Effect of substrate flexibility on solder joint reliability

Effect of substrate flexibility on solder joint reliability

Xingsheng Liu a,*,1, Shuangyan Xu b, Guo-Quan Lu a, David A. Dillard b

a Materials Science and Engineering Department and Center for Power Electronics Systems, Virginia Tech, Blacksburg, VA 24061, USAb Engineering Science and Mechanics Department, Adhension Mechanics Laboratory, Virginia Tech, Blacksburg, VA 24061, USA

Received 4 December 2001; received in revised form 5 June 2002

Abstract

Solder joint fatigue failure is a serious reliability concern in area array technologies, such as flip chip and ball grid

array packages of integrated-circuit chips. The selection of different substrate materials could affect solder joint thermal

fatigue lifetime significantly. The reliability of solder joint in flip chip assembly for both rigid and compliant substrates

was evaluated by accelerated temperature cycling test. Experimental results strongly showed that the thermal fatigue

lifetime of solder joints in flip chip on flex assembly was much improved over that in flip chip on rigid substrate as-

sembly. Debonding area of solder joints in flip chip on rigid board and flip chip on flex assemblies were investigated,

and it was found that flex substrate could slow down solder joint crack propagation rate. The mechanism of substrate

flexibility on improving solder joint thermal fatigue was investigated by thermal mechanical analysis (TMA) technique.

TMA results showed that flex substrate buckles or bends during temperature cycling and this phenomenon was dis-

cussed from the point of view of mechanics of the flip chip assembly during temperature cycling process. It was in-

dicated that the thermal strain and stress in solder joints could be reduced by flex buckling or bending and flex

substrates could dissipate energy that otherwise would be absorbed by solder joints. It was concluded that substrate

flexibility has a great effect on solder joint reliability and the reliability improvement was attributed to flex buckling or

bending during temperature cycling.

� 2002 Elsevier Science Ltd. All rights reserved.

1. Introduction

Solder joint area array technologies, such as flip chip

and ball-grid array, have emerged as low-cost, high

density, and high performance techniques for both chip-

level and board-level interconnections [1–4]. However,

solder joint fatigue is a serious concern in area array

packages. Thermal strains and stresses caused by coef-

ficient of thermal expansion (CTE) mismatch are the

main cause of failure in solder joint interconnections.

Numerous factors affect solder joint fatigue perfor-

mance, such as joint geometry, chip size, interface met-

allurgy, underfill and substrate materials. Of these

factors, mechanical properties of substrate materials

could play an important role. A proper substrate ma-

terial could reduce the thermal strains and stresses in

solder joints and improve solder joint thermal fatigue

lifetime significantly.

Flex substrates are quite popular in the electronics

industry. Flex circuits have been utilized in integrated

circuit packaging and automotive and medical elec-

tronics applications with considerable success in reduc-

ing costs, minimizing package volume, saving space

for the system, improving packaging density, and

maintaining electrical performance. The excellent per-

formance of flex substrates, especially under hostile

operating conditions, and the features of high insula-

tion, low weight, high density, and, more importantly,

the flexibility make flex attractive in power electronics

packaging applications. We have developed a three-

dimensional flip chip on flex package for power elec-

tronics applications [5–7]. In this paper, the influence of

a flex substrate on solder joint reliability is investigated

Microelectronics Reliability 42 (2002) 1883–1891

www.elsevier.com/locate/microrel

* Corresponding author. Tel.: +1-607-248-1110; fax: +1-607-

974-0440.

E-mail address: [email protected] (X. Liu).1 Current address: Corning Incorporated, Science and Tech-

nology Center, Corning, NY 14831, USA.

0026-2714/02/$ - see front matter � 2002 Elsevier Science Ltd. All rights reserved.

PII: S0026-2714 (02 )00269-X

mail to: [email protected]

and the mechanism of substrate flexibility for improving

solder joint reliability is studied.

Basically, a flex substrate consists of a metal layer

such as Cu and a polyimide layer. The metal layer could

be electroplated or laminated. In the latter case, some-

times, an adhesive layer is used between the metal and

polyimide layer to enhance the adhesion. The flex sub-

strate could be one-sided or double-sided. In order to

protect the copper layer from oxidation, Ni and/or Au

could be electroplated as the final surface finish. The flex

substrate used in this research is a commercially avail-

able, double-sided, copper-clad laminate of polyimide

film bonded to two copper foil without an adhesive. This

laminate has excellent handling characteristics for fab-

rication, outstanding dimensional stability, excellent

assembly performance over a wide range of processing

temperatures, a CTE quite similar to that of traditional

printed circuit boards (PCB), excellent thermal resis-

tance for high temperature assembly processes, and is

compatible with conventional oxide treatments and wet

chemical plated-through-hole desmear processes. The

thickness of both the polyimide film and copper foils is

50 lm (2 mil). The pattern of Cu traces can be formed byphotolithography and chemical etching. After process-

ing, the substrate retains very good bend and crease

flexibility.

In this study, accelerated temperature cycling was

imposed to evaluate the influence of the flex substrate on

solder joint fatigue lifetime. In-process electrical re-

sistance measurements, acoustic microscopy imaging

(nondestructive evaluation), and optical microscopy

were utilized to examine the integrity of the joint and to

detect cracks and other defects before and during ac-

celerated fatigue tests. The mechanism of substrate

flexibility on improving solder joint thermal fatigue was

investigated by a thermal mechanical analysis (TMA)

technique. In this paper, the influence of flex substrate

(flip chip on flex) on solder joint reliability is first sum-

marized (the details will be presented elsewhere [8]), then

the TMA study on the behavior of the flex substrate

during thermal cycling is presented and, finally, the ex-

perimental results are discussed.

2. Influence of flex substrate on solder joint reliability

The reliability of fabricated solder joints in this study

was evaluated by accelerated temperature cycling tests.

During the temperature cycling, in-process electrical

resistance measurements and nondestructive evalua-

tions, such as scanning acoustic microscopy and optical

microscopy, were conducted to monitor solder joint

failure behavior. The electrical resistance of the solder

joint interconnections, which is measured by the four-

point probe method, is used as the failure criterion. We

set the evaluation criterion such that a 20% increase in

electrical resistance is regarded as solder bump inter-

connection failure [9].

The chip dimensions [16] and solder joint configura-

tion are shown in Fig. 1(a). There are seven solder joints

on each silicon chip. The solder material used in this

study is eutectic Sn37/Pb63 lead-tin solder. All the test

samples were flip chip attached to a test vehicle as shown

in Fig. 1(b). Solder joints were fabricated using stan-

dard stencil-printing technique. Two temperature cy-

cling conditions were used. The first condition was

temperature range: 0 and 100 �C; temperature raise time:10 min; dwell time at 100 �C: 5 min; temperature falltime: 10 min; dwell time at 0 �C: 5 min. The secondcondition was temperature range: �40 and 125 �C;temperature raise time: 25 min; dwell time at 125 �C: 5min; temperature fall time: 25 min; dwell time at �40 �C:5 min. Both flex substrates and rigid PCB were used for

making the test vehicle for comparison purposes, that is,

both flip chip on flex and flip chip on rigid board were

tested under the same temperature cycling conditions.

For each case, 21 solder joints were evaluated.

Fig. 2 shows the typical electrical resistance (nor-

malized) increases of solder joints (seven solder joints of

one test chip on a PCB rigid board) during temperature

cycling (the first temperature cycling condition). Simi-

Fig. 1. (a) Dimensions and solder joint locations of tempera-

ture cycling test chips [16] and (b) thermal cycling vehicle.

1884 X. Liu et al. / Microelectronics Reliability 42 (2002) 1883–1891

larly, we have obtained electrical resistance vs. temper-

ature cycling curves for all the samples for the four cases

introduced above. From Fig. 2, it can be clearly seen

that the curves can be divided into three periods, cor-

responding to three different fatigue degradation phases

which are crack initiation, crack propagation and cata-

strophic failure. During the crack initiation phase, there

is no significant electrical resistance increase while in the

crack propagation period, electrical resistance increases

gradually. On the other hand, electrical resistance in-

creases dramatically in the catastrophic failure period.

Note that we set a 20% resistance increase as the failure

criterion and solder joint fatigue lives can be estimated

based on the normalized electrical resistance vs. tem-

perature cycle curves. Fig. 3 is a comparison of the fa-

tigue life of flip chip on flex (FCOF) and flip chip on

PCB board (FCOB) assemblies with the commonly used

barrel-shaped solder joints under the first and the second

temperature cycling conditions, where the fatigue life is

the average lifetime of 21 solder joints. We know that

the major difference between flip chip on PCB and flip

chip on flex assemblies is the substrate flexibility.

Therefore, we can attribute the solder joint fatigue life-

time difference to substrate flexibility. We can see from

Fig. 3 that the flex substrate improves solder joint reli-

ability significantly under both the first and second

thermal cycling conditions. This indicates that substrate

flexibility could affect flip chip solder joint significantly.

C-mode scanning acoustic microscopy (C-SAM)

(Sonix system) was used to detect cracks and monitor

crack propagation in flip chip solder joints during tem-

perature cycling. A 75 MHz transducer was utilized in

this research. Fig. 4 shows typical C-SAM images of the

interface between solder joints and chip of a typical flip

chip on flex assembly after 1400, 2000, 2400, 2800 and

3000 temperature cycles, respectively, under the sec-

ond temperature cycling condition. The C-SAM results

showed that the images of some parts of the solder joints

had faded or were not present at all after 1400 temper-

ature cycles. This indicated that solder joints begin to

crack after 1400 cycles, which was confirmed by elec-

trical resistance measurements. More and more areas of

C-SAM images became faded or were disappeared af-

terwards. From the C-SAM images, crack areas can be

monitored during thermal cycling and changes in bond

area can be determined. The National Instruments

IMAQ Vision Builder software is used to process the C-

SAM images and calculate crack areas of the processed

images. The image processing is based on the percent

threshold values of different colored regions within the

user-defined area of an image. The threshold was set to

the same value in processing all the C-SAM images. Fig.

5 shows an example of an acoustic image of a flip chip

Fig. 3. A comparison of the average fatigue life of FCOF and

FCOB assemblies under the first and the second thermal cycling

conditions.

Fig. 2. Typical electrical resistance increases of solder joints in a flip chip on rigid PCB assembly under the first temperature cycling

condition.

X. Liu et al. / Microelectronics Reliability 42 (2002) 1883–1891 1885

assembly and the processed picture of that image by

IMAQ Vision Builder software.

C-SAM imaging was periodically conducted on both

the FCOF and FCOB assemblies to monitor the crack

initiation and propagation process during thermal cy-

cling. Fig. 6 is a comparison of the average crack area

increase of FCOF and FCOB assemblies during tem-

perature cycling under the second temperature cycling

condition for 21 solder joints. It is obvious that the

crack area in FCOF assemblies increases slower than

that in FCOB assemblies. This is in consistent with the

thermal fatigue results shown in Fig. 3. The experi-

mental results indicated that flex substrate slows down

the crack propagation rate and thus improves overall

fatigue life.

3. Mechanism of substrate flexibility on improving solder

joint reliability

It was shown in the previous section that the flip chip

on flex assembly has a longer thermal fatigue lifetime

than the flip chip on rigid PCB assembly and lifetime

improvement was attributed to the increased compliance

of the flex substrate. This section investigates the flex

substrate behavior during temperature cycling and

studies the mechanism by which the substrate flexibility

improves solder joint reliability. Flex substrate behavior

during temperature cycling was characterized by the

displacement of flex at different locations in the assem-

bly, which was detected by a TA instruments dynamic

mechanical analyzer (DMA).

3.1. Sample preparation and flex substrate displacement

measurement

To simplify the problem and better study the flex

substrate behavior in the flip chip assembly during

thermal cycling, one-dimensional flex strip samples, il-

lustrated in Fig. 7, were prepared and investigated.

Two solder joints with the same standoff height located

at the two ends of the silicon piece were used to in-

terconnect the silicon and flex substrate strip. The same

solder and flex substrate materials were used as those

in the thermal cycling test sample. Therefore, the major

difference between the flex strip samples and the reli-

ability study samples shown in Fig. 1 is that the flex

strip samples are one-dimensional, while the reliability

study samples are two-dimensional. The flex substrate

behavior during thermal cycling in the flex strip sample

is a simplified representation of that of the flex in

thermal cycling samples. Displacement of flex at dif-

ferent locations in the assembly, illustrated by the nine

points uniformly distributed along the flex strip, during

thermal cycling was considered to be an indicator and

Fig. 4. C-SAM images of the interface between single barrel-shaped solder joints and chip of a typical FCOF assembly during

temperature cycling.


characteristic of flex behavior that would influence

solder joint reliability.

A TA instruments 2980 DMA was used to detect the

displacement of flex at different locations in the as-

sembly under thermal cycling. DMA is an analytical

instrument used to test the physical properties of many

different materials. Fig. 8 shows the TA instruments

dynamic mechanical analyzer clamp assembly [10]. The

clamp assembly is placed in a chamber which can be

programmed for temperature cycling. The TMA con-

trolled-force mode was used to measure the displace-

ment of a sample as a function of time, temperature,

and applied force. Force can be applied in one of three

manners: constant force, stepped force, or continuous

force ramp. The clamp assembly of the DMA consists

of a fixed clamp and a moveable clamp. On the

moveable clamp, there is a probe which moves with

clamp. The test sample was placed on the fixed clamp

with the flex side facing up and the moveable clamp

was above the sample, as shown in Fig. 8, and there

were not any restraints between the test sample and the

fixed clamp. To keep the probe on the movable clamp

in touch with the test sample all the way during the

test, a static force of 0.005 N, which is very small and

should play negligible effect in the stress–strain re-

sponse of the test sample, was applied to the flex in our

experiments. The DMA system monitors and measures

the probe displacements dynamically during the oper-

ation. The probe displacement resolution is 0.01 lm.Thus, we can know the vertical displacement of the flex

during temperature cycling from the displacement of

the probe.

The thermal cycling program for the DMA chamber

was programmed to be

Equilibrate at 30.00 �C, Isothermal for 3.00 min;Ramp 1.00 �C/min to 150.00 �C, Isothermal for 10.00min;

Ramp 1.00 �C/min to 50.00 �C, Isothermal for 10.00min;

Repeat for 3 cycles;

End of method.

Therefore, the sample inside the DMA goes through

three temperature cycles between 50 and 150 �C. Thedisplacement of the flex during thermal cycling was

probed only at the left five of the nine locations uni-

formly distributed along the flex strip as shown in Fig.

7(b), considering the symmetry of the nine points. Two

such samples were tested.

3.2. Experimental results

In the TMA test, the probe was placed at one point

on the flex and the sample was subjected to tempera-

ture cycling. Displacements can be measured as a

Fig. 5. (a) C-SAM image of a flip chip assembly; (b) processed

picture of the image by IMAQ Vision Builder software which is

used to calculate crack or contact area.

Fig. 6. A comparison of the average crack area increase of

FCOF and FCOB assemblies during temperature cycling under

the second temperature cycling condition.


function of temperature, time and applied force si-

multaneously. We were more interested in the flex

displacement change as a function of temperature. Fig.

9 shows a typical TMA trace of the flex displacement

at one point of a flex strip sample through three tem-

perature cycles. We can clearly see from the TMA trace

that the point of the flex at which the probe is located

has different positions at different temperatures during

one temperature cycling, that is, the flex moves up and

down during temperature cycling. Note that the origin

of the y-axis position of the DMA system is at the

upper limit of the probe range, that is, when the probe

goes up, the value of the y-axis decreases and viceversa. We can see from the typical trace of flex dis-

placement shown in Fig. 9 that the flex strip between

the two solder joints moves up (outward) as the tem-

perature increases. The flex moves down when the

temperature cools down. Therefore, during temperature

cycling, the flex substrate deforms up and down. The

initial heating cycle produces results that are different

from those of the subsequent cycles. This difference is

believed to be due to the thermal history of the flex and

also to physical aging and other viscoelastic processes.

Subsequent heating/cooling cycles are quite consistent

indicating that reproducibility of the TMA method.

The hysteresis loop is believed to be associated with the

differential thermal lag, as it tends to vanish at slower

heating/cooling rates [11].

We obtained the TMA traces of the flex displacement

at the five locations for both samples. Fig. 10 shows a

comparison of the flex displacements at different loca-

tions in the flex strip during temperature cycling for one

of the samples. Note that the TMA traces were corrected

for the DMA clamp system expansion during thermal

cycling. We can see from the TMA curves that the flex

has the largest displacement at point 5 which is at the

Fig. 7. Schematic drawing of the one-dimensional flex strip sample (in mm unit); (a) 3-D drawing; (b) cross-section.

Fig. 8. The TA instruments DMA clamp assembly [10] and schematic picture of probe and sample under test.


center of the flex strip, and has the smallest displacement

at point 1 which is right on top of the solder joint, with

the points in between having intermediate displace-

ments. Both of the samples showed the same trend that

those locations closer to solder joints have smaller dis-

placements and those locations farther away from the

solder joints have larger displacements, though the dis-

placement value at one specific location may be different

for those two samples. Fig. 11 is the summary of the

TMA test results for the one-dimensional flex strip

samples. The displacement values at different flex loca-

tions in the figure are the average values and the error

bars are the standard deviations of the two samples

(each sample had three TMA runs).

4. Discussion

The TMA results on flex behavior during thermal

cycling strongly indicate that flex buckles during tem-

perature cycling. In a temperature cycle, when the tem-

perature increases, the flex substrate expands more than

the silicon chip, and thus the flex between two solder

joints buckles up and the probe goes up accordingly. As

a result, the points that are closer to buckling center

Fig. 10. Comparison of the flex displacements at different lo-

cations in a flex strip during temperature cycling.

Fig. 11. Summary of the TMA test result for the one-dimen-

sional flex strip samples.

Fig. 9. Typical TMA trace of the flex displacement at point 2 of a flex strip sample through three temperature cycles.


have greater displacements, while those closer to the

fixed ends (solder joints) have smaller displacements. On

the contrary, when the temperature cools down, the flex

substrate shrinks more and the flex between two solder

joints becomes more and more flat and the probe goes

down.

The phenomenon of flex buckling during thermal

cycling could be explained when we consider the flip chip

on flex structure assembling process and the mechanics

of the assembly during temperature cycling process. The

flip chip attach process is at elevated temperature to

accomplish solder reflow, so the stress-free equilibrium

state for the package is at the solidification temperature

for the solder. Upon further cooling, due to the CTE

mismatch between the silicon chip and the substrate, the

substrate tends to shrink more than the chip, thus there

is strain and residual stresses built in the solder joint

interconnection after the reflow process. The solder

joints are deformed in a shear-dominant mode as shown

in Fig. 12. However, residual stress inside solder joints

can be much reduced after some time due to the creep of

soft Pb/Sn solders and/or the residual stress is forgotten

during thermal cycling [1,2,4], and residual stress is not

the dominant cause for solder joints failure in most

cases. Therefore, we can regard solder joints are in near

stress-free state at room temperature some times after

they were processed. For flip chip on rigid substrates,

there is cyclic thermal strain in the solder joints during

thermal cycling. This is the major mechanism of solder

joint fatigue failure. On the other hand, the situation is

quite different if we attach the flip chip assembly on flex

substrate. The strain and stress inside solder joints can

be much reduced by flex buckling. The buckling of flex

can be interpreted satisfactorily if we regard solder joints

are the fixed ends of the buckling and the flex in between

the solder joints was initially flat at room temperature.

In the heating process of temperature cycling, the flex

substrate expands more than silicon chip due to its

greater CTE, and thus the flex tends to push the solder

joints outward. In the other words, the solder joints try

to pull the flex substrate inward, as illustrated in Fig.

13(a). However, these compressive forces exerted by

solder joints on the flex substrate are offset from the

center of the flex substrate, and thus they induce bending

moments on the flex, which in turn cause flex bending or

buckling because of the compliant characteristic of the

flex. Fig. 13(a) illustrates the flex buckling mechanism.

Actually, the mechanics situation for the rigid substrate

is the same, but because the rigid substrate is much

stiffer than solder, solder is deformed instead of the rigid

substrate. Fig. 13(b) illustrates exaggerated thermal

bending of flex substrate in a flip chip on flex assembly.

As illustrated in Fig. 13(b), solder joint deformation

and thus strain and stress in solder joint are much re-

duced because of the flex buckling. Therefore, solder

joint reliability can be improved. This situation is very

similar to the underfill encapsulant effect on flip chip

assembly. Underfill encapsulant forces substrate (even

rigid ones) bending and thus reduce thermal strain in

solder joints [12]. Finite element analysis of flip chips on

both ceramic and FR-4 substrates has obtained the same

results that thermal strain can be reduced by bending of

substrates by underfilling [13,14]. Our accelerated tem-

perature cycling tests show that solder joint reliability is

much improved by using flex substrates. Meanwhile,

TMA results show that the compliant flex substrate

deforms under thermal cycling. We believe that a certain

amount of the thermal strain in solder joint is trans-

ferred to the compliant flex substrate during thermal

cycling and the flex substrate deformation can release

the strain and stress in the solder joints. Therefore, we

attribute the reliability improvement of flip chip on flex

assembly to flex buckling or bending under temperature

cycling.Fig. 12. Schematic of exaggerated thermal displacements in flip

chip on substrate structure.

Fig. 13. Flex buckling in flip chip on flex assembly. (a) The

buckling mechanism; (b) schematic of exaggerated thermal

bending of flex substrate in flip chip on flex package.


On the other hand, finite element modeling showed

that flip chip solder joint fatigue lifetime could be much

improved by reducing die thickness and thus reducing

the effective stiffness of the die, when die thickness

approaches zero, the fatigue life goes to infinity [15].

Similarly, we believe that making the other side (sub-

strate) of the solder joint compliant should also improve

reliability. Hence, the worst-case scenario for a solder

joint occurs when both the die and substrate have a

relatively high stiffness such that the amount of flexi-

bility in both is minimized.

5. Conclusion

The effect of substrate flexibility on solder joint reli-

ability has been studied. Temperature cycling results

showed that the thermal fatigue lifetime of solder joints

was improved by using flex substrates. TMA with static

force mode of a DAM system was used to monitor the

flex displacement during thermal cycling, and it was

found that the flex substrate buckles during temperature

cycling. It was indicated that the thermal strain and

stress in solder joints could be reduced by flex buckling

or bending that otherwise would be severe if a stiffer

substrate is used. The reliability improvement of flip

chip on flex assembly over flip chip on rigid board was

attributed to flex buckling or bending under temperature

cycling.

Acknowledgement

This work was supported in part by the ERC Pro-

gram of the National Science Foundation under award

number EEC-9731677.

References

[1] Lau JH. Flip chip technologies. McGraw Hill; 1995.

[2] Tummala RR, Rymaszewski EJ, Klopfenstein AG. Mi-

croelectronics packaging handbook. Chapman & Hall;

1996.

[3] Ray SK, Beckham KF, Master RN. Device interconnec-

tion technology for advanced thermal conduction modules.

IEEE Trans Compon Hybr Manuf Technol 1992;15(4):

432–7.

[4] Lau JH. Ball grid array technology. McGraw Hill; 1995.

[5] Liu XS, Haque S, Lu G-Q. Three-dimensional flip-chip on

flex packaging for power electronics applications. IEEE

Trans Adv Packag 2001;24(1):1–9.

[6] Liu XS, Jing X, Lu G-Q. Chip-scale packaging of power

devices and its application in integrated power electronics

modules. IEEE Trans Adv Packag 2001;24(2):206–15.

[7] Liu XS, Haque S, Wang J, Lu GQ. Packaging of integrated

power electronics modules using flip-chip technology. In:

Proceedings of the 15th Annual Power Electronics Con-

ference and Exposition, 6–10 February 2000. p. 290–6.

[8] Liu XS. Processing and reliability assessment of solder

joint interconnection for power chips. Ph.D. dissertation

(Chapter 4), Virginia Polytechnic Institute and State

University, February 2001.

[9] McCluskey P. Fatigue and intermetallic formation in lead

free solder die attach. In: Proceedings of IPACK�01, ThePacific Rim/ASME International Electronic Packaging

Technical Conference and Exhibition, Kauai, Hawaii,

USA, 8–13 July 2001. p. 1–7.

[10] TA instruments dynamic mechanical analyzer (DMA) 2980

manual, TA corporation.

[11] Dillard DA, Park TG, Zhang H, Chen B. Measurement of

residual stresses and thermal expansion in adhesive bonds.

In: Proceedings of the 22nd Annual Meeting of the Adhe-

sion Society. Panama City Beach, Florida, 21–24 Febraury

1999. p. 336.

[12] Lau JH. Chip on board technologies for multichip mod-

ules. New York: Van Nostrand Reinhold; 1994.

[13] Clementi J, McCreary J, Niu TM, Palomaki J, Varcoe J,

Hill G. Flip chip encapsulation on ceramic substrates. In:

Proceedings of 43rd ECTC, Orlando, FL, 1993. p. 175.

[14] Tsukada Y, Mashimoto Y, Nishio T, Mil N. Reliability

and stress analysis of encapsulated flip chip joint on expoxy

based printed circuit packaging. In: Proceedings of the

Joint ASME/JSME Conference on Electronic Packaging,

vol.2, Materials, Process, Reliability, Quality Control and

NDE, 1992. p. 827.

[15] Scott F. Popelar, A parameter study of flip chip reliability

based on solder fatigue modeling. In: Proceedings of IEEE/

CPMT International Electronics Manufacturing Technol-

ogy Symposium, 1997. p. 299–307.

[16] Semiconductor Data Book, IXYS Corporation, Santa

Clara, CA, 1998.


Documents

Effect of substrate flexibility on solder joint reliability