In my last post, I scrounged the parts for a very crude, but very cool, experiment you can do in your basement to demonstrate quantum entanglement. To my knowledge, it’s the cheapest and simplest such experiment ever done. It doesn’t give publishable results, but, to appropriate a line from Samuel Johnson, a homebrew entanglement experiment is “like a dog’s walking on his hinder legs. It is not done well; but you are surprised to find it done at all.”


As a warm-up exercise, I sandwich my source of entangled photons—a disk of radioactive sodium-22—between my two Geiger counters (see diagram and photo below) and leave the system to run overnight, measuring how often the Geigers click at the same time. If gamma-ray photons are indeed emerging two by two in opposite directions, the coincidence rate should vary strongly when I change the alignment of the two Geigers. And that is what I see.


When the Geigers are pointing straight at each other, each clicks about 900 times per minute and both do so in unison about 4 times per minute. This is about 40% greater than the expected rate of accidental coincidences. There are various subtleties in separating accidental and genuine coincidence rates and in estimating statistical errors, but the signal I observe is something like 10 standard deviations above the noise. When I rotate one of the Geigers out of alignment, the coincidence rate drops precipitously. For a 25° angle, it only about 15% greater than the accidental rate, which is still statistically significant, if barely. For 45° and 90°, it is equal to the expected accidental rate. So I can tentatively conclude I’m seeing pairs of gammas—one or two of them per minute! This is no mean accomplishment given how crude the equipment is.


Just because the gammas emerge in pairs doesn’t mean they are entangled, though. To check for entanglement, I measure the photons’ polarization with a technique called Compton polarimetry. A pair of aluminum cubes bought at serve as gamma-ray prisms, scattering photons in directions that depend on their polarization. The two gammas produced by the annihilation of an antielectron and electron are linearly polarized at right angles to each other, so they should scatter off the aluminum in perpendicular directions.


Here’s where the physics gets spooky. Each individual photon scatters in a random direction, yet the random direction one photon takes is related to the random direction its partner does. The gammas act in synchrony. How can they do that, if they’re truly random? Einstein concluded that the photons either are not truly random or are acting on each other at a distance.


In a first attempt to observe this effect, I sandwich the sodium-22 disk in between the two cubes and put a Geiger on one face of each cube (see photo below). I start by pointing the Geigers in the same direction and letting them sit overnight to count the coincidences. In the morning, I move one Geiger to a different face of its cube, so that the two detectors are now perpendicular to the other, and leave the system to run all day. I continue cycling through different ways to align the detectors either parallel or perpendicular to each other. Entanglement should betray itself as an asymmetry in the coincidence rate.


And indeed that’s what I see. About one coincidence occurs per minute on average, and the rate is consistently greater when the Geigers are perpendicular. It looks like entanglement in action!


A wise graduate student would hesitate to show this result to his or her faculty advisor, though. The perpendicular rate stands a couple of standard deviations above the expected accidental-coincidence rate, but the parallel rate swims in the noise. So the asymmetry might well be a fluke of statistics or a subtle bias in the setup.


To improve on the experiment, I need to beat down the accidental rate—in particular, the rate caused by gammas traveling straight from the sodium to the Geiger counter rather than scattering off the aluminum. I enclose the radioactive sodium in a so-called collimator: a lead storage canister in which I drilled a 1/2-inch hole at either end. A couple of hundred gammas per minute leak out through each hole, forming a pair of gamma-ray beams. The lead squelches off-axis radiation by a factor of about four.


With the collimator, the coincidence rate drops by a factor of 10, but now exceeds the predicted accidental rate for both orientations. The perpendicular rate is the higher of the two, again as the Compton-polarimetry theory predicts for entangled photons.


This still isn’t anything to call the Nobel committee about. At best, it implies the detection of one entangled pair of photons every 20 minutes, and with such a meager trickle, who knows what subtle bias might be operating. What was iffy for the pioneering Bleuler and Bradt experiment can only be more so for my apparatus. Then again, all I’m seeking is a suggestive demonstration, not a research-grade system.


A possible next step would be to special-order a stronger sodium-22 source, which would bring the particle rates in my experiment up to the level of Bleuler and Bradt’s, at the price of posing a greater radiation hazard. Another idea would be to try scatterers besides aluminum cubes. Beyond that, however, I think you exhaust the el-cheapo options and have to dig deeper into your wallet, starting with replacing the Geigers counters with scintillation counters, as Wu and Shaknov used. These are more efficient at picking up radiation; create shorter electrical pulses for each particle they detect, which reduces the probability of accidental coincidences; and measure particle energy, which would help to sift out annihilation-produced photons. But such instruments are pricier and fussier.


A useful guide to further refinements is Leonard Kaskay’s Ph.D. dissertation from 1972. A student of Wu, Kasday systematically went through the possible sources of error: multiple scattering, geometric misalignment, unwanted photons, and more. He was able to achieve enough precision to show that the gammas violated a mathematical inequality derived by theorist John S. Bell, confirming that he was seeing spooky action at a distance rather than some mundane effect.


These kinds of experiments are notoriously tricky, so please share your thoughts and advice—not to mention your attempts to reproduce! Wait till your friends hear that you’re an amateur quantum physicist in your spare time.

One thought on “How to Build Your Own Quantum Entanglement Experiment, Part 2 (of 2)

  1. Hi,
    at there is my replication of your quantum entanglement experiment. I also used a Na22 source (1uCi, rather weak but enough) and a couple of scintillation detectors (LYSO crystal + Silicon photomultiplier). The advantages of this type of detectors are the low price, the low voltage (with respect to traditional PMTs) and the simplified electronics, furthermore the timing is excellent (so the coincidence is more reliable) and the geometrical dimensions are ideal in order to have a good spatial resolution. I used big lead ingots as shields and iron scatterers instead of aluminium. A monte carlo simulation (performed by Prof John Clifford and Alexey) showed that Iron is a better scatterer than aluminium and this simulation gave me information about how to optimize the positions in order to maximize the counting rate.
    In order to have a small variance on the data it has been necessary to run the measure for quite a long time (several hours).
    The data are in good agreement with theory so we are quite confident – we hope so – to have a good DIY demonstration of spooky action !


Share your wisdom