Self-Limiting Behavior of Hot Carrier Degradation and Its Implication on the Validity of Lifetime Extraction by Accelerated Stress

Abstract

The time dependence of hot carrier degradation of n-channel MOSFETs and the methodology of accelerated stress have been investigated in detail. The time (T) dependence is foLund to be inconsistent with the simple expression of TN (N 0.25 ), but rather slhow a slow-down of the degradation rate. The slope of the degradation curve is also found to be dependent on the stress bias voltage. The projection of device lifetime by accelerated stress based on the TN law and the assumption of constant slope independent of stress bias is unreliable.

The dependence of hot carrier degradation on AC stress waveforms

“…This may be an indication of some sort of self-limiting behavior in the degradation, similar to that observed in D.C. stress [5].…”

Degradation Model of a Linear-Mode LED Driver and its Application in Lifetime Prediction

“…As the degradation mechanism is found to be HCI [1], the degradation rate is fast at the beginning. Due to the shift of peak electrical field and the increase in barrier potential, the degradation rate slows down after the initial period [11,13,18]. Therefore, to capture the fast initial degradation, the measurement interval is set more frequently at beginning and then less after the degradation reaches saturation as described below:…”

“…for the driver circuit is equivalent the degradation model parameter n for the single transistor [9]. This degradation slope has been reported to vary in a wide range [9,11,25]. In this work, the slope of log( 𝛥𝑉 !…”

The hot carrier degradation rate under AC stress

“…Although this power law time dependency function is widely applied for describing device degradation under hot carrier stress, it is known from literature that it yields overly pessimistic results for high stress times, as a saturation effect can be observed [7]. Various models that take this effect into account have been proposed and typically it involves a twomechanism process [8].…”

“…The main justification of this approach is, however, purely empirical [6] and only verified if the DC device degradation can be written as a power law function of time. From literature it is known that this power law may not always be applicable [7][8][9][10], and other time dependency functions for the device degradation are more appropriate. Justification for the quasi-static approach is lacking for such time dependency functions.…”