Using dipole to mimic quantum dot emission


I am using a dipole source to emulate the emission from quantum dots.When I plot the total transmission collected in all the monitors(sum) it’s going above 1.
The issues isn’t solved even after I normalized the transmission using the following

The follwing is the graph for total transmission without using the above formulae

The follwing is the graph for total transmission using the above mentioned formulae

Kindly have a look at the FDTD file attached along and suggest possible solutions to solve this issue.Any help on this from would be great.

Thanking you!
Kiran VaddiOptimized_design_transmission.fsp (283.1 KB)


Dear @me12b099

I used a transmission box with no symmetry and here is the result:

and if its normalized to dipole power:

And here is the simulation file:

Optimized_design_transmission_BK.fsp (326.2 KB)



Dear @me12b099

I am glad that you could resolve the problem.



Dear @me12b099

What version of the software are you using? I am running 2017a version 8.17.1102 and everything works just fine. Maybe you can upgrade it to the latest version and let me know if you still have problem.



I am facing an issue with the total transmission being heavily wavy in the
wavelength region 0.6-1.1 microns while running the following simulation.
Attached screenshot shows the total transmission plot.Respective files can
be found below.
I am using dt stability factor of 0.1 and TiO2 for grating layer.
Can you please tell me what the possible issue can be?It’s very urgent so
please respond

Convergence_Check_Normalized_TB_Equiv.lsf (952 Bytes)

Optimization_reflection.fsp (305 KB)


Hi @me12b099

Attached file contains a grating outside the FDTD region. Can you please double check and make sure that you have uploaded the right file? In the mean time, I think the ripples are due to early shutoff. You can find lots of useful links by performing a search on KX like these links:

Please let me know if you had further questions.



I noticed that the transmission function is wavelength dependent when I run
the following simulation which I think shouldn’t be the case.
Can you please have a look and explain why it’s the case?
Or In other words,how do I make the dipole emit such that it has same power in all directions and all wavelengths.

Please find the respective fsp files attached.

Dipole_Explore.fsp (411 KB)


Dear @me12b099

I had a few concerns regarding your simulation file:

  1. PML boundaries are close to dipoles especially in the y directions.

  2. Rectangle object is cut before it reaches the PML layer. This means that you will get reflection from edges that might affect the results.

  3. Dipoles are located right at the material edges. Please note that every dipole requires a few mesh cell for light injection (demonstrated by white box around them) and I think it will be a good idea to position them inside the material such that they inject light in material with constant refractive index.

Also, I am not quite sure if your simulations is affected by field emitted by adjacent dipoles. Please refer to this page to learn more about this effect:

Please do the modification and update me with the results.



I had incorporated the suggestions mentioned but i still have few unsolved with the simulation files.
Theoretically when I run the following simulation,I expect the Transmission(normalized to dipolepower(f)) in the side monitors to be 0.5 (average) since the refractive index of substrate is 1.5.But I notice very low transmission values in the monitors kept at the sides.

Can you please have a look and let me know what could be the problem?

Please find the respective fsp file belowDipole_Explore_100 _Reference.fsp (1.4 MB)


Dear @me12b099

Can you please clarify why you theoretically expect 0.5 transmission on the side monitors? If you refer to total internal reflection, please note that the injected light from dipoles might get reflected a few times at the facet or rectangle objects.

There is one more thing to note:
The light emitted by dipole depends on the polarization. You can check this by putting a dipole on free space and adding monitors around it (assuming they are symmetric wrt source). Here is the result for a dipole with theta = 90, and phi=0:



The following slide talks about why I expect the transmission to be approximately 0.5 without any wavelength dependency.

If you think the the above explanations will not valid in the contest of this problem,please explain why.
Also,the transmission obtained in the case with a rectangular object with dipole theta = 90, and phi=0 is very low when I kept the dipole very close to each other and it’s variable when I keep them at the scale of wavelength region.(refer slide below)

It would be great if you can clarify the following:

  1. Since I am doing a 2D simulation,should I use theta = 90, and phi=0 and theta = 90, and phi=90 but not any other dipole configurations?

  2. Why is the transmission wavelength and dipole spacing dependent(and the reason for low values also)?


Dear @me12b099

In the quantum dot case, injected light is unpolarized and light comes in all directions with equal probability. This is not the case with dipoles used in FDTD simulations.

If you are using 2D, theta should be 90 and you can choose any value for Phi (a random distribution will be a better indication of quantum dot emission I guess).

I couldn’t read the plot’s axis and I am afraid I didn’t get you r second question. Can you please clarify it?



Hey I suggest you do the following,

You keep your 2D simulation but you run 3 simulations with your Dipole, in all orthogonal directions, X,Y,Z.
Then you average the results (this is similar to equal emission in all directions).
You can adjust your theta and phi values accordingly, but you should have the blue arrows point once in all 3 directions.
When using a electric dipole, there will be almost no emission in the direction of the blue arrows.

Now regarding the theoretical values of getting that efficiency, you’re forgetting something important, geometry.
It’s much harder to calculate, and you would have to use ray-tracing/shadow-casting (tricky).

A point source emits light in a sphere like shape. Now if you want to calculate 50% you need to collect half the sphere.
To do this you would need a bigger half sphere around it or and infinitely big plane with an acceptance angle of 90.

Now you can’t make an infinitely large plane, because of simulation restriction, but even large planes give the issue that I have multimode instead of single mode. Which would make this much more complex.
The other option in THEORY would be to place the dipole directly against the Waveguide/LSC. However this is not really doable in simulations.
The best options is making a straight waveguide/LCS that extends through your PML boundaries and then placing your QD in the center. That way half the ligth(sphere) is sent to the front and the other to the back. Placing a monitor at both sides, should give you what you’re looking for. If you would add up both monitors you should get your collection efficiency, note both monitors should give the same values. When using a waveguide you need to pay close attention to losses and probably use an expansion monitor.

Yes your LCS has a ~50% collection efficiency but only of the light that hits it. So if only 1/8 of thetotalemitted light hits the LCS, then it’s ~50% of 1/8, ~6%.


Thanks both of you.That explanation is really helpful.
I am currently using a dipole source with theta=phi=0 kept at the center of
wave guide such that it creates a lambartian field inside it.However,I am
not able to reason the following results from Lumerical FDTD simulations.

  • Given that a labartian filed is generated in the field I would expect
    the efficiency to be decreasing when I increase the refractive index of
    wave guide(sin(theta) is inversely proportional to refractive index thus
    theta should decrease).But as shown in the following slide that was not the
    case.Please explain why
  • Also,I have tried placing the dipole near to the edge but inside and I
    notice that the transmission now is wavelength dependent.It would be
    great if you can give an insight into why this is happening.


Well first of I think there’s a small mistake in the theory.

Seeing how the angle is defined, at theta 0 you would never have TIR,
So the efficiency is 1 - theta/90 which gives you ~54% instead 46/47%.

At theta is 90 you always have apparent complete TIR since it will neve cross the boundary because it’s parallel.

If you take into account then the math makes sense against.

In photonics we like higher refractive index materials like silicon because it allows us to make sharper bends because there’s TIR at lower angles. (sharper bends mean smaller structures, also it’s not really TIR and a bit more complicated)

You should look into the purcell factor, and you might understand what happens when a dipole gets close to a dielectric medium interface/wall, this is quite wavelength dependent as well. (distance is relatively different for varying wavelengths)