On the Challenges of Reliable Threshold Voltage Measurement in Ohmic and Schottky Gate p-GaN HEMTs

For large scale testability of p-GaN HEMTs it is essential to investigate threshold voltage ( $\text{V}_{\mathrm{ th}}$ ) instability from the perspective of the measurement induced instability. In this paper the impact of accumulated gate bias stress during standard transfer characteristic measurements ( $\text{I}_{\mathrm{ D}}$ - $\text{V}_{\mathrm{ G}}$ ) in a p-GaN AlGaN/GaN-on-Si normally off HEMT is quantitatively analysed and modeled. This illustrates the associated threshold voltage ( $\text{V}_{\mathrm{ th}}$ ) instabilities and leads to a better understanding of the $\text{V}_{\mathrm{ th}}$ measurement challenges of a p-GaN HEMT. Upon an application of a constant gate bias close to nominal $\text{V}_{\mathrm{ th}}$ the drain current $\text{I}_{\mathrm{ D}}$ shows an initial marginal rise, followed by a short stable period ( $\text{I}_{\mathrm{ Dstable}}$ - $\text{T}_{\mathrm{ zone}}$ ) and a steep decay period. The effective bias history built on the gate stack varies when pulse on-time (Ton) or step time (Tstep), corresponding to pulsed or DC step $\text{I}_{\mathrm{ D}}$ - $\text{V}_{\mathrm{ G}}$ measurements, are varied. We find that this can lead to a $\text{V}_{\mathrm{ th}}$ variation of up to 20%. It is also observed that the choice of Ton and Tstep determines whether $\text{I}_{\mathrm{ D}}$ is measured in $\text{I}_{\mathrm{ Dstable}}$ - $\text{T}_{\mathrm{ zone}}$ or the rise or decay periods. Measurement induced $\text{V}_{\mathrm{ th}}$ instability is attributed to trapping of 2DEG electrons at the AlGaN barrier and we demonstrate that an ohmic gate contact, in comparison to the Schottky gate contact, may compensate for the trapped 2DEG electrons by hole injection across the AlGaN barrier. Experimental results of $\text{V}_{\mathrm{ th}}$ stability under the same stress conditions for a Schottky and ohmic gate contact are supported by a detailed TCAD analysis