Lately I have been investigating the physics literature on proposed photosensitive electronic devices made of graphene. Often I see responsivity values quoted as if they suppossed to impress, and there are even comparisons (e.g., "my 10 A/W detector is better than your 1 A/W detector"). An example (picked at random from a google search) is the abstract of this paper.

What I find strange is that responsivity seems, to me, to be completely arbitrary:

  • In photoconductive-mode detectors, responsivity depends entirely on the applied voltage bias.
  • Responsivity is often dependent on the device size.
  • By combining a photodetector with an amplifier, I can make any responsivity as desired.
  • Even if we are worried about detecting small currents, this also depends hugely on the impedance of the photodetector. Photodiode amplifiers for example can exploit the high impedance of the signal to amplify very tiny currents.

Am I misunderstanding something here? It seems like responsivity is basically meaningless, and so any claim of "high responsivity" is just pure fluff. Wouldn't a classic metric such as detectivity (comparing to noise level) make a far better figure of merit?

  • $\begingroup$ I don't know a thing about experimental physics, but this sounds like a really good question. +1 $\endgroup$
    – Danu
    Dec 16, 2014 at 20:55
  • 1
    $\begingroup$ This is a guess, not an answer, but I wouldn't be surprised if better responsivity has something to do with better noise performance. In a radio receiver, the noise performance of the entire system is almost entirely determined by the noise figure of the first amplifier stage connected to the antenna. Responsivity of a photodetector may be the optical equivalent. $\endgroup$
    – David Rose
    Dec 16, 2014 at 21:45
  • $\begingroup$ How many photons/s (at some wavelength) are hitting the detector for 1W of incident radiation? How many charge carriers are being generated per photon? If you think through these questions a bit, you will see why "responsivity" is a good figure of merit as long as we are NOT using amplified signals. In case of fundamentally amplifying detectors like photomultiplier tubes we need responsivity and quantum efficiency (plus a few minor pieces of information) to fully characterize the detector. $\endgroup$
    – CuriousOne
    Dec 17, 2014 at 2:56
  • $\begingroup$ Even then, responsivity is still not helping me out. In a photomultiplier the quantum efficiency can be arbitrarily high. For absorbed every photon, 1000 or 1000000 electrons may come out. Although this device indeed may be photon shot noise limited, it won't be electron shot noise limited. And hey, the detector may even have very poor performance in terms of noise equivalent power where only 1 in 10 photons actually generate signal, yet the gain can compensate to make an "amazing" responsivity and quantum efficiency. Or it may have an enormous level of dark counts. $\endgroup$
    – Nanite
    Dec 17, 2014 at 9:09
  • $\begingroup$ It is true that in some semiconductor detectors, such as ordinary type photodiodes with low dark current, the noise level is limited by electron shot noise of the signal. That seems to be the only case where the use of responsivity is justified as a figure of merit. But in a room temperature noisy detector based on a carbon resistor (graphene), noise levels are surely much much higher than shot noise. If you give me a detector that is 10x more responsive but 100x more noisy, certainly it has worse performance. $\endgroup$
    – Nanite
    Dec 17, 2014 at 9:20

1 Answer 1


I think it might be one of those things where people do something because everybody does.

I agree with you, a figure of merit that includes noise would make more sense. But, as the circuit designer that I am, I could also say that that wouldn't be the end of it.

For example, in the classic trans-impedance amplifier used for these kind of detectors the noise contributed by the op amp depends on the input capacitance. Two devices with the exact same characteristics but different capacitances will lead to different performances. With that in mind, a better figure of merit would be something like sensitivity/noise/parasitic_cap.

At this point you could also throw in area, power consumption, cost... The list of things you could add to a figure of merit are endless, so it's hard to pick one that makes everybody happy. Especially if you talk about devices that are still not completely understood and not widely used.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.