| Title |
Machine Learning Model for Ohmic Contacts in GaN Bipolar Devices |
| Authors |
함고은(Go Eun Ham) ; 김광은(Kwangeun Kim) |
| DOI |
https://doi.org/10.5370/KIEE.2024.73.10.1687 |
| Keywords |
Machine learning; GaN; Ohmic contact; Contact resistivity; Power semiconductor |
| Abstract |
There has been increasing attention on the fabrication of power semiconductors and development of device performances using machine learning(ML) model. In this work, ML model was applied to the development of ohmic contacts in GaN for power semiconductor device applications. Factors influencing contact resistivity include work function and electrical affinity of metals, annealing conditions, and doping parameter of GaN. These parameters are set as variables to calculate the contact resistivity of ohmic contacts, which can evaluate the quality of ohmic contacts with low power assumption. Using circular transmission line measurement of Ti/Al ohmic contacts on n-GaN, I-V properties are analyzed to extract the contact resistivity according to the annealing conditions as well as used to train the ML model. By classifying the contact resistivity of n-GaN using ML model, it becomes possible to apply to the fabrication of GaN power devices with low on-resistance in power switching |