1. Introduction
The silicon photomultiplier (SiPM) is a sensor that can detect weak light signals even on single photon level. Its key performance characteristics include: high photon detection efficiency (PDE), excellent resolution for single photon detection, insensitivity to magnetic field, low operating voltage and convenience for integration. SiPM is gradually replacing the traditional photomultiplier tube in various applications like high energy physics, astrophysics and nuclear medical imaging [1-3]. Most commercial SiPMs employ polysilicon or other materials to make quenching resistors. SiPM with bulk integrated quenching resistor is theoretically simulated by Semiconductor Laboratory of the Max-Planck-Society, which have some advantages like omission of polysilicon and its metal lines on top of device and the simplified production [4].
The epitaxial quenching resistor (EQR) SiPM have been developed by Novel Device Laboratory (NDL) [5-7], which employ epitaxial substrate resistor instead of polysilicon resistor to connect avalanche photodiode (APD) cells. The EQR technology can be used to obtain high geometrical fill factor and high density of micro APD cells thus the EQR SiPM feature large dynamic range and high PDE simultaneously. The N-on-P type EQR SiPM with active area of 1×1 mm2 and cell density of 21488/mm2 had PDE equal to 16% at 480 nm in previous paper [8]. Compared with the N-on-P type device, the ionization coefficient of electron is much greater than holes, that electron from the top of device traveling into bulk has a greater probability to trigger the avalanche, thus the PDE of P-on-N type SiPM can be increased obviously [9,10]. More than that, the shallow p-n junction make the depletion region rather close to the surface also enhance PDE for the blue-violet light with short penetrating depth. This paper reported the PDE of P-on-N type EQR SiPM with active area of 3×3 mm2 as 30% at 420nm.
2. Experiments
2.1 Device structure
As shown in Fig. 1, the P-on-N type EQR SiPM comprises of thousands of APD cells, each of which consists of N-enriched region forming high electric field between N-type epitaxial silicon wafer and P++ surface layer. The epitaxial region below P-on-N junction behaves as a quenching resistor. The APD cells are isolated from each other by the depletion regions. The avalanche signals from each cell are collected directly by the P++ surface layer that act as a common cathode. The SiPM was TO-5 encapsulated, which has the active area of 3×3 mm2 and the cell size of 10 μm.
2.2. Characterizing Methods
As shown in Fig.2 (a), the device had been evaluated in a temperature controlled box to control the ambient temperature. In dark conditions, the pulse shape, dark count rate, the afterpulsing ratio and the crosstalk ratio were recorded and analyzed through the digital oscilloscope (LeCroy WaveRunner 640Zi).
The PDE depending on wavelength was characterized using stochastic photon counting method that devices was irradiated with stochastic photons and PDE was quantitatively measured by considering the increased count rate for avalanche pulses when comparing with dark count rate [11]. As shown in Fig.2 (b), an integrating sphere with three ports has been used. The incident light port was connected with Xenon lamp and monochromator (Zolix LSH-X 150) through optical fiber; the other ports were cling to monitoring PIN photodiode and SiPM. The events detected from SiPM were counted by the oscilloscope (Tektronix TDS 1012) and the monitoring current from PIN were collected by Keithley 237. PDE is calculated by the following formula [12]:
where, DCR is dark count rate and CR is photon counting rate of SiPM over 0.5 p.e. threshold; h represents Planck constant, c represents vacuum light speed and λ represents wavelength; RPIN is spectral responsivity, Itotal is photocurrent and Idark is dark current of PIN; ASiPM and APIN are active areas of SiPM and PIN respectively.
Dynamic range measurement has been shown in Fig.2 (c). The pulse light from OYSL-Photonics SC-Pro source (peak wavelength~532nm, pulse width~100ps, frequency ~1MHz) was transmitted and attenuated. Under adjustable light intensity, the dynamic range of SiPM was calibrated accurately by considering the multiple photons response pulse area over single photon pulse area. Under low light intensity, the pulse area distribution spectrum can be characterized.
3. Results and Discussion
3.1. I-V Characteristics
We present the I-V characteristics of P-on-N type EQR SiPM with active area of 3×3 mm2. As shown in Fig. 3, at room temperature the breakdown voltage (Vb) is 23V and the leakage current is about 0.01nA. The voltage at the second inflexion point in the I-V curve is 28V. Thus the maximum allowable over-voltage (ΔDV) reaches 5 V.
3.2. Dark Count and Crosstalk
The rate of counts in dark conditions caused by thermally generated electrons in the active volume of silicon is usually regarded as the primary source of noise. Its secondary pulse crosstalk (CT) and the afterpulsing affect the signal-to-noise ratio of SiPM, which is supposed to be reduced. Employing the resistance capacitance filter circuit and Keithley 237 bias source, the dark count rate (DCR) were observed directly through the oscilloscope. The ratio of DCR over two threshold level could be obtained as the crosstalk probability (CT=DCR1.5p.e./DCR0.5 p.e). As shown in Fig.4, the CT and DCR both depend on over-voltage. At room temperature, under 3.5V over-voltage, the DCR of P-on-N type EQR SiPM with 3×3 mm2 active area is 7MHz and the CT is below 7%.
3.3. Temperature Related Properties
Dependence of the Vb and the DCR on temperature were characterized in Fig.5. The Vb increase linearly with temperature, which had coefficient of 19 mV/℃ that was comparatively low temperature coefficient [13]. Due to increasing ionization rates at low temperature, APD cells having higher probability to trigger avalanche make Vb decreases subsequently [14]. The DCR mainly caused by thermal generation of electrons depending on the temperature and hence it can be reduced by cooling. In addition, the DCR can be effectively reduced by selecting the silicon substrate with few defects and by increasing the process reliability.
3.4. Photon detection efficiency
Figure 6 display the PDE spectra for device under different over-voltage at room temperature. The PDE increase with the over-voltage (less than 3.5V) and the peak PDE reach over 30% at wavelength of 420 nm. We didn’t observe significant rise of the peak PDE beyond the over-voltage of 3.5 V, which indicates that the Geiger probability was probably reaching the maximum value and the DCR was too high to evaluate the PDE correctly.
For the EQR SiPM with P-on-N configuration, the triggering probability of electrons are higher than that of holes and the depletion region close to the surface benefits to detecting the blue-violet light with short penetrating depth. In applications such as Positron Emission Tomography (PET), a high sensitivity at blue and violet wavelength band is desirable due to presence of the emission spectrum of common PET scintillator (Lutetium Orthosilicate (LSO) or Lutetium-Yttrium Orthosilicate (LYSO)) peak in this part of the spectrum [15]. The P-on-N type EQR SiPM with improved blue-light detection is supposed to be used in these applications.
3.5. Dynamic range
The theoretical relationship between number of excited cells and photon numbers of a photo response pulse can be described by formula [16]:
where Nphoton represents the number of incident photons, Ntotal represents the number of total pixels in SiPM and Nfired represents the number of fired pixels.
We investigate P-on-N type EQR SiPM with pixel size of 10 μm and 90000 pixels. As shown in Fig. 7, the Nfired increase monotonously with the Nphoton. The Nfired reached to its maximum value of 85000 pixels that is very close to Ntotal (90000 pixels). The experimental results were slightly different from the theoretical results due to abnormality in micro cells. The device shows a high dynamic range that can meet most requirements for radiation detections.
3.6. Pulse area distribution
As shown in Fig. 8(a), single photoelectron peaks could be distinguish clearly that confirms good single photon resolution and uniformity for gain. The gain was calculated from the charge number of a single photon response pulse through digital oscilloscope [8]. Fig. 8(b) displayed the gain depending on over-voltage. Under over-voltage of 3.5V, the EQR SiPM had gain value as high as 2.24×105 that was equivalent to most of PMT.
3.7. Recovery time
The time needed to recharge a pixel after a breakdown which has been quenched due to the finite time taken to quench the avalanche and then to reset the diode voltage to its initial bias value is defined as the recovery time [17]. The recovery time of EQR SiPM has been characterized using double light pulse method. As shown in Fig. 9, the recovery time reach 31.1±1.8 ns when total pixels were fired. The recovery time of one APD cell was 3.1 ns [18]. The recovery time of EQR SiPM decreased with the reducing number of fired pixels. It mainly due to the influence of capacitance and resistance at readout circuit, time delay caused by transmission line. The investigation reveals that the recovery time can be shorten to decrease the operating voltages and to reduce the number of fired cells. Optimization of the surface signal collection electrode would be a feasible solution.
4. Summary
We have reported the EQR SiPM with P-on-N type diode configuration. The device with active area of 3×3 mm2 and cell size of 10μm has demonstrated gain of 2×105, dark count rate of 7MHz, crosstalk of 7%, dynamic range of 8.5×104 pixels, overall device recovery time of 32 ns at room temperature and over-voltage of 3.5V. The peak PDE is 30% at 420 nm, which perform a good balance between dynamic range and PDE. The P-on-N type EQR SiPM with improved blue-light detection can be applied in various fields where large dynamic range and high value of PDE are required simultaneously.
Characterization of a 6×6-mm2 75-μm cell MPPC suitable for the Cherenkov Telescope Array project
. Nucl. Instrum. Meth. A. 826, 31-38 (2016). doi: 10.1016/j.nima.2016.04.060Highly-Integrated CMOS Interface Circuits for SiPM-Based PET Imaging Systems
. IEEE Nucl. Sci. Conf. R, 123, 3556-3559 (2015). doi: 10.1109/NSSMIC.2012.6551814The front-end electronics and slow control of large area SiPM for the SST-1M camera developed for the CTA experiment
. Nucl. Instrum. Meth. A 830, 219-232 (2016). doi: 10.1016/j.nima.2016.05.086A novel Silicon Photomultiplier with bulk integrated quench resistors: utilization in optical detection and tracking applications for particle physics
. Nucl. Instrum. Meth. A 845, 150-153 (2017). doi: 10.1016/j.nima.2016.06.098Demonstration of a silicon photomultiplier with bulk integrated quenching resistors on epitaxial silicon
. Nucl. Instrum. Meth. A 621, 116-120 (2010). doi: 10.1016/j.nima.2010.04.040Performance of ultra-small silicon photomultiplier array with active area of 0.12mm×0.12mm
. Nucl. Instrum. Meth. A 787, 38-41 (2015). doi: 10.1016/j.nima.2014.10.068Feasibility Study on Silicon Photomultiplier With Epitaxial Quenching Resistors as the Readout for PET Detectors
. IEEE T. Nucl. Sci. 63, 17-21 (2016). doi: 10.1109/TNS.2015.2501409Research on high micro element density silicon photomultiplier based on epitaxial quenching resistance
. Aata Photonica Sinica. 46(4), 0404001 (2017). doi: 10.3788/gzxb20174604.0404001Advances in solid state photon detectors
. J. Instrum. 4, P4004 (2009). doi: 10.1088/1748-0221/4/04/P04004Enhanced blue-light sensitivity P on N silicon photomultipliers
.Performance of NUV-HD silicon photomultiplier technology
. IEEE T. Electron Dev. 63, 1111-1116 (2016). doi: 10.1109/TED.2016.2516641Precision measurements of Photon Detection Efficiency for SiPM detectors
. Nucl. Instrum. Meth. A 610, 93-97 (2009). doi: 10.1016/j.nima.2009.05.117Temperature dependence of avalanche multiplication in semiconductors
. Appl. Phys. Lett., 9, 242 (1966). doi: 10.1063/1.1754731High fill factor P-on-N silicon photomultipliers for blue light detection
.Over saturation behavior of SiPMs at high photon exposure
. Nucl. Instrum. Methods. 737, 11-18 (2014). doi: 10.1016/j.nima.2013.11.013Measurements of the recovery time of Geiger-mode avalanche photodiodes
. Nucl. Instrum. Meth. A 567, 260-263 (2006). doi: 10.1016/j.nima.2006.05.109Recovery time of silicon photomultiplier with epitaxial quenching resistors
. Instruments. 1, 5 (2017). doi: 10.3390/instruments1010005