Pump probe simulation in 3D



My structure includes a 100 nm thick metal layer on which there are periodic nanocavities. placed on a glass substrate. What I would like to see is the effect of a thin gain layer located over the metal layer on the transmission through the cavities. I started with the pump probe simulation application given in the application gallery. This is a 2D application. I first attempted to modify this 2D simulation file to 3D. (I did not put the metal layer and the substrate. Just the source gain layer and the monitors) I took the 2D simuation file and I changed the dimension from 2D to 3D, I modified the BCs so that it is periodic in x and y axes and PML in z axis I modified the source so that it is injected along z axis and I added a mesh region. And the gain layer is a 40 nm thick layer at the middle of the FDTD region. When I keep the parameters of the Laser material same as the 2D simulation, everything works fine. I get the transmission exceeding unity at the emission wavelength and I observed the population inversion. Then I changed the emission and absorption resonance wavelengths in the Laser plugin material to 860 and 500 nm respectively. (These were 1500 and 750 nm in the original file) And I reduced the pump intensity from 2e6 to 1.5e6 (Otherwise I get extremely high T) I again get nice results with these parameters.

My first question: when the pump is 1.5e6 and probe intensity is 1e4 (for abs wavelength = 500nm, em wavelength = 860 nm) I get a similar T spectrum to the one given in the application webpage (link given above) which is good. And when the pump is 150 and probe is 1, that is the same ratio with the previous trial, the story is totally different. T makes a dip instead of a peak at the emission wavelength. I really don’t understand why. What is the physical explanation for this? How exactly do I determine the order of magnitudes of the sources if I wanna change other parameters such as lifetime or time delay or linewith?

Second, now I need to change the lifetime to 10 ns (it is 0.3 ns in the original file) The parameters are such that emmission wavelength = 860 nm, absorption wavelength = 500 nm. Pump intensity = 1.5e6, probe intensity = 1e4. And the lifetime (t21 in the plugin material) should be 10 ns. (or even longer) So what other parameters do I need to modifiy to get a similar population inversion?

I attach my simulation file and the script file. Thank you!

pump_probe_3D_lifetime10ns.fsp (245.5 KB)
sc.lsf (748 Bytes)


Hi @bilge.yildizkarakul

Thank you for a detailed inquiry.

  1. These simulations are nonlinear in nature and will not follow a simple pump/probe ratio. Also the states has to be populated and fulfil the population inversion condition. When the pump energy is low, the excited states are not populated and thus it will also absorb the probe light. That’s why you are seeing a dip in the transmission. Please also note that the source amplitude should not be mistaken by source intensity.

  2. You will need to check and make sure that your gain material has population inversion. If not, increasing the pump source amplitude will be necessary. You can try this with 2D simulations that are faster and slowly increase the lifetime from 0.3ns to 10ns and adjust the parameters accordingly.

Hope this was helpful.


Thank you for the explanation. It was really helpful. I will go on with 2D simulation and try to get population inversion for 10 ns lifetime. I was wondering if the only parameters to be adjusted are pump-probe amplitudes while stepwisely increasing the lifetime. Do I need to consider other source parameters such as pulse length or offset, or plugin material parameters such as gamma a/b as well?


Hi @bilge.yildizkarakul

Sorry for a late reply, we were quite busy recently.

I have not tested the model with longer t30 and t21 values. If you have a reference that you can extract all the parameters that will be great. However this is my opinion on what might be most important:
The values of t32 and t10 are very low. I assume you can leave them as they are. The damping coefficients gamma a and gamma b are chosen to be 1e-13 but I do not have much information why this is the case. A good idea would to be to review the paper linked below and the book referred to in the pump-probe application example to find out how these parameters can be related:

It is also a good idea to monitor the population inversion and generated stimulated emission using time domain monitors to make sure that you correctly set the apodization in DFT monitors. You can find some more details in this regard in the link below:
Frequency monitor normalization on having multiple sources

I hope this was helpful.