Comparative evaluation of drying techniques for surface micromachining

Five different procedures commonly used to rinse and dry released microstructures are compared: evaporation drying with deionized (DI) water or methanol, sublimation drying with t-butyl alcohol or p-dichlorobenzene, and supercritical drying with CO2. For objective comparison, identical test structures, made by the MCNC Multi-User MEMS Processes (MUMPs), are used in evaluating the drying techniques. The test chips contain arrays of surface-micromachined polysilicon cantilevers (2 mu m thick, 2 mu m gap from the substrate) with varying widths and lengths. Some beams feature dimples or tips to quantify their anti-stiction effect. This study reveals, for the first time, that the maximum beam length obtainable increases as the beam width increases for the cases of sublimation and supercritical drying, opposite to the previously known case of evaporation drying. Both sublimation drying methods as well as supercritical drying rendered good results, releasing cantilevers up to 700 mu m in length without stiction. We also introduce a new setup that considerably improves the way sublimation is used to dry microstructures. (C) 1998 Elsevier Science S.A.