Continued miniaturisation of devices has placed greater pressure on structural integrity, especially for surface mount configurations. Reliable design requires more sophisticated approaches, and this challenge is compounded by the emergence of a new generation of solder alloys that, for environmental reasons, do not contain lead. The present paper examines the performance of Sn-37 Pb solder and two lead-free alloys during high strain cycling with and without dwell periods. 2 Experimental Details The materials used were commercial grade eutectic 63Sn-37Pb alloy, supplied in the form of rectangular ingots; the tin-3.5 silver and tin-0.5 copper were supplied as rods or bars. The alloys were heated to match the optimum reflow profile, to about 40˚C above their melting point, and cast into cylindrical aluminium moulds, preheated to the same temperature. The moulds were then rapidly quenched in water. The resultant surface finish on the specimen obtained after casting required no further treatment. Specimens were stored at -18˚C in order to minimise natural ageing. Prior to testing they were warmed to ambient for 2 hours. Their final dimensions were; gauge diameter of 11.28mm (cross sectional area 100mm 2 ) and a gauge length of 25mm. Fatigue and fatigue plus dwell tests were carried out using an Instron 1342 servo hydraulic testing machine under total strain control. Dwell cycles were timed and controlled using a computer generated wave form. The test machine was fitted with an environmental chamber operating at 75˚C. Strain was measured using a contacting extensometer attached to the gauge length with coil springs. Tests were performed at a constant strain rate of 3.33x10 -3 s -1 . The cyclic frequency therefore varied for each test depending on the strain amplitude. A triangular waveform was used with a strain ratio, R, of -1 (where R = ε min: ε max). The strain ranges employed in continuous cycling varied between 0.5 and 3.0 percent, and during cycling with dwells the strain range was 1.0 percent. Load and total displacement were continuously recorded. A tensile load reduction of 20 percent with respect to the first cycle was employed as the criterion of failure and then tests were terminated. After testing, specimens were examined using both optical and scanning electron microscopy. Samples from the gauge length of the test pieces were sectioned longitudinally using electrical discharge machining to minimise surface damage, and subsequently polished to a 1µm surface finish. 3 Results 3.1 Cyclic Response All hysteresis loops show a progressive drop in the load range necessary to maintain the constant strain limits, indicating cyclic softening. Saturation of the load range occurs at strain ranges of 1 percent and above, whereas, below this level, the plastic strain range continues to increase and saturation is not reached. A cyclic stress-strain curve, derived from the stress range during the first cycle and the total strain range may be described by the relationship, ∆σ = A∆εβ The values for the constants A and ß are presented in Table 1. TABLE 1 VALUES OF CONSTANTS IN THE CYCLIC STRESS-STRAIN EQUATION Pb/Sn Sn/Ag Sn/Cu RT 75 RT 75 RT 75 A 129 64 88.9 56 55.5 43.08 ß 25.2 22.1 30.85 20.3 4.5 8.7 A convenient means of depicting strength changes during high strain cycling is via the load drop parameter, Ø, (which is defined as Ø = 1 – (∆P/∆Pm) where ∆P is the load range at any cycle and ∆Pm is the load range during the first cycle). The variation in the load drop as a function of the fraction of fatigue life is shown in figure 1 for a strain range of 1 percent, with a dwell of 10 seconds incorporated at maximum tensile strain, at room temperature and 75˚C. The load falls (indicated by an increase in
RkJQdWJsaXNoZXIy MjM0NDE=