Effects of Auto-calibration Hysteresis

Structure Function based calibration for numerical models usually involve the heating up the device until it reaches a steady state temperature. Experiments are typically carried out in an opposite way, with the cooling down of the device from an initial powered on state. It is generally assumed that both methods will yield the same results. However, non-linear thermal effects such as the temperature dependent thermal conductivities of semiconductor materials can affect the structure function curves. This study quantified the differences when using heat-up and cool-down simulations, and the effects on the resulting calibrated models. These models were then subjected to a time varying driving cycle emulating a typical field application for the devices considered in the study. It was found that there were only minor differences in the thermal performance of the devices. Using either heat-up or cool-down simulations gave negligible differences in the thermal response and behaviour of the test devices.