86
IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 3, NO. 1, JANUARY 2013
Performance of Online and Offset Micro Pin-Fin
Heat Sinks With Variable Fin Density
Carlos A. Rubio-Jimenez, Satish G. Kandlikar, and Abel Hernandez-Guerrero
Abstract— A comparison of the performances of online and
offset micro pin-fin heat sinks with variable fin density is given
in this paper. The cooling systems generate uniform junction
temperatures, which improves the integrated circuit (IC) chip’s
performance. Water is used as a coolant in the single phase and
laminar regime. 4748 micro flat fins with rounded sides, which are
distributed in three different sections along the flow length, are
used in these configurations. The bottom wall temperature profile
along the flow length, overall thermal resistances, pressure drops,
and pumping powers for both configurations are presented. The
results indicate that the offset micro pin-fin heat sink is a good
alternative for cooling the IC chips of 2016. The cooling system
using this fin configuration is capable of achieving a thermal
resistance as low as 0.1 K/W with a pumping power requirement
of 0.45 W. Comparisons with other cooling devices reported in
the technical literature are presented.
Index Terms— Micro pin-fin heat sink, online and offset
fin configurations, uniform junction temperature, variable fin
density.
I. I NTRODUCTION
E
LECTRONIC devices are the keystone in a large part of
industrial processes and entertainment. The main component of these devices is the integrated circuit (IC) chip. In the
past decades, these microelectronic systems have shown extraordinary growth. Currently, high-tech electronic devices (e.g.,
Intel and Xeon processors) have achieved processing speeds
of 6.4 GT/s (3.2 GHz) in multicore packages [1]. This high
performance comes from deep studies upon each component,
making each single electronic device highly reliable, smaller,
lighter, and cheaper. This trend in miniaturization generates
important technological challenges: research of new materials
and manufacturing techniques, appropriate heat dissipation,
and packaging system, etc.
Manuscript received March 3, 2012; revised August 29, 2012; accepted
September 14, 2012. Date of publication December 11, 2012; date of current
version January 4, 2013. This work was supported by the Consejo Nacional
de Ciencia y Tecnología (CONACYT), Mexico. This work was conducted
at the Thermal Analysis, Microfluidics and Fuel Cell Laboratory, Rochester
Institute of Technology. Recommended for publication by Associate Editor P.
Dutta upon evaluation of reviewers’ comments.
C. A. Rubio-Jimenez was with the University of Guanajuato, Salamanca
36885, Mexico. He is now with the Department of Mechanical Engineering, Rochester Institute of Technology, Rochester, NY 14623 USA (e-mail:
[email protected]).
S. G. Kandlikar is with the Department of Mechanical Engineering,
Rochester Institute of Technology, Rochester, NY 14623 USA (e-mail:
[email protected]).
A. Hernandez-Guerrero is with the Department of Mechanical Engineering, Universidad de Guanajuato, Salamanca 36885, Mexico (e-mail:
[email protected]).
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TCPMT.2012.2225143
In 2006, the international technology roadmap for semiconductors (ITRS) forecast the scenario for the generation of
9-nm CMOS devices for 2016 [2]. According to this roadmap,
the next generation of IC chips is going to be designed
with the current maximum design temperature (< 85 °C), but
the thermal energy generated by them should be up to 288
W. This implies that the new generation of cooling systems
should have a thermal resistance ranging from 0.14 °C/W to
0.25 °C/W with high reliability and adequate size. Currently,
IC chip cooling systems are based on air-cooling technologies.
The total heat transfer surface required by these heat sinks
is directly proportional to the total thermal energy to be
dissipated. Clearly, this is an important drawback for future
applications since their size and weight are limiting factors.
Liquid cooling technologies have emerged as a viable
alternative for cooling electronic devices. Tuckerman and
Pease [3] showed that it is possible to substantially increase
the heat transfer coefficient when a fluid is passing through
microchannels. Heat sinks based on these channel dimensions
were analyzed experimentally. Their results showed that this
kind of cooling systems can achieve a thermal resistance
of 0.09 K/W, and dissipate up to 790 W/cm2 with a water
temperature rise of 71 °C. Although these results indicate that
this kind of cooling systems can be perfectly used in the near
future in IC chip packaging, there are important drawbacks
that limit their application. First, the fluid pressure drop is very
large (>200 kPa), affecting directly the energy required by the
system. Second, the temperature at the bottom wall of the heat
sink along the flow length is highly nonuniform (T ∼ 70 °C
in 10 mm), reducing seriously the chip’s performance and
lifetime [4], [5] and causing clock skew on the system [6].
Numerous studies on micro phenomena [7]–[9] and designs
of reliable microchannel heat sinks [10]–[12] can be found in
the literature. Some nonconventional techniques and methodologies have been used to design this kind of microcooling
systems [13]–[15]; however, only a few works have shown
outstanding performance that suit future needs. Among the
most outstanding works is the research of Peles et al. [16].
They proposed a micro pin-fin heat sink with constant fin
density as an alternative cooling system for IC chips. They
used the correlations provided previously in a bank of tube
analyses at the macroscale. Their experimental results showed
that this configuration was capable of dissipating up to
790 W/cm2 with a temperature rise of 30.7 °C. Thus, the
system thermal resistance is 0.039 K/W. The major drawback
of this heat sink was the large pressure drop generated in
the system (>200 kPa). Colgan et al. [17], [18] proposed
a nonconventional heat sink formed by two layers. Several
2156–3950/$31.00 © 2012 IEEE
RUBIO-JIMENEZ et al.: PERFORMANCE OF ONLINE AND OFFSET MICRO PIN-FIN HEAT SINKS
stages of fins were manufactured on the first layer. Inlet/outlet
zigzagged sections were incorporated on the second layer.
Then, both layers were bonded face to face. The results showed
that this system could achieve a thermal resistance as low
as 12 mm2 °C/W and dissipate a heat flux of 50 W/cm2
when water is flowing through the arrangement. The maximum
pressure drop was reported as 65 kPa. In the same way,
Wälchli et al. [19] analyzed a heat sink configuration formed
by three interconnected layers. Flowing water was used as the
coolant. The change of direction caused by the horizontal and
vertical channel interconnections increased strongly the heat
transfer coefficient. Their results showed that this configuration
was capable of reaching thermal resistance values as low as
0.08 K/W with a pressure drop of only 40 kPa. Recently,
Escher et al. [20] proposed a novel cooling system based on
the results of Wälchli et al. [19]. Their system was formed by
manifold channels that were placed perpendicularly. Numerical analysis showed that this kind of arrangement dissipates
up to 750 W/cm2 with a temperature difference of 65 °C and
a pressure drop of lower than 10 kPa. The thermal resistance
of this device was around 0.08 m2 K/W. These results show
an improvement in the reliability of microcooling systems
because this heat sink can dissipate the same heat flux as the
microchannel heat sink proposed in [3] but with 10 times lower
pressure drop. Although these results are very encouraging for
near-future applications, further studies are needed in order
to reduce the large temperature variation that these devices
generate in the IC chip. Moreover, several investigations are
aimed at the design of 3-D IC chip packaging. Clearly, an
appropriate thermal management is extremely required. Thus,
new designs of microcooling systems have to consider easy
stack integration.
An alternative for reducing the temperature variation on
the IC chips is the use of flow boiling [21]. However, water
at subatmospheric conditions or a refrigerant is required,
resulting in significant increase in the cost of the system.
Recently, Rubio-Jimenez et al. [22] proposed a micro pinfin heat sink with variable fin density as an alternative for
cooling the next generation of IC chips. The design of these
systems considers the reduction of the temperature variation
on the IC chip. Their numerical results show that a micro
pin-fin heat sink with 4748 flat fins placed in an online fin
configuration can achieve thermal resistance values ranging
from 0.14 to 0.25 K/W with pressure drops lower than
100 kPa. The temperature gradient at the junction wall between
the heat sink and the IC chip was proposed as a parameter of
comparison. Temperature gradients lower than 2 °C/mm were
observed.
On the other hand, Steinke and Kandlikar [23] have proposed and studied some passive enhancement techniques for
microchannels and mini-channels. Their discussion indicates
that the inclusion of some flow disruptions can serve to
trip the boundary layer and, importantly, increase the heat
dissipation. Offset strip fins were proposed as an alternative
for microchannel heat sinks. Carefully constructed geometries
were suggested.
Steinke and Kandlikar [24] and Steinke et al. [25] carried
out experimental analyses in enhanced microchannel heat
87
Fig. 1. Sketch of micro pin-fin heat sink with variable fin density for both
online and offset fin configurations.
sinks formed by offset strip fins with 250–500-μm length and
50-μm channel width. Their results showed that this device
could achieve a thermal resistance as low as 0.1 K/W and
dissipate up to 1000 W/cm2 . However, the pressure drop was
still in the range of 150–180 kPa.
Taking into account the conclusions of [23], this paper
analyzes the overall performance of micro heat sinks when
offset fin configurations with variable fin density are used
as the base for designing microcooling systems. Therefore,
the online fin configuration analyzed in [22] is used as the
starting point (MF-50 × 100 × 200 − 66). Then, the thermal
and hydrodynamic performances of both online and offset
configurations are compared. Furthermore, a comparison of
these heat sinks with other microcooling systems found in the
literature is provided.
II. M ODEL D EFINITION
In this paper, a 10 × 10 mm IC chip is considered as
the base for designing and building the offset micro pin-fin
heat sinks with variable fin density. The cooling system is
formed by 4748 flat fins placed on a 200-μm-thick silicon
substrate. The fin width, length, and height are 50, 100, and
200 μm, respectively. The rounded sides have a radius of
25 μm. The heat sink is divided longitudinally in three sections
(SI, SII, and SIII). Fig. 1 shows a sketch of these sections for
both online and offset fin configurations. Table I shows the
parameters for each section.
For comparison, a microchannel heat sink formed by
33 rectangular channels with 200 μm height, channel aspect
ratio 1.0, space between channels 100 μm, and substrate
silicon thickness 200 μm is analyzed. Only 1/66 of the heat
sinks were modeled because of their symmetry.
88
IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 3, NO. 1, JANUARY 2013
TABLE I
H EAT S INK PARAMETERS FOR B OTH M ICRO P IN -F IN H EAT S INKS
Parameter
Length (mm)
No of pin fins
No. of longitudinal rows
No. of transversal rows
S L (μm)
ST (μm)
ρpin fin (No. of pin fins/mm3 )
Section I
Section II
Section III
1.5
330
10
33
150
300
111.11
5.2
2240
34
66
150
150
217.94
3.3
2178
22
99
150
100
333.33
Symmetries
Flat-shaped fins
Fluid
flow
III. N UMERICAL A NALYSIS
A. Assumptions
The numerical analysis is based on the following assumptions.
1) The system is in the steady state.
2) Water in single phase and laminar regime enters the system at ambient temperature (25 °C) under nondeveloped
conditions.
3) The fluid and solid thermal properties remain constant,
except fluid viscosity, which is given by (1). The justification to this assumption can be found in [26] and
[27].
4) The heat flux is constant and uniform.
5) Radiation effects can be neglected
247.8 −5
10 T −140 .
(1)
μ = 2.414 × 10
Silicon substrate
Fig. 2. Mesh generated in the offset micro pin-fin heat sink model formed
by nearly 500 000 hexahedral elements.
solving the governing equations (2)–(5) for both domains
under the assumptions and boundary conditions mentioned
above. The finite-volume vertex-centered code was used for
getting the initial approach. A second-order upwind scheme
was used for discretizing the momentum equation, while the
Simple algorithm was used for the pressure–velocity coupling
[28]. The residuals were set to 1 × 10−6
∇U = 0
ρ U · ∇ U + P − = 0
∇ T = 0 for solid
ρc p U · ∇T − k f ∇ 2 T = 0 for fluid
2
B. Mesh Generation and Boundary Conditions
The mesh generated in the model is formed by approximately half a million hexahedral elements distributed
uniformly, except in the section near the fin walls where a
cell ratio of 1.025 is set. This mesh is adapted to the fin
shape. Fig. 2 shows a part of the solid domain and its mesh.
A mesh sensibility analysis was done in order to arrive at
the appropriate number of elements for achieving satisfactory
results.
The boundary conditions for the models are adjusted according to the interaction of the fluid with the surroundings.
A constant mass flow rate is considered at the fluid inlet
section of the model. It is important to highlight that the
velocity profile is not considered because of the effects of
developing the boundary layer at this inlet section (which
otherwise becomes reduced), affecting the goal of increasing
the heat transfer coefficient. A constant heat flux is imposed
on the bottom wall of the solid domain. Zero static pressure is
assumed at the fluid outlet section. Symmetry conditions for
both domains are considered at the symmetry walls. The walls
between the domains are set as interface conditions. The upper
wall of the channel and fins are set as adiabatic conditions.
C. Numerical Model
A commercial computational fluid dynamics (CFD) software (specifically, ANSYS Fluent) was used for numerically
(2)
(3)
(4)
(5)
where involves the viscous effects generated in the system
due to the relative velocity of the fluid through the channels [26]. The Reynolds number was in the range 100–200,
which was well within the laminar domain. The temperature
fields for both domains and the fluid velocity and pressure
variations are the results presented in this paper. The mesh
used in the numerical solution, as well as the accuracy of
the numerical results with theory, was studied as well. These
processes are described next.
D. Validation of the Numerical Model
A mesh sensibility analysis is carried out in order to
identify the appropriate number of elements to be used in
the numerical analysis. This analysis is based on solving
numerically the model subject to the boundary conditions and
operating parameters mentioned previously, beginning with an
initial number of elements. Once the solution is computed, the
number of elements in the mesh is increased through a specific
ratio (e.g., in the second approach, the number of elements
was doubled from the initial mesh). Thus, a second mesh
is obtained and used for computing the numerical solutions
considering the same operational parameters and boundary
conditions. Thereafter, the range of error is evaluated between
both solutions comparing the bulk temperature and pressure
RUBIO-JIMENEZ et al.: PERFORMANCE OF ONLINE AND OFFSET MICRO PIN-FIN HEAT SINKS
Analytically, these differences are given
q As
(8)
ṁcp
2Poμu L
ρu 2
P =
+K
.
(9)
2
Dh
2
The numerical and analytical fluid temperature differences
for online micro pin-fin heat sink model with operating
conditions of 1 ml/s of water and 100 W/cm2 of heat flux
are 24.07 and 23.91 K, respectively. The error is less than
0.7%. Moreover, the numerical and analytical pressure drops
in the microchannel heat sink subjected to the same operating
=
320
310
Temperature (K)
Mesh 1
Mesh 2
300
Mesh 3
290
Mesh 4
Mesh 5
280
270
0.0
0.2
0.4
0.6
0.8
1.0
z*
Fig. 3. Mesh sensibility analysis for the bulk temperature in the fluid along
the flow length.
35
30
Pressure (kPa)
drop along the channel length. This process is repeated until
the difference between the current and previous mesh becomes
negligible. In order to demonstrate this process, we present a
mesh sensibility analysis carried out on a straight microchannel heat sink.
The initial mesh was formed by 26 800 tetrahedral elements
(Mesh 1). Then, the second mesh was created by doubling
the number of elements in the initial mesh along the three
directions (x-, y-, and z-axis). This second mesh (Mesh 2) was
thus formed by 53 600 elements. After computing the numerical solution considering these two meshes, a comparison of
the bulk temperature and pressure drop was done. Figs. 3
and 4 present the curves of these variations. As can be seen
in these figures, the curves match well for a channel length
smaller than 0.4. After this point, the difference becomes
larger, mainly for the fluid temperature (∼ 1 K). Therefore, an
increase in the number of elements was considered. Another
three meshes were developed (Mesh 3, 4, and 5). The number
of elements was 145 000, 290 000, and 514 800 tetrahedral
elements, respectively. The numerical results computed with
these meshes are presented in Figs. 3 and 4.
According to this analysis, the error generated in the
hydraulic part by the mesh distribution is minimal when
Mesh 3 is used [since the difference between this one and the
next (Mesh 4) is ∼ 10 Pa]. However, this mesh significantly
affects the thermal part: making a similar comparison of the
bulk temperature between these two meshes, the difference
is ∼ 1.2 K. Moving on to the comparison between Mesh 4
and Mesh 5, it is observed that the difference computed
through each mesh is minimal for both thermal (∼ 0.1 K)
and hydrodynamic (5 Pa) aspects. Thus, the numerical analyses of the micro pin fin-heat sink models studied in this
paper are based on meshes with at least 300 000 tetrahedral
elements.
Momentum and energy balances were carried out for the
system. The fluid temperature difference in the micro pin-fin
heat sink model and the pressure drop in the microchannel
heat sink model were calculated from the numerical results and
compared with the analytical results. The average temperature
and pressure values at the fluid inlet and outlet sections were
determined by numerical surface integration
1
(6)
Tx,y d Ac |out −
Tx,y d Ac |in
T =
A c Ac
Ac
1
P =
(7)
Tx,y d Ac |out −
Px,y d Ac |in
A c Ac
Ac
89
25
Mesh 1
20
Mesh 2
15
Mesh 3
10
Mesh 4
5
Mesh 5
0
0.0
0.2
0.4
0.6
0.8
1.0
z*
Fig. 4.
length.
Mesh sensibility analysis for the fluid pressure drop along the flow
conditions are 1.65 and 1.70 kPa, respectively. The error is
around 3%. Although the hydrodynamic difference is slightly
large, the overall results show good agreement with the analytical results.
IV. R ESULTS
According to the nomenclature used in [22], the two micro
pin-fin heat sinks studied in this paper are designated as
follows:
1) MF-50 × 100 × 200 − 66 on-line (micro pin-fin configuration studied in [22]);
2) MF-50 × 100 × 200 − 66 offset (this paper).
Both configurations are described in Table I. Fig. 5 shows
the temperature profile along the dimensionless flow length
for the microchannel heat sink and the pin-fin heat sinks with
both online and offset fin configuration for 1 ml/s of water
and 100 W/cm2 of heat flux. The average temperature values
90
IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 3, NO. 1, JANUARY 2013
350
Microchannel heat sink
MF-50x100x200-66 On-line
MF-50x100x200-66 Offset
Fluid temperature
Temperature (K)
340
330
320
310
300
290
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Dimensionless flow length z/L
Fig. 5. Temperature profile at the bottom wall of the heat sink along the
dimensionless flow length for a microchannel heat sink and micro pin-fin heat
sinks with variable fin density for both online and offset fin configurations.
q = 100 W/cm2 and the flow rate is 1 ml/s.
are 334.39, 314.15, and 309.63 K, respectively. Clearly, the
offset micro pin-fin heat sink improves the heat dissipation and
reduces the IC chip surface temperature (∼ 7.5% compared to
a microchannel heat sink and 1.5% compared to the online
micro pin-fin heat sink). As mentioned in [23], the constant
development of the thermal boundary layer enhances the heat
transfer coefficient and reduces the overall thermal resistance.
In order to ratify this, Fig. 6 presents a comparison of the
contours of the velocity generated for both configurations
along the flow length for the operating conditions mentioned
above. The impact that the offset fin configuration generates
on the fluid motion, and thus on the heat dissipation, is clearly
observed in these contours. The fluid flowing through the
online fin configurations has a straight motion, and a large
part of the fluid passes through the straight gaps between the
fins. Thus, the interaction between the coolant and the walls
of the fins is significantly reduced, affecting the heat diffusion.
Moreover, when the fluid is moving through the offset fin
configuration, the thermal diffusion is enhanced due to the
fact that the fluid has better interaction with the heat transfer
areas along the flow length. Therefore, in Fig. 6 it is observed
that the offset fin configuration produces a more uniform
temperature profile and reduces the peaks/valleys caused by
the change of section (SI–SII and SII–SIII). The uniformity
of the bottom temperature impacts directly the temperature
gradients along the flow length.
The average gradient values for the online and offset fin
configurations are 1.63 °C/mm and 1.19 °C/mm, respectively.
In [22], for the online fin configuration, the average temperature gradient at the vicinity of the change of sections
(SI–SII and SII–SIII) was 2.7 °C/mm. The offset micro pinfin heat sink shows a significant reduction of these local
temperature gradients, mainly in the second zone. Temperature
gradient values of 2.29 °C/mm and 0.9 °C/mm are generated,
respectively. These results impact directly the performance and
lifetime of the IC chip.
Fig. 7 shows the thermal resistance and pressure drop
variations of the MF-50 × 100 × 200 − 66 micro pin-fin
heat sinks for online (solid line) and offset (dotted line)
configurations subject to different flow rates. Overall, the flow
rate reduces the thermal resistance of the systems; however,
the curves show a large negative slope when the flow rate
is less than 2 ml/s and a smaller decrease after this point.
In Fig. 7, the offset micro pin-fin heat sink presents almost
1.3 times lower thermal resistance than the online micro pinfin heat sink at the same operating conditions. This is a
clear manifestation of the passive thermal enhancement caused
by the offset arrangement. Otherwise, the pressure drops in
both fin configurations show an almost constant rise with the
increase of the flow rate; however, this energy lost is larger
in the offset micro pin-fin heat sink. The decrease of the
“apparent hydraulic diameter” that this fin configuration has
(the cross-sectional area is reduced by half at each section)
causes this pressure drop penalty. Overall, the offset micro
pin-fin heat sink generates almost 2.5 times larger pressure
drop than the online microcooling system.
According to these results, both micro pin-fin heat sinks
are capable of cooling the IC chips of 2016, since their
overall thermal resistances lie in the range required by the
ITRS (0.14–0.25 K/W) [2]. However, the operating conditions
required by these systems are different. The MF-50 × 100 ×
200 − 66 online requires up to 3.8 ml/s of water for obtaining
the lower value of thermal resistance (0.14 K/W), whereas the
offset fin configuration needs only 2 ml/s of coolant to get
the same thermal resistance condition. Moreover, the pressure
drops are similar in both systems (∼ 90 kPa) for this thermal
performance. Considering the pumping power that the system
requires to work as a parameter of comparison, the offset
fin configuration presents a better performance since it only
requires 0.18 W of pumping energy to get a thermal resistance
of 0.14 K/W. The online fin configuration requires almost
twice the pumping energy (0.35 W) to obtain a similar thermal
performance. In summary, the offset micro pin-fin heat sink
is more reliable for near-future applications because of the
following reasons:
1) it generates more uniform temperature profiles on the IC
chip (overall temperature gradient of 1.2 °C/W);
2) the maximum heat expected by the IC chip of 2016
(288W) can be dissipated appropriately with only
0.18 W of energy to be supplied into the system.
Table II shows a comparison between different cooling
devices found in the literature. The online and offset micro
pin-fin heat sinks working with the appropriate operating
conditions to get a thermal resistance of 0.14 have been
included in this comparative table. The water flow rate for each
configuration is 3.8 and 2.0 ml/s, respectively. Clearly, these
configurations can dissipate appropriately the heat generation
expected by the IC chips of 2016. The offset fin configuration requires a lower amount of energy than the online
arrangements and also the other heat sinks listed in Table II
(e.g., conventional microchannel heat sink and the Colgan et al. heat sink). In this line, the micro pin-fin heat
sinks are only topped by cooling systems based on boiling
cooling [29].
Moreover, a large part of our work is aimed at generating
cooling systems with thermal resistances lower than 0.1 K/W.
RUBIO-JIMENEZ et al.: PERFORMANCE OF ONLINE AND OFFSET MICRO PIN-FIN HEAT SINKS
91
(a)
(b)
Fig. 6. Contours of velocity at the middle of the channel section (y = 300 μm) for (a) online and (b) offset microchannel pin-fin heat sink with variable
fin density.
TABLE II
P ERFORMANCE OF M ICRO H EAT S INKS U SING L IQUID C OOLING T ECHNOLOGIES
Author
Description
P (kPa)
R (K/W)
Pump
Power (W)
∗
qmax
(W/cm2 )
Tuckerman and Pease (1981) [3]
Rectangular microchannel heat sink
207
0.090
2.3
> 650
Knight et al. (1992) [10]
Rectangular microchannel heat sink using turbulent flow
207
0.056
> 10.0
> 1000
Gillot et al. (2000) [11]
Rectangular microchannel heat sink for multichips modules
180
0.092
∼ 47.0
> 650
Peles et al. (2005) [16]
Micro heat sink with circular staggered pin fins
203
0.039
–
> 1500
Colgan et al. (2007) [18]
Micro heat sink with “semielliptical” staggered pin fins
< 35
0.105
< 0.9
> 500
Huscin and Kim (2008) [12]
Optimized microchannel heat sink
–
0.081
–
> 700
Daguenet–Frick et al. (2010) [29]
Radial heat sink with boiling fluid
–
0.080
0.05
750
Escher et al. (2010) [20]
Heat sink formed by manifold channels
< 10
0.087
∼ 0.15
> 680
Rubio–Jimenez et al. [22] (2011)
MF-50 × 100 × 200 − 66 on-line
90
0.14
0.34
430
This paper (2011)
MF-50 × 100 × 200 − 66 offset
86
0.14
0.18
430
*Considering a 1 × 1 cm IC chip with maximum design temperature of 85 °C and an ambient temperature of 25 °C.
0.30
300
2.5
250
2
Thermal resistance (K/W)
0.20
200
0.15
150
0.10
100
0.05
50
0
0.00
1
2
3
4
5
Flow rate (mL/s)
Fig. 7. Thermal resistance and pressure drop variation with different flow
rates for micro pin-fin heat sinks with variable fin density and both online
and offset fin configuration. q = 100 W/cm2 .
Considering this thermal performance in both micro pin-fin
heat sinks, an important increase of the pressure drop is
clearly observed in Fig. 7, mainly for the offset fin configuration. However, when the requirements of pumping power
Pumping Power (W)
MF-50x100x200-66 offset
0.25
Pressure drop (kPa)
MF-50x100x200-66 on-line
1.5
1
0.5
0
[3]
[18]
[22]
Current work
Fig. 8. Thermal resistance and pressure drop variation with different flow
rates for micro pin fin heat sinks with variable fin density and both online
and offset fin configuration. q = 100 W/cm2 .
are compared, interesting results are observed. In order to get
an overall comparison, Fig. 8 presents the pumping power
required by a conventional microchannel heat sink [3], a robust
micro cooling system [18], and both online and offset micro
pin-fin heat sink with variable fin density. All these cooling
systems are operating at specific conditions in order to obtain
a thermal resistance of 0.1 K/W. A very significant reduction
92
IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 3, NO. 1, JANUARY 2013
in the input energy is clearly observed with these last three
cooling systems compared to conventional microchannel heat
sinks (∼ 100%). Also, the robust cooling system and the offset
micro pin-fin configuration present similar values of pumping
power (∼ 0.9 W). Therefore, the assertion that the offset micro
pin-fin heat sinks are a very good alternative for cooling near
future IC chips is ratified. Furthermore, this cooling system
presents an easy integration in 2-D and 3-D IC chip packaging.
Following this comparison of the pumping power, the offset
micro pin-fin heat sink presents a better performance when
the flow rate is less than 5 ml/s (thermal resistance around
0.1 K/W and pumping power less than 0.45 W) than the
online fin configuration. Apart from this thermal resistance
value, both configurations have almost similar overall pumping
power requirements. Thus, novel and reliable micro cooling
systems capable of generating uniform junction temperature
should be looked for, for obtaining thermal resistances smaller
than 0.1 K/W.
V. F UTURE W ORK
The analysis presented in this paper clearly identifies the
benefits of using an offset micro pin-fin heat sink with variable
fin density. We are continuing with the efforts to experimentally validate these findings and also for 3-D IC packaging
integration.
VI. C ONCLUSION
An offset micro pin-fin heat sink configuration with variable
fin density was studied numerically in this paper. This heat
sink is based on Configuration III analyzed in [22]. The
results show that this offset fin configuration improves the
heat dissipation and reduces at least by 1.3 times the system
thermal resistance. This represents 1.5% and 7.5% reduction
in the overall bottom wall temperature compared to the
online micro pin-fin heat sink and a microchannel heat sink,
respectively. Furthermore, the offset fin configuration improves
the temperature uniformity since the peaks/valleys that the
online fin configuration presents at the change-of-section zones
are smoothed. This has an important impact in the overall
temperature gradient, which is reduced to 1.19 °C/W.
Hydraulically, the pressure drop rises significantly when the
offset fin configuration is used in the micro pin-fin heat sink.
However, the amount of coolant that this offset fin system
requires to obtain a specific thermal resistance is less than that
required by the online fin configuration due to the heat transfer
enhancement. The pumping power is a very good parameter
for comparison. According to this parameter, these micro pinfin heat sinks have an almost similar overall performance as
the Colgan et al. cooling device with a thermal resistance of
0.1 K/W.
The results show that the offset micro pin-fin heat sink is
highly reliable for cooling systems with thermal resistances
larger than 0.1 K/W since the pumping power is less than the
pumping energy required by the online fin configurations. For
cooling systems that require thermal resistances much lower
than 0.1 K/W, this micro pin-fin configuration with variable
fin density is not recommended because of their large increase
of the pressure drop.
R EFERENCES
[1] Intel Xeon Processor E7-8800/4800/2800 Product Families, Intel Corporation, Santa Clara, CA, 2010.
[2] S. P. Gurrum, S. K. Suman, Y. K. Joshi, and A. G. Federov, “Thermal
issues in next-generation integrated circuits,” IEEE Trans. Device Mater.
Rel., vol. 4, no. 4, pp. 709–714, Dec. 2004.
[3] D. B. Tuckerman and R. F. W. Pease, “High-performance heat sinking
for VLSI,” IEEE Trans. Electron. Devices, vol. 2, no. 5, pp. 126–129,
May 1981.
[4] M. T. Bohr, “Interconnecting scaling-the real limiter to high performance
ULSI,” in Proc. IEEE IEDM, Dec. 1995, pp. 241–244.
[5] P. Lall, M. G. Pecht, and E. B. Hakin, Influence of Temperature on
Microelectronic and Reliability. West Palm Beach, FL: CRC Press, 1997.
[6] G. O. Workman, J. G. Fossum, S. Krishnan, and M. M. Pelella, “Physical
modeling of temperature dependences of SOI CMOS devices and circuits
including self-heating,” IEEE Trans. Electron. Devices, vol. 45, no. 4,
pp. 125–133, Jan. 1998.
[7] S. G. Kandlikar and W. J. Grande, “Evaluation of single phase flow in
microchannels for high flux chip cooling-thermohydraulic performance
enhancement and fabrication technology,” Heat Trans. Eng., vol. 25,
no. 8, pp. 5–16, 2004.
[8] M. Renksizbulut and H. Niazmand, “Laminar flow and heat transfer
in the entrance region of trapezoidal channels with constant wall
temperature,” J. Heat Transfer, vol. 128, no. 1, pp. 63–74, 2006.
[9] J. P. McHale and S. V. Garimella, “Heat transfer in trapezoidal
microchannels of various aspect ratios,” Int. J. Heat Mass Transfer,
vol. 53, nos. 1–3, pp. 365–375, Jan. 2010.
[10] R. W. Knight, D. J. Hall, J. S. Goodling, and R. C. Jaeger, “Heat sink
optimization with application to microchannels,” IEEE Trans. Compon.
Hybrids Manuf. Technol., vol. 15, no. 5, pp. 832–842, Oct. 1992.
[11] C. Gillot, C. Schacffer, and A. Bricard, “Integrated micro heat sink for
power multichip module,” IEEE Trans. Ind. Appl., vol. 36, no. 1, pp.
217–221, Jan.–Feb. 2000.
[12] A. Husain and K. Y. Kim, “Optimization of a microchannel heat
sink with temperature dependent fluid properties,” Appl. Thermal Eng.,
vol. 28, nos. 8–9, pp. 1101–1107, Jun. 2008.
[13] D. V. Pence, “Reduced pumping power and wall temperature in
microchannel heat sinks with fractal-like branching channel networks,”
Microscale Thermophys. Eng., vol. 6, no. 4, pp. 319–330, 2002.
[14] C. Biserni, L. A. O. Rocha, G. Stanescu, and E. Lorenzini, “Constructal
H-shaped cavities according to Bejan’s theory,” Int. J. Heat Mass
Transfer, vol. 50, nos. 11–12, pp. 2132–2138, Jun. 2007.
[15] X.-Q. Wang, P. Xu, A. S. Mujumdar, and C. Yap, “Flow and thermal
characteristics of offset branching networks,” Int. J. Thermal Sci.,
vol. 49, no. 2, pp. 272–280, Feb. 2010.
[16] Y. Peles, A. Koşar, C. Mishra, C. J. Kuo, and B. Schneider, “Forced
convective heat transfer across a pin fin micro heat sink,” Int. J. Heat
Mass Transfer, vol. 48, no. 17, pp. 3615–3627, Aug. 2005.
[17] E. G. Colgan, B. Furman, M. Gaynes, N. LaBianca, J. H. Magerlein,
R. Polastre, R. Bezama, K. Marston, and R. Schmidt, “High performance
and subambient silicon microchannel cooling,” J. Heat Mass Transfer,
vol. 129, no. 8, pp. 1046–1051, Aug. 2007.
[18] E. G. Colgan, B. Furman, M. Gaynes, W. S. Graham, N. LaBianca,
J. H. Magerlein, R. Polastre, M. B. Rothwell, R. J. Benzema, R. Choudhary, K. C. Marston, H. Toy, J. Walkil, J. A. Zitz, and R. R. Schmidt,
“A practical implementation of silicon microchannel coolers for high
power chips,” IEEE Trans. Compon. Packag. Technol., vol. 30, no. 2,
pp. 218–225, Jun. 2007.
[19] R. Wälchli, T. Brunschwiler, B. Michel, and D. Poulikakos, “Combined
local microchannel-scale CFD modeling and global chip scale network
modeling for electronic cooling designs,” Int. J. Heat Mass Transfer,
vol. 53, nos. 5–6, pp. 1004–1014, Feb. 2010.
[20] W. Escher, B. Michel, and D. Poulikakos, “A novel high performance,
ultrathin, heat sink for electronics,” Int. J. Heat Fluid Flow, vol. 31,
no. 4, pp. 586–598, Aug. 2010.
[21] G. Hetsroni, A. Mosyak, Z. Segal, and G. Ziskind, “A uniform temperature heat sink for cooling of electronic devices,” Int. J. Heat Mass
Transfer, vol. 45, no. 16, pp. 3275–3286, 2002.
[22] C. A. Rubio-Jimenez, S. G. Kandlikar, and A. Hernandez-Guerrero,
“Numerical analysis of novel micro pin fin heat sink with variable fin
density,” IEEE Trans. Compon. Packag. Manuf. Technol., vol. 2, no. 5,
pp. 825–833, May 2012.
[23] M. E. Steinke and S. G. Kandlikar, “Single-phase heat transfer enhancement techniques in microchannel and minichannel flows,” in Proc. 2nd
Int. Conf. Microchannels Minichannels, Jun. 2004, pp. 141–148.
RUBIO-JIMENEZ et al.: PERFORMANCE OF ONLINE AND OFFSET MICRO PIN-FIN HEAT SINKS
93
[24] M. E. Steinke and S. G. Kandlikar, “Single-phase liquid heat
transfer in plain and enhanced microchannels,” in Proc. 4th Int.
Conf. Nanochannels, Microchannels Minichannels, Jun. 2006, pp.
943–951.
[25] M. E. Steinke, S. G. Kandlikar, J. H. Magerlein, E. G. Colgan, and
A. D. Raisanen, “Development of an experimental facility for investigating single-phase liquid flow in microchannels,” Heat Trans. Eng.,
vol. 27, no. 4, pp. 41–52, Aug. 2006.
[26] K. C. Toh, X. Y. Chen, and J. C. Chai, “Numerical computation of fluid
flow and heat transfer in microchannels,” Appl. Thermal Eng., vol. 25,
pp. 1472–1487, Aug. 2005.
[27] Z. Li, X. Huai, Y. Tao, and H. Chen, “Effects of thermal property
variations on the liquid flow and heat transfer in microchannel heat
sinks,” Appl. Thermal Eng., vol. 27, nos. 17–18, pp. 2803–2814, Dec.
2007.
[28] Fluent 6.2 User’s Guide, Fluent, Inc., Lebanon, NH, 2005.
[29] X. Daguenet-Frick, J. Bonjour, and R. Revellin, “Constructal microchannel network for flow boiling in a disc-shaped body,” IEEE Trans.
Compon. Packag. Technol., vol. 33, no. 1, pp. 115–126, Mar.
2010.
Satish G. Kandlikar received the Ph.D. degree
from the Indian Institute of Technology Bombay (IIT
Bombay), Mumbai, India, in 1975.
He was a Faculty Member with IIT Bombay until
1980. He is currently the Gleason Professor of
mechanical engineering with the Rochester Institute
of Technology, Rochester, NY. He is involved in
research on advanced single-phase and two-phase
heat exchangers incorporating smooth, rough, and
enhanced microchannels. He is currently involved
in a project on fuel cell water management under
freezing conditions, which is sponsored by the Department of Energy. He has
authored or co-authored more than 180 journal and conference papers. His
current research interests include heat transfer and fluid-flow phenomena in
microchannels and minichannels.
Dr. Kandlikar was the recipient of the RIT’s Eisenhart Outstanding Teaching
Award in 1997 and the Trustees Outstanding Scholarship Award in 2006.
He is a fellow of the ASME, an Associate Editor of a number of journals,
including the ASME Journal of Heat Transfer, and an Executive Editor of the
Heat Exchanger Design Handbook (Begell House).
Carlos A. Rubio-Jimenez received the B.S. and
M.S. degrees in mechanical engineering from the
University of Guanajuato, Guanajuato, Mexico,
in 2007 and 2008, respectively, where he is
currently pursuing the Ph.D. degree in mechanical
engineering, specializing in the design of reliable
microheat sinks for cooling high-tech electronic
devices.
He is currently a CONACYT Scholar. He has
authored or co-authored papers in refereed journals
and conferences.
Mr. Rubio-Jimenez was a recipient of the Fulbright Garcia-Robles
Scholarship for research at the Rochester Institute of Technology.
Abel Hernandez-Guerrero received the B.S. degree
from the University of Guanajuato, Guanajuato,
Mexico, and the Ph.D. and M.S. degrees from Oregon State University, Corvallis.
He is currently with the University of Guanajuato.
He has authored or co-authored more than 230
scientific papers in journals and international energy
conferences.
Dr. Hernandez-Guerrero was a recipient of the
2001 ASME Student Section Advisor Award and the
2006 ASME Johnson Medal. He was the President of
the Mexican Society of Mechanical Engineering from 2000 to 2002, the Chair
of the ASME Student Sections Committee from 2006 to 2010, the Chair of the
ASME Advanced Energy Systems Division from 2007 to 2008, and has been a
member of the Mexican System of Researchers (National Top Honors Society)
since 1992. He is an Associate Editor of the ASME International Journal of
Fuel Cell Technology and an Editorial Board Member for many journals,
including the International Journal of Exergy, the International Journal of
Energy Research, and the International Journal of Thermodynamics. He is a
fellow of the ASME.