robles Scholarship for developing the experimental work of his Ph.D. project with the Rochester Institute of Technology, Rochester, NY. Satish G. Kandlikar received the Ph.D. degree from the Indian Institute of Technology, Mumbai, India, in 1975. He is a Gleason Professor of mechanical engineering with the Rochester Institute of Technology (RIT), Rochester, NY. He was a Faculty Member at the Indian Institute of Technology, Bombay, before joining RIT in 1980. His research is concerned with heat transfer and fluid flow phenomena in microchannels and minichannels, and with advanced single-phase and two-phase heat exchangers incorporating smooth, rough, and enhanced microchannels. He has published more than 180 journal and conference papers. Dr. Kandlikar is a fellow of the American Society of Mechanical Engineers (ASME), an Associate Editor of a number of journals including the ASME Journal of Heat Transfer, and an Executive Editor of the Heat Exchanger Design Handbook published by Begell House. He received the RIT’s Eisenhart Outstanding Teaching Award in 1997 and the Trustees Outstanding Scholarship Award in 2006. Currently, he is involved in a project on fuel cell water management under freezing conditions, sponsored by the U.S. Department of Energy. Abel Hernandez-Guerrero received the B.S. degree from the University of Guanajuato, Salamanca, Mexico, and the Ph.D. and M.S. degrees from Oregon State University, Corvallis. He has published more than 220 scientific papers in journals and international energy conferences. Dr. Hernandez-Guerrero was the President of the Mexican Society of Mechanical Engineering from 2000 to 2002 and the Chair of the American Society of Mechanical Engineers (ASME) Student Sections Committee from 2006 to 2010 and the ASME Advanced Energy Systems Division from 2007 to 2008. He has been a member of the Mexican System of Researchers (National Top Honors Society) since 1992. He received the ASME Student Section Advisor Award in 2001 and the ASME Johnson & Johnson Medal in 2006. He was an Associate Editor of the ASME International Journal of Fuel Cell Technology, and a member of the Editorial Board of the International Journal of Exergy, the International Journal of Energy Research, and the International Journal of Thermodynamics. He is a fellow of ASME.