Prepared for submission to JINST
14th Topical Seminar on Innovative Particle and Radiation Detectors
(IPRD16)
3 - 6 October 2016
Siena, Italy
The calorimeter of the Mu2e experiment at
Fermilab
N. Atanov,
a
V. Baranov,
a
J. Budagov,
a
F. Cervelli,
e
F. Colao,
b
M. Cordelli,
b
G. Corradi,
b
E. Dané,
b
Yu.I. Davydov,
a
S. Di Falco,
e,
1
E. Diociaiuti,
b,j
S. Donati,
e,g
R. Donghia,
b,k
B. Echenard,
c
K. Flood,
c
S. Giovannella,
b
V. Glagolev,
a
F. Grancagnolo,
i
F. Happacher,
b
D.G. Hitlin,
c
M. Martini,
b,d
S. Miscetti,
b
T. Miyashita,
c
L. Morescalchi,
e,f
P. Murat,
h
G. Pezzullo,
e
F. Porter,
c
F. Raffaelli,
e
T. Radicioni,
e
M. Ricci,
b,d
A. Saputi,
b
I. Sarra,
b
F. Spinella,
e
G. Tassielli,
i
V. Tereshchenko,
a
Z. Usubov,
a
R.Y. Zhu
c
a
Joint Institute for Nuclear Research, Dubna, Russia
b
Laboratori Nazionali di Frascati dell’INFN, Frascati, Italy
c
California Institute of Technology, Pasadena, United States
d
Università “Guglielmo Marconi”, Roma, Italy
e
INFN Sezione di Pisa, Pisa, Italy
f
Dipartimento di Fisica dell’Università di Siena, Siena, Italy
g
Dipartimento di Fisica dell’Università di Pisa, Pisa, Italy
h
Fermi National Laboratory, Batavia, Illinois, USA
i
INFN Sezione di Lecce, Lecce, Italy
j
Dipartimento di Fisica dell’Università di Roma Tor Vergata, Rome, Italy
k
Dipartimento di Fisica dell’Università degli Studi Roma Tre, Rome, Italy
E-mail:
stefano.difalco@pi.infn.it
1
Corresponding author.
arXiv:1701.07975v1 [physics.ins-det] 27 Jan 2017
Abstract
: The Mu2e experiment at Fermilab looks for Charged Lepton Flavor Violation
(CLFV) improving by 4 orders of magnitude the current experimental sensitivity for the
muon to electron conversion in a muonic atom. A positive signal could not be explained in
the framework of the current Standard Model of particle interactions and therefore would be
a clear indication of new physics. In 3 years of data taking, Mu2e is expected to observe less
than one background event mimicking the electron coming from muon conversion. Achiev-
ing such a level of background suppression requires a deep knowledge of the experimental
apparatus: a straw tube tracker, measuring the electron momentum and time, a cosmic ray
veto system rejecting most of cosmic ray background and a pure CsI crystal calorimeter,
that will measure time of flight, energy and impact position of the converted electron. The
calorimeter has to operate in a harsh radiation environment, in a 10
−
4
Torr vacuum and
inside a 1 T magnetic field. The results of the first qualification tests of the calorimeter
components are reported together with the energy and time performances expected from
the simulation and measured in beam tests of a small scale prototype.
Keywords:
Calorimeter, Radiation-hard detectors
Contents
1 Introduction
1
2 The Mu2e experiment
2
3 The Mu2e electromagnetic calorimeter
3
3.1 CsI crystals
5
3.2 Photosensors
6
3.3 Read out electronics
7
3.4 Energy and time calibration
8
4 Beam test of a small matrix
8
5 Calorimeter performances predicted by simulation
8
6 Conclusions and outlook
9
1 Introduction
After the discovery of lepton flavor violation in neutrino oscillation, the search for Charged
Lepton Flavor Violation (CLFV) is one of the most important activities in particle physics.
In the Standard Model of particle interactions the occurrence of such a process is predicted
to be extremely rare, far below the possible experimental reach. On the other hand, many
extensions of the Standard Model predict CLFV rates that may be observed by the next
generation of experiments [1].
The Mu2e experiment [2] at Fermilab aims to observe the neutrinoless conversion of a
muon into an electron in the field of an Aluminum atom. In this two-body process the energy
of the emerging electron is fixed (104.967 MeV) and the possible sources of background can
be very efficiently suppressed.
In 3 years of running,
∼
10
20
protons will be delivered to Mu2e and
∼
10
18
muons will
be stopped in the Aluminum stopping target. This huge amount of data will allow for a
factor
10
4
improvement on the sensitivity to the ratio between the rate of the neutrinoless
muon conversions into electrons and the rate of ordinary muon capture in Al nucleus:
R
μe
=
μ
−
+
Al
→
e
−
+
Al
μ
−
+
Al
→
ν
μ
+
Mg
(1.1)
Even in case that no signal is observed, Mu2e will achieve a remarkable result: a limit
R
μe
<
6
×
10
−
17
at 90% confidence level, that is
10
4
times better than the current limit set
by the Sindrum II experiment [3].
– 1 –
2 The Mu2e experiment
The Mu2e experimental apparatus (figure 1) consists of 3 superconducting solenoids: the
production solenoid, where an 8 GeV proton beam is sent against a tungsten target and
pions and kaons produced in the interactions are guided by a graded magnetic field towards
the transport solenoid; a transport solenoid, with a characteristic ’S’ shape, that transfers
the negative particles with the desired momentum (
∼
50 MeV) to the detector solenoid and
absorbs most of the antiprotons thanks to a thin window of low Z material that separates
the two halves of the solenoid; the detector solenoid, where the Aluminum muon stopping
target is located and a graded field directs the electrons coming from the muon conversion
to the tracker and the calorimeter.
Figure 1
. The Mu2e experiment.
The 8 GeV proton beam has the pulsed structure shown in figure 2. Each bunch
lasts
∼
250
ns and contains
∼
3
×
10
7
protons. The bunch period of
∼
1
.
7
μs
facilitates
exploitation of the time difference between the muonic Aluminum lifetime (
τ
= 864
ns)
and the prompt backgrounds due to pion radiative decays, muon decays in flight and beam
electrons, that are all concentrated within few tens of ns from the bunch arrival: a live
search window delayed by 700 ns with respect to the bunch arrival suppresses these prompt
backgrounds to a negligible level.
Figure 2
. The Proton bunches and the live search window used for data analysis.
In order to achieve the necessary background suppression, it’s important to have a frac-
– 2 –
tion of protons out of bunch, or
extinction factor
, lower than
10
−
10
. The current simulations
of the accelerator optics predict an extinction factor better than required. The extinction
factor will be continuously monitored by a dedicated detector located downstream of the
production target.
The Mu2e Tracker consists of about 21000 low mass straw tubes oriented transverse to
the solenoid axis and grouped into 18 measurement stations distributed over a distance of
∼
3 m (figu-re 3.left). Each straw tube is instrumented on both sides with TDCs to measure
the particle crossing time and ADCs to measure the specific energy loss dE/dX, that can
be used to separate the electrons from highly ionizing particles. The central hole of radius
R
∼
380
mm precludes detection of charged particles with momentum lower than
∼
50
MeV/c (figure 3.right). The core of the momentum resolution for 105 MeV electrons is
expected to be better than 180 keV/c, sufficient to suppress background electrons produced
in the decays of muons captured by Aluminum nuclei.
Figure 3
. Left: the Mu2e tracker with its 18 stations of straw tube panels. Right: section of the
tracker showing how the large amount of low momentum particles (black circles) originated in the
stopping target (yellow circle) doesn’t cross the active detector volume.
The background due to cosmic muons (
δ
rays, muon decays or misidentified muons)
is suppressed by a cosmic ray veto system covering the whole detector solenoid and half
of the transport solenoid (figure 4.left). The detector consists of four layers of polystyrene
scintillator counters interleaved with Aluminum absorbers (figure 4.right). Each scintillator
is read out via two embedded wavelength shifting fibers by silicon photomultipliers (SiPMs)
located at each end. The veto is given by the coincidence of three out of four layers. An
overall veto efficiency of 99.99% is expected. This corresponds to
∼
1 background event in 3
years of data taking. An additional rejection factor will be provided by particle identification
obtained by combining tracker and calorimeter information
1
.
3 The Mu2e electromagnetic calorimeter
The Mu2e electromagnetic calorimeter (ECAL) is needed to:
•
identify the conversion electrons;
•
provide, together with tracker, particle identification to suppress muons and pions
mimicking the conversion electrons;
1
An irreducible background of
∼
0.1 electrons induced by cosmic muons in 3 years will nonetheless survive
to particle identification [2].
– 3 –
Figure 4
. Left: the Mu2e cosmic ray veto system covering the detector solenoid and the last part
of the transport solenoid. Right: the 4 layers of scintillators interleaved with Aluminum absorbers.
•
provide a standalone trigger to measure tracker trigger and track reconstruction effi-
ciency;
•
(optional) seed the tracker pattern recognition to reduce the number of possible hit
combinations.
ECAL must operate in an harsh experimental environment:
•
a magnetic field of 1 T;
•
a vacuum of
10
−
4
Torr;
•
a maximum ionizing dose of 100 krad for the hottest region at lower radius and
∼
15
krad for the region at higher radius (integrated in 3 years including a safety factor of
3);
•
a maximum neutron fluence of
10
12
n/cm
2
(integrated in 3 years including a safety
factor of 3)
•
a high particle flux also in the live search window.
Figure 5
. The Mu2e electromagnetic calorimeter.
– 4 –
The solution adopted for the Mu2e calorimeter (figure 5) consists of two annular disks
of undoped CsI crystals placed at a relative distance of
∼
70
cm, that is approximately half
pitch of the conversion electron helix in the magnetic field. The disks have an inner radius
of 37.4 cm and an outer radius of 66 cm. The design minimizes the number of low-energy
particles that intersect the calorimeter while maintaining an high acceptance for the signal.
Each disk contains 674 undoped CsI crystals of
20
×
3
.
4
×
3
.
4
cm
3
. This granularity
has been optimized taking into account the light collection for readout photosensors, the
particles pileup, the time and energy resolution.
Each crystal is read out by two arrays of UV-extended silicon photomuliplier sensors
(SiPM). The SiPMs signal is amplified and shaped by the Front-End Electronics (FEE)
located on their back. The voltage regulators and the digital electronics, used to digitize
the signals, are located in crates disposed around the disks.
3.1 CsI crystals
The characteristics of the pure CsI crystals are reported in table 1. These crystals have
been preferred to the other candidates because of their emission frequency, well matching
the sensitivity of commercial photosensors, their good time and energy resolution and their
reasonable cost.
Table 1
. Characteristics of pure CsI crystals.
CsI
density (
g/cm
3
)
4.51
radiation length (
cm
)
1.86
Molière radius (
cm
)
3.57
interaction length (
cm
)
39.3
dE/dX (
MeV/cm
)
5.56
refractive index (
cm
)
1.95
peak luminescence (
nm
)
310
decay time (
ns
)
26
light yield (rel. to NaI)
3.6%
variation with temperature
-1.4%/
o
C
Each crystal will be wrapped with 150
μm
of Tyvek 4173D.
Quality tests on a set of pure CsI crystals from SICCAS (China), Optomaterial (Italy)
and ISMA (Ukraine) have been performed in Caltech and at the INFN Laboratori Nazionali
di Frascati (LNF)[4]. The results can be summarized as follows:
•
a light yield of 100 p.e./MeV when measured with a 2” UV extended EMI PMT;
•
an emission weighted longitudinal transmittance varying from 20% to 50% depending
on the crystal surface quality;
•
a light response uniformity corresponding to a variation of
0
.
6%
/cm
;
– 5 –
•
a decay time
τ
∼
30
ns with, in some cases, a small slow component with
τ
∼
1
.
6
μs
;
•
a light output reduction lower than
40%
after an irradiation with a total ionizing dose
of 100 krad;
•
a negligible light output reduction but a small worsening of longitudinal response
uniformity after an irradiation with a total fluence of
9
×
10
11
n/cm
2
;
•
a radiation induced readout noise in the Mu2e radiation environment equivalent to
less than 600 KeV.
3.2 Photosensors
Figure 6 shows one of the two SiPM arrays used to read each crystal.
Figure 6
. Left: one SiPM array to be used in the Mu2e calorimeter. Right: the two series
connections in the SiPM array.
The array is formed by two series of 3 SiPMs. The two series are connected in parallel
by the Front End electronics to have a x2 redundancy. The series connection reduces the
global capacitance, improving the signal decay time to less than 100 ns. It also minimizes
the output current and the power consumption.
Each SiPM has an active surface of 6x6
mm
2
and is UV-extended with a photon
detection efficiency (PDE) at the CsI emission peak (
∼
315 nm) of
∼
30%.
Tests on single SiPM prototypes from different vendors (Hamamatsu, SENSL, Advan-
sid) have been performed at LNF and INFN Pisa.
The gain is better than
10
6
at an operating voltage
V
OP
=
V
BR
+ 3
V
, where
V
BR
is
the breakdown voltage of the SiPM. When coupled in air with the CsI crystal the yield is
∼
20 p.e./MeV. The noise correspond to an additional energy resolution of
∼
100 keV.
A test of neutron irradiation with a fluence of
4
×
10
11
neutrons/
cm
2
1 MeV equivalent
2
and a SiPM temperature kept stable at 25
o
C
, has produced a dark current increase from
60
μA
to 12 mA and a gain decrease of 50%.
A test with photon radiation corresponding to a total ionizing dose of 20 krad has
produced negligible effects on gain and dark current.
2
Since SiPMs are partially shielded by the crystals, this corresponds for the SiPMs for the 3 years of
Mu2e running to a safety factor of
∼
2.
– 6 –
In order to reduce the effects of radiation damage and to keep the power consumption
at a reasonable level the SiPM temperature will be kept stable at
0
o
C
.
The qualification tests of the SiPM array preproduction are in progress at Caltech,
LNF and INFN Pisa and will evaluate:
•
the I-V characteristics of the single SiPMs and of each series;
•
the breakdown voltage
V
BR
and the operating voltage
V
OP
=
V
BR
+ 3
V
of the single
SiPMs and of each series;
•
the absolute gain and the PDE relative to a reference sensor at
V
OP
for the single
SiPMs and for each series;
•
the mean time to failure (MTTF) through an accelerated aging test at
55
o
;
•
the radiation damage due to neutron, photons and heavy ions.
3.3 Read out electronics
In the front end electronics board, directly connected to the SiPM array, the signals coming
from the two series are summed in parallel and then shaped and amplified in order to obtain
a signal similar to the one shown in figure 7.left. This shaping aims to reduce the pileup
of energy deposits due to different particles and to optimize the resolution on the particle
arrival time.
Figure 7
. Left: example of signal produced by the shaper amplifier. Right: particles pileup
observed with the waveform digitizer. The digitization threshold used for the zero suppression is
also shown.
The shaped signals are sent to a waveform digitizer boards where they are digitized at
a sampling frequency of 200 MHz using a 12 bits ADC.
The most critical components of the waveform digitizer board are: the SM2150T-
FC1152 Microsemi SmartFusion2 FPGA, the Texas Instrument ADS4229 ADC and the
Linear Tecnologies LTM8033 DC/DC converter.
The FPGA is already qualified by the vendor as SEL and SEU free and will be tested
only together with the assembled board.
The DC/DC converter, tested in a 1 T magnetic field, still maintain an efficiency of
∼
65%. Negligible effects on output voltage and efficiency have been observed after neutron
and photon irradiation corresponding to 3 years of Mu2e running.
– 7 –