Home Forums FDTD Create a new material

Viewing 7 reply threads
  • Author
    Posts
    • #27452
      Nathalie Carrier
      Participant

      Hi everyone,

      I am using FDTD 12 (1 month evaluation) and I would like to create a new material in the FDTD profile designer based on experimental data.

      These data are stored in a .txt file and contain 3 columns (wavelenght, index real part, index imaginary part). Is it possible to import them directly ?

      Thanks for your help !
      Nathalie

    • #27481
      Damian Marek
      Participant

      The materials that are used in FDTD analysis MUST abide by the Kramers-Kronig relation. It is possible to fit the experimental data to a Lorentz drude model and then define a material with those coefficients. You can Matlab code from quick searches that will fit these problems.

      http://fdtd.kintechlab.com/en/fitting

      An example of a Matlab program can be found at the link.

      Regards

    • #27538
      Nathalie Carrier
      Participant

      Hi Damian, thanks a lot for the reply, I will fit my data with a Lorentz Drude model.
      \Nathalie

    • #27878
      Nathalie Carrier
      Participant

      Hello,

      I am using the MatLab code that you provided me.
      However fitting my data with ‘Lorentz Drude’ model appears not to be very accurate (see attached the best result I can get using 6 oscillators).
      Would it be possible to import the data directly from XML format ?

      Thank you again,
      \Nathalie

      • #27933
        Damian Marek
        Participant

        Unfortunately no, the permittivity must be provided in Sellmeier, Lorentz, Drude or Lorentz Drude format for dispersive simulations. You could choose one wavelength you are interested in and run a CW simulation with a single sample of the data or maybe there is a matlab script out there that can perform a better fit.

        Regards

    • #27947
      Nathalie Carrier
      Participant

      Thanks for the reply, I will focus more on the fitting then. It should be possible to find a better fit!
      \Nathalie

    • #27963
      Nathalie Carrier
      Participant

      One last question, I tried to set a new material (with Sellmeier model) with the oscillator parameters I found (even if the fit is not perfect) but it seems that negative amplitudes are not supported. How to overcome that ?
      I tried Lorentz-Drude, same issue.

      Thank you,
      Nathalie

      • #28223
        Damian Marek
        Participant

        The Lorentz-Drude model should allow negative strengths under the “S” parameter, I just verified that it works. It won’t work for Sellmeier model since it does not model loss. However, I am not sure what the physical meaning of a negative strength is!

    • #28853
      Nathalie Carrier
      Participant

      Hi thank you for the reply, I do not think there is a physical meaning of a negatig amplitude but the fit is good enough and in the Lorentz-Drude model form (and the fit does not converge with only positive amplitudes).

      I still have a question regarding how to fit the permittivity. The convention used by the software is: a loss correspond to a negative imaginary part. For the example shown on the attached picture on the post from November 18, 2015 at 11:38 am, it means the imaginary part should be set as negative before fitting ?

      Thank you,
      Nathalie

      • #28855
        Damian Marek
        Participant

        Yes you should use the negative of the imaginary component and then do the fitting.

    • #29235
      kasun gankanda
      Participant

      Hi
      i am a new user to this software and is planning on using it for my research work.
      currently i am making a simple simulation of SPR on a gold layer with a small slit on a glass surface .

      My problom is i cannot find how to make the glass material
      and since i am planning to use Kretschmann cofiguration for SPR excitation .
      what type of excitatio wave is good for the simulation.

      thank you in advance .

Viewing 7 reply threads
  • You must be logged in to reply to this topic.