Polarization send–return separation
A polarizing beamsplitter and a quarter-wave plate send light out and take the reflection off at a different port, without the 75% loss of a plain beamsplitter.
Open in the canvas →Background
Many instruments send a beam out and want the light that comes back along the same line: a disc reader, a reflectance microscope, a displacement interferometer, a double-passed modulator. A plain 50:50 beamsplitter does it, but wastefully — half the light is lost on the way out and half of the return on the way back, so at best a quarter of the power reaches the detector, and another quarter goes straight back into the laser.
Polarization does it without the loss. A polarizing beamsplitter (PBS) transmits one linear polarization and reflects the orthogonal one. Send the transmitted polarization through a quarter-wave plate with its axis at 45° and it becomes circular. On reflection the handedness reverses, and the second pass through the same plate turns it back into linear light — but rotated by 90°. Two passes through a quarter-wave plate are one pass through a half-wave plate[1]. The return is now the polarization the PBS reflects, so it leaves by the side port instead of retracing its way to the source.
The arrangement is the standard way to double-pass an acousto-optic modulator, where the return has to be separated from an input it exactly overlaps[2], and it is the reason a quarter-wave plate sits in front of the objective in an optical pickup.
It is not an optical isolator, although it is often used as one. It works only if the target hands the polarization back as it received it. A surface that depolarizes, or anything birefringent in the double-passed path, returns some light in the original polarization, and that part goes back to the laser. A Faraday isolator rotates the polarization non-reciprocally and blocks the return whatever happened to it on the way[3].
What this setup demonstrates
A horizontally polarized 532 nm laser passes a half-wave plate, a polarizing beamsplitter and a quarter-wave plate at 45°, and a lens focuses it on a mirror standing in for the target. The return is recollimated by the same lens, crosses the quarter-wave plate a second time, and the PBS reflects all of it to the return port, which reads 1.
Things to try. Rotate the quarter-wave plate to 0°: its axis is now along the polarization, it does nothing, and the whole return retraces to the laser — the return port goes dark. At 22.5° the split is half and half. That curve is what an alignment walks up: maximise the side port and the plate is at 45°.
The half-wave plate in front is the other common use of the same cube. Rotating it turns the input polarization, the PBS sends the unwanted part to the beam dump on the opposite side, and what is left goes out: a continuously variable attenuator. At 22.5° half the power is sent, and the return port reads 0.5; at 45° nothing is sent at all. Note that the rejected input and the return leave by opposite faces of the cube, because they arrive from opposite directions.
The polarization is carried as a Stokes vector through ideal components: the PBS has perfect extinction, the wave plates have exactly their nominal retardance at this wavelength, and the mirror reflects without changing the polarization state other than by reversing the direction of travel. A real cube leaks about a part in a thousand, a real plate is a quarter wave at one wavelength and one angle only, and a real target may depolarize — each of these sends some of the return back to the source, and none is modelled.
There is no interference between the outgoing and returning beams, no standing wave, and no feedback into the laser: light that retraces to the source simply ends there.
Related components
References
- RP Photonics Encyclopedia — Waveplates
- E. A. Donley, T. P. Heavner, F. Levi, M. O. Tataw, S. R. Jefferts, “Double-pass acousto-optic modulator system,” Review of Scientific Instruments 76, 063112 (2005)
- RP Photonics Encyclopedia — Faraday Isolators