Performance of the CMS Zero Degree Calorimeters in pPb collisions at the LHC
O. Surányi, A. Al-Bataineh, J. Bowen, S. Cooper, M. Csanád, V. Hagopian, D. Ingram, C. Ferraioli, T. Grassi, R. Kellogg, E. Laird, G. Martinez, W. McBrayer, A. Mestvirishvili, A. Mignerey, M. Murray, M. Nagy, Y. Onel, F. Siklér, M. Toms, G. Veres, Q. Wang
PPerformance of the CMS Zero Degree Calorimeters in pPbcollisions at the LHC
O. Sur´anyi , A. Al-Bataineh , J. Bowen , S. Cooper , M. Csan´ad , V. Hagopian ,D. Ingram , C. Ferraioli , T. Grassi , R. Kellogg , E. Laird , G. Martinez , W. McBrayer ,A. Mestvirishvili , A. Mignerey , M. Murray , M. Nagy , Y. Onel , F. Sikl´er , M. Toms ,G. Veres , Q. Wang MTA-ELTE Lend¨ulet CMS Particle and Nuclear Physics Group, E¨otv¨os Lor´and University,Budapest, Hungary Wigner RCP, Budapest, Hungary University of Kansas, Lawrence, USA University of Maryland, College Park, USA University of Iowa, Iowa City, USA Florida State University, Tallahassee, USA University of Alabama, Tuscaloosa, USA NRC Kurchatov Institute (ITEP), Moscow, Russia Brown University, Providence, USA* [email protected]
Abstract
The two Zero Degree Calorimeters (ZDCs) of the CMS experiment are located at ±
140 m fromthe collision point and detect neutral particles in the | η | > . Many measurements involving proton-ion and heavy-ion collisions require the knowledge ofthe centrality of the collision [1, 2]. One way to determine this is by measuring the numberof nucleons that do not participate in the collision. The SPS, RHIC, and LHC heavy-ionexperiments have measured these spectator nucleons with Zero Degree Calorimeters (ZDCs).The CMS ZDCs are two identical forward calorimeters located between the two LHC beampipes at a distance of approximately 140 m from the CMS interaction point along the beamline,on each side. There are numerous other applications of ZDC detectors, such as minimumbias triggering, study of ultraperipheral collisions, and charge exchange processes. This paperpresents results demonstrating the performance of the ZDCs in the 2016 pPb data-takingperiod. The results presented here are based on a sample of 10 million minimum bias events1 a r X i v : . [ h e p - e x ] F e b AD sectionBRANEMsection FibersPMTsLight guides
Beam directionEM sectionHAD section
HAD1 HAD2 HAD3HAD4
EM1EM2 EM3 EM4 EM5
Figure 1: The schematic side-view (left) and segmentation (right) of the CMS ZDC.collected at a center-of-mass energy of √ s NN = 8 .
16 TeV. The ZDC detects the neutralfragments of the Pb ions, and the neutrons emitted from the ions are nearly monoenergetic,thus they provide a unique opportunity to study the performance of the detector. The paperfirst introduces the structure of the ZDC detectors, then a Monte Carlo simulation study of thebehaviour of the detector is presented. Afterwards the signal extraction and the calibrationprocess is discussed. Finally, a method based on Fourier transformation is presented to correctthe measured spectrum for pileup collisions.
The ZDCs of the CMS experiment complement the main CMS detector especially for heavyion studies. They reside in special detector slots in the neutral particle absorber (TAN), whichprotects the first superconducting quadrupole magnet from radiation. A full description ofthe ZDC can be found in [3–6]. Located inside the TAN at pseudorapidity η greater than 8.3corresponding roughly to θ < . X . λ Number of channels 5 horizontal divisions 4 longitudinal segmentsModule size (W × L × H), mm 92 × ×
705 92 × × ≈ ≈ and Λ particles can also reach the ZDCs.The basic properties of the EM and HAD sections are summarized in Table 1. The EM sectionis segmented into 5 vertical strips that allows the determination of the horizontal positionof the incoming particles. Tungsten plates and fibers in the EM section run vertically. Thehadronic section is divided into 4 segments, as seen in Fig. 1. In the hadronic section, thetungsten plates are tilted by 45 ◦ to optimize the collection of Cherenkov light. The quartzfibers are clad in doped quartz, yielding a numerical aperature of 0.22. Individual fiberribbons are grouped together to form a readout bundle that is compressed and glued intoa circular shape. A light guide carries the light through radiation shielding to HamamatsuR7525 photomultiplier tubes. Between the two ZDC sections lies an ionization chamber calledBRAN (Beam RAte for Neutrals), which gives a measurement of the instantaneous luminositywhich is independent of the operation of CMS [7].For each collision event, the signal is collected over 10 time slices (TS) of 25 ns each. Thepeak of the ZDC signal always occurs in TS3, whereas due to the 100 ns bunch spacing,out-of-time pileup signals may be present four timeslices before and after the main signal.The main signal extracted from TS3. The high voltage powering the PMTs is set such thatthe analog-digital converters may saturate for larger signals. In this case, information fromthe tail of the signal is used to determine the total signal value. This preserves the excellentfew neutron resolution if the number of neutrons is low (for example in an ultra-peripheralcollision), while at the same time allowing the entire range of the ZDC to be exploited forcentrality measurement using the tail of the signal. The full ZDC geometry, including the BRAN detector, is modeled within the geant w EM weighting factor. This factor is calculated byminimizing the relative energy resolution, defined as the ratio of the standard deviation andthe mean of the measured energy values. First this calculation is performed using only theevents with the shower starting in the EM section. The best resolution is achieved by using3 AD sectionBRANEMsection
Figure 2: The ZDC detector geometry used for the geant w EM = 0 .
42, and the corresponding energy distributions are shown in the left panel of Fig.4. In the corrected ZDC response, it is found that there is a shift between the two peaks asshowers starting in the EM section are partly absorbed by the material of the BRAN detector.By the comparison of the position of the peaks it is concluded that approximately an averageof 20% of the energy is lost in those events that start to shower in the EM section. In caseof real pPb collisions, in most cases more than one neutron is produced. Some of them maystart showering already in the EM section, whereas others have the first interaction only inthe HAD section. Therefore it is not possible to treat these two cases separately as they willbe inevitably mixed, when in a real collision several neutrons hit the ZDC simultaneously.Alternatively, it is also possible to minimize the total resolution, resulting in w EM = 0 .
61. Thecorresponding energy distribution is shown in the right panel of Fig. 4. This is not the optimalfactor for the neutrons which shower in the EM, but this is the best overall resolution whichcan be achieved with the detector.The signal detected in ZDC is dominated by neutrons emitted from the colliding Pb nuclei.The three main sources of these neutrons are nuclear evaporation processes [9, 10], intranuclearcascades [9, 10] and neutrons emitted due to electromagnetic nuclear excitations such as thegiant dipole resonances (GDR) [11–13]. These neutrons are simulated in order to study ZDCacceptance and response. It is assumed that the neutrons are emitted isotropically in the restframe of the nucleus according to the Maxwell-Boltzmann momentum distribution:d N d p ∝ p exp (cid:18) − p m n T (cid:19) , (1)where p is the total momentum, m n is the neutron mass and T is the Maxwell-Boltzmanntemperature. The values of the T parameter are 1, 5, and 50 MeV for neutrons originatingfrom electromagnetic excitation [14], evaporation [10] and intranuclear cascade [10] processesrespectively. All neutrons are boosted in the z-direction by γ = 2752, which is the Lorentz-factor of the Pb ion in pPb collisions at √ s NN = 8 .
16 TeV.The effect of crossing angle, beam divergence and the smearing of the beamspot aretaken into account by applying the following procedure on all generated neutrons. First4 [TeV] had + E em E N o r m a li z ed y i e l d / . T e V All events, resolution: 21.4%Start in EM, resolution: 26.0%Start in HAD, resolution: 14.6%
CMS
Simulation [TeV] had + E em E [ T e V ] e m E N o r m a li z ed y i e l d - - - CMS
Simulation
Figure 3: The ZDC response for 2.56 TeV energy neutrons, separately plotted for neutronswhich start to shower in the EM section and the HAD section (left), and the dependence ofthe response on energy deposited in the EM section (right).the location of the interaction point ( v x , v y , v z ) is sampled from a Gaussian beamspot withposition ( x , y , z ) and size ( σ x , σ y , σ z ). The effect of beam divergence is taken into accountby introducing the M ( z ) magnification factor: M ( z ) = σ ( z ) σ (0) = (cid:115) εβ ( z ) εβ ∗ = (cid:115) z β ∗ , (2)where σ ( z ) is the transverse size of the beam at distance z from the interaction point if nofocusing is used, ε is the beam emittance and β ( z ) is the beta-function, with β (0) = β ∗ . Asthe beamspot is magnified by M (140 m) = M ZDC , the projected impact point ( v x, ZDC , v y, ZDC )on the ZDC surface is calculated as v x, ZDC = M ZDC · ( v x − x ) + x , (3) v y, ZDC = M ZDC · ( v y − y ) + y . (4)Finally the direction vector calculated from the impact point and the interaction point isrotated by half of the α crossing angle in the x-z plane. The beamspot and beam parametersare summarized in Table 2.The projected impact points for the three assumed neutron emission scenarios are shownin Fig. 5. The corresponding geometrical acceptance is larger than 98% for all processes. Thegenerated and observed energy distribution for three different types of very forward neutronsare summarized in Fig. 6. It can be concluded that in the case of evaporation and GDRneutrons, the resolution is dominated by the detector response, whereas for cascade neutronsthe energy spread dominates due to the large Maxwell-Boltzmann temperature.5 [TeV] had + E em E · N o r m a li z ed y i e l d / . T e V All events, resolution: 18.8%Start in EM, resolution: 14.6%Start in HAD, resolution: 14.6%
CMS
Simulation [TeV] had + E em E · N o r m a li z ed y i e l d / . T e V All events, resolution: 17.1%Start in EM, resolution: 17.0%Start in HAD, resolution: 14.6%
CMS
Simulation
Figure 4: The ZDC response for 2.56 TeV energy neutrons, with all EM energy depositsweighted by w EM = 0 .
42 (left) and 0 .
61 (right), which were calculated by minimizing theresolution using events that shower in the EM section and all events respectively.Table 2: Beam and beamspot parameters used in the simulation.Parameter Value β ∗
60 cm α µ rad x y z
16 mm σ x σ y σ x
47 mm
A typical signal shape in a given channel i is shown in Fig. 7. A simple way to extract the a i signal amplitude corresponding to this shape is: a i = q i [3] − q ped ,i , (5)where q i [ t ] is the charge value in the t timeslice, and q ped ,i is the pedestal calculated as q ped ,i = 12 [ q i [0] + q i [1]] . (6)The signals in TS0 and TS1 are used in the pedestal estimation to minimize the inclusion ofthe tail of the main signal. However, when a pre-pileup signal is also present, this methodbecomes unreliable as it will overestimate the pedestal value. When the signal is saturated,6 o r m a li z ed y i e l d - · x [mm] - - - y [ mm ] - - - CMS
Simulation N o r m a li z ed y i e l d - · x [mm] - - - y [ mm ] - - - CMS
Simulation N o r m a li z ed y i e l d - · x [mm] - - - y [ mm ] - - - CMS
Simulation
Figure 5: The projected impact points on the ZDC surface of neutrons originating from GDR(left), evaporation (middle) and intranuclear cascade (right) processes.
E [TeV] N o r m a li z ed y i e l d / . T e V CMS
Simulation
E [TeV] N o r m a li z ed y i e l d / . T e V CMS
Simulation
E [TeV] N o r m a li z ed y i e l d / . T e V CMS
Simulation
Figure 6: The measured energy distributions of neutrons emitted from a 2.56 TeV energy Pbion via GDR (left), evaporation (middle) and intranuclear cascade (right) processes.the ZDC signal tail is calculated, defined as a tail i = R i · q i [4] − q ped ,i , (7)where the R i factors are calculated from the distributions of ( q i [3] − q ped ,i ) / ( q i [4] − q ped ,i )values in non-saturating signals.In order to treat the events with feed-off from pre-pileup signals, a template fitting methodsimilar to [15] is used. In the following description of this method the channel indices i aredropped for the sake of simplicity and vector notation is used: all vector indices correspondto a given timeslice. Due to the 100 ns bunch spacing, out-of-time pileup signals may bepresent in TS7 (post-pileup) and in the timeslice preceding TS0 (pre-pileup). In order to beable to fully model the pre-pileup shape and eliminate all contribution from the post-pileupsignal, only the first six timeslices are used in the fit, thus all of the following vectors are6-dimensional. The measured signal values in a single event are denoted by q , whereas t and t (cid:48) stands for the main and the pre-pileup signal template respectively. The template for a givenchannel is constructed by averaging many signal shapes from which the pedestal describedby Eq. (6) is subtracted in each time slice and their integral is fixed to unity in TS3. In theaveraging those events are used, that have larger than 4000 fC signal in TS3 and have no pre-or post-pileup present. The pre-pileup events are rejected by requiring TS0 and TS1 to have7 Timeslice [25 ns] C ha r ge [f C ] HAD1pPb 8.16 TeV
CMS
Figure 7: A typical ZDC signal shape.less than 50 fC charge, whereas the post-pileup events are rejected by requiring charge valuesdecreasing monotonically from TS5. The average template shapes for the different channelsare shown in Fig. 8. The first six timeslices are denoted as t , whereas the values from TS4 toTS9 are used to construct t (cid:48) .The amplitude of the signal is then calculated by minimizing the following χ expression: χ = ( q − a t − b t (cid:48) − c ) T V − ( q − a t − b t (cid:48) − c ) , (8)where V is the covariance matrix, is a 6-element vector with all components equal to 1, a isthe main signal amplitude, b is the amplitude of the pre-pileup signal and c is the pedestal.This minimization is a fit, with a , b , and c as free parameters. There are three contributionsto V : (i) the digitization uncertainty of the measured signal, (ii) the fluctuations of thepedestal, where the off-diagonal elements should also be considered, and (iii) the uncertaintyof template shapes due to digitization and the uncertainty in the timing of the signals. Theterm corresponding to the digitization uncertainty is approximated as: V dig ,ij = δ ij · ∆ q i , (9)where ∆ q i is the width of the charge range corresponding to the measured digital valueprovided by the analog-digital converter in timeslice i and δ ij denotes the Kronecker delta.The covariance term of pedestal fluctuations is calculated from non-collision events using thesample covariance formula: V ped ,ij ≈ (cid:80) Nk =1 ( q ki − ¯ q )( q kj − ¯ q ) N − , (10)where q ki is the signal value in the timeslice i in the event k , ¯ q is the average noise level,and N is the total number of events. An example for a V ped matrix is shown in the leftpanel of Fig. 9. The large off-diagonal elements indicate a low frequency variation of the8 imeslice [25 ns] T e m p l a t e v a l ue - - -
10 1 EM1EM2EM3EM4EM5 pPb 8.16 TeV
CMS
Timeslice [25 ns] T e m p l a t e v a l ue - - -
10 1 HAD1HAD2HAD3HAD4 pPb 8.16 TeV
CMS
Figure 8: Average signal shapes of EM channels (left) and HAD channels (right). The first sixtimeslices are taken as the template of the main signal, whereas the values from TS4 to TS9are used to construct the template of the pre-pileup signals.pedestal. The pulse shape covariance matrices V shp and V (cid:48) shp , corresponding to the main andpre-pileup signal respectively, are calculated similarly using the collision events that were usedfor the determination of the template, and are shown in the middle and right panel of Fig. 9respectively. The final covariance matrix is defined as V = V dig + V ped + a V shp + b V (cid:48) shp . (11)Since the parameters a and b introduce a fourth order term in the χ expression, they areestimated as a ≈ q [3] and b ≈ R i q [0], therefore they do not spoil the linearity of the equationsderived above.The optimal parameter values can be calculated by taking the partial derivatives of thisexpression with respect to the parameters:0 = d χ d a = − tV − ( q − a t − b t (cid:48) − c ) , (12)0 = d χ d b = − t (cid:48) V − ( q − a t − b t (cid:48) − c ) , (13)0 = d χ d c = − − ( q − a t − b t (cid:48) − c ) . (14)Now A , v and x are defined as A = t T V − t t (cid:48) T V − t 1 T V − tt T V − t (cid:48) t (cid:48) T V − t (cid:48) T V − t (cid:48) t T V − (cid:48) T V − T V − , v = q T V − tq T V − t (cid:48) q T V − , x = abc . (15)and the optimal parameters can be calculated by solving the Ax = v (16)9 imeslice [25 ns] T i m e s li c e [ n s ] ] [f C ped H A D , V pPb 8.16 TeV CMS
Timeslice [25 ns] T i m e s li c e [ n s ] s hp H A D , V - · pPb 8.16 TeV CMS
Timeslice [25 ns] T i m e s li c e [ n s ] s hp H A D , V ' - · pPb 8.16 TeV CMS
Figure 9: Covariance matrices of pedestal fluctuations (left), in-time (middle) and pre-pileuppulse shape (right) of HAD1 channel. The covariance matrices of other channels look similar.The unit of the elements of pedestal covariance matrix is fC , whereas the pulse shape matricesdo not have a unit as they are calculated from normalized signal shapes.linear equation. Two example fit results are shown in Fig. 10.This method can be generalized for signals saturating in TS3 by omitting the elementsfrom all vectors and matrices corresponding to TS3 and adding a penalty term to (8). Let ˆq , ˆt , and ˆt (cid:48) be the vector of signal and template values in each time slice, except TS3 – thusthey are 5-dimensional vectors. Then the ˆ χ expression to minimize isˆ χ = ( ˆq − a ˆt − b ˆt (cid:48) − c ) T ˆV − ( ˆq − a ˆt − b ˆt (cid:48) − c ) + χ , (17)where, using the q s saturation value, the penalty term is: χ = (cid:40) − (cid:104) − erf (cid:16) q s − a · t [3] − b · t (cid:48) [3] − c √ V (cid:17)(cid:105) , if a · t [3] + b · t (cid:48) [3] + c < q s , , if a · t [3] + b · t (cid:48) [3] + c ≥ q s . (18)This term is introduced to penalize those fit functions that predict a smaller value than q s inTS3. As a result, the fit function will be closer to q s . The first part of χ is approximatedby a second order polynomial, calculated using the Maclaurin expansions of log(1 − x ) anderf( x ), therefore the minimization of ˆ χ can also be carried similarly as the minimization of(8), by solving a linear equation. There are response differences between the individual ZDC channels because of high voltagesetting, photocathode damage of the PMTs, and radiation damage. The charge of everymeasured channel i is multiplied by a w i factor to match the different gains of the individualchannels. Thus, the total energy deposited in a ZDC detector E is calculated as: E = (cid:88) i w i a i . (19)First, the whole EM section is scaled to minimize the single neutron resolution, as describedin Section 3. Then the gain matching constants for the hadron section channels are calculated10 imeslice [25 ns] C ha r ge [f C ] HAD1pPb 8.16 TeV
CMS
Timeslice [25 ns] C ha r ge [f C ] HAD1pPb 8.16 TeV
CMS
Figure 10: Examples of template fits without a pre-pileup signal (left) and with a pre-pileupsignal occurring 100 ns before the main signal (right).from the comparison of detector level and simulated per-channel energy distributions usingsingle neutron candidate events, as illustrated in Fig. 11. After the gain matching of the HADsection channels, the EM section is weighted again to match the newly calibrated HAD section.Then the HAD channel weights are refined using a more pure sample of single neutron events.Finally the EM section weights are adjusted individually by minimizing the single neutronresolution.The distribution of calibrated ZDC energies is shown in Fig. 12. The three prominent peakscorresponds to single, double and triple neutron events. The reason for this quasi-discretespectrum is that the neutrons emitted from Pb ions are approximately monoenergetic due tothe small Maxwell-Boltzmann temperature of neutrons and the large Lorentz boost of the Pbion.Assuming that the response of a single neutron can be described by a Gaussian distribution,the neutron energies are added up independently, and the zero neutron contribution is describedby the sum of two exponential functions, the low-energy part of the spectrum is fitted withthe sum of Gaussian distributions and two exponential distribution: f ( E ) = a e − λ E + a e − λ E + (cid:88) n =1 A n √ πσ n e − ( E − µn )22 σ n , (20) µ n = nµ + ν, (21) σ n = nσ , (22)where a , and λ , are the parameters of the exponential functions corresponding to thezero neutron distribution, A n is the amplitude of the n -neutron peak, and µ , ν , and σ areparameters describing the positions and widths of the neutron peaks. The relative width ofthe single neutron peak at 2.56 TeV calculated from the fit is approximately 23 . [TeV] N o r m a li z ed y i e l d / . T e V - - -
10 1 MC DataHAD1HAD2HAD3HAD4 pPb 8.16 TeV
CMS
Figure 11: The distribution of deposited energy in HAD section channels in simulation (lines)and data (dots).Finally the time-dependence of the µ and ν parameters is studied. It is found that thesequantities depend on time in the given run as shown in Fig. 13. The reasons for this areeffects that depend on the instantaneous luminosity, like the activation of the detector andthe degradation of the beam quality. Second order polynomials µ ( t ) and ν ( t ) are fitted todescribe the time-dependence. Using these, the measured total ZDC energy can be correctedon an event-by-event basis: E corr = E − ν ( t ) µ ( t ) × .
56 TeV . (23)After applying this correction, the relative width of the single neutron peak is reduced to23 . Simultaneous pPb collisions (in-time pileup) shift the ZDC energy spectrum to higher values,which causes a raise in the tail of the ZDC energy distribution. In this paper a deconvolutionmethod is applied to remove these multicollision events from the final distribution. A similarmethod was used in [16]. The probability of having k number of interactions in a bunchcrossing is distributed according to Poisson distribution: p k = µ k k ! e − µ − e − µ , (24)where µ is the mean number of collisions and the term 1 − e − µ appears in the denominator,since k ≥ f ( E ) probability density function,which is expressed using the total probability theorem as f ( E ) = g ( E ) p + ( g ∗ g )( E ) p + ( g ∗ g ∗ g )( E ) p + . . . , (25)12 [TeV] E v en t s / . T e V DataFitNeutrons0 neutron pPb 8.16 TeV
CMS
Figure 12: The measured ZDC energy distribution. The three prominent peaks correspondsto single, double and triple neutron eventswhere g ( E ) the probability density function of the energy deposit in a single collision and ∗ denotes convolution.Taking the Fourier transform of both sides: F ( ω ) = ∞ (cid:88) k =1 p k G k ( ω ) = e − µ − e − µ ∞ (cid:88) k =1 ( µ G ( ω )) k k ! = e − µ − e − µ (cid:16) e µG ( ω ) − (cid:17) , (26)where F ( ω ) and G ( ω ) are the Fourier transform of f ( E ) and g ( E ) respectively. After expressing G ( ω ), g ( E ) can be written as g ( E ) = F − (cid:20) µ log [1 + (e µ − F ( ω )] (cid:21) . (27)The method is tested with a simple model, assuming that the true ZDC energy distributionis Gaussian. First a Poisson distributed random integer k is generated. In the next step k random Gaussian variables are summed. The distribution generated in this way is displayed bythe blue curve in Fig. 14. Finally the Fourier deconvolution is applied to this distribution, andthe result (red curve) shows a good match with the true distribution (black curve), supportingthe method. This test is performed with various µ pileup values.The correction is applied to the measured data, the result is shown in Fig. 15. As asystematic study, several µ values are used to perform the correction. The plot in the rightpanel of Fig. 15 shows that choosing a too high µ value in the calculation results in a nonphysical,negative probability density function – due to the overcompensation of the tail. This providesa possibility to set an upper limit on the value of µ , in our case it is approximately 0 . µ as well, from the instantaneous luminositymeasured by the central detectors. These luminosity measurements do not include thosenuclear excitation processes, when ions are excited and emitting neutrons, but no signal isproduced in the central CMS detector. 13 [min] [ T e V ] m pPb 8.16 TeV CMS t [min] [ T e V ] n pPb 8.16 TeV CMS
Figure 13: The dependence of µ and ν parameters on time since the beginning of the run.The time-dependence of both µ and ν is fitted with a second order polynomial. E [TeV] ] - P r obab ili t y den s i t y [ T e V m CMS
Simulation
E [TeV] ] - P r obab ili t y den s i t y [ T e V m CMS
Simulation
E [TeV] ] - P r obab ili t y den s i t y [ T e V m CMS
Simulation
Figure 14: Testing pileup correction method on a toy model assuming Gaussian ZDC energydistribution with different pileup values.
The performance studies of CMS ZDC detector have been presented. The response of thedetector to neutrons originating from various physics processes was studied using a geant .
1% for 2.56 TeV neutrons. Furthermore, the ZDC has greater than 98% geometricalacceptance for neutrons produced in giant dipole resonance, evaporation and cascade processes.Then a template fitting approach was presented, which is used to extract the signalamplitudes for the individual channels. This method is based on solving a linear system ofequations and also includes the treatments of uncertainties and correlations of the pedestal,14 [TeV] ] - P r obab ili t y den s i t y [ T e V - - - - - Uncorrected = 0.15 m Corrected, = 0.17 m Corrected, = 0.20 m Corrected, pPb 8.16 TeV
CMS
E [TeV]
150 200 250 300 ] - T e V P r obab ili t y den s i t y [ Uncorrected = 0.15 m Corrected, = 0.17 m Corrected, = 0.20 m Corrected, pPb 8.16 TeV
CMS
Figure 15: Pileup correction applied on data by assuming various µ values (left). By observingthe tail of the pileup corrected distributions, upper limit on µ can be determined (right).the digitization and the template shapes. It provides an opportunity to extract signals fromevents with a pre-pileup signal without introducing a bias.The channels were gain matched by comparing to the Monte Carlo simulation of thedetector and using various data-based techniques. Peaks were observed in the ZDC energyspectrum, corresponding to single, double, and triple neutron events. It was shown, thatthe spectrum can be described by the sum of two exponential functions, describing the noisepeak and photons, and the sum of Gaussian distributions, describing the neutron peaks. Itwas found, that the parameters of the neutron peaks vary by time, because of the changein instantaneous luminosity. For this effect a simple, event-by-event correction factor wasintroduced.Finally, a method using Fourier transformation was presented to correct for the effect ofin-time pileup. The feasibility of this correction was demonstrated using a Gaussian toy model.It was shown that by examining the tail of the corrected distribution, an upper limit can bederived on the value of pileup. Acknowledgments
We congratulate our colleagues in the CERN accelerator departments for the excellent per-formance of the LHC and thank the technical and administrative staffs at CERN and atother CMS institutes for their contributions to the success of the CMS effort. In addition, wegratefully acknowledge the computing centres and personnel of the Worldwide LHC ComputingGrid and other centres for delivering so effectively the computing infrastructure essential toour analyses. Finally, we acknowledge the enduring support for the construction and operationof the LHC, the CMS detector, and the supporting computing infrastructure provided by thefunding agencies.The CMS Zero Degree Calorimeter detector is supported by the Office of Science, USDepartment of Energy. This research is supported by the ´UNKP-19-3 New National Excellence15rogram of the Ministry for Innovation and Technology, the National Research, Developmentand Innovation Office of Hungary (K 124845, K 128713, K 128786, and FK 123842), and theHungarian Academy of Sciences ”Lend¨ulet” (Momentum) Program (LP 2015-7/2015).
References [1]
CMS collaboration,
Measurement of the pseudorapidity and centrality dependence of thetransverse energy density in PbPb collisions at √ s NN = 2 . TeV , Phys. Rev. Lett. (2012) 152303 [ ].[2]
ALICE collaboration,
Centrality dependence of particle production in p-Pb collisions at √ s NN = 5.02 TeV , Phys. Rev. C (2015) 064905 [ ].[3] CMS collaboration,
Status of zero degree calorimeter for CMS experiment , AIP Conf.Proc. (2006) 258 [ nucl-ex/0608052 ].[4] O. A. Grachov et al.,
Measuring photons and neutrons at zero degrees in CMS , Int. J.Mod. Phys. E (2007) 2137 [ nucl-ex/0703001 ].[5] CMS collaboration,
Performance of the combined zero degree calorimeter for CMS , J.Phys. Conf. Ser. (2009) 012059 [ ].[6] O. Grachov, M. Murray, J. Wood, Y. Onel, S. Sen and T. Yetkin,
Commissioning of theCMS zero degree calorimeter using LHC beam , J. Phys. Conf. Ser. (2011) 012040[ ].[7] H. S. Matis, M. Placidi, A. Ratti, W. C. Turner, E. Bravin and R. Miyamoto,
The BRANluminosity detectors for the LHC , Nucl. Instrum. Meth. A (2017) 114 [ ].[8]
Particle Data Group collaboration,
Review of particle physics , Phys. Rev. D (2018) 030001.[9] A. Ferrari, P. R. Sala, J. Ranft and S. Roesler, Cascade particles, nuclear evaporation,and residual nuclei in high-energy hadron - nucleus interactions , Z. Phys.
C70 (1996)413 [ nucl-th/9509039 ].[10] F. Sikler,
Centrality control of hadron nucleus interactions by detection of slow nucleons , hep-ph/0304065 .[11] B. L. Berman and S. C. Fultz, Measurements of the giant dipole resonance withmonoenergetic photons , Rev. Mod. Phys. (1975) 713.[12] I. A. Pshenichnov, Electromagnetic excitation and fragmentation of ultrarelativisticnuclei , Phys. Part. Nucl. (2011) 215.[13] M. Chiu, A. Denisov, E. Garcia, J. Katzy, M. Murray and S. N. White, Measurement ofMutual Coulomb Dissociation in √ s NN = 130 GeV Au+Au collisions at RHIC , Phys.Rev. Lett. (2002) 012302 [ nucl-ex/0109018 ].[14] D. Gayther and P. Goode, Neutron energy spectra and angular distributions from targetsbombarded by 45 MeV electrons , Journal of Nuclear Energy (1967) 733.1615] CMS collaboration,
Reconstruction of signal amplitudes in the CMS electromagneticcalorimeter in the presence of overlapping proton-proton interactions , JINST (2020)P10002 [ ].[16] A. Laszlo, G. Hamar, G. Kiss and D. Varga, Single electron multiplication distribution inGEM avalanches , JINST (2016) P10017 [1605.06939