Penn State scientists model SiC sublimation at 3000–4500 K to improve power chip growth
Researchers developed a ReaxFF force field showing temperature-dependent vapor shifts from Si-rich to C-rich during silicon carbide sublimation.
Researchers from Pennsylvania State University have developed a specialized ReaxFF reactive force field to simulate the high-temperature sublimation of silicon carbide (SiC), according to a study published in The Journal of Physical Chemistry C acs.org
The Journal of Physical Chemistry Cacs.orgOpen source ↗. The work, conducted within the university's Silicon Carbide Innovation Alliance, directly models atomic behavior at extreme temperatures between 3000 K and 4500 K. This atomistic modeling is aimed at improving the synthesis of defect-free SiC boules used in high-voltage power electronics for electric vehicles and energy systems.
Sublimation kinetics and vapor composition
Unlike conventional semiconductors grown from melts, bulk SiC is manufactured via physical vapor transport, where a solid source sublimates inside a closed graphite furnace and recrystallizes on a seed crystal. Utilizing molecular dynamics (MD) simulations, the team identified the exact temperature-dependent distribution of gas-phase species, including Si, SiC, SiC2, Si2C, Si2C2, Si2C3, and Si3C2. The simulations demonstrated that the initial stage of sublimation at 3000 K is dominated by silicon-rich gas species, but shifts toward carbon-rich compositions as temperatures increase.
This shift in vapor stoichiometry directly influences the ratio of material reaching the growth front. Imbalances during this transfer can lead to structural defects, such as the formation of unreacted silicon droplets, unwanted polytypes, or excess graphite inclusions within the ingot. The ReaxFF force field parameters were benchmarked against quantum mechanical data and compared with established machine learning potentials to verify both calculation speed and physical accuracy.
Bridging atomistic data to macroscale modeling
To apply these microscopic observations to commercial manufacturing equipment, the authors applied Arrhenius analysis to extract kinetic rate parameters from the MD trajectories acs.org
The Journal of Physical Chemistry Cacs.orgOpen source ↗. These calculated kinetic constants can be directly integrated into macroscopic computational fluid dynamics (CFD) packages. Combining atomistic chemical transition rates with macroscale CFD enables engineers to simulate mass transport, furnace heat transfer, and gas-phase stoichiometry across large growth chambers before conducting physical synthesis runs.
