Statistical insights from inline solar cell metrology data in a PERC production environment

Abstract

The adaptation of solar cell physics models and advanced laboratorybased measurement techniques to enable their use in high-volume, inline solar cell production settings is an exciting development towards implementing Industry 4.0 compliant smart solar cell factories. This paper outlines how a blend of physics-based analysis and statistical data science methods can aid continuous improvement and yield optimization in high-volume solar cell fabrication. A specific example is provided for a passivated emitter, rear locally contacted (PERC) solar cell production environment, where four batches of 500 commercial solar cells are evaluated using I–V at one-Sun as well as both contacted and contactless spectral response techniques. The spectral response techniques revealed prominent periodic patterns in the cell measurement sequence, which could be traced to the anti-reflection coating deposition process. This process inhomogeneity led to bimodal distributions in each batch with an efficiency difference as large as 0.07% between the modes. Thus, its identification by the spectral response technique is an important first step towards improving the efficiency distribution via deposition uniformity improvement. A yieldoriented cell physics model is used to interpret the various data in the context of underlying cell parameters, forming the basis for previously impractical root cause analysis in complex adverse events, and for process optimization in order to obtain sustained yield improvement in high-volume production.