Available online at www.sciencedirect.com ScienceDirect Procedia Engineering 79 (2014 ) 333 338 37th National Conference on Theoretical and Applied Mechanics (37th NCTAM 2013) & The 1st International Conference on Mechanics (1st ICM) Reliability Assessment of the Temperature Profiles Effect on the Power Module H. C. Huang a b, T. Y. Hung a b, S. Y. Lin a b, K. H. Liao c, C. C. Wang c, K. N. Chiang a b,* a Advanced Microsystem Packaging and Nano-Mechanics Research Lab b Dept. of Power Mechanical Engineering, National Tsing Hua University, HsinChu, Taiwan, R.O.C. c DELTA Electronics, INC., Taoyuan 333, Taiwan. Abstract Power devices nowadays possess low switching and conduction loss characteristics owing to the improvement in semiconductor device manufacturing. Insulated gate bipolar transistors are widely utilized in power modules because of such characteristics. When a power module is subjected to cyclic temperature load, the thermal stress resulting from the mismatch among the coefficients of thermal expansion of materials causes fatigue to the interface of the materials. Different operating requirements cause the module to have different temperature profiles. Different dwell times and ramp rates produce different stress relaxation, which eventually affect the reliability of the power module. A 2D finite element model has been established based on an actual sample. The model was subjected to thermal cycling between -40 C and 125 C to determine the effect of dwell time and ramp rate on the creep behavior of the solder. The results indicate that dwell time produces accumulative creep strain because of stress relaxation. Furthermore, dwell time at a high temperature leads to evident stress relaxation. Ramp section produces more accumulative creep strain than dwell phase in one thermal cycle. Decreasing the ramp rate increases creep strain and diminishes reliability. 2013 2014 The Elsevier Authors. Ltd. This Published is an open by Elsevier access Ltd. article under the CC BY-NC-ND license Selection (http://creativecommons.org/licenses/by-nc-nd/3.0/). and peer-review under responsibility of the National Tsing Hua University, Department of Power Mechanical Selection and peer-review under responsibility of the National Tsing Hua University, Department of Power Mechanical Engineering. Engineering Keywords: power module; creep; ramp rate; dwell time; thermal cycling 1. Introduction * Corresponding author. Tel.: +8-863-574-2925; fax: +8-863-574-5377 E-mail address: knchiang@pme.nthu.edu.tw 1877-7058 2014 Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer-review under responsibility of the National Tsing Hua University, Department of Power Mechanical Engineering doi:10.1016/j.proeng.2014.06.352
334 H.C. Huang et al. / Procedia Engineering 79 ( 2014 ) 333 338 The development of high-power and high-reliability electronic products is a necessary trend given the current emphasis on renewable energy and environmental protection. Power modules nowadays have low switching and conduction loss characteristics because of the improvement in semiconductor device manufacturing. Power modules are widely utilized in various electronic applications, such as auto-motion control, power supply, renewable energy, hybrid gasoline-electric vehicles, and railway traction systems, because of such characteristics. In a power module, the pad is mounted on the active region of the insulated gate bipolar transistor chip, which is attached to the direct bonded copper (DBC) substrate by the solder. A DBC substrate has good electrical insulation and allows for heat dissipation. The structure is bonded to the copper plate with the solder for better heat dissipation. The wire is then bonded to the pad surface through ultrasonic wedge bonding technology (Fig. 1). The thermal stress resulting from the coefficients of thermal expansion (CTE) mismatch among materials, which functions as the power module under cyclic temperature load, causes fatigue at the interface of the materials. The power module has three main failure modes: solder fatigue, bond wire lift off, and bond wire heel cracking. Solder fatigue is investigated in this study to evaluate the reliability of the power module. 1.1. Thermal cycling test Fig. 1. The structure of power module Thermal cycling test (TCT) is a common reliability test. The ability of test samples to resist temperature variation is evaluated in TCT. Test samples are subjected to an environment that is more severe than the actual situation to accelerate destruction. Hung et al. [1] conducted a thermal cycling experiment to determine the reliability of the solder. Power modules were subjected to thermal cycling between -40 C and 125 C with 66min and 30min cycling periods to determine the relation between cycle period and reliability. The initial crack life and crack propagation rate were obtained by linear regression and slope analysis. The results are shown in Table 1. A long cycling period causes significantly shorter initial crack life and higher crack propagation rate than a short cycling period. Different cycling periods affect the reliability of the solder. These findings can be attributed to time-dependent creep behavior. Creep behavior occurs when homologous temperature, which is the ratio of ambient temperature to melting temperature, is larger than 0.5K, as shown in Eq. (1) and when the structure withstands fixed loading for some time [2]. T Th 0.5 (1) Tm where Th is the homologous temperature and Tm is the melting temperature.
H.C. Huang et al. / Procedia Engineering 79 ( 2014 ) 333 338 335 Creep behavior is related to the effect of dwell time, temperature, and strain rate. Different dwell times and ramp rates eventually affect the reliability of the power module. However, the mechanical behavior remains unclear. Therefore, this study aims to decouple the effect of dwell time and ramp rate by finite element (FE) analysis. Table 1. The experiment results [1] 2. The creep behavior of solder Period Initial crack life Crack propagation rate 30min 73cycles 0.007mm/ cycle 66min 37cycles 0.011mm/ cycle Solder fatigue induced by CTE mismatch particularly occurs in the solder layer between the DBC substrate and the copper baseplate because of geometric effect. The FE model in this research consisted of DBC, solder layer, and copper baseplate. The A-A cross section of the power module was analyzed to establish a 1/2 symmetric model (Fig. 2). The model was meshed with PLANE42 elements, and the solder was meshed with PLANE182 to investigate creep behavior. The detailed dimensions are listed in Table 2 and Fig. 3. Fig. 2. A-A cross section of power module Fig. 3. The 1/2 symmetric model Table 2. The dimension of symmetric model Material DBC,copper(Cu) DBC,ceramic SnAg solder Cu baseplate Size 14.5mm 14.5mm 14.5mm 19.5mm The material properties are listed in Table 3. The solder was designed with multi-linear material properties [3] (Fig. 4). The equation for creep behavior can be regarded as a function of temperature and stress. Solder behavior can be described by the Garofalo-Arrhenius steady-state creep model [4] as shown in Eq. (2). The parameters are listed in Table 4. Q n A [sinh( B ) ] ert (2) where is the equivalent creep strain rate, is the equivalent von Mises stress(n/mm 2 ), n is the stress exponent for dislocation creep [6], Q is the active energy for creep deformation process, R is the gas constant, and T is absolute temperature.
336 H.C. Huang et al. / Procedia Engineering 79 ( 2014 ) 333 338 Table 3. Material properties of power module Young s modulus (MPa) Poisson s ratio Coefficient of thermal expansion (ppm/ C) Ceramic 300 0.25 8 Copper 110 0.35 17 Sn 96.5Ag 3.5 Multilinear 0.4 22.36 Table 4. Garofalo-Arrhenius model[5] A (1/s) B (MPa -1 ) n Q (kj/mol) 2.46E5 0.0913 5.5 72.5 Fig. 4. Multi-linear material properties of solder [3] The temperature profiles listed in Table 5 were analyzed in terms of the effects of ramp rate and dwell time based on the JEDEC standard. Ambient temperature was set to 25 C. The temperature ranged from -40 C to 125 C. Table 5. The arrangements of Temperature profiles 3. Simulation result 3.1. Preliminary analysis Period (min) Dwell time (min) Ramp rate ( C/min) 60 15 11 40 5 11 34 15 82.5 The stress concentration on the solder near the corner of the DBC substrate is caused by CTE mismatch (Fig. 5). Solder failure mechanism is characterized by cracks initiating from the corner of the DBC substrate rather than from the edge of the solder and propagating along the upper surface of the solder [7] Comparing creep strain with plastic strain, creep impact is much greater than the impact of plastic deformation (Fig. 6). Stress is released by creep when the material reaches yield stress before entering the plastic zone. Therefore, this study will focus on the discussion of creep strain.
H.C. Huang et al. / Procedia Engineering 79 ( 2014 ) 333 338 337 Fig. 5. von Mises stress distribution (unit: MPa) Fig. 6. Comparison of creep strain and plastic strain 3.2. Temperature profile effect The equivalent creep strain in the various stages of a cycle was analyzed to observe the cumulative situation. The accumulation of creep strain was more significant when the power module was subjected to the ramp section than to the dwell phase. Considering that the power module can withstand loading in the dwell phase of the hightemperature section, creep behavior induced more obvious stress relaxation than in the dwell phase of the lowtemperature section. When the ramp rate was fixed at 11 C/min and dwell time was changed, the equivalent creep strain variations increased with the increase in dwell time because of stress relaxation (Fig. 7). Fig. 7. Comparison of ramp section and dwell phase in a cycle When the dwell time was fixed at 15min and the ramp rate was changed to 11 C/min and 82.5 C/min, the yield stress increased with the increase in ramp rate. This result indicates that creep deformation is evident at a low strain rate. Thus, the increment of creep strain increased with the decrease in ramp rate (Figs. 8a and 8b). Lefranc et al. [8] studied the effect of ramp rate on a power module subjected to TCT. The experimental results indicated that a slow ramp rate generates more damage per cycle. This finding is consistent with the trend of the simulation in the current study.
338 H.C. Huang et al. / Procedia Engineering 79 ( 2014 ) 333 338 (a) (b) Figs. 8. The equivalent creep strain under different ramp rate (a) Ramp rate in 11 C/min; (b) Ramp rate in 82.5 C/min 4. Conclusion The CTE mismatch among the materials resulted in stress concentration after the power module was subjected to TCT. The von Mises stress was concentrated on the solder layer, which was near the corner of the DBC substrate where the initial crack occurred. The simulation results indicate that the creep strain increased with the increase in dwell time because of stress relaxation. Creep strain accumulated at the dwell segment with high temperature in a cycle. The accumulation of creep strain was more significant when the power module was subjected to ramp section than to dwell phase in a cycle. Moreover, ramp rate affected yield stress. A slow ramp rate generated more damage per cycle. This result is attributed to the accumulation of creep strain with the decrease in ramp rate. Acknowledgements Acknowledgements The authors would like to thank the National Science Council for providing financial support under Project Number NSC99-2221-E-007-013-MY3. References [1] T. Y. Hung, C. J. Huang, C. C. Lee, C. C. Wang, K. C. Lu, and K. N. Chiang, Investigation of Solder Crack Behavior and Fatigue Life of the Power Module on Different Thermal Cycling Period, Microelectron. Engineering 107 (2012) 125-129. [2] F. Garofalo, Fundamentals of creep and creep-rupture in metals, Macmillan Company, New York, 1965. [3] C. C. Chiu, C. J. Wu, C. T. Peng, K. N. Chiang, T. Ku, and K. Cheng, Failure life prediction and factorial design of lead-free flip chip packge, Journal of the Chinese Institute of Engineers 30 (2007) 481-490. [4] L. Hua, S. Ridout, C. Bailey, L. S. Wei, A. Pearl, and M. Johnson, Computer simulation of crack propagation in power electronics module solder joints, International Conference on Electronic Packaging Technology & High Density Packaging, Shanghai, China (2008) 1-6. [5] R. Darveaux, K. Banerji, A. Mawer, and G. Dody, Ball grid array technology, McGraw-Hill, New York, 1995. [6] H. Ma, and C. J. Suhling, A review of mechanical properties of lead-free solders for electronic packaging, Journal of Materials Sci. 44 (2009) 1141-1158. [7] J. M. Thebaud, E. Woirgard, C. Zardini, S. Azzopardi, O. Briat, and J. M. Vinassa, Strategy for designing accelerated aging tests to evaluate IGBT power modules lifetime in real operation mode, IEEE Transactions on Components and Packaging Technologies 26 (2003) 429-438. [8] G. Lefranc, T. Licht, and G. Mitic, Properties of solders and their fatigue in power modules, Microelectron. Reliab. 42 (2002) 1641-1646.