EEMC Tower Calibrations with Electrons (L3 TPC Tracks)
Piotr A. Zolnierczuk, Indiana University Cyclotron Facility < zolnie at iucf dot indiana dot edu >

Note: The analysis presented here follows closely Alex P. Suaide analysis for Barrel EMC
So far the event sample is 7 Mevents big and spans days 120 (Apr 30, 2003) through 148 (May 28, 2003).


L3 TPC tracks

The L3 TPC tracks with at least Npoints=20 in the track are shown in the two plots below. The left one shows tracks for all STAR triggers and the right one only the ones associated with EEMC triggers (ETH1 and ETH2). The red polygones show our dE/dx and p cut used to select electron candidates. The thick red line corresponds to a "tight" dE/dx cut (approximately 0.36<dE/dx<0.44 keV/cm) and the thin red line corresponds to a "loose" dE/dx cut (approximately 0.34<dE/dx<0.46 keV/cm). The black box shows the "background" region used for comparison to the electron sample.
L3 TPC tracks for all STAR triggers (min. points in the track 20) L3 TPC tracks for EEMC triggers only (min. points in the track 20)


L3 tracks dE/dx distributions

The plots below show the dE/dx distribution for the tracks with at least Npoints (15,20,25,30) and with 1.5 GeV < p < 10.0 GeV for the two trigger conditions mentioned above (all STAR and EEMC only). What is actually plotted is a log((dE/dx)/BB), where the dE/dx is taken from L3 track info and the BB stands for the Bethe-Bloch dE/dx prediction for an electron with a given momentum.

A "tri-gaussian" fit to the log((dE/dx)/BB) distribution was then performed ( min points requirement was varied from 15 to 30).
On the plots below: the blue gaussian bell denotes electron candidates, the black and green ones - background events and the red one the overall fit.

The dE/dx cuts ("tight" and "loose") mentioned in the previous section are shown as thick blue solid vertical bars. The upper two plots show the tight cut -0.05 < log((dE/dx)/BB) < +0.15 and the lower two plots show the loose cut -0.10 < log((dE/dx)/BB) < +0.20.

The backround events were chosen with a -0.60 < log((dE/dx)/BB) < -0.20; cut (thin black solid bars)

L3 tracks dE/dx distribution
tight dE/dx cut, all STAR triggers
L3 tracks dE/dx distribution
tight dE/dx cut, EEMC triggers only

L3 tracks dE/dx distribution
loose dE/dx cut, all STAR triggers
L3 tracks dE/dx distribution
loose dE/dx cut, EEMC triggers only


Electron Sample

Based on the gaussian fits above one can calculate the following quantities:
  1. Ne - number of electron candidates, i.e. the integral of the blue gaussian bell within the dE/dx cuts
  2. Pe - electron sample purity, i.e. Ne over the total number of events within the dE/dx cuts.
  3. Ee - the efficiency of the dE/dx cut, i.e. Ne over the total number of electrons under the blue gaussian bell.

"Tight" dE/dx cut
  All STAR triggers EEMC triggers only
min points NePe [%]Ee [%] NePe [%]Ee [%]
152474635.3870.95 304036.7875.59
201466045.2273.12 158944.0678.77
25 932549.2579.13 90744.9183.42
30 765958.0179.14 84264.4182.94

"Loose" dE/dx cut
  All STAR triggers EEMC triggers only
min points NePe [%]Ee [%] NePe [%]Ee [%]
153141025.5390.05 374426.0893.09
201835131.2491.53 191929.4995.12
251119433.2994.99 105629.0297.10
30 918740.6694.92 98243.1196.75


Calibration Constant

The electron (and background) track candidates were selected according to the critetria described above. For the tracks with a matching EEMC tower values of momentum/energy were histogrammed. The momenta (p) were taken from L3 TPC tracks and the energies (E) were taken from the EEMC data according to the following formula:
   E = ( ADC - PED ) * g * k
where

  • ADC - is the ADC signal measured in a EEMC tower
  • PED - is the tower pedestal
  • g - is the tower relative gain (from MIP analysis)
  • k - is the approximate calibration constant (taken to be 50 chan/GeV)
  • The thick solid blue histogram shows electron candidate events and the thin black one the background candidates (upper left below). The background histogram was normalized to the electron one requiring that the number of events for 2.5 < p/E < 10.0 be the same. The upper left plot show the "electron - (normalized)background" histogram. Gaussian fits were performed for both "electron" and "electron-background" histograms (solid red line). The lower left distribution is eta versus p/E for the electron candidates and the lower right one shows all the matched L3 tracks that fulfill the minimum Npoints and trigger conditions.

    p/E distribution
    tight dE/dx cut, all STAR triggers
    p/E distribution
    tight dE/dx cut, EEMC triggers only

    p/E distribution
    loose dE/dx cut, all STAR triggers
    p/E distribution
    loose dE/dx cut, EEMC triggers only

    Calibration Constant (chan/GeV)

    Finally the the mean values of the gaussian fits to the p/E distributions were converted to the absolute calibration constants k = 50.0 / mean , where mean is the mean value of the gaussian fit. A table containing the resulting calibration constants is presented below:

    "Tight" dE/dx cut
      All STAR triggers EEMC triggers only
    Npoints no background subtractionwith background subtraction no background subtractionwith background subtraction
    1553.05 +/- 2.8656.39 +/- 2.6559.53 +/- 4.6657.86 +/- 3.58
    2055.58 +/- 5.5457.53 +/- 5.6170.94 +/- 16.0871.41 +/-14.69
    25 poor fit poor fit poor fit poor fit
    30 poor fit poor fit poor fit poor fit

    "Loose" dE/dx cut
      All STAR triggers EEMC triggers only
    Npoints no background subtractionwith background subtraction no background subtractionwith background subtraction
    1548.84 +/- 2.2552.98 +/- 2.0857.03 +/- 5.0053.11 +/- 3.19
    2051.21 +/- 3.4554.81 +/- 3.9861.90 +/- 9.1861.21 +/- 7.71
    25 poor fit poor fit poor fit poor fit
    30 poor fit poor fit poor fit poor fit

    A graphical presentation of the above table is below


    Conclusions

    To be written