The current noise power spectral density (NPSD) of a silicon low-gain avalanche diode (LGAD) has been measured at room temperature under different bias conditions. The device has been biased in its whole operating range up to close the junction breakdown with dark currents between 100 pA and 3 nA. The NPSDs have been measured in a bandwidth from 10 Hz up to a maximum frequency ranging from 10 kHz up to 3 MHz, depending on the noise level. The NPSDs have been found white in all the operative conditions with no evidence of a 1/ f component. An empirical law is proposed to describe the NPSD with high accuracy over four orders of magnitude from 10 ^{$-$ 28} to 10 ^{$-$ 24} A ² /Hz. The excess noise factor $\zeta$ has been experimentally determined from the NPSD and accurate measurements of the multiplication gain M . A good agreement with McIntyre’s theory has been found only at higher gain values, but for M $<$ 30, the excess noise has been found to be accurately predicted by the M ⁿ law with n $=$ 0.632 $\pm$ 0.014. An extremely low value of $\zeta$ $=$ 1.06 $\pm$ 0.01 has been measured at M $=$ 5, opening perspectives to the development of ultralow-noise radiation and particle detecting systems for both timing and spectroscopy applications.

A Monte Carlo simulator is presented to evaluate the spectroscopic response of High-Z semiconductor pixel radiation detectors starting from first principles physical laws. It performs simulation in three different domains: electrostatic fields, photonmatter interaction, and photo-generated charge-carriers transport. A sampling algorithm is proposed to address the issue of fast and accurate computations of charge cloud dynamics in presence of both diffusion and electrostatic repulsion effects, avoiding direct numerical integration of the induced current equation stated by the Shockley-Ramo theorem under a constant electric field hypothesis. The simulator is written as an Object-Oriented Programming (OOP) source code repository, relying on the integration between MATLAB® and COMSOL® Multiphysics, and can be run with one executable script. The simulator architecture is presented, followed by a detailed explanation of all the physical models and implemented simulation strategies. The simulator is validated with experimental energy spectra acquired with a radiation detection system based on CdZnTe pixel detectors and ultra low-noise front-end electronics with state-of-the-art energy resolution.

We present the performance of a X-γ ray detection systems based on a 300 µm thick silicon low-gain avalanche diode (LGAD) and of an equivalent Diode structure without the gain layer, read-out by a custom-made low-noise charge amplifier. For the LGAD structure, the multiplication gains M_s from 10.2 to 19.3 are measured, and the Equivalent Noise Charge (ENC) components have been studied in detail. As expected, a significant reduction of the white and 1/f voltage series and dielectric ENC components is observed with increasing gain, shortening the optimum peaking time to τ = 0.4÷8 μs depending on the gain value, with respect to the standard diode, while the parallel ENC component increases proportionally to the gain up to M_s = 15. Setting the gain M_s = 12.7, a minimum ENC = 34.5 electrons r.m.s. (298 eV FWHM) at τ = 1 μs is found. However, the electronic noise, evaluated on the pulser FWHM, is found to only marginally contribute to the width of the spectral lines of a radioactive source, which are dominated instead by the statistical noise of the charge multiplication within the LGAD structure. The spectral lines' excess width is found to be proportional to the signal multiplication gain M_s and increases with the energy of the photons. A minimum line width of 850 eV FWHM at 13.9 keV is measured at room temperature, τ = 2 μs and a gain M_s = 10.2.

A wideband low-noise transimpedance amplifier (TIA) is designed for the measurement of the noise power spectral density (NPSD) of the current of low-gain avalanche diodes (LGADs). The design focuses on minimization of system background noise and maximization of operating bandwidth. The diffusive line effect of the high value resistor used in the feedback network has been analyzed, and a compensation of its effect has been implemented. With a transimpedance of 200 MΩ, a background white noise of 9.5 fA/√(Hz) and an input current dynamic range up to 20 nA are achieved. The useful bandwidth ranges from 1 kHz up to 3 MHz depending on the noise level to be measured. An effective method for calibrating the system and measuring its transfer function with high accuracy over the full bandwidth is described. The designed TIA enables an accurate characterization of LGADs noise, which is a crucial step for the technology qualification and for the prediction of the performance of these devices as radiation and particle detectors.