1. Solid-State Photomultiplier for the PRIMEX PbWO4 Calorimeter
Erik Johnson1
Christopher Stapels1, Sharmistha Mukhopadhyay1, Paul Linsay1, Rory Miskimen2, Skip Augustine3,
and James Christian1
1Radiation Monitoring Devices, Inc., Watertown, MA
2University of Massachusetts, Amherst, MA
3Augustine Engineering, Encinitas, CA
Support from DOE

2. Solid-State Photomultipliers
Used for detecting light pulses from scintillation events.
Array of photodiodes readout in parallel.
Each diode has a binary response to single photons.
The response to each diode is associated with a large gain, providing good signal to noise.
The number of triggered diodes is proportional to the incident light intensity.
Radiation Monitoring Devices, Inc. has built these devices using CMOS technology, which allows integrated circuits on the same silicon die.
Explain the distributions.
Binary signal from the GPD
Charged particles will only trigger 1 pixel, maybe two.
DNP: Oct. 26, 2008

3. Geiger Photodiodes Rq - Vb + Binary signal
Optical Photon Rq - Vb
p-layer +
n-layer
SSPMs are built as an array of Geiger Photo-Diodes (GPD).
GPD is a reversed biased photodiode operated beyond the diode breakdown voltage.
Single pixel DE = Quantum Efficiency•Geiger Probability
Geiger Probability is the potential of an electron-hole pair to generate a self-sustained avalanche.
Quenching:
Passive: ballast resistor, Rq
Active: Use transistor to drop the voltage to quench the diode.
Binary signal
QE corresponds to reflection and absorption
GP corresponds to avalanche creation
DNP: Oct. 26, 2008

4. SSPM Design 2 x 2 SSPM array 1.5x1.5mm2 ea. Fill Factor: 61%
1 V: 2x106
QEmax: 48% at 520 nm
Number of Pixels: 2024
DNP: Oct. 26, 2008

5. Spectral Response Electronic noise: Small effect (large signal gain)
Energy Resolution:
Electronic noise: Small effect (large signal gain)
Scintillator: Major contribution
Detector: Various factors contribute
Photon detection (statistical fluctuations)
Thermally generated dark noise (dark counts)
Excess noise (cross-talk, after pulsing)
SSPM Statistics
DNP: Oct. 26, 2008

6. Detection Efficiency
Nuclear Instruments and Methods in Physics Research A 376 (1996)
IEEE Transactions on Nuclear Science, 55, 3 (2008)
Junction is fully depleted, QE ~ independent of excess bias.
Electron and hole ionization rates are different.
Geiger probability is dependent on whether electron or hole creates the avalanche.
DNP: Oct. 26, 2008

7. Standard Noise Sources
Cross Talk
Adjacent pixels trigger due to hot-carrier emission.
After Pulsing
Geiger pulses due to trapped carriers.
Dark Counts
Thermally excited carriers inducing an avalanche.
Magnitude of the noise sources
Fluctuations due to these sources effect the noise.
Fluctuation is proportional to the magnitude
Dark counts are handled with Poisson statistics.
Cross talk and after pulsing are handled as excess noise terms.
All sources increase with excess bias.
DNP: Oct. 26, 2008

8. Temperature Dependence
Dark Counts
Thermally excited carriers inducing an avalanche.
Cross Talk
Adjacent pixels trigger due to hot-carrier emission.
After Pulsing
Geiger pulses due to trapped carriers
Preliminary
Decreases
Dark counts: thermal excitation is suppressed.
Cross talk: hot carrier emission is reduced. (Preliminary)
Increase in after pulsing
Longer trap life times
Mitigated by two effects
Fast scintillators (integration times): sample fewer after pulses.
Noise is affected by fluctuations in output charge: the charge from after pulses is smallest near the initial pulse in time.
DNP: Oct. 26, 2008

9. Non-Linear Behavior Non-linear response of SSPM.
Non-linear transformation between number of triggered pixels and event energy.
22Na
LYSO: 1.5 x 1.5 x 3 mm3
Few Pixels
Illustration of Saturation Effect
Account for non-linear response for accurate energy resolution
An “effective integration time” may effectively increase the total number of pixels in the SSPM
DNP: Oct. 26, 2008

10. Pulsed Laser: 635-nm: ~5 ns wide
SSPM Statistics
Poisson – Good approximation
in Linear Response Region
The distribution function for the SSPM has a lower and upper bound: Use binomial statistics.
Pulsed Laser: 635-nm: ~5 ns wide
Excess Noise
Cross-talk
After Pulsing
Binomial
SSPM Statistics
Poisson
Dark Counts
Energy resolution affected by SSPM statistics when event triggers > 30% of total pixels
DNP: Oct. 26, 2008

11. Core Design Energy Resolution
2 V Excess Bias
Not Optimized
T = 0 °C
DNP: Oct. 26, 2008

12. All-Digital SSPM All-Digital SSPM
Pixel-level comparator (signal processing) & active quenching
qSSPM  Vx(T): (Linear)
Fill factor trade-off (reflected light recovery)
Provide feedback to make excess bias constant
qSSPM is constant
16% Fill Factor
400 pixels per quadrant
Testing in progress
Feedback Pixel
DNP: Oct. 26, 2008

13. Plans for PRIMEX PbWO4 SSPM PMT Cooled and Dry Vessel
Phase-I Task List
Determine the detector and readout requirements for the photodetector and ADC.
Establish a robust readout protocol for monolithic integration of detector channels.
Examine temperature dependences for SSPMs and PMTs.
Develop external ADC modules with temperature compensation.
Evaluate performance of ADC unit when coupled to PMT and SSPM.
Provide a cost analysis of design options.
Develop design concepts for SSPM integrated with an ADC.
Write Phase I report and Phase II proposal.
Key Task
Evaluate at RMD using 60Co source.
Packaging: light tight and cooled
Ship to Jefferson Lab for test beam evaluation.
Recently Started
Phase-II proposal due mid-March
PbWO4
SSPM
PMT
Cooled and Dry Vessel
DNP: Oct. 26, 2008

14. Cost Analysis Photodetector and power supply Research Prototype
Intermediate
Production
20 x 20 mm2 pixel
50k pixel device
8.2 x 8.2 mm2 Device Size 2
133
1333
Part or Process per Unit
Costs
SSPM
$5,000
$75 $8
Design and Layout
$10,000 $0
CMOS Mask Set
Included
$675 NRE
Packaging
PCB
$100
$50*
$5 - $50
Data Interface
$650 (ADI: HSC-ADC-EVALCZ)
$6* (USB Interface) $6
ADC
$220 (ADI: AD EBZ)
$60* (ADI: AD9230BCPZ-250)
$0† - $60
Power Supply Regulator
$50
$12*
$0†
Voltage Supply (AC Adapter)
$10
Assembly
$800
Total per Device
$17,130
$1013
$47 - $152
Costs are in US Dollars and are best estimates
* Bulk purchase reduction
† Inclusion in the CMOS layout
Photodetector and power supply
DNP: Oct. 26, 2008

15. Summary Solid-State Photomultipliers Future Plans Compact High Gain
CMOS: Integrated Signal Processing
Low Cost
Provide PMT-like Energy Resolution
Future Plans
Evaluation at RMD with test sources
Evaluation at JLab
Provide a more complete cost analysis (compare PMT to SSPM with integrated signal processing).
DNP: Oct. 26, 2008

16. Application with SSPMs
DNP: Oct. 26, 2008

17. Temperature Dependence
Accommodate large temperature range for useful device
Geiger Probability
Junction Capacitance
Excess Bias, Vx
Breakdown voltage, VB
Proportional to temperature (~50mV per °C)
Excess bias inversely proportional.
PG is proportional to the excess bias. (~9% per V)
Junction Capacitance, CJ (Preliminary)
Inversely proportional to temperature (0.5fF per °C)
Constant applied bias, temperature decreases:
Number of pixels increases
Output charge per pixel increases
DNP: Oct. 26, 2008

18. Device Conditions Vx V V CVx RC t t … Used for Signal Integration
Quadrant
Pixel C
(fF) R
(kW)
Recharge Time
(ns) Q1
130
160 21 Q2
270
250 67 Q3
330
290 96 Q4
170
200 34 Vx V V
CVx RC t t
Pulser
Used for Signal Integration
220 pF …
Preamp
Shaping
Amp
MCA
10 nF
substrate
2kW
- Vb +
Not Needed
DNP: Oct. 26, 2008

19. Electron/Hole Ionization
IEEE Transactions on Nuclear Science, 19, 9 (1972)
DNP: Oct. 26, 2008