RF 스퍼터링 공정의 실시간 안정화를 위한 물리 기반 인공지능 알고리즘 설계 = Design of Physics-Informed AI Algorithm for Disturbance-Robust RF Sputtering

다국어 초록 (Multilingual Abstract)

RF sputtering enables high-density thin-film deposition but suffers from chamber-pressure fluctuations that cause center–edge thickness non-uniformity, target-thickness shortfall, and occasional tool shutdowns. This study proposes a real-time control scheme that couples an analytical deposition model with AI optimization to suppress such instabilities. The model includes (i) sputter yield in the low-energy regime; (ii) gas-phase transport attenuation via mean free path as a function of pressure and flow; (iii) a mass-flow correction linked to plasma impedance; (iv) rotation/geometry effects mapping incidence-angle distributions to radial thickness; and (v) multi-zone ESC temperature control with an Arrhenius-type sticking correction. The proposed algorithm maintained a stable deposition rate even when chamber pressure was randomized, and its effectiveness was verified by comparing pre-/post-control process data in a web-based simulator.