R.S. Miller1,*, R. Freelove1, E. Layden1

1University of Alabama in Huntsville*Corresponding Author: richard.s.miller@uah.edu

Inorganic scintillators are appropriate for nuclear γ-ray spectroscopy due to their high stopping power and light yields, both of which contribute directly into excellent detection efficiency. Maximizing science return, however, depends not only on g-ray detection but also on spectroscopic performance. Our goal is to optimize these factors to produce a device capable of meeting science-based spectroscopic performance requirements relevant for astrophysics and planetary science, while simultaneously minimizing size, weight, and power resources (SWaP).

Silicon photomultipliers (SPM) are a viable opto-electronic alternative to traditional scintillator readout schemes. Integrated into a high-resolution spectroscopy system they represent an enabling technology, providing a number of key implementation benefits such as: ruggedness, compactness, low mass, insensitivity to magnetic fields, and low bias voltage (~30V) operation. While identified originally to address power challenges, SPMs facilitate the use of low-cost scintillating materials, achieve excellent spectroscopic performance, mitigates implementation complexity, and reduce instrument mass significantly - key benefits that in turn may reduce cost.

High-Resolution Gamma-Ray Spectroscopy with Silicon Photomultipliers

Abstract Silicon Photomultiplier (SPM) SPM Parameters

Spectroscopic Optimization

Spectroscopic Performance

Ongoing Development

The fundamental limit to obtain the required energy resolution is Poisson statistics - the intrinsic statistical knowledge in the number of scintillation photons detected by a suitable sensor. In general, the maximum obtainable energy resolution (FWHM) can be parameterized as

where Nphe is the photoelectron yield of the sensor used to detect scintillation photons, and v(M) is the variance in sensor gain [11]. This fundamental limit is shown in Figure 1, along with experimentally observed resolutions for inorganic scintillators using traditional photomultiplier tubes as readout sensors.

Maximizing spectroscopic resolution requires optimization of parameters:

Maximize Nphe: • Materials with high scintillation light yield• Optimize match optical emission spectrum & detection QE• Maximize Photon Detection Efficiency (PDE)

Minimize v(M): • Reduce sensor gain variations• Reduce scintillator non-proportionality• Mitigate scintillator crystal inhomogeneities

R ≡ ∆E

E= 2.354

�1 + v(M)

Nphe

Spectral characteristics. (top) The wavelength dependent SPM PDE for previous- (dark) and latest-generation (light) SPM fabrication; (middle) emission spectrum of CsI(Tl); and (bottom) effective PDE for SPM coupled to CsI(Tl). Also shown (thin middle line) is the QE for an extended-red multialkali PMT photocathode.

Figure 7. Breakdown & optical current uniformity. (Top) The distribution of breakdown voltages; (bottom) the distribution of optical currents, a proxy for response uniformity. (left panels) first generation SPM production, (right panels) the latest production results (right). Optical current variations <±10% are obtainable.

Measured energy resolution of scintillators for 662 keV γ-rays as a function of the scintillation yield, expressed as the number of photoelectrons observed with a photomultiplier tube (v(M)=0.1).The solid curve is the theoretical lower limit governed by counting statistics.

Prototype SPM Array. The array incorporates a 3×3 matrix of SPM modules for a total of 144 pixels.

Radiation ToleranceTotal Irradiation Does (TID) Test

Source:

Energy:

Dose:

60Co

≤ 1.2 MeV

150 krad total(30 krad steps)

• Idark vs. Voltage measured after irradiation •No failure of any component•5% optical responsivity drop after 150 krad•Linear increase in dark current during irradiation (<50%)

Displacement Damage (DD) Test

Proton Fluence (p/cm2): 1010, 1011, 1012

Energy: 63 MeV

• Idark vs. Voltage measured after irradiation •10/13/60% optical responsivity drop at specified fluences• Increased dark current mitigated by cooling (thermionic)

Uniformity High Uniformity Simplifies Implementation & Minimizes Systematics

SPM detectors are manufactured using standard CMOS technology which results in highly uniform microcell breakdown characteristics, typically within ±0.06V. Such a small breakdown range is significant since it simplifies the electronics requirements for biasing large numbers of detectors. Response uniformity is also good, variations are less than ±10% max/min, a 4-fold improvement in uniformity over the previous generation of devices. High uniformity makes it possible to discriminate precise numbers of photoelectrons (i.e. photon counting) detected as distinct, discrete levels upon readout, with dynamic range limited by the number of microcells.

Optical Performance High-Sensitivity Optical Photon Collection

Improvements in fabrication techniques have led directly to increases in SPM optical responsivity. Characterized by photon detection efficiency (PDE) - the product of the deviceʼs wavelength dependent quantum efficiency (QE) and the fill factor of the photosensitive area - it can be directly compared to the QE of traditional photomultiplier photocathodes. The current generation of SPMs have peak performance at wavelengths >400 nm and improvements in fabrication techniques have led to a >2-fold improvement in SPM optical responsivity over devices produced only a year ago.

Spectral ResponseResponsivity Match to Bright ScintillatorsThalium-doped cesium iodide, CsI(Tl), is an excellent spectral match to SPM PDE.• Yield: 54,000 photons/MeV• Peak Emission: 540 nm• Hygroscopic: slightly• Primary Decay Time: 1000 ns• Non-Proportionality:

<5% above 0.1 MeV Plateauing above 1 MeV

The SPM is a novel, high gain, single photon sensitive sensor based on a summed parallel array of identical and independent Geiger-mode avalanche photodiodes and quenching resistor combined into elements called microcells. SPM detectors are manufactured using standard CMOS technology which results in highly uniform breakdown characteristics.

Each microcell is:

• Structured as a p-n diode• Provides low-noise amplification of single photoelectrons (~106 gain)• Biased above the breakdown voltage with no current flow• Photon initiates avalanche breakdown

Microcells are pseudo-binary single photon detectors

On its own a system

like this is merely a

binary switch in ei-

ther an on or off state,

and can give no pro-

portional information

regarding the magni-

tude of the incident

photon flux. Thus if

two photons deposit

energy within the

same diode at the

same time, only one

photon is registered.

To overcome this limitation an SPM sums the outputs of a large number of such microcells onto

a common load. A schematic of the SPM concept is shown in Figure 6 (center). This operational

configuration allows the SPM to have a quasi-analog output in response to the number of pho-

tons that are incident on the active sensor area.

SPM detectors are manufactured using standard CMOS technology which results in highly uni-

form breakdown characteristics to within ±0.2V. Such a small breakdown range has major sig-

nificance to simplify the electronics requirements for biasing large numbers of detectors.

In addition, due to the high uniformity of the microcells, it is possible to discriminate the precise

number of photoelectrons detected as distinct, discrete levels on the output node. The ability to

measure a well resolved photoelectron spectrum is a feature of the SPM which is generally not

possible with PMTs due to the variability in the gain, or excess noise. Examples of the quality of

SPM pulse-height measurements are shown in Figure 8. The benefit of the high-resolution and

repeatable pulse-height distributions are clear.

The dynamic range is ultimately limited by the number of microcells in the SPM, since each mi-

crocell can detect only one photon per !-ray event. Multiple photons can be detected in a single

microcell if their arrival times are separated by more than the microcell dead time of ~100 ns.

Thus, for very short light pulses (<100 ns) the dynamic range is limited by the number of micro-

cells, while for longer signals it is increased in proportion to the duration of the signal acquisi-

tion.

S I L I C O N P H O T O M U L T I P L E R A R R A Y

Commercially available SPM modules will be used throughout the proposed effort. The SPMAr-

ray4 module from Sensl, Inc. is the most mature large area device (1.3 x 1.3 cm) currently avail-

able. It is configured as a 4 x 4 array of ‘pixels’, each of which consists of 3640 microcells. The

SPMArray4 can be tiled on four sides and has a rear-mounted connector enabling access to the

16 pixel signals (see Figure 3).

- S p e c t r o m e t e r D e v e l o p m e n t i n S u p p o r t o f t h e L O C O M i s s i o n C o n c e p t

R . S . M i l l e r, U A H! N A S A - A R A 2 0 0 9

9

Figure 6. Schematics of single microcell (left) and subsection of an

SPM array ‘pixel’ (center), and a photo of a similar subsection of an ac-

tual SPM (right) [21].

Schematics of single microcell (left) and subsection of an SPM array ʻpixelʼ (center), and a photo of a similar subsection of an actual SPM (right)

Measured SPM Pulse-height (pC)

Proxy for Number of Firing

Microcells

Proxy for Number of

Incident Photons

Proxy for Incident γ-ray

Energy

Single-pixel photoelectron histograms are highly discrete, owing to the SPMʼs:

• Inherent high gain• Low multiplication noise• Uniform response within a pixel

These features enable SPM pulse-height measurements to serve as a proxy for microcell counting and as a proxy for the number of photons incident on SPM.

Prototype Array Based on Senslʼs SPMArray4

• 3×3 mm2 pixels, 144 pixels per array• Signal readout for each pixel• Summed for spectroscopy applications• Integrated digital temperature sensor for gain compensation

Laboratory measurements of prototype SPM-based spectrometer module utilizing CsI(Tl). Data shown were obtained at room temperature (~23˚C) and 10˚C using laboratory radiological standards at a bias voltage 2V above breakdown. Resolution is anticipated to improve by ~20-25% at a bias voltage 4V above breakdown, with a corresponding increase in noise - impact to spectroscopic resolution under study.

Room Temp 10˚ C

1st Generation SPMArray4:

2nd Generation SPMArray4:

SPM Array Prototype

7.5% 4.7%

− 2.6%

TBD TBD

• FWHM @ 662 keV• Dark Count: ~8MHz @ 23˚C• Dark Count: ~0.5kHz @ 10˚C

All performance results validated/duplicated with analytic model of SPM functionality

• Bias Voltage: 30V• Temperature: 23.7˚C

• Front-End Electronics - Leverage Proven Space-Qualified FEE Implementation Approaches

• Module Design, Assembly, and Thermal Modeling - Evaluation of Passive & Active Cooling Approaches, Inform Assembly Design

• Additional Radiation Tolerance Testing - Derive Impact to Spectroscopic Performance

• Instrument Performance & Simulation - GEANT-based Spectrometer Module Simulation, Incorporate Into Full Instrument Model(s)

AcknowledgementSupported in part by a LUNAR NASA Lunar Science Institute research grant

References for presented work available upon requestOptimization, combined with next-generation opto-electronic readout devices provides high-resolution, cost-effective gamma-ray spectroscopy solutions