University of Heidelberg
GSI

ALICE

ISOQUANT

TPC Calibration

The ALICE TPC

The Time Projection Chamber (TPC) is the main device for tracking and identification of charged particles in ALICE. In central Pb-Pb collisions, it has to be able to reconstruct and distinguish between several thousands of particles from a single event.
The TPC is a large cylindrical detector filled with a gas mixture of Ne-CO2-N2. Particles traversing the detector lose a small part of their energy by ionizing the gas atoms. The ionization electrons drift along a uniform electric field from their point of origin towards the end caps of the detector where they induce a signal in the finely-segmented readout chambers based on GEM technology. These signals provide information about the 2-dimensional (transverse plane) coordinate of the space-points while the third coordinate (longitudinal axis) is derived from the measurement of the drift time of the ionization electrons.
A tracking and reconstruction software reconstructs the trajectories of single particles from the combination of all measured space-points of a given event. The energy loss per unit length, dE/dx, is calculated from the total charge of all signals collected at the pad plane of a given track. The measurement of the energy loss in combination with the information about the particle momentum, derived from the curvature of the particles in a magnetic field, allows to identify the particle species.



Calibration

The collection of the signals on the readout pads is subject to a number of different detector effects which distort their size, shape and position. Therefore, a fundamental understanding of the detector and these detector effects is crucial for the proper reconstruction of the space-points, tracks and their properties so that they can be used in physics analyses.
The signals and space-points are corrected for the various effects by a calibration software. Due to the recent upgrade of the TPC readout chambers, new detector effects become relevant which we study in detail. Changes in the general readout scheme require that the calibration and reconstruction software is written from scratch to provide fast and efficient methods.
Due to the complexity of some of the detector effects, we develop new calibration algorithms which also include the application of machine learning techniques and convolutional neural networks and provide multi-dimensional corrections.

Space-Charge Effects

One of the major calibration procedures is the correction of the coordinates of the measured space-points which will be distorted due to a significant amount of positive ions (space charge) inside the drift volume. These ions are created in gas amplification processes inside the readout chambers. Due to the intrinsic properties of the GEM readout chambers, a fraction of them enter the drift volume where they distort the uniform electric field and the drift path of the ionization electrons. Significant fluctuations of the space-charge density on time scales of the order of a few milliseconds add further complexity to the task.
We are heavily involved in the development of the space-charge distortion calibration, working on time-dependent multi-dimensional correction methods and algorithms.

Energy Loss dE/dx

Our group, in close collaboration with our colleagues at GSI and Frankfurt University, also works on the calibration of the dE/dx measurement on the single cluster and the track level. The energy loss is corrected as a function of several kinematic and event variables in order to provide an unbiased measurement. A robust parameterization of the corrected dE/dx is created as a function of the particle momentum for each particle species, which is then used to determine the identity of single tracks in data analyses.


events/timeframe display
Events/timeframe display from an ALICE Run 3 simulation.

Tools for Calibration and Analysis of Detector Studies

The software we use for the calibration is mostly written in modern C++ while the application of machine learning algorithms and convolutional neural networks are performed with python.
The calibration methods are implemented in the ALICE calibration and reconstruction framework called AliceO2. It is based on the ROOT software and makes use of the combined efforts of experiments from FAIR and ALICE, which provide the framework ALFA with generic methods and concepts for simulation, calibration and reconstruction.
The RootInteractive framework allows to interactively visualize and analyze multi-dimensional data.

Contact Person in the Group

Alexander Schmah (Tenure track)

Webmaster:
IT Department