Most commonly BTI is studied in terms of an equivalent threshold voltage shift which occurs when a device is subjected to stress. As the device performance changes particularly slowly at use conditions, the experiments are usually performed at significantly larger biases and temperatures which are called accelerated stress conditions. With the results obtained from voltage and temperature accelerated tests, analytical or empirical models can be derived, calibrated and then used to predict the impact of BTI on the device performance at nominal operating conditions. Therefore, sophisticated models are required which have to provide an accurate description for temperature and field dependent effects.
To illustrate the temperature dependence of the threshold voltage shift after the transistor has been subjected to NBTI stress, several recovery traces on a large-area pMOSFET are measured after the transistor has been stressed. The measured recovery of the threshold voltage is plotted after normalization to in Figure 6.1.
Most notably, a similar recovery behavior is obtained for recovery traces measured at different temperatures. As demonstrated in the following, this observation is important because by studying large-area devices the recovery seems to have only a weak temperature dependence. In contrast, a strong temperature dependence is obtained when the average emission time of single defects is studied in detail on nanoscale devices, see Figure 6.2.
Using such scaled transistors, the temperature activation of the charge trapping kinetics is directly visible, emphasizing the need for experiments on small devices.« PreviousUpNext »Contents