Discussion Forum

You can surely insert an inhomogeneous refractive index in the form of analytical expressions and may be through interpolated data as well which I never tried though. But intensity dependence is a non-linear effect and I don't know of any trick that can help you achieve that in linear simulations.

What you need here is time domain simulation in which you specify your D field with a third order non-linearity.

Hi!

I'm trying to simulate the nonlinear optic effect in wave propagation. As a result, my refractive index = n_o+n_2*(I). Because intensity I varies, I get a inhomogenous refractive index in the 2D domain. But I can only enter a constant for refractive index. I wonder if COMSOL has any tricks to solve this inhomogenous material.

Thanks,

Tina

Hello, have you solve the problem? I'm now faced with the same problem and wander whether I can insert n=n_0+2*I into the refractive index form in time-harmonic mode.

Hi
You can write as n+n2*(epsilon0*c*n*Ez^2/2) because term in the paranthesis reporesents intensity.

Hi

I'm also trying to include and simulate the nonlinear refractive index effect within optical waveguide using the RF module. Where should I include that n+n2* (epsilon0*c*n*Ez^2/2)? At material definition or at the parameter inputs.

I'll appreciate your help.

Regards
Medya

I wonder whether it's possible to model the intensity dependent refractive index by iteration. The idea is that you first simulate the field distribution in the presence of a homogeneous refractive index distribution and then calculate n(I) from the intensity distribution using n(I) = n_0 + n_2 * I. Then you solve a new field distribution in the presence of the updated refractive index distribution n(I) (that is now inhomogeneous) and update the refractive index again. You continue doing this until the intensity and the refractive index distribution do not change. At this point the solution has converged and fulfills kind of a self-consistency.

For a case that my suggestion above does not work, I have checked the Comsol Wave Optics Module user guide for several ways of modeling nonlinear refractive index. I suggest everyone interested in the topic to visit the Wave Optics Application Library. There're at least two example models, called "self focusing" (of a Gaussian beam) and "second harmonic generation", that are dealing with nonlinear optics (they exist at leat in Comsol 5). The self focusing example uses nonlinear refractive index. In general, it seems that a time dependent solver has been used quite often to solve problems like the one discussed here.

Hope I was clear enough and someone still reads this old conversation.

Added later for a case that it is problematic to model an inhomogeneous refractive index: You can define a function f under the definitions node of the model tree and use (x,y,z) as its input arguments. Then you only need to write n = f(x,y,z) in the material properties and you should have an inhomogeneous index distribution.

Added even later: If there's a strong frequency generation do to the nonlinear effects, I might guess that the above method does not work. You can still try it if you can assume that frequency generation is negligible or if you can model multiple wavelengths in Comsol. In practice this requires more simulations and can be slow if you use the frequency domain solver.

Friendly regards
VK

Hello,

There are actually several ways to address these modeling areas in the frequency domain. As noted, this example:
www.comsol.com/model/self-focusing-14639
Models a intensity-dependent refractive index in the frequency domain.

With respect to modeling second harmonic generation in the frequency domain, please see:
www.comsol.com/model/second-harmonic-generation-in-the-frequency-domain-24151

Of course you can also model in the time domain:
www.comsol.com/model/second-harmonic-generation-from-a-gaussian-beam-wave-optics-14701

Please do also keep in mind that solving in the frequency-domain is going to much faster than modeling in the time domain.

Best Regards,