Journal of Electronics Cooling and Thermal Control, 2013, 3, 85-93
http://dx.doi.org/10.4236/jectc.2013.33010 Published Online September 2013 (http://www.scirp.org/journal/jectc)
Investigation of Thermal Characterization of a Thermally
Enhanced FC-PBGA Assembly
C. F. Lin1, G. H. Wu1*, S. H. Ju2
1Department of Mechanical Engineering, National Cheng-Kung University, Tainan, Taiwan
2Department of Civil Engineering, National Cheng-Kung University, Tainan, Taiwan
Received January 1, 2013; revised February 1, 2013; accepted February 10, 2013
Copyright © 2013 C. F. Lin et al. This is an open access article distributed under the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ABSTRACT
In this paper, three-dimensional finite element analysis using the commercial ANSYS software is performed to study
the thermal performance of a thermally enhanced FC-PBGA (flip-chip plastic ball grid array) assembly in both natural
and forced convection environments. The thermally enhanced FC-PBGA assembly is a basic FC-PBGA assembly with
a lid attached on top, after which an ex truded-fin heatsink is attached on the top of the lid. The finite element model is
complete enough to include key elements such as bumps, solder balls, substrate, printed circuit board, extruded-fin
heatsink, lid, vias, TIM1 (thermal interface material 1), TIM2 (thermal interface material 2), lid-substrate adhesive and
ground planes for both signal an d pow er. Temperature fields are simulated and presented for several package configura-
tions. Thermal resistance is calculated to characterize and compare the thermal performance by considering alternative
design parameters of the polymer-based materials and the thermal enhancement components. The polymer-based mate-
rials include underfill, TIM1, TIM2, lid-substrate adhesive and substrate core material. The specific thermal enhance-
ment components are the extruded-fin heatsink and the lid.
Keywords: Polymer-Based Materials; Flip-Chip Packaging; Finite Element Methods
1. Introduction
Due to increasing demand for high density, high I/O,
high performance electrical application and flip-chip pla-
stic ball grid array (FC-PBGA) packaging are rapidly be-
coming the package of choice. FC-PBGA technology
offers many advantages over the popular PBGA. Com-
monly stated advantages of FC-PBGA are higher pack-
aging density, shorter leads, lower inductance, better
noise control, smaller device footprint and lower package
profile [1]. A typical basic FC-PBGA assembly (i.e.
without thermal enhancement) is shown schematically in
Figure 1. Heat dissipated from the chip follows two ma-
jor paths. One path goes from the top of the chip to the
ambient environment. The other path goes from the chip
to the substrate to the solder balls to the PCB and finally
to the ambient.
Due to the rapid increase of power and packaging den-
sities, thermal issues have become critical factors for
reliability. Various thermal enhancements have been de-
veloped for the basic FC-PBGA package to improve
thermal performance. Gugliermermetti and Grignaffini [2]
analyzed theoretically the thermal performance of paral-
lel stacks of in-line plate fin heat sinks. They reported
that a fin efficiency parameter, like that used for constant
cross-section fin in an isothermal fluid flow, can be in-
troduced to express both the local and overall heat trans-
fer balances. Many numerical and experimental studies
in recent years have focused on thermal enhancement
techniques for FC-PBGA packages. Joiner [3] numeri-
cally and experimentally compared the thermal perform-
ance between a FC-CBGA with a heatsink and a FC-
PBGA with a heatsink. He reported that both FC-PBGA
and FC-CBGA packages can be used for relatively high
power applications. If a heatsink is implemented, then
both package types yield comparable thermal perform-
ance. Chen el al. [4] proposed a finite element method-
ology to predict the th ermal resistance of both FC-PBGA
with a bare die and FC-PBGA with a metal cap. They
reported that the material of the metal cap slightly influ-
enced the thermal resistance when the heat dissipation
was saturated through the metal cap. Luo el al. [5] ana-
lyzed by finite element methodology, a whole mobile
phone system, by observing the system response to one
*Corresponding a uthor.
C
opyright © 2013 SciRes. JECTC

C. F. LIN ET AL.
86
source. They reported that material with thermal conduc-
tivity can be added between the chip and the bottom case
to enhance the thermal management under natural con-
vection. Menon el al. [6] studied the thermal perform-
ance of the package on package structure in natural en-
vironments using finite element modeling methodology
and analyzing the effect of die power on the logic and
memory dies.
This present paper considers three-dimensional finite
element simulation of a thermally enhanced FC-PBGA
assembly in both natural and forced air convection envi-
ronments, using the commercial software ANSYS [7].
The thermally enhanced FC-PBGA assembly is a basic
FC-PBGA assembly with a lid attached on top, after
which an extruded-fin heatsink is attached on the top of
the lid, as seen in Figure 2 and denoted in the following
as an FC-PBGA/lid/heatsink assembly. The finite ele-
ment model is complete enough to include key elements
such as bumps, solder balls, substrate, printed circuit
board, extruded-fin heatsink, lid, vias, TIM1, TIM2, lid-
substrate adhesive and ground planes for both signal and
power. Temperature fields are simulated and presented
for several package configurations. Thermal resistance is
calculated to characterize and compare the thermal per-
formance by considering alternative design parameters of
the polymer-based materials and the thermal enhance-
ment components. The polymer-based materials include
underfill, TIM1, TIM2, lid-substrate adhesive and sub-
strate core material. The specific thermal enhancement
components are the extruded-fin heatsink and the lid.
2. Finite Element Modeling
2.1. Description of FC-PBGA/Lid/Heatsink
Assembly
In the following we consider an FC-PBGA/lid/heatsink
assembly, as seen in Figure 2. This assembly has a sili-
con chip size of 7.4 × 5.4 × 0.74 mm, with 256 solder
bumps distributed over the base. The chip is first bonded
to an organic substrate with solder bump interconnects.
The space between chip and substrate is then filled with
epoxy-based underfill material. The purpose of the un-
derfill is to release the thermo-mechanical stress of the
solder bumps due to coefficient of thermal expansion
mismatch between the silicon chip and the organic sub-
strate. The 25 × 25 × 1.0 mm substrate contains six bur-
ied metal layers and also a 0.05 mm solder mask layer o n
both the upper and lower surfaces. This FC-PBGA
package is mounted on a four-layer (2s2p) PCB with
solder balls 0.77 mm in diameter and 0.45mm in height.
The solder balls are patterned as a depopulated array in
which a 9 × 9 array of central balls is removed from a
full array of 19 × 19 solder balls. The 100 × 100 × 1.52
mm PCB also contains a 0.05 mm solder mask layer on
both top and botto m surfaces and is exposed to the air in
a horizontal package-up position, as seen in Figure 2.
The core material of the substrate is FR-5 laminate, while
the core material of the PCB is FR-4 laminate. Vias con-
nect various layers within the substrate and the PCB for
reasons of electrical and/or thermal conduction and have
a much higher thermal conductivity value (300 - 400
W/m K

) than the organic core material (0.2 - 0.4
W/m K

). Heat is conducted by the vias from the chip
through the solder balls and into the PCB in the out of
plane (Z) direction. The present finite element model
assumes that each solder ball or solder bump is associ-
ated with a through-via in the substate and in the PCB.
Vias in the substrate are solid copper with a diameter of
0.05 mm. Vias in the PCB are hollow copper with outer
and inner diameters of 0.25 and 0.20 mm, respectively.
An aluminum lid is attached on top of the substrate,
covering the chip. The contact b etween any two surfaces
Figure 1. A typical basic FC-PBGA assembly.
Figure 2. Cross section of the FC-PBGA/lid/heatsink assembly.

C. F. LIN ET AL. 87
creates a resistance that may greatly reduce heat conduc-
tion efficiency across the contact surface unless a thermal
interface material (TIM) is used. A soft pad is applied
between the chip and the lid. The soft pad is silicone
elastomers filled with high conductivity particles such as
boron nitride and has a thermal conductivity of 3.7
and denoted in the following as TIM1. The lid
has a typical thickness of 0.5 mm and is bonded to the
substrate with a layer of B-stage epoxy 0.15 mm thick.
The epoxy resin, deno ted in the fo llowin g as lid-sub strate
adhesive, contains alumina filler and has a net thermal
conductivity estimated to be 1.5 .
W/m K
W/m K
Attached to the top of this aluminum lid is a 24.7 ×
27.9 × 15.24 mm aluminum extruded-fin heatsink with
eight extruded fins, as seen in Figure 3. A 50 m thick
layer of thermally conductive epoxy (thermal conduc-
tiveity = ) is used to connect the lid and the
heatsink base and denoted in th e following as TIM2. Ta-
ble 1 displays the assembly dimensions and material
conductivity.
W/m K 
2.2. Governing Equations and Boundary
Conditions
The heat diffusion equation is the governing equation for
the temperature of the items in the FC-PBGA/lid/ heat-
sink assembly. For steady state conditions, the heat dif-
fusion equation for this assembly can be written as
0
xyz
TTT
kkk
xxyyzz

  
 


 

 


g (1)
where T is the temperature of the item,
g
 is the heat
dissipation rate per unit volume in the chip and
x
k,
y
k,
z
k are thermal conductivity in the x, y and z directions.
The boundary conditions surrounding the assembly are
heat convection and radiation to the ambient air. The heat
Figure 3. Drawing of the extrude d-fin heatsink.
Table 1. Assembly dimensions and material conductivities
[8].
Item Dimension
(mm) Material Conductivity (W/m·K)
Chip 7.5 × 5.4 × 0.74 Silicon 109.0
Solder Bump0.127 diameter and
0.0762 height 90/10 Sn/Pb 36
Underfill Epoxy with
silica filler 0.6
Substrate 25 × 25 × 1.0
FR-5 core
material,
4 metal layers,
solder mask on
top and bottom
0.35 for FR-5,
398 for Cu,
0.245 for mask
Solder ball0.77 diameter and
0.45 height 40/60 Sn/Pb 50
PCB
100 × 100 × 1.52,
one top Cu trace
layer (2oz) and
two internal Cu
planes (1oz), with
thermal vias
FR-4 core
material,
4 metal layers,
solder mask on
top and bottom
0.35 for FR-4,
398 for Cu,
0.245 for mask
Extruded-fin
heatsink 24.7 × 27.9 × 15.24 Aluminum 226
loss to the ambient or outward heat flux through the ex-
posed surface is

 
crs
qhhTT

  (2)
in which
s
T is the assembly surface temperature, T

is
the ambient air temperature, r is the radiation heat
transfer coefficient and is the convection heat trans-
fer coefficient.
h
c
h
2.3. Finite Element Model
Under forced convection, the convection heat transfer
coefficients c for the FC-PBGA/lid/heatsink assembly
are calculated using the isothermal flat plate correlation
equations previously applied by Mertol [9] as follows.
h
For the basic FC-PBGA assembly external surfaces,
3.786
c
hVL, (3)
where V is the airflow velocity in m/s and L is the total
length in the flow direction in meters.
For the heatsink external surface,
4.37
c
hVL, (4)
where again is airflow velocity in m/s and L is the
average fin length in meters.
V
Under free convection, the convection heat transfer
coefficient c for the FC-PBGA/lid/heatsink assembly
is calculated as suggested by Ellison [10] for small de-
vices encountered in the electronics industry as
h

C. F. LIN ET AL.
88
n
sa
cch
T
hfL

 



, (5)
where
s
a is the temperature difference in ˚C be-
tween the surface and the ambient air,
T 
f
and are
empirical factors and ch is the package characteristic
length in meters. The constants
n
L
f
and are given as
and for a horizontal plate facing
upward, and for a horizontal plate
facing downward, and and for a
vertical plate. In the equation, ch is the characteristic
length in meters. For horizontal plates,
n
n
0.83 f
f0.33 n
f

n
0.415

0.33
09
L
1.0.35
2
ch where and are the re-
spective width and length of a plate. For vertical plates,
, where
LW
ch
LH
LWLWL
H
is the vertical height of the plate.
When radiating, the radiation heat transfer coefficient
is calculated as
r
h



22
rss
hBfeTTTT

 

, (6)
where B is the Boltzmann constant

82
5.6710W mKB 

4
, e is the surface emissivity,
f is the radiative view factor,
s
T is the assembly surface
temperature and is the ambient air temperature.
T
The FC-PBGA/lid/heatsink assembly detailed above is
represented by a three-dimensional finite element model
using ANSYS finite element code. Due to the symmetric
nature of both the heat problem and the assembly, only
one quarter of the assembly is modeled. In the finite ele-
ment analysis, an ambient temperature of 50˚C, and a
uniform chip power dissipation of 3 W is assumed. The
model is complete enough to include key packaging ele-
ments such as bumps, solder balls, substrate, PCB, ex-
truded-fin heatsink, lid, vias, TIM1, TIM2, lid-substrate
adhesive, and ground planes for both signal and power.
2.4. Grid Refinement
Choice of node (or element) density in finite element
solution procedures has a strong effect on the quality and
computational cost. In this paper, variable grid spacing is
used so that a finer spacing occurs in areas of relatively
large gradients. Moreover, a grid dependency study is
performed.
Three finite-element meshes (labeled M1, M2 and M3)
with mesh densities of 168,850, 231,392 and 323,136
elements, respectively, are used to calculate the thermal
resistance of the FC-PBGA/lid/heatsink assembly. The
calculated results are shown in Figure 4 for the cases of
air speeds from 0 to 3 m/sec. Solu tions using meshes M2
and M3 are virtually identical, indicating that M2 is suf-
ficient to obtain reasonable solutions. Mesh M1 seems
too coarse to produce adequate solutions. Since the in-
termediate density mesh M2 requires significantly less
computational effort than the highest density M3, the M2
mesh shown in Figure 5 is chosen as a good trade-off
between accuracy and computational cost.
3. Results and Discussion
The ANSYS three-dimensional finite element analysis of
the thermal characterization in the presented assemblies
yielded the following results.
3.1. Temperature Analysis
The adding of an aluminum lid/heatsink significantly in-
fluences the thermal performance of the assembly. The
temperature contours of the FC-PBGA assembly with-
out/with an aluminum lid/heatsink for an airflow of zero
are shown in Figures 6 and 7, respectively. In Figure 6,
the maximum temperature occurs in the chip. With the
addition of the lid/heatsink, as shown in Figure 7, the
Figure 4. Relation between thermal resistance and airflow
speed for FC-PBGA/lid/heatsink assembly using mesh M1,
M2 and M3.
Figure 5. Finite element mesh M2 used in simulations for
the FC-PBGA/lid/heatsink assembly.

C. F. LIN ET AL. 89
Figure 6. Temperature contours for the basic FC-PBGA as-
sembly at airflow = zero.
Figure 7. Temperature contours for the FC-PBGA/lid/heat-
sink assembly at airflow = zero.
maximum temperature still occurs in the chip. However,
the maximum temperatures drop significantly. The tem-
perature contours of the extruded-fin heatsink are shown
in Figure 8, where the maximum temperature is found at
the bottom center (remembering that the figure shows
only one quarter of the extruded-fin heatsink since the
total assembly is symmetrical), while the minimum tem-
perature occurs at the outside corner of the most distant
fin’s upper edge. The temperature distribution in the ex-
truded-fin heatsink is quite uniform due to the very high
thermal conductivity of the aluminum material.
3.2. Thermal Resistance
As commonly done, thermal resistance

j
a
R is used to
characterize the thermal performance of an assembly,
defined as

j
ja
TT
RP


 (7)
where
j
T is the predicted maximum temperature for the
assembly, P the device power and is the ambient air
temperature. T
Figure 8. Temperature distribution for the extruded-fin
heatsink at airflow = zero.
For comparison of thermal performance, four different
FC-PBGA assembly versions are considered in this sec-
tion and described as follows:
1) The basic FC-PBGA assembly (or the basic FC-
PBGA model), defined as the FC-PBGA assembly with-
out thermal enhancement shown in Figure 1;
2) Basic FC-PBGA model + lid, defined as the basic
assembly with a lid attached on the top;
3) Basic FC-PBGA model + heatsink, defined as the
basic assembly with a extruded-fin heatsink attached on
the top;
4) Basic FC-PBGA model + lid/heatsink, defined as
the basic assembly with a lid attached on top, after which
extruded-fin heatsink is attached on the top of the lid (i.e.
FC-PBGA/lid/heatsink shown in Figure 2).
Figure 9 shows the thermal performance of the above
four cases for airflows ranging from 0 to 3 m/sec. Addi-
tion of the lid to the basic FC-PBGA model (i.e. basic
FC-PBGA model + lid) causes significant improvement
in thermal performance. The high thermal conductivity
of the aluminum lid results in a larger surface area with a
temperature high enough for good convective cooling.
Addition of only the extruded-fin heatsink to the basic
model (i.e. basic FC-PBGA model + heatsink) causes
large improvement in thermal performance relative to the
basic FC-PBGA model due to the large increase in heat
transfer area. Addition of the lid between the basic FC-
PBGA model and the extruded-fin heatsink (i.e. basic
FC-PBGA model + lid/heatsink or FC-PBGA/lid/heatsink)
results in a small further improvement in thermal per-
formance. This small further increase can be attributed to
the increased cooling area made available by the sides of
the lid. Clearly, the increased cooling area presented by
the sides of lid is small relative to the large total cooling
area of the extruded-fin heatsink. The role of the lid is
interesting. If a extruded-fin heatsink is not applied to the
basic model, then addition of the lid results in significant

C. F. LIN ET AL.
90
improvement of thermal performance. On the other hand,
if a extruded-fin heatsink has been applied to the basic
FC-PBGA model, then the further addition of the lid
does not result in significant further improvement of
thermal performance.
Inspection of Figure 9 shows that, at an airflow of
zero, the predicted
j
a of the FC-PBGA/lid/heatsink is
about 57.0% lower than the basic PBGA assembly. If
progressively faster air flow is applied to the FC-PBGA/
lid/heatsink, then the improvement of the FC-PBGA/lid/
heatsink relative to the basic FC-PBGA assembly at air-
flow = zero increases with increasing air speed, reaching
a maximum improvement of 84.9% at this study’s
maximum airspeed of 3 m/s.
R
3.3. Effects of Design Parameters
Thermal performance is analyzed by considering various
design parameters of the polymer-based materials and the
thermal enhancement components. The polymer-based
materials include underfill, TIM1, TIM2, lid-substrate
adhesive and substrate core material. The specific ther-
mal enhancement components are the lid and the ex-
truded-fin heatsink. Effects of the design parameters are
studied by varying the parameter values by 30% for all
material conductivities and by 15% for all material di-
mensions in both natural and forced convection
environments. Simulation results are shown in
Figures 10-13.

0v


1v
Figure 9. Relation between thermal resistance and airflow
speed for four different assembly configur ations.
Figure 10. Relation between thermal resistance and thermal
conductivity in natural convection condition

0v

.
Figure 11. Relation between thermal resistance and thermal
conductivity in forced convection condition .

1v
3.3.1. Effect of Extruded-Fin Heatsink
Extruded-fin heatsink design has significant effect on
thermal performance of the assembly. The effects of ex-
truded-fin heatsink conductivity with regard to thermal
performance of the assembly can be found in Figures 10
and 11 for both natural and forced convection environ-
ments, respectively. Increased extruded-fin heatsink
conductivity increases the thermal performance of the
assembly since higher thermal conductivity improves the
heat spreading effect in the heatsink. Figure 14 further
shows the thermal resistance
j
a vs the thermal con-
ductivity of the extruded-fin heatsink for airflows rang-
ing from 0 to 3 m/sec. The results show little improve-
ment in performance at thermal conductivities above
R
200 WmK

. Thus, little advantage can be obtained by
using copper (thermal conductivity = 393 W/m·K) rela-
tive to aluminum (thermal conductivity = 226 W/m·K).

C. F. LIN ET AL. 91
Figure 12. Relation between thermal resistance and dimen-
sions in natural convection condition .  0v
Figure 13. Relation between thermal resistance and dimen-
sions in forced convection condition .

1v
Figure 14. Relation between thermal resistance and extru-
ded-fin heatsink conductivity.
This observation is of interest to designers with regard to
material selection.
Effects of fin length, fin thickness, fin number and base
thickness of the extruded-fin heatsink on thermal resis-
tance
j
a are presented for the FC-PBGA/lid/heatsink
assembly. Increasing the number of fins or the length of
the fins decreases the thermal resistance
R
j
a due to the
increase of heatsink cooling area, as shown in Figures 15
and 16 , for both natural
R


0v and forced convection


1v

environments.
It is seen that
j
a improves as the base thickness in-
creases because increased base thickness increases the
total cooling area of the heatsink and improves the cool-
ing efficiency of the heatsink. But the improvement is
small and the resulting material cost could be higher.
R
Figure 15. Relation between thermal resistance and design
parameters of extruded-fin heatsink in natural convection
condition


0v

.
Figure 16. Relation between thermal resistance and design
parameters of extruded-fin heatsink in forced convection
condition


1v

.

C. F. LIN ET AL.
92
Increasi fing the n thickness increases the total cooling
ar
rotect the chip and improve heat
ea of the heatsink. However this increased cooling area
is very small compared with the total area of the heat-
sink, so the results show only a slight effect for fin thick-
ness and can be neglected.
3.3.2. Effect of the Lid
The purpose of lid is to p
spreading from the chip. Effect of thermal conductivity
of the lid on the thermal performance of the assembly for
both natural and forced convection environments can
also be found in Figures 10 and 11, respectively. In-
creased lid conductivity increases the thermal perform-
ance of the assembly.
Specific variance of the thermal resistance
j
a
R
2 a rela- tive to the lid thickness is shown in Figures 1nd 13
for both natural and forced convection environments.
The results shows a small improvement in
j
a
R as thick-
ness increases.
3.3.3. Effect of Underfill fill is to release the thermo-
M2 the thermal perform-
the substrate
e assembly due to a longer thermal path
fo
ffect of Lid-Substrate Adhesive
An adhesive material is required to form a good bond
cts of the con-
finite element simulation has been
thermal performance of a thermally
gn parameters of the polymer-
ba
mal performance of the assembly. High er ther-
m
The purpose of the under
mechanical stress of the solder bumps due to CTE mis-
match between the silicon chip and the organic substrate.
Increased thermal conductivity of the underfill increases
the heat conduction in the package and is helpful for dis-
sipating heat to the ambient atmosphere. Thus, increasing
the conductivity of the underfill increases the thermal
performance of the assembly.
3.3.4. Effect of TIM1 and TI
The effects of TIM1 and TIM2 on
ance of the asse mbly for both natura l and forced convec-
tion environments can also be found in Figures 10-13.
As TIM1’s or TIM2’s conductivity incr eases, the ther-
mal performance of the assembly increases, since in-
creased conductivity of TIM1 or TIM2 reduces the ther-
mal resistance in the heat transfer path. Also, it can be
found in Figures 12 and 13 that as TIM1’s or TIM2’s
thickness decreases, the thermal performance of the as-
sembly increases, since decreased thickness of TIM1 or
TIM2 reduces the thermal resistance in the heat transfer
path. Simulation results show that TIM1 has a greater
influence on assembly thermal performance than TIM2.
3.3.5. Effect of Substrate Core Material
The effect of the thermal conductivity of
core material on the thermal performance of the assem-
bly can also be found in Figures 10 and 11. Increasing
the conductivity of the substrate core material increases
heat transfer from chip to ambient, but the amount of
heat conducted through this pathway is quite small rela-
tive to the amount conducted through the vias and heat-
sink. Thus, thermal performance increases with increas-
ing conductivity of the substrate core material, but the
effect is small.
Thicker substrate core material reduces the thermal
performance of th
r heat transfer to the PCB. Thus thermal performance
decreases with increasing thickness of the substrate core
material but the effect is small, as seen in Figures 12 and
13.
3.3.6. E
between the lid and the substrate. The effe
ductivity and thickness of the adhesive with regard to the
thermal performance of the assembly are also shown in
Figures 10-13. It is found that these effects are all rela-
tively insignificant, but higher conductivity and smaller
thickness are preferred.
4. Conclusions
Three-dimensional
used to study the
enhanced FC-PBGA assembly in both natural and forced
air convection environments. The thermally enhanced
FC-PBGA assembly is a basic FC-PBGA assembly with
a lid attached on top, after which an extruded-fin heat-
sink is attached on the top of the lid. The adding of an
aluminum lid/heatsink significantly influences the ther-
mal performance of the assembly due to the large in-
crease in heat transfer area. Also, the temperature dis-
tribution in the extruded-fin heatsink is quite uniform due
to the very high thermal conductivity of the aluminum
material. If a extruded-fin heatsink is not applied to the
basic model, then addition of the lid results in significant
improvement of the thermal performance. On the other
hand, if an extruded-fin heatsink has been applied to the
basic model, then further addition of the lid does not re-
sult in significant further improvement of thermal per-
formance, because the increased cooling area presented
by the sides of lid is little relative to the large total cool-
ing area of the heatsink.
A series of parametric simulations investigate the ef-
fects of varying the desi
sed materials and the thermal enhancement compo-
nents. Conclusions from the analysis are summarized as
follows:
1) Extruded-fin heatsink design has significant effects
on the ther
al conductivity improves the thermal performance of
the assembly. However, results show little improvement
in performance at thermal conductivities above 200
W/m·K. Thus, little advantage can be obtained by using
copper (thermal conductivity = 393 W/m·K) relative to
aluminum (thermal conductivity = 226 W/m·K). This
observation is of interest to designers with regard to ma-

C. F. LIN ET AL.
93
or TIM2 decreases, the thermal
pe
s but the improve-
m
ce of the assembly increa-
se
nd to have lesser impact o
pa
FERENCES
[1] G. H. Wu, S.HThree-Dimensional
Finite Elemenechanical Behavior
terial selection. Increasing the fin length, fin number, and
base thickness of the extruded-fin heatsink results in in-
creased thermal performance of the assembly. For the
cases studied, simulation results show that the fin length
and fin number have a stronger influence on the thermal
performance than base thickness. The effect of fin thick-
ness is negligibly small.
2) As for conductivity, when TIM1 or TIM2 increases
or the thickn ess of TIM1
rformance of the assembly increases. Simulation re-
sults show that TIM1 has a stronger influence on assem-
bly thermal performance than TIM2.
3) As conductivity of the underfill increases, thermal
performance of the assembly increase
ent is not very significant.
4) It is found that as the thickness of the substrate core
decreases, thermal performan
s. Thermal performance of the assembly can be en-
hanced by using a thinner core. Also, higher conductivity
of the substrate core is desirable for better thermal per-
formance of the assembly.
5) The effects of the lid-substrate adhesive parameters
are also considered and foun
ckage thermal performance. However, higher conduc-
tivity and lower thickness of the lid-substrate adhesive
are found preferable.
RE
. Ju and T. C. Tsein, “
t Analysis of Thermom
in Flip-Chip Packages under Temperature Cycling Condi-
tions,” Journal of Reinforced Plastics and Composites,
Vol. 24, No. 18, 2005, pp. 1895-1907.
doi:10.1177/0731684405054333
[2] F. Gugliermermetti and S. Grignaffini, “Direct Approach
Thermal performance of Flip Chip Ball Grid
hiang, “Thermal
to the Design of Plate Fin Heat Sinks Stack Cooled by
Forced Convection,” IEEE Electronic Packaging Tech-
nology Conference, San Diego, 8-10 December 1998, pp.
138-142.
[3] B. Joiner, “
Arrays Packages,” Eighteenth Annual IEEE Semiconduc-
tor Thermal Measurement and Management Symposium,
San Jose, 12-14 March 2002, pp. 50-56.
[4] K. M. Chen, K. H. Houng and K. N. C
Resistance Analysis and Validation of Flip Chip PBGA
Packages,” Microelectronics Reliability, Vol. 46, No. 2-4,
2006, pp. 440-448. doi:10.1016/j.microrel.2005.06.001
[5] Z. Luo, H. Cho and K. Cho, “System Thermal Analysis of
Mobile Phone,” Applied Thermal Engineering, Vol. 28,
No. 14-15, 2008, pp. 1889-1895.
doi:10.1016/j.applthermaleng.2007.11.025
[6] A. R. Menon, S. Karajgikar and D. Agonafer, “Thermal
evision 13.0,” Swanson Ana-
ermal Enhancement on
Design Optimization of a Package on Package,” 25th
Annual IEEE Semiconductor Thermal Measurement and
Management Symposium (Semi-Therm), San Jose, 15-19
March 2009, pp. 329-336.
[7] “ANSYS User’s Manual, R
lysis System Inc., Houston, 2010.
[8] T.-Y. Lee, “An Investigation of Th
Flip Chip Plastic BGA Packages Using CFD Tool,” IEEE
Transactions on Components, Packaging, and Manufac-
turing Technology, Vol. 23, No. 3, 2000, pp. 481-489.
doi:10.1109/6144.868847
[9] A. Mertol, “Thermal Performance Comparison of High
tations for Electronic Equip-
omenclature
Pin Count Cavity-Up Enhanced Plastic Ball Grid Array
(EPBGA) Packages,” IEEE Transactions on Components,
Packaging, and Manufacturing Technology, Part B, Vol.
19, No. 2, 1996, pp. 427-443.
[10] G. N. Ellison, “Thermal Compu
ment,” E. Robert Krieger Publishing Company, Malabar,
1989.
Nn: empirical factor.
P:it bd. device power, W.
f
: empirical factor. ll grid array. PCB: printed circuoar
FC-PBGA: flip-chip plastic ba
s
a
T


: temperature difference between the surface and
g
: heat dissipation rate per unit volume in the chip, the ambient air, ˚C.
T

: ambient air tem
3. Wm
kperature, ˚C.
: thermal conductivity, WmK.
x
k: thermal conductivity in the x direction, WmK

.
y
k: thermal conductivity in the y direction, WmK

.
z
k: thermal conductivity in the z direction, WmK

.
c
h: convection heat transfer coefficient, 2 Wm K

.
rad : radiation heat transfer coefficient, h2
Wm K

.
ch
L: characteristic length, m.
j
a
T:
R: thermal resistance of the assembly, ˚C/W.
temperature, ˚C.
j
T
C. : predicted maximum temperature of an assembly, ˚: surface temperature of the assembly, ˚C.
s
T
V: air velocity, m/s.