Performance of the partial STAR-SVT (Silicon Vertex Tracker) in the RHIC 2000 run Selemon Bekele* and Marcelo Munhoz** for the STAR SVT group * The Ohio State University ** Wayne State University/University de Sao Paulo

Abstract The Silicon Vertex Tracker(SVT), made of silicon drift detectors is one of the Subsystems of the STAR detector at RHIC. Being the innermost detector, it will help in determining the primary vertex and distinguishing particles from secondary decays. Due to its high position resolution, it will help in tracking short lived strange and multi-strange particles and will improve two track resolution which is very important to HBT studies. Also the SVT will enable STAR to cover the low momentum part of the pion spectrum, potentially on an event by event basis. Possible condensation effects, e.g. DCC and BEC, could thus be detected. For the summer 2000 run, a ladder with 7 wafers was installed and took data. Analysis of this data show that the SVT detector is working as expected.This emphasizes the fact that with the installation of the full SVT in time for the second year run, STAR will be able to improve on the physics program already initiated with the baseline detectors. In this poster, we will present analysis results of real data and also data obtained with the SVT slow simulator.

The Star Detector at RHIC consists of different sub-detectors which can take measurements simultaneously. One of the subsystems is the so-called Silicon Vertex Tracker, abbreviated as SVT. The length of the SVT is 44.1 cm and will cover the pseudorapidity range from -1 to 1. On the left is shown the half barrel of the SVT detector. The other half will soon be completed.The SVT detector consists of three concentric barrels around the beam pipe. The outer barrel is located at cm, the middle barrel at cm and the inner barrel at 5.97 cm from the beam line. There are 16 ladders on the outer barrel, 12 on the middle barrel and 8 on the inner one. The Star detector

A charged particle passing through the SDD leaves a cloud of electron hole pairs.The electron cloud is forced to the mid plane of the SDD by potentials applied to its outer faces. The electron cloud drifts towards the collecting anodes under the influence of a linearly varying potential and focused onto the anodes by a suitably adjusted electric field configuration cathodesanodes Focusing anodes To readout electronics 300 microns Electron cloud 250 microns 120 microns

While drifting to the anodes, the electron cloud expands mainly due to diffusion and coulomb repulsion. In the plot on the right, the initial widths in the electron cloud frame were taken to be equal for simplicity, and so there are two curves in each case. The signal on the anodes is then passed through a Pre Amplifier Shaper Array(PASA) circuit whose output is digitized and stored for further analysis. The Electron cloud x y X Y Cloud frame SDD frame Coulomb only Diffusion only Coulomb and diffusion (in the cloud frame)

One of the main objective of the SVT group in summer 2000 was to have an engineering run to check noise, position resolution and radiation damage. The second one was to use the real data obtained together with simulated data to improve the integrated STAR software, check the software chain that includes the SVT and TPC and to check the alignment between the two detectors.As a result, no attempt was made to do any physics analysis. In the simulation the same dE was used for all hits to investigate possible changes in the signal do to the software and/or PASA shaping. In the real data, one would get a variation in total dE due to there being different particles with varying momentum and entrance angles.

Cluster Finding A 2D plot of simulated hits. The adc counts from pixels that belong to a hit have been used to fill the histogram. The different colors for a given hit indicate the magnitudes of the count in the various pixels. A 2D plot of reconstructed hits. One can see that all the embedded hits have been located by the cluster finder. The color scheme is such that a cluster is identified by its index

The plot on the left shows real hits on a hybrid as found by the SVT cluster finder. The deep blue stripes from the hits up are the undershoot. This is from a very high multiplicity event and the average number of hits per event per hybrid was about 10 for the year 1 data Hits from real data

A plot of peak signal against the drift time shows the expected quadratic fall off as the electrons drift to the anodes. This follows from the fact that the peak has to go down as the cloud gets wider in order to maintain charge conservation. It was found through simulation that the wide variation of the peak signal at a given drift time is caused mainly by the position of the hit within a pixel. Peak versus drift time The plot on the left shows the variation of the peak adc values for the real data. The plots are per hybrid basis but still one can see that the real data shows the same behavior as the simulated data does. Drift time (in bins) Drift distance (in cm) adc counts

The plot on the right shows the variation of summed adc with drift time for simulated data. Except for small losses due to expansion and recombination the total charge, expressed here in terms of adc counts, is flat when plotted against drift time. For the simulated data, the upper two bands come from hits being embedded close to each other and becoming merged. Charge versus drift time One sees the same trend for the real data. Drift time (in bins) adc counts

Noise from run taken on June 23, 2000 (real data)

Noise from run taken on Aug 21st (real data)

The fact that the noise did not vary much from run to run implies that there was no apparent radiation damage. From the plots for the real data, one can see that the rms of adc values is between 1 and 2 channels, a channel corresponding to 4mV. This seems to be in agreement with measurements made on the bench shown on the right.

In the summer 2000 run, zero suppression was not done.In the figure on the right two plots are overlaid from data taken about 3 hours apart. They are projections onto the time bucket axis of the pedestals for one of the Wafers on the year 1 ladder. From this plot one can see that the pedestals are stable over several hours. The pedestals are flat which is important since even a small variation during the run could cause artificial hits. Pedestal subtraction will be done on line in future runs when the full SVT is in place. Pedestals (real data) Drift time (in bins)

In the course of its drift toward the anode region, the electron cloud moves with an almost constant drift velocity in the bulk of the detector while the velocity varies non linearly in the focusing region. The plot on the left shows the difference from a linear fit of a position versus time curve generated using laser data. One can use these calibration curves to get the position of a hit given the drift time and the drift velocity. Drift Velocity Calibration

The plot on the left shows matching efficiency as a function of event number and the one the right shows the efficiency as a function of drift distance. One sees that 85 to 90 % efficiency is obtained. This is reasonable considering merging effects, hits falling on the edges of the detector and signal spread at large drift distances. Hit Finding Efficiency (simulation)

40 GEANT events with the year 1 SVT configuration were processed with the SVT slow simulator and it was found that hit positions can be determined within a sigma < 25 microns. The resolution in the drift direction is not centered at zero due to the shift introduced by the PASA response function.The plots on the right show position resolution obtained using the SVT-TPC hit-track matching code for the same number of GEANT events. Monte Carlo tracks from the same events were extrapolated to the year 1 ladder position and the intersection points with the SDD were associated with hits reconstructed by the SVT software chain. In this case, the resolution in the drift direction has a sigma of about 350 microns and in the anode direction around 480 microns. It seems that the TPC resolution is playing its part in the SVT resolution in addition to the fact that the GEANT data went through the TPC slow simulator and reconstruction chain. Position Resolution (simulation)

hit - hit matching TPC track intersection point with SDD and SVT hit matching Sigma = mean = Sigma = mean = 0.017

The resolution in the time direction was found to be about 100 microns while in the anode direction it was around 300 microns. This is as good as expected given the TPC resolution and the accuracy with which the year 1 ladder was positioned and calibrated. The tails seen in the plots are found to be caused by mismatched hits and tracks. Position Resolution (real data) Delta X (in cm)

Delta Z (in cm)

Space points reconstructed from the SVT clusters were associated with space points obtained by extrapolating TPC tracks to the position of the year one ladder. A correlation between the SVT hits and those from the TPC can be clearly seen in the plot shown below. Correlation between SVT hits and TPC tracks # of SVT hits # of TPC tracks

The SVT detector performed very well in the summer 2000 run. No apparent radiation damage has been observed. The software which will be used for the full SVT was tested successfully. Half the SVT is ready to be installed and the other half is expected to be completed soon. Summary