Cryogenic Characterization of FBK RGB-HD SiPMs

TL;DR

This paper presents a detailed cryogenic characterization of FBK RGB-HD SiPMs, demonstrating their stable operation and low noise at temperatures from 40 K to 300 K, suitable for dark matter detection experiments.

Contribution

It provides the first comprehensive cryogenic performance data of FBK RGB-HD SiPMs, including gain and noise characteristics across a wide temperature range.

Findings

SiPMs operate effectively from 40 K to 300 K

Gain ranges from 10^5 to 10^6

Noise rates are a few Hz/mm^2

Abstract

We report on the cryogenic characterization of Red Green Blue - High Density (RGB-HD) SiPMs developed at Fondazione Bruno Kessler (FBK) as part of the DarkSide program of dark matter searches with liquid argon time projection chambers. A dedicated setup was used to measure the primary dark noise, the correlated noise, and the gain of the SiPMs at varying temperatures. A custom-made data acquisition system and analysis software were used to precisely characterize these parameters. We demonstrate that FBK RGB-HD SiPMs with low quenching resistance (RGB-HD-LR$_q$) can be operated from 40 K to 300 K with gains in the range $10^5$ to $10^6$ and noise rates on the order of a few Hz/mm$^2$.

Figures

Full text

1 Introduction

Silicon photomultipliers (SiPMs) are of special interest for the development of argon- and xenon-based cryogenic dark matter detectors, whose performance strongly depends on efficient detection of single scintillation photons. Operating SiPMs at cryogenic temperature ($87\text{\,}\mathrm{K}\text{/}$ for argon and $165\text{\,}\mathrm{K}\text{/}$ for xenon) introduces both challenges and advantages over room temperature operation.

Building on its strong history of SiPM development [1, 2, 3], FBK 222Fondazione Bruno Kessler, Trento, Italy has developed a new generation of devices, the Red Green Blue - High Density (RGB-HD) SiPM [4]. We evaluated RGB-HD SiPMs for possible use as photosensors in the DarkSide program of liquid argon time projection chamber dark matter searches [5, 6]. Among the features required for use in the DarkSide program of experiments are a low dark rate ($<$ $100\text{\,}\mathrm{mHz}\text{/}{\mathrm{mm}}^{2}$) and a total correlated noise probability lower than 60%. Both are necessary to maintain the detector energy resolution and pulse shape discrimination performance.

Cryogenic studies of SiPMs are already present in literature [7, 8]. This paper details the first study of the performance of FBK RGB-HD SiPMs in the temperature range from $40\text{\,}\mathrm{K}\text{/}300\text{\,}\mathrm{K}\text{/}$. Section 2 introduces the two variants of RGB-HD SiPMs that we tested; section 3 gives a brief overview of the cryogenic setup, the readout chain, and the analysis software (for a more detailed description, we refer the reader to Ref. [9]); finally, in section 4, we detail the results obtained with these devices.

2 RGB-HD SiPMs

An introduction to the performance of RGB-HD SiPMs can be found in [4]. Here we focus on the cryogenic performance of RGB-HD SiPMs. We studied two variants of RGB-HD SiPMs, the RGB-HD High quenching Resistor (RGB-HD-HRq) and the RGB-HD Low quenching Resistor (RGB-HD-LRq). The RGB-HD-HRq SiPMs reported here were fabricated with a SPAD size of $25\text{\times}25\text{\,}{\mathrm{\SIUnitSymbolMicro m}}^{2}\text{/}$ and the RGB-HD-LRq SiPMs had a SPAD size of $20\text{\times}20\text{\,}{\mathrm{\SIUnitSymbolMicro m}}^{2}\text{/}$. The capacitance per unit area is $50\text{\,}\mathrm{pF}\text{/}{\mathrm{mm}}^{2}$ in both cases. All the SiPMs tested were $5\text{\times}5\text{\,}{\mathrm{mm}}^{2}\text{/}$.

3 Setup and analysis

The cryogenic setup is contained in a stainless steel cryostat sealed with two DN 320 ISO-K flanges and pumped to a vacuum level of about ${10}^{-2}\text{\,}\mathrm{mbar}\text{/}$ with a Pfeiffer ACP15 multi-stage roots evacuation pump. A Cryomech PT90 pulse tube cryocooler, with $90\text{\,}\mathrm{W}\text{/}$ of cooling capacity at $77\text{\,}\mathrm{K}\text{/}$, is mounted to the top flange of the cryostat. The cold head of the cryocooler is equipped with a cold finger that holds the SiPM assembly under test, as shown in Figure 1. The system is optimized for fast thermal cycling: the cold finger can be cooled down to $40\text{\,}\mathrm{K}\text{/}$ in about $40\text{\,}\mathrm{min}\text{/}$. The cold finger is also equipped with a platinum RTD connected to a Lakeshore 335 temperature controller that supplies a set of high power metal film resistors mounted on the cold finger with the thermal load required for temperature regulation. The top flange also hosts feedthroughs for two optical fibers that are connected to an external LED and a laser light source. They can be used for measurements of the photon detection efficiency (PDE) of SiPMs, although this is not within the scope of this work and will be subject of a future study.

The readout chain is composed of a Keithley 2450 SourceMeter that serves as the bias source for the SiPM; a cryogenic pre-amplifier, based on a high speed, low-noise operational amplifier configured as a trans-impedance amplifier (TIA) with a feedback resistor of $500\text{\,}\mathrm{\SIUnitSymbolOhm}\text{/}$, resulting in a gain of $0.5\text{\,}\mathrm{mV}\text{/}\mathrm{\SIUnitSymbolMicro A}$; a single stage, non-inverting warm amplifier, configured for a gain of $28.8\text{\,}\mathrm{V}\text{/}\mathrm{V}$; and a CAEN V1751 $1\text{\,}\mathrm{GS}\text{/}\mathrm{s}$ $10\text{\,}\mathrm{bit}\text{/}$ digitizer configured for interleaved acquisition and operating in auto-trigger mode. The TIA was characterized over the full range of test temperatures to verify that it made no temperature dependent contribution to the SiPM performance [10].

A custom data analysis software developed at FBK reads the data saved by the digitizer and performs a detailed analysis of the SiPM response. For each event, the program calculates the peak amplitude and the time since the previous event and then generates a scatter plot of these two parameters. An example of a scatter plot from an RGB-HD-HRq operated at $40\text{\,}\mathrm{K}\text{/}$ and $4\text{\,}\mathrm{V}\text{/}$ over-voltage is shown in Fig. 2. From this figure it is possible to identify the noise sources that compose the response of the device:

• •

DCR: the main group of events is primary dark count rate (DCR), with an amplitude centered around 1 PE (Photo-Electron) and an exponential time distribution;

• •

DiCT: Direct CrossTalk (DiCT) events occur when, after a primary event, a photon triggers a second avalanche in a neighboring cell. Since the travel time is of the order of picoseconds, it is impossible to resolve the two events. As a result, DiCT events have a time distribution similar to that of DCR but a greater amplitude (2 or more PE);

• •

DeCT: Delayed CrossTalk (DeCT) is characterized by delay times of the order of a few to tens of nanoseconds. Such events occur when crosstalk photons are absorbed in the non-depleted region of a neighboring cell. The carriers then have to diffuse into the high-field region before triggering an avalanche. The resulting pulses have an amplitude of 1 PE but are delayed with respect to the previous ones by the characteristic diffusion time;

• •

AP: AfterPulsing events have intermediate delay times and an amplitude of 1 PE or lower. Such events occur when an electron produced in an avalanche is trapped by some impurity in the silicon lattice and is then released after a characteristic time, producing a second avalanche in the same cell. The time distribution is therefore correlated to the trap time constant and the recharge time constant of the microcell. If the time distance is lower than the latter, the AP event will have a reduced amplitude.

The breakdown voltage at each test temperature is calculated by analyzing the waveform amplitude using the DLED algorithm [11], this is done automatically by the software. The peak amplitude has a linear dependence on the applied over-voltage and allows a precise determination of $V_{bd}$ (see Fig. 3). This value is then used to correct the bias voltage so that the SiPMs are tested at the same over-voltages at each temperature.

4 Results

As discussed in Ref. [9], all FBK SiPMs are passively quenched using a polysilicon resistor. This resistance increases as temperature decreases, which leads to an increase in the single cell recharge time and hence the slow component of the SiPM pulse, $\tau_{s}$. Operation at cryogenic temperature therefore increases the length of the SiPM signal to several microseconds, leading to incomplete integration of the released charge within a $500\text{\,}\mathrm{ns}\text{/}$ gate. RGB-HD-LRq SiPMs were developed with a low resistance that depends weakly on temperature to overcome this problem. This reduces the temperature variation of the SPAD recharge time so that even at the $87\text{\,}\mathrm{K}\text{/}$ argon boiling point, the SiPM signal is fully contained within $500\text{\,}\mathrm{ns}\text{/}$. Figure 3 shows the SPAD recharge time for both the RGB-HD-HRq and RGB-HD-LRq SiPMs. At low temperatures, the RGB-HD-LRq SiPMs have a recharge time one order of magnitude faster. The effect of the pulse length variation on the charge collected within the $500\text{\,}\mathrm{ns}\text{/}$ gate is shown in Figure 4. The performance of the RGB-HD-LRq SiPM shows almost no variation, in contrast to the RGB-HD-HRq device. The fast peak of the pulse is almost unaffected by temperature. Its amplitude increases linearly with over-voltage and only very slowly with temperature for both devices, as shown in Figure 5.

The DCR as a function of temperature and over-voltage is shown in Fig. 6. When operated at low temperature, both variants show a DCR reduced by over five orders of magnitude relative to room temperature. The DCR for the two variants is of the same order of magnitude over the studied temperature range. The Arrhenius plot, shown in Figure 7, allows one to distinguish between the different mechanisms that give rise to the primary dark count rate. At high temperature (steep region), the dominant mechanism is thermal generation, which has an exponential dependence on temperature, while field-enhanced effects [12] dominate at low temperature, where the DCR reaches a plateau.

The two variants of RGB-HD technology have similar correlated noise levels. Direct cross talk, shown in Figure 8, has a weak dependence on the temperature and increases linearly with over-voltage. Overall, the direct cross talk probability is lower for RGB-HD-LRq devices. DeCT and AP events partially overlap in time, especially at high temperatures, making it difficult to distinguish between the two. It is therefore more convenient to measure their sum, as shown in Fig. 9. For both SiPM variants, the sum of the DeCT and AP is less than $10\text{\,}\mathrm{\char 37\relax}\text{/}$.

5 Conclusions

We compared the performance of two variants of RGB-HD SiPMs produced by FBK in the temperature range from $40\text{\,}\mathrm{K}\text{/}300\text{\,}\mathrm{K}\text{/}$. The RGB-HD-LRq SiPMs were shown to have a fast signal at low temperature that is fully contained within a $500\text{\,}\mathrm{ns}\text{/}$ integration gate, gains in the range ${10}^{5}{10}^{6}$, noise rates around $1\text{\,}\mathrm{Hz}\text{/}{\mathrm{mm}}^{2}$ in the temperature range from $40\text{\,}\mathrm{K}\text{/}300\text{\,}\mathrm{K}\text{/}$ and total correlated noise probabilities below $50\text{\,}\mathrm{\char 37\relax}\text{/}$, satisfying the requirements for DarkSide-20k. These features make the RGB-HD-LRq SiPMs attractive for use in the DarkSide family of experiments.

6 Acknowledgements

The development of the NUV-HD and NUV-HD-LF SiPM technologies was funded by the EU FP7 project SUBLIMA, Grant 241711. We acknowledge support from NSF (US, Grant PHY-1314507 for Princeton University), the Istituto Nazionale di Fisica Nucleare (Italy) and Laboratori Nazionali del Gran Sasso (Italy).