Reliability Mechanics and Modeling for IC Packaging
Transcript of Reliability Mechanics and Modeling for IC Packaging
IEEE EPTC 2019 โ Singapore December 4 - 6, 2019
Reliability Mechanics and Modeling for IC Packaging
โ Theory, Implementation and Practices
1Xuejun Fan, and 2Ricky Lee 1Lamar University
2Hong Kong University of Science and Technology
[email protected] , [email protected]
EPTC Professional Development Course
December 4, 2019
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
2
Outline
โข Introduction
โขTemperature Loading
โข Mechanical Loading
โข Moisture Loading
โข Electrical Current Loading - Multi-Physics Modeling
โข Summary
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
3
3
โขIntroduction
โขTemperature Loading
โขMechanical Loading
โขMoisture Loading
โขElectrical Current Loading -Multi-Physics
Modeling
โขSummary
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
4
4
Electronic Packaging Evolution
8
16
32
64
128
256
1000
>5000
<1985 1990 1995 2000 2015
Pin Count
QFN
DIP Dual Inline
Package
PGA
Pin Grid Array
PPGA Plastic PGA
MCP Multi-Chip Package
FLIP CHIP
Surface Mount Through Hole Fine Pitch/Thin Modules/System
QF
P
2005
SOP
SOJ Small Outline
J-Lead
(WLP)
Package-in-Package
FC
CSP
BGA
Flip chip BGA
Area Array 3D/integration
QFN w/SMD
Integrated
module
3D
module
Module
w/Embedded
Passives
1-Layer LGA
TSOP Thin SOP
MCM Multi-Chip Module
Stacked chips
PLCC Plastic Leaded
Chip Carrier
Wirebond
ed CSP
Ceramic Module Laminate Module
FEM
Stacked
package
Package-on-
Package
CoWoS (Chip-on-Wafer-on-Substrate)
InFO (Integrated Fan Out)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
5
5
Wafer, Package, and Board Levels
Electronic Packaging Wafer Fabrication &
Backend Process Surface Mounting
โข Design of a package must consider the interactions among wafer, package, and board (e.g. CPI โ chip-package-interaction).
Package
level
Wafer
level Board
level
Package
level
Wafer
level Board
level
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
6
6
Examples of Failures
10 m
Interface delamination
Substrate cracking Pad cratering
Solder brittle failure
Fatigue crack
Thin film delamination
e-
500 nm
Void
Via
e-
Organosilicate
Etch Stop
Liner
500 nm
Void
Via
e-
Organosilicate
Etch Stop
Liner
Electromigration
Film rupture
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
7
7
Thermal-mechanical Stress
Stress-free
Cooling-down
Heating-up
Thermal stress in solder
Cooling-down
Package warpage
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
8
8
Mechanical Load
โข Handheld electronic products are susceptible to drop impact failure.
PDA Handphone
MPEG-4 MP3 Player
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
9
9
Stage 1: Moisture absorption
Moisture-induced Failure
Device exposed to reflow temperature,
typically 260ยฐC
Stage 2: Initiation of delamination
Stage 3: Delamination propagation Stage 4: Package cracking and vapor release
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
10
10
Electrical Current Stressing
e-
Electron wind
Voids
Hillocks
e-
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
11
11
Organization of Course
โข Four modules
โTemperature loading
Thermal mismatch, temperature gradient, etc.
โMechanical loading
Drop and impact, bend, etc.
โMoisture loading
Moisture diffusion, vapor pressure, swelling.
โElectrical current loading - combined loading
Electrical-thermal, electrical-thermal-mechanical, and
electromigration (electrical-thermal-mechanical-mass
transport)
โข Theory, implementation, and best practices.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
12
12
โขIntroduction
โขTemperature Loading
โขMechanical Loading
โขMoisture Loading
โขElectrical Current Loading - Multi-Physics
Modeling
โขSummary
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
13
13
Outline
โข Basic concepts and analytical solutions
โข Applications โDie-level thermal stress โ thermal stress in TSV
โPackage-level thermal stress problem โ warpage
โChip-package interaction (CPI) โ sub-modeling technique
โBoard level thermal stress problems
Solder ball thermal cycling
Creep equations
โข Best practices โ Initial stress free condition โFull model vs. global/local model
โVolume averaging
โStress singularity
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
14
14
Stress, Strain, and Basic Elastic Material Properties
๐ =๐
๐ด0
ํ =๐ฟ
๐ฟ0
๐ = โํlatํlong
Poisson ratio
๐ = ๐ธํ
Hookeโs law
= L โ L0
Material Elastic Modulus (GPa) Poisson Ratio
Adhesive 0.01 โ 10 0.4 โ 0.49
Epoxy 10 โ30 0.35 โ 0.48
Copper 120 0.35
Silicon 131 0.28
Solder 25 โ 51 0.35
Substrate/PCB 11 โ 26 various
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
15
Coefficient of Thermal Expansion (CTE)
โข Material dimensions change with temperature.
โข The rate of dimensional change with temperature is called CTE.
โข Primary driver of thermo-mechanical stress in electronic packages.
Material a (ppm/ยฐC)
Aluminum
Stainless Steel
Silicon
Copper
Underfill Epoxy
FR4 (in plane)
FR4 (z-axis)
23
12
2.6
17
20, 70
15, 10
65, 180
L
T = T0
L+DL
T = T1
L
Lth D
T
th
D
a
T = T0 T > T0
Increase
internal energy
(heat) causes
atomic bond
lengths to
increase
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
16
16
Thermal Mismatch โ Two-Layer Structure
Superposition : axial loading + bending
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
17
17
Thermal-Mismatch โ Multi-Layer Structure
Superposition : axial loading + bending
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
18
18
Thermal-Mismatch : Analytical Solution
Ana Neves Vieira da Silva. Analytical Modeling of the Stress-Strain Distribution in a Multilayer Structure with Applied Bending. PhD Dissertation. Instituto Superior Tecnico of Universidade Tecnica de Lisboa, Portugal.
(effective CTE)
(warpage)
(stress in each layer)
Multi-layer structure
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
19
19
Thermal Mismatch - Shear Stress and Strain on Solder Joint
Cooling to a
lower
temperature
Solder joint
under stress
(PCB shrinks more than package)
h
TL D
)(chipPCB
aa
: shear stress in solder
: solder ball shear strain
G: elastic shear modulus
L: distance of netural point (half-die size, diagnoal)
h: solder ball stand-off
a: CTE
= G
A simplified and approximate solution
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
20
20
Thermal Mismatch โ TSV Structure
Shrink fit of compound cylinder
Single cylinder under external and internal pressure
Advanced Mechanics of Materials and Applied Elasticity, 5th Edition, by Saul K. Fenster, Ansel C. Ugural.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
21
21
Thermal Stress due to Temperature Gradient
Advanced Mechanics of Materials and Applied Elasticity, 5th Edition, by Saul K. Fenster, Ansel C. Ugural.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
22
Finite Element Analysis (FEA)
โข Analytical solutions are usually not applicable to complex problems.
โข Based on dividing the problem into discrete elements for which analytical solutions exit.
โข Analytical equations are then formed into a matrix and solved simultaneously.
โข Method may require multiple iterations to converge and may have to divide โtimeโ steps to reach a solution.
โข Used for structural, thermal, EMI, fluid dynamics, diffusion, stock analysis, etc.
Bracket meshed into 11,295
elements.
Individual element
(w/nodes in red)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
23
23
Application โ Warpage Analysis
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
24
24
Capped-Die Flip Chip Package Design - assembly process
Using an underfill-like encapsulant, a metallic cap is attached to the die.
Application โ Warpage Control
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
25
25
Comparison for Warpage Control Methods by FEA
a) Bare die on substrate
b) Package with stiffener
c) Package with lid
d) Capped-die package
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
26
26
Experimental Verification - Warpage
Shadow Moirรฉ Data
Bare Die w/ U6 Underfill
Bare Die w/ U2 Underfill
Capped-Die w/ U6 Underfill
Capped-Die w/ U2 Underfill
Shen Y, Zhang L. Zhu WH, Zhou J, Fan XJ. Finite-element analysis and experimental test for a capped-die flip-chip package design,
IEEE Transactions on Components, Packaging and Manufacturing Technology. 6(9), 1308 โ 1316. 2016.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
27
27
Chip-Package Interaction (CPI)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
28
28
Multilevel Sub-modeling Technique
package level
interconnect level
critical solder level
die-solder interface level
Xuefeng Zhang, Se Hyuk Im, Rui Huang, and Paul S. Ho, Chip-Package Interaction and Reliability Impact on Cu/Low-k Interconnects, Chapter 2, UT Austin 2008/
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
29
29
Multilevel Sub-modeling Technique
XH Liu, TM Shaw, G Bonilla - Advanced Metallization Conference, 2010 - sematech.org
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
30
30
Temperature Cycling
Board
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
31
31
Temperature Cycling
โข Temperature cycling โ uniform temperature condition assumed
-40
125
Tem
per
atu
re (ยฐC
)
Time (Seconds)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
32
32
Power Cycling
โข Temperature gradient exists.
โข Much less severe than temperature cycling.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
33
33
Fatigue Life Estimate
๐๐ = ๐๐ํ๐ ๐๐
0
๐๐ = ๐ด โ๐๐โ๐
Coffin-Manson Equation
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
34
Solder Creep Equations (ANAND Model)
โข Anandโs model (rate-dependent visco-plastic model)
๐ํ๐๐๐ก
= ๐ด sinh๐๐
๐
โ1๐
exp โ๐
๐ ๐
๐๐
๐๐ก= โ0 ๐ต ๐ผ
๐ต
๐ต
๐ํ๐๐๐ก
๐ต = 1 โ๐
๐ โ
๐ โ = ๐ ๐ํ๐/๐๐ก
๐ดexp โ
๐
๐ ๐
๐
Variable/Para
meter
Meaning Units
๐ 0 initial value of deformation resistance Stress
๐/๐ Q = activation energy
R = universal gas constant
Energy/volume
Energy/(volume temp)
๐ด pre-exponential factor 1/time
๐ multiplier of stress Dimensionless
๐ strain rate sensitivity of stress Dimensionless
โ0 hardening / softening constant Stress
๐ coefficient for deformation resistance
saturation value
Stress
๐ strain rate sensitivity of saturation
(deformation resistance) value
Dimensionless
๐ผ strain rate sensitivity of hardening or
softening
Dimensionless
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
35
35
Anand Constants for Various Solder Alloys
Parameter SAC105 SAC305 SAC387 SAC405 SnPb SnPbAg SnAg
๐ 0 2.3479 18.07 3.2992 20 12 12.41 39.09
๐/๐ 8076 9096 9883 10561 9200 9400 8900
๐ด 3.773 3484 15.773 325 4.2E6 4E6 2.23E4
๐ 0.9951 4 1.0673 10 1.5 1.5 6
๐ 0.4454 0.2 0.3686 0.32 0.3 0.303 0.182
โ0 4507.5 144000 1076.9 8E5 1.4E3 1379 3321.15
๐ 3.5833 26.4 3.1505 42.1 14 13.79 73.81
๐ 0.012 0.01 0.0352 0.02 0.071 0.07 0.018
๐ผ 2.1669 1.9 1.6832 2.57 1.31 1.3 1.82
SAC105 D. Bhate et al., Constitutive Behavior of Sn3.8Ag0.7Cu and Sn1.0Ag0.5Cu Alloys at Creep and Low Strain Rate Regimes, IEEE Transactions on Components
and Packaging Technologies, Vol. 31, No. 3, September 2008, p. 622.
SAC305 Motalab et al., Determination of Anand Constants for SAC Solders using Stress-Strain or Creep Data, ITherm 2012, pp. 909-921.
SAC387 D. Bhate et al., Constitutive Behavior of Sn3.8Ag0.7Cu and Sn1.0Ag0.5Cu Alloys at Creep and Low Strain Rate Regimes, IEEE Transactions on Components
and Packaging Technologies, Vol. 31, No. 3, September 2008, p. 622.
SAC405 W. Qiang et al., Experimental determination and modification of Anand model constants for Pb-free material 95.5Sn4.0Ag0.5Cu, in EUROSIME Conf. IEEE,
2007.
SnPb R. Darveaux, โEffect of simulation methodology on solder joint crack growth correlation,โ in Electronic Components & Technology Conf. IEEE, 2000 .
SnPbAg R. Darveaux, โEffect of simulation methodology on solder joint crack growth correlation,โ in Electronic Components & Technology Conf. IEEE, 2000.
SnAg Wang, Z. Cheng, K. Becker, and J. Wilde, Applying Anand model to represent the viscoplastic deformation behavior of solder alloys, ASME Journal of
Electronic Packaging, Vol. 123, 2001.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
36
36
Solder Creep Equations
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
37
37
Solder Creep Equations (NEW)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
38
Solder Creep Equations
TD
B
B
T
TE
5413exp
sec/11090.8
sec/11070.1
(K)eTemperatur
8856000(MPa)ElasticityofModulus
(MPa)Stress
sec)/1(RateStrainTotal
where
24
2
12
1
SnPb Creep Equation
7
2
3
1
EDB
EDB
E
SnAgCu Creep Equation
12
22
3
11
nn
DADAE
MPa1
)/7348exp(
)/3223exp(
sec/1100.1
sec/1100.4
(K)eTemperatur
667.6659533(MPa)ElasticityofModulus
)(Stress
sec)/1(RateStrainTotal
where
2
1
12
2
7
1
n
TD
TD
A
A
T
TE
MPa
Elastic
Grain Boundary Sliding (GBS)
Matrix Creep (MC)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
39
39
Best Practice (1) - Stress-Free Setting
โข Three most commonly used initial stress-free temperatures
โ The solidus temperature of solder material (e.g., for SnAgCu, this temperature
is 217ยฐC)
โ The room temperature as initial stress-free (e.g. 25C)
โ The high dwell temperature of thermal cycle or operating conditions (denoted
as Tmax, e.g. =125C for thermal cycling from -25C to 125C)
โข Tmax as stress-free condition is recommended.
Fan XJ, Pei M, Bhatti PK. Effect of finite element modeling techniques on solder joint fatigue life prediction of flip-chip BGA packages. Proc. of Electronic
Components and Technology Conference (ECTC), 972-980. 2006.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
40
40
Best Practice (2) โ One Global Model (Full Model)
โ Coarse element at most of the places
โ Refined element at critical solder joints
โ No tie element is used
โ Refined solder balls consider both SMD and MD pads
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
41
41
Best Practice (2) โ Sub-modeling (Global/Local)
Substrate/underfill interface
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
42
42
Comparison - Global/Local Model and Full Model
โ The global/local modeling approach provides satisfactory results for the volumetrically averaged and maximum values
โ For some location, the difference in maximum value can be as high as 20% (not shown here)
โข โFull Modelโ is recommended.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
43
43
Best Practice (3) - Volume Averaging
ANSYS 11.0SP1
ANSYS 11.0SP1
Critical
Layer
Elements
โ๐๐๐ฃ๐= average inelastic strain energy density accumulated per cycle for fixed
thickness layer elements
โ๐๐ = strain energy density accumulated per cycle for each element i
Vi = volume of each element i
โ๐ave= โWi Vi
Vi
โข Volume averaging over a fixed thickness of thin layer is
recommended.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
44
44
Best Practice (4) - Worst Case Solder Joint Location
โFor FC-BGA package
Itโs commonly accepted that the worst case solder joint is located in the
outermost along diagonal direction under the die shadow.
โMetrics
Von Mises stress, creep strain, creep strain energy density, peel stress
Maximum value, averaged value
โ Results
Averaged per-cycle creep strain and stress shows the worst-case solder joint
at the middle of die edge
The maximum value of these two parameters show the worst joint at the die
corner
Peel stress finds the maximum tensile stress at the joint one row inside from
the die shadow corner
โข Experimental data
โ The solder joint one row inside from the corner of die shadow usually has the highest crack
growth rate
โ All solder joints along the die edge have comparable crack growth rates
โข Correlation with experimental data is recommended. Bhatti PK, Pei M, Fan XJ. Reliability analysis of SnPb and SnAgCu solder joints in FC-BGA packages with thermal enabling preload. Proc. of Electronic Components
and Technology Conference (ECTC), 601-606. 2006.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
45
General Results: SnPb vs. SnAgCu
Maximum Von
Mises Stress (MPa)
Maximum Equiv
Creep Strain
Temperature
range -25 to +100C
0
0.01
0.02
0.03
0.04
0.05
0 30 60 90 120 150
Time (minutes)
Max E
qu
iv C
reep
Str
ain
-50
-25
0
25
50
75
100
125
Sn/Pb Sn/Ag/Cu
Temp
0
10
20
30
40
50
0 30 60 90 120 150
Time (minutes) Max
Vo
n M
ise
s S
tres
s (
MP
a)
-50
-25
0
25
50
75
100
125
Sn/Pb Sn/Ag/Cu
Temp
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
46
46
Wafer Level Package (Fan-in WLP) โ Worst Location
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
47
47
WLP - von Mises Stress Map in Solder Balls
1
MN
MX
X
YZ
1.11
9.6818.251
26.82235.393
43.96452.534
61.10569.676
78.247
ANSYS 11.0SP1
MAR 7 2008
10:04:17
1
ANSYS 11.0SP1
MAR 7 2008
10:03:05
1
MX
ANSYS 11.0SP1
MAR 7 2008
10:02:21
Critical solder ball
Package center
Package corner
1/8th model
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
48
48
WLP - Plastic Work Density Map
Fan XJ, Varia B, Han Q. Design and optimization of thermo-mechanical reliability in wafer level packaging, Microelectronics Reliability, 50, 536โ546, 2010.
1/8th model
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
49
49
Stress Singularities in Material Joints
Fan XJ, Wang HB, Lim TB. Investigation of the underfill delamination and cracking for flip chip modules under temperature cyclic loading. IEEE Transactions on
Components, Manufacturing and Packaging Technologies, 24(1), 84-91. 2001.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
50
50
Summary
โข Thermal mismatch vs. temperature gradient
โข Analytical solution โLayered structure (stress, warpage, effective CTE)
โCylindrical structure (TSV)
โข Die-level thermal stress โ thermal stress in TSV
โข Package-level thermal stress problem โ warpage
โข Chip-package interaction (CPI) โ submodeling technique
โข Board level thermal stress problem โSolder ball thermal cycling (Flip chip BGA, WLP)
โCreep equations
โBest method for practice
Initial stress free condition; full model vs. global/local model; worst solder ball location, volume averaging
โข Stress singularity of joint materials
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
51
51
โขIntroduction
โขTemperature Loading
โขMechanical Loading
โขMoisture Loading
โขElectrical Current Loading - Multi-Physics
Modeling
โขSummary
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
52
52
Outline
โข JEDEC drop test standard (old and new)
โข Simulation method โ Input G
โLarge mass
โ Input D
โDirect acceleration
โGlobal/local method
โข 4-point bending test
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
53
53
JEDEC Drop Test Set-Up
Drop
table
Base
plate
Rigid
base
Strike
surface
Guide
Rods
Test
board
6
4
2
11
9
7
5
3
1
12
10
8Drop
table
Base
plate
Rigid
base
Strike
surface
Guide
Rods
Test
board
6
4
2
11
9
7
5
3
1
12
10
8
6
4
2
11
9
7
5
3
1
12
10
8
โข Peak acceleration: 1500g
โข Impulse time: 0.5ms with half-sine shape
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
54
JEDEC Test Standard JESD22-B111 (Old)
Group Number Components
Component Location
A 4 U1, U5, U11, & U15
B 4 U2, U4, U12, & U14
C 2 U6 & U10
D 2 U7 & U9
E 2 U3 & U13
F 1 U8
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
55
JEDEC Test Standard JESD22-B111A (New, 11/2016)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
56
FEA Modeling: Input-G Method
G G
G G
PCB
M u 1 = C u 1 + K u1 = 0 (1)
with initial conditions
u1 t=0 = 0, u 1 t=0 = 2gH (2)
and boundary condition
a = 1500g sin
ฯt
tw for t โค tw , tw = 0.5
0 for t โฅ tw
(3)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
57
FEA Modeling: Input-G Method
G G
G G
PCB
Large Mass Method
Large mass 108 times mass of structure
coupled (1mm away from PCB bottom)
Acceleration = 1500 ร g ร sin (pi ร t/0.5)
Force = 108ร Acceleration
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
58
58
FEA Modeling: Input-D Method
u1 at hole =
โ
tw
ฯ
2
1500g sinฯt
tw
+ tw
ฯ 1500g + 2gH t , t โค tw
2tw
ฯ 1500g + 2gH t โ
tw
ฯ
2
1500g , t โฅ tw
M u 1 = C u 1 + K u1 = 0 (1)
with initial conditions
u1 t=0 = 0, u 1 t=0 = 2gH (2)
and boundary condition
a = 1500g sin
ฯt
tw for t โค tw , tw = 0.5
0 for t โฅ tw
(3)
G G
G G
PCB
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
59
59
FEA Modeling: Direct Acceleration Method
M u 2 + C u 2 + K u2 =
โ M 1500g sin
ฯt
tw t โค tw , tw = 0.5
0 t โฅ tw
(5)
with initial conditions
u1 t=0 = 0, u 1 t=0 = 2gH (6)
and boundary conditions u2 at hole = 0 (7)
G G
G G
PCB
Fan XJ, Ranouta AS, Dhiman HS. Effects of package level structure and material properties on solder joint reliability under impact loading. IEEE
Transactions on Components, Packaging and Manufacturing Technology. 3(1), 52-60. 2013.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
60
60
Input-G Method โ Board Vibration
โข Since acceleration impulse is given on the screw locations, the displacements
are not fixed and the board is moving in one direction.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
61
61
Direct Acceleration Method โ Board Vibration
โข Since the displacement is fixed at the screw locations, the board is vibrating as
expected.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
62
62
Input-G vs. Direct Acceleration Method
-1500
-1000
-500
0
500
1000
1500
2000
0 2 4 6 8
Acce
lera
tion
(G
)
Time(ms)
Acceleration Method
Large Mass Method
Distribution of Board Strain in x direction along
line AB at t = 1.5 ms
Comparison of acceleration time history
โข Direct Acceleration method is recommended, but all methods
provide identical results.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
63
FEA Global Model
PCB
Solder Bump
Arrays
Solder Layer
Solder Bump
Model Solder Layer
Model
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
64
FEA Local Model 1
X
Y
Z
OCT 21 2008
00:49:47
ELEMENTS
MAT NUM
1
11XY
Z
OCT 21 2008
00:29:37
ELEMENTS
MAT NUM
Cut Boundary
Distance at
2mm ร 2mm
Cut Boundary DOF Constraints
at 2mm ร 2mm
Cut Boundary
DOF Constraint
Global
Model
Local Model
Local Model 1 Local Model 2
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
65
FEA Local Model โ Solder Joint
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
66
Effect of Cut Boundaries in Sub-modeling
0
100
200
300
400
500
600
1mm x 1mm 1.5mm x 1.5mm 2mm x 2mm 2.5mm x 2mm
Max.
Peelin
g s
tress (
Mpa)
Cut Boundary Distance(mm)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
67
Mode 1: 212 Hz
Mode 2: 551 Hz
Mode 3: 915 Hz
Mode 4: 1257 Hz
Mode 5: 2193 Hz
Global Model Results โ Frequency and Mode Shape
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
68
Experimental Validation S
train
(M
icro
-
str
ain
, ฮต x
)
Time (ms)
Str
ain
(M
icro
-
str
ain
, ฮต x
)
Time (ms)
Time (ms)
Time (ms)
Str
ain
(M
icro
-
str
ain
, ฮต y
)
Str
ain
(M
icro
-
str
ain
, ฮต y
)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
69
Board Strain Analysis: Package Corner Strain(ฮตx )
-5000
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
0 2 4 6 8
Str
ain
(M
icro
-str
ain
, ฮต x
)
Time(ms)
U1
U2
U3
U6
U7
U8
Strain History in x
Direction
โขU1>U3>U8>U2>U7>U6.
โขU8 bends in opposite direction to U1.
-2500
-2000
-1500
-1000
-500
0
500
1000
1500
2000
2500
3000
0 2 4 6 8
Str
ain
(M
icro
-str
ain
, ฮต y
)
U1
U2
U3
U6
U7
U8
Strain History in y
Direction
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
70
Ball Stress Contour Plot on ยผth JEDEC Board
513 MPa
U1 U2 U3
U6 U7 U8
367 MPa
299 MPa
464 MPa
346 MPa 486 MPa
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
71
Maximum Peel Stress
0
100
200
300
400
500
600
U1 U2 U3 U6 U7 U8
Ma
x.
Pe
elin
g s
tress (
Mp
a)
Component
Overall, the stress
developed in components
can be ranked as:
U1>U8>U3>U2>U7>U6
Failure rank:
โข Group A ( U1, U5, U11 and U15)> Group F (U8) > Group E (U3, U13).
A B E B A
U15 U14 U13 U12 U11
C D F D C
U10 U9 U8 U7 U6
A B E B A
U5 U4 U3 U2 U1
1-2
-3-4
-5-1
0-9
-8-7
-6-G
-15-1
4-1
3-1
2-1
1
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
72
Correlation of Solder Joint Stress and Board Strain
Board Strain Ranking: U1>U3>U8>U2>U7>U6
Solder Joint Stress Ranking:U1>U8>U3>U2>U7>U6
โข Exact correlation does not exist for strain and stress values. Fan XJ, Ranouta AS. Finite element modeling of system design and testing conditions for component solder ball reliability under impact. IEEE
Transactions on Components, Packaging and Manufacturing Technology. 2(11), 1802-1810, 2012.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
73
73
4-Point Bending Test
aLa
hx
43
3
x
L
hx 3
12
Segment CD
Cโฒ C
D
Segment CCโ
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
74
74
โข 4PT bending test to take advantage of uniform strain BC for each component.
โข JEDEC board + 9 components with electric monitoring for each one of them.
โข Cyclic testing of freq 3Hz @ 1.5mm, 2mm and 2.5mm deflections to DC failure.
4-Point Bending Test
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
75
Cyclic Bending: Modeling Methodology
Global/local modeling methodology
โข Elastic global model.
โข Local models use EPP SAC405 solder material properties.
M Xie, Solder Reliability Models in 4 Point Bend, Intel, 2006.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
76
76
Strain Distribution
-2500.00
-2000.00
-1500.00
-1000.00
-500.00
0.00
0 10 20 30 40 50
Distance from the center
Mic
rost
rain
, LE
11
d=1.5mm
d=2.0mm
d=2.5mm
Linear moment
Constant moment
Package inside
Package edge
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
77
77
Board Side or Package Side?
Consistency with failure propagation in terms of peeling stress
โข Package side: inside-out
โข Board side: outside-in
Failure happens more likely at the package side: higher failure
parameter value.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
78
78
Failure Pattern
Solder joint crack propagation direction:
โขPackage side: inside-out
โขBoard side: outside-in Sequential Failure (1.5mm deflection)
Simultaneous Failure (2.5mm deflection)
Zhou T and Fan XJ. Effect of system design and test conditions on wafer level package drop test reliability. SMTA International. October 2013.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
79
Summary
โข JEDEC drop test standard โJESD22-B111, old one, with 15 components
โJESD22-B111A, new one, with 4 components or 1 component
โข Finite element modeling โ Input G method, large mass method, input displacement method, direct
acceleration method
โGlobal/local modeling
โPeel stress used as indicator for failure
โข Four-point bending test and modeling โGlobal/local modeling
โGlobal with linear elastic but nonlinear geometry analysis
โLocal model with elastic-plastic modeling
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
80
80
โขIntroduction
โขTemperature Loading
โขMechanical Loading
โขMoisture Loading
โขElectrical Current Loading - Multi-Physics
Modeling
โขSummary
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
81
81
Outline
โข Introduction
โข Moisture related reliability testing
โข Moisture diffusion modeling
โข Vapor pressure theory
โข Hygroscopic swelling
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
82
82
Moisture Absorption of Electronic Packages
โข Electronic packages absorb moisture in uncontrolled humid conditions prior to the surface mount on board.
storage shipment
Individual component Surface mount to board
Fan XJ, Suhir, E. (eds.). Moisture Sensitivity of Plastic Packages of IC Devices. Springer, New York, 2010.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
83
83
Polymer Materials in Electronic Packaging
โข Bulk-form โ encapsulation (e.g. mold compound)
โ substrate โฆ
โข Adhesives โ die-attach, underfill
โ thermal adhesives โฆ
โข Thick- or thin- film โ solder mask
โ passivation
Solder mask
Passivation
Die
Substrate
Solder
bump
A leadframe package
Mold compound
A ball grid array (BGA) package
Die attach
Carrier
Underfill
Thermal adhesive
โข Polymeric materials are susceptible to moisture absorption.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
84
โ Under 85ยฐC/85%RH condition, w= Csat = 2.47e-2 g/cm3 = 81.2 a (Csat:
saturated moisture concentration)
โข Moisture is condensed into liquid state.
โข Moisture exists in micro-pores or free volumes (in bulk or at interface).
โข Moisture vaporizes at reflow, possibly still at mixed liquid/vapor phases.
Moisture Absorption in Polymeric Materials
85ยฐC/85RH
a
w
Moisture condensation in a typical underfill
w: moisture density in polymeric material a: ambient moisture density under 85ยฐC/85%RH
w= 81.2 a
Fan XJ, Lee SWR, Han Q. Experimental investigations and model study of moisture behaviors in polymeric materials. Microelectronics Reliability 49, 861โ871. 2009.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
85
85
Moisture Related Testing
โข Moisture sensitivity test
Stage 1: Moisture absorption Stage 2: Soldering reflow
โข Highly accelerated stress test (HAST)
โข Biased HAST
Fan XJ, Suhir, E. (eds.). Moisture Sensitivity of Plastic Packages of IC Devices. Springer, New York, 2010.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
86
Basic Concepts of Moisture Diffusion
%100airtheofpressurevaporSaturated
airtheofpressurevaporActualRH
โข What is the Relative Humidity (RH)? โ Defined as vapor pressure ratio associated with temperature T
โข Moisture concentration, C (x, t; T, RH) โ Mass of moisture per unit volume of substance.
โข Diffusion Coefficient / Diffusivity, D(T) โ Measures the rate of mass diffusion
โ Defined as the amount of mass flux per unit concentration gradient (m2/s)
โ A function of material and temperature
โข Saturated Moisture Concentration, Csat(RH, T) โ The maximum mass of moisture per unit volume of the substance kg/m3.
โข Solubility, S(T) โ Henryโs law โ The ability of the substance to absorb moisture
โ Defined as the maximum mass of moisture per unit volume of the substance per unit
pressure (kg/(m3Pa)).
โ A function of material and temperature
P
CS sat
where P = ambient pressure in given RH
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
87
87
Moisture Diffusion Modeling
C: Concentration, kg/m3; D0: diffusivity, m2/s
โข Fickian diffusion theory
CDt
C
0
โข Discontinuity at interface
Concentration discontinuity
CA
CB
After normalization
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
88
88
An Overview of Moisture Diffusion Modeling
Normalized
field variable
Csat: temperature-
dependent
RH: time-
dependent
Non-Henryโs
law
ANSYS ABAQUS
Galloway et al.
(1997) C/S
Wong et al.
(1998) C/Csat
Jang et al.
(1993) C/M
Wong et al.
(2016) C/Csat
Markus et al.
(2016) C
Chen et al.
(2017) aw
Ma et al.
(2019) C/K
Ma L, Joshi R, Newman K, X.J. Fan, Improved Finite Element Modeling of Moisture Diffusion Considering Discontinuity at Material
Interfaces in Electronic Packages, ECTC 2019
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
89
89
Water Activity Theory
โข Water activity aw is a continuous field variable derived from chemical potential.
โข Moisture flux in terms of water activity gradient.
โข Moisture concentration is given as
โข Moisture diffusion equation
โข Water activity is continuous and no normalization is required for multi-material system.
wii aDK ,0mJ
C = K aw
K: generalized solubility, kg/m3
wiiwi aDK
t
aK
,0
Chen LB, Zhou J, Chu HW, Zhang GQ, Fan XJ. Modeling Nonlinear Moisture Diffusion in Inhomogeneous Media. Microelectronics Reliability. 75 (2017) 162โ170. 2017.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
90
90
Current Moisture Diffusion Theory in ANSYS
โข ๐ถ = ๐ถ
๐ถsat is used as field variable.
where โ ๐ท โถ diffusivity matrix
โ C (x,y,z,t) : concentration
โ G: diffusing substance generation rate per unit volume
โ ๐ป: gradient operator=๐
๐๐ฅ
๐
๐๐ฆ
๐
๐๐ง , ๐ป โโถ divergence operator
๐ถ๐ ๐๐ก๐๐ถ
๐๐ก+ ๐ถ
๐๐ถ๐ ๐๐ก๐๐
๐๐
๐๐ก= ๐ป โ ๐ท ๐ถ๐ ๐๐ก๐ปC + ๐ถ
๐๐ถ๐ ๐๐ก ๐๐
๐ป๐ + ๐บ
๐ ๐ถ๐ ๐๐ก๐ถ
๐๐ก= ๐ป โ ๐ท๐ป ๐ถ๐ ๐๐ก๐ถ + ๐บ
โข In the above formulation, ๐๐ฌ๐๐ญ is considered as temperature-dependent only.
โข However, ๐๐ฌ๐๐ญ is both time- and temperature- dependent in general.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
91
91
Henryโs Law in Moisture Diffusion
โข ๐ถsat = ๐amb๐
pamb: ambient partial vapor pressure ๐amb = RH pg T = RH p0eโEp
RT pg is saturated vapor pressure)
S : solubility ๐ ๐ = ๐0 ๐๐ธ๐ ๐ ๐
๐พ0 = ๐0๐0, ๐ธ๐ = ๐ธ๐ โ ๐ธ๐
โข In general, Csat (RH (t), T(t)) is both time- and temperature-dependent.
โข Therefore, ANSYS cannot solve the problem with varying RH as function of time.
๐ถ๐ ๐๐ก = ๐ ๐ป ๐0๐0 ๐๐ธ๐ โ๐ธ๐๐ ๐ = ๐ ๐ป ๐พ0๐
โ๐ธ๐๐ ๐ = ๐ ๐ป ๐ก ๐พ ๐
๐พ = ๐พ ๐ = ๐พ0 exp โ๐ธ๐๐ ๐
Ep = 4.01 ร 104J
mol= 0.415 ev
p0 = 3.82 ร 1010 Pa .
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
92
92
New Normalization Theory - ๐ถ ๐พ Approach
โข The governing equation is exactly identical to the diffusion equation in ANSYS.
โข ๐ถ K turns out to be the water activity aw (Chen et al.), which has been proved to be continuous at interface.
โข ๐ถ ๐พ = ๐ถ
๐พ is used as field variable.
โข K(T): generalized solubility (kg/m3)
๐ ๐พ๐ถ ๐
๐๐ก= ๐ป โ ๐ท๐ป ๐พ๐ถ ๐
โ ๐๐๐ถ ๐๐๐ก
+ ๐ถ ๐๐๐พ
๐๐
๐๐
๐๐ก= ๐ป โ ๐ท ๐พ๐ป๐ถ ๐ + ๐ถ ๐
๐๐พ
๐๐๐ป๐
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
93
93
Analogy using ANSYS
โข In using ANSYS, โ C is replaced by C ๐พ.
โ Temperature-dependent material property Csat is replaced by general solubility K.
โ Boundary condition is now changed to C ๐พ= RH.
Field Variable Material Property
C Csat
C k K
C C k
Absorption 1 RH
Desorption 0 0
C C k
Absorption 0 0
Desorption 1 (if fully saturated initially)
1 (if fully saturated initially)
โข Field variable and material property input
โข Boundary condition โข Initial condition
๐ถ๐ ๐๐ก๐พ
= ๐ ๐ป ๐ก
๐พ =๐ถ๐ ๐๐ก๐ ๐ป
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
94
94
Element Options in ANSYS
โข Since the general diffusion equation in ANSYS involves with temperature derivative with respect to time when K is temperature-dependent, the coupled element (thermal diffusion, or thermal-structural-diffusion) must be used.
โข Two different types of element option โ Diffusion element
Plane 238 (2D, Quadratic)
Solid 239 (3D, Quadratic)
Degree of Freedom (DOF) label; CONC
โ Coupled element
Plane 223(2D , Quadratic)
Solid 226(3D , Quadratic)
o Thermal-diffusion (100010)
DOF label: TEMP,CONC
o Thermal-structural-diffusion (100011)
DOF label: TEMP, UX, UY, UZ , CONC
o Structural-diffusion (100001)
DOF label: UX,UY,UZ, CONC
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
95
95
Case Study โ Varying RH and Temperature
โข A 1-D bi-material absorption problem with varying RH and Temperature โ Initial condition: dry at 300C โ Ambient temperature profile is given below to simulate a
absorption process
T t 0C = 30 +50t
200 0 < t โค 200 min
โ Relative humidity profile is given below to simulate a absorption process
RH t = 0.3 +0.7t
200 0 < t โค 200 min
โข Material properties
Properties Mat A Mat B
D0 m2 s 5.0 ร 10โ3 4.0 ร 10โ3
ED ev 0.518 0.518
S0(Kg/m/Pa) 6.0 ร 10โ10 2.0 ร 10โ10
Es ev 0.383 0.362
โข For this problem, Csat is both time and temperature-dependent.
MAT A MAT B
-1 1(mm) 0
RH(t)
T(t) โ
โ
RH(t)
T(t) โ
โ
T(t)
300C t(min)
200
800C
RH(t)
30% t(min)
200
100%
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
96
96
C๐ฌ๐๐ญ as a Function of Temperature
โข For the given material properties (Csat at 30%RH)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
97
97
Finite Element Model and Results
L๐ด = 1 mm The starting point
of the path L๐ต = 1 mm The end point of
the path
h=5 ร10โ5m
โข Coupled thermal-diffusion element presents correct results.
โข 1-D problem is constructed by a 2-D element strip model.
โข Finite difference method (FDM) is used for verification.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
98
98
C vs. ๐ ๐ค Approach
โข Since Csat is temperature-dependent and time-dependent, the results
obtained by ๐ถ are incorrect.
๐พ๐๐ถ ๐๐๐ก
+ ๐ถ ๐๐๐พ
๐๐
๐๐
๐๐ก= ๐ป โ ๐ท ๐พ๐ป๐ถ ๐ + ๐ถ ๐
๐๐พ
๐๐๐ป๐
๐ถ๐ ๐๐ก๐๐ถ
๐๐ก+ ๐ถ
๐๐ถ๐ ๐๐ก๐๐
๐๐
๐๐ก+ ๐ถ
๐๐ถ๐ ๐๐ก๐๐ก
= ๐ป. ๐ท ๐ถ๐ ๐๐ก๐ปC + ๐ถ ๐๐ถ๐ ๐๐ก ๐๐
๐ป๐
โข ๐ถ (coupled element with thermal-diffusion)
โข ๐ถ ๐ (coupled element with thermal-diffusion)
Not included in
ANSYS
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
99
99
Comparison Among Different Elements in ANSYS
โข The Diffusion element only and the coupled element with structural-diffusion option present the same incorrect results.
๐พ๐๐ถ ๐๐๐ก
+ ๐ถ ๐๐๐พ
๐๐
๐๐
๐๐ก= ๐ป โ ๐ท ๐พ๐ป๐ถ ๐ + ๐ถ ๐
๐๐พ
๐๐๐ป๐
๐พ๐๐ถ ๐๐๐ก
= ๐ป โ ๐ท ๐พ๐ป๐ถ ๐
โข Diffusion element only
โข Coupled element with thermal-diffusion
๐พ๐๐ถ ๐๐๐ก
= ๐ป โ ๐ท ๐พ๐ป๐ถ ๐
โข Coupled element with structural-diffusion
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
100
100
Vapor Pressure Model - Example: Pressure Cooker
โข Vapor phase
Single vapor phase
T1 T2
p1 p2
โ Moisture in single vapor phase;
โ < g(T), : moisture density over the total volume of cooker; g(T): saturated
moisture vapor density.
โ Ideal gas law can be used: p2=p1T2/T1
โข Liquid/vapor phase
Mixed liquid/vapor phase
T1 T2
โ Moisture in two-phases โ water/vapor mixed;
โ > g(T) - liquid-vapor phase , : moisture density over the total volume
of cooker; g(T): saturated moisture vapor density
โ Saturated vapor pressure remains regardless of water level
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
101
101
Vapor Pressure Model
โข Moisture in free-volumes โfreeโ to vaporize
โข Two distinct states โ Single vapor phase
โ Mixed liquid/vapor phase
โข Saturated moisture density g(T)
Mixed liquid/vapor phase Single vapor phase
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
102
102
Vapor Pressure Model
CfMM
RTTp
OH
2
)( when C(T) / f < g(T)
p(T) = pg(T), when C(T) / f โฅ g(T)
fCV
V
V
m
V
m
ff
/d
d
d
d
d
d
dVf : void volume in a REV (representative elementary volume)
dV : total volume of a REV
dm : total moisture mass
f : void volume fraction dVf/dV
: apaprent moisture density
REV
Fan XJ, Zhou J, Zhang GQ, Ernst LJ. A micromechanics based vapor pressure model in electronic packages. ASME Journal of Electronic Packaging 127 (3), 262-267.
2005.
Xie B, Fan XJ, Shi XQ, Ding H. Direct concentration approach of moisture diffusion and whole field vapor pressure modeling for reflow process: part I โ theory and
numerical implementation. ASME Journal of Electronic Packaging 131(3), 031010. 2009.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
103
Steam Table - pg
T(C) 20 30 40 50 60 70 80
g(g/cm210
-3) 0.017 0.03 0.05 0.08 0.13 0.2 0.29
pg(MPa) 0.002 0.004 0.007 0.01 0.02 0.03 0.05
T(C) 90 100 110 120 130 140 150
g(g/cm210
-3) 0.42 0.6 0.83 1.12 1.5 1.97 2.55
pg(MPa) 0.07 0.1 0.14 0.2 0.27 0.36 0.48
T(C) 160 170 180 190 200 210 220
g(g/cm210
-3) 3.26 4.12 5.16 6.4 7.86 9.59 11.62
pg(MPa) 0.62 0.79 1 1.26 1.55 1.91 2.32
T(C) 230 240 250 260 270 280 290
g(g/cm210
-3) 14 16.76 19.99 23.73 28.1 33.19 39.16
pg(MPa) 2.8 3.35 3.98 4.69 5.51 6.42 7.45
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
104
104
Vapor Pressure: Effect of Reflow Profiles
Moisture diffusion modeling: a) SH, b) FM Vapor pressue modeling: a) SH, b) FM
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
105
CSP with thinner substrate
CSP with thicker substrate
โ About 50% reduction on vapor pressure at 250หC in the bottom DA film
between two thicknesses of substrate
Vapor Pressure Modeling
Xie B, Fan XJ, Shi XQ, Ding H. Direct concentration approach of moisture diffusion and whole field vapor pressure modeling for reflow process: part II โ
application to 3-D ultra-thin stacked-die chip scale packages. ASME Journal of Electronic Packaging 131(3), 031011. 2009.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
106
106
Hygroscopic Swelling
85ยฐC/85%RH
~0.29%
Expansion strain
due to hygroscopic
swelling
Expansion strain
due to temperature
change of 100C
~0.25%
hygrob*C thermala*DT
bโ the coefficient of
hygroscopic swelling
C โ moisture concentration
Hygroscopic mismatch is comparable to thermal mismatch in causing mechanical
stresses
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
107
107
Temperature-Moisture-Deformation Problems
โข How to perform integrated modeling to accurately capture stress developments in each stage?
T
Time t
Stress-free
Cooling
down
Thermal stress
Hygroscopic stress
Thermal
mismatch
Hygroscopic
shrinkage
Expansion by
vapor pressure Preconditioning at 85C85%RH
Reflow
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
108
108
Material Properties to be Needed
ijijkkijij pE
CTEE
ba
)21
(1
D
a: coefficient of thermal expansion (CTE)
b: coefficient of hygroscopic swelling (CHS)
C: moisture concentration
T: temperature, p: vapor pressure
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
109
109
Implementation in ANSYS
pE
CT
ba21volu
D
))21(
(volu
E
pCT
b
ba
D
BF,,TEMP,n_temp BF,,FLUE, b1+b2
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
110
110
Body Forces in ANSYS: Temp and Flue
โข Total volume strain
= a (T-Tref )+ bC
BF,,TEMP,n_temp BF,,FLUE,n_C
nsw CD b
sw :swelling strain
b :swelling coefficient
C :a fluence
TB,SWELL,1
! C72=10 FOR ACTIVATION OF USERSW
! EPS=C67*(FLUENCE)หC68 WHERE C67=SWELLING COEFFICIENT
TBDATA,72,10
TBDATA,67,b,1
ANSYS built-in swelling function
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
111
111
Implementation in ANSYS
TB,SWELL,MAT
TBDATA,72,10
TBDATA,67,b,1
ldread,temp,,,,,moisture,rth
*do,i,1,total_node
*get,C,node,i,ntemp
BF,i,FLUE,C
*enddo
BF,,Temp,T_reflow
solve
โข Moisture diffusion (including desorption) modeling needs to be solved first and separately. โ ANSYS 17.2 or later provides coupled element
Plane 223(2D , Quadratic) Solid 226(3D , Quadratic)
o Thermal-structural-diffusion (100011) DOF label: TEMP, UX, UY, UZ , CONC
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
112
112
Example: Bi-Material Configuration
Moisture
concentration
contours
Deformed
shape
contours
von Mises
stress
Contours in MC
Time = 1hour @85C/85%RH Time = 18hour @85C/85%RH
Max stress =4.96MPa Max stress =20.99MPa
Fan XJ, Zhao JH. Moisture diffusion and integrated stress analysis in encapsulated microelectronics devices. Proc. 12th. Int. Conf. on Thermal, Mechanical and
Multiphysics Simulation and Experiments in Micro-Electronics and Micro-Systems (EuroSimE 2011). pp8. Linz, Austria. April 18-20, 2011.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
113
113
Example: Reflow Profile Setting
T
Time t
85C
0.5C/s
200C
260C
2C/s
230s 30s 30s
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
114
114
Moisture Concentration and Vapor Pressure Contours
Moisture concentration contours
Vapor pressure contours
Max p = 1.55MPa
T=200C
Max p = 5.84MPa
T=260C Max p = 0.5e-3MPa
T=265C
Max C = 3.54mg/mm3
T=200C
Max C = 2.11mg/mm3
T=260C
Max C = 1.05mg/mm3
T=265C
Cu
MC
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
115
115
Results of Integrated Stress Modeling
Von Mises Stress in
Epoxy (MPa)
Von Mises Stress in
Cu (Mpa)
T 2.4 8.7
T + H 4.9 17.3
T + H + V 6.3 22.9
Von Mises Stress in
Epoxy (MPa)
Von Mises Stress in
Cu (Mpa)
T 6.0 21.9
T + H 7.2 27.5
T + H + V 11.2 38.9
Von Mises Stress in
Epoxy (MPa)
Von Mises Stress in
Cu (Mpa)
T 6.0 21.9
T + H 6.3 24.7
T + H + V 6.3 24.7
260C
260C after
30sec hold
200C
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
116
116
Summary
โข For a general moisture diffusion problem with temperature-dependent Csat and varying ambient RH and temperature with time โ C k must be used and the coupled element with thermal-diffusion or thermal-
structural-diffusion option must be applied at the same time.
โข ANSYS built-in ๐ช approach cannot solve the problem with varying RH correctly. โ ANSYS diffusion element only
Csat must be temperature-independent. Temperature gradient is not considered.
โ ANSYS coupled element with structural-diffusion option.
Csat must be temperature-independent. Temperature gradient is not considered.
โ ANSYS coupled element with thermal-diffusion option (or thermal-structural-diffusion) option
If ๐ถ is used, RH must be constant.
If C k is used, no restriction for any diffusion problems.
โข Vapor pressure model
โข Thermal-hygro-mechanical modeling
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
117
117
โขIntroduction
โขTemperature Loading
โขMechanical Loading
โขMoisture and Humidity
โขElectrical Current - Multi-Physics
Modeling
โขSummary
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
118
118
Outline
โข Electrical-Thermal-Mechanical Modeling
โข Electromigration
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
119
119
Electrical-Thermal Modeling
โข Governing equations
โข Current flow generates joule heating at any location.
โข Joule heat is applied as heat source in heat diffusion in solids.
โข Joule heat term:
โข We solve this multi-physics problem one step at a time โ Both problems are diffusion-type
โ Two problems are sequentially coupled if electric conductivity does not vary with temperature
02
2
2
2
2
2
z
V
y
V
x
V02 V
21VVj
0
1 22 VTkT
2
0
1Vq
Current flow:
Heat conduction: Joule heating
(electrostatics)
(steady-state)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
120
120
Example: A Microresistor Beam in MEMS
โข Current flow modeling
COMSOL Multi-Physics Modeling Tutorial.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
121
121
Thermal Modeling
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
122
122
Thermal-Mechanical Modeling
โข Governing equations
โข Temperature load is applied in the form of โbody forceโ in mechanical analysis
t
TcTk p
][
0]21
[,)( ,
2
ijii XTE
eGuG a
ui : the component of displacement vector
e : the total volumetric strain e=ui,i
Xi : the component of body force vector.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
123
123
Electrical-Thermal-Mechanical Modeling
โข MEMS: microresistor beam modeling
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
124
124
Mechanical Modeling
โข Results, Deformation
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
125
125
What is Electromigration (EM)?
โข Electromigration is the most persistent reliability problem in interconnect technology in semiconductor device.
Void Hillock
K.N. Tu: EPTC 2008 short course
Very high current density flows through
โข Electromigration is a process of mass transport in the current-carrying metal
under the driving forces generated by electric field.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
126
126
Similar to River Flowโฆ
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
127
Driving Forces
โข The atomic transport is caused by a combination of interacting
driving forces that can generate voids at different locations.
Atomic
Transport
Electron
Wind
Temperature
Gradient
Co
ncen
tratio
n
Gra
die
nt S
tress
Gra
die
nt
Electromigration
Thermomigration
Se
lf-diffu
sio
n Str
es
s-m
igra
tio
n
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
128
Blechโs Theory (1976)
Atomic
Transport
Electron
wind
Str
ess
gra
die
nt
Electromigration
Thermomigration
Str
es
s-m
igra
tio
n
Self-d
iffusio
n
I. Blech and C. Herring, APL, 29, 131,
1976.
โข EM flux is entirely balanced by the stress-induced counter flux.
โข Blech Product, jL, provides a threshold condition of maximum stress, below which, electromigration failure will not occur - mechanical failure.
๐โ๐๐๐ + ฮฉ๐๐
๐๐ฅ= 0 ๐๐ฟ =
๐๐๐๐ฅโ๐๐๐๐ ฮฉ
๐โ๐๐
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
129
Electromigration โ a Coupled, Multi-Physics Problem
โข EM is a multiphysics coupled field problem involved with electron
wind, chemical potential, stress gradient and temperature gradient.
๐ฑv = ๐ท๐ฃ๐ถ๐ฃ๐๐ต๐
๐โ๐๐๐ โ๐๐ต๐
๐ถ๐ฃ๐ป๐ถ๐ฃ โ ๐ฮฉ๐ปฯ +
๐โ
๐๐ป๐
Total flux in terms of Cv (vacancy concentration)
Cv vacancy concentration (m-3), ฮฉ volume of per atom (m3)
Dv diffusivity j current density (A/m2)
kB Boltzmannโs constant (J/K), e elementary charge (C)
Z* effective charge number(>0) , Q* heat of transport (kJ/mol)
f volume relaxation ratio
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
130
An Overview in Literature
Flux by self-
diffusion
Flux by
stress
Source/sin
k term
Constraint
condition
Stress
equilibrium
EM strain in
stress/strain
Shatzkes and
Lloyd (1986) N/A N/A N/A
Kirchheim (1992)
Korhonen et al.
(1993)
Clement and
Thompson (1995)
Sarychev et al.
(2000) N/A
Suo et al. (2003,
2011, 2014)
Sukharev et al.
(2004, 2007) N/A
Maniatty et al.
(2016)
โข Inconsistent and incomplete solutions appear in literature.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
131
New Theory โ General Coupling Model
Flux by self-
diffusion
Flux by
stress
Source/sin
k term
Constraint
condition
Stress
equilibrium
EM strain in
stress/strain
Cui et al. (2019)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
132
Puzzle One
โข Equilibrium (๐๐๐
๐ ๐= ๐) is violated in the existing literature theory and
solutions.
๐ = 0 (volume strain equals to zero)
anywhere.
๐ข ๐ฟ = ๐ข 0 = 0 โ ํ๐ฅ๐๐ฅ๐ฟ
0
= 0 โ ๐๐๐ฅ๐ฟ
0
= 0
Literature Present solution
Confined conductor configuration
๐๐๐
๐ ๐ ๐
โข Volume Strain in Confinement
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
133
Puzzle Two
โข With the exact -Cv equation, the stress level is significantly lower.
Literature Present solution
๐ =๐๐ต๐
ฮฉln
๐ถ๐ฃ๐ถ๐ฃ0
๐ =2๐ธ๐ด
9 1 โ ๐ฃ
1 + ๐ฃ
2 1 โ 2๐ฃ ๐ฟ ln
๐ถ๐ฃ๐ถ๐ฃ0
๐ฟ
0
๐๐ฅ + ln๐ถ๐ฃ๐ถ๐ฃ0
Confined configuration
C. Herring, J. Appl. Phys. 21, 437 (1950)
โข Hydrostatic Stress vs. Vacancy Concentration Cv
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
134
Puzzle Three
โข When concentration gradient is considered, the stress becomes
further lower.
Literature Present solution
๐๐ฃ = โ๐ท๐ฃ๐ถ๐ฃ๐๐ต๐
๐ฮฉ๐๐
๐๐ฅ+ ๐โ๐๐๐ ๐๐ฃ = โ
๐ท๐ฃ๐ถ๐ฃ๐๐ต๐
๐๐ต๐
๐ถ๐ฃ
๐๐ถ๐ฃ๐๐ฅ
+ ๐ฮฉ๐๐
๐๐ฅ+ ๐โ๐๐๐
๐ฑv = ๐ท๐ฃ๐ถ๐ฃ๐๐ต๐
๐โ๐๐๐ โ๐๐ต๐
๐ถ๐ฃ๐ป๐ถ๐ฃ โ ๐ฮฉ๐ปฯ
+ ๐โ
๐๐ป๐
โข Effect of Self-Diffusion
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
135
Puzzle Four
Present solution
โข Stress-Strain Relation (Constitutive Equation)
๐บ = ๐บ๐ + ๐บ๐ + ๐บ๐ธ๐ = 0
Literature
Electromigration-induced volume strain
ํ๐๐๐ธ๐ = โ
๐
๐ต
๐ = 2๐บ๐บ + ฮปtr ๐บ ๐ โ ๐ตtr ๐บ๐ ๐ ๐ ๐ถ๐ฃ = โ๐ด ln๐ถ๐ฃ
๐ถ๐ฃ0 ๐ด =
๐๐ต๐
๐ตฮฉ
ํ๐๐๐ธ๐ = ๐ ๐ถ๐ฃ
๐บ = ๐บ๐ + ๐บ๐ + ๐บ๐ธ๐
๐บ๐ = ๐ผโ๐๐, ๐บ๐ธ๐ =๐ ๐ถ๐ฃ
3๐, ๐ ๐ถ๐ =
โ ๐ด ln ๐ถ๐ฃ
๐ถ๐ฃ0
๐ = 2๐บ๐บ + ฮปtr ๐บ ๐ โ ๐ตtr ๐บ๐ ๐ โ ๐ตtr ๐บ๐ธ๐ ๐
โข A 3-D, general and self-consistent stress-strain constitutive
equation is obtained.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
136
Puzzle Five
Literature Present solution
๐๐
๐๐ก= ฮฉ ๐ป โ ๐ฑ๐ฃ
๐ = ํ๐๐
๐ + ํ๐๐
๐ + ํ๐๐
๐ธ๐
๐ํ๐๐๐
๐๐ก+๐ํ๐๐
๐
๐๐ก+๐ํ๐๐
๐ธ๐
๐๐ก= ฮฉ๐ป โ ๐ฑ๐ฃ ๐ํ๐๐
๐ธ๐
๐๐ก= ฮฉ๐ป โ ๐ฑ๐ฃ
โข Atomic Transport Equation
๐๐ถ๐ฃ๐๐ก
+ ๐ป โ ๐ฑ๐ฃ = G
G = ๐๐ถ
๐๐ก
๐๐ถ๐ฃ๐๐ก
โช๐๐ถ
๐๐ก
๐๐ถ
๐ถ= ๐ํ
๐๐
๐ธ๐
with
approximation
without any approximation
(sink/source
term)
โข Mass conservation equation is used to describe the atomic
transport. Thus, the sink/source term is considered naturally.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
137
New Theory โ General Coupling Model
๐๐
๐๐ก= ฮฉ ๐ป โ ๐ฑ๐ฃ
๐ฑ๐ฃ = โ๐ท๐ฃ๐ป๐ถ๐ฃ + ๐ท๐ฃ๐ถ๐ฃ๐โ๐๐๐
๐๐ต๐ โ ๐ท๐ฃ๐ถ๐ฃ
ฮฉ
๐๐ต๐๐ป๐ + ๐ท๐ฃ๐ถ๐ฃ
๐โ
๐๐ต๐๐ป๐
๐ = tr ๐บ , ๐บ = ๐บ๐ + ๐บ๐ + ๐บ๐ธ๐
๐บ๐ = ๐ผโ๐๐, ๐บ๐ธ๐ =๐ ๐ถ๐ฃ
3๐, ๐ ๐ถ๐ = โ๐ด ln
๐ถ๐ฃ
๐ถ๐ฃ0
๐ = 2๐บ๐บ + ฮปtr ๐บ ๐ โ ๐ตtr ๐บ๐ ๐ โ ๐ตtr ๐บ๐ธ๐ ๐
๐ = tr ๐
3
๐ป โ ๐ + ๐ญ = 0 ๐บ =1
2 ๐ป๐ + ๐๐ป
๐ป โ ๐ = 0, ๐ =๐ฌ
๐= โ
๐ปV
๐
๐๐ป2๐ + ๐ โ ๐ฌ = 0
Mass conservation equation:
Constitutive equation:
Field equations:
Cui Z, et al., JAP, 125,
2019
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
138
138
Totally fixed configuration โ 1-D problem
โข Assumptions
โ1-D problem: ํ๐ฆ = ํ๐ง = ํ๐ฅ๐ฆ= ํ๐ฆ๐ง= ํ๐ฅ๐ง= 0
โCurrent density j is constant and in negative x-axis direction.
โTemperature is constant โ no temperature gradient.
โPerfectly blocking condition for Cv: Jv (0,t)= Jv (L,t)= 0.
โMechanically totally fixed:
๐ข ๐ฟ, ๐ก = ๐ข 0, ๐ก = 0
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
139
139
1-D Governing Equations
๐๐ถ๐ฃ๐๐ก
โ1
๐ฟ
๐๐ถ๐ฃ๐๐ก
๐ฟ
0
๐๐ฅ =3 1 โ ๐ฃ ๐ท๐ฃ๐ถ๐ฃฮฉ
1 + ๐ฃ ๐ด1 +
2๐ธ๐ดฮฉ
9 1 โ ๐ฃ ๐๐ต๐
๐2๐ถ๐ฃ๐๐ฅ2
+๐โ๐๐๐
๐๐ต๐
๐๐ถ๐ฃ๐๐ฅ
๐ =2๐ธ๐ด
9 1 โ ๐ฃ
1 + ๐ฃ
2 1 โ 2๐ฃ ๐ฟ ln
๐ถ๐ฃ๐ถ๐ฃ0
๐๐ฅ๐ฟ
0
+ ln ๐ถ๐ฃ๐ถ๐ฃ0
๐๐
๐๐ก+ ๐ท๐ฃฮฉ
๐๐๐ถ๐ฃ๐๐ฅ
+๐ฮฉ๐ถ๐ฃ๐๐ต๐
๐๐๐๐ฅ
+๐โ๐๐๐๐ถ๐ฃ๐๐ต๐
๐๐ฅ = 0
๐ = โ๐ด ln ๐ถ๐ฃ๐ถ๐ฃ0
+3 1 โ 2๐ฃ ๐
๐ธ
๐ =2๐ธ๐ด
9 1 โ ๐ฃ ln
๐ถ๐ฃ๐ถ๐ฃ0
+1 + ๐ฃ
3 1 โ ๐ฃ ๐๐ฅ
๐๐ฅ = 1 โ ๐ฃ ๐ธ
1 + ๐ฃ 1 โ 2๐ฃ ํ๐ฅ +
๐ธ๐ด
3 1 โ 2๐ฃ ln
๐ถ๐ฃ๐ถ๐ฃ0
๐๐๐ฅ๐๐ฅ
= 0
ํ๐ฅ๐๐ฅ๐ฟ
0
= 0 ๐ข ๐ฟ, ๐ก = ๐ข 0, ๐ก = 0 โ ํ๐ฅ๐๐ฅ๐ฟ
0
= 0 โ ๐๐๐ฅ๐ฟ
0
= 0
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
140
140
Parameters and Physical Properties
Length of interconnect (L) 50 mm Temperature (T) 500K
Youngโs modulus (E) 70109 Pa Poisson ratio (ฮฝ) 0.3
Atomic diffusivity (Da) 310-16 m2/s Vacancy diffusivity (Dv) 310-9 m2/s
Atomic volume (ฮฉ) 1.6610-29 m3 Initial vacancy
concentration (Cv0)
6.021021 m-3
๐ท๐ฃ๐ถ๐ฃ = ๐ท๐๐ถ๐, ๐ถ๐ = 1/ฮฉ Electrical resistivity () 4.8810-8 Ohmยทm
Current density (j) 1010 A/m2 Elementary charge (e) 1.6010-19 C
Effective charge number (Z*) 3.5 Boltzmann constant (kB) 1.38 ร 10โ23J/K
Coefficient of
electromigration strain (A) 0.0071
Cui Z, et al., JAP, 125, 2019
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
141
Numerical Results ()
โข The present results show significant differences with the previous
results, particularly in the magnitude of hydrostatic stress. (113
MPa vs. 350 MPa: 30% of the stress value obtained).
350 MPa
113 MPa
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
142
Numerical Results (Cv)
โข Vacancy concentration is comparable with the previous results, even
though stress is significantly lower.
2.4
2.1
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
143
Effect of Self-Diffusion
โข Self-diffusion plays more important role than stress in electromigration.
โข Normalized flux term,
๐ฝ1 =๐น๐ธ ๐ฅ, ๐ก
๐โ๐๐๐, ๐ฝ2=
๐น๐ ๐ฅ, ๐ก
๐โ๐๐๐, ๐ฝ3=
๐น๐ ๐ฅ, ๐ก
๐โ๐๐๐
where
๐น๐ธ = ๐โ๐๐๐, ๐น๐ = โฮฉ๐๐
๐๐ฅ, ๐น๐ =
โ๐๐ต๐
๐ถ๐ฃ
๐๐ถ๐ฃ
๐๐ฅ
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
144
Numerical Results: Stress-Free Condition
โข Electromigration happens on the same time-scale.
โข Electromigration is resisted by the counter force due to concentration
gradient.
๐๐ถ๐ฃ๐๐ก
=๐ท๐ฃ๐ถ๐ฃฮฉ
๐ด
๐2๐ถ๐ฃ๐๐ฅ2 +
๐โ๐๐๐
๐๐ต๐
๐๐ถ๐ฃ๐๐ฅ
โข Governing equations,
๐ = 0
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
145
Stress-Free Condition (Lloydโs Results)
โข Electromigration happens in miliseconds.
โข 107 orders difference compared to the confined configuration.
โข Governing
equations
M. Shatzkes and J. Lloyd, JAP, 59, 3890, 1986.
๐๐ถ๐ฃ๐๐ก
= ๐ท๐ฃ
๐2๐ถ๐ฃ๐๐ฅ2 +
๐โ๐๐๐
๐๐ต๐
๐๐ถ๐ฃ๐๐ฅ
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
146
Revisit โ Blechโs Theory
โข The effect of thermomigration is not taken into the consideration for
the purpose of comparison.
Blechโs theory (1976) Cui, Fan, Zhang (2019)
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
147
Blech Theoryโs Extension
โข The product of jL, is obtained for both confined and stress-free conditions.
โข The vacancy concentration is used as threshold condition for failure
criterion.
๐๐ฟ๐ต =2๐ธ๐ด๐ฮฉ + 9 1 โ ๐ฃ ๐๐ต๐
9 1 โ ๐ฃ ๐โ๐๐ln
๐ถ๐ฃ,๐๐๐ฅ
๐ถ๐ฃ,๐๐๐
1 +2๐ธ๐ด๐ฮฉ
9 1 โ ๐ฃ ๐๐ต๐ ๐2๐ถ๐ฃ๐๐ฅ2 +
๐โ๐๐๐
๐๐ต๐
๐๐ถ๐ฃ๐๐ฅ
= 0
โข Confined Configuration
๐2๐ถ๐ฃ
๐๐ฅ2+
๐โ๐๐๐
๐๐ต๐
๐๐ถ๐ฃ
๐๐ฅ= 0
๐๐ฟ๐ต =๐๐ต๐
๐โ๐๐ln
๐ถ๐ฃ,๐๐๐ฅ
๐ถ๐ฃ,๐๐๐
โข Stress-free Configuration
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
148
Threshold Product Curve
โข The confined metal line can sustain the higher current density than
that of stress-free condition.
โข The experimental results are consistent with the present theoretical
predictions. โ In Blechโs experiment, the measured threshold product for the uncovered metal line is lower
than that for the confined metal line. I. A. Blech, JAP. 47, 1203,1976.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
149
149
Summary
โข Electrical-thermal-mechanical modeling โ sequential approach
โข Electromigration โ A 3-D general coupling model for electromigration is developed.
โ 1D solution for the confined configuration is obtained.
โ Mechanical stress-based failure criterion may not be valid anymore.
โ Blechโs theory is revisited and reanalyzed.
โ Vacancy concentration, instead of hydrostatic stress, used for failure criterion โ a departure from the original view.
โ The new predictions are consistent with experimental observations.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
150
150
โขIntroduction
โขTemperature Loading
โขMechanical Loading
โขMoisture Loading
โขElectrical Current Loading - Multi-Physics
Modeling
โขSummary
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
151
151
Module 1 โ Temperature Loading
โข Thermal mismatch vs. temperature gradient
โข Analytical solution โLayered structure (stress, warpage, effective CTE)
โCylindrical structure (TSV)
โข Die-level thermal stress โ thermal stress in TSV
โข Package-level thermal stress problem โ warpage
โข Chip-package interaction (CPI) โ submodeling technique
โข Board level thermal stress problem โSolder ball thermal cycling (Flip chip BGA, WLP)
โCreep equations
โBest method for practice
Initial stress free condition; full model vs. global/local model; worst solder ball location, volume averaging
โข Stress singularity of joint materials
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
152
Module 2 โ Mechanical Loading
โข JEDEC drop test standard โJESD22-B111, old one, with 15 components
โJESD22-B111A, new one, with 4 components or 1 component
โข Finite element modeling โ Input G method, large mass method, input displacement method, direct
acceleration method
โGlobal/local modeling
โPeel stress used as indicator for failure
โข Four-point bending test and modeling โGlobal/local modeling
โGlobal with linear elastic but nonlinear geometry analysis
โLocal model with elastic-plastic modeling
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
153
153
Module 3 โ Moisture Diffusion and Vapor Pressure
โข For a general moisture diffusion problem with temperature-dependent Csat and varying ambient RH and temperature with time โ C k must be used and the coupled element with thermal-diffusion or thermal-
structural-diffusion option must be applied at the same time.
โข ANSYS built-in ๐ช approach cannot solve the problem with varying RH correctly. โ ANSYS diffusion element only
Csat must be temperature-independent. Temperature gradient is not considered.
โ ANSYS coupled element with structural-diffusion option.
Csat must be temperature-independent. Temperature gradient is not considered.
โ ANSYS coupled element with thermal-diffusion option (or thermal-structural-diffusion) option
If ๐ถ is used, RH must be constant.
If C k is used, no restriction for any diffusion problems.
โข Vapor pressure model.
โข Thermal-hygro-mechanical modeling.
IEEE EPTC 2019 โ Singapore December 4 โ 6, 2019 Xuejun Fan and Ricky Lee
154
154
Module 4 โ Electrical Current Loading
โข Electrical-thermal-mechanical modeling โ sequential approach
โข Electromigration โA 3-D general coupling model for electromigration is developed.
โ1D solution for the confined configuration is obtained.
โMechanical stress-based failure criterion may not be valid anymore.
โBlechโs theory is revisited and reanalyzed.
โVacancy concentration, instead of hydrostatic stress, used for failure criterion โ a departure from the original view.
โThe new predictions are consistent with experimental observations.
IEEE EPTC 2019 โ Singapore December 4 - 6, 2019
Questions?
Thank you for your attention.