Predictive process design: a theoretical model of atomic layer deposition

We present a theoretical framework for the predictive design of atomic layer deposition (ALD). ALD is the leading process for the controlled deposition of thin films in a variety of technologies. For example, insulating alumina layers are fabricated by ALD for electroluminescent flat-screen displays and for node DRAM, and are under investigation as high-k dielectrics for the MOSFET gate. To develop and optimise an ALD process for a new material requires knowledge of the reaction mechanism. By combining structures computed at the quantum mechanical level and literature data from in situ experiments, we develop a quantitative model of the ALD reaction cycle for alumina deposition. We are thus able to identify the intrinsic limits on ALD growth and explain how the growth rate depends on process conditions.