|Thesis abstract: |
In the last years the analysis of extremely fast and faint luminous signals has taken a decisive role in many areas, like medicine, biology and chemistry. In all these fields the technique known as TCSPC (Time-Correlated Single Photon Counting) has become of utmost importance. It consists of periodically stimulating a sample with a LASER pulse and measuring the delay between the arrival time of a photon emitted from the specimen and the reference pulse in order to build a histogram of the measured times.
Some example of analysis which benefits from TCSPC are fluorescence lifetime imaging microscopy (FLIM), Förster resonance energy transfer (FRET) and fluorescence correlation spectroscopy (FCS).
These noninvasive techniques allow to collect information about protein structure, DNA sequencing or tissue growing, making possible the detection of pathologies and tumors.
The device of choice for all these applications is nowadays represented by single photon avalanche diodes (SPAD), that have gradually replaced photo multiplier tubes (PMT), thanks to the higher photon detection efficiency (PDE), lower power consumption, less encumbrance, and higher immunity to magnetic fields.
With the evolution of applications a system with high number of parallel measurement channels has become increasingly required. Indeed it allows not only to speed up the experiments, but also to record multiple measurements simultaneously and permitting to vary some parameters like the wavelength of the source or the spatial coordinates.
In recent years many arrays of photodetector with dedicated electronic have been proposed, integrating thousands of SPADs on the same chip, each one equipped with a dedicated measurement chain, which relies on a TAC (Time to Amplitude Converter) followed by an ADC or on a TDC (Time to Amplitude Converter). Despite this structure definitely presents considerable advantages, a real exploitation of a high number of measurement channels in parallel is nowadays not viable because of the too high data rate that would occur at the output of the whole system.
Ideally the maximum data rate at which every channel can operate depends only on the frequency of repetition of the laser pulses which excite the sample, but despite the relatively low average excitation rate of a single channel, the presence of a high number of sensors leads to a huge bit stream. Therefore a tradeoff between array dimensions and performances is observed.
A state-of-the-art multichannel system that overcomes the bottleneck of data transfer would be of great interest in all the applications listed above.
A possible way to reach this goal is the realization of a system based on a smart routing logic and only few high-performance acquisition chains, which will be will be investigated in detail.
The use of only few acquisition and conversion chains, shared with the whole array, allows to make the best use of the available resources. In fact the number of converters can be chosen regardless of the number of detectors, in order to obtain the maximum throughput that can be handled at the output.
Furthermore, since only few external converters are necessary, low power consumption and small dimensions are not the most stringent constraints and it is possible to consider low jitter, high resolution and high linearity as main figures of merit.
Since the data transfer bottleneck affects also system specifically designed for counting and imaging applications a possible extension of this solution to such systems will be investigated, providing that no information can be lost during the measurement.
In conclusion, the target of this project is the design of the integrated circuits that can lead to the building of a SPAD-based acquisition system capable of breaking the tradeoff between number of channels and performance that now sets a limit to the available systems.