10. Bibliografía
11.1.3. Transcripción clase 3: 24 de Febrero de 2015
When designing a hadronic calorimeter the integration time, i.e. the time in which the detector is sensitive and accumulating signal, has to be tuned according to the time structure of hadronic showers. In theory the integration time would ideally cover the entire energy deposition time of hadronic showers in order to reduce statistical errors on the measurement. In potential real world applications such as the ILD detector at CLIC where bunch crossings occur every 0.5 ns, however, the overlap with subsequent events significantly limits the integration time window. A study on which fraction of
Shower Radius [cm]
0 10 20 30 40
Mean time of First Hit [ns]
0 5 10 15 Data + π 180 GeV 60 GeV 80 GeV
Figure 4.17: The mean Time of first Hit at different shower radii.
the total energy is deposited after what time helps to find a good compromise on the integration time.
This quantity can be obtained by first creating a histogram filled with all Time of first Hits, weighted with their respective amplitude. This gives the average energy deposition of hadron showers for every point in time. Thus the second step is the integration of this histogram.
However, as T3B is only a single strip, some correction factors need to be included in order to correctly represent the energy deposition fraction of an actual calorimeter. As discussed in subsection 4.2.2, the amplitudes are normalized to the sensitive geometric area of the respective tile. Furthermore, differences in the frequency of the shower starting positions are taken into account by normalizing the amplitude with the number of events with shower starts in the different W-AHCal layers.
Integrating over several µs is infeasible in a real detector and typically also not necessary to achieve a good energy reconstruction. The shaping time of the W-AHCal prototype, for example, was in the order of 100−150 ns. Thus a maximum integration time of 200 ns was assumed for the present study, so that the energy deposited at a certain time is considered relative to the total deposition in 200 ns. Depositions at later times were not included in this analysis.
Before the results obtained from the simulation of different physics lists can be compared to the data acquired at the testbeam, the method used to reconstruct the time distribution of the energy deposition fraction is validated. Studying the changes to the distribution due to the analysis is done by comparing the raw simulation output of Geant4 with the fully digitized and reconstructed simulation of the same physics
lists. Figure 4.18 shows the resulting energy deposition fraction curves of a 60 GeV
Time [ns]
0 50 100 150 200
Energy Deposition Fraction / 0.1 ns
0.9 0.92 0.94 0.96 0.98 1 1.02
Energy Deposition Fraction
60 GeV QBBC TofH 60 GeV QBBC Undigitized
Figure 4.18: Comparison of the fraction of deposited energy over time. One set was generated from data of the T3B strip obtained from a full detector simulation including digitization and reconstruction using Time of First Hit. The other set is the pure
Geant4 output of the entire detector without digitization or reconstruction.
of first Hit (TofH) data (including all mentioned corrections) of a simulated and fully digitized T3B strip, and it is once obtained from pure Geant4 data of the entire
detector without digitization (undigitized).
Clearly the reconstructed, Time of first Hit based data from the T3B strip over- estimates the response speed, as it reached 98% already after about 10 ns, while the undigitized, actual energy deposition reaches that value only after 20 ns. This effect can be explained mainly with the time definition of the Time of first Hit, which uses the arrival time of the second photon which is obviously earlier than the average arrival time of all photons. This definition of the Time of first Hit intended to compensate the delaying effects such as the photon travel time in the scintillator and the decay constant of the scintillator, both of which were implemented in the digitization. Obviously the comparison to the raw simulation data reveals that this definition overcompensates the delaying effects, shifting the Time of first Hit based energy deposition fraction curve to earlier times.
However, it is expected that calorimeters such as the one within the ILD concept mimic the behaviour of the Time of first Hit definition, as the read out electronics will only assign a single timestamp to a hit in contrast to the full waveform of the oscilloscopes used at T3B.
A comparison of the energy deposition fraction between the different physics lists and testbeam data is shown in Figure 4.19. As before the results are obtained from the 60 GeV π+ dataset. One can clearly see that the high precision physics lists
Time [ns]
0 50 100 150 200
Energy Deposition Fraction \ 0.80 ns
0.9 0.92 0.94 0.96 0.98 1 1.02 + π Fraction - 60 GeV dep E data QBBC QGSP_BERT_HP QGSP_BERT
Figure 4.19: Comparison of the fraction of deposited energy over time between data and simulation with different physics lists.
depositions than the QGSP BERT physics list. However, the deviation of the latter is only in the order of 1%. Furthermore the plot shows that over 90% over the energy are deposited almost instantaneously, and all simulations as well as the testbeam data reach a fraction of 98% after about 10 ns when using the Time of first Hit definition. According to Figure 4.18 at the 98% level a 10 ns delay in the Time of first Hit based energy deposition fraction is consistent with a 20 ns delay in the undigitized case, thus limiting the need for longer integration times in calorimeters, also when tungsten is used as absorber.