IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 23, NO. 18, SEPTEMBER 15, 2011 1319 Avalanche Current Measurements in SPADs by Means of Hot-Carrier Luminescence Antonino Ingargiola, Mattia Assanelli, Ivan Rech, Member, IEEE, Angelo Gulinatti, Member, IEEE, and Massimo Ghioni, Senior Member, IEEE Abstract—A growing number of applications require arrays of single-photon avalanche diode (SPAD) detectors with low timing jitter. In order to improve jitter without compromising other performance parameters, a clear understanding of avalanche dynamics and statistics is necessary. In this work, a noninvasive electro-luminescence technique has been employed to investigate the current growing in a SPAD device. The obtained results let us assess, for the first time experimentally, the avalanche spreading speeds and also confirmed our assumption that the current growth pace and its statistics critically depend on the space charge effects. Index Terms—Electro-luminescence, photon timing, single- photon avalanche diode (SPAD), time-correlated single-photon counting (TCSPC). I. INTRODUCTION S INGLE photon avalanche diodes (SPADs) have nowadays gained wide acceptance as solid state alternatives to Pho- tomultiplier Tubes (PMTs) in particular in TCSPC applications. SPADs not only provide the typical advantages of solid state devices compared to vacuum tubes, but they also offer higher photon detection efficiency. In order to detect the incident photons with the lowest timing jitter, a suitable current pickup circuit [2] that senses the rising edge of the avalanche is employed. By using this technique, photon-timing jitter down to 35 ps Full Width at Half Maximum (FWHM) can be obtained even in large area SPAD detectors (with diameters up to 200 m [3]). These results can be achieved employing a very low threshold in the current pickup circuit in order to detect the avalanche when it is still confined within a small volume around the seed point. In Fig. 1 we report a comparison of the photon-timing jitter performances obtained with three different device families: two full custom devices (S44 and S62) and a SPAD made in stan- dard CMOS technology, called SCMOS 1 . It is worth noting that different device families present a different photon-timing jitter dependence on the avalanche discriminating threshold. In the last years a growing interest has been observed in SPAD-arrays for a variety of applications ranging from parallel fluorescence correlation spectroscopy [4] to spectrally resolved Manuscript received March 31, 2011; revised June 06, 2011; accepted June 18, 2011. Date of publication June 23, 2011; date of current ver- sion August 26, 2011. This work was supported by EC grant agreements n.248095 (Q-ESSENCE) FP7-ICT-2009-4 and n.232359 (PARAFLUO) FP7-SME-2008-1. The authors are with Dipartimento di Elettronica e Informazione—Politec- nico di Milano, 32 Milan, Italy. Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LPT.2011.2160533 1 For more details about the studied devices see [1]. Fig. 1. Photon-timing jitter as a function of the avalanche discrimination threshold obtained with three different kinds of devices. Red solid line: SCMOS device. Green dashed line: S44 device. Blue dashed–dotted line: S62 device. All the devices have the same diameter: 50 m. Fluorescence Lifetime Imaging Microscopy (FLIM) and to 3-D imaging with subcentimeter depth resolution [5]. In some SPAD-arrays the need arises to operate with high threshold voltages in order to avoid electrical crosstalk between devices. In these applications the degradation of the photon-timing per- formances at high thresholds should be small. SCMOS device (Fig. 1) has this attractive feature but lacks of others significant characteristics. In fact, if compared to custom devices, SCMOS device is characterized by a higher after-pulsing probability and a smaller photon detection efficiency. Beside this, dark counts are higher and can not be efficiently decreased with the reduction of the temperature. Therefore, a further evolution of SPAD technology requires improving the photon-timing jitter without sacrificing the other performances. For this purpose it is mandatory to discover the key element that determines such a weak dependence of the photon timing jitter on the sensing threshold in SCMOS de- vice. So far the main statistical phenomena that afflict the cur- rent growing in a device are not totally clear. Nonetheless, in our previous work [1] we presented a figure of merit PT that evidenced the importance of the space charge resistance and the buildup current growth time constants in determining the timing performances of a device. In particular we were able to ascribe the good timing performances of SCMOS device to its very low space charge resistance: 35 k m compared to a value higher than 100 k m for both S44 and S62 devices. Such a small space charge resistance results in a faster growing avalanche in particular in the first instants after avalanche injec- tion. The knowledge of these ultrafast phenomena derive from indirect observations and simulations. In this work we want to experimentally evaluate the current in SPAD devices by means of a noninvasive electro-luminescence characterization. 1041-1135/$26.00 © 2011 IEEE