Fatigue and dwell-fatigue behavior of nano-silver sintered lap-shear joint at elevated temperature

Microelectronics Reliability 54 (2014) 648–653

Contents lists available at ScienceDirect

Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel

Fatigue and dwell-fatigue behavior of nano-silver sintered lap-shear joint at elevated temperature Yansong Tan a, Xin Li b,⇑, Xu Chen a a b

School of Chemical Engineering and Technology, Tianjin University, Tianjin, PR China School of Material Science and Engineering, and Tianjin Key Laboratory of Advanced Joining Technology, Tianjin University, Tianjin, PR China

a r t i c l e

i n f o

Article history: Received 30 August 2013 Received in revised form 2 December 2013 Accepted 2 December 2013 Available online 3 January 2014

a b s t r a c t Load-controlled fatigue and dwell-fatigue tests were conducted at elevated temperature to describe the high temperature mechanics behavior of nano-silver sintered lap-shear joints. The results showed that the shear strength of nano-silver sintered lap-shear joints was strongly temperature dependent, and almost halved at the temperature of 325 °C. The Basquin model was used to assess the fatigue life of the joints at elevated temperature and the constants in the model were figured out, which yielded good prediction for experimental data. In dwell-fatigue tests, at the temperature of 325 °C, creep was found the dominant factor that resulted in failure acceleration and cyclic life reduction. With the temperature decreasing to 225 °C, the creep played a less important role to the total deformation. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction First level packaging involves the interconnect technology, which interconnects the chip to a carrier and provides mechanical continuity, physical protection, electrical connection, as well as thermal cooling for the integrated circuit [1–3]. The performance of interconnect technology is crucial to the integrity of integrated circuit, which in turn is vital to the overall functioning of the assembly [4,5]. The existing traditional interconnect technologies consist of wire bonding, solder reflowing and conductive adhesive curing. In recent years, the ever-smaller feature size of integrated circuit imposes increasingly stringent requirements on weight reduction, size miniaturization as well as high thermal dissipation capability of integrated circuit [6–10]. The appearance of high power density systems like wide-band gap semiconductors makes the operating temperature higher, which is beyond the capability of traditional die-attaching materials and interconnect technologies. The present situation of electronic packaging is promoting the introduction of a superior die-attach technology for highpower electronic packaging to the market place. In the 1970s, the viewpoint of diffusion welding silver film was firstly introduced by O’brien et al. [11], and thus a new die-attaching technology named low-temperature sintering technology was remarkably promoted. Under the mechanical pressure of about 40 MPa, micro-sized silver powder could be sintered at temperature below 300 °C [12]. From then on, silver was used widely in microelectronic packaging as a promising interconnection material between substrates and chips because of its superior electrical/ ⇑ Corresponding author. Tel./fax: +86 22 27405889. E-mail address: [email protected] (X. Li). 0026-2714/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.microrel.2013.12.007

thermal conductivity, high melting temperature (Tm = 960 °C), and good reliability. However, for brittle silicon chips and ceramic substrates, the aided pressure might be destructive even slightest irregularities [13,14]. To raise the sintering driving force of this interconnection material, Bai and Lu introduced a paste which was mainly formed by nano-sized silver powders, and gained the close attention of both scientists and power electronics engineers [15]. Before the nano-silver sintering technology coming into practical application, both the sintering process and the mechanical properties of nano-silver paste have been studied. In recent years, a low-temperature sintering profile with sintering temperature of 285 °C, heating rate of 10 °C/min, and holding time of 60 min was introduced by Wang et al. [16]. Yu et al. studied the tensile behavior of low-temperature sintered nano-silver films and proved that accumulation of plastic strain took place in silver-bonding layer during thermal cycling, which might lead to the final failure of the chip-attachment [1]. Wang et al. stated that the fatigue failure of sintered silver paste was dominated by ratcheting response, especially for elevated temperature [7]. These studies only revealed the properties of sintered silver film, which in turn imposed a limitation on consideration of the thermally induced strain. However, the service environment of the electronic equipment was very complex with frequently mechanical vibration and temperature change. Both cyclic loads and stress concentration were induced by the thermal expansion coefficient mismatch between substrates and chips in actual application [17–20]. As a result, Li et al. [21–23] constructed a lap-shear structure to study the mechanical properties of sintered nano-silver as a joint. Based on this lap-shear structure, fatigue and dwell-fatigue tests were conducted in this article in order to understand the cyclic failure

649

Y. Tan et al. / Microelectronics Reliability 54 (2014) 648–653

mechanism of nano-silver sintered lap shear joint at elevated temperature. The relationship between creep damage and fatigue damage in the low-cycle regime was studied.

2. Sample preparation and experiment procedure The nano-silver sintered lap shear joints were employed as specimens in this study. The pastes with nano-silver particles smaller than 50 nm were provided by NBE Tech, LLC [24]. Widely used Cu with purity of about 99.9% was chosen as substrates [25]. After sintered according to a recommended heating profile for chip attachment as shown in Fig. 1 [21–23], the specimens of 2 mm  1 mm and thickness of 50 lm were obtained as exhibited in Fig. 2. All the tests were conducted on Micro Uniaxial Fatigue Testing System (MUT-1020) provided by CARE Measure & Control Co., Ltd. as given in Fig. 3. In order to obtain the shear stress–strain relation of sintered lap shear joints as the base for cyclic tests, a series of shear tests were conducted at four different ambient temperatures of 25 °C, 125 °C, 225 °C and 325 °C. The loading conditions of fatigue and dwell-fatigue tests are listed in Table 1. All the shear and cyclic tests were conducted under stress-controlled mode. Three samples were conducted for every loading condition. In the present study, the shear strain was determined by dividing the joint displacement over the lapped joint thickness. In order to remove the effects of substrate and machine compliance, the joint displacement, which was used to transform into shear strain, was revised by elastic theory calculating and non-contact measurement, respectively.

3. Results and discussions

Fig. 2. Prepared sintered nano-silver lap-shear structure.

Fig. 3. The testing apparatus.

3.1. Shear behavior 3.2. Fatigue tests of fully reversed at elevated temperature As shown in Fig. 4, the effect of temperature on strain of nanosilver joints can be concluded. At room temperature of 25 °C, the failure strain is less than 1.5%, which contains a small part of plastic strain and thus a brittle failure. With the temperature increasing, the shear modulus decreases and the plastic flow of the joint is more obvious [21]. The average shear strength under four ambient temperatures is given in Fig. 5. With the gradually increasing of temperature, the shear strength decreases. At room temperature of 25 °C, the shear strength is as high as nearly 28 MPa, which almost halved at the temperature of 325 °C (13.3 MPa).

Fig. 1. Sintering profile of nano-silver paste.

3.2.1. Fatigue behavior Fig. 6 shows shear stress–strain hysteresis loops under different loading amplitudes at 325 °C. From Fig. 6 it can be found that the shear strain amplitude increases with the increasing of loading amplitude under fully reversed loading situation. The increase of hysteresis loops with loading amplitudes demonstrates the larger energy dissipating per unit volume during a cycle, which results in shorter cyclic life [26,27]. The plot of shear strain range versus number of cycles as shown in Fig. 7 illustrates that the initial shear strain range increases as the increase of the loading amplitude. The evolution of shear strain range can be divided into three stages. The first stage takes the longest duration of the fatigue life at constant shear strain amplitude. When the fatigue test steps into the second stage, the shear strain slowly increases, which can be considered as the crack propagation. In the third stage, the fatigue damage accumulation accelerates and results in ultimate failure. The strain–stress hysteresis loops of the nano-silver joint at load amplitude of 6 MPa and 8 MPa are taken as examples in Fig. 8. The enclosed area of the hysteresis loops represents the cyclic plastic energy consumed in each cycle. At the given loading amplitudes, the enclosed area gradually increases with the number of cycles. The hysteresis loop experiences a rapid increase before the nano-silver joint coming to the final failure. The temperature effect on hysteresis loops under 7 MPa and 7 MPa loading amplitude is shown in Fig. 9. It can be found that at higher temperature of 325 °C, the loops are wider and thus more plastic strain is accumulated in one cycle. As a result, the fatigue life at the temperature of 325 °C is shorter than that at the temperature

650

Y. Tan et al. / Microelectronics Reliability 54 (2014) 648–653

Table 1 Loading conditions of fatigue and dwell-fatigue tests. Spec. ID

Temperature (°C)

Loading rate (MPa/s)

Stress amplitude (MPa)

Mean stress (MPa)

Dwelling time (s)

A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11 B1 B2 B3 B4 B5 B6 B7 B8 B9

325 325 325 325 325 325 325 325 325 325 325 225 225 225 225 225 225 225 225 225

2 2 2 2 2 2 3 3 3 3 3 2 2 2 2 3 3 3 3 3

10 9 8 7 6 5 3 3 3 3 0 10 9 8 7 3 3 3 3 0

0 0 0 0 0 0 3 3 3 3 6 0 0 0 0 3 3 3 3 6

0 0 0 0 0 0 0 1 5 9 Creep 0 0 0 0 0 1 5 9 Creep

Fig. 4. Shear stress–strain relationship of sintered nano-silver joint.

Fig. 6. Comparison of hysteresis loop under different loading condition.

Fig. 5. Shear strength of sintered nano-silver joint.

Fig. 7. Comparison of strain amplitude under different loading condition at the temperature of 325 °C.

of 225 °C. A point should be put forward that higher temperature could mobilize the dislocation in nano-silver joints and increase

the stress concentration of the joints interface, which leads to a significant reduction of fatigue life.

651

Y. Tan et al. / Microelectronics Reliability 54 (2014) 648–653

Fig. 8. The strain–stress hysteresis loops of the nano-silver joint at load amplitude of (a) 6 MPa and (b) 8 MPa.

Fig. 9. The temperature effects on hysteresis loops under (a) 7 MPa and (b) 9 MPa.

3.2.2. Fatigue life prediction The stress–life (S–N) data can be plotted linearly in a log–log scale which is firstly observed by Basquin [28]. Here the formula can be expressed as:

Ds=2 ¼ r0ft ð2Nf Þb where Ds/2 is the shear stress amplitude, Nf is reversals to failure, and 1rev = 1/2 cycle. As shown in Fig. 10, the relationship between the shear stress amplitude and the fatigue life for sintered nano-silver lap shear joint at 325 °C and 225 °C are linearly fitted as Basquin model in the logarithm coordinate system. The fatigue strength exponent b and fatigue strength coefficient r0ft for sintered lap shear joint are given in Table 2. It can be obviously observed that the fatigue strength coefficient increases with decreasing temperature. The fatigue strength exponent b has little change at different temperatures as presented in Fig. 10. The fatigue life of the nano-silver sintered lap-shear joints at the temperatures of 325 °C and 225 °C is well predicted.

Fig. 10. S–N curve for nano-silver sintered lap-shear joints.

3.3. The dwell-fatigue behavior of nano-silver sintered lap-shear joints Presently study shows little difference on the fracture surface for the cases of fatigue, dwell fatigue, creep, and simple shear tests at the temperature of 25 °C and 325 °C. The cross-section cracking pattern of cyclic tests had been studied in Ref. [22]. Under cyclic load, lots of micro crack nucleation occurred between grains

Table 2 The constants (r0ft and b) of Basquin model for sintered lap shear joints. Temperature (°C)

r0ft

b

225 325

19.2 15.3

0.1056 0.1061

652

Y. Tan et al. / Microelectronics Reliability 54 (2014) 648–653

gradually, and no obvious plastic flow was induced during cyclic loadings. When the micro crack density reached the level where formation of one or several macro cracks occurred, the silver joints would become unstable and the brittle fracture would occur in the nano-silver paste layer. That is to say that, the observation of the inner crack propagation is better than that of fracture surface for fracture mechanism investigation. The non-destructive X-ray computer tomography (CT) is intended to be used to observe the damage evolution inner the joint under cyclic loading. However, since no in situ observation is available, the specimens have to be repeatedly discharged from the testing apparatus and thus significantly affected on the cyclic life. Currently, the set-up of in situ X-ray Micro-Uniaxial Fatigue testing system is still under preparation. And the damage evolution under cyclic loading will be evaluated in further studies. Hence, in this research, only the study of macroscopic mechanical behavior of nano-silver sintered joints under cyclic loading was presented. 3.3.1. Shear strain evolution A series of dwell-fatigue tests (A7-A11, B5-B9) with peak stress holding, were conducted at the temperatures of 325 °C and 225 °C in this part. Fig. 11 shows the trapezoidal stress wave (take sample A9 for example) that used in the dwell-fatigue tests. For the sintered nano-silver joint, the shear strain evolution processes during the dwell-fatigue tests exhibit similar characteristics for all the specimens. According to the accumulation rate of shear strain in Fig. 12, the low cycle dwell-fatigue process can be divided into three stages. During the initial few cycles, the strain increases rapidly. However, as dwell-fatigue tests progressing, the sintered joints are strain hardening and thus the strain accumulation rate decreasing gradually [29,30]. After strain hardening reaching a saturated state, the strain increases linearly, which indicating that the creep fatigue process has entered a steady stage. This section takes the longest duration of the whole life. When the strain increases to a certain value, plastic deformation in the joint is aggravated and then the third stage starts. During this stage, the fatigue and creep failure greatly accelerate and ultimately result in final fracture. According to their features, the three stages are defined as the strain hardening stage, the steady deformation stage and the accelerating fracture stage, respectively [30].

Fig. 12. The shear strain–time relationship of the joint under dwell-fatigue loading condition.

3.3.2. Temperature effect on the dwell-fatigue process Dwell-fatigue and fatigue tests are conducted under 3 MPa mean shear stress and 3 MPa shear stress amplitude at temperatures of 225 °C and 325 °C. In addition, creep tests under the peak

stress of the previous cyclic loading, which is 6 MPa, are carried out for comparison. Fig. 13 shows the shear strain evolution under these three test conditions. Compared the figures of 225 °C and 325 °C, it can be concluded that the nano-silver sintered lap-shear joints are significantly temperature dependent and the elevated temperature strongly accelerates the creep damage on the failure progress. An interesting phenomenon should be pointed out that at the temperature of 325 °C, The shear strain rate of creep test is larger than that of the dwell-fatigue test (peak stress holding for 1s) and fatigue test. But at the temperature of 225 °C, the shear strain rate is lower under constant loading of creep. Reason for this phenomenon remains to be further studied. In this study, the life of the sintered joint is calculated in two ways, the fatigue life and the creep life. The fatigue life is defined as number of cycles to final failure and the creep life is the sum of the peak stress holding time. At the temperature of 225 °C, when the joints are tested under cyclic loading of 0–6 MPa (B5), the fatigue life is more than 100,000 cycles, and the introducing of peak stress holding (B6B8) has little effect. The strain evolution in fatigue and dwell-fatigue tests at 325 °C is given in Fig. 14. From Table 3 we can find that, when the temperature increases to 325 °C, the creep life is little affected by the various cyclic loading and thus the creep damage is more dominant for the joint life. Furthermore, the cyclic life of the joint is significantly reduced by the peak stress holding. Therefore, at higher temperature of 325 °C, creep failure is

Fig. 11. The loading condition of dwell-fatigue tests.

Fig. 13. Shear strain evolution at 225 °C and 325 °C.

Y. Tan et al. / Microelectronics Reliability 54 (2014) 648–653

653

References

Fig. 14. The average shear strain-cycles relationship of fatigue and dwell-fatigue tests.

Table 3 The dwell-fatigue tests result under different dwelling time at the temperature of 325 °C. Spec. ID

Dwelling time (s)

Fatigue life (cycles)

Creep time (s)

A7 A8 A9 A10 A11

0 1 5 9 Creep

555 264 180 51 0

0 264 900 459 241

found to be the dominant factor that resulted in the joint ultimate failure. 4. Conclusions In order to study the mechanics behaviors of nano-silver sintered lap-shear joint at elevated temperature, a mass of shear tests, fatigue tests as well as dwell-fatigue tests were conducted at the temperatures of 325 °C and 225 °C. The following conclusions are obtained. (1) Shear behaviors are temperature dependent. The shear strength decreases with the increase of temperature. The shear strength at room temperature is as high as 28 MPa, which halved at the temperature of 325 °C. (2) The Basquin model is presented to predict the fatigue life of the nano-sintered lap-shear joints at different temperatures, where the fatigue strength exponent b is a constant and the fatigue strength coefficient r0ft decreases with the elevating temperature. (3) At the temperature of 225 °C, the introducing of peak stress dwelling has little effect on the fatigue life. When the temperature elevates to 325 °C, the creep damage dominates the failure process of the joints.

Acknowledgements The project is supported by National Natural Science Foundation of China (Nos. 10802056 and 11072171) and Natural Science Fund of Tianjin, China (No. 13JCQNJC02400).

[1] Yu DJ, Chen X, Chen G, Lu GQ. Applying Anand model to low-temperature sintered nano-scale silver paste chip attachment. Mater Des 2009;30:4574–9. [2] Mei YH, Chen G, Lu GQ, Chen X. Effect of joint sizes of low-temperature sintered nano-silver on thermal residual curvature of sandwiched assembly. Int J Adhes Adhes 2012;35:88–93. [3] Li X, Chen X, Lu DQ. Reliability of high-power light emitting diode attached with different thermal interface materials. J Electron Package 2010;132:031011-1–1-5. [4] Li Y, Wong CP. Recent advances of conductive adhesives as a lead–free alternative in electronic packing: materials, processing, reliability and application. Mater Sci Eng R 2006;51:1–35. [5] Abtew M, Selvaduray G. Lead-free solders in microelectronics. Mater Sci Eng A 2000;27:95–141. [6] Kim HK, Liu HK, Tu KN. Three-dimensional morphology of a very rough interface formed in the soldering reaction between eutectic SnPb and Cu. Appl Phys Lett 1995;66(18):2337–9. [7] Wang T, Chen G, Wang YP, Chen X. Uniaxial ratcheting and fatigue behaviors of low-temperature sintered nano-scale silver paste at room and high temperature. Mater Sci Eng, A 2010;527:6714–22. [8] Lu GQ, Calata JN, Zhang ZY. A lead-free, low-temperature sintering die-attach technique for high performance and high-temperature packaging. In: Proceeding of the 6th IEEE CPMT conference on high density microsystem design and packaging and component failure analysis, China; 2004. [9] Abtew M, Selvaduray G. Lead-free solders in microelectronics. Mater Sci Eng 2000;27:95–141. [10] Bai JG, Zhang ZZ, Lu GQ. Low-temperature sintered nanoscale silver as a novel semiconductor device-metalized substrate interconnect material. IEEE Trans Comp Package Technol 2006;29(3):589–93. [11] O’brien M, Rice CR, Olson DL. High strength diffusion welding of silver coated base metals. Weld J 1976;55:25–7. [12] Zhang Z, Lu GQ. Pressure-assisted low-temperature sintering of silver paste as an alternative die-attach solution to solder reflow. IEEE Trans Comp, Package, Manuf 2002;25:279–83. [13] Knoerr M, Kraft S, Schlitz A. Reliability assessment of sintered nano-silver dieattachment for power semiconductors. In: Proceedings of the electronics packaging technology conference (EPTC), integrated syst. and device technol., Fraunhofer, Germany; 2010. [14] Lei TG, Calata JN, Lu GQ. Low-temperature sintering of nanoscale silver paste for attaching large-area (>100 mm2) chips. IEEE Trans Comp Package Technol 2010;33(1):98–104. [15] Bai JG, Lu GQ. Thermo mechanical reliability of low-temperature sintered tired silver die-attached SiC power device assembly. IEEE Trans Dev Mater Reliability 2006;6:436–41. [16] Wang T, Chen X, Lu GQ, Lei G. Low-temperature sintering with nano-silver paste in die-attached interconnection. J Electron Mater 2007;36:1333–40. [17] Chen G, Sun XH, Nie P, Chen X, Lu GQ. High-temperature creep behavior of low-temperature-sintered nano-silver paste films. J Electron Microsci 2012;41:782–90. [18] Evans JW. A guide to lead–free solders, physical metallurgy reliability. London: Springer Inc.; 2005. p. 128–43. [19] Ohguchi K, Sasaki K, Ishibashi M. A quantitative evaluation of timeindependent and time-dependent deformations of lead-free and leadcontaining solder alloys. J Electron Mater 2006;35(1):132–9. [20] Hertzberg RW. Deformation and fracture mechanics of engineering materials. 4th ed. New York: Wiley; 1996. [21] Li X, Chen G, Chen X. Mechanical property evaluation of nano-silver paste sintered joint using lap-shear test. Solder Surf Mt Tech 2012;24:120–6. [22] Li X, Chen G, Chen X. High temperature ratcheting behavior of nano-silver paste sintered lap shear joint under cyclic shear force. Microelectron Reliab 2013;53:174–81. [23] Li X, Chen G, Chen X. Creep property of low temperature sintered nano-silver lap shear joints. Mater Sci Eng 2013;579:108–13. [24] NBE Tech, LLC., . [25] Lang F, Tanaka H, Munegata O, Taguchi T, Narita T. The effect of strain rate and temperature on the tensile properties of Sn–3.5Ag solder. Mater Charact 2005;54:223–9. [26] Bannantine JA, Corner JJ, Handrock JL. Fundamentals of metal fatigue analysis. New Jersey: Prentice Hall press; 1990. [27] Lim CB, Kim KS, Seong JB. Ratcheting and fatigue behavior of a copper alloy under uniaxial cyclic loading with mean stress. Int J Fatigue 2009;31:501–7. [28] Kun F, Carmona HA, Andrade Jr JS, Herrmann HJ. Universality behind Basquin’s law of fatigue. Phys. Rev. Lett. 2008;100:094301. [29] Kanchanomai C, Miyashita Y, Mutoh Y. Low cycle fatigue crack growth behaviour of Sn–Ag eutectic solder. Solder Sur M Technol 2002;14(3):30–6. [30] Zhang QK, Zhang ZF. In situ observations on creep fatigue fracture behaivor of Sn–4Ag/Cu solder joints. Acta Mater 2011;59:6017–28.