6. CENTRAL TRACKER The central tracking system will play a major role in all physics searches. It is also a very challenging and complex sub-detector based on substantial progress achieved in gas and silicon detector technology. Russian scientists participate in construction of Microstrip Gas Chambers for the forward-backward parts of the Tracker and in detector simulations. R&D. In 1996 we continued research on the origin of discharges in MSGC. The study of discharge limits at different gas mixtures and for different materials of strips in the chambers was performed. It showed that the difference between Ne and Ar in terms of discharge limits depends on particular MSGC technology very much and that in general gold chamber has lower discharge limits than one of chromium. We plan to continue this work including the study of the influence of heavily ionizing particles on the discharge probability ("Study of ageing and gain limits of MSGC at high rates", B.Boimska et.al, CERN-PPE/96-201; "Gain limits in different types of MSGC with several gases in the presence of strongly ionizing radiation", B.Bouclier et.al, CMS TN/96-018). In order to have an alternative technology for the electronically conductive substrate we made a search of technologies of thin conductive coating available in Russia. The appropriate technology was found which allows to produce electronically conductive layers on glass substrates with the surface resistivity in the range of $1014-1016$ Ohm/sq. This technology, namely aluminum nitride coating, was used to produce test chambers. The performance of these first MSGC with AlN coating was investigated. General stability was shown and rate capability of about $106$ Hz/$mm2$ was demonstrated (Fig. 6.1). We continue study of general features of this coatings, compatibility with existing technology of MSGC production, resistance to discharges and ageing. ("MSGC with AlN coating", A.Bondar et.al, accepted for publication in Nucl.Instr.&Meth.) Preparation of milestone prototype In 1996 we started preparation of the milestone prototype F1 which will consist of several modules each containing 8-9 wedge MSGC fully equipped with readout electronics. One of these modules is being produced in Novosibirsk. In 1996 we produced the mask for the wedge MSGC and tested the first two samples produced with this mask. The uniformity of gain along the strip of wedge MSGC was measured within 5 % (Fig. 6.2). The absolute gain can be obtained safely up to 2000. The production of all MSGC substrates is expected to be completed in May 1997. Novosibirsk group is responsible for the development of the readout electronics for the milestones prototype and manufacturing of the Slow Control Module which will be used for monitoring of low voltages and currents delivered to the front-end electronics and for the control of calibration of the front-end electronics. All the elements of the readout system were designed by the engineer from BINP in the first half of 96. The whole readout chain of one detector module was debugged and corrected in October-November 1996. The production of the necessary amount of the Slow Control Modules is foreseen to be finished in March 1997. General review of the work was made by L.Shekhtman in his report "Tracking with Micro-strip gas chambers" at the International Conference on Instrumentation for e+e- colliders (March 1996, Novosibirsk, Russia). Simulations. In order to exploit higher performance of the CMS central tracker system, dedicated reconstruction software has to be developed. Huge number of detector channels and the event complexity (up to $105$ measured 3d points) together with tough time requirements 0.1 sec/ev make this problem really challenging. Approaches used so far in HEP turn out to be unapplicable in the LHC environment due to the severe combinatorics. The actual amount of "interesting" signals in the system appears to be 2-3% of the total number at higher luminosity. The simulation and reconstruction software must provide efficient tools for optimization of the tracker system in the present phase of the experiment. Our investigations show that it seems attractive to divide the pattern recognition task in two major stages, the first stage to be the preselection of interesting patterns and second one - local track finding within these patterns. At present two algorithms based on this approach (GTF and Connection Machine) are developed and being intensively used for tracker optimization studies. The example of LHC event reconstructed by this software is shown in Fig. 6.3. The algorithms demonstrate good performance in reasonable time scale of 1 min. and are perfectly suited for present tasks. However more work has to done in order to develop faster reconstruction software and try different model approaches the pattern recognition problem (neural networks, genetic algorithms, cellular automation etc.). Found tracks have to be combined together to form vertices. To do this, a novel algorithm was developed and evaluated. The vertex finding problem is formulated as a discrete - continuous optimization problem and solved using neural networks technique. Current tasks include optimization of the detector layout, performance and occupancy studies. The investigation of physics channels most crucially dependent on the tracker performance is on the way. In 1996 we started to work on the problem of the effects of misalignments on the CMS Central Tracker performance and the development of the global software alignment procedure. The first study was performed using local track finding algorithm available in CMS simulation program. This algorithm was used to reconstruct simulated tracks of isolated muons. It was shown that in terms of momentum resolution the performance of tracker does not deteriorate if the r.m.s. misalignment of individual detector modules is less than half of the intrinsic spatial resolution. Also a simple procedure for correction of the misalignment of large elements of tracker was demonstrated, using a maximum likelihood method.