A poster that was displayed in a conference in Seattle that summarizes some of the work that is done by the research group that I am a part of.

1. Characterization of SiPM
for CTA Medium-Sized Telescopes
Jonathan Biteau1,3
, A. Bouvier1,2
, D. Chinn1
, D. Dang1
, K. Doyle1
, C. Johnson1
, A. Kuznetsov1,2
, D. Williams1
, W. Yeung1
for the CTA Consortium
1
Santa Cruz Institute for Particle Physics and Department of Physics, University of California
2
Now at InVisage, Inc.
3
jbiteau@ucsc.edu
Abstract
We are part of an effort to build medium-sized Schwarzschild-
Couder telescopes (SCT) exploiting silicon photomultipliers
(SiPM) for the Cherenkov Telescope Array (CTA).
We report on measurements of 1660 SiPM pixels of 6x6 mm2
that will populate the camera of a prototype SCT. For CTA, we
aim to build many similar telescopes with >11,000 pixels per
camera. We consider various competitive SiPM devices and
compare their impact on the CTA-SCT capabilities.
Outlook for the Prototype
With a collected charge scaling as PDE x Gain x (1+Xtalk), the
~4% dispersion measured in the flash responses of the 1660
pixels is consistent with the relative dispersion measured in the
PDE (~3%), gain (~2%), and crosstalk (~2%) of the sample.
No faulty pixel has been identified among the 1660 tested ones
and the assembly has started in Sep. 2014. All the components
of the camera are expected to be complete by Apr. 2015, with a
forecast commissioning on the SCT prototype in fall 2015.
SiPM to Observe the Violent Universe
Ground-based gamma-ray
telescopes observe the
Cherenkov emission
generated by gamma-ray
showers in the atmosphere.
The past years have seen
an increasing interest in
SiPM to detect Cherenkov
photons, an advantageous
alternative to PMT [1].
Prototype SCT Camera
The central sector of the prototype camera will be fully
equipped with 400 SiPM tiles from Hamamatsu Photonics,
model MPPC S12642-0404PA-050.
As in [4], we have fully characterized the properties of a sample
of these devices, by measuring their pulse shape, gain, dark
rate, and optical crosstalk, for the sixteen 3x3 mm2
chips that
compose a tile and for groups of 4 chips connected in parallel
(6x6 mm2
CTA pixels). We also measured the photodetection
efficiency (PDE) of this sample and characterized the response
to flashes of 415 tiles (1660pixels), including 15 spare devices.
Absolute Spectral Response of a Sample
Our PDE measurements are performed at a temperature stable
within ~0.1°, using a method based on zero-peak Poisson
statistics [4]. Five pulsed LEDs probe the wavelength range
375-600nm.
The PDE of 14 tiles are shown as dashed lines in Fig. 1 at four
different operating voltages. The averages and rms are shown as
solid curves. A relative dispersion of ~3% is observed at all
wavelengths and voltages. Better sensitivity below 400 nm
would be desirable considering the Cherenkov spectrum.
Relative Response of 1660 Pixels
A full characterization of the 1660 pixels being impracticable,
we have designed a setup to estimate their breakdown voltage
from current-voltage characteristics, as well as their response to
LED flashes at 365 nm. We chose an operating voltage of 3 V
above breakdown voltage, to maximize the PDE while keeping
the dark noise at a reasonable level.
The charge distributions resulting from 90 p.e. flashes (per
pixel) are stored simultaneously for the four pixels of a tile, as
well as for a reference pixel. After spatial- and temperature-
dependent calibration, based on repeated measurements of a
single tile, the normalized response of the calibration pixels
show an rms of 2.1% (see Fig. 1, dashed blue histogram),
indicative of the accuracy of the devised test.
The responses of the 1660 pixels show an rms of 4.4%,
equivalent to a 3.8% intrinsic dispersion after correction for the
accuracy of the test.
Towards the SCT contribution to CTA
The selection of Hamamatsu MPPC S12642-0404PA-050 for
the prototype was motivated by a 30-35% PDE in the blue and
a high packing fraction implied by the use of the Through
Silicon Via technology. Despite being the best choice at the
time, its level of crosstalk, ~50% at the operating point, and
pulse rise time, ~20ns before shaping for a 6x6mm2
pixel, are
not ideal, and newer devices are showing performance more
favorable for the many SCT that will be built for CTA.
With pulse widths < 10 ns, PDE in the blue reaching 40% and a
crosstalk level below 20%, latest SiPM models start to match
the technical goals of ground-based gamma-ray astronomy. An
improved spectral response below 400 nm, where the
Cherenkov spectrum peaks, and a decreased sensitivity above
550nm, where the contribution of the NSB becomes significant,
would further guide the selection of the SiPM model that will
equip the SCT.
We will continue our partnership with manufacturers to
evaluate the improvements in the SiPM technology, that will be
highly beneficial to gamma-ray astronomy and to our search for
the most energetic phenomena in the Universe.
Selecting the Photosensor for CTA
We have characterized in detail a new set of SiPM devices. For
the sake of brevity, we focus here on their spectral response.
Figure 3 shows the PDE of seven devices developed by the
Excelitas, Hamamatsu, and SensL companies, including the
best devices studied in [4] (Hamamatsu LCT-3x3-100C,
Excelitas C30742-33-050-C). Hamamatsu MPPC LCT4-3x3-
50C and SensL MicroFB-SMA-30035, with their 35-40% PDE
in the blue, their maximum optical crosstalk of 10-15%, and
their fast pulses (FWHM of 3-6 ns, for 3x3 mm2
pixels), are
particularly promising devices.
Simulated Impact of Latest SiPM
We processed simulated NSB photons and Corsika-generated
gamma-ray showers through a detector model of CTA,
accounting for the SiPM's PDE and crosstalk. Given a pixel
threshold, corresponding to an array trigger rate on the night
sky background (NSB) of 100 Hz, the image intensity at which
half of the showers are detected is defined as the trigger
threshold of the telescopes.
Crosstalk and PDE are both increasing function of overvoltage,
with PDE plateauing at large voltages. A trade off between
these increases can be found for most devices, resulting in a
minimum trigger threshold (see Fig. 4). The dark rate of the
device, which at low overvoltage is an order of magnitude
below the NSB, is not accounted for by the simulation at this
stage, and it is expected to increase the threshold at the highest
overvoltages.
The most recent SensL and Hamamatsu devices improve the
threshold by 20-30% with respect to the protoype device.
Acknowledgments
We gratefully acknowledge support from the agencies and organizations listed under Funding Agencies at this website:
http://www.cta-observatory.org/, and in particular the U.S. National Science Foundation award PHY-1229792 and the University of California.
We also acknowledge the generosity of the SiPM manufacturers Excelitas, Hamamatsu, and SensL, who provided samples tested in this work.
Bibiliography
[1] R. Mirzoyan & E. Popova, SPIE Conference Series, vol. 8621, Mar. 2013.
[2] R. A. Cameron, SPIE Conference Series, vol. 8444, Sep. 2012.
[3] J. Vanderbroucke for the CTAConsortium, sub. to Technology in Particle Physics, 2014
[4] A. Bouvier, et al., SPIE Conference Series, vol. 8852, Sep. 2013.
IEEE NSS/MIC 2014 – ID N22-3
CTA is undergoing a prototyping
phase, including construction of
an SCT of medium size with a
9.7 m-diameter primary mirror
[2]. This design provides a wide
field of view of 8° and a 30-40%
improved angular resolution
with respect to single-mirror
telescopes. This configuration
also allows for smaller cameras.
The optimal pixelation for the
prototype is about 6x6 mm2
,
matched to the point spread
function in the focal plane [3].
~ 100 m
~ 10 m
~ 1 m
~ 1 cm
~ 100 μm
Artist's concept of two gamma-ray showers
observed by CTA.
Structure of the
SCT prototype.
Exploded
view of the
prototype
camera.
SiPM tiles, Hamamatsu
MPPC S12642-0404PA-050,
chosen for the prototype.
SiPM's APD cells.
Prototype SCT camera. Only the central sector is equipped with SiPM (1600px). The
camera is divided in 9 sectors of 13 to 25 modules, in turn composed of 64 pixels.
64px
Fig. 1. Top: PDE of 14 prototype SiPM tiles as a function of wavelength
at four operating voltages. Bottom: Typical spectra of the Cherenkov
signal (blue) and irreducible noise (night sky background, red).
Fig. 2. Distribution of the response of the 1660 pixels (solid green
histogram). Multiple measurements of a single device, serving
calibration purposes, are shown by the dashed blue histogram.
Fig. 3. Top: PDE as a function of wavelength for 7 SiPM. The onset of
the PDE plateau as a function of voltage is set as the operating point.
Fig. 4. Trigger threshold of the SCT for 7 SiPM models, compared
to the threshold of the prototype (at an overvoltage of 3-3.5 V).