International Journal of Heat and Mass Transfer 133 (2019) 1219–1229

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier .com/locate / i jhmt

Realizing highly coordinated, rapid and sustainable nucleate boiling inmicrochannels on HFE-7100

https://doi.org/10.1016/j.ijheatmasstransfer.2018.12.1620017-9310/� 2019 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: li01@cec.sc.edu (C. Li).

Jiaxuan Ma, Wenming Li, Congcong Ren, Jamil A. Khan, Chen Li ⇑Department of Mechanical Engineering, University of South Carolina, Columbia, SC 29208, United States

a r t i c l e i n f o

Article history:Received 24 September 2018Received in revised form 22 December 2018Accepted 27 December 2018

Keywords:Flow boilingElectronics coolingMicrochannelsMicro-slotHFE-7100

a b s t r a c t

Flow boiling in microchannels using dielectric fluids is one of the most desirable cooling solutions forhigh power electronics. However, it is challenging to promote the flow boiling performances, particularlycritical heat flux (CHF), due to their unfavorable thermophysical properties. Flow boiling in parallel andisolated microchannels has been extensively studied. In this study, five parallel microchannels(W = 200 mm, H = 250 mm, L = 10 mm) are interconnected by 4 � 28 micro-slots (20 mm wide and250 mm deep) starting from the middle section to the channel outlet. Our visualization study shows thatthese micro-slots designed as artificial nucleation sites can enable high frequency nucleate boiling bydrastically reducing the bubble waiting time and remaining fully activated, simultaneously. More impor-tantly, such rapid switch on–off harmonically coordinated nucleate boiling in the neighboring channelscreates a highly desirable periodic rewetting mechanism to substantially delay CHF conditions andenhance heat transfer rate. Flow boiling in this innovative microchannel configuration has been system-atically characterized with mass flux ranging from 462 kg/m2∙s to 1617 kg/m2∙s. Compared to plain-wallmicrochannels with inlet restrictors (IRs), flow boiling heat transfer coefficient (HTC) is enhanced up to�172% at a mass flux of 462 kg/m2�s primarily owing to the enhanced latent heat transfer includingnucleate boiling and thin film evaporation. The peak value of effective HTC is ~60 kW/m2�K in the fullydeveloped boiling regime. Moreover, CHF is substantially enhanced by �76% at a mass flux of1155 kg/m2�s owing to the rapid and periodic rewetting enabled by these micro-slots. Such drasticenhancements have been achieved without compromising two-phase pressure drop.

� 2019 Elsevier Ltd. All rights reserved.

1. Introduction

Compared to deionized (DI) water, properties of dielectric fluidsare more stable and reliable for applications of electronics cooling[1–4], which make dielectric fluid flow boiling in microchannels amore favorable cooling solution for high power electronics. How-ever, considering their unfavorable thermophysical properties,enhancing flow boiling heat transfer coefficient (HTC) and criticalheat flux (CHF) in microchannels is a challenging task. For instance,the thermal conductivity of HFE-7100 is 0.069 W/m�K, approxi-mately �10 times lower than that of water (�0.6 W/m�K). Besides,the latent heat of vaporization of HFE-7100 is about 111.6 kJ/kg,nearly 20 times smaller than that of water. In addition, the surfacetension is 13.6 mN/m at room temperature, about 5 times lowerthan that of water, making it difficult to form and sustain long thinliquid films on heating surfaces. Thus, it is challenging to promoteefficient thin film evaporation and nucleate boiling.

Many efforts have been taken to enhance flow boiling HTC ondielectric fluids. For example, nanowires [3], reentrant cavities[4,5], porous graphite [6,7] and other porous surfaces [1] can effec-tively enhance nucleate boiling by increasing nucleation site den-sity. In addition, sustaining thin film through enhanced capillaryflows induced by micro/nano-structures can result in highly effi-cient thin film evaporation [5]. Other methods including divergingmicrochannels [8], inclining microchannels [9], and mixing gener-ation [5] have been developed to significantly promote HTC onHFE-7100.

However, it is more challenging to enhance CHF of dielectricfluid flow boiling in microchannels. CHF crisis in a closedmicrochannel system can be triggered by three main factors, suchas explosive boiling, two-phase flow instabilities, and local dry-out.The low thermal conductivity of dielectric fluids is more likely totrigger explosive boiling and then result in premature CHF condi-tions. Local dry-out in the form of a stable vapor film on heatingsurfaces is highly likely to occur on low surface tension dielectricfluids due to its weak capillarity. Moreover, severe two-phase flowinstabilities [10,11] can also lead to local dry-out by, particularly,

Nomenclature

A base area, m2

Ac cross section area, m2

Cp heat capacity at constant pressure, J/kg�Kf frequency, HzG mass flux, kg/m2�s�h overall heat transfer coefficient, W/m2�Khfg latent heat of vaporization, kJ/kgH channel height, mI electrical current, Ak thermal conductivity, W/m�KL length, mm parameter for fin efficiency_m mass flow rate, kg/sNa nucleation site density, sites/cm2

Dp pressure drop, kPap pressureP power, Wq00 heat flux, W/cm2

R electrical resistance, XK slope of linear function, X/Kt thickness between heater and microchannel base, mt time, sT temperature, �C�T overall temperature, �CV electrical voltage, VV volume of bubble, m3

W microchannel width, m

Greek symbolsg efficiencyq density, kg/m3

v vapor quality

Subscriptsa ambientb baseCHF critical heat fluxd departuree exiteff effectiveeva evaporatef fing growingHTC heat transfer coefficienti inletnuc nucleates surfacesat saturatedsp single-phasetp two-phasev vaporw waiting

1220 J. Ma et al. / International Journal of Heat and Mass Transfer 133 (2019) 1219–1229

more pronounced near the outlet section. Enhancing CHF on DI-water with numerous technologies, such as regulating bubble slugs[12,13], suppressing flow instability [14,15], modifying surfaceproperties [16–21], and promoting liquid rewetting [22–25] hasbeen studied for decades. Note that the research to enhance CHFon dielectric fluids at room temperature is few. For instance, theenhancement of HTC in microchannels with nanowires [3], reen-trant cavities [4] has been achieved, but not on CHF. One of themain reasons for the limited enhancement of CHF is lack of liquidsupply, despite that capillary flow is promoted by integrating reen-trant cavities [4] and nanowires [3]. As aforementioned, Lee et al.has reported a high CHF of �700 W/cm2 at a high mass flux of5550 kg/m2 s on pre-cooled HFE-7100 in microchannels with theinlet temperature of a �30 �C [26]. In general, it is extremely chal-lenging to enhance CHF on dielectric fluid flow boiling withoutprecooling inlet temperature.

In this study, an innovative microchannel configuration hasbeen developed to enhance flow boiling in terms of HTC and CHFon HFE-7100. In this design, five parallel microchannels are inter-connected by 28 micro-slots on each intermediate wall. Thesemicro-slots were designed as nucleation sites to enhance and sus-tain nucleate boiling. More importantly, nucleate boiling in neigh-boring microchannels can be harmonically coordinated for the firsttime, leading to significantly higher bubble growth rates andrewetting frequency and eventually, greatly enhanced HTC andCHF.

2. Design and experimental procedures

2.1. Design and microfabrication of interconnected microchannels

In plain-wall microchannels, it is challenging to enhance heattransfer performances of flow boiling due to a long waiting time

of bubble nucleation and bubble departure time. To promote flowboiling heat transfer performance of HFE-7100, an innovativemicrochannel configuration is designed, as shown in Fig. 1. Thetested microdevice consists of five parallel microchannels(W = 200 mm, H = 250 mm, L = 10 mm) interconnected by 28micro-slots (width: 20 mm, height: 250 mm) starting from the out-let with a pitch of 180 mm on each wall. More details about thedimensions of microdevice have been discussed in our previousstudy [28]. These micro-slots can serve as nucleation sites togreatly enhance nucleate boiling. Fig. 2 shows experimental setupand the test package module. The test setup consists of an opticalimaging system, a data acquisition unit, and an open coolant loop.The working fluid of HFE-7100 is pumped by pressurized nitrogen(N2). All test processes are executed at room temperature �19 �Cand 1 ATM. Details of experimental setup were reported in our pre-vious studies [5,27].

Fabrication of the microchannel device: Fig. 3 shows thedetailed flow chart to fabricate the microfluidic device. First, a1 ± 0.01 lm thick thermal oxide layer was grown on both side ofp-type h1 0 0i silicon wafer to provide electric insulation for themicro heaters and serve as a hard mask for deep reactive ion etch-ing (DRIE). A 1 ± 0.05 lm thick thin film micro-heater was fabri-cated through a lift-off process on the backside of the wafer.Silicon oxide serving as an etching mask was etched off using reac-tive ion etching (RIE). Then five parallel micro-channels wereetched by DRIE. The depth of the channel is 250 ± 3 lm. The DRIEprocess also creates deep vertical sidewalls with a root meansquare (RMS) roughness of �300 nm. A Pyrex glass wafer wasanodically bonded to the silicon substrate to seal the microfluidicdevice. The individual microchannel test chips (length30 ± 0.005 mm; width 10 ± 0.005 mm; thickness 1 ± 0.005 mm)were cut from the wafer by a dice saw. More detailed microfabri-cation was elucidated in our previous study [28].

Fig. 1. (a) Configuration of the interconnected microchannels. (b and c) Scanning electron microscopy (SEM) images of interconnected microchannels and dimensions.

Fig. 2. (a) An exploded 3D model of the test package module [27], and (b) experimental setup.

J. Ma et al. / International Journal of Heat and Mass Transfer 133 (2019) 1219–1229 1221

2.2. Data reduction

In experiments, the input power was calculated by multiplyingthe DC voltage (V) with current (I). Then the effective heat flux wascalculated after subtracting the heat loss, Qloss, (pre-calibratedbetween the ambient environment and the test device) from thetotal input power, P, as follow,

q00eff ¼

P � Qloss

Að1Þ

where A is the base heating area. Based on the pre-calibrated linearrelationship between temperature and electrical resistance, theaverage temperature of the micro heater (on the backside of thedevice) was calculated as,

Theater ¼ K R� Rað Þ þ Ta ð2Þwhere Ra is the resistance of the micro heater at ambient tempera-ture Ta and K is the slope of pre-calibration. The average tempera-ture on the bottom wall of microchannels was then derived as,

Twall ¼ Theater �q00eff t

ksð3Þ

where t and ks are the substrate thickness, thermal conductivity ofsilicon, respectively.

The fin efficiency, gf, of a finite fin was estimated from

Twall ¼ Theater �q00eff t

ksð4Þ

The equation of gf is used to characterize fin performance and tocalculate the average effective HTC, where the parameter m wascalculated as,

m ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2h LþWð Þ=ksWL

qð5Þ

Then, the effective two-phase HTC considering fin efficiency, htp,is evaluated by,

htp ¼ QlatentPWLþ 2HLgf

� �Twall � Tsat� �� � ð6Þ

where Tsat is the saturated temperate of working fluid. The latentheat contributed to boiling heat transfer was derived as,

Qlatent ¼ P � Qloss � Qsensible ð7Þwhere Qsensible is the sensible heat due to the liquid temperature asfollows.

Qsensible ¼ GAcCp Tsat � Tið Þ ð8Þ

Fig. 3. Microfabrication process of the tested microdevice.

Table 1Major thermo-physical properties of HFE-7100.

Name HFE-7100

Latent heat of vaporization (kJ/kg) 111.6Specific heat, 25 �C, 1 ATM (J/kg�K) 1183Liquid thermal conductivity (W/m�K) 0.069Boiling point, 1 ATM (�C) 61Liquid Density, 25 �C (kg/m3) 1520Kinematic viscosity (cSt) 0.37Surface tension, 25 �C (m N/m) 13.6Vapor pressure, 25 �C (kPa) 26.9

Fig. 4. (a) Effective HTC as a function of effective heat flux. (b) Overall HTC as afunction of effective heat flux.

1222 J. Ma et al. / International Journal of Heat and Mass Transfer 133 (2019) 1219–1229

Major physical properties of dielectric fluid HFE-7100 are givenin Table 1. Tsat is a function of working pressure (p) in the middle ofa microchannel.

The vapor quality was calculated as [28],

v ¼ Peff � Qsensible

_mhfgð9Þ

In addition, the overall heat transfer performance of the device

was estimated as �h ¼ PeffA �Twall�Tsatð Þ.

2.3. Uncertainty analysis

The measurement uncertainties of flow rate, pressure, voltage,current, temperature, and microfabrication resolution are ±0.1%,±1.5%, ±0.5%, ±0.5%, ±1 �C, and 3 mm, respectively [29]. And theuncertainties of bubble diameter and departure frequency are±6% and ±7%, respectively, leading to an uncertainty of estimatedheat flux contributions of ±20%. Uncertainty propagation is calcu-lated using methods developed by Kline and McClintock [30].The uncertainty of effective HTC has been estimated to be less than±2 kW/m2 K.

3. Results and discussion

3.1. Flow boiling curves

In this study, the flow boiling performance in terms of effectiveHTC considering all effective heat transfer areas is characterizedwith mass flux ranging from 462 kg/m2 s to 1617 kg/m2 s. Fig. 4(a) shows that the effective HTCs decrease with the increase of heatflux. Fig. 4(b) shows the overall HTCs based on the heater area as afunction of effective heat flux. The overall HTCs based on the heat-ing area share almost the same trend with the effective HTCs, but�2 times larger. An overall HTC of �120 kW/m2 K is achieved afterthe onset of nucleate boiling (ONB) at mass flux of 1155 kg/m2 s.Fig. 4 also indicates that there are two distinct regimes of HTCcurves. For mass flux ranging from 462 kg/m2 s to 924 kg/m2 s,the curves of HTC become flat after ONB. The sustainable nucleateboiling may contribute to the stable heat transfer rate. At mass fluxof 1155 kg/m2 s and 1617 kg/m2 s, the HTCs gradually decline withthe increase of heat flux. The high contribution of convection aswell as occurrence of heating surface dry-out near outlet sectionshould be two main reasons for the decrease of HTC, especiallynear CHF conditions.

3.2. New nucleate boiling phenomena

3.2.1. Bubble dynamicsA complete bubble dynamic cycle includes bubble growing time

and bubble waiting time as schematically shown in Fig. 5. The bub-

Fig. 5. Schematic of a bubble growth cycle including waiting and bubble growingperiods for (a) interconnected microchannel configuration and (b) plain-wallmicrochannels.

J. Ma et al. / International Journal of Heat and Mass Transfer 133 (2019) 1219–1229 1223

ble growing time is the duration from bubble nucleation to depar-ture or collapse on the heating surface. Additionally, the waitingtime is the time of the initiation of nucleus on a nucleation site

Fig. 6. Comparisons of bubble dynamics between the present design and plain-wall mifunction of heat flux, (b) bubble diameter as a function of time and (c) bubble waiting t

after previous bubble departure or collapse from the same site.The bubble departure frequency is defined by [31]:

f d ¼1

tw þ tgð10Þ

Bubble dynamics in terms of bubble departure diameter (Dd),bubble growth rate (dDdt ), bubble waiting and growth times are char-acterized and compared with those in plain-wall microchannels atdifferent mass fluxes as shown in Fig. 6. Compared to those on theplain-wall microchannels, the bubble departure diameter (Dd) andbubble growth rate (dDdt ) on the present microchannel configurationare �2.3 times larger and �3 times faster, respectively, as shown inFig. 6(a, b). As depicted in Fig. 6(c and d), bubble waiting time (t) onthe present microchannel configuration is less than half of that onthe plain-wall microchannels and the growing time of the presentchannel is�2.6 times longer than that in plain-wall microchannels.

On highly wetting fluids, it would be challenging to meet tworequirements for bubble nucleation on a plain wall: residual vaporand superheated fluid. The superior wetting liquids could com-pletely occupy the nucleation sites by removing residual vaporafter a bubble departure on the plain wall. Another factor, it wouldtake longer time to make the refilled liquid in the nucleation site tobe superheated. These two factors would lead to a significantlylonger bubble waiting time. In contrast, in the interconnectedmicrochannels with micro-slots, owing to the induced capillarypressure, a small amount of vapor could remain inside themicro-slots after the bubble departure. Moreover, liquid can be

crochannels at a mass flux of 693 kg/m2 s. (a) Bubble departure diameter (Dd) as aime as a function of heat flux. (d) Bubble growing time as a function of heat flux.

Fig. 9. A bubble growth process at a mass flux of 462 kg/m2 s and a heat flux of40 W/cm2. (a) Initial stage for interconnected channel, (b) bubble growing frommicro-slots, (c) bubble collapsing or departing from channel, and (d) bubblemerging together.

1224 J. Ma et al. / International Journal of Heat and Mass Transfer 133 (2019) 1219–1229

quickly sucked into the slots during the bubble departure processand heated up by walls after then, which keeps these micro-slotsactive all the time and greatly reduces the bubble waiting timeas illustrated in Fig. 6(c and d). A visualization study has been con-ducted to validate these two hypotheses and are discussed in thenext section.

Fig. 7 shows three sequential frames of liquid jetting from themicro-slots. These images are selected from image sequences at asample rate of 7000 frames per second (fps). The observed liquidjetting should be pumped out primarily by bubble expansioninside the slot and assisted by pressure imbalance between theneighboring channels owing to unstable non-uniform distributionsof two-phase flow [32]. More importantly, this jetting flows wellvalidate the existence of residual vapor and liquid inside the slots,which can significantly shorten the bubble growing time.

For a comparison, Fig. 8 depicts the waiting period of nucleateboiling in plain-wall microchannels at a mass flux of 693 kg/m2 sand a heat flux of 65 W/cm2. The bubble departure frequency inthe plain-wall microchannels is �355 Hz as shown in Fig. 11(a),which is lower than �452 Hz at similar working load of intercon-nected channels. Besides, the bubble departure diameter is�45 mm, around 2.3 times smaller and waiting time of the plainwall microchannels is �1.3 times longer than that in intercon-nected microchannels, as shown in Fig. 6(a) and (c). Fig. 8(a) and(d) clearly show that the bubble nucleation on the plain wall ran-domly occurs and the bubble departure frequency in the plain-wallmicrochannels is much lower than that of the present microchan-nels owing to the shorter bubble waiting time.

Fig. 7. Liquid jetting from these slots at 40 W/cm2 and a mass flux of 462 kg/m2 s.(a) Initial stage of non-fluid jetting status. (b, c) Fluid jetting observed in twosequential images.

Fig. 8. Waiting time in plain-wall microchannels at a mass flux of 693 kg/m2 s and aheat flux of 65 W/cm2. (a) Bubble departs from wall surface, (b, c) no bubblenucleation appears with a long waiting period, and (d) new bubble appears at samenucleation site.

Fig. 9 shows the bubble growth process in the interconnectedmicrochannels at a mass flux of 462 kg/m2 s and a heat flux of40W/cm2. Compared to the plain-wall microchannel, the bubblegrowth and departure or collapse in the interconnected channelhas been harmonically coordinated. The bubble growing time ismeasured at �220 ls as shown in Fig. 6(d).

Another parameter in characterizing bubble dynamics is activenucleation site density. The active nucleation site density was esti-mated by dividing the measured number of active nucleation sitesto the total area using images captured by a high speed camera[33]. As shown in Fig. 10, the active nucleation site density of inter-connected microchannels is a constant of 2240 sites/cm2 owing tothe 100% active nucleating sites, which is �1.86 times higher than

Fig. 10. (a) Designs of different configurations. (b) Active nucleation site density asa function of heat flux with different configurations.

J. Ma et al. / International Journal of Heat and Mass Transfer 133 (2019) 1219–1229 1225

the cavity structure at similar working condition of a mass flux693 kg/m2 s and �1.7 times higher than that of plain-wallmicrochannels.

The micro-slots in the interconnected channels were 100% acti-vated during the boiling but reentry-cavities in our previous designwere only 22.2% activated according to experiment data [5]. Highactive nucleation site density can facilitate efficient nucleate boil-ing heat transfer.

Fig. 11(a) shows the effects of mass flux and heat flux on bubbledeparture frequency, which increases with the increase of heat fluxand decreases with the increase of mass flux on the presentmicrochannel configuration. In addition, Fig. 12(c) demonstratesthat the switching frequency increases as the heat flux increasingat different mass fluxes. The trends of bubble switch frequencywere nearly overlapped in two different mass fluxes, meaning thatit should be primarily determined by nucleate boiling. Bubbledeparture frequency and departure diameter are highly affectedby the superheat. A reported study has suggested the bubbledeparture diameter is closely related to departure frequency inthe form of f Dd = constant [34]. In this study, a fit curve is fDd = 0.07. It is obtained from curve-fitting experimental results inFig. 11(b).

3.2.2. Coordinated nucleate boilingIn conventional microchannels, nucleate boiling in individual

channel is usually isolated. It would be highly desirable to harmon-ically coordinate these usually isolated boiling processes toenhance nucleate boiling heat transfer. On the present microchan-nel configuration, bubble nucleation switches between two sides ofeach wall at a high frequency. More importantly, the bubble nucle-ation and departure process on one side of the wall can enhance

Fig. 11. (a) Bubble departure diameter increases with the increase of heat flux atdifferent mass fluxes. (b) f Dd as a function of heat flux.

Fig. 12. A schematic of a bubble growth cycle in two neighboring channels (a)schematic of two neighboring channel walls 1–4. (b) Bubble growth diameter as afunction of time on two neighboring walls. (c) Nucleate boiling switching frequencyas a function of heat flux.

nucleate boiling on the other side of the wall, which greatlyenhances bubble nucleation. Thus, for the first time, nucleate boil-ing has been harmonically coordinated through these well desig-nated micro-slots as illustrated in Fig. 12(a).

The bubble growth and switching between two neighboringchannels has revealed in Fig. 12(a and b). In according to the visu-alization study, the bubble first nucleates on the upper surface ofchannel #1 and after the bubble collapse or departure, a bubblestartst to nucleate and grow on the top surface of channel #2. Thenthe bubble nucleates on the bottom surface of channels #1 and #2,sequentially. This bubble switching period, ts, is the durationbetween the bubble growth on the top surface of channel #1 and#2 to it occurs on the top surface of channel #1 and #2. The wholeswitching bubble nucleation process has been illustrated in Fig. 12(b). The bubble switching frequency is calculated by f = 1/ts.

As illustrated in Fig. 12(c), the switching frequency increasessmoothly with the heat flux increasing. As the mass flux increasedat different working conditions, the switch frequency decreased atthe same heat flux. For instant, the switch frequency at heat flux of60W/cm2 at 462 kg/m2 s is around 425 Hz, and for 693 kg/m2 s is410 Hz approximately.

1226 J. Ma et al. / International Journal of Heat and Mass Transfer 133 (2019) 1219–1229

3.3. Enhanced HTC

Compared to the plain-wall microchannel configuration, theeffective HTCs by considering all effective heat transfer areas inthis study are substantially enhanced, as shown in Fig. 13(a, b). Sig-nificant enhancements of �175% and �156% are achieved, respec-tively, at a mass flux of 462 kg/m2 s and 693 kg/m2 s. The enhancedmechanism of HTC is explored as following. On the plain wall sur-face, it is difficult to form large areas of thin liquid film due to thelow surface tension of HFE-7100. The explosive boiling would pos-sibly take place and blow off the liquid film from the heating sur-face, resulting in local dry-out spots. Meanwhile, the suppressednucleate boiling due to the highly wetting ability of HFE-7100would deteriorate heat transfer rates.

The interconnected microchannel configuration developed inthis study can overcome these aforementioned challenges. Nucle-ate boiling has been drastically enhanced by increasing activenucleation site density and bubble growth/departure rates. The28 micro-slots can serve as nucleation sites to greatly enhancenucleate boiling. More importantly, highly desirable periodicrewetting is enabled in each channel and coordinated betweenchannels by these micro-slots.

Comparisons between the current design and the plain-wallmicrochannels at different mass fluxes were conducted as shownin Fig. 13. Two types of effects including heat flux and vapor qualitywere compared to indicate the heat transfer mechanism as shown

Fig. 13. Significant enhancements of HTC are achieved on the present design compared tod) Effective HTC as a function of vapor quality.

in Fig. 13(c, d). The interconnected channels can enable signifi-cantly higher vapor quality (enhanced up to �55% at a mass fluxof 462 kg/m2�s) than the plain-wall microchannels do. This isbecause the new microchannel configuration increases highnucleation-site-density and keeps them activated, which greatlyenhances contributions of latent heat transfer as indicated by highexit vapor quality. Note that HTC remains stable as a function ofthe vapor quality.

During the coordinated nucleate boiling process, heat is primar-ily transferred by advection, nucleate boiling, and evaporation. Tobetter understand the enhanced mechanisms, these three heattransfer modes have been analyzed. The advection heat flux(q00

a = q00total – q00

nuc – q00eva) including mixing and convection

contributions, which results from the disruption of the thermalboundary layer during bubble growth, lift-off and/or collapse[35]. The periodic disruption of thermal boundary layer can reducethe thermal resistance [36–40]. The nucleate boiling heat flux(q00

nuc = qvhfgVbNaf) is defined as the amount of heat carried awayby these bubbles nucleated on the heating surface. On the presentmicrochannel configuration, the whole growth process of eachindividual bubble can be accurately measured by high spatiotem-poral images captured by high-speed camera (Phantom v7.3)owing to the large bubble departure size, extended growth time,and fully activated sites as illustrated in Fig. 6. The evaporationheat flux can be estimated by excluding nucleate boiling fromthe total latent heat flux q00

eva = v _mhfg/Aheater � q00nuc.

plain wall microchannels. (a, b) Effective HTC as a function of effective heat flux. (c,

J. Ma et al. / International Journal of Heat and Mass Transfer 133 (2019) 1219–1229 1227

Fig. 14 summarizes the ratio of three major heat transfer modecontributions to the total effective heat flux at different massfluxes between the two configurations. In the plain-wallmicrochannels, the advection heat transfer at a mass flux of693 kg/m2 s dominates heat transfer, where heat transferred bynucleate boiling and evaporation only accounts a small fraction.Fig. 14(a, b, c) shows the nucleate boiling contributes 20–40% of

Fig. 14. Comparison of proportional energy. (a) Bar graph of nucleation heat flux. (b) Nucheat flux. (d) Evaporation heat flux ratio as a function of effective heat flux. (e) Bar graph o

the total heat transfer rate in the overall heat transfer process;while evaporation accounts for 21–46% of total heat removal, morethan doubled compared to these in plain-wall microchannels. Inthose testing conditions, the contribution of the latent heat hasbeen substantially enhanced, which should be the primarily reasonbehind the drastically enhanced HTCs enabled by the presentmicrochannel configuration.

leation heat flux ratio as a function of effective heat flux. (c) Bar graph of evaporationf advection heat flux. (f) Advection heat flux ratio as a function of effective heat flux.

Fig. 16. Thin liquid film can be inhaled in micro-slots when bubble grows in thepresent study at a mass flux of 462 kg/m2 s and a heat flux of 44 W/cm2.

1228 J. Ma et al. / International Journal of Heat and Mass Transfer 133 (2019) 1219–1229

3.4. Enhanced CHF with reduced pressure drop

Fig. 15 compares the enhanced CHF for mass flux ranging from462 kg/m2 s to 1617 kg/m2 s in the present design with that of theplain-wall microchannels. A significant enhancement of �76% hasbeen achieved at a mass flux of 1155 kg/m2 s by enhancing liquidsupply and rewetting through the micro-slots. Moreover, a slightdecrease of �6% of two-phase pressure drop was found comparedto the plain-wall microchannels (Fig. 15b).

The enhancement of CHF indicates that micro-slots play a vitalrole in enhancing CHF. The rapid and periodic rewetting enabled bythe highly coordinated nucleate boiling in neighboring channelsare the main factor leading to the enhanced CHF.

Usually, premature CHF conditions can be triggered by explo-sive boiling, flow instabilities, and partial dry-out. For plain-wallmicrochannels with IRs, when heat flux increases, explosive boilingis likely to occur because the low thermal conductivity of HFE-7100. Then, thin liquid film would be expelled away from wallsand eventually, vapor layer forms on the heating surface becauseof its low surface tension. The integration of micro-slots has suc-cessfully prevented the occurrence of explosive boiling by.

Heating surface dry out is one of the main factors resulting inpremature CHF conditions. Enhanced liquid spreading by increas-ing capillary flows is essential to increase CHF. Normally, surfacerewetting is highly influenced by the properties of working fluid,surface conditions of heating surfaces and two-phase transportregimes. However, it is challenging to increase CHF on low surfacetension HFE-7100. In conventional plain-wall microchannels, it isdifficult to maintain liquid film on heating surface on HFE-7100because of its low surface tension. The potential solution for this

Fig. 15. Comparison of (a) CHF and (b) two-phase pressure drop between thepresent design and plain wall configuration.

issue is to enhance capillary pressure through surface modificationsuch as nanowires to compensate the low surface tension.Although local rewetting is enhanced in microchannel coveredwith nanowires, the global liquid supply greatly suffers from highflow resistance resulting from dense nanowires. As a result, theenhancement of CHF is not significant using nanowires. In thisstudy, individual microchannel interconnected by micro-slotswould greatly promote liquid rewetting. Fig. 16 shows that thinliquid film can be sustained between the micro-slots, effectivelydelaying local surface dry out at a mass flux of 462 kg/m2 s and aheat flux of 44 W/cm2 in the present study. The high frequencynucleation boiling in the present design can also improve liquidsupply to the neighboring channels.

4. Conclusions

In this study, flow boiling of HFE-7100 has been systematicallycharacterized on an innovative microchannel configuration. Rapidand sustainable nucleate boiling has been well coordinatedthrough designated micro-slots for the first time. Compared tothe plain-wall microchannels, HTC and CHF have been consider-ably enhanced without compromising two-phase pressure drop.The primary HTC enhancement mechanism is the higher contribu-tion of latent heat transfer on the present microchannel configura-tion. The highly coordinated and high frequency nucleate boilingprocess also greatly delays CHF conditions. The bubble dynamicshave been systematically characterized in terms of bubble growthrates, bubble departure diameter, and bubble departure frequencyand explain enhanced nucleate boiling.

Acknowledgments

This work was supported by the U.S. Department of Defense,Office of Naval Research under the Grants N000141210724 andN000141612307 (Program Officer Dr. Mark Spector). Devices were

J. Ma et al. / International Journal of Heat and Mass Transfer 133 (2019) 1219–1229 1229

fabricated at Institute of Electronics and Nanotechnology (IEN) inGeorgia Tech, which is supported by the National Science Founda-tion under the Grant ECS-0335765. SEM images were taken at USCMicroscopy Center.

References

[1] J.Y. Chang, S.M. You, Boiling heat transfer phenomena from microporous andporous surfaces in saturated FC-72, Int. J. Heat Mass Transf. 40 (1997) 4437–4447.

[2] T. Chen, S.V. Garimella, Measurements and high-speed visualizations of flowboiling of a dielectric fluid in a silicon microchannel heat sink, Int. J. Multiph.Flow 32 (8) (2006) 957–971.

[3] F. Yang et al., Flow boiling heat transfer of HFE-7000 in nanowire-coatedmicrochannels, Appl. Therm. Eng. 93 (2016) 260–268.

[4] C.J. Kuo, Y. Peles, Flow boiling of coolant (HFE-7000) inside structured andplain wall microchannels, J. Heat Transfer 131 (12) (2009) 121011.

[5] W. Li et al., Flow boiling of HFE-7100 in silicon microchannels integrated withmultiple micro-nozzles and reentry micro-cavities, Int. J. Heat Mass Transf.123 (2018) 354–366.

[6] M.S. El-Genk, J.L. Parker, Enhanced boiling of HFE-7100 dielectric liquid onporous graphite, Energy Convers. Manage. 46 (15–16) (2005) 2455–2481.

[7] J.L. Parker, M.S. El-Genk, Enhanced saturation and subcooled boiling of FC-72dielectric liquid, Int. J. Heat Mass Transf. 48 (18) (2005) 3736–3752.

[8] K.-S. Yang et al., Heat transfer and flow pattern characteristics for HFE-7100within microchannel heat sinks, Heat Transfer Eng. 32 (7–8) (2011) 697–704.

[9] C.-C. Wang et al., Effect of inclination on the convective boiling performance ofa microchannel heat sink using HFE-7100, Exp. Therm. Fluid Sci. 36 (2012)143–148.

[10] G. Hetsroni et al., Explosive boiling of water in parallel micro-channels, Int. J.Multiph. Flow 31 (4) (2005) 371–392.

[11] L.V.Z. Yusheng Dou, Nicholas Winograd, Barbara J. Garrison, Explosive boilingof water films adjacent to heated surfaces – a microscopic description, J. Phys.Chem. A 105 (2001) 2748–2755.

[12] A. Kos�ar, C.-J. Kuo, Y. Peles, Suppression of boiling flow oscillations in parallelmicrochannels by inlet restrictors, J. Heat Transfer 128 (3) (2006) 251–260.

[13] G. Wang, P. Cheng, A.E. Bergles, Effects of inlet/outlet configurations on flowboiling instability in parallel microchannels, Int. J. Heat Mass Transf. 51 (9–10)(2008) 2267–2281.

[14] C.J. Kuo, Y. Peles, Local measurement of flow boiling in structured surfacemicrochannels, Int. J. Heat Mass Transf. 50 (23–24) (2007) 4513–4526.

[15] A. Kalani, S.G. Kandlikar, Combining liquid inertia with pressure recovery frombubble expansion for enhanced flow boiling, Appl. Phys. Lett. 107 (18) (2015)181601.

[16] N. Singh et al., Flow boiling enhancement on a horizontal heater using carbonnanotube coatings, Int. J. Heat Fluid Flow 31 (2) (2010) 201–207.

[17] A.K.M.M. Morshed et al., Enhanced flow boiling in a microchannel withintegration of nanowires, Appl. Therm. Eng. 32 (2012) 68–75.

[18] C. Li et al., Nanostructured copper interfaces for enhanced boiling, Small 4 (8)(2008) 1084–1088.

[19] A. Sitar, I. Golobic, Heat transfer enhancement of self-rewetting aqueous n-butanol solutions boiling in microchannels, Int. J. Heat Mass Transf. 81 (2015)198–206.

[20] M.S. Sarwar, Y.H. Jeong, S.H. Chang, Subcooled flow boiling CHF enhancementwith porous surface coatings, Int. J. Heat Mass Transf. 50 (17–18) (2007) 3649–3657.

[21] S.M. You, J.H. Kim, K.H. Kim, Effect of nanoparticles on critical heat flux ofwater in pool boiling heat transfer, Appl. Phys. Lett. 83 (16) (2003) 3374–3376.

[22] H. O’Hanley et al., Separate effects of surface roughness, wettability, andporosity on the boiling critical heat flux, Appl. Phys. Lett. 103 (2) (2013)024102.

[23] B.S. Kim et al., Interfacial wicking dynamics and its impact on critical heat fluxof boiling heat transfer, Appl. Phys. Lett. 105 (19) (2014) 191601.

[24] D.E. Kim et al., Enhanced critical heat flux by capillary driven liquid flow on thewell-designed surface, Appl. Phys. Lett. 107 (2) (2015) 023903.

[25] K.-H. Chu, R. Enright, E.N. Wang, Structured surfaces for enhanced pool boilingheat transfer, Appl. Phys. Lett. 100 (24) (2012) 241603.

[26] J. Lee, I. Mudawar, Fluid flow and heat transfer characteristics of lowtemperature two-phase micro-channel heat sinks – Part 1: Experimentalmethods and flow visualization results, Int. J. Heat Mass Transf. 51 (17–18)(2008) 4315–4326.

[27] F. Yang et al., Enhanced flow boiling in microchannels by self-sustained highfrequency two-phase oscillations, Int. J. Heat Mass Transf. 58 (1–2) (2013)402–412.

[28] W. Li et al., Enhanced flow boiling in microchannels using auxiliary channelsand multiple micronozzles (I): characterizations of flow boiling heat transfer,Int. J. Heat Mass Transf. 116 (2018) 208–217.

[29] W.M. Li et al., Experimental and theoretical studies of critical heat flux of flowboiling in microchannels with microbubble-excited high-frequency two-phaseoscillations, Int. J. Heat Mass Transf. 88 (2015) 368–378.

[30] S.J. Kline, F.A. McClintock, Describing uncertainties in single-sampleexperiments, Mechanical Engineering 75 (1) (1953) 3–8.

[31] D. Euh et al., Characteristics of bubble departure frequency in a low-pressuresubcooled boiling flow, J. Nucl. Sci. Technol. 47 (7) (2010) 608–617.

[32] T. Zhang et al., Stability analysis and maldistribution control of two-phase flowin parallel evaporating channels, Int. J. Heat Mass Transf. 54 (25–26) (2011)5298–5305.

[33] N. Basu, G.R. Warrier, V.K. Dhir, Onset of nucleate boiling and active nucleationsite density during subcooled flow boiling, J. Heat Transfer 124 (4) (2002) 717–728.

[34] C.J. Kuo et al., Bubble dynamics during boiling in enhanced surfacemicrochannels, J. Microelectromech. Syst. 15 (6) (2006) 1514–1527.

[35] G.R. Warrier, V.K. Dhir, Heat transfer and wall heat flux partitioning duringsubcooled flow nucleate boiling—a review, J. Heat Transfer 128 (12) (2006)1243.

[36] P. Naphon, Effect of wavy plate geometry configurations on the temperatureand flow distributions, Int. Commun. Heat Mass Transfer 36 (9) (2009) 942–946.

[37] T.-C. Hung et al., Optimal design of geometric parameters of double-layeredmicrochannel heat sinks, Int. J. Heat Mass Transf. 55 (11–12) (2012) 3262–3272.

[38] J. Zhang, J. Kundu, R.M. Manglik, Effect of fin waviness and spacing on thelateral vortex structure and laminar heat transfer in wavy-plate-fin cores, Int.J. Heat Mass Transf. 47 (8–9) (2004) 1719–1730.

[39] Y. Sui, C.J. Teo, P.S. Lee, Direct numerical simulation of fluid flow and heattransfer in periodic wavy channels with rectangular cross-sections, Int. J. HeatMass Transf. 55 (1–3) (2012) 73–88.

[40] W. Li et al., A numerical parametric study of flow and heat transfer in circularand zig-zag, Square Microchannel Heat Sinks 50336 (2016), V002T15A015.