test run report of the p348 collaboration · goal of the test run during sept.23-oct.7 was to test...
TRANSCRIPT
CER
N-S
PSC
-201
5-03
7/
SPSC
-SR
-172
14/1
0/20
15
Prepared for submission to SPSC
Test Run Report of the P348 Collaboration
D. Banerjeeh, Y. Chen h, P. Crivellih, E. Depero h, A.V. Dermenevb, S.V. Donskove,
R. Dusaevf , F. Dubininc, A. Gardikiotisd, S.N. Gninenkob1, A.E. Karneyeub, M.M.
Kirsanovb, N.V. Krasnikovb, V. Kromarenkoa, S.V. Kuleshovg , V.E. Lyubovitskijf , V.
Lysana, V.A. Matveeva,b, V. Mialkovskiia, Yu.V. Mikhailove, V.D. Peshehonov a, V.A.
Polyakove, A. Rubbiah, V.D. Samoylenkoe, V. Tikhomirovc, D. Tlisovb, A.N. Toropinb,
B. Vasilishinf , G. Vasquez Arenasg, P. Ulloa g, and K. Zioutasd
The P348 Collaboration
aJoint Institute for Nuclear Research, 141980 Dubna, RussiabInstitute for Nuclear Research, Moscow 117312, RussiacLebedev Physical Institute, Moscow 117312, RussiadPhysics Department, University of Patras, Patras, Greecee State Scientific Center Institute for High Energy Physics of National Research Center
’Kurchatov Institute’, 142281 Protvino, RussiafTomsk Polytechnic University, Lenin Avenue 30, 634050 Tomsk, RussiagUniversidad T’ecnica Federico Santa Mar’ia, 2390123 Valpara’iso, ChilehETH Zurich, Institute for Particle Physics, CH-8093 Zurich, Switzerland
October 14, 2015
1Contact person, [email protected]
Abstract:
The P348 Collaboration preformed test measurements in the H4 SPS beam line during
a period of 2 weeks in September-October. The measurements were designed with two main
objectives. The first was to test the new detector components such as Micromegas and
straw tube tracker, high-energy electron tagging system, ECAL, and HCAL. The second
was to explore the feasibility of searching for invisible decays of dark photons in the sub-
GeV mass range. Here we report various results of the test measurements. The main
outcome of the run is that by using simple selection criteria, without good track definition,
we reject background in the signal region down to the level of a few ×10−8 per incident
electron. As we do not see any other issues that represent fundamental problems for the
experiment, our conclusion is that after small modifications the P348 detector is ready to
start data taking in the year 2016. We recall that the A‘ with a mass in sub-GeV range is
still one of the favorable explanation of the muon g-2 anomaly. This mass range is not yet
probed by the direct experimental searches. Accumulation of about ≃ 5 × 109 e−s would
allow either to observed A‘ or completely exclude this explanation for the discrepancy next
year.
– 1 –
Contents
Table of Contents 1
1 Introduction 2
2 The P348 detector 2
2.1 The vacuum vessel 5
2.2 The SPS H4 beam line 5
2.3 Sc counters, Veto and Sc hodoscopes 5
2.4 The e−-tagging system based on synchrotron radiation 6
2.5 The shashlik electromagnetic calorimeter and preshower. 7
2.6 Hadronic calorimeter 9
2.7 Data Acquisition System (DAQ) 11
3 Preliminary data analysis 11
4 Micromegas tracker performance . 17
5 Expected sensitivity 18
6 Summary 19
– 1 –
1 Introduction
The main goal of P348 is to search for invisible decays of dark photons, A‘ → invisible in
a still unexplored region of mixing strength 10−5 . ǫ . 10−3 and A′ masses MA′ . 500
MeV by using the H4 electron beam from the CERN SPS. The A‘ with a mass in sub-GeV
range and dominant A′ → invisible is still one of the favorable explanation of the muon
g-2 anomaly which is not yet excluded by the direct experimental searches.
The process of the dark photon production and subsequent invisible decay A′ →
invisible is a rare event. For the previously mentioned parameter space, it is expected
to occur with the rate . 10−9 with respect to the ordinary photon production rate. Hence,
its observation presents a challenge for the detector design and performance. The main
goal of the test run during Sept.23-Oct.7 was to test the main conceptual design ideas of
the experiment, e.g. as synchrotron radiation tagging of the high-energy electrons, and
show that such decays could be observed by looking for events with large missing energy
at a level below 10−9 per incident electron.
2 The P348 detector
The experimental setup specifically designed to search for the A′ → invisible decays is
schematically shown in Fig. 2. The photo of the detector located at the H4 beam line
is presented in Fig.3. The experiment employs the clean 100 GeV e− H4 beam. The
χ
Z
e−e− A’
γ
χ
Figure 1. Diagram illustrating the massive A′ production in the reaction e−Z → e−ZA′ of
electrons scattering off a nuclei (A,Z) with the subsequent A′ decay into a χχ̄ pair.
admixture of hadrons in the beam (beam purity) depends on the beam intensity and varies
from ≃ 0.8 to a few 10−2. The detector is equipped with a low material budget tracker, 4
Micromagas(MM) and 4 straw tubes (ST) chambers. The 3 MM chambers were assembled
inside vacuum pipe of 159 mm in diameter and inserted inside the 1.5 T magnet upstream
of the setup. The ST chambers, with the straw diameter of 6 mm (2 ST) and 2 mm (2
ST) were delivered for tests outside magnetic field. Three scintillation counter S1,S2,S3
were used for beam definition. The setup also uses high density shashlik electromagnetic
calorimeter ECAL to detect e− primary interactions, high efficiency veto counter V, two
scintillating hodoscope H1, H2 and a hermetic hadron calorimeter HCAL located at the
downstream end of the setup in order to detect all final state products from the primary
reaction e−Z → e−ZA′. The 15 m long vacuum vessel between the magnet and the ECAL
– 2 –
Figure 2. Schematic illustration of the setup to search for A′ → invisible decays with 100 GeV e−
at H4 beam. The incident electron energy absorption in the ECAL is accompanied by the emission
of bremsstrahlung A′s in the reaction eZ → eZA′ of electrons scattering on nuclei, due to the
γ−A′ mixing. The part of the primary beam energy is deposited in the ECAL, while the remaining
fraction of the total energy is transmitted by decay light dark matter particles χ through the rest
of the detector. The χ penetrates the ECAL, veto V and the HCAL without interactions resulting
in the missing-energy signature in the detector.
Figure 3. Photo of the P348 detector at the H4 beam line.
was installed to avoid absorption of the synchrotron radiation photon detected at the
downstream end of the vessel by the array of 8 BGO crystals.
The method of the search is the following. The incident electron energy absorption
– 3 –
in the ECAL is accompanied by the emission of bremsstrahlung A′s in the reaction eZ →
eZA′ of electrons scattering on nuclei, due to the γ − A′ mixing. The diagram for the A′
production in the reaction
e−Z → e−ZA′, A′ → invisible (2.1)
is shown in Fig. 1. The reaction (2.1) typically occurs in the first few radiation length
(X0) of the calorimeter. The part of the primary beam energy is deposited in the ECAL,
while the remaining fraction of the total energy is transmitted by light dark matter decay
particles χ through the rest of the detector. The χ penetrates the ECAL, veto V and
the HCAL without interactions resulting in the missing-energy signature in the detector.
The occurrence of A′ → invisible decays produced in e−Z interactions would appear as an
excess of events with a single e-m showers in the ECAL1, Fig. 2, and zero energy deposition
in the rest of the detector, above those expected from the background sources. The signal
candidate events have the signature:
SA′ = H1 × H2 × ECAL(EECAL < E0) × V × HCAL, (2.2)
and should satisfy the following selection criteria:
• The starting point of (e-m) showers in the ECAL should be localized within a few
first X0s.
• The lateral and longitudinal shapes of the shower in the ECAL are consistent with
an electromagnetic one. The fraction of the total energy deposition in the ECAL is
f . 0.5.
• No energy deposition in the V and HCAL.
To improve the primary high energy electrons selection and additionally suppress back-
ground from the possible presence of low energy electrons in the beam typically with energy
Ee . 0.5E0 (see below), one use a high energy e−-tagging system utilizing the synchrotron
radiation (SR) from high energy electrons in a dipole magnet, as schematically shown in
Fig. 4. The basic idea is that, since the critical SR photon energy is (~ω)cγ ∝ E30 (here E0
is the beam energy) the low energy electrons in the beam could be rejected by using, e.g.
the cut on Eγ > 0.3(~ω)cγ in a gamma-ray detector, shown, for example in Fig. 4. In this
scheme, the electrons and radiation photons are detected separately. The total length of
the vacuum line is about 15 m. We consider the scheme shown in Fig. 4 for detection of the
synchrotron radiation photons in vacuum by utilizing inorganic BGO crystal with a high
light yield, see below 2.4 Note that electrons with the energy . 10 GeV which are present
in the beam before the dipole magnet will be deflected by it at a large angle, so they do
not hit the ECAL. However, such electrons could appear in the beam after the magnet due
to the muon µ → eνν or pion π → eν decays in flight in the vacuum beam pipe. Since
µs and πs do not radiate in the magnet, this source of the background is supposed to be
suppressed.
– 4 –
2.1 The vacuum vessel
The vacuum vessel is a tank of 26 cm in diameter and about 15 m length. The volume
is evacuated to the pressure below 10−3 mbar, to minimize SR photon absorption and
secondary hadronic interactions in air. The in- and output flanges made of thin Mylar
layers, about 20 mg/cm2.
2.2 The SPS H4 beam line
The experiment employs the CERN SPS H4 e− beam, which is produced in the target
T2 of the CERN SPS and transported to the detector in an evacuated beam-line tuned to
a freely adjustable beam momentum from 10 up to 300 GeV/c. A concern of the SPSC
and the technical review of the P348 experiment is whether the properties of the H4 beam
are sufficient for the proposed experiment. One of the main issue was whether the beam
intensity is sufficient to accumulate the total number of electrons in the range 1011 − 1012.
Note that to exclude the still unexplored parameter space suggested by the muon g-2
anomaly we would need to collect about 5 × 109 electrons We find that this not an issue.
The typical beam intensity we operate the setup at selected beam energy of ≃ 100 GeV, is
of the order of 5 − 6 × 105 e− for one typical SPS spill with a few 1012 protons on target.
This was obtained without any tuning of the beam line. Assuming that a typical total
number of the SPS cycles per day is 4 × 103, one get ne ≃ 3 × 109/day.
To provide as maximal as possible coverage of still unexplored area of the mixing
strength 10−5 . ǫ . 10−3 and masses MA′ . 1 GeV, and also to have realistic beam
exposure time, we plan to take measurements in 2016 with a beam of 100 GeV with the
total number of accumulated electrons on the ECAL target N etot & 1011 e−’s. Reaching this
goal requires an average beam intensity of ∼ 7×105 e− per SPS spill which is confirmed by
our measurements. Because, there are no special requirements for the small beam size at
the entrance to the detector, which could be within a few cm2, the beam intensity can be
possibly increased further by a factor, say 2 by tuning the beam line optics and collimators.
However, it would be useful to keep the hadron contamination still within a few times 10−2.
Thus, we can assume that for a realistic scenario, the total number of electrons accumulated
during one month of data taking is N etot ≃ 7 × 105 × 4500 spills/day × 30days h 1011. In
a less optimistic case, this number could be N etot h 5× 1010 electrons with the data taking
period of at least 1 month.
2.3 Sc counters, Veto and Sc hodoscopes
The e− beam is defined by :
• S1 with diameter of 42 mm, veto V1 with a hole of 40 mm against upstream beam
interactions, and X-Y scintillation hodoscope H1 with the spacial resolution of 1 mm
in the upstream the magnet
• S2 and S3 with diameter of 32 and 35 mm, respectively, and the second X-Y scin-
tillation hodoscope H2 with the spacial resolution of 1 mm in the upstream of the
ECAL.
– 5 –
The main task of these system is to measure precisely the time of arrival of particles in
order to allow the matching with hits detected in the S1 and S2 and to reject pile-up events.
The electron tracks were measured by scintillating fiber hodoscopes, arranged upstream and
downstream of the magnet. The thickness of the hodoscope planes is of order 3 mm. This
is the thickness seen by the beam electrons which will produce bremsstrahlung photons
and e+e− pairs, and has to be kept as low as possible, compatible with the proposed
performance of this sub-detector.
2.4 The e−-tagging system based on synchrotron radiation
One of the main background sources in the experiment is related to the possible presence
of the low-energy tail in the energy distribution of beam electrons. Indeed, this tail was
immediately observed during irradiation of the setup by the 100 GeV electron beam without
switching on the deflecting magnet (see below Fig.9). This tail is caused by the electron
interactions with a passive material, e.g. as entrance windows of the beam lines, residual
gas, etc... Another source of low energy electrons is due to the pion or muon decays in flight
in the beam line. The uncertainties arising from the lack of knowledge of the dead material
composition in the beam line are potentially the largest source of systematic uncertainty
in accurate calculations of the fraction and energy distribution of these events. Hence,
the sensitivity of the experiment could be determined by the presence of such electrons in
the beam, unless one takes special measures to suppress this background. To reject these
background sources at high energies by using standard techniques, e.g. threshold Cerenkov
counters, is practically impossible.
PMT
Vacuum beam pipe
Dipole magnet
e−
V
γSc
S
e−
CAL1
Figure 4. The scheme of the additional tagging of high energy electrons in the beam by using the
electron synchrotron radiation in the banding magnetic dipole. The synchrotron radiation photons
are detected by a γ - detector by using the scintillator BGO crystals (Sc). The crystal is viewed by
a high quantum efficiency PMT. The beam defining counters are also shown.
To improve the high energy electrons selection and suppress background from the pos-
sible admixture of low energy electrons, we use a tagging system utilizing the synchrotron
radiation (SR) from high energy electrons in a dipole magnet, installed upstream of the
detector. The proposed method is schematically illustrated in Fig. 4. The basic idea is
that, since the critical SR photon energy is (~ω)cγ ∝ E30 , the low energy electrons in the
– 6 –
beam could be rejected by using the cut, e.g. Eγ > 0.3(~ω)cγ , on the energy deposited in
the BGO detector shown in Fig. 4. For detection of the SR photons in vacuum we utilize
the inorganic BGO crystal with a high light yield. The crystal were arranged into the 2x4
matrix. Each crystal has hexagonal shape with the diameter of 61 mm and 200 mm long
and was viewed by EMI PMT 9603. Note, that electrons with the energy . 10 GeV will
be deflected by the magnet at an angle which is larger than those for 100 GeV e−, and,
hence do not hit the ECAL. However, low energy electrons could appear in the beam after
the magnet due to the muon µ → eνν or pion π → eν decays in flight. Since µs and πs
do not emit SR photons with energy above the cut, this source of background will also be
suppressed.
In Fig. 5 the total energy EBGO deposited in the BGO array is shown. The crystal
were calibrated by comparing the peak position from the energy deposited by minimum
ionizing particles with the one from Monte Carlo simulation, which corresponds to about
64 MeV. After imposing cut on EBGO shown in Fig.5, we observed significantly improved
Figure 5. Distribution of the total energy deposited in all BGO crystals by synchrotron radiation
photons. The detected spectrum is well agreed with the simulated one. The blue area shows events
selected with the cut 10 < EBGO < 80 MeV.
shape of the energy distribution in the ECAL from 100 GeV electrons, see Fig.6. Namely,
the low-energy tail of the distribution is completely vanished. Thus the use of the SR
tagging system will also significantly enhance the MM tracker capability.
2.5 The shashlik electromagnetic calorimeter and preshower.
The electromagnetic calorimeter shown in Fig.7 is the shashlik type detector with the
following characteristics:
• it a matrix of 6x6 cells, each with dimensions 38.2 × 38.2 × 490 mm3.
– 7 –
Figure 6. Distribution of the energy deposited in the ECAL by 100 GeV electrons with(bottom)
and without(top) cut 10 < EBGO < 80 MeV on the energy deposition in BGO array. The significant
reduction of the low-energy tail in the spectrum is clearly seen.
• each cell is (1.50 mm Pb + 1.50 mm Sc) x 150 layers or 40 radiation length (X0).
Each cell is longitudinally subdivided at preshower section of 4 X0 and the rest 36
X0 ECAL part.
• The energy resolution measured is σE/E ≃ 1.5% at 100 GeV. This is in agreement
with expected value given by σE/E ≃ 9%/√
E(GeV ) + 0.7
• The e/π rejection is . 10−3 at 100 GeV
The important characteristic feature of the ECAL is that the WLS fibers are inserted
in a spiral in order to avoid energy leak through them. Timing and energy deposition
information from each PS/ECAL module is digitized for each event. To improve the
e/π rejection factor the each calorimeter module is subdivided at two, preshower and
calorimetry, parts. A good overall e/π suppression factor . 10−3 is obtained based on
– 8 –
Figure 7. Photo of the shashlik ECAL of 6x6 modules. WLS fibers pass through the module in a
spiral and are read out at the back of the module with either a photomultipliers
analysis of first data. The use of the electromagnetic and hadronic shower profiles, both
lateral and longitudinal, and their fluctuations in the calorimeters allow to improve it
further by a factor 5-10. The results of this work will be used for further development of
the method, including event-by-event shower shape fluctuations.
2.6 Hadronic calorimeter
The hadronic calorimeter is a set of four modules. Each module is a matrix of 3x3 cells.
Each cell is a sandwich of alternating layers of Fe and scintillator plates with thicknesses of
25 mm and 4 mm, respectively, and with a lateral size of 194×192 mm2. Each cell consists
of 48 such layers and has a total thickness of ≃ 7λint. Light read-out is provided by Bicron
WLS-fibers BCF91a with diameter of 1 mm, embedded in round grooves in the scintillator
plates. The WLS-fibers from each cell (there is also a possibility to make longitudinal
granularity by collecting a few consecutive scintillator tiles) are collected together in a
single optical connector at the side of the module. Each of the 9 optical connectors per
module at the side of the module is read-out by a single photomultiplier. The 9 PMTs per
module are placed at the side of the module with the front-end-electronics. The expected
dependence of the HCAL energy resolution on beam energy in the energy range below 100
GeV is approximately given by σE
E= 0.60√
E+ 0.037. This is in agreement with measured
value of 7.4 % for 80 GeV hadron beam. The number of photoelectrons produced by a
MIP crossing the module is in the range ≃ 150-250 ph.e., obtained from the distribution
of the energy deposited in the HCAL by traversing muons with energy Eµ = 100 GeV.
The width of the muon peak is defined by the fluctuations of the total number of collected
photoelectrons nph.e. & 600 ph.e. with a rms ≃ 25 ph.e. corresponding to the total energy
of ≃ 10 GeV deposited in the HCAL. This should be compared with the effective HCAL
threshold ≃ 0.3 GeV, or ≃ 30 ph.e., for the zero-energy signal. Thus, one can see that the
– 9 –
Table 1. Main characteristics of sub-detectors used in the experiment
1. Electromagnetic calorimeters ECAL + Preshower
• module design: (1.5 mm Pb + 1.5 mm Sc) × 150 layers
• purpose: energy measurements, shower profile measurements, e/π separation
• performance: energy resolution ∆E/E ≃ 0.18/√
E, X,Y resolution ≃ 3 mm,e/π . 10−2
• event rate: up to 106 e− per spill, 1012 − 1013 e− in total run
2. Hadronic Calorimeter HCAL, 4 modules
• module design: (25 mm Fe+ 4 mm Sc) × 48 layers
• purpose: secondary energy detection, π, p, n detection
• performance: energy resolution ∆E/E ≃ 0.62/√
E, π-hermeticity ≃ 10−9
• event rate: up to 106 π per spill, 1010 − 1011 in total run
3. Beam counters S1,S2 and Hodoscopes H1,H2
• design: Sc 1mm hodoscopes
• purpose: e−e+ pair hits, track detection and T0
• performance: spacial resolution ≃ 1 mm, 2 tracks separation ∆R & 1 mm
• event rate: up to 106 e− per spill
4. Veto counter V
• design: plastic scintillator, 5 cm x 60x 60 cm3
• purpose: low energy charged track detection
• performance: mip inefficiency . 10−3
• event rate: up to 105 hits per spill
5. Synchrotron radiation counter, BGO array
• design: diam 60x200 mm thick BGO crystal
• purpose: γ ray energy mesurements in the range 1-30 MeV
• performance: energy resolution ∆E/E ≃ 3% at 1 MeV, time resolution ≃ 2 ns.
• event rate: up to 106 1-30 MeV γ per spill, ≃ 1011 for full run
6. Decay volume
• design: diameter ≃30 cm × 15 m length, filled with He or vacuum . 10−5 Torr
• purpose: minimize secondary particles interactions
7. Micromegas tracker, 4 chambers
• design: diam 80 mm
• purpose: e- track measurement
• performance: momentum resolution at 100 GeV ∆P/P ≃ 2%.
• event rate: up to a few 105 per spill
8. Straw tube chambers, 4 chambers
• design: 200x200 mm2, thickness ≃ 200 µm thick, straw tube diameter 2 and 6 mm
• purpose: e- track measurement
• performance: momentum resolution at 100 GeV ∆P/P ≃ 2%.
• event rate: up to 106 per spill
probability for an event with the MIP energy deposited in the HCAL to mimic the signal
due to fluctuations of nph.e. is negligible. The hadronic energy resolution of the HCAL
– 10 –
Figure 8. Photo of the Fe-Sc-WLS fiber HCAL modules consisting of a stack of Fe and scintillator
plates of the thickness 25 mm and 4 mm, respectively. Wavelength shifting fibers pass laterally
through the plates and are read out at the side of the module with PMT.
calorimeters as a function of the beam energy is not yet optimized due to absence of the
accurate HCAL cells calibration with hadronic beam. This procedure would take about a
couple days, so we gave it up. Finally, the description of the detector components is briefly
summarized in Table1.
2.7 Data Acquisition System (DAQ)
The DAQ system of the DM experiment is downscaled version of COMPASS FDAQ system.
The FDAQ features hardware based event builder with 4GB of DDR3 memory, PCIEx-
press interface to comercial computer, dedicated software for configuration and monitoring.
The system also includes Run control GUI, Message browser and web based interface for
configuration data base. The DAQ consists of one DHCmx module which combines data
from 3 front-end modules via SLink interfaces and combines assembled event to SpillBuffer
PCI card plugged to ReadOut Engine PC. The DAQ, in such configuration, is capable to
collect 120MBytes of data per second in sustained mode and 480 MB/s during spill. The
DAQ can receive data from up to 15 front end modules.
3 Preliminary data analysis
The occurrence of A′ → invisible decays produced in e−Z interactions would appear as
an excess of events with a single e-m showers in the ECAL , see Fig. 2, and zero energy
deposition in the rest of the detector, Veto and the HCAL. After debugging of the setup and
DAQ, in the run 2015 we recorded a run about 106 e−s at 100 GeV at a rate ≃ 4.5×105/spill
with the trigger :
Tr = S1 · S2 · S3 · V1 (3.1)
and with magnetic field off.
– 11 –
Table 2. 2D events distribution in the (EHCAL; EECAL)) plane collected with the trigger (3.1).
Source of background Rejection by
e−’s low-energy tail, Ee . 80 GeV reduction of upstream material,
SR BGO tag, good track selection
Admixture of hadrons at the level ≃ 1% preshower, SR BGO tag
Admixture of muons SR BGO tag
Bremsstrahlung γ‘s B-field on
Broad HCAL energy spectrum from e−-hadroproduction accurate HCAL calibration
The 2D distribution of collected events in the (EHCAL;EECAL)) plane is shown in
Fig.9. Immediately, we recognize several severe problems which are indicated in Fig.9 and
summarized in Table 2 with its possible solutions.
Figure 9. Distribution of 100 GeV e- events in the (EHCAL; EECAL)) plane .
At the next step to reduce the amount of parasitic beam interactions we remove from
the upstream part of the beam the hodoscope H1, beam-monitor counters, and beam Delay
chamber. Then, to check the e- beam composition at 100 GeV we recorded a sample of
events with trigger Tr = S1 · V1 and with magnet on at 0.5 T. Here V 1 is a ∼ 100 × 100
mm2 veto counter with a hole of 30 mm installed in front of S1 to suppress hallo from
upstream beam interactions. The ECAL was positioned in such a way, that the magnetic
field displaced the beam from the edge ECAL cell 6 of the central raw (corresponding to
the beam position when magnet is off) to the position between the cells 4 and 5 of the
same raw.
– 12 –
Figure 10. Distribution of the energy deposited in the ECAL cells ( by 100 e− beam deflected by
the magnetic field ≈0.5 T. Blue circle indicate the beam position at the ECAL cells ( rear view).
See text.
In Fig.10 the observed distribution of energy deposited in the ECAL cells of the central
raw is shown. It gives clear evidence of the presence of bremsstrahlung photons and low-
energy electrons in the beam. One can see that the smaller the electron energy, the farther
they are form the cell 6. These low energy tail originates presumably from the electrons
interactions with the residual gas in the beam line and also with material of the S1 counter.
The fraction of these events is . 10−3. Note that this measurement can be used as an
effective tool for tuning the beam with the minimal low-energy tail.
Then, a sample of 100 GeV e−s events with the trigger condition of (3.1), but with
magnetic field on was recorded. The B-field (≃ 1.3 T) results in displacement of the beam in
at the ECAL position by ≃ 15 cm. This required displacement of all downstream detectors:
H2, ECAL, and HCAL. The measured X-Y beam profile in the hodoscope H2 is shown in
Fig.12.
For this sample we apply the following selection cuts
• the e-m shower maximum should be in the one of the central ECAL cells.
• the total energy deposited in the BGO array should be in the range 10 < EBGO < 100
MeV (see Fig.5).
No selection of a track are used.
– 13 –
Figure 11. Distribution of events in the (EHCAL; EECAL)) plane obtained without (top plot) and
with (bottom plot) selection cuts on the value EBGO.
These simple constrains results in significant improvement of the spectra. The low-
energy tail in the ECAL vanished, as was previously reported (see comparison of distri-
butions shown in Fig.6). The 2D events distribution in the (EHCAL;EECAL)) plane is
significantly cleaned up, see Fig.11. The hadron suppression factor is & 103.
Being encouraged by these results, at the next step we record a sample of 100 GeV
events with the typical e− rate 4.5 × 105 e−/spill and the event recording rate reduced to
– 14 –
Figure 12. Distributions of e− events in X-Y planes of the hodoscope H2 upstream of the ECAL.
The X-, Y-axis are given in number of Sc strips.
Figure 13. Distributions of events in the (EHCAL; EECAL)) plane obtained with no (lhs plot) and
applied (rhs plot) selection criteria (SR BGO tag + H2 single hit). The ECAL, HCAL energy scales
are given in msADC counts, and corresponds approx. 100 GeV ≃ 3500 ch for the ECAL energy,
and 100 GeV ≃ 3000 ch for the HCAL one.
≃ 2.5 × 103/spill by the trigger:
Tr = S1 · S2 · S3 · V1 · PS(& 0.5 GeV) · ECAL(. 90 GeV). (3.2)
– 15 –
Figure 14. Photo of the Micromegas tracker which was inserted inside the magnet.
The sample corresponds to ≃ 4 × 107 incident e−s. This trigger suppressed the event rate
by a factor ≃ 200.
For this sample we apply the following selection criteria:
• the e-m shower maximum should be in the one of the central ECAL cells.
• the total energy deposited in the BGO array should be in the range 10 < EBGO < 100
MeV, see Fig.5.
• only a single hit in both X and Y H2 planes within 1 < X < 7 and 10 < Y < 15
should be detected, see Fig.12.
– 16 –
The effect of these selection cuts is clearly seen from comparison of lhs and rhs plots shown
in Fig.13. No events are observed in the signal region (EECAL . 50 GeV ;EHCAL .
0.3 GeV ) at the level, roughly a few ×10−8.
Figure 15. Distributions of difference between the measured and predicted hit position in a MM
chamber. The spacial resolution is σ ≃ 100µ m for both X and Y planes.
4 Micromegas tracker performance .
The use of Micromegas tracker, see photo in Fig.14in the run was complicated due to the
fact that MM was not integrated to the main DAQ, but used its own low rate DAQ with
the speed of 1 kHz. For comparison, the speed of DAQ used in the run is an order of
magnitude greater than that. The main DAQ system allows to operate setup with more
than 10 kHz speed. We plan to modify FE APV chip of the MM tracker in order to adapt it
with the high speed. This modification will take a few weeks. Several runs at the e− beam
energy 50 and 100 GeV were taken with the MM DAQ system in order to evaluate the
MM performance. In Fig.15 the difference between predicted and measured coordinates in
a MM chamber is shown. The predicted hit position was calculated from the extrapolation
of the track measured by two MM to the third one. The distribution of the reconstructed
beam momentum is shown in Fig. 16. As expected, the accuracy is better for the lower
– 17 –
Figure 16. Distributions of reconstructed momentum of 50 (top plot) and 100 GeV (bottom plot)
electrons in three MM chambers.
momentum tracks. For 50 and 100 GeV electrons sigma is σP ≃ 1.36 and σP ≃ 5.4 GeV ,
respectively.
5 Expected sensitivity
It has been shown previously that we are able to reject background in the signal region
to a level of a few ×108 without using information from the MM tracker. In order to
estimate the background rejection capability of the MM tracker we perform simulation of
the low-energy electron tail observed in the beam ( see Fig.9) assuming that it is originated
form the primary electron interactions in the S1 counter. In Fig.17(top) the momentum
distribution of secondary electrons form simulated 1010 interactions is shown. This number
approximately corresponds to the number of interactions in the S1 counter per 1012 passing
primary electrons.
In Fig.17(bottom) the distribution of simulated events in the plane Ptrue;Preco is
shown, where Ptrue, Preco are the simulated and reconstructed by the MM tracker elec-
tron momenta, respectively. The dangerous events are those with the Ptrue . 70 GeV but
reconstructed as electrons with Preco ≃ 100 GeV. One can see that out of 1010 interactions,
no such fake events are observed. Thus, we conclude that by using the MM tracker with
the measured performance we are able to reject low-energy electron background down in
– 18 –
Figure 17. Simulated momentum distributions of low-energy electron in the beam as observed in
data (top plot). The shaded area is the potential source of background. Bottom plot shows distri-
bution of low-energy electrons in the (Preco; Ptrue) plane, where Preco and Ptrue is the reconstructed
and true e− momentum, respectively. No fake events are observed. See, text.
the beam to the level ≃ 10−11−10−12. It is clear that using additionally the e- synchrotron
radiation tagging enhance the background rejection capability of the experiment.
6 Summary
During period Sept.23 -Oct. 7 the P348 collaboration undertook a beam study at the H4
to check whether this beam line is suitable for a sensitive search for invisible decays of dark
photons in sub-GeV mass range. Several of the potential issues suggested to us by the
SPSC were carefully checked and turn out to be solvable. Several difficulties have arisen
that we do not believe are issues in principle. The positive results reported here include
the following:
– 19 –
• The conceptual idea and design of the setup worked well and all sub-detectors worked
as expected. We are very satisfied with this.
• The new high-speed DAQ system, including the software, and the control and mon-
itoring functions works well. The new FPGA card are being tested and the control
and monitoring software is being prepared for the real hardware.
• It is found that measurements of the energy deposited in the cells of the ECAL
irradiated by electrons deflected by the magnetic field can be used as an effective tool
in order to tune the beam to the minimal low-energy tail.
• The proposed method of high-energy electron tagging by means of their synchrotron
radiation works well. By imposing cut on energy deposited in the BGO array, we
observed significant improvement in the shape of the energy distribution in the ECAL
from 100 GeV electrons. Namely, the low-energy tail of the distribution is completely
vanished. Thus, using the SR tagging system will also significantly enhance the MM
tracker capability.
• One of our concerns was that the H4 e− beam contamination by pions will affect the
sensitivity due to their decays in flight into eν pairs. We believe this is not an issue
anymore. The SR tagging combined with the spectrometer and PS+ECAL detector
capability enhances pion and other heavy particle rejection by a factor more than
1000 resulting to negligible contribution (below 10−12) from this background source.
The total background is conservatively at the level . 10−12, and is dominated by the
admixture of hadrons in the electron beam. This means that the search accumulating
up to ≃ 1012 e− events, is expected to be background free.
• One concern expressed was that the H4 cannot provide enough electron intensity
for the signal search. We observed that even without tuning of the channel we can
operate the setup at intensity in the range (5−7)×105 e−s per SPS spill. Taking into
account that the typical average number SPS spills is in the range (4000-5000)/day
results in the number of electrons accumulated ne ≃ (3− 4)× 109 e−/day. Thus, this
turns out to be not an issue.
• One concern was about the HCAL hermeticity. We found that the MIP particle
produces more than 600 ph.e. in the HCAL modules. The comparison of this number
with the HCAL threshold of ≃ 30 ph.e. ( 0.3 GeV) shows that the probability to lose
a MIP signal due to photo-statistics is well below 10−9, i.e. 20 sigma level. Taking
into account that the fraction of hadrons due to electro-production in the ECAL is
. 10−3 the background from the HCAL non-hermeticity is negligible.
On the other hand, there are several issues needing work, some of which might require
additional test time. These include the following:
• It would be important to tune the H4 to minimize the amount of upstream inter-
actions in the beam line. For example, using collimators of a definite shape can be
studied.
– 20 –
• Hadronic contamination in the beam needs to be better understood, in particular at
high electron intensity.
• While the electron intensity is sufficient, it would be good to maximize it.
• Our Micromegas detector DAQ design has to be reworked, to have improved data
recording speed and decreased ”busy” time. The best would be change the Front
End APV chip to allow full compatibility with the present DAQ configuration.
• The hodoscope H2 must be carefully re-checked. The test run shows that the effi-
ciency for the H2-Y plane is less than 70%, while the efficiency of the H1 is more
than 90% for both planes.
• The BGO array box must be redesigned to have easier access to the crystals and
electronics.
We do not believe that any of these difficulties represents fundamental problems for the
experiment. The test beam run shows encouraging results. The main goal of the dedicated
test run at H4 is achieved: the P348 detector is capable to search for invisible decays
A′ → invisible with a high sensitivity.
After testing the detector in the run 2015, we would like to slightly modify the detector
and start data taking in the next year 2016. We recall that the A‘ with a mass in sub-GeV
range and decaying mostly invisibly is still one of the favorable explanation of the muon g-2
anomaly which is not yet excluded by the direct experimental searches. Accumulating of
about (6− 7)× 109 e−s would allow either to observe of completely exclude such A‘ as the
explanation for the discrepancy next year. Accumulation of higher statistics would allow to
cover a significant fraction of the yet unexplored parameters space for the A′ → invisible
decay mode. We plan to perform a sensitive search for the decay A′ → invisible during the
period of 2016-2017(or 2018). It would require to collect data during the period of several
months assuming average electron rate of & (6 − 8) × 105 per SPS spill.
Modifications of the detector and running it in 2016 will require resources in terms
of man power, equipment and consumables which can be obtained only due to timely
applications of participating Institutes to their funding agencies. At this stage, it is crucial
for the P348 Collaboration to get positive signal from the SPSC to go ahead with these
applications.
Acknowledgments
We would like to thank Vice-DG S. Bertolucci for his support of this research, H. Wilkens,
A. Fabich, H. Fischer, I. Konorov, V. Frolov, D.-L. Lazic, M. Jeckel, J. Novy who helped
us in many ways, the CERN LCD group and the SPS crew.
– 21 –