SPAD Quantum Camera: The Owner's Manual

For those of you following my travails in quantum information processing, our most recent work just appeared in the prestigious open-access journal Optics Express, published by the Optical Society of America, under the title: "On The Application Of A Monolithic Array For Detecting Intensity-Correlated Photons Emitted By Different Source Types." (PDF)

I gave a brief account of the motivation for our new quantum imaging device back in January of this year, which led to it getting slash-dotted. The current paper is a kind of compendium of all the prats and pitfalls we ran into during the course of doing various quantum imaging experiments with this new device.


This computer-generated image gives an impression of what our camera "sees" looking at correlated photons emitted from the end of the glass fiber described in our paper. The concentric rings correspond to weaker correlation maxima. The scale is in microns and inner grid corresponds to the 4×4 pixel array of our prototype SPAD (single-photon avalanche diode) imager.

When I wrote this sentence for the Abstract,

It is not widely appreciated that many subtleties are involved in the accurate measurement of intensity-correlated photons; even for the original experiments of Hanbury Brown and Twiss (HBT)...

I wasn't being modest. The principal of detecting correlated photons, is more than 50 years old (long before SPADs), and if you read either the original papers by Robert Hanbury Brown (all one person) and Richard Twiss or the HBT concept in physics textbooks, it looks very straightforward to repeat—especially with the advent of modern equipment. In fact, one paper by Hanbury Brown and Twiss describes how they did an experiment quite literally on a kitchen table. This all turns out to be rather misleading. Although none of the literature is written with any malice in mind, one is left with the impression that it is easier to do than it really is.

For example, the original HBT table-top experiment only required measuring a single data point to establish proof-of-concept. We, on the other hand, had to characterize an entire curve—the exact shape of which we didn't know beforehand. This demands sampling 1000s of data points. Combined with various measurement inefficiencies, the detection of correlated photons is on the order of a "million to one shot (Doc)" and thus, it takes several weeks for sufficient data acquisition (at 100 picosecond resolution) to generate an image like the one above.

As we learned in the process of carrying out the measurements discussed in our paper, this kind of physics is no stroll in the park and definitely not for those who need instant gratification.