Track reconstruction at the third level trigger will allow particle identification (PID). Two complementary methods are used in STAR for PID: dE/dx measurements (TPC and SVT) and time of flight techniques with the TOF detector. The dE/dx study of the PID capabilities at L3 requires a detailed simulation of cluster formation and finding. This is outside the scope of the present analysis. Therefore only the TOF PID capabilities will be investigated.
Two quantities enter in the determination of the particle mass with a TOF detector: the trajectory length and time of flight. In order to understand how PID will behave at L3, the resolution in the determination of those two quantities was studied for the different pad-row configurations.
Figure 5 shows the difference between the real track length and the value obtained from the reconstruction. As in previous distributions, the measurement of the track length deteriorates when only short pad-row configurations (21-45 and 30-45) with no vertex are used. However the vertex inclusion improves the resolution substantially. Using even or third pad-rows gives quite acceptable results even without the vertex constraint. Table 7 contains the numerical values of the length resolutions obtained by fitting gaussians to the distributions.
Figure 6 shows the time of flight resolution for the different configurations. It is to be noted that the time resolution depends on the tracking quality. This is due to the correction applied to the raw time provided by the TOF. This correction is a function of the distance of the particle impact in the TOF counter to the its photo-multiplier. The quality of the tracking will determine the accuracy in the impact measurement. As can be observed, the time resolution does not seem to depend on the vertex constraint presence. In addition configuration 21-45 and 30-45 give better results than even and third rows. This is not difficult to understand since the more information available in the region close to the TOF the better the track hit position will be. Table 7 contains the numerical values of the time resolutions obtained by fitting gaussians to the distributions.
Figure 7 shows the particle mass distribution, m,
obtained combining tracking with TOF information in the
range 
GeV.
Tracks were required to be long enough (>20 pad-rows for
the full TPC, even rows and third rows and >10 for the
other configurations).
The kaon and proton peaks are clearly visible except for the
configurations 21-45 and 30-45 with no vertex. However when
the vertex is included clear signals can be observed for all
configurations. This plot shows that particle identification with
the TOF at L3 should be possible.