**3. Bidimensional arrays of SPADs**

As discussed in the previous sections, SPAD arrays fabricated by standard planar silicon production processing, are at the moment considered an interesting topic of research because many photonic applications, not only scientific, would become available at much more reasonable cost.

The main requirement in the array fabrication is to guarantee electrical and optical isolation among pixels. In fact, each pixel must be operated independently from the others, independently on neighboring pixels and without disturbing them.

Fig. 6. *a)* PDE of ST-Microelectronics SPAD at 5 and 10% of EBV; *b*) Photon timing of

The measurement principle adopted for the evaluation SPAD PDE is a direct comparison with a calibrated photodiode that receives the same photon flux of SPAD. An example of experimental set-up where, with high level of accuracy, are possible PDE measurements is discussed by Bonanno et al. 2005. In figure 6a is reported the typical PDE measured for the

The SPAD performance in *Photon Timing* is characterized by its time resolution curve, which is the statistical distribution of the delay between the true arrival time of the photon and the measured time, marked by the onset of the avalanche current pulse. These can be obtained by means of a time-correlated photon counting apparatus (Privitera at al. 2008), where optical pulses with short time duration (less than 30 - 40 ps) are used to illuminate the devices. After a careful check that the contribution of the laser pulse and the eventual electronic jitter to the SPAD response are negligible, by using a digital oscilloscope it is possible to build the histogram of the time delay between the laser reference pulse and the

 In figure 5b is reported the time resolution curve obtained by using laser pulses of 408 nm wavelengths, 35 ps FWHM and 1 kHz repetition rate, on an ST-Microelectronic SPAD. It is

As discussed in the previous sections, SPAD arrays fabricated by standard planar silicon production processing, are at the moment considered an interesting topic of research because many photonic applications, not only scientific, would become available at much

The main requirement in the array fabrication is to guarantee electrical and optical isolation among pixels. In fact, each pixel must be operated independently from the others,

consistent with the typical values reported in the literature for other devices.

independently on neighboring pixels and without disturbing them.

passively quenched ST-Microelectronics SPAD.

ST-Microelectronic devices.

**3. Bidimensional arrays of SPADs** 

SPAD pulse.

more reasonable cost.

A typical way to achieve electrical isolation in silicon planar technology is to employ isolation diffusion, i.e. a thus electrically isolating the epitaxial layer, of the opposite doping sufficiently deep diffusion that reaches the substrate of the same doping type, type, as shown in Figure 7. However, the hardest problem to solve in SPAD array design is optical coupling. This effect is peculiar to the above-breakdown operation of SPADs.

Fig. 7. Optical cross-talk representation between two neighbouring devices. The avalanche current in one pixel can trigger the process in surrounding pixels, due to photons emitted by hot carriers.

During the avalanche process each ionization impact causes high deceleration in electron drift, so the Bremsstrahlung process enables a secondary photons production inside the junction (the emission probability is estimated in 10-5 photons per carrier crossing the junction).

The optical cross talk is then activated by the emission of secondary photons, that can travel along the silicon bulk, transparent to the optical wavelength, and cross another diode that is triggering a correlated avalanche (Fig. 7). The time delay between the two produced signals is less than the measurable one and induced signals cannot be distinguished from the generated one. A further contribute to the optical cross talk was found in the indirect optical pat enabled by the optical reflection by the back part of the device.

In order to avoid the optical cross-talk, it is possible design and fabricate arrays that are optically and electrically isolated by deep thin trench technology. The trench process starts with a vertical etch (about 10 μm deep and 1μm large), a subsequent oxide deposition for complete electrical isolation. The process continues with tungsten fill to avoid optical crosstalk and ends with planarisation.

Nevertheless, all the cross-talk contributions previous mentioned, become important when the density of implemented elements is higher and the distances between neighbouring devices are smaller as required by imaging applications or by silicon PhotoMultiplier technology. They influence the total Dark counting rate of the device.

## **3.1 Silicon photomultiplier concept**

A photomultiplier based on the silicon technology represents the new frontier of photodetection. SPADs integrated on the same substrate, with a common read-out, could satisfy such expectations. The main limitation of a single diode working in Gaiger mode is

Time Resolved Camera: The New Frontier of Imaging Devices 317

In low light level applications one can use an event-driven readout, where the detector initiates and drives a column-wise detection process directly (Niclass et al. 2006). The drawback of this approach is that multiple photons cannot be detected simultaneously on the same column. In addition, the bandwidth of the column readout mechanism limits saturation levels of the entire column. These limitations are not problematic if the expected

An alternative approach for no low light level situations is the use of a latchless pipeline scheme. In this approach, the absorption of a photon causes the SPAD to inject a digital signal into a delay line that is then read externally. The timing of all injected pulses is evaluated so as to derive the time of arrival of the photon and the pixel of origin. This method allows detection of photons simultaneously over a column even though some restrictions apply on the timing of the optical set up. Note that in this design an effective gating mechanism is necessary to prevent photons detected outside a certain time window being interpreted as originating in a pixel other than the one responsible for that detection. The pixel access problem can be overcome if the photon time of arrival is performed in situ,

The main problem of the all discussed solutions is the physical size of the on-board ancillary circuitries (digital signal extractor, counters, memory elements, buffer readout, etc.), which drastically decrees the fill factor and increase the complexity of the chip manufacturing.

The alternative approach is to bring out the chip all the readout circuitries in order to increase the fill factor, simplify the matrix design and the manufacturing process (Tudisco et

A cross-wire readout scheme could be adopted to access every pixel independently; the avalanche signal of each diode is twist extracted from two contacts, these are shared one for all diodes of the same row and the other for all diodes of the same column. Figure 9 reports the scheme of two possible configurations. In such a way it is possible to deduce the hit

Fig. 9. Cross-wire readout scheme. In *a*) the avalanche signal of each pixel is collected from both side of the junction. In *b*) each pixel has two quenching resistors and a common anode

If one of the devices "fires" in response to light applied to that device, it will generate a current pulse in one of corresponding row and column outputs. This allows the specific

device to be determined unambiguously by its unique row-column signature.

position with a readout complexity which grows as *2n* instead of *n2*.

photon flux hitting the sensor is low.

i.e. on the pixel itself.

al. 2009).

contact.

Fig. 8. *a*) Schematic representation of SiPM working principle. The amplitude of the out signals is proportional to the number of detected photons. In *b*) is reported the distribution.

that the output signal is the same regardless of the number of interacting photons. In order to partially overcome this limitation, the diode can be segmented in tiny micro-cells (each working in Gaiger mode) connected in parallel to a single output. Each element, when activated by a photon, gives the same current response, so that the output signal is proportional to the number of cells hit by a photon. The number of elements composing the device limits the dynamic range, and the probability that two or more photons hit the same micro-cell depends on the size of the micro-cell itself. This structure is called Silicon PhotoMultiplier (SiPM). Some of the advantages offered by the solid-state solution are: insensitivity to magnetic fields, ruggedness, compactness, low operating voltage and long lifespan. In addition, this technology facilitates the interconnection between the detector and the read-out electronics.

An interesting application of SiPMs is the detection of the light emitted by scintillators. Among the various types of scintillators, particular attention has recently been given to lutetium oxyorthosilicate (LSO) for its high light yield, short decay time and relatively good mechanical properties. These features are extremely useful, for example, in positron emission tomography (PET).

#### **3.2 Image sensor**

The identification of the fired pixel and the time at which the event occurred are the fundamental requirements in order to realize the time resolved imaging. These two features are almost simultaneously incompatible with the traditional reading techniques (CCD etc.). High-definition images require a large number of pixels and reading techniques designed to minimize the number of connections. On the other hand, these are almost incompatible with the information on the photon arrival time.

To avoid missing photon counts, a counter should be used for each pixel. However, large counters are not desirable due to the fill factor loss and/or extra time required to perform a complete readout of the contents of the chip. A possible solution to this problem is to access every pixel independently but sequentially using a digital random access scheme (Niclass et al. 2005).

Fig. 8. *a*) Schematic representation of SiPM working principle. The amplitude of the out signals is proportional to the number of detected photons. In *b*) is reported the distribution.

the read-out electronics.

emission tomography (PET).

the information on the photon arrival time.

**3.2 Image sensor** 

al. 2005).

that the output signal is the same regardless of the number of interacting photons. In order to partially overcome this limitation, the diode can be segmented in tiny micro-cells (each working in Gaiger mode) connected in parallel to a single output. Each element, when activated by a photon, gives the same current response, so that the output signal is proportional to the number of cells hit by a photon. The number of elements composing the device limits the dynamic range, and the probability that two or more photons hit the same micro-cell depends on the size of the micro-cell itself. This structure is called Silicon PhotoMultiplier (SiPM). Some of the advantages offered by the solid-state solution are: insensitivity to magnetic fields, ruggedness, compactness, low operating voltage and long lifespan. In addition, this technology facilitates the interconnection between the detector and

An interesting application of SiPMs is the detection of the light emitted by scintillators. Among the various types of scintillators, particular attention has recently been given to lutetium oxyorthosilicate (LSO) for its high light yield, short decay time and relatively good mechanical properties. These features are extremely useful, for example, in positron

The identification of the fired pixel and the time at which the event occurred are the fundamental requirements in order to realize the time resolved imaging. These two features are almost simultaneously incompatible with the traditional reading techniques (CCD etc.). High-definition images require a large number of pixels and reading techniques designed to minimize the number of connections. On the other hand, these are almost incompatible with

To avoid missing photon counts, a counter should be used for each pixel. However, large counters are not desirable due to the fill factor loss and/or extra time required to perform a complete readout of the contents of the chip. A possible solution to this problem is to access every pixel independently but sequentially using a digital random access scheme (Niclass et In low light level applications one can use an event-driven readout, where the detector initiates and drives a column-wise detection process directly (Niclass et al. 2006). The drawback of this approach is that multiple photons cannot be detected simultaneously on the same column. In addition, the bandwidth of the column readout mechanism limits saturation levels of the entire column. These limitations are not problematic if the expected photon flux hitting the sensor is low.

An alternative approach for no low light level situations is the use of a latchless pipeline scheme. In this approach, the absorption of a photon causes the SPAD to inject a digital signal into a delay line that is then read externally. The timing of all injected pulses is evaluated so as to derive the time of arrival of the photon and the pixel of origin. This method allows detection of photons simultaneously over a column even though some restrictions apply on the timing of the optical set up. Note that in this design an effective gating mechanism is necessary to prevent photons detected outside a certain time window being interpreted as originating in a pixel other than the one responsible for that detection. The pixel access problem can be overcome if the photon time of arrival is performed in situ, i.e. on the pixel itself.

The main problem of the all discussed solutions is the physical size of the on-board ancillary circuitries (digital signal extractor, counters, memory elements, buffer readout, etc.), which drastically decrees the fill factor and increase the complexity of the chip manufacturing.

The alternative approach is to bring out the chip all the readout circuitries in order to increase the fill factor, simplify the matrix design and the manufacturing process (Tudisco et al. 2009).

A cross-wire readout scheme could be adopted to access every pixel independently; the avalanche signal of each diode is twist extracted from two contacts, these are shared one for all diodes of the same row and the other for all diodes of the same column. Figure 9 reports the scheme of two possible configurations. In such a way it is possible to deduce the hit position with a readout complexity which grows as *2n* instead of *n2*.

Fig. 9. Cross-wire readout scheme. In *a*) the avalanche signal of each pixel is collected from both side of the junction. In *b*) each pixel has two quenching resistors and a common anode contact.

If one of the devices "fires" in response to light applied to that device, it will generate a current pulse in one of corresponding row and column outputs. This allows the specific device to be determined unambiguously by its unique row-column signature.

Time Resolved Camera: The New Frontier of Imaging Devices 319

Fig. 11. *a*) CCD image from the Light Emission Diode. *b*) The same image obtained by 10x10

The possibility to adjust the photon flux with a fine precision is a fundamental requirement in order to characterize a TRC. The use of LED source guarantees this possibility through the current limitation on it. With this regulation capability it is possible to collect image with a normal CCD camera (fig. 11a) at medium photon rate and test the SPAD matrix capability in all the sensor dynamic range, from single photon up to the saturation level. By using time modulated current signals it is also possible to test the time dependent characteristic of

TRC doesn't have frame time limitations like frame duration, frame rate, minimum time distance between frames etc. Every collected photons give his contribution to a continuous streaming of the photon source, the pictures can extracted putting together all the photons during a defined time, or can be collected in groups forming a sequence of video frames without any dead time between frames. The time and spatial positions of all detected photons are collected, and can be used for many correlation technique that use time and/or

The solution sketched in figure 9b is very promising. Arrays of several size and configuration (up to 170x170 elements) has been designed and manufactured in

As previous mentioned, this solution guarantees the collection of the total OR signal (like in SiPM configuration) simplifying the ancillary readout circuitry. From the back, through the anodic contact, you can extract the photon arrival time information (e.g. through a single channel of Multi-Hit TDC) while the position information can be extracted from the front by

An alternative readout strategy currently under development is reported in figure 12a. A resistive path is realised on the row contacts and column contacts. By using the charge repartition principle it is possible to determine the row and column fired just sampling the

collaboration with FBK-IRST and are currently under test a characterization.

array. *c*) Schema Readout electronic.

whole system.

spatial correlation analysis.

**3.2.1 Current status and prospective** 

using, for example, the scheme reported in figure 12b.

The only real limitation of this spatial recognition approach is the uncertainty in the simultaneous detection of two photons. This uncertainty comes from events tagging and is related to the time jitter of row and column signals.

In the configuration of figure 9a the avalanche signal of each pixel is collected from both side of the junction, anode and cathode while the solution of figure 9b requires the integration of two quenching resistors for each diode, with advantages and disadvantages related to the existence of a common anode contact. The latter gives the possibility to collect the total OR-signal (like in SiPM configuration). Figure 10 shows the layout of some arrays made in collaboration with the FBK-IRST and currently under test and characterization.

Fig. 10. Layout of FBK–IRST arrays.

Adopting the first solution was recently realised in collaboration with ST-Microelectronics an elementary first demonstrator of 10x10 elements (Tudisco et al. 2009).

The array layout has been designed with common anode and cathode bus. Metal strips used to contact the cathodes in each column are continuous while those used to contact the anodes in each row are interrupted in proximity of each cathode metal bus in order to avoid short-circuits with cathode metal strips. A high-doped silicon underpass guaranties the conductivity. The operating conditions were assured by the use of a single external quenching resistor for row (or column).

In such simplified configuration, respect to the figure 9a, the main disadvantage is the quenching of all devices in the row (or column) join to the slow recharge time coming from to use of externals quenching resistor. Nevertheless it remains a valid concept device.

In order to realize a first working prototype of "Time Resolved Camera" (TRC), row and column signals can be treated by standard nuclear electronics. For example, by using commercial VME modules (Constant Fraction Discriminators and multi-hits Time to Digital Converters, fig. 11c) it is possible to acquire and reconstruct the information concerning position and arrival time of photons.

An example of the achievable performance is reported in Figure 11b, in which the head of a Light Emission Diode (LED) was focus (trough a set of lens) on the top, alternative, of ccd/SPAD-array surface. The image was reconstructed after the subtraction of the dark count rate of each pixel. To maximize the image contrast we fix the black level of the colour scale at counting rate less than 500 Hz.

The only real limitation of this spatial recognition approach is the uncertainty in the simultaneous detection of two photons. This uncertainty comes from events tagging and is

In the configuration of figure 9a the avalanche signal of each pixel is collected from both side of the junction, anode and cathode while the solution of figure 9b requires the integration of two quenching resistors for each diode, with advantages and disadvantages related to the existence of a common anode contact. The latter gives the possibility to collect the total OR-signal (like in SiPM configuration). Figure 10 shows the layout of some arrays made in collaboration with the FBK-IRST and currently under test and

Adopting the first solution was recently realised in collaboration with ST-Microelectronics

The array layout has been designed with common anode and cathode bus. Metal strips used to contact the cathodes in each column are continuous while those used to contact the anodes in each row are interrupted in proximity of each cathode metal bus in order to avoid short-circuits with cathode metal strips. A high-doped silicon underpass guaranties the conductivity. The operating conditions were assured by the use of a single external

In such simplified configuration, respect to the figure 9a, the main disadvantage is the quenching of all devices in the row (or column) join to the slow recharge time coming from to use of externals quenching resistor. Nevertheless it remains a valid concept device.

In order to realize a first working prototype of "Time Resolved Camera" (TRC), row and column signals can be treated by standard nuclear electronics. For example, by using commercial VME modules (Constant Fraction Discriminators and multi-hits Time to Digital Converters, fig. 11c) it is possible to acquire and reconstruct the information concerning

An example of the achievable performance is reported in Figure 11b, in which the head of a Light Emission Diode (LED) was focus (trough a set of lens) on the top, alternative, of ccd/SPAD-array surface. The image was reconstructed after the subtraction of the dark count rate of each pixel. To maximize the image contrast we fix the black level of the colour

an elementary first demonstrator of 10x10 elements (Tudisco et al. 2009).

related to the time jitter of row and column signals.

characterization.

Fig. 10. Layout of FBK–IRST arrays.

quenching resistor for row (or column).

position and arrival time of photons.

scale at counting rate less than 500 Hz.

Fig. 11. *a*) CCD image from the Light Emission Diode. *b*) The same image obtained by 10x10 array. *c*) Schema Readout electronic.

The possibility to adjust the photon flux with a fine precision is a fundamental requirement in order to characterize a TRC. The use of LED source guarantees this possibility through the current limitation on it. With this regulation capability it is possible to collect image with a normal CCD camera (fig. 11a) at medium photon rate and test the SPAD matrix capability in all the sensor dynamic range, from single photon up to the saturation level. By using time modulated current signals it is also possible to test the time dependent characteristic of whole system.

TRC doesn't have frame time limitations like frame duration, frame rate, minimum time distance between frames etc. Every collected photons give his contribution to a continuous streaming of the photon source, the pictures can extracted putting together all the photons during a defined time, or can be collected in groups forming a sequence of video frames without any dead time between frames. The time and spatial positions of all detected photons are collected, and can be used for many correlation technique that use time and/or spatial correlation analysis.

#### **3.2.1 Current status and prospective**

The solution sketched in figure 9b is very promising. Arrays of several size and configuration (up to 170x170 elements) has been designed and manufactured in collaboration with FBK-IRST and are currently under test a characterization.

As previous mentioned, this solution guarantees the collection of the total OR signal (like in SiPM configuration) simplifying the ancillary readout circuitry. From the back, through the anodic contact, you can extract the photon arrival time information (e.g. through a single channel of Multi-Hit TDC) while the position information can be extracted from the front by using, for example, the scheme reported in figure 12b.

An alternative readout strategy currently under development is reported in figure 12a. A resistive path is realised on the row contacts and column contacts. By using the charge repartition principle it is possible to determine the row and column fired just sampling the

Time Resolved Camera: The New Frontier of Imaging Devices 321

Schwille, P. et al., (1999). *Molecular dynamics in living cells observed by fluorescence correlation* 

Strickland, D. et al., (1985). *Compression of amplified chirped optica pulses*, Opt. Comunic. Vol.

McPhate, J. et al., (2005) *Noiseless kilohertz-frame-rate imaging detector based on microchannel* 

Etoh, T. G. et al. (2005) *Design of the PC-ISIS: photon-counting in-situ storage image sensor,* IEEE

Kawai, N. et al. (2005) *A low-noise signal readout circuit using double-stage noise cancelling* 

Hynecek, J. et al. (2001) *Impactron—a new solid state image intensifier,* IEEE Trans. on Electron

Cova, S. et al. (1981) *Towards picosecond resolution with single-photon avalanche diodes*, Rev. of

Cova, S. et al (1983) *A semiconductor detector for measuring ultraweak fluorescence decays with 70* 

Cova, S. et al. (1989) *Trapping phenomena in avalanche photodiodes on nanosecond scale*, IEEE

Cova, S. et al. (1996) *Avalanche photodiodes and quenching circuits for single photon detection*,

Mc Intyre, R.J. (1961) *Theory of Microplasma Instability in Silicon,* J. Appl. Phys. Vol. 35 pp.

Mc Intyre, R.J. (1966) *Multiplication noise in uniform avalanche diodes*, IEEE Trans. on Electron

Mc Intyre, R.J. 1985 *Recent developments in silicon avalanche photodiodes,* Measurement Vol. 3

Hurkx, G.A.M. et al. (1992) *A new analytical diode model including tunneling and avalanche breakdown*, IEEE Trans. Electron Devices, Vol. 39, pp ISSN 0018-9383 Oldham, W.G. et al. (1972) *Triggering phenomena in avalanche diodes," IEEE Trans. Electron* 

Zappa F et al. (1997) *Integrated array of avalanche photodiodes for single-photon counting*, IEEE

Kindt, W.J. (1999) *Geiger mode avalanche photodiode arrays for spatially resolved single photon* 

Sciacca, et. al. (2003) *Silicon planar technology for single-photon optical detectors*, IEEE Trans. on

Ripamonti, G.et al. (1985) *Carrier diffusion effects in the time-response of a fast photodiode*, Solid

Lacaita, A. et al. (1989) *Double epitaxy improves singlephoton avalanche diode performance*,

*counting,* PhD Thesis Delft University Press, ISBN 90-407-1845-8

*ps FWHM resolution,* IEEE J. Quantum Electronics Vol. 19 pp. 630-634 ISSN 0018-

Available on from http://www.max.hi-ho.ne.jp/teranishi/

Devices Vol. 48 pp. 2238–41 ISSN 0018-9383

Scie. Instrum. Vol. 52 pp. 408–412 ISSN 0034-6748

Appl. Opt. Vol. 35, pp. 1956–1976 ISSN 0003-6935.

*Devices*, Vol. 19, pp. 1056–1060 ISSN ISSN 0018-9383

Devices Vol. 13 pp. 164-168 ISSN 0018-9383

ESSDERC (Stuttgart, Germany) pp 600–603

Elect. Dev. Vol. 50, pp. 918-925 ISSN 0018-9383

Electron. Lett., Vol. 25, no. 13 ISSN 0013-5194

State Electron., Vol. 28, pp. 925–931, ISSN 0038-1101

Electron Device Lett. Vol. 12, pp. 685–687 ISSN 0741-3106

ISSN 0006-3495

9780819458865

ho.ne.jp/teranishi/

1370 ISSN 0021-8979

pp. 146–152 ISSN 0263-2241

9197

56 pp. 219-221, ISSN 0030-4018

*spectroscopy with one and two-photon excitation,* Biophys. J. Vol. 77 pp. 2251–2265,

*plates readout Medipix2 CMOS pixel chip, Proc. SPIE Vol.* 5881 pp. 88–97 ISBN

Workshop on CCDs and Advanced Image Sensors (Karuizawa, Japan) pp. 113–116

*architecture for CMOS image sensors* IEEE Workshop on CCDs and Advanced Image Sensors (Karuizawa, Japan) pp. 27–30 Available on from http://www.max.hi-

two signals at the end of the chain. This solution drastically reduces the number of reading channels up to two for rows and two for columns.

Fig. 12. *a*) Sketch of the alternative readout strategy through the rows resistive path and signals sampling. *b*) Possible readout schema of fig. 9b array.
