i simulate the simple grating coupler which have been proved in a famous paper to varificate fdtd solutions is practicable in caculation of coupling coefficient. but the results seen to be wrong, almost double. so i modify the structure，replacing the grating by a simple waveguide ，coupling coefficient is very big, e+8。i caculate the coefficient by mode expansion,please check for my lsp.thanks.simple_example.fsp (4.5 MB)
Are you able to tell me the information of the paper so I can try to take a look close look at it?
I apologize for the delays. It has been a very busy week.
Thank you for providing more information on this problem. I have briefly taken a look at the thesis, but I was not completely sure which design/figure you are hoping to reproduce. If you can let me know, I can try to compare the results with it.
I have edited you file based on some general best practices of a grating coupler simulation. Such as
- reduced simulation region to focus on the coupling region. There is no obvious advantage to include that much of the fiber region like you initially did.
- the mesh looks a bit too coarse, I have added a mesh override there to better resolve the grating structure.
- The silicon substrate looks too thin, I think you should make sure the dimension of it. Right now it looks too thin to believe.
- The spectral range looks too broad for a typical grating couple, maybe you can reduce the range a bit. But it should not affect the simulation results too much.
Some grating couple workflow
- You should make use of the “above” monitor to find the far field angle of the emission. It can tell you the best angle of the fiber to collect light.
- Once you found the angle, then you can run a parameter sweep for the fiber x position. The parameter sweep can tell you the best position for the fiber to collect the most light.
I hope these suggestions can give you some idea. I am going to visit your university on April 24-26th, you can come speak to me if you follow up questions, or by replying this post.
Edited file attached. Note that you will need to reselect the mode in the mode expansion monitor. I have cleared the mode to reduce the size of the file to allow me to upload it.
simple_example (1).fsp (294.0 KB)
I am hoping to reproduce the vertical coupling with air in the paper.
thanks for your reply.i have learned much from your reply.
since the mesh accuration dont have much influence on coefficient but more run time,i just set at 2.
i altered the document,ran and found the cofficient not change much,bigger than papers as past.
i have used comsol to simulate the structure recently.the cofficient by comsol is 18%(ie. max),which is much the same as paper does.
i dont exactly know what cause the diffierence between comsol and FDTD.if i cant make any obvious progress in FDTD solutions, i probably use comsol to do my next research.
finaly, thanks for replying my question.
Sorry for the slow response, I am travelling for business right now. Did you manage to make any progress? I will be in SYSU in April 24-26th, please come find me if you would like to discuss your case in person.