:)<-\\'J^

RADC-TR-89-281
In-House Report
November 1989

RELIABILITY ASSESSMENT USING

FINITE ELEMENT TECHNIQUES

Gretchen A. Bivens

APPROVED FOR PUBLIC RELEASE; DISTRIBUTION UNLIMITED.

ROME AIR DEVELOPMENT CENTER

Air Force Systems Command
Griffiss Air Force Base, NY 13441-5700
 This report has been reviewed by the RADC Public Affairs Office (PA) and is
releasable to the National Technical Information Service (NTIS). At NTIS it will be
releasable to the general public, including foreign nations.

RADC TR-89-281 has been reviewed and is approved for publication.

™„
^ 3 L ^ V U ^ - - W / ~ \ j 5 dkjLLXK

ANTHONY J FEDUCCIA
Chief, Reliability & Engineering Division
Directorate of Reliability and Compatibility

APPROVED:

\ s u ^ v ij<x4^r
JOHN J. BART
Technical Director
Directorate of Reliability and Compatibility

FOR THE COMMANDER:

rn^p
JAMES W. HYDE III
Directorate of Plans and Programs

If your address has changed or if you wish to be removed from the RADC mailing list,
or if the addressee is no longer employed by your organization, please notify RADC
(RBES) Griffiss AFB NY 13^1-5700. This will assist us in maintaining a current
mailing list.

Do not return copies of this report unless contractual obligations or notices on a

specific document requires that it be returned.
 UNCLASSIFIED
SECURITY CLASSIFICATION OF THIS PAGE
Form Approved
REPORT DOCUMENTATION PAGE OMB No. 0704-0188

1a. REPORT SECURITY CLASSIFICATION 1b. RESTRICTIVE MARKINGS

UNCLASSIFIED N/A
2a. SECURITY CLASSIFICATION AUTHORITY 3. DISTRIBUTION/AVAILABILITY OF REPORT
NJA.
2b. DECLASSIFICATION /DOWNGRADING SCHEDULE
Approved for public release;
distribution unlimited.
N/A
4. PERFORMING ORGANIZATION REPORT NUMBER(S) 5. MONITORING ORGANIZATION REPORT NUMBER(S)
RADC-TR-89-281 N/A

6a. NAME OF PERFORMING ORGANIZATION 6b. OFFICE SYMBOL 7a. NAME OF MONITORING ORGANIZATION
Rome Air Development Center (If applicable) Rome Air Development Center (RBES)
RBES
6c ADDRESS {City, State, and ZIP Code) 7b. ADDRESS (City, State, and ZIP Code)

Griffiss AFB NY 13*41-5700 Griffiss AFB NY 13**1-5700

8a. NAME OF FUNDING /SPONSORING 8b. OFFICE SYMBOL 9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER
ORGANIZATION (If applicable)
Rome Air Development Center RBES N/A
8c ADDRESS (City, State, and ZIP Code) 10. SOURCE OF FUNDING NUMBERS
PROGRAM PROJECT TASK WORK UNIT
Griffiss AFB NY 13**1-5700 ELEMENT NO. NO. NO ACCESSION NO.
62702F 2338 02 2X
11. TITLE (Include Security Classification)
RELIABILITY ASSESSMENT USING FINITE ELEMENT TECHNIQUES
12. PERSONAL AUTHOR(S)
Gretchen A. Bivens
13a. TYjPE OF REPORT 13b. TIME COVERED 14. DATE OF REPORT (Year, Month, Day) 15. PAGE COUNT
-[ouse
Tn-Hor-- FROM .Ian 8 8 T O l u l 8 8 November 1989 76
16. SUPPLEMENTARY NOTATION
N/A

17. COSATI CODES 18. SUBJECT TERMS (Continue on reverse if necessary and identify by block number)
FIELD GROUP SUB-GROUP Reliability Prediction Techniques
1* 0* Finite Element Analysis Material Failure
20 13
19. ABSTRACT (Continue on reverie if necessary and identify by block number)

Finite element analysis (FEA) is a computer simulation technique that is used to predict the thermal
and mechanical response of a structure. Rome Air Development Center (RADC) has used FEA to
predict the response in various electronic assemblies. This particular study was concerned with the
reliability of surface mounted microwire boards. A FEA of this assembly was simulated and the results
used to estimate its useful l i f e . This report gives a detailed acccount of the factors to be considered
when performing a FEA and the procedure used to transfer the results to a reliability figure-of-merit.

20. DISTRIBUTION/AVAILABILITY OF ABSTRACT 21. ABSTRACT SECURITY CLASSIFICATION

C UNCLASSIFIED/UNLIMITED D SAME AS RPT. D OTIC USERS UNCLASSIFIED
22a. NAME OF RESPONSIBLE INDIVIDUAL 22b. TELEPHONE (Include Area Code) 22c. OFFICE SYMBOL
GRETCHEN A. BIVENS (315) 330-4891 TJABCi
•"""""
DD Form 1473. JUN 86 Previous editions are obsolete. SECURITY CLASSIFICATION OF THIS PAGE

UNCLASSIFIED
UNCLASSIFIED

UNCLASSIFIED
 TABLE OF CONTENTS

Chapter Title Page

0.0 EXECUTIVE SUMMARY 1

1.0 INTRODUCTION 3

2.0 RELIABILITY CONCERNS IN ELECTRONIC ASSEMBLIES 5

3.0 STUDY METHOD 10

4.0 FINITE ELEMENT MODELING 20

4.1 MODELING OF COMPLIANT CONNECTIONS 21

4.1.1 SYSTEMS MODELS 21

4.1.2 BOARD MODEL 32

4.1.3 COMPLIANT CONNECTION MODELS 32

4.2 MODELING OF COMPLEX CIRCULAR GEOMETRIES 40

4.3 LESSONS LEARNED 46

5.0 TRANSFER OF FINITE ELEMENT RESULTS 48

5.1 S-LEAD CONNECTION 49

5.2 LEADLESS CONNECTION 52

5.3 GULL-WING CONNECTION 54

5.4 MICROWIRE/PTH CONNECTION 57

6.0 CONCLUSIONS 63

7.0 REFERENCES 66

i l l
 LIST OF ILLUSTRATIONS

Figure Title Page

1 Four Surface Mounted Package Designs 6
2 Microwire Construction 7
3 Microwire Construction 8
4 Microwire Construction 8
5 3-D Solid Elements 11
6 Typical One-Quarter Geometry 12
7 Stress-Strain Curve for a Ductile Material 14
8 Bounded Plastic Region 15
9 Coffin-Manson Model 16
10 Coffin-Manson Curve for Copper 18
11 Coffin-Manson Curve for Solder 19
12 S-Lead System FEM Generation 23
13 S-Lead System Superelement Model 24
14 S-Lead System FEM 25
15 Leadless System FEM Generation 26
16 Leadless System Superelement Model 27
17 Leadless System FEM 28
18 Gull-Wing System FEM Generation 29
19 Gull-Wing System Superelement Model 30
20 Gull-Wing System FEM 31
21 Microwire Board FEM 32
22 S-Lead Superelement Model 33
23 S-Lead FEM 34
24 Solder Joint: X-Y Plane 35
25 Solder Joint: X-Z Plane 36
26 Solder Joint: Y-Z Plane 37
27 Leadless FEM 38
28 Gull-Wing Superelement Model 39
29 Gull-Wing FEM 39
30 Wire Within a Block 41
31 Meshed Block 41
32 Mierowire/PTH FEM Generation 43

IV
33 Microwire/PTH FEM 44
34 Modified Microwire/PTH FEM (20 Mil Via) 45
35 Modified Microwire/PTH FEM (10 Mil Via) 45
36 Stress Contours for S-Lead FEM 50
37 Stress Contours for Leadless FEM 53
38 Stress Contours for Gull-Wing FEM 55
39 Stress Contours for Microwire/PTH (20 Mil Via) FEM 58
40 Stress Contours for Microwire/PTH (10 Mil Via) FEM 59

v
0.0 EXECUTIVE SUMMARY

Finite element analyses (FEA) are increasingly being used to predict

the mechanical response of an electronic assembly exposed to thermal and

vibrational environments. The results of these analyses are used to esti-

mate a component's life. This technique provides engineers with a method

of assessing the reliability of emerging technologies associated with

electronic systems. This particular investigation was concerned with the

reliability of surface mounted solder interconnections and microwire

board's plated-thru-hole (PTH) connections when exposed to severe thermal

environments. These two interconnections were analyzed using FEA and an

estimate of the number of cycles to failure was made.

There are many factors that must be considered before performing a

finite element analysis. Are the materials relatively well known and the

proper material properties available for this analysis? Are the boundary

conditions, input conditions and environment of the system known? How

complex is the geometry to be modeled? Can any assumptions be made to

simplify the model? It is important that the engineer performing the

analysis be aware of the information available in order to decide whether

an analysis is feasible and if so, what the best approach to performing the

study would be.

When modeling a new, complex system, preliminary modeling becomes

important. Investing this initial effort reduces the overall analysis

time and cost, and increases the results' accuracy. Examples of

1
preliminary modeling which were performed in this investigation are docu-

mented in the body of this report.

The results of the FEA were plotted on either the Coffin-Manson curve

for copper or for solder, depending on the material being analyzed. The

Coffin-Manson curve relates the material's strain to the number of cycles

to failure. This cycles to failure value gives an estimation of the life

of these connections. This method of transferring FEA results to an

estimated life allows the design and reliability engineer to locate areas

of reliability concern and make the corresponding corrections early in the

program. This reduces the overall cost of the program and allows the

reliability of the system to be built in. This technique is especially

important for military systems incorporating new technologies.

2
1.0 INTRODUCTION

RADC has used finite element analysis (FEA) as a way to assess the

reliability of emerging technologies. In the past, the resulting stress

values were taken from the FEA results and compared with the strength of

the corresponding materials. It was soon realized, however, that a tech-

nique was needed to transfer the numerical results from the FEA to a

measure of useful life for the system being analyzed. This report docu-

ments a reliability assessment of the areas of reliability concern in

electronic assemblies through the use of FEA and a transfer of FEA results

to an estimation of the useful life for the areas of concern.

Two areas of reliability concern were identified for the electronic

assembly analyzed. The first area was the solder connection between a

surface mounted device and its connecting board. Current use of these

devices has shown that these connections tend to fail soon after the system

has been installed. Three different lead designs were evaluated, the S-

lead, leadless, and gull-wing designs. The second area of concern was the

interface connection between the microwire and the PTH. Microwire layer-

ing is a new technology that is being used to replace conventional signal

layers in a surface mounted board. The connection with the PTH is critical

to the use of microwire technology.

Once the areas of reliability concern are identified, the evaluation

engineer determines the best way to develop the finite element model. A

geometric model representing the physical system is generated and a data

3
file which describes the materials used, the boundary conditions, and the

input conditions is created. This data file is read by the finite element

simulation code and a thermal analysis performed. From this analysis, the

strain range for the material of concern is obtained. Once the strain

range in a monotonic thermal cycle is known, theoretical relations are used

to estimate the number of thermal cycles this material would survive before

failure.

4
2.0 RELIABILITY CONCERNS IN ELECTRONIC ASSEMBLIES

As technology has advanced and electronic assemblies become more com-

plex, there is an increasing concern pertaining to the reliability of these

systems. A highly specialized system that cannot perform its function is

useless to the overall system. This is very critical when the system is a

military aircraft or spacecraft. However, in order to implement many of

the emerging technologies, unique packaging designs must be developed. If

these unique designs are analyzed in the initial phase of a program,

considerable time and money are saved.

The military was concerned with the reliability of surface mounted

multilayer assemblies investigated in this study. These boards are using

new mounting techniques and large packaging densities which have led to

problems in the reliability of the interconnects. In particular, the

failure of the leads of the surface mounted components is currently a

driving failure mechanism in the reliability of these boards. Figure 1

shows typical surface mounted packages for four different lead designs.

Factors which have been shown to effect the solder connection failure

include the lead design, package size, board material and heat removal.

In order to minimize the solder connection failure mode, board mate-

rials which reduce the board's horizontal expansion were used. This reduc-

tion has caused an increase in the board's vertical expansion which has led

to an increase in PTH connection failures. This assembly used microwire

5
Jn B
23
S
n
J=L
n UJ

II

si en
a
•H
en

<u
bO
CO
o
(0
&<

£
UJ
*U CJ
CO
4-1

I
2 I W
1
SM s &4

<
<U
CO
•H
&4
1
o
<
i
CO

tn
boards which have a reduced vertical height of the board and thus a reduc-

tion in the overall vertical expansion of the board. There is a relia-

bility concern, however, with the expansion that occurs at the

microwire/PTH connection interface.

Microwires are insulated copper wires which are ultrasonically bonded

to a dimensionally stable core (Figure 2 ) . These wires are then encap-

sulated in a resin to form the microwire boards. PTHs, known as vias, are

then laser drilled and the insulating material and encapsulating resin

surrounding the copper wire is evaporated (Figure 3 ) . The hole is then

plated with copper and the interconnect between the PTH and the microwire

is formed (Figure 4 ) . This interconnect region is the second area of

reliability concern.

Figure 2: Microwire Construction

 Laser Via

Figure 3: Microwire Construction

Surface Pad

PLATED INTERCONNECTION

Figure 4: Microwire Construction

8
 Before these interconnect problems became so prevalent, the Air Force

assumed that adequate heat transfer and low package thermal resistance

values guaranteed reliable assemblies. Studies at RADC, other DoD organi-

zations and contractors have shown that this is not sufficient criteria for

surface mounted assemblies due to the interconnect failure mode. The

overriding problem with the reliability of interconnects is cyclic

fatigue; that is, as these devices and connecting boards expand and con-

tract, the interconnects between them are forced to expand and contract as

well. Since these interconnects have become extremely small in size, it

does not take very much expansion to cause a significant amount of strain

to occur. Once the device and board return to the original state, so does

the interconnect. A problem occurs because the materials used in these

interconnects, solder and copper, tend to fail rapidly even when exposed to

small amounts of cyclic strains.

9
3.0 STUDY METHOD

FEA is the technique currently used at RADC to evaluate the relia-

bility of new microcircuit packaging designs. Over the past five years,

RADC has used FEA to perform reliability evaluations on VHSIC, gallium

arsenide, wafer scale integration, and surface mounted devices. The find-

ings in these evaluations have had excellent correlation with the results

from laboratory testing and field demonstration. Furthermore, FEA is

increasingly being used in industry and other DoD organizations to design

and evaluate microcircuit packages.

The FEA code that is used at RADC is Numerically Integrated Elements

for System Analysis (NISA). NISA was developed by Engineering Mechanics

Research Corporation, Troy MI. This program is used to perform a wide

range of analyses including static stress, dynamic, buckling and heat

transfer. In this study, NISA's static stress module was utilized to

simulate the response of the surface mounted boards to varying temperature

conditions.

In order to predict the physical response of the surface mounted

microwire boards, finite element models (FEMs) were generated and a tem-

perature change was simulated. FEMs are geometrical models which repre-

sent the actual physical system. The FEM is defined by two physical

parameters, nodes and elements.

10
 A node is a grid point located in space. In this study, the models

were created in three dimensions and, therefore, each node was defined by

three coordinates, (x, y, z) . The second parameter, the element, is

defined by a material index number and a set of nodes. Each index number

has specific material properties associated with it. The properties can be

defined to be constant or to vary with temperature. Two of the materials

used in this analysis had temperature-dependent properties and an equation

relating the material property versus a temperature change was assigned to

the material index number corresponding to those materials. The elements

used in this study were three-dimensional (3-D), six-sided, solid ele-

ments. Two types of 3-D elements were utilized (Figure 5 ) . The first type

was identified by eight corner nodes. These elements are usually rectan-

gular in shape, but do not have to be. This type of element uses linear

equations to define the lines connecting the nodes. The second type of

element is defined by eight corner nodes and twelve midside nodes. This

type uses parabolic equations to define the lines connecting the nodes.

This allows curved surfaces to be accurately modeled. Using 20 noded

elements significantly increases the complexity of the model and the simu-

lation.

Figure 5: 3-D Solid1 Elements

11
 Figure 6: Typical One-Quarter Geometry-

It is important when designing a FEM to keep its complexity to a

minimum. One common way of reducing the model's complexity is to take

advantage of the physical device's symmetrical geometry and boundary con-

ditions. The FEMs developed in this study represent one quarter of the

actual assembly. The boundaries at these symmetrical "cut" surfaces had to

be constrained in the direction perpendicular to these surfaces. This

prevented movement into the missing quarters. An example of a one-quarter

geometry is displayed in Figure 6.

12
 When choosing the temperature cycling range, many factors were con-

sidered. The most important was the environment that the microwire assem-

blies were experiencing in the field. The second was the temperature at

which the assemblies are manufactured. This value was considered the

stress free temperature of the assembly. All of the models were cycled

twice, once from the stress free temperature to the maximum temperature

(high temperature cycle) and once from the stress free temperature to the

minimum temperature (low temperature cycle). The output strain from each

of these analyses were added together to obtain the material's total

strain.

When calculating the total strain in an analysis, the following pro-

cedure is used. First, the maximum Von-Mises stress in the ductile mate-

rial (copper or solder) is tabulated. The Von-Mises stress value is a

combined stress value used when predicting failure. If this stress is

greater than the yield stress, then the elastic strain and the plastic

strain must both be calculated. A typical stress/strain curve for a

ductile material is shown in Figure 7. The strain in the elastic region is

the elastic stress divided by the modulus of elasticity. The strain in the

plastic region is then estimated using two different approaches. One

approach underestimates the plastic strain, the other approach overesti-

mates the strain and the actual strain is assumed to be in between these

two values. Figure 8 shows how the first approach estimates the plastic

strain by extending the straight stress/strain line from the elastic

region into the plastic region. The second approach takes twice the

13
plastic strain that is estimated in the first approach and adds this to the

elastic strain, thus giving the lower bound of the region.

ELASTIC
A DEFORMATION

CO
CO
LU
CL
I-
co

STRAIN

Figure 7: Stress-Strain Curve For A Ductile Material

14
 STRAIN
Figure 8: Bounded Plastic Region

The total strain is calculated by adding the elastic and plastic

strains for the two temperature cycles. This total strain value is trans-

ferred into a number of cycles to failure value using the Coffin-Manson

Curve for the specific material. This curve is generated from the Coffin-

Manson model displayed in Figure 9. In order to use this formula, the

coefficients associated with the material of interest must be known. The

15
 A± = JIL <2Nf) + €f ( 2 N f ) c

ELASTIC PLASTIC
STRAIN STRAIN

PARAMETER COPPER SOLDER

a
f 50,050 psi 7,350 ps
.3 .325
6
E 16xl0 psi 1.09x10*
b -.05 -.05
c -.6 -.5

FATIGUE STRENGTH COEFFICIENT

FATIGUE DUCTILITY COEFFICIENT
E MODULUS OF ELASTICITY
b FATIGUE STRENGTH EXPONENT
FATIGUE DUCTILITY EXPONENT

Figure 9: Coffin-Manson Model

16
values for the copper and solder are listed in Figure 9. These values were

inserted into the model and the corresponding curves were generated,

Figures 10 and 11. Each figure has two lines and one curve. The lines

represent the elastic strain component and the plastic strain component

and the curve represents the combined strain. In order to determine the

number of cycles to failure for a given strain, the total strain value for

this copper or solder is plotted onto the corresponding curve and the

number of cycles to failure value is extrapolated.

17
 IT3S
^yg^ja^g^ii "S|
m &
- i — > - £ •-••
! ' | . ' ' \ / ' - • — •• :

:
- - - | - -
-(• iBf—-, •' -•:- i s ~- • - - ^

-77r-_
~:
: it-:-. .: . : •
-^-+- _4 - - — - - " )II• * r-^- f ; ^ i i J 3 . - -~ -p :^ -
' — • - . .
- - ' • • -
-i--
_ ^-
-i • -------f-
1 • --.
; • | | :

l£- = a; H : ::-ji-4 -- • . :
" : - • - -
r^ "A^-^ :.. .L-:^-ZE
! , ,' ; -i • • £
-
' V ' T - ^ - + | .... • - '
- f -. ' b - =
:': : i
'
,.
••r^rr . :--'--;.-|
J •."_i^=r.:"

M~
^ \zX T_^_
j.

_
i

:
;
:

:
i

;
•;":••

• --t- t 1 • - I —
if].
1
af
-.. 1 ±r
-4— — r ) 1 - i
i •
:rrz
•C~

^ ~
f
" ^ •• s i t
•

^
•H-— -vur.
-| ' : : • ; — j ~ - " t
H-r-W H ' : |- | - j
s
' : ' . . : • ' i -'.":- ' . : - - ' - i - V " "4 — : ._.,., i
^=.-^= rvl
J
:
i • ! ; ] - • :
=\]- - . = . . • : - : -
~ " i | ""-i — _^ ^ = E -
::'tl~ r.F~
: 1 : 1- : - : ^ : | -1 r - j " • • I — - ;

j / —
':"...[•..
—U._.
> ._ -.- -- £ • — r " 1

." ~Ayy^r-
, ' - -.---, .: - - " T - '-'•" k---^= ^ ' - T ^
; ' i ,. ~=fe j4 i
..:,:-« - T / - ^ =
_-.; _. rrrj^ ; -!— u. .i_ -JhX • • - ! = - _

1
• ' I • - • • " •
A
:. ;• ; : j ... .-- — - :J7^^~- . : £ ^ z r r ZJ rX ~ 4 ™ r

i ' : i
4lTL__t_r^_f.
1
' '-i i j - •-""- i — j
j ;

r
•.. . - . . - - •
^
f-1
^^ =.. 4—
E c " ^ ~ ' ! : — - - — • • " • f ~ ^ L ~ + •~c=^
-.. ! ! I j- ; ! i !• s f - j:, CJ f ^ - j = - - , . | _ C ! •= .|.J t_c^ p- =L~
=i
-
^
. L.
[1^
P . ! . • • - - •
V t ~ : l t — -.=: ~r+zz r- -
- . . . . ^

" . " " • " _

1 i ~ -
= = f c __
o

\.\ ' i
: C J j-"H]

X
- -HE=:-±
^Z JT:
•_:L -

b-n. ._
i l f/^ _pJ-tr^|'<
:
••" •• - - ' t : 7 :
S» h .i. . . . . ..- 1 •--—.. m
4 ! . "f "J" fc^.r- --~."t.^TT^

n \ \ ,:.-}=----• A i'1=^= r~' ^1=- ^=~^"--^7-

i f ! .
:" '- —:-—-_-:
v -f4f - j :-. . :•:?• - = ^ - - - V / rfe E"" - • ;

. • i : - : -:— _J_-zr~_;_
; ; : ! -• ; : : |_L , - , - -,.-.<cvn +- =i - X--
: :
t

: ':—.-. .. . ~~"• ."_

-: i-.._ -n_-:
§ £ *4_±£i -
_L.: '"
/ - i

' jir -J
::r_r 1-1' " I
'•

" i -it 1
/ • ,
|- ! ' " - - — - - • -

H- i (___^„ j
!• ._-, . - -=^=_—
: i (/ * ' • • ' • / " • N
--. •]-...,:-",
/_ •1 ' ; -

; j
• i ,•/ i l l •X-k. I
i '-*-•/i 1 1 • : i . : f . -.+ :
- l i i / i 1 =• • / • " • • M l • X - - ; - •'-'-" - _
! ! i / ! > j/ U4 — r ---
. f ' ^ - ' • - : - :

'' XX f
..1 .. . ;
. i : :i ;
i ;1
:
. | ; l-Jr'; .1. ..^._ i ""-L._
[z
-
-r::.

F0 t\ '
/ J! , f
/'
1 ' •
r i

/ i t\:
/ '• \f i I
/ / ' UJ ' ' . Ii
/ /; S
77^ :
r" i

L • • : - r- - - ; • - :

/ / ^ ._.

/ / :
1 ' ' !
1

§
&

W/uD NIVdIS 1V101

1 3
 •:T.7T]-—
r- J—T-f--- ^ - • - " - "

I.- 1 --- --N-I- 1

1 - - - | _ i - !
l
• • , - - - - j - - ' i'"F - • " • • • T T !••••—
• ;—rH—r-^-_.j-_ - t - •• i ' ' '
if • "-' - .." "• ! ! ' : ' i . 1 i

' l-- r .--r--4==j--- : J " = . I_. - --. : ;- u - • r ' i • • -

' r=£.^-.. - ^.T:":-.-'- - •• ; ' ; - ^ - • ! • : ; • \ " , - . \ - i • • : ' • • • - J.--v.-==;
: - • 1
: f ! : - l i' j , . : : - : '
1—
:• .1 • . ' / —:
• J f '•;.. 7 — -
• 1 —' ;—;— • /

- - -- 4 ; ; J : -;-:--!:•• r - " r 7i

:
/
: : . 1 i- i .. ! . j . . . .: . ' ..: . V-
i r r i " V..\\\/ .... :

--- .
• . .
i • J • i /
;
• ' • ' • / - ]
1 ! "1 \ • \ •. 1 • 1 : • • 1 / ' !
;
. : ' -— ' ._
;
-•-" - i ' 4: ! . - • - : ! .
:
! • - ! - : / j

: . !
! •
. -j_-- L i . !" i - " '! " ' : i ! / • i - i -;: ( • - ; ' • . —

Mil! • i- r - - i : ,-" +a• ' " " 1 • • • ! • ' - " ! : : / ! ! : - .

_ > • : !
i
•
-: ; J -'--hi— -=-- -
•" " i i r
r - ' 1:/, ---=
^f- A
• • i •
i l l : i f - " -r:" j ; - 1
j ! l-'i-l 1 - s.:': 1 r / : . T s- " _f z5:: —
?!-!.
; j i i ! •
i,L|•••!••-

I- 1 j - 1 " ' -:-i--- -iLrz±:Jr i! 1 i " ' ~ • ' -

1
1 ' -^ . :—'. U J —i t t -" "1— ;i T-
:
•)- - H
i! t- -; — oe 1 • i ^ ,
1

1
•

: | i : | ;. i !_. .-.- = - 1 - :""" i ^ / ~ -

t -
-r-—
- J.-:|_ C J ~ " | ' •
«/^
• - 1

' "i t ' !" f " V / • 1 ''•

I-!r ji-; ; . - - - - c^ .. -J J - - .- ;
-p.-- : ^ / > 1
" - - • -

L .-.-.-:
- ^ t-~i=-._-
L_t=

- ^ Refill i - .! -' -' — t -- • • : • —

J.'.'- 1 - -. - i . . 1
'•_.-J H
.-.-
^3-r*ii r
^ ^ - :
£2
H.*
. ,-:=™- + •**!=* -
i

= '1 "j " 1 " •-••• ^ - .=-|.| {] ~ • 7

i' j

j
" \[ ?s= P ^ --^^s^-s:
= - ' • : • / •

|
.-_- ._—.-SI"
^ P=^«=M
— = r ^ = . • 1 •
. - . ' • -

1 -;
>-
' - - L— ^ § -]t p . -^~- ^ . - r ^ r
::.". ~r'r.~
. - _ - . ; •

. • • ! . , . !-L-

w - --1-
- - - - j - —
-h-
. ^ U5 T.J-. : - -7-.I -: - 7 _ -77
LlJ^ --

:
H-f"r
'•- • - " - -
:T_- :
:-;~"ri^r.: - ::.-y ..'i:- ..:[:-.•: . - , ! • •
.11 —
j , , ::
E?—~E:
:

1 ~-rir:., 7
j i
. - J.—.
--*---
- -

r : -j r
-t^

- - • - 1
1- !
1
V,, •. , ^ ^ r ~-\

j - -A- ----
if f-
. _ • : _

• 1 1
' - : ' = • •

1
' ~r ~'~] =--'- i l l !
J • 1S£ -;% . . ' ; • . • - . . I

1
1 ' • • . •'

- i
i
• :
F~
r.:_ • ' " ! , • i [_
Y 1 --1 L : : 1
.!_•.(._._;..., |. - i - -

i hi. M i P-^- 4 \t ' --t - • " • I --

" - = :!.: -..;. •fr~'---•': 3- -=i7£

. - - • • ! • - J .
- -1

11 ' - [ i I------ k-.::;_ — E I T ^

r -J^^ri :_ i "i i " '
•_
F-
< . -77-7: T)
;:.-
ps^y f
- - . — -T-
- -: 7-

I ! I •
1 1 '
:. -••.tEB -—^-/ - .; .: -;
1 I —| T i: i ; .
"1" i 7i —y— T—^_ " :
j
• • - , • .

- •
!

1
-

: • - • - i !
' ' • i
•! 1 i ! '" /
• •§ '• l • \ \

X T "~l
1- / ..;_••- -"U—J^ 1- H I--: i !
! i 1 ! " ! , " : •

i-1—"-_-- - • : • / —

-"*' - j J 1 1 1
! !'! , 17- 1I i _'
- • • -

' • • •• !

.- :-
- „ , . : :
-
-f^f ~ " U j : tr—: ^ ^
i '--L- '•-•3-:-
—--——(4-!— __..4..-.^ j •
- : • — :
-

3-/HH ; CO J -: -!- ::. -—^-

trh
ii-t-U . _ / / : .1- 5& : : •
.^1. j

Tj-i M / / ,,-htr ".."_.^~" 11 i 1 1 : i

ii\iML.L:^J>. _4_- • -
:
i'ii
1. . - - - •

i ,:"""T:
f \

lr<\ 2 1 '..' _ i !jl ! I" ill i .'." ..'.'-".

8
("!/"!) NIVUiS 1V101

19
H.O FINITE ELEMENT MODELING

When the engineer performing the FEA has an extensive knowledge in the

area being analyzed, assumptions about the geometry, materials, and bound-

ary and input conditions can be made. This is not always the case. An

engineer may be given the basic information and asked to investigate the

reliability of the system. Engineering judgment can be used to form some

assumptions and other assumptions can be made as a result of preliminary

modeling. Preliminary modeling is very important because incorrect

assumptions can have a significant effect on the results of the analysis.

These models are simple and are analyzed in a short time period compared to

the overall time of the analysis. If the preliminary models result in

assumptions that simplify the model, the time spent performing the anal-

ysis is reduced. Since most systems are too complex to be modeled in exact

detail, assumptions have to be made so that the analysis can be performed

in a timely manner.

Another reason preliminary modeling is useful is in reducing the

chances of round-off error when modeling complex geometries. Round-off

errors can result from several different factors including elements that

are too large, are improperly formed, and have unusual geometries. Simple

models representing the general shape of the complex geometry can indicate

to the engineer whether the model's basic design will have problems before

the more complex model is developed. The finite element model of the

microwire and PTH connection interface is an example of this.

20
4.1 MODELING OF COMPLIANT CONNECTIONS

Three different compliant connections were modeled using the same

general procedure. A model representing the package, board, and connection

system was modeled. This is referred to as the system model. The deflec-

tions resulting from a thermal analysis of this model were then inputted

into a model representing the compliant connection. Because the system has

symmetrical geometries and boundary conditions, only one quarter of the

system was modeled. The compliant connection did not have symmetrical

boundary conditions, therefore, the entire connection was modeled.

During the generation of the system model, it became apparent that the

model had too many elements for RADC's computer capabilities. To reduce

the large number of elements, the eight layers of microwire board were

represented by a separate model and the resulting deflections inputted

into the system model. Only the top layer of the microwire board was

represented in the system model. Once both analyses were simulated, a

comparison was made between the deflections along the top surface of the

board for the board-alone model and for the system model. The results

showed that the differences in the deflections were negligible and, there-

fore, this procedure was accurate.

4.1.1 SYSTEM MODELS

When generating a system finite element model, the following proce-

dure was used. The figures given will represent the generation of the S-

21
lead system model. The S-lead package was a .45 x .55 inch package with 32

leads. First, a two-dimensional drawing representing a segment of the

package and board and the S-lead was generated (Figure 12a). This model

was then extended in the third dimension (Figure 12b). Now a second two-

dimensional set of elements was created in a direction perpendicular to the

first set. This set represented the same configuration as the first set

and the nodes along the left side were taken from the original set of

elements (Figure 12c). This second set of elements was then extended in

the third dimension in the direction opposite to the first set of elements

(Figure 12d). Now two sets of three-dimensional elements existed. These

sets of elements were defined with different orientations and, therefore,

the second set of elements was redefined to conform with the first set of

elements.

22
Figure 12: S-Lead System FEM Generation

23
 The remaining package and board elements were generated (Figure 13),

thus completing the superelement model. Because there was a small number

of leads, the overall model was meshed so that the node points corresponded

to the location of the leads. The elements representing the air spaces in

between the leads were then deleted. The final finite element model is

shown in Figure 14. The proper boundary conditions and temperature condi-

tions were inputted into the data file and the analysis executed.

Figure 13: S-Lead System Superelement Model

14
Figure 14: S-Lead System FEM

25
 Since the l e a d l e s s chip c a r r i e r (LCC) was a l s o a .45 x .55 inch

package with 32 c o n n e c t i o n s , t h e same procedure was used to g e n e r a t e t h e

system model f o r t h e LCC as shown i n F i g u r e s 15, 16, and 17. The g u l l - w i n g

package was a .94 x .94 i n c h package with 172 l e a d s . The superelement

model was generated u s i n g the same procedure as used f o r the S-lead system.

1
\

I _—x.

L
-

( a ) 2-D FEM (b) 3-D FEM

(c) Additional 2-D FEM (d) Additional 3-D FEM

Figure 15: Leadless System FEM Generation

 "V

Figure 16: Leadless System Superelement Model

27
Figure 17: Leadless System FEM

28
Due to t h e l a r g e package s i z e and t h e l a r g e number of l e a d s , t h e model

could not be meshed so t h a t the node p o i n t s would correspond to t h e l o c a -

t i o n of t h e l e a d s . T h e r e f o r e , t h e f i n a l model had approximately t h e same

meshing c o n f i g u r a t i o n as the o t h e r models and t h e m a t e r i a l p r o p e r t i e s of

t h e s o l d e r and a i r m a t e r i a l s were combined to give a s e t of equivalent

properties. The g u l l - w i n g system model i s r e p r e s e n t e d i n Figures 18, 19,

and 20.

N>

( a ) 2-D FEM (b) 3-D FEM

(c) Additional 2-D FEM (d) Additional 3-D FEM

Figure 18: Gull-Wing System FEM Generation

29
Figure 19: Gull-Wing System Supereleraent Model

30
Figure 20: Gull-Wing System FEM

31
4.1.2 BOARD MODEL

A model representing the eight board layers was generated (Figure

21). This model contained node points on its horizontal axis which corres-

ponded to the locations on the board where the leads connected to the

board. The dimensions of this model were the same as the largest package,

the gull-wing package. A thermal analysis of the board model was simulated

and the results were inputted into the system models.

Y RX- 20

Figure 21: Microwire Board FEM

4.1.3 COMPLIANT CONNECTION MODELS

FEMs were generated for the S-lead, leadless solder joint and gull-

ing lead. The leaded models contained the lead and solder elements and

32
the solder joint model contained only solder elements. Each model has a

region representing the lead's contact with the edge of the package and a

region representing the lead's contact with the top surface of the board.

Deflections from the system models were inputted along these regions and a

static analysis simulated. The resulting stress values in the solder were

then used to estimate the lead's useful life.

The generation of the S-lead model was straightforward. The entire

model was completed in a day. First, a two-dimensional (2-D) model of the

lead and the attached solder was created. Eight noded 2-D elements were

used in order to obtain the circular geometry, Figure 22a. This model was

then extended into the third dimension and 20 noded, 3-D elements were

generated, Figure 22b. This model was then meshed and the final model is

displayed in Figure 23.

(a) 2-D FEM 3-D FEM

Figure 22: S-Lead Superelement Model

33
 y
RX= eg
X R2= 0

Figure 23: S-Lead FEM

34
 The generation of the leadless solder joint was a more complex

process. This model took approximately two weeks to complete. Drawings

representing the x-y, x-z and y-z planes were used as guides in the genera-

tion process, Figures 24, 25, and 26. This model represents a j o i n t used

on a package with 50 mil spacing between j o i n t s . The package has c a s t i l l a -

tions and, therefore, the solder curves into the package. The segment

representing the solder base was simple to generate since the x-z plane i s

uniform in the y - d i r e c t i o n . The f i l l e t element, however, had to be

inputted individually since these elements vary in a l l three dimensions.

The base elements were meshed and then the f i l l e t nodes and elements were

defined. The r e s u l t i n g model i s shown in Figure 27.

..- - .028._.... .QM . .QtfQ. 050

Figure 24: Solder Joint: X-Y Plane

35
 ! 1
1 J
i
I i

i
i
Ps
A j
1 i i
1

1 y.
1 [ ; \
1
1 1
1 1

| 1
1 1
r 1 —j_

1 1
1
n
—
^ 1
r 1
*
1
1 (

1
1
1 •

1
1
1 1
1

1 1
•

1
1
1 1
1
-• J1 1
J-- —

X{ 1 1
1 1
2 1 1
o .0 OVi» » 02. OS .O 25

Figure 25:. Solder Joint: X-Z Plane

36
 ,Y
ii

1
i 1
1
\ ^
f\ V?
]\ 1 1

• Si "*>H\
•
»(/ i>/s»

"

.0*11

1
1

.ohri i

I
1

.a M£ I

"-€ 9
i
i t i
1
i
i
i

~
00} ?
1
Figure 26: Solder Joint: Y-Z Plane

37
 y RX= ea
I -~x RY= - 3 0
RZ= 0

Figure 27: Leadless FEM

38
 The g u l l - w i n g model was a l s o simple t o g e n e r a t e . A 2-D model was

c r e a t e d u s i n g e i g h t noded elements r e p r e s e n t i n g t h e s o l d e r , l e a d and b r a z -

i n g m a t e r i a l s , Figure 28a. This model was then extended i n t o t h e third

dimension and 20 noded elements were g e n e r a t e d . This model was then

meshed, Figure 28b. A segment of t h e l e a d had t o be extended i n t h e z-

d i r e c t i o n before the f i n a l model was completed, Figure 29.

Figure 28: Gull-Wing Superelement Model

Figure 29: Gull-Wing FEM

39
4.2 MODELING OF COMPLEX CIRCULAR GEOMETRIES

When generating a FEM, the analysis will create a superelement model

outlining the general shape of the system. Normally the different elements

in the system will be of the same shape, i.e. circular, rectangular,

triangular, etc. In the system that was analyzed, however, there were not

only circular geometries embedded within rectangular geometries, but also

circular geometries intersecting with other circular geometries. Instead

of attempting to use a combination of rectangular and circular elements,

only quadralateral elements were used. Because these elements have corner

and midside nodes, rectangular and circular geometries can be defined.

It was obvious that the creation of this model would be complex and

take a considerable amount of time. Therefore, preliminary models were

created in order to check for round-off error problems in the approach

used. The first model created was a wire within a block (Figure 30). An

analysis was simulated and no problems occurred. This model was then

meshed in all three directions and elements were removed from the center

area (Figure 31). Once again, the simulation indicated no problems with

this basic design. From this model, it was obvious that only one quarter

of the system would need to be modeled for the ideal case where the wire

passed directly through the center of the hole. Generation of the actual

model now begins.

40
Figure 30: Wire Within A Block

Figure 31: Meshed Block

41
 Using the proper dimensions for the system, another 2-D superelement

model using eight noded elements was created (Figure 32a). This model was

meshed and a finer model generated (Figure 32b). This finer model was then

extended back i n t o the t h i r d dimension and 20 noded elements were gener-

ated. This model was then meshed (Figure 32c) and the materials were

properly i d e n t i f i e d for a l l the elements. Two sections had to be removed.

The f i r s t represented the a i r elements surrounding the top of the plating

and the second represented the a i r elements within the PTH (Figure 32d).

Figure 32e displays an enlarged view of the cavity s e c t i o n . The nodes

along the cavity surface were reshaped into a c i r c u l a r hole (Figure 33).

Since t h i s surface contained corner and midside nodes, i t was necessary to

change the location of both s e t s of nodes. When reshaping a surface such

as t h i s one, i t i s important to be sure that new surfaces do not overlap

any e x i s t i n g surfaces.

When the analysis of this was simulated, some round-off errors

occurred. This was due to improper meshing by the preprocessor program and

was e a s i l y corrected by a l t e r i n g the location of specific nodes. The

corrected analysis showed t h a t the r e s t r a i n i n g core elements restrained

the board so well that zero deflection could be considered along the top of

t h i s surface. The r e s t r a i n i n g core and the elements below i t were elimi-

nated and the board material was expanded in both horizontal d i r e c t i o n s .

Once these elements were expanded, the location of the midside nodes had to

be altered to correspond to the new dimensions. Figure 34 shows the FEM

for the 20 mil via, PTH. This model was modified to represent a 10 mil via,

PTH. The 10 mil via FEM i s displayed in Figure 35. The second model was

42
(a) (b)

a) 2-D Superelenent Model

b) 2-D Finite Element Model

c) Extended 3-D. Model

d) Finite Element Model with Cavity

e) Enlarged View of Cavity

(e)
Figure 32: Microwire/PTH FEM Generation

43
\

Y RX= 20
X RY= 39
<Z RZ= 0

Figure 33: Microwire/PTH FEM

44
 w*m^^' ;\ •<****

Fi^ire 34: Modified Microwire/PTH FEM (20 Mil Via)

Figure 35: Modified Microwire/PTH FEM (10 Mil Via)

45
analyzed and the results were used to estimate a useful life of the micro-

wire interface connection.

4.3 LESSONS LEARNED

After completing a FEA, it is important to check the results to see if

any other simplifications are possible. This information can then be

utilized in the future generation of FEM. Two observations were made in

regard to this study.

The first observation concerned the package/board/lead system models.

The resulting deflections from the three system models showed that the lead

design had a negligible effect on the expansion of the package and the

board. This indicates that a more complex system model is unnecessary.

Only a FEM of the package and a FEM of the board would be generated and two

separate analyses on the model performed. Since both the package and the

board are rectangular in shape, generation of these models would be

extremely simple. In addition, a single model could be generated to

represent several different package and board sizes. The resulting

deflections along different locations could be inputted into the indivi-

dual lead models. An assessment of the effect package size has on the

estimated life could be made.

The second observation concerned the microwire/PTH interface connec-

tion. The results showed that the board material away from the region of

concern had negligible stress and that the stresses in the region of

46
concern had l i t t l e variance between the two different models. This i n d i -

cates that a much simpler model representing only a small segment in the

area of i n t e r e s t would have been s u f f i c i e n t . This reduction in complexity

would also be beneficial in the modeling of a wire that was off center

since, in t h i s case, a one-quarter model could not be used and the e n t i r e

geometry would have to be modeled.

47
5.0 TRANSFER OF FINITE ELEMENT RESULTS

When stress caused by a static load is a concern, the engineer per-

forms a finite element analysis and compares the resulting stresses to the

yield stresses for the different materials. What can be done when the load

is not static, but varies with time? The specific problem dealt with in

this analysis was stresses caused by thermal cycling conditions. These

stresses eventually resulted in fatigue of the ductile materials. A tech-

nique was needed to transfer the results of the FEA to a value of merit

which could be used to assess the reliability of the design.

The method used followed a simple procedure. First, a FEM was gener-

ated. Static analyses were simulated for the temperature change of stress

free temperature to maximum temperature (high temperature cycle) and for

the temperature change of stress free temperature to minimum temperature

(low temperature cycle). The resulting maximum Von-Mises stress value for

the material of concern was obtained. In this study, the two materials of

concern were the ductile materials, copper and solder. The strain range

for this cyclic condition was calculated by adding the individual strains

for the low temperature and the high temperature cycles. These strains had

to include both the elastic strain and the plastic strain. Since the exact

plastic strain could not be determined without performing a nonlinear

analysis, a range for this strain was determined and it was assumed that

the actual plastic strain was in between these two values (see section 3.0

for a more detailed explanation).

48
 Once the strain range value was obtained, this value was plotted onto

a curve representing the strain range versus the number of cycles to

failure for the particular temperature range that was used (minimum tem-

perature to maximum temperature). This curve was generated from the

Coffin-Manson equation as described in section 3.0. Two values of cycles

to failure were obtained for each analysis. One presented the low plastic

strain and the other the high plastic strain. The resulting range of the

number of thermal cycles to failure gave an indication of the useful life

of the component. This procedure will now be applied to the FEM developed

in section 4.0.

5.1 S-LEAD CONNECTION

The high and low temperature cycle Von-Mises stress contours in the S-

lead connection are displayed in Figure 36. The overall maximum stress

does not occur in the solder material, but in the lead material. This

stress value is well below the yield point for the lead material. The

solder's maximum linear stress and plastic stress are:

SIMULATION MAXIMUM LINEAR STRESS PLASTIC STRESS

High Temp Cycle 6,000 psi 2,800 psi

Low Temp Cycle 3,900 psi 700 psi

The elastic strain is .0029 in/in (3,200 psi/1,090,000 psi). The

estimated plastic strain and the total strain are:

49
 STRESS COtlTOURS
UdM-rirSES STRESS
VIEW : 1.67E"92
RANGE : 7.62E*03

S9S.S

429.4

14.99
Er«c-ms(VDrsPLAY

T RX- SB
R r - -3e
sLx
RZ- e

(a) Low Temperature Cycle

STRESS COtlTOURS
UON-MISES STRESS
VIEW ;• 4.67E>91
RBMGE : S.85E<-83

S20.9

f H 456.4

391.8

327.3

26E.8

198.2

4.668
EMRC-NISfVDISPLflY

Y RX- ae
J y RY- -38

(b) High Temperature Cycle

Figure 36: Stress Contours for S-Lea'd FEM

50
High Temperature Cycle

METHOD PLASTIC STRAIN TOTAL STRAIN

1 2,800/1.09 x 106 = .0026 in/in .0029 + .0026 = .0055 in/in

2 .0026 x 2 = .0052 in/in .0029 + .0052 = .0081 in/in

Low Temperature Cycle

METHOD PLASTIC STRAIN TOTAL STRAIN

1 700/1.09 x 10 = .0006 in/in .0029 + .0006 = .0035 in/in

2 .0006 x 2 = .0012 in/in .0029 + .0012 = .0041 in/in

The overall total strain for high and low temperature cycles for the

S-lead connection is:

Method 1 .0055 + .0035 = .0090 in/in

Method 2 .0081 + .0041 = .0122 in/in

These overall strain values were plotted on the elastic plus plastic

line of the Coffin-Manson curve for solder (Figure 11) and the following

number of cycles to failure range was obtained: 11,500 to 60,000 cycles.

51
5.2 LEADLESS CONNECTION

The leadless solder connection model consists entirely of solder ele-

ments. Figure 37 displays the high and low temperature cycle Von-Mises

stress contours for this analysis. The solder's maximum linear stress and

plastic stress are:

SIMULATION MAXIMUM LINEAR STRESS PLASTIC STRESS

High Temp Cycle 23,400 psi 20,200 psi

Low Temp Cycle 7,698 psi 4,498 psi

These values are both beyond solder's yield stress of 3,200 psi. The

elastic strain is .0029 in/in (3,200 psi/1,090,000 psi). The estimated

plastic strain and the total strain are:

High Temperature Cycle

METHOD PLASTIC STRAIN TOTAL STRAIN

1 20,200/1.09 x 10 = .0185 in/in .0029 + .0185 = .0214 in/in

2 .0185 x 2 = .0370 in/in .0029 + .0370 = .0399 in/in

52
 mess commas
ISH-HISES STOESS
WEU : S.96E<«
Z.3««M

S.9H
QWC-MISIVDISPLAY

Y HX- 39

* ^ RZ- 8

(a) Low Temperature Cycle

UJN-MTSES STHESS
WEU : 7.77E<K
7.7ae»«

231.5

77. m
Er«c-rilS(vtiKPLfiY

L R f -38

(b) High Tmperature Cycle

Figure 37: Stress Contours for Leadless FEM

Low Temperature Cycle

METHOD PLASTIC STRAIN TOTAL STRAIN

1 4,498/1.09 x 106 = .0041 in/in .0029 + .0041 = .0070 in/in

2 .0041 x 2 = .0082 in/in .0029 + .0082 = .0111 in/in

The overall total strain for high and low temperature cycles for the

S-lead connection is:

Method 1 .0214 + .0070 = .0284 in/in

Method 2 .0399 + .0111 = .0510 in/in

These overall strain values were plotted on the elastic plus plastic

line of the Coffin-Manson curve for solder (Figure 11) and the following

number of cycles to failure range was obtained: 120 to 260 cycles.

5.3 GULL-WING CONNECTION

The high and low temperature cycle Von-Mises stress contours in the

gull-wing connection are displayed in Figure 38. Once again, the overall

maximum linear stress does not occur in the solder, but in the lead

material and this stress value is well below the yield point for this

material. The solder's maximum stress and plastic stress are:

54
 STRESS cartTWjRS
UOfl-IIISES STRESS
VIEU : 3.s«»ee
RflMGE : 1.37E*64

' P . i o . i >r I AFP •

_ - 136.7

HL 121.S

yi 196.3

••
_ 91.12

75.94

•
60.78

15.58

mprj
38. IB

15.21
^5'
3E-82
:r«c-riisfi''DrsPLPY

Y RX—168

(a) Low Temperature Cycle *L* RY- 39

RZ- 188

STRESS CONTOURS
UOtl-rllSES STRESS
V1EU 7.7iE*ee
RAf«E 3.2SE»84

ip,j„.i ¥ 1 HF;' -

n
_ - 32S.2

289.1

ii 253.8

••
— 216.9

189.7

••
144.6

1E8.S

72.34
i * -^
36.21

8E-92
erac-msiVDrsPLflY

Y RX--168
x
(b) High Temperature Cycle
sX 2 R»Y"
z- ise

Figure 38: Stress Contours for Gull-Wing FEM

55
STIMULATION MAXIMJM LINEAR STRESS PLASTIC STRESS

High Temp C y c l e 2,700 psi

Low Temp Cycle 4,250 psi 1,050 psi

Only the low temperature cycle value is beyond solder's yield stress

of 3,200 p s i . The elastic strain, the estimated plastic strain, and the

total strain are:

High Temperature Cycle

The elastic strain is .0025 in/in (2,700 psi/1,090,000 p s i ) . Since

there is no plastic strain, then the total strain is .0025 i n / i n .

Low Temperature Cycle

The e l a s t i c strain is .0029 in/in (3,200 psi/1,090,000 p s i ) .

METHOD PLASTIC STRAIN TOTAL STRAIN

1 1,050/1.09 x 106 = .0010 in/in .0029 + .0010 = .0039 in/in

2 .0010 x 2 = .0020 in/in .0029 + .0020 = .0049 in/in

The overall total strain for high and low temperature cycles for the

gull-wing connection i s :

56
Method 1 .0025 + .0039 = .0064 in/in

Method 2 .0025 + .0049 = .0074 in/in

These overall strain values were plotted on the Coffin-Manson curve

for solder (Figure 11) and the following number of cycles to failure range

was obtained: 400,000 to 2,000,000 cycles.

5.4 MICROWIRE/PTH CONNECTION

Four analyses were performed on the two connection models. Figure 39

displays the Von-Mises Stress contours for the 20 mil via model and Figure

40 displays the contours for the 10 mil via model. The majority of the

board has very little stress with the high stresses occurring in the plated

copper and the copper microwire. The highest stresses occurred at the

plated barrel where the pad contacts the board. These values were not used

in this analysis because that area was not the area of reliability concern.

The maximum linear stress found in the copper wire/PTH interface is listed

below for the four different analyses.

SIMULATION LINEAR STRESS PLASTIC STRESS

10 Mil Via, High Temp Cycle 19,000 psi 11,000 psi

10 Mil Via, Low Temp Cycle 19,000 psi 11,000 psi

20 Mil Via, High Temp Cycle 10,000 psi 2,000 psi

20 Mil Via, Low Temp Cycle 8,000 psi 0 psi

57
 STRESS CCMTEURS
I W I - t l l S E S STRESS
UICU i 1.816-66
BOHSE :• < t t c «

= .,,,,] < • . • ; * . • ? •

^__. «.63

18.33

6.198

et-ae
EJ-FC -MISa'D ISPLBY

|y RX- 2»
I X RY> 38
\Z RZ- 9

(a) Low Temperature Cycle

STUESS corrrouo
WM-ruses STRESS
UICU i 3.07C-4C
*nnsE ! 3 5BE-aa

*W- 37.83

3E-88
ETRC -rt KB-'O I3T.BY

RX- 2*
RY- 38
^ RZ- 8

(b) High Temperature Cycle

Figure 39: Stress Contours for Microwire/PTH (20 Mil Via)FEM

58
 srwcs* cawnws

UIEO I 1 . * « - «
WM6E : 4.USE-**

6C-W

2E-W

t- KZ- «

(a) Low Temperature Cycle

STRESS COITtMB
icn-mses STBESS

BAItSC s S.27E-81

SE-W

iiiis;::;.. SE-BS
grKC-nisn-'Dia'Lflr

k
UK- 2»
RT- 3»
RZ- «

(b) High Temperature Cycle

Microwire/PTH (10 Mil Via) FEM

40: Stress Contours for
Figure
59
 Since t h e e l a s t i c l i m i t f o r copper i s 8,000 p s i and each s i m u l a t i o n

met or exceeded this value, the elastic strain is the elastic stress

divided by the modulus of elasticity, which comes to .0005 in/in

C8,000psi/16,000,000psi). The e s t i m a t e d p l a s t i c s t r e s s i s then t h e d i f -

f e r e n c e between t h e maximum l i n e a r s t r e s s and t h e y i e l d s t r e s s as shown

above. The e s t i m a t e d p l a s t i c s t r a i n and the t o t a l s t r a i n a r e c a l c u l a t e d

bel ow:

10 Mil V i a , High Temperature Cycle

METHOD PLASTIC STRAIN TOTAL STRAIN

1 11,000/16 x 10 = .0007 i n / i n .0005 + .0007 = .0012 i n / i n

2 2 x .0007 = .0014 i n / i n .0005 + .0014 = .0019 i n / i n

10 Mil Via, Low Temperature Cycle

METHOD PLASTIC STRAIN TOTAL STRAIN

1 11,000/16 x 10° = .0007 i n / i n .0005 + .0007 = .0012 i n / i n

2 2 x .0007 = .0014 i n / i n .0005 + .0014 = .0019 i n / i n

The o v e r a l l t o t a l s t r a i n f o r high and low t e m p e r a t u r e c y c l e s f o r t h e

10 mil v i a i s :

60
METHOD OVERALL STRAIN

1 .0012 + .0012 = .0024 i n / i n

2 .0019 + .0019 = .0038 i n / i n

20 Mil Via, High Temperature Cycle

METHOD PLASTIC STRAIN TOTAL STRAIN

1 2,000/16 x 10 = .0001 i n / i n .0005 + .0001 = .0006 i n / i n

2 2 x .0001 = .0002 i n / i n .0005 + .0002 = .0007 i n / i n

20 Mil Via, Low Temperature Cycle

METHOD PLASTIC STRAIN TOTAL STRAIN

1 0 in/in * .0005 + 0 = .0005 i n / i n

2 0 in/in * .0005 + 0 = .0005 i n / i n
*Note: The Von-Mises s t r e s s value did not exceed t h e e l a s t i c limit,

t h e r e f o r e , t h e r e was no p l a s t i c s t r a i n .

The o v e r a l l t o t a l s t r a i n f o r the high and low temperature cycles for

t h e 20 mil via i s :

61
METHOD OVERALL STRAIN

1 .0006 + .0005 = .0011 in/in

2 .0007 + .0005 = .0012 in/in

These overall strain values were plotted on the plastic plus elastic

Coffin-Manson curve for copper (Figure 11) and the following number of

cycles to failures ranges were obtained'.

10 mil via: 110,0000 to 10,000,000

20 mil via: greater than or equal to 10,000,000

62
6.0 CONCLUSIONS

Several Finite Element Analyses (FEA) were performed and the results

used to estimate the useful life of the various connections. For the

surface mounted lead connections, the results indicated that the leadless

connections would only survive a minimal number of thermal cycles before

problems occurred (less than 260 cycles). The S-lead and gull-wing leads,

however, showed very good survival rates. It was estimated that the S-lead

would survive between 11,500 and 60,000 thermal cycles and that the gull-

wing lead would survive between 600,000 and 2,000,000 thermal cycles. This

indicates that the two leaded designs will have superior reliability per-

formance.

The results from the analysis on the microwire and Plated-Through-

Hole (PTH) connection indicate that, under ideal conditions, this connec-

tion will survive a minimum of 110,000 thermal cycles. These results must

be carefully considered. Due to insufficient information and time, an

analysis of nonideal configurations was not performed. There are two types

of microwire connections that would be of serious reliability concern. The

first type is where the wire does not pass straight through the PTH (180

degrees), but is rotated at an angle (135 or 90 degrees). The second type

of connection is where the microwire is allowed the maximum variation from

the ideal condition. An analysis under these conditions would predict the

reliability of the worst acceptable microwire connection.

63
 In addition to the reliability assessments, conclusions can be made

about the finite element modeling and how this process can be improved.

Listed below are four observations that have been identified:

1. The results of the FEA on the microwire connection indicate that

a much smaller, less-refined model might have been sufficient to

analyze this connection. This alternate model would have reduced the

model generation time by at least 50 percent and would have provided

the time needed to generate a model of the nonideal case where the

wire is off center.

2. The resulting stress values in the center of the microwire in the

microwire connection FEM were not as harmonious with one another as

these values should have been. This indicates that instead of using

quadralateral elements only and distorting some of the elements to a

considerable degree, a mixing of element types would be preferential.

In this case, the quadralateral elements were so distorted in the

center of the microwire that the elements took on the approximate

shape of a triangle. This would explain the inconsistency of stress

values. Instead, these center elements should have been triangles.

3. In the FEMs of the individual leads, the deflections across the

width of the leads (25 mils) were essentially constant. These results

indicated that the leads could have been modeled in two dimensions

versus three dimensions.

64
 4. A comparison of the FEA results for the three different

package/board/lead models showed only a slight variation in package

and board deflections for the three models. This occurrence implies

that the lead type has little effect on the package and board deflec-

tions. Consequently, a single package FEM and a single board FEM

could be generated and used for the analysis of many different lead

types.

The work documented in this report is just a stepping stone in the

development of a method to transfer FEA results to a useful estimate of

reliability. There are several areas in this analysis, in both the finite

element modeling and the transfer of results, that deserve further inves-

tigation. Two of these areas are outlined below."

1. The validity of performing a purely linear, elastic analysis and

using the results to estimate the amount of plastic strain should be

verified. A question arises as to the ability to predict the response

of the material once it goes beyond the linear range. There is some

indication that, in fact, a nonlinear, elastic and plastic analysis

should be performed instead.

2. If these connection models could be significantly simplified, a

full-view model, rather than a one-quarter model, could be generated

without exceeding the computer's limit. The full-view model would

then be used to perform a vibration analysis.

65
7.0 REFERENCES

1. Sandor, Bela I., "Fundamentals of Cyclic Stress and Strain", The University of
Wisconsin Press, 1972.

2. Bivens, Gretchen A. and Bocchi, William J., "Reliability Analysis of a Surface

Mounted Package Using Finite Element Simulation", October, 1987, RADC-TR-87-177.

3. Bivens, Gretchen A., "Reliability Assessment of Surface Mount Technology

(SMT)", March, 1988, RADC-TR-88-72.

66
e«oe*3<oe*s<^8oejeoe*3^t<sQW3ws^^

MISSION
of
Rome Air Development Center

RADC plans and executes research, development, test and

selected acquisition programs in support of Command, Control,
S
Communications and Intelligence (C I) activities. Technical and
engineering support within areas of competence is provided to
ESD Program Offices (POs) and other ESD elements to
3
perform effective acquisition of C I systems. The areas of
technical competence include communications, command and
control, battle management information processing, surveillance
sensors, intelligence data collection and handling, solid state
sciences, electromagnetics, and propagation, and electronic
reliability/maintainability and compatibility.

tt&ZOSfZtAQey&rxjs^^^