TE and TM in MZ interferometer


Hi everyone:

I want to get familiar with the elements in the INTERCONNECT library, and I built a MZ interferometer. But there are some problems confused me.

In one arm of the MZ interferometer, I added a phase shifter (phase angle=0rad) and a polarization rotator (angle=pi/2) which change the TE mode into TM mode. For convenience, I wrote the OOSC results in the figure. I found that there is no TM mode at the output.

To my knowledge, the TE mode and TM mode cannot interfere so the intensity (or optical power) should be the sum of the two modes. There should be 2.5W TE mode and 2.5W TM mode at each output ports. But the simulation shows no TM mode. Please help me find my wrong settings!

I have tried to disconnect one line, as shown below, and the TM mode appears!!!

I upload the INTERCONNECT file and the SCRIPT file here.
mz.icp (424.2 KB)
mz.lsf (590 Bytes)

Thank you very much!



Please look into the example for MZ interferometer:




In the link below it describes how to simulate the electro-optical behavior of a Mach-Zehnder interferometer (MZI) fabricated with a silicon-on-insulator (SOI) process using the multi-physics simulation environment of DEVICE. In particular, we will calculate the spatial carrier concentration as a function of applied voltage using the CHARGE solver in DEVICE. From the carrier concentration, we can estimate the device capacitance. The carrier concentration can also be exported to the FEEM solver in DEVICE where we can calculate how the optical properties of the waveguide change as a function of applied voltage.


I would suggest that you could add a Y-branch element. The Y Branch adds up the electrical fields but not the intensity of the signals (based on the coupling coefficients), and the equation is (E1+E2)/sqrt(2)The example file.optical_y_branch.icp (263.7 KB) has some optical Oscilloscopes connected to the lasers and to the output of the Y Branch and measure the complex transmissions of the signals.

If you look at the complex transmission real and imaginary parts of the signals, they all follow the equation (E1+E2)/sqrt(2).
Hope this helps!


Thank you for your reply.
I have tested the attached files and got the same results in the pictures.
I have three questions.

(Q1) It seems that the two CW Lasers are coherent perfect, i.e., the amplitude of the E filed is added together rather than the power of them when two beams are superposition at the output port of the Y branch. But to my knowledge, two lasers cannot interferes, their power added rather than the E filed amplitude. I am confused about this.
e.g. The two CW Lasers are both 3W. At the output port of the Y branch, the power should be 3W if the power is added, since the Y branch introduces 3dB loss for each input light.

(Q2) I am looking for an optical component which can add TE and TM together. I tested the “polarization combiner”, it can add TE and TM correctly.

But waveguide couplers and Y branches don’t do this correctly as shown below. The simulation file optical_y_branch2.icp (1.3 MB)
Does either the coupler and the Y branch can deal with only one polarization mode?

(Q3) How to simulate the polarization cross-coupling in optical fiber? I plan to first separate the TE and TM modes, then rotate their polarization, then combine them together. Which component can I use for the “?”
Is there a simple way to do the simulation?


Hi @zhanghaisheng,

Thank you for all the testings. Indeed it seems like the couplers and y branches don’t combine the polarization. I will confirm on this and get back to you. You can put another polarization combiner for the ? place and it should combine the signals fine. Here is an example file: pol_test.icp (4.5 MB)


Thank you for your reply.
I have looked into the “pol_test.icp” in your relpy. But I found that the element - polarization splitter and combiner - work as a mode filter. Only TE mode can pass through the port 1 of the polarization combiner, and only TM mode can pass through the port 2 of the polarization combiner. The TM component at the port 1 is filtered out. So, I am afraid that a polarization combiner can not realize combining two lights, both of which have TE and TM modes.


Hi @zhanghaisheng,

I few change the combiner “angle” to pi/4, the combiner will pass both modes for all the branches. However, in the example we have, the TM modes pass the upper branch and the lower branch are in opposite phase, so they added up destructed. I changed the optical oscilloscopes “plot format” to “rectangular” so that they measures the phase information. The modified file is here: pol_test_modified.icp (393.2 KB)

Upon your request, I think we can make a compact model to build the polarization combiner/splitter you want. Please let me know whether this could be an option?