Aluminum-based one- and two-dimensional micro fin array structures: high-throughput fabrication and heat transfer testing

Aluminum based one and two dimensional micro fin array structures high throughput fabrication and heat transfer testing This content has been downloaded from IOPscience Please scroll down to see the f[.]

This content has been downloaded from IOPscience Please scroll down to see the full text.

Download details:

IP Address: 150.216.68.200

This content was downloaded on 24/02/2017 at 11:09

Please note that terms and conditions apply

Aluminum-based one- and two-dimensional micro fin array structures: high-throughput fabrication and heat transfer testing

View the table of contents for this issue, or go to the journal homepage for more

2017 J Micromech Microeng 27 025012

(http://iopscience.iop.org/0960-1317/27/2/025012)

Home Search Collections Journals About Contact us My IOPscience

You may also be interested in:

Cu microchannel devices: fabrication and analysis of flow and heat transfer characteristics

Fanghua Mei, W A Phillips, B Lu et al

Fabrication, assembly and heat transfer testing of low-profile copper-based MHEs

Bin Lu, Ke Chen, W J Meng et al

MEMS silicon-based micro-evaporator

M Mihailovic, C M Rops, J Hao et al

Effects of structural parameters on flow boiling performance of reentrant porous microchannels

Daxiang Deng, Yong Tang, Haoran Shao et al

Subcooled convective boiling in structured surface microchannels

Shou-Shing Hsieh and Chih-Yi Lin

Active control of flow and heat transfer in silicon microchannels

Guohua Liu, Jinliang Xu, Yongping Yang et al

Internal passivation of Al-based microchannel devices by electrochemical anodization

Paul J Hymel, D S Guan, Yang Mu et al

Pool boiling heat transfer on artificial micro-cavity surfaces in dielectric fluid FC-72

Chih Kuang Yu, Ding Chong Lu and Tsung Chieh Cheng

Investigation of change in surface morphology of heated surfaces upon pool boiling of magnetic

fluids under magnetic actuation

Mostafa Shojaeian, Melike-Mercan Yildizhan, Ömer Cokun et al

1 Introduction

Metal-based microscale fin array structures are of interest for microscale heat transfer applications While the classic paper

by Tuckerman and Pease showed the advantages of microscale structures with increased surface to volume ratio in single-phase convective heat transfer applications [1], silicon-based structures, such as those employed by Tuckerman and Pease

in their original paper, are not optimal from the perspective

of heat transfer performance and mechanical integrity Microscale compression molding was used successfully to replicate one- and two-dimensional (1D/2D), microscale features directly onto surfaces of high thermal conductivity and high ductility metals, such as aluminum (Al) and copper (Cu) [2–4] Subsequent bonding of Al and Cu caps to open 1D microchannel arrays led to the formation of Al- and Cu- based, enclosed, microchannel heat exchangers (MHEs) [5] Such Al and Cu MHEs can be made with low profile and high

Journal of Micromechanics and Microengineering

Aluminum-based one- and two-dimensional micro fin array structures: high-throughput fabrication and heat transfer testing

Philip A Primeaux1, Bin Zhang1, Xiaoman Zhang1, Jacob Miller1,

W J Meng1, Pratik KC2 and Arden L Moore2

1 Mechanical and Industrial Engineering Department, Louisiana State University, Baton Rouge,

LA 70803, USA

2 Mechanical Engineering Department, Louisiana Tech University, Ruston, LA 71272, USA E-mail: wmeng1@lsu.edu

Received 12 July 2016, revised 6 December 2016 Accepted for publication 14 December 2016 Published 9 January 2017

Abstract

Microscale fin array structures were replicated onto surfaces of aluminum 1100 and aluminum

6061 alloy (Al1100/Al6061) sheet metals through room-temperature instrumented roll molding Aluminum-based micro fin arrays were replicated at room temperature, and the fabrication process is one with high throughput and low cost One-dimensional (1D) micro fin arrays were made through one-pass rolling, while two-dimensional (2D) micro fin arrays were made by sequential 90° cross rolling with the same roller sleeve For roll molding of 1D micro fins, fin heights greater than 600 µm were achieved and were shown to be proportional to the

normal load force per feature width At a given normal load force, the fin height was further shown to scale inversely with the hardness of the sheet metal For sequential 90° cross rolling, morphologies of roll molded 2D micro fin arrays were examined, which provided clues to understand how plastic deformation occurred under cross rolling conditions

A series of pool boiling experiments on low profile Al micro fin array structures were performed within Novec 7100, a widely used commercial dielectric coolant Results for both horizontal and vertical surface orientations show that roll molded Al micro fin arrays can increase heat flux at fixed surface temperature as compared to un-patterned Al sheet

The present results further suggest that many factors beyond just increased surface area can influence heat transfer performance, including surface finish and the important multiphase transport mechanisms in and around the fin geometry These factors must also be considered when designing and optimizing micro fin array structures for heat transfer applications

Keywords: microscale roll molding, aluminum micro fin arrays, pool boiling heat transfer (Some figures may appear in colour only in the online journal)

P A Primeaux et al

Printed in the UK

025012

JMMIEZ

© 2017 IOP Publishing Ltd

27

J Micromech Microeng.

JMM

10.1088/1361-6439/aa53c9

Paper

2

2017

1361-6439

doi:10.1088/1361-6439/aa53c9

J Micromech Microeng 27 (2017) 025012 (9pp)

P A Primeaux et al

2

cooling capacity in the single-phase, forced flow, convective

heat transfer regime [6]

Compression molding of metals at the microscale involves

a mold insert containing a microscale surface pattern inverse

to the final desired pattern on the metal work piece When the

mold insert is placed onto the metal work piece with a

compres-sion load applied, patterns on the mold insert are transferred

onto the work piece surface through plastic deformation [7]

Because the key parameter controlling this compression

molding process is the contact pressure [7], the compression

force required to achieve replication of a given pattern scales

linearly with the area of the pattern This fact, combined with

the requirement on alignment accuracy between the mold

insert and the molded metal substrate, makes scaling up of

microscale compression molding to larger pattern footprints

more difficult

We have shown previously that roll molding can be used to

generate 1D microchannel arrays with large depths on sheet

metal surfaces in a parallel manner [8] In roll molding, metal

sheets are passed through a device analogous to that used for

flat rolling of sheet metals [9], in which one or both rollers can

contain microscale surface protrusions Plastic deformation is

induced in the metal sheet when it comes into contact with

the patterned roller(s), thereby creating microscale patterns on

one(both) surfaces of the metal sheet [8] Roll molding offers

much increased fabrication throughput as compared to

com-pression molding, and reduces the requirements on alignment

The potential of using roll molding to generate 1D/2D surface

patterns on metal substrates and the use of 1D/2D micro fin

array structures for heat transfer applications provide

motiv-ation for the present study

In addition to convective heat transfer in single-phase, forced

liquid flow situations, enhancing two-phase heat transfer

effi-ciency using metals with microscale 1D/2D surface patterns is

also of interest for pool boiling environments [10–16], which

have broad industrial processing applications including power

plants, refrigeration systems, and food production Two-phase

direct immersion cooling has become increasingly attractive

as a means of achieving efficient thermal management of high

density electronics systems and IT hardware [17–22] Besides

facilitating lower operating temperatures or higher power/per-formance capabilities, an ideal two-phase immersion cooling surface enhancer should also be low profile so as not to require large printed circuit board spacing and negatively affect com-puting density Thus the types of low cost, high throughput micro fin array structures produced in this work may hold promise for facilitating enhanced heat transfer and achieving widespread industrial adoption Here we present preliminary results on the heat transfer performance of select Al micro fin array structures within pool boiling environments that mimic those encountered in direct immersion cooling of electronics,

as a demonstration of potential and as a guide towards future work

2 Experimental setup and procedures

2.1 Roll molding of Al strips and characterization

Roll molding of Al sheet metals was carried out on a custom- designed and built machine, analogous to a sheet metal roll [9] Rotation of the lower steel roller of the machine, with

a nominal outer diameter (OD) of 108 mm (4.25 inches), was computer controlled with an angular speed range of

0–6 rpm and instrumented so the total input torque was mea-sured The upper steel roller, with the same nominal OD, was attached to a hydraulically driven assembly, actuated to move in the vertical direction, and instrumented to measure the normal loading force and the upper roller displacement [8] Both steel rollers could accommodate roller sleeves con-taining microscale protrusions on their external surfaces In the present experiments, shown schematically in figure 1

the upper roller sleeve was made of hardened 52 100 steel and contained an array of circumferential micro-protrusions made by wire electrical discharge machining The protrusion cross sections were trapezoidal in shape, with a sidewall taper

of ~7° [8] The total width of the roller sleeve was 25.4 mm (1 inch) The OD of the roller sleeve was 108.7 mm (4.28 inches), slightly larger than the steel roller OD Roll molding with such a roller sleeve produced 1D micro fin arrays with a similar sidewall taper [8]

Cylindrical roller

Sheet metal

Cut-away view of the roller sleeve with an array of circumferential microprotrusions

Figure 1. A schematic of the roll molding process for replicating straight, 1D micro fin arrays on sheet metals.

J Micromech Microeng 27 (2017) 025012

Commercial Al1100 (99% + Al) and Al6061 sheet metal

strips were used as substrates The initial thickness of the Al

strips was 6.35 mm (0.25 inch) The initial width of the Al

strips was 31.8 mm (1.25 inches) Al strips were used in the

as-received condition, and after annealing at 300 °C and 500 °C

for various durations Because the Al strip widths were larger

than the pattern width on the roller sleeve of 25.4 mm, micro

fin array patterns were imprinted onto strip surfaces with two

untouched rims on the outside

During roll molding, the lower steel roller without surface

pat-tern and the upper steel roller sleeve with a surface patpat-tern were

rotated at 0.25 rpm such that a constant rolling surface speed of

1.4 mm s−1 was achieved Al strips were placed in between the

bottom and top rollers Once rotation started, the normal load

was increased until a pre-set load level was reached The metal

strip was then rolled in steady state, with the normal loading

force continuously recorded Multiple experiments were carried

out by varying the normal loading force applied to the Al strips,

while keeping all other param eters unchanged No damage was

observed on the roller sleeve after multiple roll molding runs

Microhardness measurements on as-received and annealed

Al strips were conducted on a Future-Tech® FM-1E tester,

using a diamond Vickers indenter The height of roll molded

features was measured on a VanGuard optical microscope with

a calibrated focal depth dial by focusing on the microchannel

top and bottom, and confirmed with additional scanning

elec-tron microscopy (SEM) measurements For one specimen, at

least five independent depth measurements were carried out

at random locations on the specimen, from which the average

feature height and its standard deviation were calculated

Morphological examinations of roll molded Al specimens

were conducted through SEM on a FEI Quanta3D FEG

Daul-Beam focused ion beam (FIB) instrument

2.2 Pool boiling heat transfer performance testing

To determine how roll-molded Al alloy micro fin arrays

perform as heat sinks within a pool boiling environment,

a series of experiments were performed on un-patterned

Al as well as 1D- and 2D- patterned samples fabricated by roll molding An annotated illustration of the experimental setup is given in figure 2 The walls, bottom, and top of the cube-shaped test tank were stainless steel plates with inter ior surfaces possessing a #8 mirror finish to minimize bubble nucleation from surfaces other than the sample under study Three circular polycarbonate (Lexan) viewports were mounted over cutouts within the sidewalls with a gasket ring

in between These viewports were utilized for visualization during the experiment The lid located at the top of the tank

is mechanically clamped along with an interfacing edge gasket to minimize vapor escape A passive pressure valve fitted into the lid maintained the vessel at atmospheric con-ditions A pair of sealed feedthroughs allowed for cartridge heater power leads and Type K thermocouples (Omega) to pass through the setup’s lid and access the variable power source and data acquisition hardware One thermocouple was inserted within the heated stage in close proximity to the base

of the mounted sample The remaining thermocouple was used to measure the temper ature of the pool away from the sample stage Two coils of thin-walled copper tubing occu-pied interior space within the tank One coil resided within the pool and was used to maintain the temperature of the pool at the desired subcooled condition In subcooled pool boiling, the temperature of the pool away from the heat

source is below the saturation temper ature (Tsat, more com-monly referred to as the boiling point) of the medium at the operating pressure The second copper coil was located above the free surface of the pool and served as a condenser for the coolant vapor produced during boiling After condensing, the coolant would fall back into the pool and thus a constant fluid level was maintained throughout the experiment Each coil was connected to its own water-based closed-loop flow system with dedicated liquid-to-air heat exchangers, pumps, and reservoirs

The sample stage was mounted on a pair of adjustable mechanical supports made of low thermal conductivity Nylon The connection between the stage and the supports allowed the stage to rotate such that data could be obtained for arbitrary

Figure 2. Annotated illustrations of (a) the pool boiling experimental setup and (b) the sample stage Not to scale.

P A Primeaux et al

4

surface orientation angles The interior of the sample stage was

a 25 mm × 32 mm × 6 mm C101 oxygen-free Cu block with

four embedded cartridge heaters (Omega) connected in parallel

to a 2000 VA variable AC power transformer (Philmore) A

smaller 10 mm × 13 mm × 6 mm C101 block was situated atop

this main heater block with solder at the interface One of the

thermocouples was embedded within the center of the smaller

C101 block via a drilled cavity and held in place with thermally

conductive adhesive The sample of interest was mounted on

top  of the mesa using a thin layer of thermally conductive

graphite/polylactic acid composite All surfaces except the

mounted sample were surrounded by 6 mm thick low thermal

conductivity Teflon to provide high temperature-compatible

insulation from the surrounding pool All seams/crevices

around the sample were sealed with a high temperature RTV

silicone to prevent spurious heat transfer/bubble nucleation

At the maximum heater power (~200 W), the upper limit

of uncertainty on heat flux originates from heat loss through

the Teflon layers and via thin insulated metallic lead wires

This upper limit is <10% of total power generation within the

cartridge heaters Limitations on accuracy of the two

thermo-couples is ±2.1 K per the manufacturer The temperature at

the base surface of the sample (Tbase) was obtained by taking

the thermocouple-measured temperature and, via Fourier’s

law, projecting the temperature differences through the copper

and composite materials with known thermal conductivity

and thickness values To directly relate the performance of the

tested samples to direct immersion cooling applications, the

liquid medium for all experiments was 3M™ Novec™ 7100

which has a saturation temperature of Tsat = 61 °C at sea-level

atmospheric pressure For similar reasons, tests were

per-formed at horizontal (0°) and vertical (90°) stage orientations

and with the pool maintained at room temperature With the

pool being maintained at a temperature below the saturation

temperature of the working fluid, this experiment

repre-sents what is known as a ‘subcooled’ pool boiling condition

For Tsat = 61 °C, by holding the temperature of the pool at

Tpool = 21 °C the experiments are at a subcooling (defined as

ΔTsub = |Tpool–Tsat|) of 40 K relative to the saturation

temper-ature of the pool

3 Results and discussion

3.1 Formation of 1D/2D micro fin array structures by roll molding

Figure 3(a) shows an optical overview of a typical 1D micro fin array structure prototype produced on the surface of an annealed Al1100 sheet by room temperature roll molding The approximately square prototype was cut from an origi-nally longer roll molded Al strip, and was limited in width by that of the roller sleeve The actual duration to replicate this 1D micro fin array structure was less than 20 s, demonstrating the high throughput of the roll molding process Figure 3(b) shows a low-magnification cross-sectional SEM image of a portion of the same prototype Figure 3(b) shows clearly the micro fin array structures created by the roll molding process Each fin cross section  is trapezoidal in shape The fin base and top widths are ~700 µm and ~530 µm, respectively The

fin height is ~640 µm, and the fin-to-fin spacing is ~960 µm

The taper angle of the fin sidewalls is ~7.6°, and mimics that

of the micro protrusion sidewalls on the roller sleeve The total thickness of the prototype is ~900 µm, showing that the

rolling molding technique can be used to make low profile, high fin height, Al micro fin array structures

Figure 4 shows additional data documenting the process

of roll molding 1D micro fin array structures onto surfaces

of Al strips Figure 4(a) shows the feed-in end of the roll molding machine For the present set of experiments, two steel patterned roller sleeves, each 25.4 mm in total width, were placed symmetrically on the upper roller with respect to its center (figure 4(a)) Two Al strips, 31.8 mm in total width, were rolled respectively under the two roller sleeves (figure

4(a)), and reached steady-state rolling when the total normal load force reached the set value Because of the symmetrical arrangement of the two roller sleeves, the measured total normal loading force was assumed to be shared equally among them Figure 4(b) shows the average height of the micro fins formed on surfaces of annealed Al1100 strips as a result of the roll molding process versus the total normal loading force, or more appropriately, the load force normalized by the total fea-ture width of 50.8 mm As expected from the assumed equal

Figure 3. A typical 1D micro fin array structure fabricated by room temperature roll molding: (a) an optical overview, with markings on the ruler in mm; (b) a low magnification cross-sectional SEM image, with micro fin dimensions measured and marked.

J Micromech Microeng 27 (2017) 025012

load sharing between the two roller sleeves, the average fin

heights measured from the left and right Al strips after roll

molding are consistent, and appear to scale linearly with the

normal force per unit feature width, consistent with previous

results [8]

Another set of roll molding experiments were executed

at a fixed total normal load of 20000lbf or a normal force

per unit feature width of ~1800 N mm−1 Four different sets

of aluminum sheet metal strips were prepared A set of

as-received Al1100 strips was annealed at 500 °C for 12 h One

as-received Al6061 long strip was cut into three separate

strip sets One Al6061 strip set was used in the as-received

condition The two other Al6061 strip sets were respectively

annealed at 300 °C for 2 h and at 500 °C for 2 h The

hard-ness of the four sets of Al strips were measured before the

rolling experiments The measured hardness values are

repre-sented by the abscissa of figure 4(c), with the lowest hardness

value (~25 kgf mm−2) from the annealed Al1100 strip and the

highest hardness value (~110 kgf mm−2) from the as-received

Al6061 strip Figure 4(c) plots measured average fin height

versus the hardness of the Al strips The dotted line through

the data points represents a non-linear least squares fit to a

power law function with a power law exponent of −1, and

shows that, at a given normal load, the average fin height

scales inversely with the hardness of the Al strip The

consist-ency of the data points shown in figure 4(c) further suggests

that, to first order, the fin height achieved under the same

roller condition is controlled primarily by the hardness of the aluminum sheet metal and does not depend significantly on the actual alloy composition

Roll molding with roller sleeves containing an array of circumferential micro-protrusions was used to fabricate 2D micro fin array structures by sequential 90° cross rolling One

Al strip was passed under the roller sleeve under a set normal load, turned 90°, and passed through the same roller sleeve a second time under a different normal load force Figure 5(a) shows an optical image of a typical annealed Al1100 micro fin array prototype formed by sequentially cross rolling The foot print of the cross rolled prototype is ~25 mm × 25 mm, again limited by the width of the roller sleeve Figure 5(b) shows a low-magnification SEM image of the same prototype taken along the direction of the second rolling, and shows that the second rolling produces clean sidewalls and flat bottoms between different fin columns Figure 5(c) shows another low-magnification SEM image of the same prototype in a rotated view, and shows that the sequential roll molding process led

to the formation of 2D micro fins with consistent shapes and heights It should be noted that, depending on the normal loads, fin heights formed through the two rolling passes, i.e the distances from the top of fin structures to the flat bottoms formed due to the initial roll pass and the second roll pass, are not necessarily the same

Additional cross rolling experiments were conducted to better understand the process of plastic deformation in the

Figure 4. Room temperature roll molding of 1D micro fin array structures: (a) an optical image of the feed-in side of the roll molding machine; (b) average fin height versus normal load force; (c) average fin height versus the hardness of the Al strip at a fixed normal load The arrows in (a) point to the roller sleeves and the Al sheet metal strips.

Figure 5. Room temperature formation of 2D micro fin array structures by sequential 90 ° cross roll molding: (a) an optical image of one 2D micro fin array structure; (b) an SEM image of the 2D micro fin array in the direction of the second rolling; (c) an SEM image showing the morphology of sequentially cross rolled micro fins, with the arrows indicating the first and second rolling directions.

P A Primeaux et al

6

sequential 90° cross rolling geometry While the first rolling

occurs on a flat, non-patterned surface of an Al sheet metal, the

subsequent 90° cross rolling occurs on an Al surface with

previ-ously formed microscale features, namely, a straight 1D micro

fin array At three initial total normal load forces of 15000lbf,

20000lbf, and 23000lbf or load force per feature width values

of 1314 N mm−1, 1752 N mm−1, and 2014 N mm−1, 90° cross

rolls with various secondary normal loads were performed

Measurements of the difference in fin heights formed through

the two rolling passes were made through optical microscopy

and recorded as a function of the ratio of the normal load forces

in the secondary pass to the initial pass Figure 6(a) shows one

such data set, in which the difference in fin heights, i.e the

distance from the top of fin structures to the flat bottom formed

due to the second roll pass minus that for the initial pass, is

plotted versus the ratio of the normal load during the second

roll pass to that during the initial pass The entire data set shown

in figure 6(a) was obtained at the normal load of 20000lbf for

the initial roll pass Figure 6(b) shows an analogous data set

obtained at the normal load of 23000lbf for the initial roll pass

Despite the large scatter in the data points, especially for those

shown in figure 6(a), the data shown in figures 6(a) and (b)

clearly highlight the fact that the ratio of normal loads needed

to achieve zero difference in fin height in sequential 90° cross

rolling is over 90% This result would at first appear somewhat

counter intuitive as the second rolling pass occurs on a straight

1D micro fin array interrupted by empty micro channels, so it

may be argued that contact between Al and the protrusions on

the roller sleeve in the feed direction occurs at a fraction of that

for the initial roll pass on flat sheet metal

Additional SEM imaging was carried out and dimensional

measurements were made from test samples with initial

normal loads of 15000lbf and 20000lbf and various values

of the secondary to initial normal load ratio, between 60% and 100% When the first rolling occurs on flat, non-patterned sheet metal surfaces, material flow in the direction parallel with the feed is not high That is, the dominant plastic flow occurs in the plane perpendicular to the feed direction [8] In contrast, SEM images shown in figure 7 demonstrate that sig-nificant plastic flow during the 90° cross roll occurs in the feed direction This is because the flow constraint during the 90° cross roll no longer exists due to the existing channels formed after the initial rolling Figure 7(a) shows clearly that mat-erials from the 1D micro fin array have flowed into the open channels under compression loading during the secondary cross roll pass At a certain secondary normal load determined

by the geometry of the 1D micro fin array, e.g fin height, fin width, and fin-to-fin spacing, the feed-direction material flow before one fin will contact material flow behind the next

An example of this is shown clearly in figure 7(b) After this contact, additional material flow occurs perpendicular to the feed direction and begins to pile up on both sides of the pro-truding roller sleeve features, as shown in figure 7(c) Such additional material flow during the secondary roll pass occurs under conditions similar to those during the initial roll pass on

a non-patterned surface, as the circumferential protrusions on the roller sleeve are now making 100% contact with solid Al

In light of this qualitative description of the plastic deforma-tion process during 90° cross roll, the observation that a ratio

of secondary normal load to initial normal load close to 1 is needed to achieve consistent fin heights in the initial and sec-ondary roll passes appears reasonable

In addition, images in figure 7 clearly show that the plastic deformation in the feed direction due to the 90° cross roll is largely symmetric, i.e there is not a significant difference between deformations ‘in front’ or ‘behind’ the feed direction

Figure 6. Difference in fin height in 90 ° cross rolling: ratio of normal loads versus fin height difference at the initial normal load of (a) 20000lbf; (b) 23000lbf The lines drawn through the data points are guides for the eye.

J Micromech Microeng 27 (2017) 025012

Thus the present experiments indicate that, at the present ratio

of roll molded feature height to roller diameter of ⩽0.01

(1000 µm/100 mm), effects of the roller curvature appear to be

secondary Plastic deformation occurring due to roll molding

is largely similar to that occurring during a normal

compres-sion, albeit at a short contact length

3.2 Heat transfer testing on 1D/2D micro fin array structures

Data sets of heat flux versus surface excess temperature were

obtained for an un-patterned Al1100 sample as well as the 1D

and 2D roll molded micro fin array structures in both vertical

and horizontal surface orientations, with the results given in

figure 8 using the base area of 10 mm × 13 mm for the heat

flux calculation Here, heat flux refers to the rate at which heat

is being transported through the base of the sample per unit

area Surface excess temperature ΔT refers to the different in

temper ature between the sample and the saturation temperature

of the working fluid (ΔT = Tbase–Tsat) Since the experiments

were carried out under subcooled conditions, the obtained

results include heat transfer in the natural convection, nucleate

boiling, and film boiling regimes For low surface heat flux

values (<~5 W cm−2 in this work), no phase change occurs and the medium remains entirely liquid Since this is a pool boiling scenario, there is no forced fluid motion by pumps or other external movers However, the heated surfaces of the Al samples create local density differences near the surface that induce buoyancy-driven movement of the liquid from which natural convection, single-phase heat transfer occurs As heat flux is increased, nucleation sites become active and individual vapor bubbles begin to form and grow on the sample surface The combination of bubble growth and the local convection that occurs during bubble release can lead to large increases

in heat flux as more and more nucleation sites become active This nucleate boiling regime is generally considered the optimal operating condition for phase change heat transfer systems and, in this work, can be seen as the nearly vertical regions on the respective boiling curves shown in figure 8

Factors determining the degree to which this nucleate boiling regime continues as well as its effectiveness for cooling are complex, and depend on both the nucleation site density of the surface as well as the ability for individual vapor bubbles to freely move away from the surface As heat flux is increased further, eventually the rate of vapor production at the surface

Figure 7. Morphology of plastic deformation in the 90 ° cross rolling geometry: (a) an SEM image of a 2D micro fin array structure formed with a total initial normal load of 15000lbf and a total secondary normal load of 11250lbf; (b) an SEM image of a 2D micro fin array structure formed with a total initial normal load of 15000lbf and a total secondary normal load of 13125lbf; (c) an SEM image of a 2D micro fin array structure formed with a total initial normal load of 20000lbf and a total secondary normal load of 20000lbf.

Figure 8. Base area heat flux versus excess temperature for the three sample types tested in both horizontal and vertical orientations.

P A Primeaux et al

8

becomes large enough that individual bubbles begin to merge

and a semi-contiguous vapor film forms at the surface This

so-called film boiling regime results in large increases in

sur-face temperature for small increases in heat flux and is thus

undesirable as an operating condition

In the present experiments, the straight, 1D micro fin array

sample showed measurably better heat transfer performance

than the un-patterned Al in both orientations, but especially

in the vertical alignment due to its apparent ability to

sus-tain nucleate boiling behavior to higher heat flux levels The

2D micro fin array sample, while possessing ~40% greater

surface area, demonstrated markedly lower heat transfer

per-formance compared to the 1D sample If surface/contact area

alone were the sole deciding factor in pool boiling

perfor-mance of modified surfaces, the 2D micro fin array would be

expected to have superior performance compared to the 1D

micro fin array and unpatterned cases However, the work of

Cooke et al [12, 13] has demonstrated a complex relationship

between bubble nucleation within microchannels,

migra-tion to fin top surfaces, and subsequent growth along with

microchannel-facilitated liquid replenishment While these

preceding studies were conducted in water with higher surface

tension than the Novec™ 7100 used here, they do demonstrate

the complex multiphase dynamics that must be considered in

two-phase heat transfer from engineered surfaces In

addi-tion, surface roughness and the availability of nucleation sites

have been shown to be important factors in nucleate boiling

heat transfer [23] Thus, structures generated by microscale

molding techniques which may affect surface roughness and

nucleation site density during the fabrication process need to

have their heat transfer performance evaluated relative to that

of a planar surface of comparable processing and surface

con-ditions Further, the ability of the modified surface to facilitate

rewetting of dry-out regions should be considered This fact

may be directly related to the relatively low performance of

the 2D micro fin array given that the specific fin arrangement/

shape used in the present experiments may actually obstruct

fluid flow to dry-out regions Hence, a parametric examination

of fin array geometry to determine the optimum combination

of surface area and beneficial fluid rewetting behavior may be

required in future work in order to realize the full potential of

2D micro arrays operating under the conditions tested here

Data shown in figure 8 show the potential of using metal

sur-faces with microscale patterns for enhancing two-phase heat

transfer performance in the pool boiling scenario At the same

time, these data suggest that future work should include

com-paring heat transfer enhancement offered by increased surface

area of micro fin array structures to that of un-patterned

surfaces of comparable surface conditions Further

consid-erations should also be given to microscale fluid dynamics to

facilitate efficient rewetting at fin tips

4 Summary

We have shown that room temperature roll molding offers one

avenue for fabricating Al-based, 1D/2D micro fin array

struc-tures In particular, the sequential 90° cross rolling technique

is a convenient way for producing 2D micro fin arrays with good height and shape consistency Preliminary heat transfer testing shows that Al micro fin array prototypes produced via room temperature roll molding can indeed produce superior heat transfer characteristics in pool boiling environments, while additional considerations related to wettability, multi-phase transport dynamics, and surface finish/nucleation site density should be included in optimizing microscale engi-neered surfaces for thermal management applications The present results show the potential of the roll molding tech-nique for fabricating metal-based micro fin array structures with low profile and large footprint at high throughput and low cost, and motivate further exploration of the use of such low-cost, low-profile, micro fin array structures for heat transfer applications

Acknowledgments

The authors gratefully acknowledge partial project support from NSF and Louisiana State Board of Regents through pro-gram OIA-1541079 and contract LEQSF(2013-16)-RD-B-01

References

[1] Tuckerman D B and Pease R F W 1981 High performance

heat sinking for VLSI IEEE Electron Device Lett 2  126 – 9

[2] Cao D M and Meng W J 2004 Microscale compression molding of Al with surface engineered LiGA inserts

Microsyst Technol.10  662 – 70

[3] Cao D M, Jiang J, Meng W J, Jiang J C and Wang W 2007 Fabrication of high-aspect-ratio microscale Ta mold inserts

with micro-electrical-discharge-machining Microsyst

Technol.13  503 – 10

[4] Jiang J, Mei F and Meng W J 2008 Fabrication of metal-based high-aspect-ratio microscale structures by compression

molding J Vac Sci Technol A 26  745 – 51

[5] Mei F, Parida P R, Jiang J, Meng W J and Ekkad S V

2008 Fabrication, assembly, and testing of Cu- and

Al- based microchannel heat exchangers IEEE/ASME

J. Microelectromech Syst.17  869 – 81

[6] Lu B, Chen K, Meng W J and Mei F 2010 Fabrication, assembly, and heat transfer testing of low-profile

copper-based microchannel heat exchangers J Micromech

Microeng.20  115002

[7] Jiang J, Meng W J, Sinclair G B and Lara-Curzio E 2007 Further experiments and modeling for microscale compression molding of metals at elevated temperatures

J. Mater Res.22  1839 – 48

[8] Lu B and Meng W J 2014 Roll molding of microchannel arrays

on Al and Cu sheet metals: a method for high-throughput

manufacturing ASME J Micro Nano Manuf 2  011007

[9] Groover M P 2010 Fundamentals of Modern Manufacturing:

Materials Processes, and Systems 4th edn (Hoboken, NJ: Wiley)

[10] Guglielmini G, Misale M and Schenone C 2002 Boiling of

saturated FC-72 on square pin fin arrays Int J Therm Sci

41  599 – 608

[11] Yu C K and Lu D C 2007 Pool boiling heat transfer on

horizontal rectangular fin array in saturated FC-72 Int J

Heat Mass Transfer50  3624 – 37

[12] Cooke D and Kandlikar S G 2011 Pool boiling heat transfer and bubble dynamics over plain and enhanced

microchannels J Heat Transfer 133  052902

J Micromech Microeng 27 (2017) 025012

– [14] Kalani A and Kandlikar S G 2012 Pool boiling of fc-87 over

microchannel surfaces at atmospheric pressure Proc of the

ASME 2012 10th Int Mechanical Engineering Congress

and Exposition (Houston, Texas, USA , 9 –15 November

2012)

[15] Kalani A and Kandlikar S G 2012 Pool boiling heat transfer

over microchannel surfaces with ethanol at atmospheric

pressure Proc of the ASME 2012 10th Int Conf on

Nanochannels, Microchannels, and Minichannels

(Rio Grande, Puerto Rico , 8 –12 July 2012)

[16] Lee W, Son G and Yoon H Y 2012 Numerical study of bubble

growth and boiling heat transfer on a microfinned surface

Int Commun Heat Mass Transfer39  52 – 7

[17] Anderson T M and Mudawar I 1989 Microelectronic cooling

by enhanced pool boiling of a dielectric fluorocarbon liquid

J Heat Transfer111  753 – 9

– [19] Tuma P E 2010 The merits of open bath immersion cooling

of datacom equipment 26th IEEE SEMI-THERM Symp

pp 123 –31 [20] Campbell L and Tuma P E 2012 Numerical prediction of the junction-to-fluid thermal resistance of a 2-phase

immersion-cooled IBM dual core POWER6 processor 28th IEEE

SEMI-THERM Symp. pp 36 –44 [21] Nakayama W 2012 Heat in computers: applied heat transfer in

information technology J Heat Transfer 136  013001

[22] Eiland R, Fernandes J, Vallejo M, Agonafer D and Mulay V

2014 Flow rate and inlet temperature considerations for

direct immersion of a single server in mineral oil 14th IEEE

ITHERM Conf. pp 706 –14 [23] Jones B J, McHale J P and Garimella S G 2009 The influence

of surface roughness on nucleate pool boiling heat transfer

J Heat Transfer131  121009