跳到主要内容
版本：5.0 (最新)

Model-Material Refractive-Index Fitting

Introduction​

Model-material fitting uses parameterized optical models (such as Cauchy, Sellmeier, or other selectable model forms) to fit refractive-index behavior. It is used when a physically meaningful model is preferred over pointwise fitting.

Procedure​

  1. Select a material model that matches expected optical behavior.
  2. Load measured spectra and set initial model parameters.
  3. Configure parameter bounds, constraints, and optimization stop criteria.
  4. Run fitting and inspect model parameter convergence.
  5. Verify model validity in the target wavelength range and save parameters.

Import and Export​

Import an existing model material​

Use the model-material import button in the model parameter editor.

After import:

  1. The model type is switched to the imported type.
  2. The parameter editor is rebuilt to match that model.
  3. Parameters that do not belong to the new model are removed from the current layer UI.

Save the current model as material​

Use the upload button in the model parameter editor when you want to keep:

  1. The model type
  2. The model coefficients
  3. The ability to re-import the material as a model later

Do not use the table-editor upload path if your goal is to preserve the model structure itself.

How model materials are used​

Model materials are used in two different ways:

  1. Structure-sensitive usage
  2. Calculation-oriented usage

Structure-sensitive usage​

When the software needs the model structure itself, it must use the model-material path.

Examples:

  1. Refractive-index fitting model editor
  2. Restoring fitting configurations that rebuild the model UI
  3. Reusing a saved dispersion model for further parameter editing

Calculation-oriented usage​

When the software only needs refractive-index values for calculation, a model material can still be used.

The system evaluates the model at the requested wavelengths and then continues calculation.

This means model materials are not limited to fitting-only scenarios.

Example​

A transparent dielectric layer is fitted with a Cauchy-type model. Starting from empirical initial parameters, optimization converges and outputs model coefficients that reproduce measured spectra within process tolerance.

Notes​

  • Choose the simplest valid model first, then increase complexity only when needed.
  • Over-parameterized models may fit noise and reduce transferability.
  • Always check both spectral residuals and parameter physical plausibility.
  • If you want to reuse the fitted result as a reusable model, save it from the model editor or from the fit-result save path, not from the table editor.

If you are unsure whether the current result should be kept as a tabulated material or a model material, check this page first: