CLAS Cerenkov detector efficiency

Some results of the CLAS Cerenkov detector efficiency analysis based on February, 1998 runs.

The goal of this analysis is to obtain the mean number of photoelectrons for ALL Cerenkov detector acceptance. 1.6 GeV data with 38% and 57% of full magnetic field ( Torus current ~ 1500 A and 2250 A, Mini-torus current 6000 A) from February,1998 data ( ~21,700,000 events total ) were analyzed.

Selection procedure and applied cuts

RECSIS time-based tracking was used to select events with negative track.
Missing mass distribution was used to select elastically scattered electrons.
The cut on the fraction of the particle energy, deposited in EC was applied.
The cuts on X and Y deviations from the tracking matching point in EC were applied as well.

Typical number of selected electrons as a function of angles Theta and Phy is presented on the Fig. 1

Theta and Phy are accordingly polar and azimuthal angles of the entrance point on the surface of the Cerenkov detector in the sector reference system.

Results :

Sector 1
Sector 2
Sector 3
Sector 4
Sector 5
Sector 6

The white line shows the fiducial region, according to GEANT simulations.

The region of low number of photoelectrons in Sector 4 ( at angle Theta ~ 37 degrees and Phy < 0 ) appeared to be a result of the incorrect single photoelectron position for PMT #29 in the data map : it was 480 instead of ~100, so Sector 4 actually is not bad.

There is some region of low number of photoelectrons in sector 5 at angles Theta ~ 17 degrees. It should be noted that the number of events (electron candidates) also is small at this region. There are few possible reasons for that : Low DC efficiency, dead SC paddle or CC inefficiency. Evidently, this should be tested again.

Direct efficiency estimations.

We need to see the efficiency estimation as a function of detected photoelectrons, to be sure, that this efficiency is consistent with the Poisson distribution, and that we define the number of photoelectrons correctly. Unfortunately, we have to use December runs without CC in the trigger for this goal. ( Better use same runs for both. ) Runs number 7841, 7843 and 7880 were used for this estimation.

First,the electron candidates were selected, using missing mass distribution ( elastic peak position ), matching points in EC and SC. The plot of mean number of photoelectrons as a function of projective angles Theta, Phy for sector 1 were used to plot the number of electron candidates as a function of Nphe. ( For every event Theta, Phy was defined and respective Nphe was derived from this plot). Fig. 2 shows the number of ALL electron candidates and the number of inefficient ones (when there are no CC respond in the proper place) as a function Nphe
Fig. 3 shows the electron inefficiency as the ratio of number of inefficient to all events. The curve on the Fig. 3 is the inefficiency, expected from Poisson distribution, when taken in mind that in calculation of the mean number of photoelectrons only events with non-zero Nphe were used.
One can see from Fig.3 that

The electron inefficiency as a function of Nphe is close to those expected from Poisson distribution.
The difference could be a result of SPE drift, (we used different runs for this estimation) , small difference in Nphe plot because of different magnetic field values ( 38% and 57% magnetic field) or SPE uncertainty.

Test of magnetic field influence.

We used projective angles Theta and Phy for Nphe or inefficiency plot because we hope, that this plot will be the same for ALL electron momenta. To test this assumption, we made some GEANT calculation and compare Nphe plots for two different magnetic fields : 38% and 57% (1550 and 2250 A torus current) at 1.645 GeV/c initial momentum.

GEANT estimations.

The problem is that for the electrons with different momenta coming to the same point at the inner surface of Cerenkov detector, the input angle is different, so different could be the Nphe .

Fig. 4 and Fig. 5 the GEANT estimations for Nphe plot for 0.8 GeV/c and 3.0 GeV/c electrons emitted from the target (FULL magnetic field). This is about the upper estimation of possible difference in Nphe , because actually the momentum - angular dependence always presents and it is unlikely to find electrons with very different momenta coming to the same point.
Fig. 6 and Fig. 7 show respectively the Nphe difference and inefficiency difference for these cases. Althow this difference is ruther small, one can see the systematic difference in electron inefficiency with period of about the width of the mirror (2 degrees on Theta). The reason probably is that for different input angles different mirrors and PMT could work, so this is a result of small shift of ( Nphe / inefficiency ) plot for other momentum. The possible decision could be to use Theta and Phy not for entring point, but for EC plane point, where we saw the self-focusing effect and used it for eid0 programm. In that case the inefficiency pattern should be the same for all momenta.

Two magnetic fields data comparing.

Fig. 8 shows inefficiency plot for 38% and 57% magnetic field. Most of phase space has low inefficiency, some inefficiency region is near the sector middle plane ( Phy = 0 ) and at large phy near Theta = 28 and 33 degrees.
Fig. 9 shows the plot of inefficiency difference for 38% and 57% . One can see that it it close to those expected from GEANT estimations.

The questions so far:

After fixing SPE positions (J. Price is working on it) for all February runs it will be good to remake this analysis in terms of Theta and Phy at some point near EC plane. (The best is to have the distance to inner surface of Cerenkov detector to be equal to the ray tracing distance) And after that the plot of Cerenkov detector efficiency could be ready for use for physics estimations.

Mail to me!