510
J. Apostolakis
11.5.2 Hadronic Calorimeters: Comparisons with LHC Test
Beam Results
For the hadronic calorimeters, after muons and electrons, the test beam results
with pions were compared to simulation. This served to validate the simulation of
hadronic interactions, since the conversion of the energy deposition into signal is
common with the electron test beams. Additional observables compared included
the longitudinal and lateral shower shape.
The ATLAS Hadronic Barrel Calorimeter’s (TileCal) use of scintillator requires
the accurate simulation of neutrons for those interactions that contribute significantly to energy deposition and which occur within its time window of 150 ns. This
is treated in the simulation.
ATLAS has also undertaken extensive test beam measurements of the response
of the TileCal to pions. The most recent comparisons in the Combined Test Beam
setup (2004) involved pion energies up to 350 GeV. These tests [112] examined
the energy response and resolution of the calorimeter, and compared them with the
predictions of a simulation based on GEANT4 (version 9.1).
Events were selected based on several criteria, including the energy deposition
in a cryostat scintillator (SC1) placed before the TileCal. This cut was made in
order to enable comparisons with the previous test beam. Potential biases from this
selection cut on the response and resolution were studied. An approximation of
this cut was used in the simulation: the energy deposition in the surrounding dead
material. The change in the energy response due to this cut ranged from −2.5% to
+0.5% depending on the energy and eta value (0.25–0.65) of the pion beam. This
was reproduced within 0.5% at low energy (20 GeV) and within 1.5% at high energy
(300 GeV).
The energy resolution was affected in a range from +10% to −10% between low
and high energies respectively and reproduced within 2% for a large combination of
angles and energies, except for a 4% deviation at one angle at 20 GeV. Comparing
the final results for the energy response, the agreement obtained is within 3% for
the full energy range studied (20–350 GeV). Typically, agreement at the 1–2% level
is achieved for beam energies of 50–250 GeV; greater deviations are seen at 20
and 300 GeV (the latter, in particular, remains to be understood). For the energy
resolution agreement at the 10% level is obtained.
The measurements of the reconstructed energy in the Atlas TileCal at energies
from 20 GeV to 180 GeV have also been compared with simulation [113]. Figure
11.2 shows the ratio of simulation and data for the energy for the case of incident
pions using different configurations of physics models in 2010 with GEANT4
version 9.2. Agreement for the mean energy between simulation and data ranges
from −10% (the legacy LHEP physics modeling) to +3% (the production physics
list QGSP_BERT). The root mean square deviation (RMS) of the reconstructed
energy agrees within −4% (QGSP_BERT) to about +15% (LHEP).
Since a 1 MeV neutron could travel only 3 cm within 100 ns, even if it never
interacted, the propagation of low energy neutrons—and thus their contribution to
J. Apostolakis
11.5.2 Hadronic Calorimeters: Comparisons with LHC Test
Beam Results
For the hadronic calorimeters, after muons and electrons, the test beam results
with pions were compared to simulation. This served to validate the simulation of
hadronic interactions, since the conversion of the energy deposition into signal is
common with the electron test beams. Additional observables compared included
the longitudinal and lateral shower shape.
The ATLAS Hadronic Barrel Calorimeter’s (TileCal) use of scintillator requires
the accurate simulation of neutrons for those interactions that contribute significantly to energy deposition and which occur within its time window of 150 ns. This
is treated in the simulation.
ATLAS has also undertaken extensive test beam measurements of the response
of the TileCal to pions. The most recent comparisons in the Combined Test Beam
setup (2004) involved pion energies up to 350 GeV. These tests [112] examined
the energy response and resolution of the calorimeter, and compared them with the
predictions of a simulation based on GEANT4 (version 9.1).
Events were selected based on several criteria, including the energy deposition
in a cryostat scintillator (SC1) placed before the TileCal. This cut was made in
order to enable comparisons with the previous test beam. Potential biases from this
selection cut on the response and resolution were studied. An approximation of
this cut was used in the simulation: the energy deposition in the surrounding dead
material. The change in the energy response due to this cut ranged from −2.5% to
+0.5% depending on the energy and eta value (0.25–0.65) of the pion beam. This
was reproduced within 0.5% at low energy (20 GeV) and within 1.5% at high energy
(300 GeV).
The energy resolution was affected in a range from +10% to −10% between low
and high energies respectively and reproduced within 2% for a large combination of
angles and energies, except for a 4% deviation at one angle at 20 GeV. Comparing
the final results for the energy response, the agreement obtained is within 3% for
the full energy range studied (20–350 GeV). Typically, agreement at the 1–2% level
is achieved for beam energies of 50–250 GeV; greater deviations are seen at 20
and 300 GeV (the latter, in particular, remains to be understood). For the energy
resolution agreement at the 10% level is obtained.
The measurements of the reconstructed energy in the Atlas TileCal at energies
from 20 GeV to 180 GeV have also been compared with simulation [113]. Figure
11.2 shows the ratio of simulation and data for the energy for the case of incident
pions using different configurations of physics models in 2010 with GEANT4
version 9.2. Agreement for the mean energy between simulation and data ranges
from −10% (the legacy LHEP physics modeling) to +3% (the production physics
list QGSP_BERT). The root mean square deviation (RMS) of the reconstructed
energy agrees within −4% (QGSP_BERT) to about +15% (LHEP).
Since a 1 MeV neutron could travel only 3 cm within 100 ns, even if it never
interacted, the propagation of low energy neutrons—and thus their contribution to