APC is very important to improve throughput, yield and performance of products in modern semiconductor manufacturing. It is difficult for universal APC to reduce error automatically and effectively. A novel anti-systematic-error APC: Dual Proximate Inverse System Feedback Control (DIFC) is established in this paper。It can decrease systematic error by amending property variation of equipments. In DIFC, two loops are used to inverse both target output and real output. Control law will analyze the difference between them to amend input.Sufficient condition that makes DIFC asymptotically stable and convergent to target output is given and proved. DIFC is simulated using several virtual semiconductor tools, e.g. RTP, ion implanter and CMP tool. Different control laws (e.g., Proportional, Proportion-Integral-Differential -PID, Dynamic Proportional) and Inverse System (e.g., Radial Basic Function–RBF network, Linear Approximation Model) are used in the simulations. It shows that DIFC can deal with systematical error both continuous (during manufacturing) and step (after maintenance), and DIFC do reduce these errors by several orders of magnitude rapidly. |