Special ChBE Seminar: Simon D. Elliott
Friday, October 14, 2011
12:00 p.m.-1:00 p.m.
Room 1146 A.V. Williams Building
Professor Ray Adomaitis
Reaction Modeling for Atomic Layer Deposition of High-k Dielectrics Onto III-V Substrates
Simon D. Elliott
Senior Staff Researcher and Cluster Co-Principal Investigator
Tyndall National Institute
University College Cork, Dyke Parade, Cork, Ireland
Atomic layer deposition (ALD) is a technique for the controlled growth of ultrathin films, which is increasingly being applied in high-tech sectors like nanoelectronics, photovoltaics, displays, but also being used to coat textiles and packaging material. The success of an ALD process depends more than anything else on the chemistry of the precursor molecules and substrate, and so an understanding of chemical mechanism is crucial. We are applying density functional theory (DFT) to a variety of ALD processes with the aim of elucidating the reaction mechanism, explaining experimental findings and designing new ALD processes.
Here we focus on important examples of the ALD of high-k dielectrics onto semiconductor substrates. ALD-HfO2 is probably the material to have received most interest as a gate dielectric for transistors over the last ten years and is now in commercial production. There have been many computational studies of the underlying reaction steps. Nevertheless, a higher-level understanding is lacking of how the reactions link together and how a layer of HfO2 film grows. To address this, we are developing a kinetic Monte Carlo model of HfO2 growth from Hf(NMe2)4+H2O, and in doing so, we have had to revisit the atomic-scale reaction steps at the DFT level. We present results showing the evolution of thickness, roughness and impurity concentration in time.
The second example is in a related area. There is a drive for future transistors to be based on high mobility III-V semiconductors, such as InGaAs, but this raises new challenges for materials processing. Defects at the interface between III-V and dielectric are an obstinate problem, but the situation has been improved by discovery of the clean-up effect during deposition of Al2O3 dielectric onto oxidised III-V substrates. Clean-up is the observation that the ALD precursor trimethylaluminum removes arsenic oxides from the substrate before growth of the Al2O3 layer. We have simulated these reactions at the atomic scale using DFT, with a view to explaining clean-up, and we find that a few important elementary steps (such as methyl transfer) lead to a surprising range of by-products. More generally, this helps us understand the initial steps of ALD onto semiconductor substrates.
We are grateful to Science Foundation Ireland for funding under the strategic research cluster FORME Functional Oxides and Related Materials for Electronics. http://www.tyndall.ie/forme