Novel challenges for root cause investigation in semiconductor manufacturing

Due to the constantly growing automation in the semiconductor manufacturing and the associated increasing amount of data, the question of data handling and automated decision making becomes more and more important. Advanced measurement techniques in high-precision and high-tech manufacturing processes lead to more data and hence, enable more monitoring possibilities. Nevertheless, recorded data is only meaningful if evaluated. Common root cause analyses, performed by experts, are mainly based on physical relations and lower order correlations aiming to keep the manufacturing process stable. To investigate higher order correlations, especially if there is an increased data landscape available, needs automated evaluation procedures because manual effort is not feasible anymore. Therefore, novel investigation procedures are needed by means of data- and knowledge-driven analysis methods to support the expert´s routine in order to perform root cause analysis.