Regression analysis of temperature-dependent alumina atomic layer deposition growth per cycle using trimethylaluminum and water as precursors

Temperature-dependent alumina atomic layer deposition (ALD) growth per cycle (GPC) data were collected from studies of the trimethylaluminum (TMA)/water ALD process. The data were used in three regression approaches based on fitting the data to globally defined polynomials in temperature T, and two linear piecewise-continuous representations of the fitted data, each patterned after the hypothesized existence of an ALD window exhibiting constant or linear GPC(T) dependence. All three regression approaches identified a low-T region characterized by rising GPC with T, reaching a maximum of approximately 1.1 Å cycle−1 slightly below 200 °C, followed by a slower decline in GPC with T after this point. The results indicate that a temperature-independent ALD window for the TMA/water system may be relatively small (with lower and upper limits of 177 and 208 °C, respectively), if it exists at all.