|Thesis abstract: |
In recent years, the biomedical and chemical research has focused on optical analysis to better understand biological processes, indeed these non-invasive measurements are the best solution for in vivo experiments and medical diagnostic tests. This trend has pushed the research in the electronic field towards the development of high-performance photodetectors, in order to meet the strict requirements imposed by applications. As an example, in the last decades single-molecule spectroscopy (SMS) has rapidly grown in the biomedical and biochemical field. The technique involves examining very low concentration samples, therefore it requires photodetectors with extremely high sensitivity and low noise, and nowadays the best answer to these requirements is represented by single-photon detectors. In particular, great achievements in single-photon avalanche diode (SPAD) arrays have been recently made, pushed by a growing demand for parallel experimental setups. Indeed a multispot approach significantly reduces the measurement time, allowing the study of fast dynamic processes in analysis such as SMS, single-molecule Förster resonance energy transfer (smFRET) and fluorescence lifetime imaging microscopy (FLIM). Moreover, the temporal response of the single-photon detector is of utmost importance when the time-correlated single-photon counting (TCSPC) technique is employed to obtain fluorescence decay curves with subnanosecond time resolution.
Nowadays, state-of-art imaging sensors integrate thousands of single-photon detectors on the same chip. As an example, devices designed in CMOS technology make possible to integrate the detector and circuits for time-to-digital conversion within the pixel, thus allowing the fabrication of large 2-D SPAD arrays for high frame-rate time-resolved imaging applications, like laser ranging (LIDAR) or FLIM. However, from the detector point of view, these systems suffer from poorer performance with respect to the best in literature, which is represented by SPADs (and SPAD arrays) developed in a customized technology process. As a drawback, the latter limits the number of pixels that can be integrated in a single chip, due to on-chip signal routing issues. Moreover, to exploit the best detector performance, a hybrid CMOS-custom technology pixel structure has to be implemented, which leads to off-chip signal routing issues and increased area occupation.
To break the strong tradeoff between performance improvement and parallelization, in our research group a complete 32-channel system for TCSPC measurements, based on a custom-technology 32x1 linear SPAD array, has been designed and fabricated with the aim to maximize the detector performance. Part of this complete system constitutes the main subject of this thesis work, which involved the design and fabrication of a compact 32-channel time-resolved single-photon detection instrument, capable of performing single-photon counting analysis as a stand-alone module, and TCSPC analysis when connected to a multichannel TCSPC instrument. The basis of this project are represented by a 8-channel detection head, developed in a previous thesis work, and the experience acquired from the development of a complete single-channel TCSPC instrument.
The 32-channel time-resolved detection head exploits the extremely high performance of the 32x1 custom-technology SPAD array, which works in conjunction with a 32x1 CMOS AQC array. However, this hybrid architecture has led to many design issues to deal with, like signal routing, electrical crosstalk between parallel channels, connectivity, power management, heat dissipation and the development of suitable packaging solutions. In particular, many efforts have been required to implement a SPAD temperature control that allows to significantly reduce the dark counts.
A complete and parametric characterization of the detection head has been performed. Results are in good agreement with SPADs developed in a custom technology process, in terms of dark count rate, afterpulsing probability, photon-detection efficiency and optical crosstalk. Moreover, the system response function to a photon-detection event features a time jitter as low as 60ps FWHM at a mean count rate up to 1Mcps per channel, hence suitable for multidimensional TCSPC measurements.
Besides the time-resolved module, during this thesis work a 64-channel single-photon detection head for single-photon counting measurements has been designed and fabricated. Indeed, 2-D SPAD arrays and SPADs with enhanced efficiency in the red region of the visible spectrum have been recently developed in our research group. The new detection head has been developed with the aim to make these new detectors employable in an experimental setup, thus meeting a growing demand for high performance and high measurement throughput in biochemical applications. In the new detection head, the channel number increase has been obtained at the cost of a reduced time resolution performance. However, the two systems developed in this thesis work are complementary, each one targeting a specific range of applications. Moreover, they actually represent an attempt to investigate the limit of the tradeoff between performance and parallelization for custom-technology-based single-photon detection systems.