Quantum Entanglement, Dark Counts, Coincidence Detection

[caraher]

34. I'm slowly figuring out what you're doing...

First, I should mention that the coincidence circuit isn't such a big deal these days. Mine cost a few hundred bucks; I'm using the FPGA setup Mark Beck describes based on the Altera DE2 development board. You can get a coincidence window under 10 ns wide with this unit. You probably don't want to use LabVIEW unless you have a ton of money or an educational affiliation (or, of course, already have it), but all you really need to use the DE2 is a serial connection and the correct command strings.

One downside of fiber is that coupling to the fiber tends to be lossy, so if you only want to sample a tiny part of your interference pattern the bare detector might be better. You really can't get around the need to filter, if only because your laser is bright and the detectors burn themselves out at count rates above 15 million counts per second - easily achieved with room light for a bare detector! (That's another advantage of the fiber - the small acceptance angle of the fiber tends to protect a bit better against such catastrophes.)

Injection into the fiber is mainly a matter of focusing the beam onto the fiber's entry port. How you focus will affect the efficiency of the coupling; you'll want a fiber coupler on a mirror mount to make small angle adjustments. If you establish the "flight path" of your photons with a regular laser beam you can do much of the adjustment "by eye" just looking at the light coming out of the fiber; when it looks good, you put the fiber on your detector.

Buy some green LED lighting. This will let you run your experiment with the detectors on and also be able to see what you're doing. You should have some filters to block anything under maybe 750 nm in any case, and since your eyes are most sensitive to green light you get the most ability to see with the fewest photons (and ~530 nm is below the detectors' peak sensitivity).

I just skimmed the progress reports on Cramer's web site as well as a few recent popular articles. It sounds like he's really made some things more difficult than he needed to by cobbling together his own detectors. You can buy APDs for much less money than SPCMs, but the latter are engineered to take out all the hassles he went through.

It's not clear what laser he's using now from what I've read; the Sacher system was probably overkill. You do want the wavelength to be stable, but it probably doesn't need to be 405 nm on the nose - there's no particular physics selecting that wavelength. The main reasons to have exactly that value is that the commercially-available filters will be centered on wavelengths like 800 nm, 810 nm, etc. Current, temperature and optical feedback all affect the lasing wavelength, and his Sacher system had all three. Usually grating feedback is important only if you're trying to excite a particular atomic transition, so all you really need is a steady current source and either a big heat sink or active temperature control (which is what you don't typically get with a $50 eBay laser pointer).

For wavelength measurement in a DIY experiment, I would recommend making a spectrometer with a diffraction grating and calibrating it by comparing with known reference lines from a mercury lamp (or a neon discharge tube). This could also be good practice using optical fibers; fiber-coupling the light to a spectrometer will help ensure a consistent illumination of your grating. Or you might be able to borrow one from a chemistry department somewhere for a one-time calibration of a homemade setup. You could use a double slit (or for that matter, a DVD or CD) to do the measurement, but you also need to estimate the uncertainty in your value. If you measure 407 +/- 5 nm is that good or bad?

Half-wave plates are generally used to rotate linear polarization. So if you want to pump a pair of BBO crystals with light polarized at 45 degrees, you set a half-wave plate at 22.5 degrees between the pump laser and the crystals. They work with birefringence - the polarization along a particular axis passes through the crystal faster than the perpendicular polarization. It's often important to use what's called a "zero order" waveplate, which generally have better performance, less wavelength sensitivity, and, most important of all, shift the two polarizations by exactly one-half wave relative to one another. ("Multiple order" waveplates will also rotate polarization, but advance one polarization relative to the other by n+1/2 waves, which can be a problem when you're trying to overlap single photons in time, for instance!)