Statistical analysis in MRSM approach of computing and experimental results for integrated circuit technology

With increasing integrated circuit packing density, manufacturing processes are becoming more and more complex and tolerances of process parameters more critical for production yield and product reliability, and thus the. economic viability of the integrated circuit (IC) manufacturing(1). Therefore statistical many-factorial IC technology design and optimization becomes more and more important. Modified Response Surface Methodology (MRSM) had been proposed for the approximation of computer simulation results and natural experiments. Here we present next step and results for the one important problem in statistical analysis of integrated circuit technology when we have a set of incorrect experimental data. Input database for the analysis was obtained by means of natural experimental results provided in the real conditions of integrated circuit manufacturing. Detailed inspection for adequacy criterions of statistical results was performed and illustrated duly.