Direct-simulation Monte Carlo modeling of reactor-scale gas-dynamic phenomena in a multiwafer atomic-layer deposition batch reactor

2021 
Atomic layer deposition (ALD) using multiwafer batch reactors has now emerged as the manufacturing process of choice for modern microelectronics at a massive scale. Stringent process requirements of thin film deposition uniformity within wafer (WiW) and wafer–wafer (WTW) in the batch, film conformity along submicrometer wafer features, thin film quality, and the utilization of expensive precursors in the reactor dictate ALD reactor design and process parameter optimization. This paper discusses a particle-based direct-simulation Monte Carlo (DSMC) of the full reactor scale simulation that overcomes the low Knudsen number limitation of typical continuum computational fluid dynamics approaches used for modeling low-pressure ALD reactors. A representative industrial multiwafer batch reactor used for the deposition of Si-based thin films with N2 and Si2Cl6 (hexachlorodisilane) as process feed gases with pressures in the range 43–130 Pa and a uniform reactor temperature of 600 °C is simulated. The model provides detailed insights into the flow physics associated with the transport of the precursor species from the inlets, through wafer feed nozzles, into the interwafer regions, and finally through the outlet. The reactor operating conditions are shown to be in the slip/transitional flow regime for much of the reactor volume and especially the feed gas nozzle and interwafer regions (where the Knudsen number approaches ∼0.2), justifying the need for a high-Knudsen number DSMC approach as in this work. For the simulated conditions, the nonuniformity of precursor species immediately above the wafer surface is predicted to be within <1% for a given wafer and <2% across the entire multiwafer stack. Results indicate that higher pressure degrades WiW and WTW uniformity. A reactor flow efficiency is defined and found to be ∼99%, irrespective of the chamber pressure.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    4
    References
    0
    Citations
    NaN
    KQI
    []