We present a multi-scale computational approach to model the gas-phase chemical kinetics for Metal-Organic Chemical Vapor Deposition (MOCVD) of WSe2 using W(CO)6 and H2Se as gas-phase precursors. This framework combines Quantum Mechanical (QM) methods based on Density Functional Theory (DFT), ReaxFF-based reactive molecular dynamics, and Computational Fluid Dynamics (CFD) to efficiently model the gas-phase physiochemical processes leading to WSe2 growth in a cold-wall horizontal MOCVD chamber. A detailed gas-phase chemical kinetic reaction model is developed to describe all major chemical reaction pathways from the precursors W(CO)6 and H2Se to the most thermodynamically stable molecules, with quantified kinetic rate constants. First QM calculations are performed to suggest key reaction types and to provide the necessary training set to determine ReaxFF reactive force-field parameters for the W/H/C/O/Se system. Using the developed force field, ReaxFF simulations are performed to identify all major chemical reaction pathways and determine their associated activation energies. Other kinetic parameters, together with the thermal and transport properties of all species involved, are estimated using well-established theories or correlations. This chemical kinetic model with thermal and transport parameters is then integrated into a reacting flow solver for full-scale CFD simulations of the MOCVD chamber under realistic operating conditions, to demonstrate its capabilities in predicting the major processes in the gas phase and qualitatively estimating thin film growth behavior. The predicted gas-phase concentrations of tungsten chalcogenides at the growth substrate correlate well with experimental measurements of average film thickness across the substrate, which suggests that crystal growth may result from surface deposition reactions of these species. This computational framework for the gas-phase chemical kinetics in MOCVD (prior to surface deposition and subsequent crystal growth) can thus test experimental MOCVD conditions, generate insights into more effective growth protocols, shed light on the significance of reactor geometry, and improve the reproducibility of MOCVD results across different experimental growth chambers. The computational framework is also transferable to other CVD chemistries.
All Science Journal Classification (ASJC) codes
- Condensed Matter Physics
- Inorganic Chemistry
- Materials Chemistry