PROCESS MONITORING USING KEY SENSITIVITY INDEX: APPLICATIONS TO SEMICONDUCTOR MANUFACTURING

Process monitoring is essential to maintain product quality in semiconductor manufacturing and the batch-wise processing nature makes data-based monitoring attractive. However, unlike chemical processes, the semiconductor manufacturing process exhibits the following characteristics: (1) much shorter (minutes) and often variable (deliberately adjusted) batch time, (2) multiple processing steps (10-20) in each batch, (3) only some particular processing steps (not the entire trajectory) constituting the quality determining steps, (4) mixed products for the same batch processing. In this work, instead of incorporating large number of trajectory data with variable batch time and possibly “missing” data for some process variables using multivariate statistics, a process-insight based approach, key sensitive index (KSI), is taken. From process knowledge, the key sensitive time-slot (KST) in the recipe is identified. Next, possible key sensitive process variables (KSV) are selected and validated according to the process trend correlation. Then, an index for these variables (key sensitive index, KSI) is sought. The KSI's can be classified into: variable-based, model-based, and data-based measures. Two integrated circuit processing examples from real fab data are used to illustrate the KSI-based approach and results clearly indicate that process trend is captured using KSI-based approach.