Birefringent components fall into two main classes: beamsplitters and waveplates. A beamsplitter splits the beam into orthogonally polarized components, and a waveplate rotates the polarization of the beam. Beamsplitters give two beams out for one beam in, waveplates give one beam out for one beam in.

These two component classes are modeled using the mode flag of the "Birefringent In" surface. Modes 0 and 1 are used to trace the ordinary or extraordinary ray in a beamsplitter. Because the input ray splits into these two components, we must use two configurations, one for the o-ray and one for the e-ray. The number of configurations needed goes as 2n, where n is the number of birefringent components. To compute the transmission and extinction of the beamsplitter, we need to add the field amplitudes, not intensities, of each configuration. We give a full discussion in the "How to design a birefringent polarizer" article, including a discussion of the transmission/extinction calculation in section Calculating Extinction Ratio.

For waveplates, one needs only a single configuration, because the beam is not split. We use modes 2 and 3 to model the input beam as either the o- or e-ray, and we rotate the polarization vector of the ray. In the attached sample file, we trace the ordinary ray through a thin quartz waveplate. Try varying the thickness of the waveplate, and note how the retardation changes.