Explore customizable surface chemistry models in Ansys Fluent for optimizing chemical vapor deposition with enhanced predictive insights and process efficiency.
Growing a uniform and stoichiometric layer by Chemical Vapor Deposition (CVD) requires balancing complex interactions between precursor chemistry, reactor flow, and operating conditions. In this blog, we explore the case of Trimethyl Gallium (TMGa) and Arsine (AsH₃) undergoing reactions on a heated, rotating substrate, and the outcome is sensitive to disk rotation, gas flow, and temperature. Without a predictive framework, achieving consistent layer quality can be costly and time-intensive.
The CFD model includes surface reactions describing the exchange processes between adsorbed species (Ga_s, As_s) and gas-phase precursors (AsH₃, TMGa) that lead to deposition of solid GaAs. This model is based on Ansys Fluent tutorial.
|
Reaction |
Role in Growth |
Description |
|---|---|---|
| AsH₃ + Ga_s → Ga(s) + As_s + 1.5H₂ | Arsine decomposition and As incorporation | Gas-phase AsH₃ (arsine) interacts with surface Ga sites. The As from AsH₃ replaces Ga atom on the surface, releasing H₂. This forms surface As atoms. |
| Ga(CH₃)₃ + As_s → As(s) + Ga_s + 3CH₃ | Trimethylgallium (TMGa) decomposition and Ga incorporation | TMGa interacts with As-terminated sites. The Ga from TMGa replaces As atom on the surface, releasing CH₃ radicals. This forms surface Ga atoms. |
These reactions are implemented in a Ansys Fluent CFD model where an inlet gas containing Arsine, TMGa and hydrogen gas is injected at a constant velocity in the geometry shown on the side.
A customizable surface chemistry model has been developed. It explicitly incorporates the two surface reactions governing Ga and As deposition and couples them with Arrhenius-based rate expressions.
In this application, the species transport model is employed, with fluid species, solid species, and site species explicitly defined within the mixture. Reaction rates are specified directly through the Fluent graphical user interface, as well as via User Defined Functions for more advanced kinetics. The simulation setup and resulting data are presented below. Notably, the majority of TMGa is consumed, particularly near the center of the disk.
PyFluent
We can carry out parametric studies in PyFluent by varying parameters to evaluate their effects on how the surface deposition rates of Ga and As:
Rotational Velocity: Determines boundary layer thickness and radial mass transport.
Gas Inlet Velocity: Balances precursor delivery with utilization efficiency.
Disk Temperature: Modulates reaction rates through activation energy.
The values of each parameter is associated with an input variable defined in the Fluent case file, and can be easily modified as shown in the code below:
The code provided enables modeling of CVD across a range of operating conditions and supports the generation of results such as those illustrated below:
User Defined Functions
For complex reaction kinetics that are not available within the standard Reactions window, users can implement custom reaction rate models by developing User Defined Functions (UDFs). The following workflow outlines the process for integration:
| #include "udf.h" real arrhenius_rate(real temp, real Pre_exp, real Ea) { return Pre_exp*exp(-Ea/(8.314*temp)); } DEFINE_SR_RATE(my_rate,f,t,r,mw,yi,rr) { Thread *t0=t->t0; cell_t c0=F_C0(f,t); real T = C_T(c0,t0); real ash3 = yi[0]; real game3 = yi[1]; real ga_s = yi[7]; real as_s = yi[6]; real rho_w = 1.0, site_rho = 1.0e-8; real T_w = F_T(f,t); //get the temperature on the wall rho_w = C_R(c0,t0)*C_T(c0,t0)/T_w; // get the gas density near the wall ash3 *= rho_w/mw[0]; game3 *= rho_w/mw[1]; ga_s *= site_rho; as_s *= site_rho; real rate_arrhenius = 0.0; real rate_transport = 0.0; real effective_rate = 0.0; if (STREQ(r->name, "ga-dep")) { rate_arrhenius = arrhenius_rate(T_w, 110000000,9000); rate_transport = 1000000*pow(T_w,0.5); effective_rate = (rate_transport*rate_arrhenius)/(rate_arrhenius+rate_transport); *rr = effective_rate*ash3*ga_s; } else if (STREQ(r->name, "as-dep")) { rate_arrhenius = arrhenius_rate(T_w, 8e13,9000); rate_transport = 1.0e12*pow(T_w,0.5); effective_rate = (rate_transport*rate_arrhenius)/(rate_arrhenius+rate_transport); *rr = effective_rate*game3*as_s; } } |
Once the UDF is properly loaded and linked, Fluent will automatically use the rate generated by the user defined function as a source term.
Surface Deposition Thickness
Details about the simulation setup are shown in this video:
Mesh and python scripts: [Link]
By capturing both reaction kinetics and process dynamics, surface chemistry modeling in Fluent equips engineers and researchers with powerful tools to design and refine surface reaction processes efficiently and reliably. Leveraging Fluent's reaction can bring about several advantages, including the following:
Predictive Insights: Anticipates how changing rotation, flow, or temperature impacts growth rate and uniformity.
Reduced Experimentation: Lowers reliance on trial-and-error by mapping process sensitivities computationally.
Process Optimization: Identifies windows of operation for uniform, stoichiometric GaAs films.
Scalability: Applicable across single-disk research systems and multi-wafer industrial reactors.
Unlocking the Power of ANSYS Simulation for CVD [Link]
Unlocking the Potential of ANSYS Simulation for Plasma Enhanced Chemical Vapor Deposition [Link]