Temperature determination methods on copper material for highly accelerated electromigration tests (e.g. SWEAT)

Highly accelerated electromigration tests conducted at wafer level require meaningful measurements of the stress temperature of the test line. Estimates of the test temperature can be made with the use of the temperature coefficient of resistance, TCR, before significant electromigration damage occurs and with the use of the thermal resistance of the test line as electromigration damage progresses to line failure. Each approach depends on an assumption of linearity; for the use of TCR it is the linear dependence of resistance on temperature, and for the use of the thermal resistance it is the linear dependence of temperature on power dissipation. For testing aluminum test structures, these assumptions are adequately well obeyed to approximately 300°C. This is not the case for testing copper test lines. Stress temperatures as high as 600°C are encountered in highly accelerated electromigration tests. At such high temperatures, the assumptions of linearity are not valid, and large overestimates of the stress temperature will result. We show how to correct TCR estimates of the stress temperature by the use of recommended values for the resistivity of pure, bulk copper with temperature. We obtain a suitable value for the thermal resistance of the test structure from the slope of the temperature-versus-power-dissipation curve at temperatures near the desired stress temperature. Use of the approaches are demonstrated.