Smart Readout of Nondestructive Image Sensors with Single Photon-Electron Sensitivity

Image sensors with nondestructive charge readout provide single-photon or single-electron sensitivity, but at the cost of long readout times. We present a smart readout technique to allow the use of these sensors in visible light and other applications that require faster readout times. The method optimizes the readout noise and time by changing the number of times pixels are read out either statically, by defining an arbitrary number of regions of interest in the array, or dynamically, depending on the charge or energy of interest in the pixel. This technique is tested in a Skipper CCD showing that it is possible to obtain deep subelectron noise, and therefore, high resolution of quantized charge, while dynamically changing the readout noise of the sensor. These faster, low noise readout techniques show that the skipper CCD is a competitive technology even where other technologies such as electron multiplier charge coupled devices, silicon photo multipliers, etc. are currently used. This technique could allow skipper CCDs to benefit new astronomical instruments, quantum imaging, exoplanet search and study, and quantum metrology.

Figure 1

(a) Conceptual design; (b) number of samples per pixel $N$ for the same fractional contribution from readout noise to photon uncertainty as a function of $s_f$; assuming $k=0.95$, $\mathrm{QE}=0.9$, and ${\sigma }_0=2$ electrons, or, equivalently, photons.

Figure 2

Baseline of raw video signal in volts. Part of one row, the first 3039 pixels use $N=1$ and the others $N=100$. At the top, insets illustrate clock sequences and their voltage swings. At the bottom, enlarged regions of the baseline corresponding to the samples of a single pixel read with $N=1$ at $t\approx 0.05$ and with $N=100$ at $t\approx 0.215$.

Figure 3

Measurement using ROI technique. Pixels in the words have $N=500$ (right scale); pixels outside the words have $N=1$ (left scale). $s_f$ was zero in most pixels, with some pixels having $s_f=1$, 2, 3 or very large values for the two muon tracks that are observed.

Figure 4

(Top) Image using EOI technique. (Bottom) $N$ for each pixel.

Figure 5

Pixel histograms for EOI technique. $N=500$ for charge ranges $0e^{-}$ to $42e^{-}$ (top) and $15e^{-}$ to $19e^{-}$ (bottom).

Figure 6

Experiment combining ROI and EOI. Two color scales: on the left pixels with $N=1$ and a noise of $3e^{-}$, and on the right pixels with $N=500$ and a readout noise of $0.13e^{-}$ (and therefore quantized charge).