The PADME detector

To search for the production of a dark photon ( A′ ) in the process e+e−→A′γ , the PADME apparatus has been built athe INFN Laboratori Nazionali di Frascati. It is a small-scale detector consisting of an active target, a beam monitor system, a spectrometer to measure charged particle momenta in the range 50–400 MeV, a dipole magnet to deflect the primary positron beam out of the spectrometer and allow charged particle momentum analysis and an electromagnetic calorimetric system to detect signal and background photons produced in the annihilations with high accuracy. Each element has specific requirements that are stringent and sometimes at the limit of present technology. This proceeding will give an overview of each component and a description of the technical solutions implemented to accomplish the experiment needs. Results of the commissioning data taking, performed from October 2018 to February 2019, will be illustrated.


Introduction
The Standard Model (SM) of particle physics is one of the greatest achievements of last century. Nevertheless, some phenomena, such as neutrino oscillations and matter-antimatter asymmetry, cannot be explained in the context of SM. Some cosmological phenomena, like galaxy rotational velocity [1], gravitational lensing [2], or cosmic microwave background anisotropies [3], can be explained introducing a new kind of matter, which interacts gravitationally with regular SM matter. This new entity is commonly known as dark matter (DM). Over the last few years many experiments have tried to detect DM in different ways, however no signal of a DM candidate has yet been reproduced by independent experiments. A possible solution to this problem is that DM could live in a separate sector with respect to the one where SM particles live. This different sector, called the dark sector (DS) is described by many theories in different ways. The simplest theory introduces a new interaction mediated by a vector boson particle, called the dark photon ( ¢ A ) in analogy with the SM photon [4]. The mediator could also act like a portal between the DS and the SM particle sector. The coupling constant ò of the new interaction should be quite small (ò = 1), in order to take into account the elusiveness of DM. The simplest theory of this kind introduces a DP that only couples to leptons [5].
A DP could be produced in different processes. Assuming that the ¢ A couples to leptons, there are three possible ways of producing it: (iii) meson decay, for example from a π 0 , p g  ¢ A 0 .
Depending on its mass ( ¢ m A ), the DP could decay into SM particles or into DM particles. If the ¢ A is the lightest existing DM particle, it will only be able to decay in SM particles. In this case, the DP will only be detectable through its decays to SM particles, through the visible search technique. If DM particles χ with < c ¢ m m 2 A exist, the DP will mainly decay to DM particles. In this case, the DM decay products will be undetected, and DP will only be identified through the invisible search technique.
2. Detection of a dark photon in PADME PADME (Positron Annihilation into Dark Mediator Experiment) [6][7][8] is searching for a DP which couples to leptons by studying the annihilation of a positron beam with electrons in a fixed target. In the following, the physics case of the experiment is briefly described, and the detector is illustrated in full.
The process under study is: where the e + comes from a positron beam whose energy can be up to 550 MeV, the e − is the electron of a fixed target, ¢ A is the DP and γ is the SM photon. Measuring the four-momentum of the SM photon P γ , the kinematics of the annihilation can be fully reconstructed, since the electrons are at rest and the four-momentum of the positrons P beam is determined by magnetic selection and measured to be 0.3% by the width of the outcoming beam at the TimePix detector downstream of the PADME dipole. The existence of DP could be then inferred from a peak in the missing mass-squared distribution M Miss 2 (figure 1) of the process: A signal event in PADME is represented by a single SM photon in the electromagnetic calorimeter, and no other particles in the other subdetectors. The main backgrounds are SM annihilation e + e − → γγ, which can mimic the signal if one photon is undetected, and Bremsstrahlung e ± N → e ± Nγ.  The nominal beam used by PADME is a 550 MeV positron beam, with a multiplicity of ∼20 k particles per bunch, 49 Hz repetition rate, and bunch duration ∼300 ns ( [10]). DP masses 23.7 MeV can be reached using a beam with these properties.

The detector
In figure 3 a top view of the detector is given. The positron beam enters the experimental hall, passes through two dipoles and two quadrupoles, and hits the active diamond target where the annihilation takes place. After the collision, charged particles are deflected by a magnetic field (∼0.5 T) in vacuum (10 −5 mbar) towards the veto system, placed on the two sides of the vacuum chamber. The most energetic charged particles are bent towards the high energy veto and the beam monitor, which detects particles from the beam that have not interacted in the target, while photons reach the two calorimeters of the experiment.
In the following, each part of the detector is described.
2.2.1. The active diamond target PADME is the only experiment to date to explore the dark sector using an active fixed target, and assuming DP couples to leptons. The target is an active part of the detector, being able to give information about the beam characteristics. It consists of Chemical Vapour Deposition 20 × 20 × 20 mm 3 polycrystal diamond, on which 16 × 16 graphitic strips (x and y coordinates) have been created by INFN Lecce (figure 4, left) [11]. Carbonʼs low atomic number reduces the contribution to the background from Bremsstrahlung (σ Brems ∝ Z 2 ). The target is placed in vacuum (figure 4, right) and a remotely controlled stepper motor allows it to be moved on/off beam. The target gives information on beam intensity and measures the interaction point. The linearity between the collected charge with respect to the number of positrons per bunch was estimated through dedicated measurements (see section 3.1).

Figure 2.
Top view of the BTF: PADME is located in front of BTF-1 line, while BTF-2, still under construction, is reserved for external users. PADME is highlighted with a blue frame.  figure 5, left, and can be moved on/off beam remotely to provide beam characterization. In this configuration, it is possible to measure the divergence of the incoming beam as well as its position (figure 5, right) and the multiplicity. MIMOSA is usually placed on beam before a physics run in order to have information about the beam.

The charged-particle veto system
Bremsstrahlung radiation e ± N → e ± Nγ is one of the main sources of background to the experiment. For this reason, the detector must be equipped with a veto system able to identify these events. The PADME veto system is located in three different regions of the detector: the Positron Veto (PVeto) and the Electron Veto (EVeto) are placed on the two sides of the vacuum chamber, inside the magnetic field (∼0.5 T), while the High Energy Positron Veto (HEPVeto) is placed near the beam exit. Each veto subdetector consists of 10 × 10 × 178 mm 3 plastic scintillating bars (90 for PVeto, 96 for EVeto, 16 for HEPVeto, figure 6, left). Each bar is connected to a wavelength shifter (WLS) fiber inserted into a groove along the bar length and coupled to a Hamamatsu S13360 3 × 3 mm 2 (25 μm pixel width) SiPM, two at the opposite extremities at each bar in HEPVeto. Bremsstrahlung rejection can be obtained if good efficiency and a time resolution better than 1 ns are guaranteed. Since the small angle calorimeter was designed to detect γ from Bremsstrahlung (see section 2.2.5), time coincidence measurements between the calorimeter and the PVeto are performed. The time difference distribution is reported on the right side in figure 6. True coincidences (Gaussian peak) are superimposed on a flat continuum of accidentals. The fit of the Gaussian peak give an estimate of the uncertainty in measuring this coincidence.

The electromagnetic calorimeter
The electromagnetic calorimeter (ECal) is the subdetector in charge of the detection of the SM photon released in the process A . This is crucial to fully reconstruct the kinematics of the process, in order to produce the missing mass-squared distribution.
The ECal is a cylindrical segmented calorimeter ( figure 7, left), with a square hole in the center. It consists of 616 21 × 21 × 230 mm 3 BGO crystals, optically glued to HZC Photonics XP1911 photomultipliers (figure 7, right). The shape of the calorimeter was designed to minimize the contribution from Bremsstrahlung radiation (e ± N → e ± Nγ), which is emitted with a narrow angle with respect to the beam direction. This background could   mimic the ¢ A signal if the e ± leaves the detector undetected. The hole at the center of the ECal allows the Bremsstrahlung radiation to reach the much faster small angle calorimeter, which was designed to cope with the high rate of Bremsstrahlung (see section 2.2.5).
BGO crystals were recovered from the L3 experiment. They were polished and then cut in a parallelepiped shape, then underwent an annealing procedure in order to recover their transparency to factory condition. The annealing process consists of heating crystals to 200°C, maintaining them at this temperature for 6 hours, and then allowing them cool down to room temperature. According to studies done by the L3 collaboration, the energy resolution of the crystals σE/E was expected to lie in the range (1-2)%/ for electrons and photons of energy 1 GeV [12]. This value was confirmed by measurements performed by the PADME collaboration (see section 3.2).

The small angle calorimeter.
Bremsstrahlung radiation can reach a rate of ∼100 MHz, which is unsustainable for the decay time of BGO crystals (∼300 ns). For this reason, a faster calorimeter is needed, in order to efficiently reject this background. The small angle calorimeter (SAC) is placed behind the central hole in the ECal (figure 8, right) to fulfil this role. The SAC consists of 25 PbF 2 crystals, coupled to Hamamatsu R13478UV photomultipliers ( figure 8, left). PbF 2 crystals emit Cherenkov radiation, which is much faster than scintillation radiation, allowing double-peak separation ∼2 ns [13].

The TimePix3 beam monitor
The non-interacting beam can give online information regarding its characteristics. For this reason, a TimePix3 silicon pixel detector [14] is placed at the end of the experiment, monitoring the exiting beam ( figure 9, left). The detector consists of 12 sensors, each made of a 256 × 256 pixel matrix (14 × 14 mm 2 ). The total surface covered by the detector is 8.4 × 2.8 cm 2 . The detector provides information about beam spread and helps in the  reconstruction of the number of positrons on target, beam size (x and y), time and beam structure (figure 9, right).

PADME trigger and DAQ
The trigger in PADME is distributed to each subdetector by two kinds of custom boards: the CPU trigger board and the trigger distribution boards.
The CPU trigger board provides three kind of trigger signals: (i) the accelerator trigger, given by the beam entering the experimental hall; (ii) the cosmic-ray trigger, which is given by the coincidence of plastic scintillators placed above and below the two calorimeters (used for calibration measurements); (iii) the random trigger, used for pedestal studies.
The three trigger signals are used at the same time during standard data-taking. This makes it possible to select data corresponding to specific trigger type offline in order to perform dedicated measurements.
30 CAEN V1742 boards digitize signals for each subdetector (excluding TimePix3 and MIMOSA, which have a dedicated DAQ system). Each board has 32 channels, and is equipped with 4 DRS4 chips. The dynamic range of each channel is 1 V, with a 12 bit precision (1024 capacitor cells per channel). The sampling rate can be set to 1 GS/s, 2.5 GS/s or 5 GS/s. Depending on the detector, different board configurations are used. Data are collected by a two-level readout system. Level 0 PCs collect data from each board, performing zero suppression if required. Level 1 PCs perform event merging and process raw data from Level 0 into ROOT [15] files.

Data taking and calibration studies
The first PADME data-taking run started on October the 4th 2018 and lasted approximately 4 months. The integrated number of particles was 10 12 positrons on target. The main technical purposes of the run were: • calibration of each subdetectors; • optimization of beam configuration; • background studies for the data analysis;

Detector calibration
In order to calibrate the target response, measurements with different beam multiplicity were performed.
During the target calibration measurement, the beam multiplicity was changed in order to study the response of the detector. The average of the total charge collected by the X and Y views is shown in figure 10, as a function of the multiplicity provided by the BTF calorimeter. The target response was extremely linear.
The first ECal calibration was performed on the scintillating units before assembly of the detector, using a 22 Na source. Once the detector was assembled, cosmic rays were used to test the reliability of this first calibration. Two plastic scintillators were placed above and below the ECal, and the charge distributions obtained from cosmic rays crossing the scintillating units were analyzed. Variations of the response of the scintillating units to minimum ionizing particles are within 11%.
The SAC calibration was performed firing a single-positron beam on each of the 9 central crystals of the detector (those reachable through ECal central hole). A set of dimensionless multiplicative calibration constants was extracted from these measurements, that was later compared with a second set obtained from cosmic-rays measurements. The results, in figure 11, show a good agreement between the two sets (10% for all the units, except one that was substituted later). The calibration constants for the crystals in the external region will be evaluated with different cosmic-ray data sets.

ECal: reconstruction and energy resolution
The performance of the ECal was analyzed in terms of clusters. The seed is the crystal where the majority of the energy is released, and the cluster is the 5 × 5 matrix around the seed. Scintillating units inside a cluster must be in time (±10 ns around seed time) and the measured energy must be greater than 1 MeV. In order to measure the energy resolution of ECal, a special run without the target and with a single-positron beam of 490 MeV was performed. The beam was fired directly on the calorimeter crystals. In figure 12 a first result for the energy resolution of the ECal (black dot) is shown with respect to the energy resolution of a prototype (red and blue   The curve is the best-fit to the red and the blue points. The parameters a, b, c in the statistics box are respectively the stochastic term, the noise term and the constant term of the energy resolution formula for calorimeters dots) [16]. The prototype was a 5 × 5 matrix of scintillating units. The prototype had slightly shorter BGO crystals (20mm versus the final 21 mm), that were coupled to PMT with grease, instead of optical glue, and were wrapped with PTFE tape, while the calorimeter crystals are painted with reflective paint. Because of these differences, that improve the energy containment in the calorimeter, the energy resolution of the detector is clearly better than that of the prototype being 2.62 ± 0.05(stat)% at 490 MeV [17].

Beam optimization
During the July 2019 data-taking run, an unexpected source of beam induced background was observed. The main cause turned out to be the beam hitting the beryllium window which separated the BTF vacuum from the PADME vacuum. A MC simulation including the beryllium window, the magnet geometry of the beamline and the target support was performed. The background energy distribution of the data was very similar to the distribution from the Monte Carlo when using a beam with an energy spread of 1.5 MeV (see figure 13). In order to reduce beam induced background, the beryllium window separating the BTF vacuum from the PADME vacuum was removed to be placed in a different place in the beam line at the end of July 2019. Unfortunately, the window broke during this operation, requiring the whole beamline to be fully decontaminated. This accident caused the collaborationʼs planned activities to be delayed by approximately one year. After the decontamination, the beryllium window was substituted with a thinner Mylar window. This material needs to be changed more often than beryllium, but itʼs not toxic. In addition, the window was placed further upstream along the beamline to reduce the background still further.
A new technical run was performed in July 2020, with the primary goal of finding the best beam configuration to reduce the background as much as possible. For this purpose, the way positrons are obtained in BTF was also changed. The configuration used for the first data taking exploited a 'secondary beam' where positrons are produced by the collision of accelerated electrons on a target and then selected in energy (maximum energy: 550 MeV). In July 2020, the collaboration moved to a 'primary beam' configuration, where positrons are accelerated inside the LINAC to the desired energy after the production (maximum energy: 490 MeV). Preliminary studies using the primary beam have been performed, showing significantly reduced beam-induced background with respect to the secondary beam. For this reason, the collaboration intends to use the primary beam configuration for the second data-taking run.

Conclusions
PADME is searching for a dark photon, which could be the mediator of a new interaction between Standard Model particles and dark matter particles. One possible production process of a dark photon is the annihilation g  ¢ +e e A , where the Standard Model photon in the final state must be detected in order to fully reconstruct the kinematics of the process and produce missing mass-squared distributions, where the existence of a dark photon would appear as a peak in the region of ∼400 MeV 2 . A 550 MeV positron beam colliding on an active diamond target is used for this purpose. The commissioning data-taking run started in October 2018 and the data collected show beam-induced background which was reproduced by the Monte Carlo simulation of the experiment once the whole beamline was introduced into the simulation. A failure during an intervention on the beamline to reduce the background caused a delay of ∼1 year on the collaborationʼs planned activities. Figure 13. Energy deposited in the ECAL due to beam induced background: real data (left) and Monte Carlo (right). On the left-hand plot, DHSTB002 is the last magnet of the beamline before the beam arrives to the target (different colours in this plot refer to two different days). The beryllium window in the simulation gave a distribution more similar to the data energy distribution, but still not in perfect agreement.
Calibration data were collected during July 2020, and a second data-taking run was performed at the end of 2020.

Data availability statement
No new data were created or analysed in this study.