Characterisation and performance of the PADME electromagnetic calorimeter
P. Albicocco, J. Alexander, F. Bossi, P. Branchini, B. Buonomo, C. Capoccia, E. Capitolo, G. Chiodini, A.P. Caricato, R. de Sangro, C. Di Giulio, D. Domenici, F. Ferrarotto, G. Finocchiaro, S. Fiore, L.G. Foggetta, A. Frankenthal, G. Georgiev, A. Ghigo, F. Giacchino, P. Gianotti, S. Ivanov, V. Kozhuharov, E. Leonardi, B. Liberti, E. Long, M. Martino, I. Oceano, F. Oliva, G.C. Organtini, G. Piperno, M. Raggi, F. Safai Tehrani, I. Sarra, B. Sciascia, R. Simeonov, A. Saputi, T. Spadaro, S. Spagnolo, E. Spiriti, D. Tagnani, C. Taruggi, L. Tsankov, P. Valente, E. Vilucchi
PPrepared for submission to JINST
Characterisation and performance of the PADMEelectromagnetic calorimeter
PADME collaboration
P. Albicocco, a J. Alexander, b F. Bossi, a P. Branchini, c B. Buonomo, a C. Capoccia, a E.Capitolo, a G. Chiodini, d A.P. Caricato, d , e R. de Sangro, a C. Di Giulio, a D. Domenici, a F.Ferrarotto, f G. Finocchiaro, a S. Fiore, f , g L.G. Foggetta, a A. Frankenthal, b G. Georgiev, h , a A.Ghigo, a F. Giacchino, a P. Gianotti, a S. Ivanov, h V. Kozhuharov, h , a E. Leonardi, f B. Liberti, i E.Long, j , f M. Martino, d , e I. Oceano, d , e F. Oliva, d , e G.C. Organtini, j , f G. Piperno, f , j , M.Raggi, j , f F. Safai Tehrani, f I. Sarra, a B. Sciascia, a R. Simeonov, h A. Saputi, a T. Spadaro, a S.Spagnolo, d , e E. Spiriti, a D. Tagnani, c C. Taruggi, a , k L. Tsankov, h P. Valente, f and E. Vilucchi a a INFN Laboratori Nazionali di Frascati,Via E. Fermi 54, 00044 Frascati b Department of Physics, Cornell University,109 Clark Hall, Ithaca, NY 14853 c INFN Sezione di Roma 3, Via della Vasca Navale, 84, 00146 Roma RM d INFN Sezione di Lecce,Via Provinciale per Arnesano, 73100 Lecce e Dipartimento di Matematica e Fisica, Università del Salento,Via Provinciale per Arnesano, 73100 Lecce f INFN Sezione di Roma,p.le Aldo Moro 5, 00185 Roma g ENEA centro ricerche Frascati,Via Enrico Fermi 45, 00044 Frascati (Roma) h University of Sofia St. Kl. Ohridski,15 Tsar Osvoboditel Blvd, 1504 Sofia i INFN Sezione di Roma Tor Vergata,Via della Ricerca Scientifica 1, 00133 Roma j Dipartimento di Fisica, Sapienza Università di Roma,P.le Aldo Moro 5, 00185 Roma k Dipartimento di Fisica Univerisità di Roma Tor Vergata,Via della Ricerca Scientifica 1, 00133 Roma
E-mail: [email protected] Corresponding author. a r X i v : . [ phy s i c s . i n s - d e t ] A ug bstract: The PADME experiment at the LNF Beam Test Facility searches for dark photonsproduced in the annihilation of positrons with the electrons of a fixed target. The strategy is to lookfor the reaction e + + e − → γ + A (cid:48) , where A (cid:48) is the dark photon, which cannot be observed directlyor via its decay products. The electromagnetic calorimeter plays a key role in the experiment bymeasuring the energy and position of the final-state γ . The missing four-momentum carried awayby the A (cid:48) can be evaluated from this information and the particle mass inferred. This paper presentsthe design, construction, and calibration of the PADME’s electromagnetic calorimeter. The resultsachieved in terms of equalisation, detection efficiency and energy resolution during the first phaseof the experiment demonstrate the effectiveness of the various tools used to improve the calorimeterperformance with respect to earlier prototypes.Keywords: Calorimeters, Gamma detectors, Detector design and construction technologies andmaterials ontents Na 54.2 Cosmic rays 74.3 Calorimeter performance with positron beam 10
A possible explanation for the elusiveness of dark matter (DM) is that it interacts with standardmodel (SM) particles only by means of a mediator that feebly couples to them. One of the simplesttheoretical models based on this assumption introduces a new U ( ) gauge symmetry with a vectorboson mediator, the dark photon (DP), or A (cid:48) [1, 2]. SM particles are neutral under this newsymmetry but the A (cid:48) could faintly interact with them because of a kinetic mixing with the ordinaryphoton. The strength of this interaction is given by an effective charge ε q , where q is the electriccharge of the particle and ε is the kinetic mixing coefficient.The Positron Annihilation into Dark Matter Experiment (PADME), ongoing at the LaboratoriNazionali di Frascati (LNF) of the Istituto Nazionale di Fisica Nucleare (INFN), aims to explore thistheoretical scenario. For this purpose, a dedicated detector was built, whose central component is anelectromagnetic calorimeter made of crystals of bismuth germanium oxide (BGO). The calorimeterhas high granularity and excellent detection efficiency for photons with energy between 100 and500 MeV. Sensitivity to the coupling constant down to 1 · − could be reached for values of m A (cid:48) ≤ . collecting 10 positron on target. In the following sections, a short descriptionof the experimental technique is given (section 2), followed by a detailed description of the designsolutions adopted for the calorimeter (section 3), and finally a report of the performance reached interms of efficiency and energy resolution in the 2018-19 PADME commissioning run (section 4).– 1 – alorimetersmall angle calorimeter high energypositron veto positron veto spectrometer activetargetelectron veto spectrometerimaging veto v a c u u m v e s s e l e γ R out n o t i n t e r a c t i n g b e a m + BTFvacuum Figure 1 . Top view of the PADME detector (not to scale). From right to left: active target, dipole magnet(with e + / e − vetoes inside), high-energy positron veto and the calorimeters. PADME is hosted at the Beam Test Facility of LNF [3, 4]. It looks for a DP produced via thereaction e + + e − → γ + A (cid:48) using positrons accelerated to 550 MeV by the laboratory’s LINAC,impinging on the electrons of a 100 µ m thick active diamond target, able to measure beam positionand intensity [5].The experimental technique relies on the measurement of the missing mass of the final stateswith a single photon. By measuring its energy and the direction of flight with a granular electro-magnetic calorimeter (ECal), it is possible to measure m A (cid:48) as the square of the missing mass.In addition to the ECal, the setup consists of a dipole magnet to deflect the beam away from thecalorimeter, a charged particle veto system (made of a positron veto, electron veto and high-energypositron veto [6, 7]) to identify positrons that lose part of their energy through Bremsstrahlung, anda fast small angle calorimeter (SAC) [8] to detect and veto Bremsstrahlung photons in the forwarddirection. A schematic drawing of the detector is shown in figure 1.A more detailed description of the PADME detector can be found in [9]. The segmented PADME electromagnetic calorimeter was built by reshaping BGO crystals recoveredfrom the endcaps of the calorimeter of the L3 experiment [10]. The following sub-sections presenta general description of the calorimeter, the production procedure of the scintillating units (SUs)(i.e. the assembly of a crystal and its photomultiplier tube (PMT)), the solution adopted for thesignal digitisation, and finally the trigger systems used to collect data.
ECal consists of 616 BGO crystals (2 . × . × . ) arranged in a cylindrical shape ( ≈
29 cmexternal radius) and features a central square hole (5 × igure 2 . Rear view of the ECal without back closing panel. The blue structure is the mechanical support,and the black case is the cover of the PMTs and their cables, which are shown only for few units. The purplecomponents are plastic fillers (see text for more detail). The size of the calorimeter is a compromise between large acceptance, good angular resolution,and the space available in the experimental hall. The distance between the target and the ECal is3 .
46 m, determining an angular coverage of [ . , . ] mrad. For the SAC, the covered regionis [ , . ] mrad. The numbers in brackets report respectively half of the minimum and of themaximum opening angle of a cone with vertex at the target and base contained by the front face ofthe corresponding calorimeter.The main requirements for the ECal crystals are a small Molière radius, and an energy resolutionof roughly 2% /√ E . BGO was chosen because it satisfies these conditions and because BGO crystalswere available for reuse from the L3 experiment [10].The crystal transverse dimensions were chosen to provide excellent position resolutions, whilethe longitudinal dimensions of the crystals, slightly more than 20 radiation lengths, were imposedby the original L3 crystal length.The central square hole allows the passage of forward emitted Bremsstrahlung photons, detectedby the SAC. The high rate of these photons would flood the inner ECal crystals. On the contrary,being based on Cherenkov radiator (PbF ), the SAC is able to sustain particle rates up to hundredsof MHz, thanks to a signal length of only 3 ns, much shorter with respect to BGO scintillation lightdecay time of ∼
300 ns [11].A CAD drawing of the calorimeter is presented in figure 2. The metallic support structure hasa square shape chosen to ease the detector assembly. The remaining free space between the frameand the crystals is filled with plastic elements. This also puts BGO in contact with a low-densitymaterial instead of metal where photons could produce showers with higher probability. All HVand signal cables exit from the back in groups of 64, passing through two holders, the inner onelight-tight.Due to the low energy of the photons in PADME, any single crystal support structure could spoilthe energy resolution by introducing dead materials. In assembling the calorimeter special care hasbeen taken to control size differences among the SUs which were individually selected on the basisof their dimensions. 50 µ m black Tedlar (cid:114) [12] strips were used to compensate for differences intheir heights. To reduce light cross-talk, additional Tedlar (cid:114) strips were inserted vertically between– 3 – igure 3 . BGO transparency as a function of wavelength, before (left) and after (right) annealing. Theimprovement is visible especially at lower wavelengths. adjacent crystals, while foils were places horizontally between consecutive layers. The process of building scintillating units for the ECal required different steps. First, the oldreflective paint and photodiodes were removed from the original L3 crystals. To recover performancedeterioration due to radiation damage, crystals underwent accelerated annealing at CERN’s LAB27:1. crystals were heated from room temperature to 200 °C over 3 h;2. they were kept at 200 °C for 6 h;3. finally, they were cooled down from 200 °C back to room temperature over about 1 day.Figure 3 shows measurements of the transparency before (left) and after (right) annealing. Theimprovement in transparency at lower wavelengths is clearly visible.Since the original shape of the crystals was tapered pyramidal, they were cut at Gestione SILO(Italy) [13] to have rectangular faces. The same firm also glued the PMTs (using the ELJEN EJ-500optical cement [14]) and coated the BGO with three layers ( ≈ µ m) of ELJEN EJ-510 (a brightwhite diffusive paint with titanium dioxide pigments [15]).The PMTs used in the experiment are the HZC XP1911 type B [16], with a quantum efficiencyof 21% at 480 nm (the BGO maximum emittance wavelength) and a 19 mm diameter. The gaincurve of each PMT was studied before gluing by flashing a blue LED in front of the photocathodeusing different PMT HV values, and measuring the corresponding charge collected by each tube. ECal PMT signals are digitised using CAEN V1742 boards [17, 18]. These host 4 DRS4 ASICs(a switched-capacitor array sampling chip) and provide a total of 32 channels. Each channel has adynamic range of 1 V with 12-bit precision.The board has selectable sampling frequency (1 GS/s, 2 . µ s long digitisation window able to match the long decay time of BGO scintillation light. Figure 4illustrates a typical calorimeter pulse. – 4 – ime [ns]0 200 400 600 800 1000 A DC c oun t s Figure 4 . A typical ECal signal digitized with a sampling frequency of 1 GS/s.
For each digitisation window, some samples before the BGO signal leading edge are collectedto evaluate the pulse base-line, which is obtained as the mean of the first 50 pre-pulse samples. Thetotal collected charge is then calculated summing over all the acquired samples and subtracting thebase-line value. Algorithms to correct for possible cuts on the pulse tail and for variation of theparticle energy loss are also implemented.During the data-taking two different kind of triggers are used, calibration trigger and physicstrigger:1. Cosmic ray (CR) trigger: provided by two scintillating slabs, placed above and below theECal (see section 4.2). Its purpose is to check the SUs stability using signals from minimumionizing particles (MIPs);2. Beam trigger: generated by the accelerator complex for each bunch, and used in data-takingfor physics.
SUs were calibrated with a Na source before mounting in the calorimeter. This was done todetermine the charge vs HV curve used to set the SU voltage to the desired gain. After theinstallation of ECal in the experimental hall, the CR trigger was also used to check the response ofthe units to MIPs. Using CR events, it is possible to validate the Na calibration and to assess andimprove the SUs equalisation. Na To select and equalise the response of each SU, a dedicated setup was built exploiting the twoback-to-back 511 keV photons emitted by a Na source. It allows the scan of a 5 × igure 5 . Setup for SUs calibration. Left: sketch of the test stand (not to scale), with the Na sourceindicated with a yellow disk (possible motions are also shown). A LYSO crystal is placed on the oppositeside of the source to provide the trigger signal. Right: a photograph of the test stand, with a stack of SUs. / ndf c ped A 17.8 – ped m – ped s – A 4.1 – m – s – – Charge [pC]100 150 200 250 300 C oun t s / ( . p C ) / ndf c ped A 17.8 – ped m – ped s – A 4.1 – m – s – – HV [V]1100 1200 1300 1400 1500 C ha r ge [ p C ] / ndf c - – - – c - – - – Figure 6 . Left: charge distribution of a SU at 1400 V fit with a double Gaussian plus a flat function toreproduce the pedestal, the 511 keV peak, and the constant background. Right: Measured charge as a functionof HV with the gain curve fit Q = A · V s superimposed. path highlighted and figure 5 (right) shows a photograph of the test stand. A 3 × ×
20 mm LYSO crystal, read out by a SiPM, is placed on the opposite side of the source with respect to theBGO and produces the trigger signal. Data were collected at 10 different voltages in the interval [ , ] V in steps of 50 V, acquiring about 6000 events per HV value.An example of the charge distribution obtained for a unit at 1400 V is presented in figure 6(left). The pedestal and the 511 keV signal are clearly visible and both are fitted with a Gaussian,while the approximately flat continuum is modelled with a constant function.The collected charge, for each value of the HV, was determined as the difference between the511 keV peak and pedestal position. Figure 6 (right) presents the charge vs HV behaviour with thebest-fit curve superimposed (same SU of figure 6 (left)). The fitting curve has the form Q = A · V s ,where Q is charge, V is voltage, and A and s are free parameters.In figure 7, the distribution of voltages needed to obtain a gain of 15 . ean 1186Std Dev 52.97 HV [V]1050 1100 1150 1200 1250 1300 1350 1400 C oun t s / ( . V ) Mean 1186Std Dev 52.97
Figure 7 . Distribution of PMT voltages used to set the gain of the corresponding SU to 15 . Mean 0.5878Std Dev 0.6724
HV relative difference [%]3 - - - C oun t s / ( . % ) Mean 0.5878Std Dev 0.6724
Figure 8 . Relative difference between voltage measurements for the 135 SUs that were tested a second timeto assess gain curve reproducibility. The required gain is 15 . identical, measurement campaign. The relative difference between the two optimal HVs, givenby V − V ( V + V )/ , remains below a few percent as shown in figure 8. This difference also incorporatespossible temperature variations between the two measurements. Since the start of data-taking, the calorimeter has been operated with voltages corresponding to again of 15 . igure 9 . Structure and logic of the ECal CR trigger. It consists of two plastic scintillator slabs, one aboveand one below the ECal. Each bar is read out by two PMTs, one per side, set in logic AND. The logic OR ofthe two ANDs gives the trigger signal. C ha r ge [ p C ]
297 346 178 256 378 256 377 324 123 260 220 269 279 131 192 376 171 239 262 169 370 406 344 500 414 183 300 277 292
Figure 10 . A cosmic ray passing through the ECal. The color scale and numbers inside the squares representthe charge collected by the SU. vertically are considered. This is defined by three conditions:• a cosmic ray passes through three SUs aligned in a column;• there are no other signals in the three rows of the SUs under consideration;• only the pulse from the central SU is considered for the calculation.This ensures that the cosmic ray releases little energy in adjacent crystals. For example, infigure 10, only SUs in position ( , ) and ( , ) satisfy these requirements. For peripheral units– 8 – / ndf c –
375 MPV 1.1 – – Charge [pC]0 200 400 600 800 1000 1200 1400 1600 C oun t s / C / ndf c –
375 MPV 1.1 – – Figure 11 . Example of charge distribution given by CRs passing vertically through a SU. A Landau fit issuperimposed. without any SUs directly above or below (68, corresponding to 11% of the total), the two crystalsbelow or above are used, respectively.An example of the charge distribution produced by this selection, with a superimposed Landaufit, is given in figure 11. Landau parameters are those defined by the ROOT package [19] and thealgorithm is from CERNLIB G110 DENLAN, which is based on [20].Figure 12 presents the most probable values (MPVs) from Landau fits performed on chargedistributions obtained from CR data collected over three days. On the left, the 2D distribution(with values and color scale) is shown and on the right two histograms are superimposed: the bluehistogram corresponds to all active SUs while the red one excludes the 68 peripheral units. Themeans of the two Gaussian fits are compatible: ( . ± . ) pC and ( . ± . ) pC, respectively.Taking into account the Gaussian for all the units, the ECal equalisation achieved using the gaincurves from the Na source indicates a signal spread of ( . ± . ) %, where the spread isexpressed as the ratio of the sigma and of the mean of the Gaussian fit.Empty squares on the left plot of figure 12 indicate 4 non-operational channels. If one of thesebelongs to the triplet needed to define the verticality condition, it is skipped and the one aboveor below is used. This study also helps to improve the calorimeter energy resolution. Assumingthat, on average, CRs release the same amount of energy in all crystals, the MPV of each chargedistribution can be used as a normalisation term, by simply dividing each pulse integral by the MPVof the corresponding ECal channel.Finally, CRs are also used to evaluate the SU efficiency. With the same triplets defined in theprevious analysis, the efficiency of the central SU can be determined. The efficiency is definedas the ratio between the number of events with hits in all three cells and that with hits only inthe two external ones. Figure 13 (left) shows the 2D distribution of ECal efficiencies. They havebeen evaluated using a CR data set taken over three days. Figure 13 (right) reports the measuredefficiencies with a zoom on the interval [ , ] % in the inset. A reverse Landau fit is superimposedon both distributions.Figure 14 reports the cumulative efficiency distribution of SUs. The percentage is given without– 9 – C ha r ge [ p C ]
238 257 368 357 256492 328 563 432 489 665 423 308 509 236 266264 222 277 261 397 284 269 251 455 307 256 300 263 252 305279 300 302 280 279 286 289 274 262 317 298 248 339 269 327 320 341 256 341255 287 231 286 273 247 277 300 298 216 304 243 262 272 262 282 240 281 277 294284 381 331 277 237 285 236 252 294 294 345 248 336 276 236 287 243 257 273 253 349 369 209280 256 240 223 268 244 240 242 286 278 277 291 224 308 240 233 249 276 280 228 272 260 283251 257 237 285 250 273 228 318 268 275 215 252 273 258 221 289 247 247 339 246 240 289 234 250 404262 258 232 230 244 287 304 330 255 237 284 250 278 205 286 374 281 238 259 294 302 203 318 266241 306 248 235 355 226 224 238 280 310 293 191 291 255 248 240 203 227 290 248 232 252 271 251 261 254 354285 250 263 233 279 258 279 275 226 280 287 232 331 265 256 289 266 203 274 259 269 258 231 276 266 304252 275 228 262 276 220 336 260 442 291 243 270 252 232 220 237 246 229 268 283 248 248 248 251 298 251 288291 239 291 363 253 268 259 275 266 280 270 294 241 313 203 234 269 246 286 314 263 337 288 296327 280 285 259 249 298 277 273 268 276 281 282 272 246 297 281 263 275 279 257 242 271 337 273282 292 247 225 254 255 224 288 281 235 290 234 221 263 252 215 284 229 243 293 275 216 410 279276 281 249 331 263 238 276 250 211 308 325 246 264 242 232 249 245 261 251 313 230 256 275 273303 288 268 310 326 290 268 262 244 252 262 237 304 260 243 236 353 245 256 265 211 240 233 317268 341 280 274 382 278 311 296 199 278 274 282 342 210 282 268 200 253 269 286 224 246 231 313 388 286 270251 238 257 270 263 254 247 308 302 328 274 248 257 262 259 231 239 226 250 318 222 271 251 375 276 254 293343 260 298 293 245 337 308 256 239 233 243 248 276 329 275 256 259 469 215 264 275 216 265 282 255 317 272271 290 311 258 277 318 263 248 278 333 229 271 294 212 256 228 315 249 291 239 261 251 238 268 267290 315 340 304 250 298 323 293 302 271 262 269 321 232 274 277 238 266 272 238 206 265 276 264 349304 247 442 299 381 221 269 117 262 265 248 266 275 267 261 280 276 329 195 284 319 294 293241 284 314 245 262 237 215 348 231 282 270 228 297 278 244 246 231 301 250 234 242 271 272201 362 272 284 276 233 270 240 269 226 271 242 290 250 236 279 259 271 278 269 232376 260 253 224 318 237 302 244 267 291 257 241 252 266 219 398 273 320255 238 320 281 317 316 273 233 277 315 351 288 416 287 283296 416 421 263 303 255 332 277 123 355 428274 245 228 314 251 / ndf χ all A 1.75 ± all µ ± all σ ± Charge [pC]0 100 200 300 400 500 600 700 C oun t s / ( p C ) / ndf χ all A 1.75 ± all µ ± all σ ± χ inner A 1.67 ± inner µ ± inner σ ± χ inner A 1.67 ± inner µ ± inner σ ± Figure 12 . MPVs obtained from Landau fits to the charge distributions produced by CRs passing verticallythrough SUs. Left: map of charge measurement in the ECal. The 4 white squares correspond to non-operational SUs. Right: MPV distributions considering (blue) and not considering (red) the 68 edge crystals. E ff i c i en cy [ % ] / ndf χ ± ± σ ± Efficiency [%]0 20 40 60 80 100 C oun t s / ( . % ) / ndf χ ± ± σ ± Efficiency [%]
97 97.5 98 98.5 99 99.5 100 100.5 101 C oun t s / ( . % ) Figure 13 . Efficiency of SUs, evaluated using CR data. Left: ECal efficiency map. Right: efficiencydistribution with a reversed Landau fit superimposed (see text). In the inset, a zoom on the region [ , ] %.Both plots exclude the 4 non-operational units and the 68 edge units. considering the 4 non-operational units and the 68 edge SUs. Before ECal construction, several tests were performed on a prototype consisting in a 5 × . × . × . ) read out by XP1912 HZC PMTs. In particular, single-positronbeams of different energies were fired at the central crystal of the matrix. These measurementsreveal an energy resolution compatible with the desired performance [21]: σ ( E ) E = . (cid:112) E [ GeV ] ⊕ . E [ GeV ] ⊕ . . – 10 – fficiency [%]0 20 40 60 80 100 C u m u l a t i v e [ % ] -
10 110 Figure 14 . Cumulative efficiency distribution of figure 13, in percent. The percentage is given withoutconsidering the 4 non-operational units and the 68 edge units.
Energy (MeV)200 300 400 500 600 700 800 900 1000 1100 ( E ) / E σ / ndf χ ± − ± − ± χ ± − ± − ± Figure 15 . Energy resolution measured with the ECal prototype (red and blue points) [21] compared withthe ECal result on special run data set (black point).
Figure 15 presents the energy resolution measured during these tests. Blue (red) squaresidentify results obtained with a 250 MeV (450 MeV) beam energy. During the same test the chargeresponse as a function of the deposited energy has been shown to be linear in within 2% up to1 GeV.To evaluate the energy resolution of ECal a special run was performed with a 490 MeV positronbeam energy (multiplicity ≈
1) fired directly at the calorimeter. A 5 × . ± . ( stat ) %, as shown by the black circle in figure 15. Thisimprovement was partially expected since the prototype detector was slightly different from thefinal calorimeter. Several factors contributed to improve the performance of the ECal with respectto the prototype:• The ECal has longer crystals and therefore a better energy containment;• PMTs are glued to the BGO while in the prototype they were optically coupled with optical– 11 –rease;• ECal crystals are painted and not simply wrapped with PTFE tape;• The ECal SUs responses are equalised a priori while at the test beam all SUs were operatedat the same voltage of 1100 V.The beam characteristics also contributes to the measured energy resolution: the beam usedfor the prototype test was not optimised as in the standard data-taking. In fact, in order to be able tochange its energy, electrons produced on a secondary target were used. This was necessary since themain user of the LINAC primary beam was the DA Φ NE LNF collider requiring a fixed energy (550MeV). A primary beam, 10 particles/bunch, was hitting a secondary target and electrons of desiredenergy, produced by the electromagnetic showers, were then selected with a dipole magnet. Dueto the high multiplicity, the showers also produced many photons that could reach the prototypeunder test, inducing background. Conversely, in the ECal test, the beam was a primary beamconsisting only of 490 MeV positrons. In this case the LINAC gun was off and only few electronswere accelerated, the ones produced by the gun’s dark current in phase with the accelerating field.This very low multiplicity helped reducing the beam-induced photon background reaching the ECalimproving the measured energy resolution. The primary goal of the PADME experiment is to measure with high efficiency and precision theSM photons produced by the reaction e + + e − → γ + A (cid:48) , where the dark photon A (cid:48) signal appearsas missing mass. This is obtained by means of an electromagnetic calorimeter made of 616 BGOcrystals.Each SU was calibrated with a Na source before installation in order to set the working pointto 15 . ( . ± . ) %, while the efficiency is ≥
98% for 99 .
1% of the channels.Finally, to check the energy resolution, a dedicated measurement was performed by firing asingle positron beam of 490 MeV energy at the calorimeter. The obtained resolution is 2 . ± . ( stat ) %, with an evident improvement compared to the values obtained with a calorimeterprototype. Acknowledgments
The PADME collaboration wants to warmly thank the BTF and LINAC teams, for the excellentquality of the beam provided during the test run. The PADME collaboration wants to thanks thetechnical staff of LNF for the work done. Among the others a special mention goes to R. Lenci, andG. Papalino for their help in assembling and cabling the detector.University of Sofia group is partially supported under BG-NSF DN08-14/14.12.2016 and theMoU SU – LNF-INFN 70-06-497/07-10-2014.– 12 – eferences [1] P. Galison and A. Manohar,
Two Z’s or not two Z’s? , Phys. Lett. B (1984) 279[2] B. Holdom,
Two U(1)’s and (cid:15) charge shifts , Phys. Lett. B (1986) 196[3] G. Mazzitelli et al.,
Commissioning of the DAΦNE beam test facility , Nucl. Instrum. Methods A (2003) 524[4] P. Valente et al.,
Linear Accelerator Test Facility at LNF Conceptual Design Report , INFN-16-04-LNF , arXiv:1603.05651 (2016)[5] R. Assiro et al.,
Performance of the diamond active target prototype for the PADME experiment at theDAΦNE BTF , Nucl. Instrum. Methods A (2018) 105[6] F. Ferrarotto et al.,
Performance of the Prototype of the Charged-Particle Veto System of the PADMEExperiment , IEEE Trans. Nucl. Sci. (2018) 8[7] S. Ivanov and V. Kozhuharov, The charged particle veto system of the PADME experiment , AIP Conf.Proc. (2019) 080005[8] A. Frankenthal et al.,
Characterization and performance of PADME’s Cherenkov-based small-anglecalorimeter , Nucl. Instrum. and Methods A (2019) 89[9] M. Raggi and V. Kozhuharov,
Proposal to Search for a Dark Photon in Positron on Target Collisionsat DAΦNE Linac , Adv. High Energy Phys. (2014) 959802[10] B. Adeva et al.,
The construction of the L3 experiment , Nucl. Instrum. Methods A (1990) 35[11] P.A. Zyla et al. (Particle Data Group), to be published in Prog. Theor. Exp. Phys. 2020, 083C01(2020).[12] [13] [14] [15] [16] [17] [18] E. Leonardi et al.,
Development and test of a DRS4-based DAQ system for the PADME experiment atthe DAΦNE BTF , J. Phys. Conf. Ser. (2017) 032024[19]
ROOT Data Analysis Framework , https://root.cern.ch// [20] K.S. Kölbig and B.Schorr, A program package for the Landau distribution , Computer Phys. Comm. (1984) 97 [Erratum-ibid. 178 (2008) 972][21] M. Raggi et al., Performance of the PADME Calorimeter prototype at the DAΦNE BTF , Nucl.Instrum. Methods A (2017) 31[22] E. Leonardi, G. Piperno and M. Raggi,
Evaluation of clustering algorithms at the < GeV energyscale for the electromagnetic calorimeter of the PADME experiment , J. Phys. Conf. Ser. (2017)072019(2017)072019