Pi/e separation from shower-shape variable by 2016 CERN 8-module data8-module data

In the HL-LHC era, the pileup environment will produce considerable number of high energy pions in the forward region. A small percentage of the charged pion can interacts electromagnetically, which would deposit the same signal in the upstream tracker as elec-tron does. These pions is tagged as e-like pion later.) This analysis aims to provide a tool to separate the electrons with those pions.

As HGCAL is designed as a sampling detector, the information of shower development is recorded. Based on the particle type and energy, the shower-shape varies from particle to particle. As a first trial of using shower-shape information to identify the particle type, the lateral containment variables^† are defined and tested using the 2016 CERN config2 setup.

The result further shows that the HGCAL prototype has a potential to tag the delta-ray. The delta-ray is a secondary electron with enough energy to escape a significant distance away from the primary beam. The delta-ray can be generated from high energy transfer from the scattering process of the primary beam (the pion), the knocked out electron will further ionize or creates shower inside the calorimeter. A preliminary study to the delta-ray candidates will also be covered.

†One of the example using the containment variable in current ECAL is the R9 variable.[25]

9.2.1 Data set

Data: 20, 32, 70, 100GeV electron and 125GeV pions.

MC: 125GeV pion, 100 GeV and 125GeV electrons.

Since 125 GeV pion is the only pion energy taken in the period, the closest energy of the electron beam (100 GeV) is chosen to compare with the pion in the study.

9.2.2 2D re-weighting of MC beam profile

Before all study begin, the first task is to re-weight the beam distribution in MC to match the Data distribution. The profiles are extracted from X_imp and Y_imp in the first layer^‡, and filled into a 21 × 21 array. Each slice correspond to 10mm in the detector. The ratio of each array component is then applied on such position. As a result, the 2D re-weighting shifts the MC beam profile to match the data distribution. The 1D distribution of the beam before and after the 2D re-weighting are shown in Figure9.9 for 125 GeV pion and 9.10 for 100 GeV electron.

Figure 9.9: 125GeV pion 2D re-weight

‡The X_imp and Y_imp stands for impact position for X and Y, which is basically the energy center of gravity of such layer. Same definition can be deduced from8.1.

Figure 9.10: 100GeV electron 2D re-weight

9.2.3 Selection

The hit selection is basically the same as 7.1. However, for the central event selection is loosen in order to include more pion in the data. Since the study aims to separate e-like pion and electrons, the concept of other selections is to cut away the events which are obvious pion events and keep all the electron events.

• Event selection

1. R_imp of first layer < 30 mm

2. E8 < 2.5 MeV, where E8 is the energy sum of the last silicon layer.

3. E_visible > 0.01 GeV

4. SHD < 16 X0 and SHD > 6 X0

• Hit selection (Apply to all cells) 1. Energy in the cell > 2MIP

The E_visible cut is to remove pion with small energy deposition which is not interesting, while electrons would always leave higher energy than the threshold.

The E8 cut perform as a ”VETO”, which rejects the event with energy higher than 25 MeV as all electrons should finish showering before 27 X₀.

The SHD cut is based on the observed SHD for all electron samples, since conserving all electrons is the first priority. See Figure 8.6.

Figure9.11 shows some observable of electron and pion in the same energy (125 GeV) before the selection, where the reason of applying these cuts is obvious. The statistics of the data will be shown in Table 9.3.

The E_visibleof the pion data is again 10 % lower than the MC. A 1.105 factor is applied to scale up the data to match the MC behavior as in electron case.

Figure 9.11: Observables before cut

In the plots both electron and pion are 125 GeV, where the 125 GeV electron behaves almost the same in E1,E8 and SHD.

Profile Total events Passed events

100GeV e MC 27198 27004

100GeV e Data 93997 82299

125GeV e MC 21996 21682

125GeV π MC 639987 204424 125GeV π Data 29545 8511

Table 9.3: Table of statistics - πe separation

9.2.4 Electron contamination in pion data

During this study, a second peak in pion data in Evis is observed when comparing data to MC. This peak locates at the same energy where the 125 GeV electron MC peaks, which indicates a contamination of 125 GeV electron in pion.^† See Figure 9.12. However, the contamination is only 3%, which should not affect the over all behavior of the pion data.

Figure 9.12: Electron contamination in pion data

In the plot the pion MC is re-normalized based on data in order to match the same scale after the E_vis > 0.01 GeV cut. The electron MC is scaled down by a factor of 2 in order to empha-size the peak in pion data. The bottom histogram (labeled as realecandidate|isthedirectsubtractionbetweenpiondataandM C.

†Due to the beam selecting mechanism, particles with the same energy can pass the beam line and incident the prototype.

9.2.5 Lateral shower-shape variables

The definition of the shower-shape variable is simple. First, search the cell with the highest energy in a layer, which is labeled as E_seed or E_max in previous chapter. Then the first neighbour of the E_seed is called the first ring, which has 6 members. The cell outside the first ring then compose a second ring, which has 12 members. Due to the geometry of the hexagonal cell, the outer ring has 6 members more than the inner ring. ^†

The energy of the E_seed is defined as E_r0 and the energy of the E_seed plus the first ring is defined as E_r01 (with maximum 7 members after 2 MIP hit selection). Follow this rule, E_r012 (with maximum 19 members) and E_r0123 (with maximum 37) members are also defined.

Different combination of the containment ratio are tested and here the variables with best performance are shown.

The single layer containment variables are shown in Figure 9.13. All the layers are studied with a good performance but layer 2 and 3 are selected to be displayed here since these layers locate around electron shower maximum. The results of the containment sum of multiple erengy of elctron and 125 GeV pion MC are compared in Figure 9.14, and for the 100 GeV electron and pion Data/MC is shown in Figure 9.15.

The results of the containment sum not only show the power of separating the e-like pion with electron, but they also indicates the presence of the delta-ray candidates with the containment sum ratio (

PEr01

PEr012) > 0.9.

Thus, a further study is done on selecting these events. The theoretical highest energy of the electron can be knocked out by 125 GeV pion is around 100 GeV, calculate from the T_max = 2m_ec²β²γ² in Equation 1.1. The expected energy deposition corresponds to 100 GeV electron is 0.17 (GeV). The first test, is again the Data/MC comparison for the pions after the additional selection of the containment sum. The results again show a perfect agreement in Figure 9.16. This result also shows the selected events can be well describe by the 10 ∼ 100 GeV electron.

After the 10 ∼ 100 GeV delta-ray candidates are selected. The second step is to see if the prototype can separate the delta-ray candidates with certain electron energy. To

†Of course this rule breaks when the ring reaches the module boundary. However, thanks to the central cut in event selection the Eseed usually locates around the middle of the sensor. Hence, the ring is ensured complete at least in the third ring.

‡ E_r0

Er01 is studied but is shown too sensitive to the initial position of the incident beam.

Figure 9.13: Single layer containment variable

The plots show that e-like pion events create generally broader shower than the electron events, while a bump at 1 indicates the MIP events or the delta-ray candidate events.

Figure 9.14: Containment sum MC comparison

The plot not only shows the different behavior in pion and electron, but also indicates that the electron shower is broader when the energy raise.

Figure 9.15: Containment sum for electron and pion

Figure 9.16: delta-ray candidates Data/MC

The shower depth on the top right is the shower depth calculate by E_max.

realize the idea, the events with 0.01 (GeV) < E_visible < 0.05 (GeV) are selected. These pion events leaves the same energy as the 20 GeV electrons do.^† Figure9.17 shows that the longitudinal shower shape is different in the delta-ray candidates and the real 20 GeV electron events, which proves the ability of the separation.

Figure 9.17: E1/E_visand SHD between 20 GeV e⁻ and delta-ray candidates The plots show these 2 variables are useful to separate the delta-ray candi-dates with the real electron.

The shower-shape study shows the lateral containment variables have the power to separate e-like pion and electron. For the higher containment sum region, the delta-ray candidates can also be separated from the real electron events by longitudinal shower-shape variable such as E1/E_vis and the SHD.

This study is actually a first try to proof the idea of the design of HGCAL as a calorimeter which is capable of the particle identification (PID). More complicate meth-ods can be used to boost the performance of these variables, while on the other hand there is still plenty of space to find new variables to achieve PID from the shower-shape information. The prospects is though encouraging in the field.

†The selection is related to Figure8.2.