Identification and etch depth control of an ion milling process

The etching process is a critical manufacturing step in the fabrication of microelectronic devices. Ion milling applications in the micro-electronics industry are becoming more challenging due to decreasing feature sizes and increasing aspect ratios. Current industrial practice is to operate the ion milling process in the open-loop mode. However, due to the batch nature and fairly fast run times of the milling process (typically 5–10 minutes) it is possible that a fairly large quantity of nonconforming parts can be fabricated prior to the detection of a process fault. Closed-loop operation of the ion milling process is highly desirable. A closedloop strategy facilitates the early detection of process faults and provides adequate information allowing the final milling time to be adjusted online from batch to batch to tightly regulate final etch depth. An automatic control system for the control of ion milling etch depth is described herein. The heart of the control system is a three-dimensional surface evolution model which is used to predict etch rates and dynamic surface profiles during the milling process. It is shown that the modified surface evolution model predictions match experimental results with a high degree of accuracy.; A real-time measurement of etch rate and/or etch depth is required to correct for mismatch between the model predictions and the state of the physical system. A heterodyne laser interferometer is used as the measurement device. An interferometric measurement can allow for the in-situ measurement of etch depth and etch rate in the system. Once the process has begun, the interferometer is used to measure the difference in etch rates between the photoresist layer and the substrate layers.; An extended Kalman filter is developed to determine optimal estimates of the ion beam current density, voltage and angle of incidence, as well as material-specific etch rate model parameters, based on the measurement of the differential etch rate. Using these values of the estimated states and parameters, the process model predicts the evolution of the surface in three dimensions. (Abstract shortened by UMI.)...