Simultaneous extraction of multiple buffer-trap parameters in GaN HEMTs via measurement-window sweeping: A TCAD study
Abstract The duration of the measurement window strongly affects the observed current collapse in GaN HEMTs and can obscure the individual contributions of buffer traps. Using TCAD simulations of output measurements following off-state stress, this work tracks trap emission and recapture and develops a multi-exponential time-domain deconvolution method based on measurement-window sweeping. The drain current sampled at the knee voltage is analyzed as a function of the measurement-window duration to extract characteristic time constants and recovery-current amplitudes. Two-trap tests resolve the distinct recovery steps associated with Fe and deeper trap components and quantify their individual contributions. For multi-component responses, the root-mean-square error (RMSE) and time-constant-collapse criteria are used jointly to determine the number of resolvable components and identify overfitting. In the uniform-profile multi-trap test, the fitted time constants of the dominant recovery components agree with the prescribed values within approximately 16%. A further test with a nonuniform Fe profile resolves distinct fast E1 and slow Fe recovery components, with time-constant deviations of −0.85% and −29.3%, respectively. By simultaneously extracting characteristic time constants and relative contributions to the recoverable current-collapse amplitude, the method provides a quantitative basis for assessing the impact of individual trap components and guiding epitaxial and process optimization.
Authors
- Fangzhou Du (ORCID: https://orcid.org/0000-0002-8151-8092)
- Chenkai Deng (ORCID: https://orcid.org/0000-0001-9585-353X)
- Hao Xiong (ORCID: https://orcid.org/0000-0001-7156-6587)
- Hongyu Yu (ORCID: https://orcid.org/0000-0002-5756-868X)
- Yuhao Wang
- Zhongrui Wang
- Peiran Wang
Institutions
- Shenzhen Polytechnic University (CN)
- Southern University of Science and Technology (CN)
Publication Details
- Journal
- Semiconductor Science and Technology
- Published
- 2026-11-11
- DOI
- https://doi.org/10.1088/1361-6641/aec0c4
- Primary Topic
- GaN-based semiconductor devices and materials
- Type
- article
- Field-Weighted Citation Impact
- 0.00