In recent years a growing number of applications demands always better timing resolution for Single Photon
Avalanche Diodes. The challenge is pursuing the improved timing resolution without impairing the other device
characteristics such as quantum efficiency and dark counts. This task requires a clear understanding of the
physical mechanisms necessary to drive the device engineering process.
Past studies state that in Si-SPADs the avalanche injection position statistics is the main contribution to the
photon-timing jitter. However, in recent re-engineered devices, this assumption is questioned. For the purpose
of assessing for good this contribution we developed an experimental setup in order to characterize the photontiming
jitter as a function of the injection position by means of TCSPC measurements with a laser focused
on the device active area. Results confirmed not only that the injection position is not the main contribution
to the photon-timing jitter but also evidenced a radial dependence never observed before. Furthermore we
found a relation between the photon-timing jitter and the specific resistance of the devices. To characterize
the resistances we studied the avalanche current density distribution in the device active area by imaging the
photo-luminescence due to hot-carrier emission.
In this paper we present a physically-based model aimed at calculating the Photon Detection Efficiency (PDE) and the
temporal response of a Single-Photon Avalanche Diode (SPAD) with a given structure. In order to calculate these
quantities, it is necessary to evaluate both the probability and the delay with which a photon impinging on the detector
area triggers an avalanche. Three tasks are sequentially performed: as a first step, the electron-hole generation profile
along the device is calculated according to the silicon absorption coefficient at the considered wavelength; successively,
temporal evolution of the carriers distribution along the device is calculated by solving drift diffusion equations; finally,
the avalanche triggering probability is calculated as a function of the photon absorption point.
Validation of the model has been carried out by comparing simulation and experimental results of a few generations of
detectors previously realized in our laboratory. Photon detection efficiency has been measured and calculated for
wavelengths ranging from 400nm to 1000nm and for excess bias voltages ranging from 2 to 8V. Similarly, temporal
response has been investigated at two different wavelengths (520 and 820nm). A remarkable agreement between
experimental and simulation results has been obtained in the entire characterization domain simply starting from the
measured doping profile and without the need of any fitting parameter. Consequently, we think that this model will be a
valuable tool for the development of new detectors with improved performances.
Improving SPAD performances, such as dark count rate and quantum efficiency, without degrading the photontiming
jitter is a challenging task that requires a clear understanding of the physical mechanisms involved. In this
paper we investigate the contribution of the avalanche buildup statistics and the lateral avalanche propagation to
the photon-timing jitter in silicon SPAD devices. Recent works on the buildup statistics focused on the uniform
electric field case, however these results can not be applied to Si SPAD devices in which field profile is far from
constant. We developed a 1-D Monte Carlo (MC) simulator using the real non-uniform field profiles derived
from Secondary Ion Mass Spectroscopy (SIMS) measurements. Local and non-local models for impact ionization
phenomena were considered. The obtained results, in particular the mean multiplication rate and jitter of the
buildup filament, allowed us to simulate the statistical spread of the avalanche current on the device active area.
We included space charge effects and a detailed lumped model for the external electronics and parasitics.
We found that, in agreement with some experimental evidences, the avalanche buildup contribution to the total
timing jitter is non-negligible in our devices. Moreover the lateral propagation gives an additional contribution
that can explain the increasing trend of the photon-timing jitter with the comparator threshold.