Optimization of thermal field design for SiC epitaxial furnace (hot-wall CVD reactor)

The core objective is to achieve wafer surface temperature uniformity (≤±0.5–5℃) and temperature/flow field stability, thereby improving epitaxial layer thickness uniformity (<3%), doping uniformity (<8%), reducing defect density, and increasing growth rate (>60 μm/h).

Recent advances in SiC epitaxy process optimization have focused on thermal management, multi-parameter optimization, AI-assisted simulation, gas-flow regulation, and reactor structure upgrades. These developments aim to improve epitaxial layer uniformity, growth efficiency, defect control, and large-wafer industrial scalability.

Thermal Conductivity Modeling of Insulation Materials

One important research direction is the thermal conductivity modeling of fibrous graphite felt used in epitaxy reactors. Advanced analytical models have been developed to evaluate the apparent thermal conductivity while considering gas composition, chamber pressure, and operating temperature. Under hydrogen-rich carrier gas conditions, gas-phase heat transfer becomes the dominant heat-transfer mechanism. Studies show that reducing chamber pressure from 100 mbar to 1.5 mbar significantly decreases the required heating power. These models also enable more accurate prediction of temperature distribution throughout different reactor regions, helping prevent deposition non-uniformity caused by temperature variations outside the wafer area even when substrate temperature remains constant.

Multi-Objective Parameter Optimization Using FEM and Machine Learning

Another major breakthrough combines finite element modeling (FEM) with machine learning algorithms for multi-objective optimization. Key process parameters include total gas flow rate, growth temperature, chamber pressure, susceptor rotation speed, and gas distribution design. Optimization approaches such as MOPSO, NSGA-II, and SVM surrogate models have been widely adopted. Results demonstrate that thickness uniformity can be improved by approximately 30%, while Pareto-front optimization achieves both high growth rates and low coefficient of variation simultaneously. Optimal process windows are typically found at growth temperatures of 1450–1500°C, chamber pressures of 80–100 mbar, susceptor rotation speeds above 60 rpm, and asymmetric gas inlet ratios such as 5:16:5.

Transient Multiphysics Simulation Combined with Machine Learning

Recent studies also integrate transient CFD simulations with machine learning techniques to accelerate process optimization. Thermal-flow-chemical coupled CFD models combined with ACO-BPNN neural networks are used to optimize deposition temperature, inlet gas flow, rotation speed, and chamber pressure. Experimental validation shows excellent agreement between simulation and practical results, with prediction deviations of only 4.03% for growth rate and 0.49% for uniformity. This approach significantly shortens development and optimization cycles and is particularly suitable for horizontal hot-wall CVD reactors.

Gas Flow and Temperature Field Optimization

Optimization of gas-flow and thermal-field distribution remains critical for high-quality SiC epitaxy growth. Under optimized conditions, including an H₂ flow rate of 100 slm, flow split ratio of 20:60:20 (side:center:side), C/Si ratio of 0.95, growth temperature of 1610°C, and susceptor rotation, researchers achieved a highly stable parallel flow field and uniform temperature distribution. The wafer surface temperature gradient was reduced to only 19.3°C. In addition, nitrogen doping uniformity reached 3.35–4.85%, while crystal defects were significantly reduced to 28 total defects, including only 8 triangular defects and 6 basal plane dislocations (BPDs).

Equipment Structure Iteration and Industrialization

Industrial-scale reactor upgrades between 2023 and 2026 mainly focus on vertical split gas injection systems, multi-zone induction heating, compatibility with both single-wafer and dual-wafer configurations for 6–12 inch wafers, and graphite component redesign with automated preventive maintenance (PM). These structural improvements have enabled 8-inch and 12-inch SiC epitaxy processes to achieve thickness non-uniformity below 3% and doping variation below 8%. Furthermore, particle contamination has been reduced by approximately 50%, maintenance downtime shortened by 30%, and temperature variation controlled within ±5°C in dual-wafer systems.

Three Key Conclusions

1. Simulation + Machine Learning Has Become the Mainstream Method for Thermal Field Optimization: By coupling the thermo-fluid-chemical field through CFD/FEM, and combining it with ACO-BPNN or MOPSO/NSGA-II, the optimal Pareto parameters can be found within weeks (rather than traditional trial and error), significantly improving thickness/doping uniformity by more than 30% and reducing experimental costs. This is an essential tool for the large-scale epitaxial growth of 8–12-inch SiC.

2. The Influence of the Gas Phase (H₂ Pressure/Composition) Inside the Insulation Felt on the Apparent Thermal Conductivity Cannot Be Ignored: At high H₂ temperatures, gas phase heat transfer is dominant, and changes in pressure/precursor flow rate will alter the overall temperature distribution of the reactor. The latest analytical models can be directly embedded into CFD to achieve accurate power prediction and closed-loop thermal field control, which is the core of high efficiency, energy saving, and uniformity in thermal fireplaces.

3. Transition to larger sizes (8–12 inches) requires structural innovation: Domestic equipment has achieved wafer surface temperature ≤ ±0.5℃ and dual-wafer temperature difference ≤ 5℃ through vertical split air intake, multi-zone temperature control, and susceptor optimization. Thickness/doping uniformity has reached the international leading level, directly supporting cost reduction and doubling of production capacity. Horizontal hotwall + rotating susceptor is still the mainstream and there is no obvious controversy.

Semicorex offers high-quality components in epitaxial process. If you have any inquiries or need additional details, please don't hesitate to get in touch with us.

Contact phone # +86-13567891907

Email: sales@semicorex.com