ZEMAX Users' Knowledge Base - http://www.zemax.com/kb
How To Model a Gradient Index version of a Sellmeier1 Glass using the Gradient 5 Surface
http://www.zemax.com/kb/articles/274/1/How-To-Model-a-Gradient-Index-version-of-a-Sellmeier1-Glass-using-the-Gradient-5-Surface/Page1.html
By Brian Bauman
Published on 22 June 2009
 

This article explains how to convert Sellmeier 1 dispersion coefficient data into dispersion coefficient values for the Gradient 5 surface. After doing so, the Gradient 5 surface may then be used to model the variation of index with position for a Sellmeier 1 glass.


Calculating Gradient 5 Dispersion Constants
The Gradient 5 surface type has parameters to model variations of the refractive index in r and z. Other gradient surface types can model these variations as well, but what makes Gradient 5 different is that it includes dispersion parameters which mimic the often-used Sellmeier 1 formula. The dispersion formula for the Gradient 5 surface type is:

Gradient 5 dispersion formulae

while the dispersion given by the Sellmeier 1 formula is:

Sellmeier 1 dispersion formula

If one is modeling a Sellmeier 1 glass that also has r and/or z variations (say, because of temperature gradients), then only the first K term and the first L term are necessary, since the Sellmeier K and L parameters do not vary with the index at the reference wavelength. In other words, the only terms needed for the Gradient 5 surface will be K11, K21, K31, L11, L21, and L31. Note that the convention that I’m using here is that K’s and L’s with one index are Sellmeier constants, and those with two indices are Gradient 5 constants.

The Li1’s are defined in the same way as the Li terms in the Sellmeier formula, which makes sense because the index formulae should have singularities at the same wavelengths (assuming small changes in index). However, the Ki1’s (which generally indicate the strength of the absorption line) are not defined the same way as the Ki terms in the Sellmeier formula. If one evaluates the Sellmeier formula for a wavelength λ and a wavelength λref, takes the difference between the expressions, and compares to the Gradient 5 formula, one finds that:

Conversion from Sellmeier to Gradient 5 dispersion coefficients

with the other Kij and Lij terms being zero.

One item to note is that it does not matter what wavelength is used as λref as long as the Ki1’s are calculated using the same wavelength, and that same wavelength is used in finding the base index n0 in the Gradient 5 surface type.

If one has a glass that is not described by a Sellmeier 1 formula, then one can use the index fitting feature (Tools > Catalogs > Glass Fitting) to generate a Sellmeier 1 model. More information on the Glass Fitting Tool may be found in the article entitled “Fitting Index Data in ZEMAX”.

Example
This example uses fused silica as the glass to be modeled by the Gradient 5 surface type. The Glass Catalog entry for fused silica is shown below, followed by the relevant 10 lines from the SGRIN.DAT file. Note that the K’s are different, but that the L’s are the same. Also, note that the K’s and L’s are in different orders.
 
Glas catalog information for Silica

Gfusedsilica   ← name of Gradient 5 glass
0.21 3.71      ← wavelength range
0.58756180     ← reference wavelength
1 1            ← # of Kij and Lij terms
-0.00956531    ← K11
-0.01661702    ← K21
0.900654310    ← K31
0.004679148    ← L11 (same as L1 in Sellmeier formula)
0.013512060    ← L21 (same as L2 in Sellmeier formula)
97.93400000    ← L31 (same as L3 in Sellmeier formula)

To check that the model is correct, put both the original glass and the Gradient 5 glass in your lens file and plot the index versus wavelength for the two surfaces using the Universal Plot 1D feature (the usual dispersion plot feature found under the Analysis > Glass and Gradient Index > Dispersion Diagram menu will not work because it will not accommodate the Gradient 5 feature). Be sure that the Gradient 5 surface has zeros for the r and z parameters and that the n0 term is set according to the reference wavelength in the SGRIN.DAT file. If the model is correct and the horizontal/vertical scales are the same, then overlaying the two index plots will show no difference between the indices of the two surfaces:

Comparison between Sellmeier 1 and Gradient 5 dispersion results

The lens file which illustrates this comparison is provided as an attachment at the end of this article.