Atomic layer deposition (ALD) is evolving beyond binary compounds to complex oxides and doped structures, taking advantage of the nanometer precision ALD provides. In practice, the development of complex ALD-processes usually means performing many ALD-runs, as success at first attempt is unlikely. One factor at a time methods, where only one factor is altered and the rest are kept constant, are most often chosen due to their intuitive communication of control. However, they do have several drawbacks, being slow, neglecting secondary effects, and are usually not randomized—meaning that errors that arise over time can easily be overlooked. We here dig into our statistical toolbox and show how design of experiments (DoE) can be used to efficiently develop an ALD-process to deposit crystalline, luminescent CaMoO4—a proposed material for optoelectronic applications, like light emitting diodes or as a host for solar down-converters. Using DoE enables screening for a wider range of deposition temperatures, pulsed composition, and annealing parameters, by only performing nine ALD-runs in our case. We moreover look into how these parameters affect crystallinity, composition, and the photoluminescence properties and use DoE to show which factors have the greatest effects on these properties. The work also lays out the basic theory of the DoE-field and how to implement DoE in developing ALD-processes, in general, to ease the usage of DoE for the ALD-community.

Supplementary Material

You do not currently have access to this content.