Enhanced flow boiling in microchannels through integrating multiple micro-nozzles and reentry microcavities Wenming Li, Xiaopeng Qu, Tamanna Alam, Fanghao Yang, Wei Chang, Jamil Khan, and Chen Li
Citation: Appl. Phys. Lett. 110, 014104 (2017); doi: 10.1063/1.4973495 View online: http://dx.doi.org/10.1063/1.4973495 View Table of Contents: http://aip.scitation.org/toc/apl/110/1 Published by the American Institute of Physics
APPLIED PHYSICS LETTERS 110, 014104 (2017)
Enhanced flow boiling in microchannels through integrating multiple micro-nozzles and reentry microcavities Wenming Li,1 Xiaopeng Qu,1 Tamanna Alam,1 Fanghao Yang,2 Wei Chang,1 Jamil Khan,1 and Chen Li1,a) 1
Department of Mechanical Engineering, University of South Carolina, Columbia, South Carolina 29208, USA Princeton Plasma Physics Laboratory, Princeton, New Jersey 08540, USA
2
(Received 7 September 2016; accepted 19 December 2016; published online 4 January 2017) In a microchannel system, a higher mass velocity can lead to enhanced flow boiling performances, but at a cost of two-phase pressure drop. It is highly desirable to achieve a high heat transfer rate and critical heat flux (CHF) exceeding 1 kW/cm2 without elevating the pressure drop, particularly, at a reduced mass velocity. In this study, we developed a microchannel configuration that enables more efficient utilization of the coolant through integrating multiple microscale nozzles connected to auxiliary channels as well as microscale reentry cavities on sidewalls of main microchannels. We achieved a CHF of 1016 W/cm2 with a 50% less mass velocity, i.e., 680 kg/m2s, compared to the two-nozzle configuration developed in our previous studies. Two primary enhancement mechanisms are: (a) the enhanced global liquid supply by four evenly distributed micronozzles, particularly near the outlet region and (b) the effective management of local dryout by the capillary flow-induced sustainable thin liquid film resulting from an array of microscale cavities. A significantly improved heat transfer coefficient of 131 kW/m2 K at a mass velocity of 680 kg/m2s is attributed to the enhanced nucleate boiling, the established capillary/ thin film evaporation, and the induced advection from the present microchannel configuration. All these significant enhancements have been achieved with a 55% lower two-phase pressure drop. Published by AIP Publishing. [http://dx.doi.org/10.1063/1.4973495] In flow boiling systems, indicated by a sharp increase of wall temperature, critical heat flux (CHF) conditions define the maximum working heat flux and could lead to severe and permanent damages to equipment and devices if not properly managed. A higher CHF is highly desirable to increase the safety margin of flow boiling systems such as two-phase heat exchangers and thermal-hydraulic systems in nuclear power plants. Additionally, high power electronics cooling becomes even more challenging due to the shrinking device size with an increased power density. Two-phase cooling using closed microchannels has been considered as one of the most promising solutions owing to its advantages in packaging and potentially high transport performance compared to singlephase microchannel cooling techniques.1–4 CHF conditions can be triggered in these two-phase microchannel cooling systems by the formation of a stable vapor film on heating surfaces, resulting in a tremendous reduction of heat transfer rate and a spike of surface temperature.5 In closed microchannel systems, CHF crisis can be triggered by explosive boiling, two-phase flow instabilities, and local dry-out without a timely liquid supply on heating surfaces. Explosive boiling can lead to the premature CHF because the rapidly growing bubbles prevent the heating surface from rewetting.6,7 On the other hand, two-phase flow instabilities can also result in severe reverse flows, leading to liquid supply crisis. Compared to regular-sized systems, it is more challenging to regulate twophase flows to enhance flow boiling in closed microchannels due to the well-known bubble confinement effect.8 In the last few decades, numerous techniques have been developed to a)
Author to whom correspondence should be addressed. Electronic mail:
[email protected]. Tel: 803-777-7155
0003-6951/2017/110(1)/014104/5/$30.00
enhance CHF through promoting surface wettability as well as manipulating two-phase flows. These include, but not limited to, regulating bubble slugs,8,9 suppressing flow instability,10,11 modifying surface properties,12–17 and promoting liquid rewetting.18–21 Inlet restrictors were considered as one of the most effective ways to improve CHF,8 however, at a cost of pressure drop. The state of the art CHF enhancement techniques have been reviewed by Li et al.22 A CHF higher than 30 kW/cm2 has been achieved in microtube flow boiling at a mass velocity of 38 111 kg/m2 s with an exit vapor quality less than 0.1.23 Most recently, a tapered microchannel configuration has been developed and demonstrated a CHF of 1070 W/cm2 at an inlet mass velocity of 2624 kg/m2 s with a low pressure drop of 30 kPa owing to the increased liquid inertia force and vapor removal capability.11 It appears that a high mass velocity can result in a relatively higher CHF in a designated microchannel configuration, but could lead to an inefficient unitization of the coolant and hence, a lower exit vapor quality. Another challenge is to maintain wall temperature in the proper working range under high heat flux conditions, which demands a high heat transfer coefficient (HTC) near CHF conditions in a closed microchannel system, equally important, without significantly elevating the pressure drop. Principally, this could be achieved through reducing mass velocity and improving the contribution of phase change heat transfer, i.e., achieving a higher exit vapor quality. In the taper gap microchannel, HTC based on the heating area increases with working heat flux and impressively reaches a maximum HTC of 295 kW/m2 K at a wall superheat of 35 K.11 Different from the tapered microchannels, flow boiling HTC in a closed microchannel system was observed to be lower than that of the tapered microchannel and decrease
110, 014104-1
Published by AIP Publishing.
014104-2
Li et al.
with increasing working heat flux.1,3,8,24–28 In plain-wall microchannels with inlet restrictors, HTC based on the heating area is 75 kW/m2 K at a mass flux of 389 kg/m2 s under a CHF of 614 W/cm2.8 To enhance HTC, reentrant cavities were developed to enhance HTC to 98 kW/m2 K at a CHF of 643 W/cm2 and a mass flux of 303 kg/m2 s.10 However, a further enhancement of HTC was hindered by the persistent vapor slug. Recently, we have developed a two-nozzle microchannel configuration that can create mixing resulting from the microbubble-excited high frequency two-phase oscillations by passively cracking confined bubbles,29,30 achieving a HTC of 105 kW/m2 K at a CHF of 1020 W/ cm2 and a mass flux of 1350 kg/m2 s. However, the confined bubble cracking mechanism is only effective in the downstream of the two-nozzle microchannel configuration. In this study, we proposed an improved microchannel configuration as schematically illustrated in Figure 1(a). Four evenly distributed micronozzles are designed to generate jetting flows to rapidly crack bubbles in the main channels aiming to improve the global liquid supply of the main channel. Microscale reentry cavities, which target at enhancing nucleate boiling, inducing highly deserved capillary flows, are also integrated on the main channel sidewalls. Simultaneously, mixing generated by high frequency twophase oscillations can be extended to the entire channel through the proposed multiple-nozzle configuration. As a result of the rapid collapse of confined bubbles, two-phase flow instabilities and two-phase pressure drop due to the bubble confinement can be well managed. With the improvement of liquid supply at both local and global levels, a significant delay of CHF crisis can be expected in the present microchannel configuration with a reduced two-phase pressure drop. HTC would be also substantially enhanced, considering the enhanced nucleate boiling and the thin film evaporation enabled by the cavity array, and the extended mixing range resulting from multiple-nozzles. A microdevice was fabricated on a 500 lm thick silicon wafer. The dimensions (length width depth) of the five
FIG. 1. Concept and structure of the microchannel with integrated multiple microscale nozzles and reentry cavities. (a) Improved global liquid supply to main channels through auxiliary channels via nozzles, the bubbles nucleate from nozzles and cavities, and local liquid spreading by microscale reentry cavity-induced capillary flows. (b) SEM (scanning electron microscope) image of a top view of the main channel integrated with embedded reentry cavities (cavity dimensions: diameter ¼ 30 lm, opening ¼ 6 lm, distance ¼ 100 lm). (c) SEM of a micronozzle from aerial view of 45 , which connects both auxiliary and main channels. (d) SEM of two reentry cavities inside the sidewall. (All scale bars are 40 lm).
Appl. Phys. Lett. 110, 014104 (2017)
main channels, auxiliary channels, and restrictors are 10 mm 200 lm 250 lm, 8 mm 60 lm 250 lm, and 400 lm 20 lm 250 lm, respectively. To achieve a more compact design, two neighboring main channels share an auxiliary channel. Four converging micronozzles with a 20 lm throat are evenly distributed along the channels at a pitch of 2 mm. Microscale inlet restrictors were fabricated to trap bubbles and guide the direction of bubble expansion in the main channel. An aluminum film resistor (10 mm 2 mm 1.2 lm) was deposited on the backside of the chip to provide uniform heat flux and to serve as a thermistor to measure the wall temperature. A 500 lm thick Pyrex glass wafer was bonded to the silicon substrate to seal the channels and form a visualization window. Deionized (DI) water at room temperature is used as the working fluid. The inlet temperature varies significantly with both the mass velocity and working heat flux as shown in Fig. S3(a) in the supplementary material. Further details of fabrication processes and experimental details have been discussed in our previous studies.29,30 An overview of the present design is given in Fig. S1 in the supplementary material, and it is compared to our previous configuration. Using the results in the two-nozzle configuration29,30 as a baseline, Figure 2 summarizes the major enhancements obtained in this study. Figure 2(a) shows that a peak CHF of 1016 W/cm2 is obtained at a moderate mass velocity of 680 kg/m2 s, meaning a 63% enhancement compared to the baseline at a mass velocity of 750 kg/m2 s.22 Considering the pumping power budget, the CHF enhancement is more pronounced, approximately 95% higher for a given pressure drop, as illustrated in Figure 2(b). Figure 2(c) shows that a sustained HTC of 131 kW/m2 K at a mass velocity of 680 kg/m2s is achieved, 30% higher than that at a mass flux of 1350 kg/m2 s on the baseline near CHF conditions.29,30 The CHF is also compared to the reported CHF values on DI-water, as shown in Table I and plotted in Fig. S2 in the supplementary material.
FIG. 2. Comparisons of flow boiling performances between the present microchannel configuration and the two-nozzle one in our previous studies. (a) A significant enhancement of CHF is obtained with higher mass velocity. (b) CHF enhancement becomes more significant with increasing pressure drop. (c) Average HTC by considering the base area is substantially enhanced compared to previous two-nozzle microchannels. An enhancement of 59% is achieved at a mass flux of 430 kg/m2 s, respectively. (d) Drastically reduced pressure drops compared to two-nozzle microchannels.
014104-3
Li et al.
Appl. Phys. Lett. 110, 014104 (2017)
TABLE I. Comparison of CHF values on DI-water.
CHF data from Qu and Mudawar32 Kuan and Kandlikar33 Kuo and Peles10 Two-nozzle configuration22 Present study
Substrate material
The length of heating area (mm)
Cross section of channel W H (lm)
Tin( C)
G (kg/m2s)
q00 CHF (W/cm2)
Copper Copper Silicon Silicon Silicon
44.8 63.5 10 10 10
Plain wall 215821 Plain wall 1054157 Reetrant cavities 200 253 Micronozzles 200250 Micronozzles þ reentry cavities 200250
30 25.4 22 25 25
86–368 46.4–171.9 83–303 200– 1350 250–680
107–216 206–545 161.2–643 226–1020 419.6–1016
Although copper microchannels show higher CHF than Si microchannels for a given mass velocity in their working range, the present microchannel configuration appears to outperform these microchannels in terms of peak CHF and most Si microchannels at the same mass velocity. At a mass velocity of 430 kg/m2 s, HTC in the present configuration is 59% higher than that from the baseline. Note that pressure drop either in single phase or two-phase flow regimes is substantially reduced in the present configuration compared to the baseline as illustrated in Figure 2(d). For example, the two-phase pressure drop can be reduced by approximately 55% at a mass velocity of 430 kg/m2 s and heat flux of 518 W/cm2. The exit vapor quality in the present configuration remains higher than that from the baseline with the mass velocity increasing (Fig. S3(b) in the supplementary material). Particularly, the exit vapor quality at the peak CHF value in the present configuration is 0.457, which is more than 2.5 folds higher than that in the two-nozzle one, indicating a significantly higher portion of phase change heat transfer contribution near CHF conditions. The measurement uncertainties of mass flux, pressure, voltage, current, temperature, and microfabrication resolution are 61%, 61.5%, 60.5%, 60.5%, 61 C, and 3 lm, respectively.22 Uncertainty propagations are calculated using methods developed by Kline and McClintock.31 Uncertainties of the present CHF, average HTC, and pressure drop have been estimated to be less than 610 W/cm2, 60.3 kW/m2 k, and 5 kPa, respectively. In our visualization study, three primary enhancement mechanisms have been identified in the present study. First, as illustrated in Figure 2, nucleate boiling is significantly enhanced under both low and high heat flux working conditions. At a heat flux of 121 W/cm2 and a mass velocity of 750 kg/m2 s, the whole bubble nucleation, growth, and collapse process from the designated reentry cavities and microscale nozzles were directly visualized and are illustrated in Figures 3(a)–3(d). As a result, explosive boiling can be well managed by the stable bubble growth-collapse process under low heat flux conditions, leading to a significant delay of premature CHF crisis. As heat flux further increases to 403 W/ cm2 at a mass velocity of 325 kg/m2 s, nucleate boiling can be still enhanced and favorably induces jetting flows at a frequency of 111 Hz owing to the fast bubble growth-collapse process (Figures 3(e)–3(h)). The enhanced nucleate boiling can relief the two-phase flow instabilities resulting from the explosive boiling and hence, contribute to enhanced heat transfer rate and delay CHF. The second critical enhancer is the four evenly distributed micronozzles that substantially improve the global
liquid supply in the main channel. Figure 4(a) shows that the coolant can be pumped into the main channels as high frequency jetting flows through micronozzles once the main channels are blocked by vapor slugs. However, as a comparison, Figure 4(b) shows that rewetting is limited in the twonozzle microchannel, especially near the middle of the main channel where there is a significant dryout developing in the upstream channel. Hence, more micronozzles can improve global liquid supply compared to the two-nozzle configuration. The main reason is that the bypass created by auxiliary channels and four-nozzle configuration enables liquid supply to the entire main channel, especially near outlet areas that are usually prohibited in traditional microchannel systems. As a result, global liquid supply along the channel has been achieved in the present configuration. However, with an improved global liquid distribution, the local dryout still cannot be fully prevented. Due to the vigorous evaporation under high heat flux conditions, it is extremely challenging to timely rewet the high temperature surfaces, especially near the outlet region. To address this issue, microscale cavities are integrated into the microchannels to induce capillary-assisted liquid spreading as illustrated in Figure 5. With this improved design, highly desirable thin liquid films can be formed and refreshed on the sidewalls with a high frequency, e.g., 67 Hz under a heat flux of 131 W/cm2 and a mass velocity of 250 kg/m2 s. Because of this effect, the periodic local dry spots can be
FIG. 3. Working mechanisms of micro-cavities and multiple-nozzles. (a)–(d) Enhanced nucleation at a heat flux of 121 W/cm2 and a mass velocity of 750 kg/m2 s (refer movie 1, supplementary material). (e)–(h) Periodic jetting flow at a frequency of 111 Hz from microcavities at a heat flux of 403 W/cm2 and a mass velocity of 325 kg/m2 s near CHF conditions in the outlet section (refer movie 2, supplementary material). (All scale bars are 100 lm).
014104-4
Li et al.
FIG. 4. Enhancement of global liquid supply and mixing by a combination of auxiliary channel and reentry cavities in the present study. (a) Liquid was supplied through jetting flows in auxiliary channels when main channels were blocked by vapor slugs at a heat flux of 375 W/cm2 and a mass velocity of 325 kg/m2 s (refer movie 4, supplementary material). (b) Limited mixing and insufficient liquid supply in two-nozzle microchannels30 at a heat flux of 399 W/cm2 and a mass velocity of 430 kg/m2 s (refer movie 5, supplementary material). (c) Strong mixing and global liquid supply in the present design at a heat flux of 791 W/cm2 at a mass velocity of 600 kg/m2 s (refer movie 6, supplementary material). (All scale bars are 100 lm).
timely rewetted, leading to an effective management of local dryout. Microcavity-induced capillary flows are believed to be the primary mechanism to form and sustain the thin liquid film under high heat flux working conditions as illustrated in Figure 5(c). Another interesting phenomenon is that the liquid residue inside microcavities can delay the dryout process as highlighted in Figure 5. Figure 5(e) shows that even the
FIG. 5. Enhanced local liquid spreading enabled by the microcavity-induced capillary flows (refer movie 3, supplementary material). (a) Three wet cavities are labeled. (b) and (c) Thin liquid film became thinner and one cavity dryout. Partial dry-out in the main channel. (d) Complete dry-out in the main channel. The #1 and #2 cavity dryout. (e) Jetting flow from the auxiliary channel and #3 cavity dryout. (f) The surface and cavity are rewetted by jetting flow in the main channel. The local liquid spreading is at a frequency of around 67 Hz under a heat flux of 131 W/cm2 and a mass velocity of 250 kg/m2 s. (Scale bar is 100 lm).
Appl. Phys. Lett. 110, 014104 (2017)
bottom wall is fully dryout, many cavities still remain wet. Besides its capabilities in enhancing nucleate boiling, generating, and sustaining thin liquid film, the group of cavities also serve as a liquid reservoir to delay CHF conditions. The gradually cavity drying out process can be also evidenced by the heat transfer curve at the mass flux of 680 kg/m2s, where the HTC approaching CHF conditions is not reduced sharply, instead it slightly drops to 109 kW/m2 K from a relatively stable value of 131 kW/m2 K as illustrated in Figure 2(c). These three primary enhancers can also explain the enhanced HTC (Fig. S4 in the supplementary material) since all three major heat transfer modes such as nucleate boiling, advection resulting from mixing, and thin film evaporation have been significantly enhanced. Specifically, as shown in Figure 2, it appears that all microscale cavities are active. We assume the average active nucleate site density NA in the present microchannel shall be larger than 4 103 sites/cm2, considering the additional activated nucleation sites on plain surfaces at high superheat. Compared to the experimental study,34 NA at a superheat of 30 K and mass velocity of 302 kg/m2 s was measured at 900 sites/cm2 in silicon microchannels. A model based on copper microchannels35 predicts that NA at a superheat of 30 K is 6.2 102 sites/cm2. This comparison confirms that the enhanced nucleate boiling is significant, particularly, dominant under low superheat conditions where the high frequency two-phase oscillations have not been established yet. To better understand the contributions of each enhancement mechanism, as illustrated in Fig. S5 in the supplementary material, an evolution of three microchannel configurations from #1 to #3 is used to quantitatively determine the impact of the aforementioned three enhancers. The overall HTCs in the two-nozzle configuration (#1)30 and four-nozzle configuration (#2)36 are compared in Figs. S6(a)–S6(c) in the supplementary material. The overall HTC in the configuration #2 is significantly enhanced from 64% to 131% corresponding mass velocities of 325 kg/m2 s and 150 kg/m2 s, respectively. Local HTC has been used to understand the heat transfer enhancement mechanisms and plotted in Figs. S6(d)–S6(f) in the supplementary material. It appears that prior to the full establishment of high-frequency two-phase oscillations, which is indicated by the ascending HTC section until reaching the peak value, the major enhancer is the nucleate boiling owing to more cavities, i.e., high NA . The enhancement percentage drops with the increase of mass velocity, i.e., from 63% at 150 kg/m2 s to 30% at 430 kg/m2 s in this study. At a high superheat or working heat flux, the enhancement is reduced to approximately 30% until approaching the peak local HTC values. We believe that both nucleate boiling and mixing play a role there, but hard to quantify individual contributions. With a further increase of heat flux, the local HTC enhancement is minor at a mass velocity of 150 kg/m2 s and even reduced at 430 kg/m2 s. To better understand the role of the micro-cavities in the enhanced boiling performance, we compared configurations #2 and #3. At a mass velocity of 150 kg/m2 s, it is surprising to observe a reduced overall HTC after integrating microcavities (Fig. S6(a) in the supplementary material). A visualization study shows that collapse of vapor slug doesn’t occur
014104-5
Li et al.
as frequently as we expected (Figs. S7–S9) in the supplementary material). However, an additional 28% enhancement at low superheat is still obtained owing to the enhanced nucleate boiling. At a medium mass velocity of 325 kg/m2 s, the highest enhancement of 78% as illustrated in Fig. S6(e) (in the supplementary material) has been achieved. An additional 30% enhancement in nucleate boiling regime is also realized. The 78% enhancement should be primarily caused by the enhanced mixing. Furthermore, the integration of micro-cavities enhances bubble generation and hence the mixing, as indicated by the additional 20% enhancement at a superheat of 27 K (Fig. S6(e) in the supplementary material). With the increase of mass velocity, enhancements of nucleate boiling and mixing are reduced to 20% and 28%, respectively, as illustrated in Fig. S6(f) in the supplementary material. The enhancement in the upstream is because of the induced mixing and enhanced nucleate boiling (Fig. S10 in the supplementary material). While, in the downstream as shown in Fig. S11 in the supplementary material, the liquid is observed to be separated from sidewalls at high working heat flux and two-phase oscillations only occur at a lower frequency. Short period thin films or droplet evaporation should be the major enhancement mechanism in this situation according to visualization study (Fig. S11 in the supplementary material). Compared to the two-nozzle configuration, pressure drop is further reduced owing to the enhanced pumping effect resulting from the enhanced bubble growth-collapse process as well as the additional effective flow area from the longer auxiliary channels and increased micro-nozzles.36 In conclusion, the present microchannel configuration through integrating multiple micronozzles and reentry cavities has been demonstrated to drastically enhance flow boiling at a reduced mass velocity. The continuous rewetting and sustainable global liquid supply have been substantially promoted by generating high frequency jetting flows through four evenly distributed micronozzles. Equally important, the capillary flows induced by the microcavities can form and sustain a thin liquid film on the sidewall to effectively manage local dryout. Highly efficient heat transfer modes including nucleate boiling, thin film evaporation, and convection have been substantially promoted as well, resulting in highly efficient flow boiling heat transfer. All these enhancements have been obtained passively without elevating pressure drop. The present microchannel configuration is promising in implementing two-phase cooling of high-power electronics. See supplementary material for Figs. S1–S11 and movie #(1–6). A comparison of the present configuration and twonozzle configuration is presented. The CHF of the present study is compared to the reported existing studies. A more comprehensive analysis on the enhancement mechanisms has been conducted based on a parametric study. The evolution of three microchannel configurations is presented. They are used to quantitatively determine the impact of the aforementioned three enhancers. This work was supported by the U.S. Department of Defense, Office of Naval Research under the Grant No.
Appl. Phys. Lett. 110, 014104 (2017)
N000141210724 (Program Officer Dr. Mark Spector). Devices were fabricated at Institute of Electronics and Nanotechnology (IEN) in Georgia Tech, which are supported by the National Science Foundation under the Grant No. ECS-0335765. SEM images were taken at USC Microscopy Center. 1
B. Agostini, J. R. Thome, M. Fabbri, and B. Michel, IEEE Trans. Compon. Packag. Technol. 31(3), 691–701 (2008). 2 M. A. Ebadian and C. X. Lin, J. Heat Transfer 133(11), 110801 (2011). 3 S. G. Kandlikar, J. Heat Transfer-Trans. ASME 134(3), 034001 (2012). 4 J. R. Thome, Heat Transfer Eng. 27(9), 4–19 (2006). 5 J. E. Galloway and I. Mudawar, Int. J. Heat Mass Transfer 36, 2527–2540 (1993). 6 G. Hetsroni, A. Mosyak, E. Pogrebnyak, and Z. Segal, Int. J. Multiphase Flow 31(4), 371–392 (2005). 7 L. V. Z. Y. Dou, N. Winograd, and B. J. Garrison, J. Phys. Chem. A 105, 2748–2755 (2001). 8 A. Kos¸ar, C.-J. Kuo, and Y. Peles, J. Heat Transfer 128(3), 251–260 (2006). 9 G. Wang, P. Cheng, and A. E. Bergles, Int. J. Heat Mass Transfer 51(9–10), 2267–2281 (2008). 10 C. J. Kuo and Y. Peles, Int. J. Heat Mass Transfer 50(23–24), 4513–4526 (2007). 11 A. Kalani and S. G. Kandlikar, Appl. Phys. Lett. 107(18), 181601 (2015). 12 N. Singh, V. Sathyamurthy, W. Peterson, J. Arendt, and D. Banerjee, Int. J. Heat Fluid Flow 31(2), 201–207 (2010). 13 A. K. M. M. Morshed, F. Yang, M. Y. Ali, J. A. Khan, and C. Li, Appl. Therm. Eng. 32, 68–75 (2012). 14 C. Li, Z. Wang, P. I. Wang, Y. Peles, N. Koratkar, and G. P. Peterson, Small 4(8), 1084–1088 (2008). 15 A. Sitar and I. Golobic, Int. J. Heat Mass Transfer 81, 198–206 (2015). 16 M. S. Sarwar, Y. H. Jeong, and S. H. Chang, Int. J. Heat Mass Transfer 50(17–18), 3649–3657 (2007). 17 S. M. You, J. H. Kim, and K. H. Kim, Appl. Phys. Lett. 83(16), 3374–3376 (2003). 18 H. O’Hanley, C. Coyle, J. Buongiorno, T. McKrell, L.-W. Hu, M. Rubner, and R. Cohen, Appl. Phys. Lett. 103(2), 024102 (2013). 19 B. S. Kim, H. Lee, S. Shin, G. Choi, and H. H. Cho, Appl. Phys. Lett. 105(19), 191601 (2014). 20 D. E. Kim, S. C. Park, D. I. Yu, M. H. Kim, and H. S. Ahn, Appl. Phys. Lett. 107(2), 023903 (2015). 21 K.-H. Chu, R. Enright, and E. N. Wang, Appl. Phys. Lett. 100(24), 241603 (2012). 22 W. Li, F. Yang, T. Alam, J. Khan, and C. Li, Int. J. Heat Mass Transfer 88, 368–378 (2015). 23 € A. Kaya, M. R. Ozdemir, and A. Kos¸ar, Int. J. Therm. Sci. 65, 70–78 (2013). 24 S. G. Kandlikar, Exp. Therm. Fluid Sci. 26(2–4), 389–407 (2002). 25 A. Kosar, C. J. Kuo, and Y. Peles, Int. J. Heat Mass Transfer 48(23–24), 4867–4886 (2005). 26 D. Li, G. S. Wu, W. Wang, Y. D. Wang, D. Liu, D. C. Zhang, Y. F. Chen, G. P. Peterson, and R. Yang, Nano Lett. 12(7), 3385 (2012). 27 F. Yang, X. Dai, Y. Peles, P. Cheng, J. Khan, and C. Li, Int. J. Heat Mass Transfer 68, 703–715 (2014). 28 F. Yang, X. Dai, Y. Peles, P. Cheng, and C. Li, Appl. Phys. Lett. 103(4), 043122 (2013). 29 F. Yang, X. Dai, and C. Li, Appl. Phys. Lett. 101(7), 073509 (2012). 30 F. Yang, X. Dai, C.-J. Kuo, Y. Peles, J. Khan, and C. Li, Int. J. Heat Mass Transfer 58(1–2), 402–412 (2013). 31 S. J. Kline and F. A. McClintock, Mech. Eng. 75(1), 3–8 (1953). 32 W. Qu and I. Mudawar, Int. J. Heat Mass Transfer 47(10–11), 2045–2059 (2004). 33 W. K. Kuan and S. G. Kandlikar, J. Heat Transfer 130(3), 034503 (2008). 34 C. J. Kuo, A. Kosar, Y. Peles, S. Virost, C. Mishra, and M. K. Jensen, J. Microelectromech. Syst. 15(6), 1514–1527 (2006). 35 T. Hibiki and M. Ishii, Int. J. Heat Mass Transfer 46(14), 2587–2601 (2003). 36 W. Li, F. Yang, T. Alam, and C. Li, in ASME 2016 5th International Conference on Micro/Nanoscale Heat and Mass Transfer (Singapore, 2016), Vol. 1.
