Thermal Management for Electronic Packaging 03/02/2006
Guoping Xu Sun Microsystems
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Outline Introduction Heat transfer theory Thermal resistance in electronic packaging Thermal design Thermal modeling Thermal measurement
Page 2
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Introduction Functions of Electronic Packaging Package protection Signal distribution Power distribution Heat dissipation
Page 3
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Introduction Packaging Hierarchy Chip Package Board System Rack Room
Page 4
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Introduction High end chip power trend 600
CPU Power, (W)
500 400
UltraSparc [1] Power 4 [2] Itanium 2 [3] ITRS 2002 [4] ITRS 2005
300 200 100 0 1990
1995
2000
2005
2010
2015
2020
Year
Page 5
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Introduction Cost performance chip power trend 160 140
ITRS 2005
CPU Power, (W)
120 100 80 60 40 20 0 2004
2006
2008
2010
2012
Year
Page 6
2014
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Introduction Power density in datacom equipment
Page 7
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Introduction Power density in datacom equipment Total power: 24KW Footprint: 15 sq. ft Power density: 1600W/sq. ft
Sun Fire E25K
Page 8
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Introduction Impact of Device junction temperature Computing performance Reliability Fire hazard and/or Safety issues
Page 9
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Heat Transfer Theory Conduction Definition: Conduction is a mode of heat transfer in which heat flows from a
region of higher temperaure to one of lower temperature within a medium (solid, liquid, or gases) or media in direct physical contact
Fourier's law: Q = -KA(dT/dX) 1-D conduction: Q = -KA (T1-T2)/L Thermal resistance: R = (T1-T2)/Q = L/(KA)
Page 10
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Heat Transfer Theory Conduction Contact thermal resistance
Page 11
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Heat Transfer Theory Thermal conductivity of various packaging materials Material Aluminum (pure) Aluminum Nitride Alumina Copper Diamond Epoxy (No fill) Epoxy (High fill) Epoxy glass Gold Lead Silicon Silicon Carbide Silicon Grease Solder
W/mK 216 230 25 398 2300 0.2 2.1 0.3 296 32.5 144 270 0.2 49.3
Page 12
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Heat Transfer Theory Convection Convection: is a mode of heat transport from a solid surface to a fluid and occurs due to the bulk motion of the fluid. Newton's law: Q= hA (Tw- Tf) Convective thermal resistance: R= 1/(hA) Effects of heat transfer coefficient Convetion mode: Natural convection, Foreced convection, phase change Flow regime: Laminar, Turbulent flow Flow velocity y Tf Fluid y Surface condition Fluid Velocity distribution
u(y)
Heated surface
Temperature distribution
Tw T(y)
Page 13
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Heat Transfer Theory Typical values of the heat transfer coefficient
Page 14
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Heat Transfer Theory Radiation Definition: Radiation heat transfer occurs as a result of radiant energy emitted from a body by virtue of its temperature.
Page 15
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Resistance -Package without heat sink Ta
Tb
Chip
Tt Tj Package PCB
Rja: Junction to air thermal resistance Rja= (Tj-Ta)/P Low value is good thermal perfromance
Rjc: Junction to case thermal resistance Rjc = (Tj-Tc)/P
Ψjt: Thermal characterization parameter: Junction to package top, NOT thermal resistance. Ψjb: Thermal characterization parameter: Junction to board Page 16
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Resistance -Package with heat sink Ta Heat sink Package Die
Ts Tc Tj
Rja: Junction to air thermal resistance Rja= (Tj-Ta)/P=Rjc +Rcs+Rsa
Rjc: Junction to case thermal resistance Rjc = (Tj-Tc)/P
Rsa: External heat sink thermal resistance Rsa= (Ts-Ta)/P
Page 17
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Resistance -PBGA package example
Tair Page 18
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Resistance -Impact factors for package without heat sink Die size Package size, lead count Packaging material thermal condunctivity Material thickness in major heat flow path Number of vias Heat spreader or heat slug Air velocity and temperature PC Board size Board configuration and material Board layout
Page 19
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Design Conduction application Single material Q Die Substrate
Tc Tj
Tc = Tj - QL/(KA)
ti
Composite material Kin-plane Layer 1 Layer 2 Layer i Layer N KTthrough
Uniform heating on the die
Page 20
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Design Conduction application Heat spreader Die Substrate
Ab As kb Ab R ba tanh H b Rsb kb Ab As 1 kb Ab R ba tanh H b
23 1 Ab As Ab: Heat spreader base area As: Heat source area Hb: Heat spreader thickness Kb: Heat spreader thermal conductivity Page 21
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Design Convection application-Heat sink design
Hf
b H
tf
Fins Hb L
Heat sink base W
Page 22
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Design Convection application-Heat sink design Thermal resistance:
Heat transfer:
Rba
Tb Tinlet Q
hDh Nu 7.54 k air
Fin efficiency:
o 1
At
1 f
.
.
m c p (1 e
hAto m c p
Laminar flow
hDh Nu 0.024 Re 0.786 Pr 0.45 k air
Af
1
Turbulent flow
f
tanh( mH f ) mH f Page 23
)
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Design Convection application-Heat sink design Total static pressure loss: U ch 2 L P K c 4 f app K e Dh 2 Culham and Muzychka (2001) Apparent friction factor fapp calculation
Fully developed flow friction factor f
f app Re 3.44 L*
L Dh Re
2
f Re
2 40.829b H f 3 22.954b H f 4 5 6.089b H f
f Re 24 32.527 b H f 46.721 b H f
Contraction loss coefficient Kc Expansion loss coefficient Ke L*
0.5 2
David Copeland (2000) 12
1
f app
Re 3.2 L*
0.57 2
2
f Re
b H f 2 1 f Re 4.7 19.64 b H f 12
K c 0.421
K c 0.8 0.4 2
K e 1 2
K e 1 2 0.4
Nt f W
b tf b
12
Page 24
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Design Convection application- Heat sink design Impact factors Air flow rate Available space Heat sink base and fin material Fin pitch and fin thickness Heat flux Heat sink technologies
Page 25
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Thermal Design Design methodology Define requirements Analyze given package design Identify major heat paths and paths for improvements Consider and assess potential improvements Detail analysis/modeling Build prototypes Thermal testing
Page 26
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Modeling Finite Element Method (FEM) Software: ANSYS Solve conduction problem within package or board Require input data: material propoerties, package/board construction/geometry Boundary conditions: Heat source distribution on the die or board Effective convective heat transfer coefficient on the surface of the package or board
Page 27
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Modeling Finite Element Method Procedure: Create package/board geometry or import from CAD file Mesh Input material properties and assign boundary conditions Solve Post-process
Page 28
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Modeling Finite Difference Method (FDM ) Computational Fluid Dynamics (CFD) Commercial software: Flotherm, Fluent Solve the temperature field and flow field Not only solve the conduction, also on convection, radiation and phase change Required input: Geometry, flow conditions, material properties including fluid Mesh dependent on the chosed model
Page 29
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Modeling Finite Difference Method (FDM ) Example
Page 30
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Measurement Packaging thermal parameters Mount package on a standard test board Mount thermocouple on top of the package center Mount thermacouple on board at the edge of package Put package in a standard test environment Wind tunnel to vary the air speed
Apply known amount of power Measure temperature of Tj, Ta, Tb, Tt Calculate Rja, Ψjt, Ψjb Ta Tt Chip Tj Tb Package PCB test board Page 31
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Measurement Packaging thermal parameters
Page 32
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Measurement Packaging thermal resistance Rjc All heat is removed from top of the package
Page 33
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Measurement Thermal interface material resistance Rcs
Page 34
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Measurement Thermal interface material resistance Rcs
Page 35
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Measurement p et
tl
T ou
p ir flow
A
t
le
in
T
2
4
T
Heat sink thermal resistance Rsa
wer
lo
B
le
zz
No
er
b
m
ha
stc
te
low
irf
A rs
te
ea
Rsa
Ts Tinlet Q
Q k s As
d
le
a
c
ts
o
n
re
a
s
ion
ns
e
im
D
tion
sula
In
H
k
loc
b
r
pe
o
C
3
T
al
ri
ate
em
fac
2
T
al
rm er
he
T int
s
T
Heat sink
T3 T2 L
Heat source size: 25 mm x 25 mm
Page 36
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Measurement Heat sink thermal resistance Rsa The rmal Re s is tanc e , ( o C/ W )
0.600 Test data Present method (Eqs. 6-8) Teertstra [1] Copeland [2]
0.500 0.400 0.300 0.200 0.100 0.000 0
0.01
0.02 0.03 3 Flow rate (m /s )
0.04
0.05
Page 37
CSE291: Interconnect and Packaging, UCSD, Winter 2006
Q&A
Page 38
Guoping Xu
[email protected]