Sie sind auf Seite 1von 10

A Systematic Methodology for

Weilin Qu
Graduate Research Assistant
Student Member ASME
Optimal Design of Two-Phase
e-mail: quw@purdue.edu
Micro-Channel Heat Sinks
Issam Mudawar1
Professor and Director This study provides a comprehensive methodology for optimizing the design of a two-
Fellow ASME phase micro-channel heat sink. The heat sink parameters are grouped into geometrical
e-mail: mudawar@ecn.purdue.edu parameters, operating parameters, and thermal/fluid parameters. The objective of the
proposed methodology is to optimize micro-channel dimensions in pursuit of acceptable
Purdue University International Electronic values for the thermal/fluid parameters corresponding to a given heat flux, coolant, and
Cooling Alliance (PUIECA), overall dimensions of the heat generating device to which the heat sink is attached. The
Boiling and Two-Phase Flow Laboratory, proposed optimization methodology yields an acceptable design region encompassing all
School of Mechanical Engineering, possible micro-channel dimensions corresponding to a prescribed coolant flow rate or
Purdue University, pressure drop. The designer is left with the decision to select optimum channel dimen-
West Lafayette, IN 47907 sions that yield acceptable values of important thermal/fluid parameters that are easily
predicted by the optimization procedure. 关DOI: 10.1115/1.2056571兴

Introduction Application of this two-step process is clearly evident in recent


single-phase micro-channel heat sink development efforts. Early
Thermal engineers in the electronics industry are facing unprec-
studies in this area were mostly focused on experimental investi-
edented challenges of removing enormous amounts of heat from
gation of pressure drop and heat transfer characteristics 关1–7兴.
devices and packages, which are brought about by aggressive cir-
Heat sinks with different substrate materials and geometries were
cuit integration and miniaturization. In fact, breakthroughs in
fabricated and tested with different liquid coolants. Experimental
many high-performance electronic systems are becoming increas-
data were compared to predictions of models or correlations that
ingly dependent upon the ability to safely dissipate the waste heat,
were developed earlier for macro-channel flows. Most studies,
as the reliability and life span of electronics are both strongly
especially the more recent ones, revealed macro-channel predic-
influenced by temperature. Among only a handful of available
tive tools are highly effective at predicting the pressure drop and
high-performance cooling techniques, micro-channel heat sinks
heat transfer characteristics of single-phase micro-channel heat
have recently emerged as a highly effective thermal solution for
sinks 关7兴. Later, studies on optimal design of these heat sinks
next generation high-power-density electronics.
became more widely available 关8–11兴. Using macro-channel mod-
Micro-channel heat sinks utilize a series of small parallel chan-
els, heat sink geometry was optimized by minimizing the overall
nels as liquid flow passages and are especially suited for applica-
thermal resistance of the heat sink.
tions involving the dissipation of large amounts of heat from a
Research on two-phase micro-channel heat sinks is fairly re-
small area. Key merits of these heat sinks are low thermal resis-
cent. Most published studies in this area concern the fundamental
tance, small coolant inventory, and small heat sink mass and vol-
understanding of micro-channel flow boiling, including boiling
ume. Depending on whether the liquid coolant maintains liquid
incipience 关12,13兴, dominant flow patterns 关14–16兴, hydrody-
state or undergoes flow boiling inside the micro-channels, heat
namic instability 关17–20兴, pressure drop 关20–26兴, heat transfer
sinks can be classified as either single phase or two phase. Com-
关21,26–30兴, and critical heat flux 共CHF兲 关31–33兴. Experimental
pared to their single-phase counterparts, two-phase heat sinks pro-
investigations were performed and experimental data compared to
duce much higher convective heat transfer coefficients, reduce
the predictions of macro-channels models or correlations. These
coolant flow rate requirements, and provide greater streamwise
studies revealed channel size has a significant influence on flow
temperature uniformity.
boiling in micro-channels and, unlike single-phase heat sinks, pre-
Effective design of any novel thermal/fluid system is often a
dictions of macro-channel models and correlations showed appre-
two-step process with different objectives achieved in each step.
ciable deviation from experimental data. Several new predictive
In the first step, research efforts are concentrated on acquiring a
tools specifically tailored to micro-channel flow boiling were pro-
fundamental understanding of the transport characteristics of the
posed thereafter 关20,29,30,33兴. So far, no studies have been pub-
system. Where computational methods are in question, a thorough
lished on optimal design of a two-phase micro-channel heat sink.
experimental study is often undertaken, and experimental data are
The present study was motivated by the lack of a systematic
combined with mechanistic models to refine existing models or
methodology to guide two-phase micro-channel heat sink design.
develop new predictive tools for accurate description of the trans-
A new methodology is proposed wherein key heat sink parameters
port characteristics of the system. Once reliable predictive tools
and the best available predictive tools are first identified. Different
are established and verified, attention is shifted to optimal design
performance characteristics of the heat sink are then examined to
of the system, which is the main objective of the second step of
optimize micro-channel dimensions in pursuit of acceptable val-
the design process. This step involves selecting a system geometry
ues for the thermal/fluid parameters corresponding to a given heat
that yields optimum performance under prescribed input condi-
flux, coolant, and overall dimensions of the heat generating device
tions.
to which the heat sink is attached.

1
Author to whom correspondence should be addressed. Tel. 共765兲 494-5705; fax
共765兲 494-0539 System Parameters and Predictive Tools
Contributed by the Electronic and Photonic Packaging Division of ASME for
publication in the JOURNAL OF ELECTRONIC PACKAGING. Manuscript received February
Figure 1 illustrates the construction of a two-phase micro-
25, 2004; final manuscript received December 17, 2004. Review conducted by: Guo- channel heat sink. The heat sink is composed of a base substrate
Quan Lu. and a cover plate. The base substrate is usually made from silicon

Journal of Electronic Packaging Copyright © 2005 by ASME DECEMBER 2005, Vol. 127 / 381
there is always a practical minimum value for Hw that is set by
machining and/or structural limitations. Both Hc and Hw are ex-
cluded from present optimization methodology.
Aside from the planform dimensions 共W ⫻ L兲, the channel
height Hch, channel width Wch, and wall thickness Ww between
neighboring channels constitute the primary geometrical param-
eters for heat sink design. The ultimate goal here is to select
values for Hch, Wch, and Ww that yield the most desirable heat sink
performance. From Fig. 1, the number of channels in a heat sink
can be evaluated from
W
N= . 共1兲
Wch + Ww
Operating parameters represent conditions under which the heat
sink is expected to operate. They include substrate material, type
of coolant, inlet temperature Tin, outlet pressure Pout, total coolant
volume flow rate Qt, and heat flux q⬙ef f . q⬙ef f represents the heat
removal requirement of the heat sink and is based on the heat
sink’s planform area:
PW
⬙f=
qef . 共2兲
WL
In heat sink design, another parameter q⬙p is often used to describe
heat flux along the conducting walls of the micro-channels. Re-
ferring to Fig. 1, q⬙p can be related to q⬙ef f by
⬙ f 共Wch + WW兲
qef
q⬙p = . 共3兲
Wch + 2Hch
Like heat sink overall dimensions, operating parameters also serve
Fig. 1 Construction of typical two-phase micro-channel heat as input parameters in heat sink design, and their values are often
sink
specified beforehand.
Thermal/fluid parameters are dependent transport parameters
that determine the performance of a heat sink under given geo-
or a high-thermal-conductivity metal such as copper or aluminum. metrical and operating parameters. Key thermal/fluid parameters
A series of parallel micro-slots with characteristic dimensions include pressure drop ⌬P, highest heat sink temperature Tmax,
ranging from 10 to 1000 ␮m are cut into the top surface of the minimum heat flux qmin⬙ , and maximum heat flux qmax⬙ . Given be-
base substrate. The cover plate is attached atop the base substrate low is a discussion on how to evaluate each of these thermal/fluid
to form rectangular micro-channels. While other channel geom- parameters using the best available predictive tools.
etries are possible, the rectangular geometry is often preferred for
ease of fabrication. The cover plate is made from a low-thermal- Pressure Drop ⌬P. Pressure drop ⌬P indicates the total pres-
conductivity material such as glass or high-temperature plastic to sure drop across a two-phase micro-channel heat sink. The mag-
ensure that all the heat is removed by the coolant. The heat sink is nitude of this parameter dictates the pumping power required to
attached to the top surface of a high-power-density electronic de- operate the heat sink and, in many cases, pump size and type.
vice as shown in Fig. 1. Heat generated by the device is conducted For most two-phase micro-channel heat sink applications, liq-
through the base substrate and removed by the coolant flowing uid coolant is supplied into the heat sink in subcooled state 共Tin
through the micro-channels. Figure 1 shows a unit cell containing ⬍ Tsat兲. The coolant maintains liquid state along the channel up to
a single micro-channel and surrounding solid. Taking advantage a location where thermodynamic equilibrium quality,
of symmetry between micro-channels, the cooling performance of
the heat sink can be derived from analysis of the unit cell alone. h − hf
xe = , 共4兲
A variety of system parameters must be examined when design- h fg
ing a two-phase micro-channel heat sink. As shown below, these
parameters can be grouped into 共1兲 geometrical parameters, 共2兲 reaches zero. Thereafter, the flow changes into a saturated two-
operating parameters, and 共3兲 thermal/fluid parameters. phase mixture. Figure 2共a兲 illustrates the flow regions of a micro-
Geometrical parameters include heat sink and micro-channel channel, a single-phase liquid region,
dimensions and are illustrated in Fig. 1. The overall heat sink ␳Qtc P,f 共Tsat − Tin兲
dimensions are length L, width W, and height H. When attaching Lsp = , 共5兲
a heat sink to an electronic device, the planform dimensions 共W qef⬙ fW
⫻ L兲 of the heat sink 共excepting the inlet and outlet plenums兲 are and a two-phase region, Ltp = L − Lsp. Tsat in Eq. 共5兲 is the satura-
often set equal to those of the device. Therefore, L and W are tion temperature at the location of xe = 0, which is evaluated using
specified beforehand, and serve as input parameters for heat sink the given heat sink outlet pressure Pout, assuming a small pressure
design. H is the sum of cover plate thickness Hc, channel height drop across the heat sink.
Hch, and distance between channel bottom wall and heat sink bot- The single-phase region can be further divided into developing
tom wall Hw. Since the cover plate can be treated as thermally and fully developed subregions. Since two-phase micro-channel
insulating, Hc has no bearing on the overall heat sink perfor- heat sinks feature both low flow rate and small channel size, the
mance. Hw, on the other hand, is proportional to the thermal con- flow in the single-phase liquid region is often well within the
duction resistance between the electronic device and micro- laminar range. The following equations can be employed to evalu-
channels and should be made as small as possible. In practice, ate the length of the two single-phase sub-regions 关34兴:

382 / Vol. 127, DECEMBER 2005 Transactions of the ASME


Table 1 Predictive tools for two-phase micro-channel heat
sinks

Fig. 2 „a… Flow regions in a micro-channel and „b… temperature


profile along stream-wise direction

Lsp,d = 共0.06 + 0.07␤ − 0.04␤2兲Rein dh , 共6兲


and
Lsp,f = Lsp − Lsp,d . 共7兲
⌬P is the sum of pressure drops across single-phase and two-
phase regions, as well as pressure losses and recoveries associated
with the contraction and expansion, respectively, at channel inlet
and outlet. The total pressure drop can be expressed as
⌬P = ⌬Pc + ⌬Psp,d + ⌬Psp,f + ⌬Ptp + ⌬Pe , 共8兲
where ⌬Pc is the inlet contraction pressure loss, ⌬Pe is the outlet
expansion pressure recovery, ⌬Psp,d and ⌬Psp,f denote pressure
drops in the single-phase developing and fully developed subre-
gions, respectively, and ⌬Ptp is the pressure drop across the two-
phase region. ⌬Ptp can be further expressed as the sum of accel-
erational and frictional components:
the two-phase region, however, Tc maintains a fairly constant
⌬Ptp = ⌬Ptp,a + ⌬Ptp,f . 共9兲 value equal to Tsat, and Tw and Ts increase only slightly with
Predictive tools for evaluating all the pressure drop components in distance. This behavior is attributed to the large value of the flow
Eqs. 共8兲 and 共9兲 are summarized in Table 1. boiling heat transfer coefficient htp 关29兴. The highest heat sink
temperature is always encountered in the heat sink’s bottom wall
Highest Temperature Tmax. The performance of an electronic immediately below the micro-channel outlet. Assuming one-
device to which a heat sink is attached is highly temperature de- dimensional heat conduction between the heat sink’s bottom wall
pendent. For a given q⬙ef f , the highest temperature in the device is and micro-channel bottom wall, Ts,out can be related to the micro-
encountered immediately below the location of highest tempera- channel bottom wall temperature at the channel outlet by
ture in the heat sink. This latter temperature should therefore be
kept as low as possible. ⬙ f Hw
qef
Ts,out = Tw,out + . 共10兲
Figure 2共b兲 shows the profiles of the coolant mean temperature ks
Tc, micro-channel bottom wall temperature Tw, and heat sink bot- Since Hw is not part of the optimization methodology and should
tom wall temperature Ts along the streamwise direction. Defini- be made as small as possible, Tw,out is used in the present study to
tions of these three temperatures are indicated in Fig. 1. In the represent Tmax:
single-phase region, Tc increases fairly linearly with increasing
streamwise distance, and Tw and Ts also increase accordingly. In Tmax = Tw,out . 共11兲

Journal of Electronic Packaging DECEMBER 2005, Vol. 127 / 383


The fin analysis method is used to evaluate Tw,out. This method ⬙ and qmin
Figures 3共a兲–3共c兲 show the variations of qmax ⬙ , ⌬P, and
models the solid walls separating micro-channels as thin fins and Tw,out with Qt for q⬙ef f = 100, 500, and 1000 W / cm2, respectively.
adopts approximations such as one-dimensional heat conduction On the left side of each figure are the results for the heat sinks
and a constant convective heat transfer coefficient along the entire with Hch = 1000 ␮m, and on the right side Hch = 500 ␮m.
heated perimeter 关29兴. Applying the fin analysis method to the unit Figures 3共a兲–3共c兲 show qmin ⬙ is a function of Qt alone and in-
cell shown in Fig. 1 yields the following energy balance at the creases with increasing Qt. qmax ⬙ , on the other hand, depends not
heat sink outlet, only on Qt but micro-channel dimensions 共Hch, Wch, and Ww兲 as
⬙ f 共Wch + Ww兲 = htp共Tw,out − Tsat兲共Wch + 2␩Hch兲,
qef 共12兲 well. The dependence of qmax ⬙ on the micro-channel dimensions
will be discussed in a later section. For given micro-channel di-
where htp and ␩ are the flow boiling heat transfer coefficient and
fin efficiency, respectively. The left-hand side of Eq. 共12兲 repre-
⬙ increases with increasing Qt. As discussed earlier,
mensions, qmax
sents heat input to the unit cell, and the right-hand side the rate of ⬙ and qmax
qmin ⬙ set lower and upper limits, respectively, for q⬙ef f , i.e.,
heat removal by flow boiling from the channel bottom wall and ⬙ 艋 qef
qmin ⬙ f 艋 qmax
⬙ 共16兲
sidewalls. Fin efficiency ␩ is evaluated by applying the thin fin
approximation to the channel sidewalls, ⬙ and qmax
for a given Qt. Since both qmin ⬙ increase monotonically
with increasing Qt, Eq. 共16兲 can be used to determine to minimum
tanh共mHch兲 and maximum values for Qt for each set of micro-channel dimen-
␩= , 共13兲
mHch sions:
where m is the fin parameter, Qt,min 艋 Qt 艋 Qt,max , 共17兲

m= 冑 2htp
k sW w
. 共14兲
where Qt,min and Qt,max can evaluated from the following rela-
tions, respectively:

htp is evaluated using the Warrier et al. correlation for micro- ⬙ 共Qt,max兲 = qef
qmin ⬙f 共18兲
channel flow boiling 关21兴. Other more accurate models can be and
used to evaluate htp 共e.g., 关30兴兲. However, the Warrier et al. cor-
relation is used here because of its relative ease of implementation ⬙ 共Qt,min兲 = qef
qmax ⬙ f. 共19兲
in the proposed design methodology as well as its reasonable pre- ⬙ , ⌬P and Tw,out are functions of both Qt and micro-
Like qmax
dictive capability 关29兴. Predictive tools for Tw,out are summarized channel dimensions as illustrated in Figs. 3共a兲–3共c兲. For fixed
in Table 1. micro-channel dimensions, ⌬P decreases with increasing Qt in the
low Qt region, but could either increase or decrease in the high Qt
Minimum Dissipative Heat Flux qmin ⬙ . The minimum dissipa- region depending on micro-channel dimensions. Tw,out increases
⬙ is the lowest heat flux that can sustain flow
tive heat flux qmin with increasing Qt in the low Qt region, and decreases slightly in
boiling in the micro-channels. It represents the input heat flux the high Qt region. The dependence of ⌬P and Tw,out on micro-
value that causes flow boiling to first occur at the channel outlet channel dimensions will be discussed in a later section.
and below which any q⬙ef f value would yield single-phase liquid Comparing the left and right sides of each figure reveals that for
flow throughout the micro-channel. Therefore, qmin⬙ is set equal to the same input parameters as well as Wch and Ww, heat sinks with
the dissipative heat flux value corresponding to zero thermody- deeper micro-channels 共larger Hch兲 always yield better perfor-
namic equilibrium quality at the channel outlet, xe,out = 0. From mance by providing lower values for both ⌬P and Tw,out.
Eq. 共5兲, Comparing Figs. 3共a兲–3共c兲 for different values of q⬙ef f shows a
␳Qtc P,f 共Tsat − Tin兲 higher q⬙ef f demands a higher Qt range and yields higher values for
L = Lsp = . 共15兲 both ⌬P and Tw,out.
⬙ W
qmin
The dependence of qmax ⬙ on micro-channel dimensions is illus-
⬙ given in Table 1.
Rearranging Eq. 共15兲 yields the relation for qmin ⬙
trated in Figs. 4共a兲 and 4共b兲. These two figures show constant qmax
Maximum Dissipative Heat Flux q⬙max. The maximum dissi- lines in a Wch-Ww plane at Qt = 120 ml/ min for Hch = 1000 and
⬙ represents maximum value of device heat
pative heat flux qmax 500 ␮m, respectively. qmax⬙ increases with decreasing Wch and Ww
flux that can be removed by the heat sink. For a two-phase heat and reaches the highest value at the left lower corner of each plot,
i.e., for the lowest Wch and Ww values.
⬙ is set by the critical heat flux 共CHF兲 in the micro-
sink, qmax
channels since exceeding this limit would precipitate a sudden ⬙ = 667 W / cm2, inde-
For Qt = 120 ml/ min, Eq. 共15兲 yields qmin
large increase in the heat sink temperature, which can lead to pendent of Wch or Ww. The dashed line corresponding to
permanent device failure. A CHF correlation recently developed ⬙ = qmin
qmax ⬙ = 667 W/cm2 共20兲
by the present authors for two-phase micro-channel heat sinks
关33兴 is used to evaluate qmax⬙ . At very low coolant flow rates, in Figs. 4共a兲 and 4共b兲 sets an upper limit for the region of the
however, the CHF correlation yields a heat flux value higher than ⬙ 艌 qmin
Wch-Ww plane where Eq. 共16兲 is valid, i.e., where qmax ⬙ .
the input heat flux value required to convert all the liquid to satu- Comparing Figs. 4共a兲 and 4共b兲 reveals that for fixed values of
rated vapor at channel outlet, xe,out = 1. Under these conditions, the Wch and Ww a deeper micro-channel always yields a higher qmax ⬙
⬙ . A sum-
input heat flux corresponding to xe,out = 1 is used for qmax value. This observation further proves that micro-channels should
be made as deep as possible, while being mindful of machining
mary of the predictive relations for qmax⬙ is given in Table 1.
and structural limitations.

Characteristics of Thermal/Fluid Parameters Heat Sink Optimal Design


A water-cooled copper two-phase micro-channel heat sink with Design Under Fixed Flow Rate. Flow rate or pressure drop are
planform dimensions 共W ⫻ L兲 of 1 ⫻ 1 cm2 is used to illustrate the key constraints in the design of micro-channel heat sinks since
general trends of the aforementioned thermal/fluid parameters. such heat sinks often demand specialized micro-pumps whose
Two channel heights are compared, 500 and 1000 ␮m. For each performance is dictated by either flow rate or pressure drop.
Hch value, four 共Wch , Ww兲 combinations 共50⫻ 50, 50⫻ 100, 100 In this section, optimal design of a two-phase micro-channel
⫻ 50, 100⫻ 100 ␮m2兲 are examined. Tin and Pout are set at 25° C heat sink under fixed total volume flow rate Qt is explored. The
and 1.2 bar, respectively. water-cooled copper two-phase micro-channel heat sink is again

384 / Vol. 127, DECEMBER 2005 Transactions of the ASME


Fig. 3 Variations of minimum effective heat flux, maximum effective heat flux, pressure drop, and channel bottom tem-
⬙ = 100 W / cm2, „b… q eff
perature at heat sink outlet with total volume flow rate for „a… q eff ⬙ = 500 W / cm2, and „c… q eff

= 1000 W / cm2.

Journal of Electronic Packaging DECEMBER 2005, Vol. 127 / 385


Fig. 4 Performance map for maximum effective heat flux for Fig. 5 Performance map for „a… pressure drop and „b… channel
„a… Hch = 1000 ␮m and „b… Hch = 500 ␮m ⬙ = 500 W / cm2 and
bottom temperature at heat sink outlet for q eff
Qt = 60 ml/ min

employed to illustrate the design methodology. Input parameters


are summarized as follows: heat sink planform dimensions 共W
⫻ L兲 of 1 ⫻ 1 cm2, Tin and Pout of 25° C and 1.2 bar, respectively, fabrication technique; the values used here are intended only to
device heat flux of q⬙ef f = 500 W / cm2, and a flow rate of Qt illustrate the use of the optimization methodology.
= 60 ml/ min. Although Qt is an input parameter, it is necessary to make cer-
Machining and/or structural limitations should always be exam- tain Qt satisfy Eq. 共17兲. Otherwise, a different Qt value must be
ined first to identify maximum channel height Hch,max, minimum assigned from the start. When Qt,min is evaluated from Eq. 共19兲,
width, Wch,min, and minimum wall thickness, Ww,min. Another con- Hch,max, Wch,max, and Ww,min should be used for micro-channel
sideration in setting a minimum channel width is the need to avoid dimensions as this combination yields the highest possible qmax ⬙
clogging. value, and therefore lowest Qt,min.
As discussed in the previous section, deeper micro-channels Figures 5共a兲 and 5共b兲 show the output of the optimization meth-
always produce better performance. Therefore, Hch,max is em- odology in a Wch-Ww plane. For a fixed flow rate of Qt
ployed for Hch:
= 60 ml/ min and a device heat flux of q⬙ef f = 500 W / cm2, the ac-
Hch = Hch,max . 共21兲 ceptable range of two-phase operation is confined to the region
Once Hch is fixed, only two other design parameters, Wch and Ww, below the dashed line corresponding to qmax ⬙ = q⬙ef f = 500 W / cm2
need to be determined, which should satisfy the following rela- since only 共Wch , Ww兲 combinations that are located in this region
tions, satisfy the requirement q⬙ef f ⬍ qmax
⬙ . In the case where a factor of
safety on CHF needs to be assigned, the acceptable design region
Wch 艌 Wch,min 共22兲 is obtained in a similar manner using the line corresponding to
and ⬙ = ⌽q⬙ef f , where ⌽ is the factor of safety on CHF.
qmax
The next step involves evaluating ⌬P and Tw,out corresponding
Ww 艌 Ww,min . 共23兲 to micro-channel dimensions in the acceptable design region. Pre-
In the present study, Hch,max is set at 1000 ␮m, and both Wch,min dictive tools given Table 1 are employed to conduct the evalua-
and Ww,min at 50 ␮m. It is important to note that these values tion. Figures 5共a兲 and 5共b兲 show lines of constant ⌬P and constant
generally depend on heat sink material, coolant type, and heat sink Tw,out, respectively. Figure 5共a兲 shows ⌬P is low for large Wch and

386 / Vol. 127, DECEMBER 2005 Transactions of the ASME


Fig. 6 Performance map for „a… pressure drop and „b… channel
Fig. 7 Performance map for „a… total volume flow rate and „b…
⬙ = 500 W / cm2 and
bottom temperature at heat sink outlet for q eff
channel bottom temperature at heat sink outlet for q eff ⬙
Qt = 40 ml/ min
= 500 W / cm2 and ⌬P = 0.1 bar

small Ww and increases with decreasing Wch and increasing Ww. timal design of a two-phase micro-channel heat sink must be per-
Figure 5共b兲, on the other hand, shows Tw,out is low for small Wch formed under the requirement of a fixed pressure drop ⌬P, instead
and small Ww and increases with increasing Wch. of a fixed flow rate Qt. This section outlines the design procedure
Figures 6共a兲 and 6共b兲 show similar plots corresponding to a corresponding to a fixed ⌬P. Input parameters discussed here are
lower flow rate of Qt = 40 ml/ min. The general trends in these identical to those in the previous section with the exception that
figures are similar to those of Figs. 5共a兲 and 5共b兲, respectively, Qt is now an output variable and pressure drop is fixed at ⌬P
⬙ limit is shifted towards smaller channel di-
except that the qmax = 0.1 bar.
mensions. Comparing Figs. 5共a兲 and 6共a兲 reveals that for the same The design procedure is very similar to that discussed in the
micro-channel dimensions, ⌬P increases with decreasing Qt. This previous section. As indicated in Figs. 7共a兲 and 7共b兲, the accept-
is because the pressure drop in the two-phase region is higher than able design region under fixed ⌬P is enveloped by lines corre-
in the single-phase region. With a lower flow rate, two-phase flow sponding to Qt = Qt,max = 90 ml/ min and qmax ⬙ = q⬙ef f = 500 W / cm2.
occupies a larger portion of the micro-channel length, resulting in Lines corresponding to constant values of Qt and Tw,out are con-
higher total pressure drop. A comparison of Figs. 5共b兲 and 6共b兲 structed in Figs. 7共a兲 and 7共b兲, respectively, within the acceptable
shows Tw,out decreases with decreasing Qt. These trends are valid design region.
only for low values of Qt; high Qt values are expected to yield the Figures 8共a兲 and 8共b兲 show similar plots for a higher pressure
opposite trend as shown in Figs. 3共a兲–3共c兲. drop of ⌬P = 0.2 bar. The general trends in these plots are similar
Figures such as 5共a兲, 5共b兲, 6共a兲, and 6共b兲 are the final output of to those of Figs. 7共a兲 and 7共b兲 except that the acceptable design
the numerical optimization procedure. They provide the end user region is shifted to lower Wch and high Ww values.
with an acceptable range of 共Wch , Ww兲 combinations which can Figures such as 7共a兲, 7共b兲, 8共a兲, and 8共b兲 are the final output of
safely remove the heat for a given flow rate. The heat sink de- the numerical optimization corresponding to a fixed ⌬P. The heat
signer is left with the decision to select an acceptable combination sink designer is left with the decision to select an acceptable com-
of ⌬P and Tw,out; this latter combination dictates the dimensions bination of Qt and Tw,out based upon which the channel dimen-
共Wch and Ww兲 of the micro-channel. sions 共Wch and Ww兲 are determined.
Design Under Fixed Pressure Drop. In some applications, op- Summary of Optimization Procedure. Figure 9 shows a flow

Journal of Electronic Packaging DECEMBER 2005, Vol. 127 / 387


Fig. 8 Performance map for „a… total volume flow rate and „b…
channel bottom temperature at heat sink outlet for q eff ⬙
= 500 W / cm2 and ⌬P = 0.2 bar Fig. 9 Flow chart of numerical procedure for optimal design of
two-phase micro-channel heat sink

chart that summarizes the numerical procedure for optimal design


of a two-phase micro-channel heat sink under fixed total volume heat sink and whose magnitude is used to guide the designer
flow rate or fixed pressure drop. Figures 10共a兲 and 10共b兲 illustrate in selecting optimum micro-channel dimensions.
the general trends of output parameters within the acceptable de- 2. With the exception of minimum dissipative heat flux re-
sign region for fixed total volume flow rate and fixed pressure quired to initiate boiling at the channel outlet, all other
drop, respectively. thermal/fluid parameters are sensitive to micro-channel di-
mensions.
3. A design procedure for two-phase micro-channel heat sinks
Conclusions under the conditions of a fixed device heat flux and fixed
This study concerns optimal design of two-phase micro-channel total volume flow rate is developed. This procedure yields an
heat sinks. System parameters and predictive tools were first sum- acceptable design region encompassing all possible micro-
marized and are followed by a discussion of the characteristics of channel dimensions and provides predictions for pressure
thermal/fluid transport parameters. Finally, a systematic optimiza- drop and highest heat sink temperature from which a de-
tion methodology for heat sink design was developed for the con- signer can select optimum channel dimensions.
ditions of either fixed total volume flow rate or fixed pressure 4. A similar design procedure is proposed for the conditions of
drop. Key findings from the study are as follows: fixed device heat flux and fixed pressure drop. This proce-
dure is used to select optimum channel dimensions that yield
1. System parameters associated with two-phase micro-channel acceptable values for flow rate and highest heat sink
heat sinks can be grouped into geometrical parameters, op- temperature.
erating parameters, and thermal/fluid parameters. Geometri-
cal parameters include heat sink and micro-channel dimen-
sions. In heat sink design, heat sink dimensions and Acknowledgment
operating parameters are used as input parameters and are The authors are grateful for the support of the Office of Basic
often specified beforehand. Thermal/fluid parameters are Energy Sciences of the U.S. Department of Energy 共Award No.
output parameters that describe the transport behavior of the DE-FG02-93ER14394 A7兲.

388 / Vol. 127, DECEMBER 2005 Transactions of the ASME


Hw ⫽ distance from micro-channel bottom wall to
heat sink bottom wall
k ⫽ thermal conductivity
K共⬁兲 ⫽ entrance loss coefficient
Kc ⫽ contraction loss coefficient
Ke ⫽ expansion recovery coefficient
L ⫽ length of heat sink
Lsp ⫽ length of single-phase region
+
Lsp,d ⫽ nondimensional length of single-phase devel-
oping subregion
Lsp,d ⫽ length of single-phase developing subregion
Lsp,f ⫽ length of single-phase fully developed
subregion
Ltp ⫽ length of two-phase region
m ⫽ fin parameter
N ⫽ number of micro-channels in heat sink
Nu3 , Nu4 ⫽ Nusselt number for laminar fully developed
flow for three and four wall heat transfer
P ⫽ pressure
Pout ⫽ outlet pressure
Pw ⫽ total rate of heat supplied to heat sink
⌬P ⫽ pressure drop across heat sink
⌬Pc ⫽ contraction pressure loss
⌬Pe ⫽ expansion pressure recovery
⌬Psp,d ⫽ pressure drop in single-phase developing
subregion
⌬Psp,f ⫽ pressure drop in single-phase fully developed
subregion
⌬Ptp ⫽ pressure drop in two-phase region
⌬Ptp,a ⫽ accelerational two-phase pressure drop
⌬Ptp,f ⫽ frictional two-phase pressure drop
q⬙ef f ⫽ dissipative heat flux based on heat sink plan-
form area
⬙ ⫽ maximum dissipative heat flux
qmax
⬙ ⫽ minimum dissipative heat flux
qmin
q⬙p ⫽ mean heat flux over heated inside area of
micro-channel
Qt ⫽ total volume flow rate
Fig. 10 General performance map trends for „a… fixed total vol- Qt,max ⫽ maximum total volume flow rate
ume flow rate and „b… fixed pressure drop Qt,min ⫽ minimum total volume flow rate
Re ⫽ Reynolds number
Re f ⫽ Reynolds number based on local liquid flow
rate
Nomenclature Rein ⫽ Reynolds number based on inlet liquid
Ach ⫽ cross-sectional area of micro-channel conditions
Ap ⫽ cross-sectional area of plenum Resp ⫽ Reynolds number based on properties at mean
Bo ⫽ boiling number temperature in single-phase liquid region
C ⫽ parameter in pressure drop relations T ⫽ temperature
cP ⫽ specific heat at constant pressure Tc ⫽ coolant mean temperature
de ⫽ heated equivalent diameter Tin ⫽ inlet temperature
dh ⫽ hydraulic diameter of micro-channel Tmax ⫽ highest temperature in optimization procedure
E ⫽ parameter in heat transfer relations 共Tmax = Tw,out兲
f ⫽ friction factor Ts ⫽ heat sink bottom wall temperature
f app ⫽ apparent friction factor for developing single- Tsat ⫽ saturation temperature
phase liquid flow Tw ⫽ micro-channel bottom wall temperature
ff ⫽ friction factor based on local liquid flow rate v ⫽ specific volume
f sp ⫽ friction factor in single-phase fully developed v fg ⫽ specific volume difference between saturated
region liquid and saturated vapor
G ⫽ mass velocity W ⫽ width of heat sink
h ⫽ enthalpy Wch ⫽ width of micro-channel
hf ⫽ enthalpy of saturated liquid We ⫽ weber number
h fg ⫽ latent heat of vaporization Ww ⫽ thickness of wall separating micro-channels
hsp ⫽ single-phase convective heat transfer xe ⫽ thermodynamic equilibrium quality
coefficient xe,out ⫽ thermodynamic equilibrium quality at channel
htp ⫽ flow boiling heat transfer coefficient outlet
H ⫽ height of heat sink Xvv ⫽ Martinelli parameter for laminar liquid-laminar
Hc ⫽ cover plate thickness vapor flow
Hch ⫽ height of micro-channel z ⫽ stream-wise distance

Journal of Electronic Packaging DECEMBER 2005, Vol. 127 / 389


Greek Symbols 关14兴 Zhang, L., Koo, J. M., Jiang, L., Banerjee, S. S., Ashegi, M., Goodson, K. E.,
Santiago, J. G., and Kenny, T. W., 2000, “Measurement and Modeling of
␣ ⫽ void fraction Two-Phase Flow in Microchannels with Nearly-Constant Heat Flux Boundary
␤ ⫽ aspect ratio of micro-channel Conditions,” Micro-Electro-Mechanical Systems (MEMS)—2000, A. Lee et al.,
␾2f ⫽ two-phase frictional multiplier based on local eds., ASME, MEMS-Vol. 2, pp. 129–135.
liquid flow rate 关15兴 Jiang, L., Wong, M., and Zohar, Y., 2001, “Forced Convection Boiling in a
Microchannel Heat Sink,” J. Microelectromech. Syst., 10, pp. 80–87.
⌽ ⫽ factor of safety on CHF 关16兴 Qu, W., and Mudawar, I., 2004, “Transport Phenomena in Two-Phase Micro-
␩ ⫽ fin efficiency Channel Heat Sinks,” ASME J. Electron. Packag., 126, pp. 213–224.
␮ ⫽ viscosity 关17兴 Kandlikar, S. G., Steinke, M. E., Tian, S., and Campbell, L. A., 2001, “High-
␳ ⫽ density Speed Photographic Observation of Flow Boiling of Water in Parallel Mini-
Channels,” Proceedings of National Heat Transfer Conference, ASME, pp.
␴ ⫽ surface tension 675–684.
关18兴 Hetsroni, G., Mosyak, A., Segal, Z., and Ziskind, G., 2002, “A Uniform Tem-
Subscripts perature Heat Sink for Cooling of Electronic Devices,” Int. J. Heat Mass
f ⫽ liquid Transfer, 45, pp. 3275–3286.
g ⫽ vapor 关19兴 Wu, H. Y., and Cheng, P., 2003, “Visualization and Measurements of Periodic
in ⫽ inlet Boiling in Silicon Microchannels,” Int. J. Heat Mass Transfer, 46, pp. 2603–
2614.
max ⫽ maximum 关20兴 Qu, W., and Mudawar, I., 2003, “Measurement and Prediction of Pressure
min ⫽ minimum Drop in Two-Phase Micro-Channel Heat Sinks,” Int. J. Heat Mass Transfer,
out ⫽ outlet 46, pp. 2737–2753.
p ⫽ inlet or outlet plenum 关21兴 Warrier, G. R., Dhir, V. K., and Momoda, L. A., 2002, “Heat Transfer and
Pressure Drop in Narrow Rectangular Channel,” Exp. Therm. Fluid Sci., 26,
s ⫽ base substrate pp. 53–64.
sp ⫽ single-phase liquid 关22兴 Bowers, M. B., and Mudawar, I., 1994, “High Flux Boiling in Low Flow Rate,
sp, d ⫽ single-phase developing subregion Low Pressure Drop Mini-Channel and Micro-Channel Heat Sinks,” Int. J. Heat
sp, f ⫽ single-phase fully developed subregion Mass Transfer, 37, pp. 321–332.
关23兴 Bowers, M. B., and Mudawar, I., 1994, “Two-Phase Electronic Cooling Using
tp ⫽ two phase Mini-Channel and Micro-Channel Heat Sinks: Part 1—Design Criteria and
Heat Diffusion Constraints,” ASME J. Electron. Packag., 116, pp. 290–297.
References 关24兴 Bowers, M. B., and Mudawar, I., 1994, “Two-Phase Electronic Cooling Using
Mini-Channel and Micro-Channel Heat Sinks: Part 2—Flow Rate and Pressure
关1兴 Tuckerman, D. B., and Pease, R. F. W., 1981, “High-Performance Heat Sink- Drop Constraints,” ASME J. Electron. Packag., 116, pp. 298–305.
ing for VLSI,” IEEE Electron Device Lett., EDL-2, pp. 126–129. 关25兴 Tran, T. N., Chyu, M. C., Wambsganss, M. W., and France, D. M., 2000,
关2兴 Kishimito, T., and Ohsaki, T., 1986, “VLSI Packaging Technique Using “Two-Phase Pressure Drop of Refrigerants During Flow Boiling in Small
Liquid-Cooled Channels,” IEEE Trans. Compon., Hybrids, Manuf. Technol., Channels: An Experimental Investigation and Correlation Development,” Int.
CHMT-9, pp. 328–335. J. Multiphase Flow, 26, pp. 1739–1754.
关3兴 Rahman, M. M., and Gui, F., 1993, “Experimental Measurements of Fluid 关26兴 Lee, H. J., and Lee, S. Y., 2001, “Heat Transfer Correlation for Boiling Flows
Flow and Heat Transfer in Microchannel Cooling Passages in a Chip Sub- in Small Rectangular Horizontal Channels With Low Aspect Ratios,” Int. J.
strate,” ASME J. Electron. Packag., EEP-4, pp. 495–506.
Multiphase Flow, 27, pp. 2043–2062.
关4兴 Ravigururajan, T. S., Cuta, J., McDonald, C. E., and Drost, M. K., 1996,
关27兴 Peng, X. F., and Wang, B. X., 1993, “Forced Convection and Flow Boiling
“Single-Phase Flow Thermal Performance Characteristics of a Parallel Micro-
Heat Transfer for Liquid Flowing Through Microchannels,” Int. J. Heat Mass
Channel Heat Exchanger,” Proceedings of National Heat Transfer Conference,
Transfer, 36, pp. 3421–3427.
ASME HTD-329, Vol. 7, pp. 157–166.
关28兴 Ravigururajan, T. S., 1998, “Impact of Channel Geometry on Two-Phase Flow
关5兴 Kawano, K., Minakami, K., Iwasaki, H., and Ishizuka, M., 1998, “Micro
Channel Heat Exchanger for Cooling Electrical Equipment,” Proceedings of Heat Transfer Characteristics of Refrigerants in Microchannel Heat Exchang-
the ASME Heat Transfer Division, ASME HTD-361–3/PID-3, pp. 173–180. ers,” ASME J. Heat Transfer, 120, pp. 485–491.
关6兴 Harms, T. M., Kazmierczak, M. J., and Cerner, F. M., 1999, “Developing 关29兴 Qu, W., and Mudawar, I., 2003, “Flow Boiling Heat Transfer in Two-Phase
Convective Heat Transfer in Deep Rectangular Microchannels,” Int. J. Heat Micro-Channel Heat Sinks-I. Experimental Investigation and Assessment of
Fluid Flow, 20, pp. 149–157. Correlation Methods,” Int. J. Heat Mass Transfer, 46, pp. 2755–2771.
关7兴 Qu, W., and Mudawar, I., 2002, “Experimental and Numerical Study of Pres- 关30兴 Qu, W., and Mudawar, I., 2003, “Flow Boiling Heat Transfer in Two-Phase
sure Drop and Heat Transfer in a Single-Phase Micro-Channel Heat Sink,” Int. Micro-Channel Heat Sinks-II. Annular Two-Phase Flow Model,” Int. J. Heat
J. Heat Mass Transfer, 45, pp. 2549–2565. Mass Transfer, 46, pp. 2773–2784.
关8兴 Knight, R. W., Goodling, J. S., and Hall, D. J., 1991, “Optimal Thermal Design 关31兴 Roach, G. M., Jr., Abdel-Khalik, S. I., Ghiaasiaan, S. M., Dowling, M. F., and
of Forced Convection Heat Sinks—Analytical,” ASME J. Electron. Packag., Jeter, S. M., 1999, “Low-Flow Critical Heat Flux in Heated Microchannels,”
113, pp. 313–321. Nucl. Sci. Eng., 131, pp. 411–425.
关9兴 Knight, R. W., Hall, D. J., Goodling, J. S., and Jaeger, R. C., 1992, “Heat Sink 关32兴 Yu, W., France, D. M., Wambsganss, M. W., and Hull, J. R., 2002, “Two-Phase
Optimization with Application to Microchannels,” IEEE Trans. Compon., Hy- Pressure Drop, Boiling Heat Transfer, and Critical Heat Flux to Water in a
brids, Manuf. Technol., 15, pp. 832–842. Small-Diameter Horizontal Tube,” Int. J. Multiphase Flow, 28, pp. 927–941.
关10兴 Lee, D. Y., and Vafai, K., 1999, “Comparative Analysis of Jet Impingement 关33兴 Qu, W., and Mudawar, I., 2004, “Measurement and Correlation of Critical
and Microchannel Cooling for High Heat Flux Applications,” Int. J. Heat Mass Heat Flux in Two-Phase Micro-Channel Heat Sinks,” Int. J. Heat Mass Trans-
Transfer, 42, pp. 1555–1568. fer, 47, pp. 2045–2059.
关11兴 Choquette, S. F., Faghri, M., Charmchi, M., and Asako, Y., 1996, “Optimum 关34兴 Shah, R. K., and London, A. L., 1978, Laminar Flow Forced Convection in
Design of Microchannel Heat Sinks,” Micro-Electro-Mechanical Systems Ducts: A Source Book for Compact Heat Exchanger Analytical Data, Suppl. 1,
(MEMS)—1996, DSC-Vol. 59, ASME, pp. 115–126. Academic Press, New York.
关12兴 Kennedy, J. E., Roach, G. M., Jr., Dowling, M. F., Abdel-Khalik, S. I., Ghi- 关35兴 Blevins, R. D., 1984, Applied Fluid Dynamics Handbook, Van Nostrand Re-
aasiaan, S. M., Jeter, S. M., and Quershi, Z. H., 2000, “The Onset of Flow inhold Company, New York.
Instability in Uniformly Heated Horizontal Microchannels,” ASME J. Heat 关36兴 Collier, J. G., and Thome, J. R., 1994, Convective Boiling and Condensation,
Transfer, 122, pp. 118–125. 3rd ed., Oxford University Press, Oxford.
关13兴 Qu, W., and Mudawar, I., 2002, “Prediction and Measurement of Incipient 关37兴 Zivi, S. M., 1964, “Estimation of Steady-State Steam Void-Fraction by Means
Boiling Heat Flux in Micro-Channel Heat Sinks,” Int. J. Heat Mass Transfer, of the Principle of Minimum Entropy Production,” ASME J. Heat Transfer,
45, pp. 3933–3945. 86, pp. 247–252.

390 / Vol. 127, DECEMBER 2005 Transactions of the ASME