This paper describes experiments on the effect of using contrast enhancement material, CEM-420, on photoresist imaging over an aluminum substrate on thermal oxide steps. Wafers with oxide step heights ranging from 0.4μm up to 1.04μm were used to explore the potential advantage in using CEM-420. Various thicknesses of CEM and photoresist were prepared to determine the combination that provides the best process parameters. A full factorial ex-periment with three levels of each of the three variables, PR, CEM, step, was performed. A minimum of three levels is required in order to reveal the curvilinear effects. Each combination received a different range of ten level exposure energies. The CD's at the top and bottom of the step were measured with a SEM. The data from experiments were analyzed by RSM, a software program from CompuServe. Regression analysis was used for model fitting. The effect of resist thickness, CEM thickness, step height, and exposure energy on CD is discussed. The required minimum print bias under different conditions for lμm line/space feature sizes on mask is discussed.