Modeling the PCS in ZEMAX is relatively simple. The rhomb may be modeled using a polygon object or a rectangular volume object. Here I choose to use the polygon object.

The polarizing beamsplitter coating may be designed in a thin-film package and exported into ZEMAX. ZEMAX can read the prescription data of any thin-film coating and accurately model transmission, reflection, absorption, retardance, diattenuation etc. However coating manufacturers are often reluctant to give out prescription information, and so ZEMAX supports a TABLE coating, which is a tabulated listing of coating peformance, and an IDEAL coating which simply lets us tell ZEMAX what we want it to do. In this example, I will use two IDEAL coatings.

The first will be for the broadband anti-reflection coating that is used to maximize transmission of the optics. This will be an ideal, 100% transmissive coating. The second will be the polarizing beamsplitter coating, which will be set to reflect the S polarization with 100% efficiency and to transmit the P polarization with 100% efficiency. Remember that ZEMAX is capable of modeling the real coating prescription, and therefore of giving a much more accurate and detailed treatment of coating performance than I will give here.

The following lines are added to the coating file used by ZEMAX:

! Define some idealized coatings for Polarization Conversion System
! First define a 100% transmissive coating
! Format: IDEAL <TRANSMITTED intensity> <REFLECTED intensity>

IDEAL transmit 1.0 0.0

! Now get 100% reflection, but only for the S
! The P should see 100% transmission 
! Use IDEAL2 coating. Defines reflected and transmitted 
! complex amplitude for S and P, rest is assumed to be absorbed
! Format: IDEAL2 s_rr s_ri s_tr s_ti p_rr p_ri p_tr p_ti no_pi_flag

IDEAL2 S_Reflect 1 0 0 0 0 0 1 0 0

The S_Reflect coating is then placed on the interface between the two prisms. How is this done? Remember that when two objects touch, as these two polygon objects do, the optical properties of the interface are set by whichever object is last in the Non-Sequential Component Editor. I then put the transmit coating on all other faces of the optics. I set a source object to have Jx = 1, Jy =1 so that there are equal amounts of light polarized in x and y, and then trace rays accounting for polarization and splitting:

The two beams have equal energy (the lower one has traveled slightly further and so has expanded more, hence its irradiance is lower). Now, Detector objects have a Polarization property that allws them to detect only polarized light. Setting this property to 0 means the detector detects all light, irrespective of polarization, as in the screenshot above. Setting it to 1 means the detector sees only the x-polarization, and 2 only the y-polarization. With the polarization control set to 2, so that only y-polarized light is detected, we get:

Adding a Jones Matrix object and setting it to be a half-wave plate in y gives:

and we now have 100% energy transmitted and a 100% P-polarized beam. The Jones matrix object is shown as a red circle in the screenshot, for clarity. The real half-waveplate is a rectangle which is the same size as the exit face of the bottom rhomb. A user-defined aperture can be added to the object to make it the desired shape.