Mass Transport in Confined Micro Geometry in Reciprocating Paddle Plating Cell by Numerical Simulation and Neural Network Analysis

The mass transport in confined geometry in a paddle plating cell is studied using numerical simulation with an attempt to extrapolate an explicit correlation to understand the transport physics and to predict the electrodeposition rate of metal microstructures. A moving boundary in conjunction with mapped mesh is used to allow the reciprocating movement of the flow. A correlation is obtained based on a generalized additive model using multivariant linear regression. Neural networks are also used to analyze the efficacy of such correlation and to determine the descriptor characterizing the error in prediction. A two-step convectional mass transport process, one in the bulk electrolyte outside the patterns and the other inside the micro-trenches, is demonstrated to better describe the overall transport physics and improve the correlation.