A hybrid computational model for precise SiC epitaxial thickness determination: combining Sellmeier dispersion modeling with Fourier transform

This work presents a single-beam, interference-based framework for precise thickness measurement of silicon carbide (SiC) epitaxial layers, augmented by multi-method cross-validation and multi-beam interference compensation. We formulate the core model by combining Sellmeier dispersion with Snell’s law and apply it to angle-resolved reflectance acquired over 400–2500 nm. Thickness is independently inverted via least squares (LS), extrema-interval analysis, and fast Fourier transform (FFT); the three estimates exhibit strong agreement. To mitigate practical multi-beam effects, we detect interference using the Secondary Peak Ratio (SPR) criterion and correct the phase/thickness using an Airy-based model. Experiments show a representative thickness of 7.435 μm after correction, with an overall deviation reduction of ~0.04 μm and a minimum standard deviation of 0.0099 μm, demonstrating high accuracy and stability of the proposed approach. The framework achieves precise, prior-free thickness metrology for SiC epitaxial layers and is readily extensible to other substrates.