Effects of thermodynamic loading
10/01/2000
Study indicates that vibration prolongs thermal fatigue life of BGA solder joints.
By C. BASARAN, Y. ZHAO, A. CARTWRIGHT AND T. DISHONGH
 Figure 1, Optical table. |
It is well-known that the dominant failure mode for solder joints is low-cycle thermal fatigue; this is caused by the thermal cycling of semiconductor devices and the coefficient of thermal expansion (CTE) mismatch between the soldered layers. The literature regarding the contribution of vibration damage to the overall fatigue life of solder joints is scarce. The common opinion is that vibrations only cause elastic material response.1-6 Recent tests have revealed that vibration loading at room temperature can be more damaging than similar vibration combined with thermal loading.7,8 Linear damage super-position rule (Miner's Rule) is used in practice as a first-order approximation for damage calculation and fatigue-life prediction for concurrent loading.9,10 Miner's Rule ignores the thermal and vibration interaction, and material microstructure and material property evolution, which may cause a prediction to deviate from an actual situation.
In this project, concurrent thermal cycling and vibration tests are conducted to explore the cyclic behavior of solder joints and plasticity in vibration coupled with thermal cycling. Actual ball grid array (BGA) packages are used for testing. High sensitivity moiré interferometry (MI) is used to capture the in-plane cyclic inelastic deformation field with submicron resolution. MI has been used in the past for measuring inelastic deformations only up to a few thermal cycles. A new optical grating replication technology and opto-mechanical techniques allow for measuring inelastic deformations for hundreds of thermal cycles.
Experimental Setup
Moiré Interferometry: Moiré interferometry setup used here is similar to the one described in other studies.11 The optical table of MI is shown in Figure 1, and the interferometry layout is shown in Figure 2 in only one direction for simplicity. Two coherent beams lie in the same plane to create an interferometry pattern. U-field is a term that refers to the fringe field generated by the two beams in the horizontal plane, and V-field denotes the fringe field generated by the vertical two beams. The fringe contour represents the equal deformation points, and each fringe is given an order number. An increase of the order number with regard to a reference order number represents increasing of deformation from the reference point.
 Figure 2, above. Optical setup of moiré interferometry. |
The relationships between the fringe order and actual in-plane deformation can be represented by the equations:
 |
where U is the displacement in the x direction; V is the displacement in the y direction; and f = 2fs, where fs is the frequency of the specimen diffraction grating.12 In this study, fs = 1,200 lines/mm, Nx is the horizontal fringe order, and Ny is the vertical fringe order. Differentiation of the displacement with respect to the two basic directions - horizontal (x) and vertical (y) - gives the strain equations:
 |
Specimen Preparation: The BGA package used for testing is cut through the center of a row of solder balls (Figure 3). Because of the structural symmetry of the test sample, only half of the specimen is plotted. All the solder joints are Sn63/Pb37 eutectic solder alloy. The solder joints selected for the measurement of the thermal response are at the bottom layer, which bonds the FR4 PCB layer and polymer connector layer. With a of total 30 joints in each row, these solder joints have been observed to be the most vulnerable interconnections of the entire device under thermal loading because of the significant thermal expansion mismatch between the two layers.
 Figure 3. Cross-section of BGA package specimen. |
For the convenience of analysis, solder joints are numbered consecutively from the free edge to the center of the package (Figure 3). The specimen is symmetrically clamped at the two ends of the middle FR-4 layer during environmental cycling; therefore, it is reasonable to assume that the stresses and strains caused by cycling in the sample are symmetric. The specimen is environmentally cycled together with the fixture to ensure the consistency of the boundary conditions through the entire cycling procedure.
Concurrent Vibration and Thermal Cycling: The concurrent test is conducted using an environmental chamber unit and an electrodynamic shaker (Figure 4a). The in-plane acceleration is applied on the specimen.
 Figure 4a. Concurrent cycling unit. |
In order to register the initial reference fringe field, the un-deformed specimen grating is used to tune the optical setup to a null field. The specimen is then taken to the loading unit for cycling. At the end of cycles, the specimen is taken out, the plastic deformation at that stage is measured using the MI and the optical fringe field is digitally recorded on a computer. After measurement, the sample is put back in the loading unit for further concurrent (thermal plus vibration) cycling.
 Figure 4b. Temperature history. |
Because MI measurement relies on the initial reference field, the optical setup needs to be carefully protected from any disturbance (i.e., to be kept in the initially aligned manner until testing ends). Furthermore, a special position register is designed to ensure that the specimen under test always occupies the same optical space after each cycling. The temperature profile applied during testing is shown in Figure 4b. Temperature range is -55 to 125°C, with ramping rate of 20°C/min, dwelling time of 12 minutes at highest and lowest temperature respectively, and a total cycle period of 42 minutes. The module is subjected to a harmonic vibration that has a frequency of 100 Hz and an acceleration of 50 g.
Discussion of Results
Figures 5a and 5b show the MI fringe pattern of the deformation fringe field after the first combined loading cycle. The irreversible shear strain distribution of the observed row of solder joints is shown in Figure 6. Because of the symmetry of the structure and boundary conditions, strain field is measured in only half of the structure. The largest strain occurs at free-edge joint number one. Inner joints show smaller values of irreversible shear strain. This deformation trend is in agreement with thermal cycling results, and is also discussed using finite element method (FEM).11,13 It is clear that free-edge joints will yield first as the maximum shear strain develops from the free-edge joint toward inner joints. De-bonding or fracture may initiate at the free-edge solder joint. Figure 7 shows shear strain accumulation versus number of thermal cycles under combined loading conditions. It can be seen that the fastest increment of the irreversible shear strain is induced in the first cycle. As the number of cycles increases, the amount of the increment of plastic deformation in each cycle is reduced. This curve shows that Coffin-Manson curves or tin/lead constitutive models based on Coffin-Manson curves for fatigue life predictions yield inaccurate estimations of fatigue life.
 Figure 5. Deformation fringe patterns after the first combined loading cycle: (a) U-field, (b)V-field. |
According to the Coffin-Manson equation, cyclic plastic strain is constant from one cycle to another. This plateau trend in the shear strain vs. the number-of-cycles curve is similar to thermal cycling testing results previously reported.11 Virtually every solder joint presents this characteristic. This demonstrates the stabilization process in cyclic loading and is in line with reports that damage accumulation is not linear.10
 Figure 6. Irreversible shear strain distribution. |
Concurrent Loading vs. Pure Thermal Cycling: As shown in Figure 8, the average plastic strain at the end of the first cycle under concurrent loading is compared to the results obtained from pure thermal cycling with the same temperature profile.11 In Figure 9, the irreversible shear strain distribution curves obtained from the two loading conditions are also plotted for the fourth cycle. The difference in response shows that vibration has significant effect on the inelastic behavior of solder joints, especially when it is coupled with thermal cycling. The same solder joints respond elastically at room temperature, and plastically at elevated temperature to the same vibration.14 The response under concurrent thermal and vibration testing cannot be explained by simple superposition of thermal cycling and dynamic cycling. These results show that Miner's accumulative damage rule cannot be used for BGA solder joints.
 Figure 7. Shear strain accumulation vs. number of thermal cycles under combined loading conditions. |
Irreversible shear strain at the end of cycles is smaller in the concurrent loading case than in the case of pure thermal cycling. It seems that dynamic force, which is vibrating in a high frequency, restricts the creep mechanism and reduces the overall cyclic plastic strain accumulation. A large number of short vibration cycles are imposed on a long thermal cycle. The total plastic energy dissipated during concurrent cycling is still large because creep cannot be ignored during low-frequency high-temperature vibration.14 The energy dissipated in vibration cycles can vary according to different values of vibration frequency and acceleration. Low-frequency high-temperature vibrations usually induce significant plastic deformations. High-frequency low-temperature vibrations are usually in the elastic region.
Cyclic Stabilization: The irreversible strain and number-of-cycles curves are plotted in Figure 10 for pure thermal cycling and concurrent thermal-vibration cycling. It is obvious that the irreversible shear strain per cycle tends to stabilize as the number of cycles increases. However, in the latter case, this stabilization process is approximately two cycles - much faster than the former case, where it is approximately eight cycles. This is probably related to the effect of straining to microstructure stabilization, which in turn relates to the stabilization of creep flow behavior of eutectic solder alloy at high temperatures.
 Figure 9. Irreversible shear strain distribution curves for the fourth cycle. |
The as-cast tin/lead solder joint is usually microstructurally unstable, and exhibits unstable flow behavior during the first few cycles. The stabilization is represented by approaching a constant value of irreversible strain per cycle. After that, microstructure will still evolve, but at a much slower pace; straining makes stable solder microstructure.15 When vibration is coupled with thermal cycling, the solder material is strained much faster by dynamic force back and forth in a high frequency, which contributes to the faster stabilization of solder microstructure.
Conclusions
To study the contribution of vibration to thermal fatigue life of microelectronics solder joints, a series of concurrent thermal cycling and vibration tests are conducted on actual BGA packages. Although thermal fatigue is the dominant failure mode, vibration significantly modifies the total behavior of solder joints in the concurrent testing. Miner's Rule and Coffin-Manson curves are not accurate for concurrent environments because of their lack of consideration of thermal and vibration interaction.
 Figure 10. Irreversible strain and number-of-cycles curves for pure thermal cycling and concurrent thermal-vibration cycling. |
The cyclic stabilization behavior of solder joints is reflected in the stabilization of cyclic inelastic strain. This behavior is possibly related to the microstructure evolution process of the eutectic solder joints, and vibration contributes to the evolution process by straining the material in a faster pace.
Tests described here are aimed at better understanding the material behavior of actual solder joints under realistic thermal and vibration loading and they provide a solid basis for more accurate material modeling and fatigue-life prediction.
Acknowledgements
This research project is sponsored by the Department of Defense Office of Naval Research Young Investigator Award to the first author. Helpful discussions with Dr. Roshdy Barsoum, director of the solid mechanics program at ONR, are greatly appreciated.
References
- H. S. Blanks, "Accelerated Vibration Fatigue Testing of Leadless and Solder Joint," Microelectronics and Reliability, Pergoman Press, UK, Vol. 15, 1976, pp. 213-219.
- H. W. Markstein, "Designing Electronics for High Vibration and Shock," Electronic Packaging & Production, April, 1987, pp. 40-43.
- D. S. Steinberg, Vibration Analysis for Electronic Equipment, John Wiley & Sons, N.Y., 1988.
- E. Suhir and Y. C. Lee, "Thermal, Mechanical, and Environmental Durability Design Methodologies," Electronic Material Handbook, Vol. 1, ASM International, New York, 1998.
- D. B. Barker and K. Sidharth, "Vibration Induced Fatigue Life Estimation of Corner Leads of Peripheral Leaded Components," Proceedings, ASME International Mechanical Engineering Congress and Exposition, ASME, New York.
- K. Darbha, S. Ling, K. Upadhyayula and A. Dasgupa, "Stress Analysis of Surface-Mount Interconnects Due to Vibrational Loading," paper presented at the ASME International Mechanical Engineering Congress and Exposition, Atlanta, Ga., 1996.
- K. Upadhyayula and A. Dasgupta, "An Incremental Damage Superposition Approach for Reliability of Electronic Interconnects Under Combined Accelerated Stresses," ASME International Mechanical Engineering Congress & Exposition, Dallas, Texas, Nov.16-21, 1997.
- R. Chandaroy and C. Basaran, "Damage Mechanics of Surface Mount Technology Solder Joints Under Concurrent Thermal and Dynamic Loading," Journal of Electronic Packaging, Vol.121, June 1999, pp. 61-68.
- D. Barker, J. Vodzak, A. Dasgupta and M. Pecht, "Combined Vibrational and Thermal Solder Joint Fatigue - A Generalized Strain Versus Life Approach," Journal of Electronic Packaging, Vol. 112, June1990, pp. 129-134.
- C. Basaran and R. Chadaroy, "Mechanics of Pb40/Sn60 Near Eutectic Solder Alloys Subjected to Vibrations," Applied Mathematical Modeling, Vol. 22, 1998, pp. 601-627.
- Y. Zhao, C. Basaran, A. Cartwright and T. Dishongh, "Thermomechanical Behavior of Micron Scale Solder Joints: An Experimental Observation," Journal of the Mechanical Behavior of Materials, Vol.10, No.3, 1999, pp. 135-146.
- D. Post, B. Han and P. Ifju, High Sensitivity Moiré, Springer-Verlag, 1994.
- C. Basaran and Y. Zhao, "Closed Form vs. Finite Element Analysis of Laminated Stacks," International Journal of Finite Elements in Analysis & Design, accepted for publication, 1999.
- Y. Zhao and C. Basaran, "Thermomechanical Behavior of Micron Scale Solder Joints Under Dynamic Loads," Mechanics of Material, Vol.32, Iss. 3, Feb. 16, 2000, pp. 161-173.
- B.P.Kashyap and G.S.Murty, "Superplastic Behavior of the Sn-Pb Eutectic in the As-Worked State," Metallurgical Transactions, Vol. 13A, Jan. 1982, pp. 53-58.
C. BASARAN is associate professor and director, Y. ZHAO is a Ph.D. candidate, and A. CARTWRIGHT is assistant professor at the Electronic Packaging Laboratory at the State University of New York, Buffalo. T. DISHONGH is manager of packaging Q and R at Intel Corp. For more information, contact C. Basaran at SUNY-Buffalo, 212 Ketter Hall, North Campus, Buffalo, NY 14260; 716-645-2114; Fax: 716-645-3733; E-mail: [email protected].
null