Copyright 28 Year IEEE. Reprinted from Journal of ELECTRONIC MATERIALS, Vol. 37, No. 6, 28. Such permission of the IEEE does not in any way imply IEEE endorsement of any of Institute of Microelectronics products or services. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution must be obtained from the IEEE by writing to pubspermission@ieee.org.
Journal of ELECTRONIC MATERIALS, Vol. 37, No. 6, 28 DOI: 1.17/s11664-8-43-x Ó 28 TMS Regular Issue Paper Stress Strain Characteristics of Tin-Based Solder Alloys for Drop-Impact Modeling E.H. WONG, 1,6 C.S. SELVANAYAGAM, 1 S.K.W. SEAH, 1 W.D. VAN DRIEL, 2,3 J.F.J.M. CAERS, 3 X.J. ZHAO, 3 N. OWENS, 4 L.C. TAN, 4 D.R. FREAR, 4 M. LEONI, 4 Y.-S. LAI, 5 and C.-L. YEH 5 1. Institute of Microelectronics, 11 Science Park Rd, Singapore 117685, Singapore. 2. NXP Semiconductors, Gerstweg 2, 6534 AE Nijmegen, The Netherlands. 3. Philips Applied Technologies, 56 MD Eindhoven, The Netherlands. 4. Freescale Semiconductor, Inc, Tempe, AZ, USA. 5. Central Labs, ASE, Inc, Kaohsiung, Taiwan. 6. e-mail: eehua@ime.a-star.edu.sg The stress strain properties of eutectic Sn-Pb and lead-free solders at strain rates between.1 s -1 and 3 s -1 are required to support finite-element modeling of the solder joints during board-level mechanical shock and productlevel drop-impact testing. However, there is very limited data in this range because this is beyond the limit of conventional mechanical testing and below the limit of the split Hopkinson pressure bar test method. In this paper, a specialized drop-weight test was developed and, together with a conventional mechanical tester, the true stress strain properties of four solder alloys (63Sn- 37Pb, Sn-1.Ag-.1Cu, Sn-3.5Ag, and Sn-3.Ag-.5Cu) were generated for strain rates in the range from.5 s -1 to 3 s -1. The sensitivity of the solders was found to be independent of strain level but to increase with increased strain rate. The Sn-3.5Ag and the Sn-3.Ag-.5Cu solders exhibited not only higher flow stress at relatively low strain rate but, compared to Sn-37Pb, both also exhibited higher rate sensitivity that contributes to the weakness of these two lead-free solder joints when subjected to drop impact loading. Key words: Material characterization, constitutive properties, electronic packaging, drop impact, mechanical shock, finite-element modeling INTRODUCTION Portable electronic products such as mobile phones are susceptible to accidental drops and impacts during use. The mechanical stress that accompanies these impacts can result in severe damage to the product. Cracking of the liquid-crystal display (LCD) panel and separation of the plastic housing leading to battery displacement are common modes of damage. The integrated circuit (IC) components that are attached to the printed circuit board (PCB) through compliant copper leads are rarely damaged by these mechanical shocks. However, the use of surface-mount packaging eliminated (Received November 4, 27; accepted January 25, 28; published online March 6, 28) these compliant leads and solder joints became the mechanical and electrical connection between the devices and the PCB. The loss of compliance and the increasing drive towards reducing the crosssection of the solder joints has significantly increased the susceptibility of the solder joints to drop impact failure. Additionally, the move to Pb-free solders to comply with legislative requirements has resulted in increased susceptibility of solder joint failure under drop conditions. 1 3 The solder joint within a mobile product that is experiencing a drop impact does not, in general, experience direct stress from the impact. Rather, the mechanical shock induces flexing in the PCB; and the solder joints deform plastically to accommodate the imposed strain. 4 This results in a significantly lower strain rate in the solder joint than 829
83 Wong, Selvanayagam, Seah, van Driel, Caers, Zhao, Owens, Tan, Frear, Leoni, Lai, and Yeh stress strain data for tin-based solder alloys at high strain rates obtained using the SHPB technique, including 63.%Sn-37.%Pb, 96.5%Sn-3.5%Ag, 95.5%Sn-3.8%Ag-.7%Cu, 96.5%Sn-3.%Ag-.5%Cu, and 96.2%Sn-2.5%Ag-.8%Cu-.5%Sb, at strain rates of 5 s -1 to 3 s -1 and a temperature range from -4 C to6 C. This paper presents the stress strain characteristics of four solder alloys for the range from.5 s -1 to 3 s -1. A brief review of the experimental techniques for high strain rate characterization is presented in the following section as a prelude to the other sections. Fig. 1. Typical plastic strain distribution in a 3 lm solder joint. what would occur under direct impact. The typical plastic strain distribution in a.5-mm-thick solder joint when the PCB is subjected to 3 1-3 strain is illustrated in Fig. 1. The plastic strain distribution in the solder joint can be converted to a strain rate distribution by multiplying the plastic strain by 2pf, where f is the frequency of flexing of the PCB, which ranges from 2 Hz to 5 Hz for a test board used in JEDEC JESD22-B111 or the mother board in a portable mobile phone. 5 The bulk of the solder joints experience a strain rate of less than 1 s -1, but the corner of the solder joints, where failure typically initiates, have a maximum plastic strain of.72, which corresponds to a strain rate above 2 s -1. The strain rates of interest for solders used in mobile applications range from.1 s -1 to 3 s -1. Computation modeling using finite-element analysis can be used to design and analyze portable electronic products and their subassemblies under drop-impact conditions. Rate-dependent stress strain constitutive properties of solders are required to support this modeling. It is especially important to describe the magnitude of the stress and plastic strain in the solder joints to be able to predict failure. While there is abundant data available on the constitutive properties of eutectic 63%Sn-37%Pb solder at many creep strain rates, 6 9 there are far fewer reports on the constitutive properties of solder at higher strain rates. Early work by Clyens and Campbell 1 reported the stress strain characteristic of 62%Sn-38%Pb solder over the strain range from 2 s -1 to 4 s -1 in shear. More recently, Wang and Yi 11 reported compressive constitutive properties of 63%Sn-37%Pb solder over the range from 34 s -1 115 to s -1 using the split Hopkinson pressure bar (SHPB) technique. However, the reported stress strain curves exhibited dramatic strain softening after yield, casting doubt on the quality of the data. Siviour et al. 12 published compressive BASIC DEFINITIONS AND REVIEW OF EXPERIMENTAL TECHNIQUES Strain rate is defined as the rate of change of strain _e ¼ de dt ¼ dl Ldt ¼ V L ; (1) where L is the instantaneous length of the specimen and V is the instantaneous relative velocity of the two faces of the specimen separated by a distance L. If V is the initial velocity and L the initial length of the specimen then _e ¼ V =L is the engineering strain rate. The two strain rates are related by Eq. 2 _e ¼ V L ¼ V ð1 þ e v Þ L ð1 þ eþ ¼ _eð1 þ e vþ expðeþ ; (2) where e is engineering strain, e the true strain, and e v = DV/V. A stress wave is a form of acoustic wave that travels at finite velocity in a solid. Any applied stress will induce disequilibrium, leading to particles moving and adjusting themselves to the disequilibrium stress. This particle adjustment is propagated at a certain speed. At low strain rate (relative to the propagation velocity and dimension of the object), the stress wave will make multiple reflections and equilibrate in the solid; then a quasistatic condition may be assumed. At a high strain rate, there is insufficient time for stress equilibrium to be achieved within the medium so the stress wave phenomenon occurs. The simplest stress wave equation is a one-dimensional (1D) longitudinal wave described by the wave equation @ 2 u @t 2 ¼ @ 2 u c2 L @x 2 ; (3) where u is the displacement of particle on a longitudinal p medium along the longitudinal direction, c L ¼ ffiffiffiffiffiffiffiffiffiffiffiffi ðe=qþ is the propagation velocity of the stress wave, and E and q are the elastic modulus and density of the medium, respectively. Reference 1 categorizes strain rate into four regimes based on the appropriate experimental techniques; this data is summarized in Table I. Details of the various experimental techniques used
Stress Strain Characteristics of Tin-Based Solder Alloys for Drop-Impact Modeling 831 Table I. Strain Rate Regimes and Experimental Characterization Techniques Strain Rate Regime Experimental Technique Low rate _e < :1 s 1 Standard test procedures using conventional mechanical tester Medium rate :1 s 1 _e 2 s 1 Mechanical tester with ultra capacity Cam plastometer Drop weight High rate 2 s 1 _e 1 5 s 1 Hopkinson pressure bar Taylor rod impact Very high rate _e 1 5 s 1 Flyer plate impact to derive this data can be found elsewhere, 13,14 but the following is a summary of the methods used. Conventional Mechanical Tester Mechanical testers are typically screw driven or servohydraulic driven and offer flexibility in force and displacement control. However, at high speeds, the accuracy of load and displacement measurements are affected by the dynamics of the machine structure and the measuring instruments. This limits usage to low-strain-rate characterization. Split Hopkinson Pressure Bar (SHPB) A schematic of the split Hopkinson pressure bar (SHPB) technique is depicted in Fig. 2. A striker bar strikes the incident bar squarely at a given velocity V and sets up stress wave propagation in the incident bar, the specimen, and the output bars. The stress strain response of the specimen can be evaluated from measurements of the incident, reflected, and transmitted pulse through strain gauge measurements on the incident and output bars. The general solution to Eq. 3 is of the form: u ¼ f ðx c L tþþgðx þ c L tþ; (4) where f(x - c L t) and g(x + c L t) describe waves that propagate in the positive and negative x-direction, respectively. Differentiating Eq. 4 with respect to x and t gives the strain and velocity, respectively. The velocity and strain are related by the constant c L with appropriate sign to indicate the direction of propagation. The strain, stress, and velocity at interface À and interface ` of Fig. 2 are given in Eqs. 5 and 6, respectively. e 1 ¼ e I ðtþþe R ðtþ r 1 ¼ E I ½e I ðtþþe R ðtþš V 1 ¼ c L ½ e I ðtþþe R ðtþš e 2 ¼ e T ðtþ r 2 ¼ E T e T ðtþ V 2 ¼ c L e T ðtþ where E I and E T are the elastic modulus of the incident bar and transmission bar, respectively. In most cases, E I = E T = E. Note that the specimen does not experience uniform stress/strain across its length. The average stress, strain rate, and strain in the specimen are given by r ave ðtþ ¼ r 1ðtÞþr 2 ðtþ A 2 A _e ave ðtþ ¼ V 2ðtÞ V 1 ðtþ L (5) (6) (7) (8) (a) V o Strain gauge e ave ðtþ ¼ c L L Z t ½e I ðtþ e R ðtþ e T ðtþšdt; (9) Striker bar (b) ε I ε R u 1 L ε T Incident bar u 2 S tra i n Specimen Reflected pulse ε R (t) Incident pulse ε I (t) Output bar (c) Time Transmitted pulse ε T (t) Fig. 2. (a) Schematic of SHPB, (b) enlarged view of specimen, and (c) strain wave across interfaces. where A and A are the original and instantaneous cross-sectional area of the specimen, respectively. The absence of the machine structure and measuring instruments allows measurement at high strain rate. However, the assumption of a 1D wave, ignoring the effects of radial inertia due to the Poisson ratio, limits the upper strain rate of the SHPB technique. The lower limit of strain is the regime of interest that the SHPB can manage. This is limited by the available kinetic energy from the striker bar to deform the specimen to the desired plastic strain. For example, it would require a striker bar of identical diameter as the test specimen and more than 1 m in length to introduce 1%
832 Wong, Selvanayagam, Seah, van Driel, Caers, Zhao, Owens, Tan, Frear, Leoni, Lai, and Yeh H M Before impact Drop mass Specimen Load cell plastic strain in a material that has a flow stress of 1 MPa at a strain rate of 2 s -1. Drop-Weight Technique The stress strain characteristics of solders in drop-impact applications can be characterized using an ultracapacity mechanical tester, a cam plastometer, or a drop-weight tester (Table I). The first two require expensive specialized tools. However, the drop-weight tester is low cost and provides highquality test data. The drop-weight technique uses a falling weight to provide a compressive load to the specimen. A typical set up of drop-weight testing is illustrated in Fig. 3. The load is measured with a load cell, and the displacement is obtained either through direct measurement using external instrument or through integration of the acceleration of the drop mass. Assuming stress equilibrium in the specimen, the force that retards the drop weight is the same as that measured by the load cell. The acceleration, velocity, displacement, true strain, and true strain rate of the drop weight can then be evaluated: aðtþ ¼ FðtÞ Z t Z t M ; VðtÞ ¼ aðsþds; uðtþ ¼ VðsÞds eðtþ ¼ln L uðtþ ; _eðtþ ¼ VðtÞ L L uðtþ : ð1þ The strain rate and maximum plastic strain can be adjusted by varying the height and mass of the drop weight. The minimum strain rate is limited by the available kinetic energy of the drop mass but this restriction is much less severe than for SHPB as the drop mass need not be a longitudinal bar. The maximum strain rate is limited by the lack of stress equilibrium in the test specimen and the dynamic noise from the load cell. The former can be minimized by reducing the length of the specimen. The mechanics of the dynamic noise of the load cell is explained below: a typical load cell has a natural frequency between 5 khz and 5 khz. The load cell behaves as a spring-mass when subjected to an impact pulse. Figure 4 illustrates a load cell of natural frequency x subjected to a half-sine force pulse of F = F sin Xt. The ratio of the maximum dynamic force, F d, to the maximum static force, F, M F(t) After impact Fig. 3. Schematic of drop-weight test technique. a(t), V(t), u(t) Fd / F o 2 1.8 1.6 1.4 1.2 1.8.6.4.2 F ω Fo rc e F=F o sinω t Time 2 4 6 8 1 Frequency Ratio, ω/ω Fig. 4. Illustrations of dynamic noise from the load cell. is shown over a range of frequency ratio, x/x. Depending on the frequency ratio, the load cell may register a different magnitude of load from the actual magnitude of load applied to the load cell. Figure 4 shows that an ideal load cell should have a natural frequency that is more than five times that of the highest-frequency component of the impact pulse. The noise from load cell resonance may be minimized using a low-pass filter. CHARACTERIZATION EXPERIMENT The strain rate range of interest for a solder joint subjected to typical drop-impact conditions is between.1 s -1 and 3 s -1 (the medium strain rate in Table I). Four tin-based solder alloys, 63.%Sn-37.%Pb (Sn-37Pb), 98.5%Sn-1.% Ag-.1%Cu (SAC11), 96.5%Sn-3.5%Ag (Sn-3.5Ag), and 95.5%Sn-3.%Ag-.5%Cu (SAC35) were selected based on the following considerations. The SAC35 solder (and to a lesser extent the Sn-3.5Ag solder) have been widely adopted by the industry as lead-free replacements for the Sn-37Pb solder due to the superior creep-fatigue resistance. However, the two solder alloys were subsequently found to be susceptible to brittle fracture when subjected to drop-impact loading. 1 3 There is also increased interest in the industry to explore solder alloys with lower silver content. 15 In this work, the strain rates between.5 s -1 and 12 s -1 were characterized using a standard Instron Micro-Force tester, using cylindrical specimens of diameter of 4 mm and length 4 mm. The strain rate at 12 s -1 used specimens of reduced length and diameter, both 2 mm, to achieve the desired strain rate. The range of strain rate between 5 s -1 and 3 s -1 was characterized using a dropweight tester that was developed for this purpose. All specimens were machined from extruded bars for SAC35, Sn-3.5Ag, and Sn-37Pb, and cast bars for SAC11.
Stress Strain Characteristics of Tin-Based Solder Alloys for Drop-Impact Modeling 833 D i s p l a c e m e n t ( m m ) 1.8 1.6 1.4 Las e r m e as urem ent 1.2 1.8 Integration, eq(1 ).6.4.2 SAC1 1 at st ra in rate 1s -1 1 2 3 4 Time (ms) Fig. 5. Comparison of the direct integration method and laser measurement. The drop-weight tester consists of two parallel frames along which a weight of desired mass is elevated and lowered through a motorized drive. At the desired height, the weight is released and falls squarely onto the specimen of diameter 4 mm and length 4 mm, compressing it between the weight and a load cell. The load cell is attached rigidly to an anvil of substantial mass to prevent rigid-body movement of the load cell. The drop height was varied to achieve the desired strain rate while the mass of the drop weight was varied according to the strain rate to attain a minimum plastic strain of.25. The magnitude of the plastic strain shown in Fig. 1 appears to be low but this increases significantly in repeated cycling. The load was measured with a model 981A load cell from Kistler with a maximum load of 65 N and a resonant frequency of 5 khz. The specimen was placed asymmetrically on the load cell. The displacement was measured with a laser (model LK-G87 from Keyence) that has a resolution of.2 lm and a dynamic data rate of 2 ls. The accuracy of the measurement obtained from direct integration (Eq. 1) compared well with the laser measurement (Fig. 5). The effect of friction is minimized through the application of graphite lubricant across the two contact surfaces of the specimen. RESULTS AND DISCUSSIONS Visual inspection of the deformed specimens of 63.%Sn-37.%Pb (Sn-37Pb), 98.5%Sn-1.%Ag-.1%Cu (SAC11), 96.5%Sn-3.5%Ag (Sn-3.5Ag), and 95.5%Sn-3.%Ag-.5%Cu (SAC35) showed that the different materials have distinct characteristics. The surface of the deformed Sn-37Pb specimens showed wrinkling (Fig. 6a) possibly due to interaction of slip bands, while that of SAC11 exhibited a crumpled appearance (Fig. 6b). On the other hand, cracks have been observed on some of the Sn-3.5Ag (Fig. 6c, d) and the SAC35 (Fig. 6e, f) specimens. The Sn-3.5Ag specimens typically exhibit shallow circumferential crack, which may be accomplished Fig. 6. Distinct characteristics of deformed specimens: (a) Sn-37Pb, (b) SAC11, (c) and (d) Sn-3.5Ag, and (e) and (f) SAC35. by a few parallel cracks (Fig. 6d) on its top or bottom surface, and the cracks do not propagate to its cylindrical surface. The SAC35 specimen tends to exhibit a vertical crack along its cylindrical surface, which propagates inward in the circumferential direction (Fig. 6f). The presence of cracks in the deformed specimens suggests that the Sn-3.5Ag and SAC35 solders are less ductile than the Sn-37Pb and the SAC11 solders. The true stress strain characteristics of Sn-37Pb, SAC11, Sn-3.5Ag, and SAC35 are presented in Fig. 7 as a function of strain rate (V /L ). The highfrequency noise from the load cell was removed using a low-pass filter set at 1 khz. Strain softening was observed for Sn-3.5Ag and SAC35 solders. The softening could be attributed to the formation of microcracks within these two solders. The peak flow stress for the Sn-37Pb, SAC11, Sn-3.5Ag, and SAC35 solders at 1 s -1 are 61 MPa, 64 MPa, 81 MPa, and 87 MPa, respectively. Plotting the flow stress at 1% plastic strain across the range of strain rate in the ln-ln scale (Fig. 8), the strain rate sensitivity of the four
834 Wong, Selvanayagam, Seah, van Driel, Caers, Zhao, Owens, Tan, Frear, Leoni, Lai, and Yeh (a) (c) 12 1 8 6 4 2..1.2.3.4.5 18 16 14 12 1 8 6 4 2..1.2.3.4.5 (b) 16 14 (d) 12 1 8 6 4 2..1.2.3.4.5 True strain 18 16 14 12 1 8 6 4 2..1.2.3.4.5.5 s -1 6. s -1 1 s -1.5 s -1 12 s -1 2 s -1 1. s -1 5 s -1 3 s -1 Fig. 7. Stress strain characteristics of (a) Sn-37Pb, (b) SAC11, (c) Sn-3.5Ag, and (d) SAC35 expressed in curves of constant engineering strain rate. ln (flow stress) 5.5 5 4.5 4 3.5 Sn37Pb SAC11.11.1 Sn3.5Ag SAC35.12.7 3-6 -4-2 2 4 6 8 ln (strain rate) Fig. 8. Strain rate sensitivities of solders at 1% strain. solders, as expressed as a gradient of the straight line, are.7,.12,.1, and.11 for Sn-37Pb, SAC11, Sn-3.5Ag, and SAC35, respectively. This is similar to the gradient of.8 obtained by Clyens 1 for Sn-38Pb and between the values of.7 and.9 obtained for a number of lead-free solders by Shohji. 6 If the strain rate sensitivity of the solders is independent of strain, it would be possible to project the strain rate sensitivity from a single stress strain characteristic obtained at any strain rate. The strain-rate sensitivity of the solders characterized at 5%, 1%, 15%, and 2% strain are tabulated in Table II. The data is more consistent for the higher strains (1%, 15%, and 2%) than the 5% strain. This is due to the resonance of the load cell, which is at a maximum at lower strain. For the higher strains, the strain rate sensitivity of the solders is independent of strain. The strain rate sensitivity is provided in three strain rate regimes: the full range that covers.5 s -1 to 3 s -1, the low-strain-rate regime that covers.5 s -1 to 12 s -1, and the medium-strain-rate regime that covers 5 s -1 to 3 s -1. The strain rate sensitivity of the Sn-37Pb, SAC11, Sn-3.5Ag, and SAC35 solders at 1% strain are (.8,.17,.7), (.11,.19,.12), (.9,.16,.1), and (.9,.13,.11), respectively, where the first, second, and the third number in the bracket indicate the low- and medium-strain-rate, and full regime, respectively. The lead-free solders exhibit higher rate sensitivity than the Sn-37Pb solder and there are distinct differences in the rate sensitivity between the low- and medium-strainrate regimes, with the latter almost double that of the former. However, it is unclear if the difference could be attributed to the two different test techniques used for the two regimes. The gradient of.17 for Sn-37Pb at medium strain rate is marginally smaller than the value of.19 obtained by Siviour 12 at higher strain rate. The stress strain data in Fig. 7 are plots of constant engineering strain rate (V /L = constant), but not constant true strain rate (V/L = constant). For the low-strain-rate region using conventional uniaxial testing, the length of the specimen decreases with increased strain, leading to increasing strain rate. For the medium-strain-rate region using the drop-weight test, both the velocity and the length of the specimen decrease, but at different rates. With increased strain, the true strain rate may increase or decrease. The stress strain characteristics of the
Stress Strain Characteristics of Tin-Based Solder Alloys for Drop-Impact Modeling 835 Table II. Stain Rate Sensitivities of Solders at Strains Between 5% and 2% Strain (%) Strain Rate Regime (s -1 ) Strain Rate Sensitivity Gradient (N mm -2 s) Sn-37Pb SAC11 Sn-3.5Ag SAC35 5.5 12.7.11.9.9 5 3.16.13.13.13.5 3.6.9.9.1 1.5 12.8.11.9.9 5 3.17.19.16.13.5 3.7.12.1.11 15.5 12.8.12.1.9 5 3.15.19.17.14.5 3.8.12.11.11 2.5 12.9.12.1.8 5 3.16.18.18.14.5 3.8.12.11.11 (a) 12 (c) 1 8 6 4 2..1.2.3.4.5 18 16 14 12 1 8 6 4 2..1.2.3.4.5 (b) (d) 16 14 12 1 8 6 4 2..1.2.3.4.5 True strain 18 16 14 12 1 8 6 4 2..1.2.3.4.5.5 s -1 6. s -1 1 s -1.5 s -1 12 s -1 2 s -1 1. s -1 5 s -1 3 s -1 Fig. 9. Stress strain characteristics of (a) Sn-37Pb, (b) SAC11, (c) Sn-3.5Ag, and (d) SAC35 expressed in curves of constant true strain rate. solders at the true strain rate have been constructed from the data at the engineering strain rate through interpolation and are shown in Fig. 9. Compared with Fig. 7, there is a difference between the engineering strain rate and the true strain rate. The true strain rate sensitivity is tabulated in Table III. The strain rate sensitivity of the Sn-37Pb, SAC11, Sn-3.5Ag, and SAC35 solders for the lowand medium-strain-rate regimes at 1% strain are (.7,.2,.7), (.11,.22,.12), (.9,.2,.1), and (.8,.19,.1), respectively. Comparing to Table II, there is little difference in solder behavior at the low strain rate and full range, but there is a significant increase in the sensitivity gradients for the solders at the medium strain rate. This increase is attributed partly to the interpolation procedures that effectively increase the flow stress range for the medium-strain-rate regime. SAC35 and Sn-3.5Ag were initially identified as lead-free replacements due to their superior creepfatigue resistance, but susceptibility to fracture in drop impact 3 has raised issues on these alloys. Based on the data in this paper, the Sn-3.5Ag and the SAC35 solders exhibit higher flow stress at relatively low strain rates compared to Sn-37Pb and both exhibit higher rate sensitivity. The solder
836 Wong, Selvanayagam, Seah, van Driel, Caers, Zhao, Owens, Tan, Frear, Leoni, Lai, and Yeh Table III. True Stain Rate Sensitivities of Solders at Strains Between 5% and 2% Strain (%) Strain Rate Regime (s -1 ) Strain Rate Sensitivity Gradient (N mm -2 s) Sn-37Pb SAC11 Sn-3.5Ag SAC35 5.5 12.6.1.8.7 5 3.2.14.18.19.5 3.6.9.9.1 1.5 12.7.11.9.8 5 3.2.22.2.19.5 3.7.12.1.1 15.5 12.8.11.9.8 5 3.2.23.21.2.5 3.8.12.11.1 2.5 12.8.11.1.7 5 3.2.22.21.21.5 3.8.12.11.1 forms a metallurgical joint with the copper pad on the printed circuit substrate in the form of an intermetallic compound, typically Cu 6 Sn 5. An increase in the flow stress (strength) of the ductile solder induces stress in the brittle intermetallic compound and results in fracture because the solder does not deform. Therefore, the weakness of these two lead-free solder joints when subjected to dropimpact loading is not surprising. The SAC11 solder has a comparable flow stress to Sn-37Pb and is a good candidate for lead-free replacement in conditions where drop-impact reliability is important. CONCLUSIONS The true stress strain properties of four solder alloys (63Sn-37Pb, Sn-1.Ag-.1Cu, Sn-3.5Ag, and Sn-3.Ag-.5Cu) have been generated for strain rates of.5 s -1 to 3 s -1 using a standard loadframe and the drop-weight tester. Several conclusions have been drawn from these results. Firstly, the strain rate sensitivity of the solders was found to be independent of strain. Hence, armed with a single stress strain characteristic obtained at a specific strain rate, it is possible to extrapolate to obtain stress strain characteristics at any other strain rate. Secondly, the solders displayed distinct rate sensitivity between the low (.5 s -1 to 12 s -1 ) and medium (5 s -1 to 3 s -1 ) strain rate regimes with the latter having a rate sensitivity of almost double the former. It was also found that the stress strain characteristics under constant true strain rate rather than constant engineering strain rate have a significantly higher rate sensitivity for the medium-strain-rate regime. Finally, the Sn-3.5Ag and the SAC35 solders exhibited not only higher flow stress at relatively low strain rate but higher rate sensitivity compared to Sn-37Pb. This is the reason for the higher susceptibility of these two lead-free solder joints when subjected to dropimpact loading. REFERENCES 1. M. Date, T. Shoji, M. Fujiyoshi, K. Sato, and K.N. Tu, Proceedings of the 54th IEEE Electronic Components & Technology Conference (Las Vegas, NV, 24), pp. 668 674. 2. K. Newman, Proceedings of the 55th IEEE Electronic Components & Technology Conference (Orlando, FL, 25), pp. 1194 121. 3. E.H. Wong, R. Rajoo, Y.-W. Mai, S.K.W. Seah, K.T. Tsai, and L.M. Yap, Proceedings of the 55th IEEE Electronic Components & Technology Conference (Oralndo, FL, 25), pp. 122 129. 4. E.H. Wong, K.M. Lim, N. Lee, S. Seah, C. Koh, and J. Wang, Proceedings of the 4th Electronic Packaging & Technology Conference (Singapore, 22), pp. 327 333. 5. E.H. Wong, Y.-W. Mai, S.K.W. Seah, R. Rajoo, T.B. Lim, C.T. Lim, and J. Field, Proceedings of the High Density Pack., Shanghai (25), pp. 27 217. 6. I. Shohji, T. Yohshida, T. Takahashi, and S. Hioki, J. Mater. Sci.: Mater. Electron. 15, 219 (24). 7. K. Kawashima, T. Ito, and M. Sakuragi, J. Mater. Sci. 27, 6387 (1992). 8. H.D. Solomon, J. Electron. Mater. 19, 929 (199). 9. P.J. Wray, Metall. Trans. 4, 2475 (1973). 1. S. Clyens and J.D. Campbell, Inst. Phys. Conf. Ser. 21, 62 (1974). 11. B. Wang and S. Yi, J. Mater. Sci. Lett. 21, 697 (22). 12. C.R. Siviour, S.M. Walley, W.G. Proud, and J.E. Field, J. Phys. D: Appl. Phys. 38, 4131 (25). 13. P.S. Fallansbee, Metals Handbook, vol. 8 (ASM, 1985). 14. J.E. Field, S.M. Walley, W.G. Proud, H.T. Goldrein, and C.R. Siviour, Int. J. Impact Eng. 3:725 115 (24). 15. N. Tanaka, T. Sasaki, T. Kobayash, and K. Tatsumi, Proceedings of the 56th IEEE Electronic Components & Technology Conference (San Diego, CA, 26), pp. 78 84.