Science
In reply to the discussion: Quantum Entanglement, Dark Counts, Coincidence Detection [View all]caraher
(6,278 posts)but I'm kind of thinking about my own more intensively right now
In my current experiment both detectors are inside a light-tight box whose only entry point is through a longpass filter. I'm using the ThorLabs FELH0750, which is $150 but has MUCH higher transmission than their regular filters (97% at 810 nm; you can download an Excel file with the specs from their site). I generally also have a bandpass filter in front of that for a variety of reasons, but I've run my system with just the longpass filter without harming anything. (I also have a blue-reflecting dichroic mirror at 45 degrees after my downconversion crystal, so there are two elements that provide big losses to my pump beam before the detectors. But my experiment is collinear Type II downconversion, so I need to work harder to eliminate the pump. If you're doing Type I with the 3-degree cone the pump beam is spatially separated from the downconverted photons.)
By extension, I'm thinking about what Cramer is up to as well. It seems to me that the kind of interferometry he is using (and perhaps you) makes the experiment harder than it needs to be. Cramer seems determined to work with photon momentum, which is nice for making the connection to the Young double slit experiment, but there are so many other properties to work with. The reason polarization is used so much is that it's comparatively easy to manipulate and you can often use "bucket" detectors rather than fiddle with anything highly position-sensitive. I also do now have a better appreciation of why Cramer is having so much trouble with dark counts - he's set up an experiment with a very small signal.
Schematically, if I understand what you (Cramer and you) want to do, it's essentially a more elaborate version of a delayed choice experiment. Your effort to send "retrocausal" signals rests on somehow using the choice of measurement to send a signal that affects a previous measurement (or at least a spacelike-separated measurement), right? If I were trying to do that I think I'd go for something like a variant of Aspect's implementation of delayed choice. I'm not sure what exactly the best thing to measure would be, but I'm thinking you'd want a setup where the choice of which measurement you make on one element of the biphoton system affects the behavior of the other element of the system. So, for instance, if you feed the signal into something like Aspect's setup, you'd need to set up a measurement of some property of the idler such that the subsequent measurement of the signal "tells" the idler what the choice was, but in a way that avoids the possibility of the causal "story" working in the other time direction (i.e. the measurement of the idler doesn't somehow "drive" the later choice of signal measurement). I'd be surprised if a two-slit welcher weg ("which path" or "which way" measurement had any unique properties that make it the measurement of choice here.
One more thing you should think about before paying Newlight for BBO - what kind of entanglement do you really need? The 2-crystal Type I setup is designed to create polarization entanglement, but it sounds like the Cramer Mach-Zehnder with the D-shaped mirror on the input is really all about momentum entanglement. If you really don't care about polarization, you might do better with a single BBO crystal (still cut for Type I phase matching). This also lets you filter out stray pump photons with a polarizer (since in Type I the downconverted photons have the same polarization, which is perpendicular to the polarization of the pump beam). You also don't need a phase shifter before the BBO if you're just pumping one crystal (that's necessary to get polarization entanglement).