Theory Guide

1. Electromagnetic Theory of Thin Films

1.1 Maxwell's Equations in Dielectric Media

The propagation of light through dielectric thin films is governed by Maxwell's equations. In a non-magnetic, isotropic medium with refractive index n, the electric field satisfies:

For plane waves with angular frequency ω and wavevector k, the solution is:

with dispersion relation: |k| = nω/c = 2πn/λ

1.2 Fresnel Coefficients

At an interface between media with refractive indices n₁ and n₂, the reflection and transmission coefficients for normal incidence are:

These coefficients represent the amplitude ratios of reflected/transmitted to incident fields. The reflectance (intensity ratio) is: R = |r₁₂|²

1.3 Phase Accumulation in Thin Films

Light propagating through a film of thickness d and refractive index n accumulates a phase shift:

This is the optical phase thickness. The round-trip phase difference between top and bottom surface reflections is δ = 2β = 4πnd/λ.

1.4 Airy Formula for Single Films

For a single thin film between air (n₀ = 1) and substrate (n₂), the reflectance is given by the Airy formula:

where r₀₁ and r₁₂ are the Fresnel coefficients at the air-film and film-substrate interfaces.

2. Transfer Matrix Method

2.1 Matrix Formalism

The transfer matrix method provides a systematic approach for multi-layer systems. The electric field amplitudes in forward (+) and backward (-) directions are related by 2×2 matrices.

Interface Matrix

At an interface, the field transformation is:

Propagation Matrix

Through a layer of thickness d:

2.2 Total System Matrix

For a stack of N layers, the total transfer matrix is the product:

The total reflection coefficient is extracted as:

And the reflectance: R = |r_total|²

3. Dispersion Models

3.1 Cauchy Dispersion

The refractive index wavelength dependence for transparent dielectrics is modeled by the Cauchy equation:

where λ is in micrometers, and A, B, C are material-dependent constants. This is valid in the transparent region, far from absorption bands.

Material Parameters

Material	A	B (μm²)	n @ 550nm
SiO₂	1.4508	0.0036	1.46
Si₃N₄	1.9963	0.0068	2.02
TiO₂	2.2716	0.0179	2.35
Ta₂O₅	2.0772	0.0145	2.12

3.2 Sellmeier Formula

For higher accuracy near absorption features:

This accounts for resonances at wavelengths √Cᵢ with strengths Bᵢ.

4. FFT-Based Thickness Analysis

4.1 Fringe Frequency Analysis

The spacing between adjacent maxima in the reflectance spectrum encodes the optical thickness. Transforming to optical frequency space (ν = 1/λ), the fringes become periodic with frequency:

4.2 Fourier Transform Method

The discrete Fourier transform of the reflectance R(λᵢ) is:

The power spectrum |F(k)|² exhibits a peak at frequency k_peak corresponding to the optical thickness. The physical thickness is:

where n̄ is the average refractive index over the measured spectrum.

4.3 Signal Processing Enhancements

Detrending

Remove linear baseline drift to isolate oscillatory component:

Windowing

Apply window function to reduce spectral leakage. Hann window:

Zero-Padding

Extend signal with zeros to increase frequency resolution without adding information:

5. Optimization Algorithms

5.1 Levenberg-Marquardt Algorithm

This gradient-based method minimizes the sum of squared residuals:

The parameter update is:

where J is the Jacobian, r is the residual vector, and λ is the damping parameter that interpolates between gradient descent (large λ) and Gauss-Newton (small λ).

Convergence Criteria

• |d_k+1 - d_k| < ε (parameter tolerance)
• |χ²_k+1 - χ²_k| < ε (cost tolerance)
• Maximum iterations reached

5.2 Differential Evolution

A stochastic global optimizer using population-based mutation and selection:

Algorithm Steps

1. Initialize: Random population of NP parameter vectors in search space
2. Mutation: For each individual x_i, create mutant:
   v_i = x_r1 + F(x_r2 - x_r3) (20)
3. Crossover: Mix parent and mutant with probability CR
4. Selection: Keep offspring if fitness improves
5. Iterate: Until convergence or max generations

Hyperparameters

• Population size: NP = 10-20 (for 1D problem)
• Mutation factor: F = 0.8
• Crossover rate: CR = 0.9

6. Uncertainty Analysis

6.1 Sources of Uncertainty

Source	Typical Magnitude	Mitigation
Refractive index error	1-5%	Accurate material characterization
Spectral noise	0.5-2%	Multiple measurements, averaging
Wavelength calibration	0.1-0.5 nm	Periodic calibration with standards
Non-uniformity	2-10%	Multi-point mapping

6.2 Error Propagation

For independent variables with uncertainties, the combined uncertainty in thickness is:

6.3 Fitting Uncertainty from Jacobian

The covariance matrix of fitted parameters is:

The standard error in thickness: σ_d = √(Cov₁₁)

References

1. Born, M. & Wolf, E. Principles of Optics, 7th ed. (Cambridge University Press, 1999).
2. Tompkins, H. G. & Irene, E. A. Handbook of Ellipsometry (William Andrew Publishing, 2005).
3. Heavens, O. S. Optical Properties of Thin Solid Films (Dover Publications, 1991).
4. Stenzel, O. The Physics of Thin Film Optical Spectra (Springer, 2005).
5. Nocedal, J. & Wright, S. Numerical Optimization, 2nd ed. (Springer, 2006).
6. Press, W. H. et al. Numerical Recipes, 3rd ed. (Cambridge University Press, 2007).