I am using the FDTD and Device software to simulate the VPD, I have several question and I would appreciate your help on that,
In the webpage that has an example of the VPD, The simulated responsivity is larger than 1.35 A/W for reveres voltage larger than 0 while the maximum responsivity is 1.25 A/W for Non-avalanche PD at wavelength of 1550 nm. why it shows large value?
In order to generate the file for Device simulation, we need to define source power. what is the optimum range of the source power and why the responsivity is different for various power?
I simulated a 25um length VPD in FDTD and then I run analysis for various source power that generate 100uA and 1mA and I then I simulate the responsivity for each source power, for low power the responsivity is larger.
The Germanium that is used for FDTD simulation has absorption coefficent related to the bulk germanium but for the thin germanium this can be several times larger, I have tried to increase the absorption by changing the imaginary weight to 8 but I didn’t see difference in results. Do you have any suggestion on that?
The DEVICE simulation uses a length equal to 50 micron in the Y direction. This length is 5 times the length in FDTD simulation. The length was initially made longer in DEVICE to make the current value comparable with the experimental data. However, making the length 5 times larger compared to the FDTD simulation means that the effective input optical power is 5 times of what we defined in FDTD. The input optical power used in the DEVICE script should therefore be 5 times larger which would make the responsivity to be slightly below 0.3 A/W, much less than the reported value. This smaller number is primarily because we are simulating a smaller length of the photodetector (100 micron length in the experimental device) and as a result, a large portion of the incident light remains unabsorbed. If you make the length (in Y direction) larger in the FDTD simulation, the responsivity will start to get closer to the theoretical maximum.
Normally, I would expect the responsivity to be independent of the input power value. Can you tell me how different the values are between the two cases?
The “imaginary weight” variable only affects the fitting of the FDTD index model with the experimental data. It does not change the value of the index of the material. If you want to change the index of the material you will need to create a new material model. For example, if you simply want to increase the imaginary mart of the index, you can write a script to first get the index of the bulk material from the material database (use getindex). Next scale the values of the imaginary part. Finally create a new material with the updated index value using the sampled data material.
About question 1, what is the correct value for responsivity? I was confused about the curve that is reported in the webpage for the responsivity, even if we simulate the PD with 100um length we don’t expect that the responsivity become larger than 1.25A/W so I think the reported curve in the webpage
"https://kb.lumerical.com/en/pic_photodetectors_vertical_photodetector.html"can not be correct.
About question 2: the responsivity of the lower power is 10 times larger, it seems that the generated file for both power number is the same, although I see that when I run the analysis the simulated current is proportional to the source power. It seems that the calculated photocurrent for both cases are the same so with lower optical power the calculated responsivity is larger.
To get the correct value, we’ll need to simulate a 100 micron long device in FDTD and then perform a DEVICE simulation of it.
So are you getting the same current for both the generation rates (with different power) in DEVICE? If that is the case then may be somehow you are using the same data in DEVICE for the two simulations. When you calculate the generation rate in FDTD for different power, remember to change the name of the .mat file so that the files do not get overwritten. Next use the same DEVICE file to load the two .mat files one after the other to make sure that all other parameters remain the same. Let me know if you still get the same photocurrent.