dimension sweep and optical generation object




I am trying to sweep x dimension of my cell, and check Jsc and Voc (generally efficiency) versus different values of x.
my question is, should I do the x sweep in FDTD and DEVICE both? or it is enough to sweep only in the device?
I am asking because in solar generation analysis we define a unit cell and a period number to have solar generation object match with the dimension of the cell (here it is x).
when I sweep x, how this period number can change automatically to have proper generation object to cover the whole dimension of x?

Thanks a lot in advance


The answer to this questions is dependent on the device geometry. For your structure, since a change in the value of x does not affect the FDTD unit cell, you can only vary x in your DEVICE simulation. The effect of a larger x would be that the photogenerated carriers will need to travel a longer distance to reach the emitter and thus will have a greater possibility of recombining with each other before reaching the contact.

In your simulation setup (DEVICE) you need to only simulate half of the electrical unit cell (region between two emitters) because of the symmetry of the structure. If you do this then you can very easily sweep for different values of x by using a generation rate that will cover the device for your largest value of x. Take a look at the cartoon below:


Hi Aaalam

Thanks a lot for the help.
what is the maximum x (limit of the DEVICE) to get reliable results?


The maximum value of x is going to be limited by the computational resources. As you make x larger and larger, the simulation will require more and more memory. This is going to be the limiting factor. If you have access to a system with very large memory then you can make x very large. The memory requirement will almost scale linearly with x so you can run a few simulations with different values of x to get a feel of what the maximum limit is for your system.


Hi Aalam

I am very thankful for your great help.
The problem of the suggestion is that my design is one side contact and I cannot simulate half of the cell. Both base and emitter are at the top.


In case I have enough memory resources if one of the dimensions of the device is in mm or cm, is the results of Lumerical reliable?


Hi. If you have to simulate the entire width of the structure then you will also have to use generate multiple generation data for the different x values. The work flow in that could be,

  1. Run the simulation in FDTD.

  2. Run the analysis group multiple times each time changing the filename and the value for “periods” to create G for different length.

  3. In DEVICE use a script to load the different data (G) and run the simulation for different values of x.


The result from the solver should be reliable even at larger scale provided that you have the resources to simulate it. Please keep in mind though that in order to simulate a mm x mm structure you will probably need more that a few hundred gigabytes of memory.