Novel Optimization Method using Machine-learning for Device and Process Competitiveness of BCD Process

The novel optimization method for BCD(Bipolar-CMOS-DMOS) process development based on Machine Learning(ML) and statistical process modeling considering the entire wafer variation is proposed to improve the device and process competitiveness. The self-align PBODY process is used for high-performance N-type Lateral Diffused Metal Oxide Semiconductor(NLDMOS) in BCD process and it also is related to stability in PMIC operation. The process modeling embracing the performance and the stability of LDMOS is performed with TCAD using inline data. For the development of BCD process, the PBODY process parameters are optimized through the ML algorithms and the condition is verified with TCAD and silicon test. Finally, we can secure new low voltage NLDMOS with the improved performance and stability respectively for without any degradation in the new 0.13μm BCD process.