A multivariate approach to utilizing mid-sequence process control data

While the measurement of cell efficiency is still considered one of the primary assessments of a cell´s quality, a modern photovoltaic manufacturing facility will also include a range of metrology to assess the performance at various steps during the manufacturing sequence. These measurements can be used to control individual processes and ensure reliable process interactions, but they are at their most powerful where they can be correlated to the final performance of the cell. Such a relationship is not always easy to establish, particularly when the data collected during the process cannot be parametrized entirely with a single variable. This paper shows how two multivariate approaches can be used to form a relationship between cell lifetime data collected early in a fabrication sequence, and the final cell Voc. While building a model with a high level of predictive accuracy is rarely feasible, it is possible to identify a higher proportion of under-performing product and provide insight into how material type interacts with the manufacturing sequence.