Nuclear Instruments and Methods in Physics Research A

Transcription

1 Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]] ]]] Contents lists available at SciVerse ScienceDirect Nuclear Instruments and Methods in Physics Research A journal homepage: Single photoelectron timing resolution of SiPM as a function of the bias voltage, the wavelength and the temperature V. Puill n, C. Bazin, D. Breton, L. Burmistrov, V. Chaumat, N. Dinu, J. Maalmi, J.F. Vagnucci, A. Stocchi Laboratory of Linear Accelerator (LAL), CNRS In2p3, Orsay, France article info Keywords: Silicon photomultipliers (SiPM) Single photoelectron timing resolution (SPTR) Time-of-flight (TOF) Picoseconds level abstract This work reports on Silicon Photomultipliers (SiPM) timing resolution measurements performed at the picosecond level at Laboratory of Linear Accelerator (LAL), In2p3- CNRS. The dependence of Single Photoelectron Timing Resolution (SPTR) with the applied voltage, wavelength of the light and the temperature was measured for detectors from Hamamatsu Photonics, AdvanSiD and Sensl with an active area of 1 and 9 mm 2. The SPTR improves with the bias voltage increase. No significant variation of SPTR was observed with the temperature change. We also observed a weak variation of it as a function of the wavelength of the light. The best SPTR measured was about 120 ps (FWHM). & 2012 Elsevier B.V. All rights reserved. 1. Introduction Time-of-flight (TOF) technique is used in High Energy Physics experiments to perform particle identification. TOF systems based on SiPM detectors coupled to quartz Cherenkov radiators could be an option for upgrading the Particle Identification system capabilities. It is assumed that few photons (less than 10) reach the photodetector at the quartz output; SiPM should then be evaluated at a weak light level in order to determine its contribution to the total timing resolution of the detection chain. In the present article, we report on the study of the single photoelectron timing resolution (SPTR) of different SiPMs. This measurement is performed in blue light to match with the wavelength of Cherenkov detectors but also in red to study its variation with the wavelength of the light. We also study the SPTR at 0, 10 and 20 1C to determine if an accurate stabilization of the temperature is mandatory to keep the SPTR at a constant level. This work was carried out in the framework of the Forward PID collaboration of the SuperB experiment with funding of IN2P3 and INFN. 2. Experimental 2.1. Description of the tested devices and experimental set-up Devices from Hamamatsu Photonics (MPPC), Sensl (SPM) and AdvanSiD (ASD, produced by F.B.K) were characterized in the same n Corresponding author. address: puill@lal.in2p3.fr (V. Puill). experimental conditions. These detectors are listed in Table 1. The two MPPC with the references 10-50S BK-4S and F FS are prototypes (also called wide trace MPPC ) designed to improve the timing resolution of the standards MPPC [1]. All the characterizations were performed inside a climatic chamber that gives a high stability (70.1 1C) of the temperature. The temperature of the SiPM is monitored by a Pt100 sensor mounted very close (2 mm) to it and read by an acquisition unit (Keithley 2700). The Fig. 1 presents the experimental set-up used for the SPTR measurements. Optical pulses from Pilas laser diodes are sent via a semi-reflective mirror on the SiPM and on a reference PMT (Hamamatsu R7400U-01) that checked the time stability of the laser intensity. Three different laser diodes (405, 467 and 635 nm) are driven at a repetition rate of 500 khz. The light intensity is controlled with neutral density filters placed between the semireflective mirror output and the SiPM. The spectral width of the laser pulses does not exceed 3 nm and, depending on the laser diode head, the pulse timing width is between 38 and 50 ps. The read-out electronics for the SiPM signal consists in a 500 MHz MITEQ voltage amplifier (gain¼350) with an input impedance of 50 O. The amplifier output is sent via a SMA cable and sampled either by a Wavepro 750ZI LeCroy digital oscilloscope (40 GSamples/s, 4 GHz of analog bandwidth) or by the 3.3 GSamples/s WaveCatcher ASIC-based waveform digitizer developed at LAL [2] Timing measurement method A first set of measurements was performed in order to determine the working range of each SiPM: its breakdown voltage (V BD ), /$ - see front matter & 2012 Elsevier B.V. All rights reserved. doi: /j.nima

2 2 V. Puill et al. / Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]] ]]] gain and dark count rate (DCR) were measured at 0, 15 and 20 1C. The temperature variation coefficient of the breakdown voltage of each device was then calculated. For more precisions about the employed set-up and the principle of measurements, refer to [3]. Then, and for each measurement, we checked that the SiPM works in single photo-electron mode: on the histogram of the signal amplitude, we observe the pedestal peak with at least 80% of the events (signals with amplitude 0), then a second peak due to signals with an amplitude of 1 p.e and between 15% and 20% of the population of this peak with amplitude 2 p.e (Fig. 2). The events of the 2 p.e peak include the cross-talk of the device. Since the PDE SiPM increases with the bias voltage, we adapted the attenuation of the light when changing the bias voltage (Vbias) of the SiPM in order to stay in the single photoelectron detection condition. The timing resolution to single photon was studied by measuring the fluctuations of the difference in time (Dt) between the SiPM amplified signal and the laser driver synchronization output (laser trigger). The tuning of the instruments (oscilloscope and Wavecatcher) is a very important matter: as the incident flux is very low (1 or 2 photons/pulse) and the SiPM PDE is between 10 and 40% (depending on the device and the wavelength), a lot of photons are not converted. In order to avoid those lost events for timing resolution measurements, we performed a coincidence window of around 10 ns between the laser trigger and the SiPM signal taken with a threshold above the electronic noise of the chain (around 20 mv). This tuning improves the single photo-electron acquisition rate. as the rising time of the SiPM signal changes when its amplitude changes (e.g. due to cross-talk event, variation of Vbias), we use a constant-fraction threshold instead of a fixed one to perform the measurements of Dt. With this method, all the detectors are characterized in the same experimental conditions. This threshold is set to 50% of the peak of the SiPM signal and of the laser trigger one even if this last is very stable. In all measurements discussed below, we report timing resolutions as the FWHM of the timing (Dt) distribution and refer to them as SPTR. The SPTR is not corrected with the laser pulse width. The contributions to the SPTR from the electronics were measured at LAL: it is about 8 ps in the case of the use of the Wavecatcher and 1 ps for the Wavepro 750ZI. Optical contributions were measured by Advanced Laser Diode Systems: as the laser trigger jitter is around 3 ps, the main contribution from the light source comes from the pulse width (38 50 ps FWHM). The systematic errors on the SPTR measurement were estimated at 75% (10% for the 9 mm 2 ). The values of Dt measured by the Wavecatcher and the oscilloscope are in agreement within 5%. Table 1 Characteristics of the tested SiPMs (V BD : breakdown voltage). Producer Ref SiPM Area (mm 2 ) Pixel size (lm) VBD (V) at 20 1C 3. SiPM SPTR measurements 3.1. SiPM SPTR as a function of the bias voltage AdvanSiD ASD-SiPM1S-M HAMAMATSU S HAMAMATSU S HAMAMATSU S HAMAMATSU 10-50S BK-4S HAMAMATSU S FS Sensl SPM1020X Sensl SPM1035X AdvanSiD ASD-SiPM3S-M HAMAMATSU S HAMAMATSU S HAMAMATSU S Sensl SPM1035X The study of the SiPM SPTR with the bias voltage was performed at 467 nm and at a temperature of 20 1C. Fig. 3 shows the results of these measurements as a function of the overvoltage DV (Vbias Vbreakdown). We observe an improvement of the timing resolution with an increasing overvoltage for all the devices till a maximum value of the bias voltage above which we cannot distinguish dark noise pulses from the true pulses due to the detection of light. For the MPPC, we observe an improvement of the SPTR with the pixel size increase (at the same DV). The size of the pixel does not affect the SPTR of the SPM. Fig. 1. Experimental setup for the SiPM SPTR measurement.

3 V. Puill et al. / Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]] ]]] 3 Fig. 2. Laser and SiPM signals on the oscilloscope with the CFD ratios, coincidence window and histograms of the measurements (SiPM amplitude and Dt between the SiPM and the laser trigger signals). The SiPM signal, in this example, arrives before the laser trigger due to the delay put between the laser driver and the oscilloscope. Fig. 3. 1mm 2 SiPM SPTR as a function of the over-voltage. The best SPTR was measured on the detector ASD (FWHM¼ 120 ps), the wide trace MPPCs and the two SPMs show approximately the same behavior ( ps). The wide trace MPPCs show better SPTR than the standard one SiPM SPTR as a function of the temperature The variation of the temperature is a critical parameter for the behavior of SiPM as it implies the change of its breakdown voltage, gain and DCR [4,5]. To study its effect on the timing resolution, we performed measurement between 0 and 20 1C at constant DV (which is calculated at 20 1C for a bias voltage value that gives the best SPTR). Only 9 mm 2 MPPC could be measured, the SPM and ASD showing too much dark count rate (E10 MHz at 20 1C). Fig. 4 shows the results at 0, 10 and 20 1C and for different wavelengths. As the MPPC and ASD show a stable behavior with the temperature increase, we observe a small trend to the degradation of the SPTR (10 15%) for the SPM. We assume that this is due to the fact that the difference of temperature is too low to affect in a significant way the mobility of the charge carriers [6]. These variations with the temperature are independent of the pixel size and on the wavelength of the light detected by the SiPM. Fig. 4. 1mm 2 and 9 mm 2 SiPM SPTR as a function of the temperature at constant DV and for different wavelengths SiPM SPTR as a function of the wavelength Fig. 5 shows the SPTR variations with the wavelength at a temperature of 20 1C. We ob serve 2 different behaviors: a trend to the improvement of the SPTR when the wavelength increases for the MPPCs whereas the contrary is observed for the SPMs and the ASD. For all these detectors, SPTR and PDE (photon detection efficiency) do not achieved their best value at the same wavelength [2] (as the MPPC PDE is best in blue and the SPM s is best in red).

4 4 V. Puill et al. / Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]] ]]] In the case of the SPMs, it seems that the wavelength increase has a more important effect on larger pixel size (the SPTR variation between 405 and 635 nm is negligible for the 20 mm whereas it is about 16% for the 35 mm). We observe also this phenomenon with the MPPCs: even if the variation is very weak (almost within the measurement errors), the trend is stronger for the 100 mm than for the 25 mm for the 9mm 2 for example. To explain this observation is not trivial; for MPPC for example, we would have expected a better SPTR in blue than in red: if we consider the simple detector geometry shown in Fig. 6 (p/n junction on a n-type substrate) and the absorption length of light in Silicon, blue photons are absorbed in half a mm and the red one are stopped deeper, at a depth of 2 3 mm. The electrons created by the absorption of blue photon reach quickly the high field region (if we consider that the junction is at a depth of mm) whereas the holes created by the absorption of red photon reach this region hundred of picoseconds later (as they are farther and their drift velocity is twice less than that of the electron [7]). We would then have expected more important fluctuations on the arrival time of these carriers due to the phenomenon of lateral spreading by diffusion [8] and therefore a higher SPTR value. This reasoning is too simple and other parameters like the field profiles and the depth of the junction have to be taken into account in order to explain what we observe (these information are not disclosed by the producers) Delayed events The timing profile (Dt histogram) in the single photon regime is well fitted by a Gaussian function plus another one corresponding to delayed events with a mean value ps latter than the mean value of the main Gaussian fit (Fig. 7). We observed this behavior, in a more or less pronounced way, with all the SiPMs, at the three different wavelengths (with 2 laser drivers and 3 different laser diode heads), in two test benches with different geometries. The proportion of delayed events increases with the overvoltage. We checked, cutting the events with an amplitude pulse of 2 photo-electrons and fitting the resulting timing distribution histogram that these events do not come from the cross-talk inherent to the detector. They cannot be explained either by after-pulses as we use a narrow coincidence widow (between 5 and 10 ns). The study of this tail as a function of the detector pixel size, the bias voltage and the wavelength of the light is on progress and its results will be reported in a future article. 4. Conclusions Fig. 5. 1mm 2 and 9 mm 2 SiPM SPTR as a function of the wavelength at 20 1C. The single photoelectron timing resolution of different SiPMs was measured over each detector bias voltage range in blue light, then at 2 different wavelengths of pulsed light at a fixed overvoltage for temperatures from 0 to 20 1C. The increase of the bias voltage improves significantly the SPTR. However, this effect is limited by the parallel increase of the DCR that prevents the detection of single photon when the overvoltage is too high. The best values of SPTR for MPPC and SPM are around 150 ps (FWHM) whereas the SPTR of ASD shows Fig. 6. Simplified structure of a MPPC and absorption depth of photon in intrinsic Si as a function of the wavelength (reproduced from [9]).

5 V. Puill et al. / Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]] ]]] 5 Fig. 7. Dt distribution of the MPPC S at 2 bias voltages (left: 69.3 V, right: 70 V) at 20 1C and 467 nm. 120 ps. MPPC SPTR is slightly better in red light than in blue whereas the contrary (with amplitude of variation more pronounced) is observed on the SPM and ASD. The variation of the temperature (from 0 to 20 1C) does not affect in a significant way the SPTR; nevertheless, the cooling of the device can improve the single photon discrimination efficiency by decreasing the DCR. More work is in progress to understand the shape of the timing resolution (delayed events) and the SPTR variations with the wavelength and the pixel size. SiPMs exhibit good single photoelectron timing resolution for particle identification system in comparison with MCP-PMTs (multi channels plate photomultiplier) with SPTR quoted at 70 ps in sigma, around 160 ps in FWHM. However, their weak radiation hardness does not permit to use them, for the moment, in hostile environments. References [1] S. Kamakura, S. Ohsuka, K. Yamamura, K. Sato, Production and Development status of MPPC, PoS(PD09)017. [2] J. Va vra, D. Breton, E. Delagnes, J. Maalmi, K. Nishimura, L.L. Ruckman, G. Varner, High resolution photon timing with MCP-PMTs: a comparison of commercial constant fraction discriminator (CFD) with ASIC-based waveform digitizers TARGET and WaveCatcher, NIM A 629 (2011) [3] N. Dinu, Z. Amara, C. Bazin, V. Chaumat, C. Cheikali, G. Guilhem, V. Puill, C. Sylvia, NIM A 610 (2009) 423. [4] N. Dinu, C. Bazin, V. Chaumat, C. Cheikali, A. Para, V. Puill, C. Sylvia, J.F. Vagnucci, NSS Conference Records (2010) 215. [5] G. Collazuol, M.G. Bisogni, S. Marcatili, C. Piemonte, A. Del Guerra, NIM A 628 (2010) 389. [6] M. Sze, Physics of Semiconductor Devices, third ed. Simon. [7] E.J. Ryder, Physical Review 90 (1953) 766. [8] A. Lacaita, M. Mastrapasqua, M. Ghioni, S. Vanoli, Applied Physics Letters 57 (1990) 489. [9] K. Rajkanan, R. Singh, J. Shewchun, Solid-State Electronics 22 (1979) 793.