Thursday, October 23, 2008

RIT Team to Develop Photon Counting Sensor

Democrat and Chronicle: Donald Figer and his colleagues at Rochester Institute of Technology got $2.8M grant to develop a light detector that would count individual photons for astronomy applications.

Current technology uses sensors that can be triggered by electronic "noise" within the electronic device itself, resulting in a grainy, speckled image, especially in low-light conditions.

"In our detector we're doing something radically different. Each photon of light is being counted," said Figer, director of the Rochester Imaging Detector Laboratory at RIT's Chester F. Carlson Center for Imaging Science.


  1. a couple of notables
    "Not only will Figer's team have to tinker with the basic engineering of the prototype circuits so they can detect a single photon, but they must stuff those circuits into a ridiculously small space — eventually a 20-by-20-micron square pixel. A micron is one millionth of a meter."

    "But there are still quite a few important astronomy problems involving extremely low background-light levels, for which the devices Don describes would be a healthy improvement over the state of the art," Watson said."

    ? How exactly can you imrove on 1 -2 e electron detection capability that has been present in scientific CCD's (cooled) for years now?

  2. Well, let me try to defend Figer's concept. Once we talk about 1-2e noise, there usually rms noise. Figer promises photon counting, which means if you get 1 count, it means 1 photon has detected with zero noise. Among other things it means close to 100% QE, otherwise there is no sense in zero noise electronics.

  3. Fair enough, but shot noise would be dominant so it would be very hard to seperate out this since in a scientific, cooled CCD the general rule of thumb is that the readout noise has to be significantly below the noise of converting a single electron. Scientific CCD's of this ilk are also thinned and backside illimuniated, so FF is a moot point. I'd not necessarily assume that he'd get 100% QE becasue these types of devices typically work on coloumb blocking effects, which means it works on electrons, there is still the photon to electron conversion that hasn't been discussed. The main thing this MIGHT bring to the table is better MTF vs. a CCD. But next generation CMOS sensors will be in this regime sooner (i.e. single electron, good MTF).

    Granted, if this device works on plasmonics, then the QE will be better. so an improvement of 25%? But way better in the blue.

  4. I understand what you are saying, but I meant something different.
    The readout and shot noises are additive, while photon counter noise is multiplicative. This might or might not be an advantage, depending on application.
    Lat's assume a cooled CCD has 1e of RMS readout noise with gaussian distribution. It means that about 3 pixels out of every thousand have noise of 3e in every single frame, right? Now, suppose the sensor integrates an image of a weak star, which has a total integrated signal of 3e. Then 1MP CCD would have 3,000 "false stars".

    Now, let's look on the photon counter. If it really counts photons, its noise is multiplicative: when there is no signal, there is no noise. In that case 3e star would be visible, albeit with large shot noise of sqrt(3).

    So, photon counter can have some useful properties putting it apart of low-noise CCDs.

  5. One correction: shot noise is multiplicative, in a sense that if there is no signal, the noise is zero. Ideal photon counter has a shot noise, but no additive readout noise, while CCDs have both.

  6. Anyone know what their technical approach is?

    -Detector material?
    -Detector structure?
    -Gain mechanism?

    Seems like the issues of QE, noise and false counts hinges on these details. For all I know from the article they are building an ultra low noise CCD or CMOS image sensor with very high conversion gain (e.g., 10,000 uV/e-).

  7. As for me, I have no idea how their photon counter works. My guess is that they are designing some kind of squelching avalanche multiplier.


All comments are moderated to avoid spam.