The MPP group developed algorithms based on the tracks of charged particles for the positional calibration of the silicon modules in the pixel and strip detectors. The positional calibration, i.e. the knowledge of the exact positions of all modules, is essential to achieve the best possible reconstruction of the particle trajectories.

The first step of the calibration procedure uses optical measurements to determine the positions of the individual sensors on each detector module. The accuracy achieved for the position of the sensors in the silicon strip detector is better than 5 micrometers, for example. The MPP researchers also analyzed where exactly the modules are installed on the mounting frame of the silicon strip detector in comparison to their intended position. The positional accuracy of the end-cap modules on the disks is about 20-50 micrometers perpendicular to the beam direction and approx. 100 micrometers parallel to it. This is still insufficient to achieve the ATLAS objectives in the field of physics, however.

Taking these external measurements of the sensor positions as starting points, an iterative method using particle trajectories in the detector was therefore used to determine the positions of the sensors more accurately. The method developed by the MPP is based on a χ² minimization, whereby the position of each module in terms of its six degrees of freedom is determined as accurately as possible. The method uses the residuals, i.e. the distances between the extrapolated hit and the currently assumed locations of the read-out structure (pixel or strip). It also minimizes the widths of the residual distributions. This alignment algorithm has been applied successfully to simulated events, to cosmic radiation data, test beam data and ATLAS proton-proton collision data.

The procedure was optimized continuously, for example by giving special consideration to the regions in the detector where two modules overlap, since these provide a particularly good determination of detector position, or by incorporating the constraint that all tracks have a common origin (vertex) into the algorithm.

After the successful initial calibration of the ATLAS silicon detector, this line of research is not being pursued further at the moment.