A study of yield optimization work which do not use the design of experiments technique

ABSTRACTIn both industrial and scientific work, process variables and their effect on the product is an important area of study. Two approaches to finding the set of parameters for maximizing yield are OFAT (One Factor At a Time) and the Design of Experiments(DOE). In OFAT, one factor is varied until yield is maximized, while others are kept constant. This is repeated for all factors and it is then assumed that this set obtained by individual maximization is also the set that gives the absolute maximum for the yield. In the DOE approach, a set of points in the parameter space of the problem is selected. Yield is experimentally determined for each point and an analytical model is fit to the point set to extract a response surface from available data. The maximum of the multi dimensional yield surface is then calculated. In this thesis, published data of OFAT some authors is used for calculating DOE type yield surface by using least squares. Approach in this thesis indicated yield improvements of up to 11% relative to the OFAT results reported.