Figure 35: Master-curve of SAC creep model.

• Tensile creep data from five independent tests show consistency over a wide temperature range, -55°C to 150°C:

• The results from three datasets and SAC alloys with 2.5 to 3.9% Ag and 0.5 to 0.8% Cu contents were used to develop a first-order hyperbolic sine creep model. The master-curve of the model is shown as the centerline of the correlation band in Figure 35 where the 37 "x" symbols represent the original data that was used for curve fitting purposes.
• The 24 other data points (shown as triangular symbols in Figure 35) represent creep results for other SAC type alloys that were rapidly cooled (15 data points) and 9 data points from compression tests ran on specimens of the Sn-4.7Ag-1.7Cu NCMS alloy. I.e., although the model was derived based on tensile creep data, it seems to fit compression creep data as well, at least to a first order.
• The 61 total data points shown in Figure 35 fall within or close to the arbitrary "lower" or "upper" bounds of the correlation band. These bounds are set a factor 10 or approximately 3.16 times below and above the centerline of the model. To a first order, the creep model applies to bulk SAC solders of slightly different compositions in the range: 95.5 to 96.5%Sn with 2.5 to 3.9% Ag and 0.5 to 0.8% Cu contents, including Castin^TM (96.2Sn-2.5Ag-0.8Cu-0.5Sb).
• Most of the data was at stress levels greater than 10 MPa. More data is needed to test the model (or any other constitutive model) below 10 MPa since solder joints of electronic assemblies are likely to experience such stress levels in use.

• Comparison of the model to creep rate data for slowly cooled specimens showed that, under those conditions, the model is offset from the data by a large factor (~ 100X). Thus, the effect of coolingrates appears to be an important parameter that needs to be investigated further.

• This strong effect of slow versus rapid cooling rates was observed in two independent experiments (Kim et al., 2001; and Joo et al., 2002). Fast cooling rates result in a more creep resistant, finer microstructure with -Sn globules of approximate size 5-10 µm.

• The comparison of the tensile-creep, SAC bulk solder model to a shear-creep, SAC flip-chip solder joint model showed that the two models follow different trends. Such a discrepancy is not new and has been observed with Sn-Pb creep data in the past. Nevertheless, this is thought to be a significant issue that needs to be resolved.
• As for Sn-3.5Ag, most of the SAC mechanical properties are provided in the range of stresses above 10 MPa. By the same token, only secondary (or steady-state) creep data are reported on since secondary creep is the overwhelmingly dominant deformation mode under high stress conditions.

• Deformation modes at lower stress levels may include significant, or at least not-negligible primary creep, as was found by Darveaux et al. (1995) and Yang, H. et al. (1996) for Sn-3.5Ag.