Learning Forum

I wanted to simulate a metal-coated dielectric device in 3D FDTD but it took too long and huge memory. Why does it take too long? Is there another way to simulate metal-coated dielectric structures or any examples? Here the metal, I want to use Ag or Au, is for a reflector. I want to see the transmittance of the light at 900nm.

[Amrita Pati]

Hi,

In the default non-uniform mesh in FDTD, the number of mesh points depends upon the effective wavelength of source in different materials in the simulation region. Since, metals have large permittivites, the mesh resolution is typically very high in metals (finer mesh), which requires more memory and time to simulate. At the intial stages, you can use a lower mesh accuracy and you may also use the uniform mesh to have more control over the mesh size. Would you be able to share a screenshot of the geometry? This will help assess if the geometry can be simulated faster in any of our other solvers. Thanks!

Regards,
Amrita

[Joohyung Song]

Hi,

Thank you for the reply. I am simulating the parabolic reflector with a dipole source at 900-1200nm. I want to see the extraction efficiency that emits to the bottom(see the structure below), so I am using far field analysis in FDTD. In the below figure, I haven't added metal coating but I added the 100nm Au on the reflector for the other simulation file. 

[Amrita Pati]

Hi Joohyung,

Thanks for sharing additional information! Looks like you need to use FDTD to simulate this, other approximate methods won't work well for this geometry. But I will recommend you to use Symmetric/Anti-Symmetric BCs, they can leverage the symmetry of the structure and fields to reduce the computational domain and thus the resource requirements. Please use this link to learn more about them: Symmetric and anti-symmetric BCs in FDTD and MODE. As for the mesh accuracy, you can try a lower mesh accuracy for the intial tests, but eventually, I will recommend you to perform convergence testing with respect to the mesh resolution to confirm the accuracy of the simulation. 

Regards,
Amrita