Virtual Metrology Scheme for Predictive Plasma Uniformity Diagnosis of Plasma-Enhanced Atomic Layer Deposition Process using Optical Emission Spectroscopy

Plasma is widely used in etching, deposition, ashing, and ion implantation processes in the manufacturing processes of semiconductor and display devices. Recently, the role of plasma in semiconductor processes is becoming important because the pattern with high step coverage and aspect ratio is required as the semiconductor process is scaled. In particular, plasma-enhanced atomic layer deposition (PE-ALD) is recently highlighted with excellent deposition uniformity compared to chemical vapor deposition (CVD) and physical vapor deposition (PVD) technologies ¹. The density and the state of plasma during the process cannot always remain constant and can be changed by process variables and unpredictable process random fluctuations. Since this can cause yield drop and decrease in productivity, the methodology for monitoring and diagnosing the abnormality of plasma in real time during the process is essentially required. In this paper, the plasma monitoring using the optical emission spectroscopy (OES) in PE-ALD equipment to analyze the characteristics of the process variables. Based on the diagnosis results, the optimal process variable values were proposed, and a process performance predictive model is introduced by analyzing the correlation between variables. Based on the modeling results, the modeling output for the final deposited thickness can be predicted by the initial deposition rate, which can be developed as a virtual metrology system application. In addition, by applying this proposed virtual metrology scheme for the advanced process control (APC), it is possible to estimate the process output without whole wafer measuring, thereby improving the time and cost losses and the wafer-to-wafer variation in the semiconductor manufacturing industry.