Updates on protoDUNE detector calibration based on cathode piercing cosmic muons Ajib Paudel 1 , Tingjun Yang 2 , Glenn Horton-Smith 1 1 Kansas State University 2 Fermi National Accelerator Laboratory 6/27/2018 1
Br Brie ief revi view: : dQ/dx cali alibratio ion su summary ry dQ/dx values vary throughout the TPC due to a number of factors namely, misconfigured or shorted TPC channels, Space Charge Effect, attenuation due to impurities like CO 2 and O 2 present in liquid argon, diffusion etc. We have corrected dQ/dx values to make detector response uniform throughout the TPC using through going cathode crossing cosmic muons . Our corrected dQ/dx value is, Where C(y,z) is the correction factor for the Y-Z plane, and C(x) is the correction factor along drift direction, , Details of which can be found in DUNE collaboration meeting talk, link for which is https://indico.fnal.gov/event/14582/session/36/contribution/125/material/slides/0.pdf 6/27/2018 2
dE dE/dx calib libratio ion: Throughout the analysis we are using 3ms lifetime Monte Carlo MCC10 sample with SCE turned ON We used T0 tagged stopping muon tracks which have a well known dEdx vs residual range characteristic curve Once we estimate dQ/dx correction factor (which we have already done), we move to dE/dx from calibrated dQ/dx using modified box model. Using the standard relation between dE/dx and residual range in liquid Argon for stopping muons and comparing it with the relation we obtain from reconstructed stopping muon tracks, we will determine the calibration constant (for converting ADC/cm into number of electrons/cm) for protoDUNE detector. 6/27/2018 3
St Stoppin ing muon se selec lectio ion: Major challenge for the analysis was selecting a highly pure sample of stopping muons Cuts used: Fiducial volume cuts: Track start point should be within 20 cm or less from a TPC boundary Track end point should be more than 50cm from any TPC boundary Removing broken tracks: Looked for other tracks in the vicinity of our T0 tagged track which is within certain distance and makes a certain minimum angle with our T0 tagged track. (details in next slide) 6/27/2018 4
The figure below describes our minimum distance cut, Here (y1, z1) and (y2, z2) are the Y and Z start and end points (not in any particular order) of our T0 tagged track and (y3, z3) and (y4, z4) are the start and end points of any arbitrary track in the same event, If △ y=y2- y2 calculated <30cm and the cosine of the minimum angle between the two cos( 𝞲 )>0.97 or 𝞲 <14deg where, y2 calculated =((z4-z3)/(y4-y3))*(z2-z3)+y3 In addition we removed those tracks with △ y<50cm and the minimum angle between the two tracks is less than 3 deg. (This eliminates parallel tracks within 50 cm from T0 tagged tracks) 6/27/2018 5
APA boundary cuts: Removed tracks whose end or start Z coordinate lies within 5 cm of APA boundaries. Significant number of tracks were found to be broken in that region. Hit peak time cuts : Removed tracks with minimum hit peak time (Hit::PeakTime(), time of the signal peak in tick units) less than 2000 ticks. We achieved a purity ≈ 94% and efficiency ≈ 14.35% . On calculating the purity by varying the peak Time cut we found that purity is maximum if we remove tracks with minimum peak time less than 2000 ticks. 6/27/2018 6
Angular cuts : We are removing tracks which are moving straight towards the wire planes or parallel to the wire planes, as in the case of dQ/dx calibration details of which can be found in my earlier talks. Reversing residual range values: For tracks whose end and start points were wrongly labelled in reconstruction we reversed the residual range values such that the actual end point (with lower y-coordinate for cosmics) always has the lower residual range value. Plots for dQ/dx vs residual range after all the cuts described in earlier slides using pmtrack : Left: uncalibrated dQ/dx vs residual range, Right: calibrated dQ/dx vs residual range(tracks used for plot=1769) 6/27/2018 7
We then scanned some individual tracks for dQ/dx values. The reconstruction near the end of some tracks was not so good with pmtrack. Then we tried pandoraTrack reconstruction. Using the same selection cuts for stopping muons (as used with pmtrack), we obtained a purity of 95.6 .6% and efficiency ≈ 15% 15% . dQ/dx vs residual range for pandoraTrack. 6/27/2018 8
Following plots shows dQ/dx vs residual range for plane 1(showing the comparison): Left ft: pm pmtr track(1367 tr trac acks use used), Rig Right:pandoraTrack(1640 tracks used) in both the plots we have used calibrated dQ/dx and SCE on sample 6/27/2018 9
Following plots shows dQ/dx vs residual range for plan lane 0(showing the comparison): Le Left: pmtrack(1331 tr tracks use sed), Ri Right:pandoraTrack(1509 tracks used) in both the plots we have used calibrated dQ/dx and SCE on sample 6/27/2018 10
Here are some plots giving an idea on how well the end points have been reconstructed (ba (based on on pand pandoraTrack ): ): Top Left: No o SCE SCE cor orrectio ion ap applied , Top Right: SCE SCE cor orrection ap applied to to True En Endpoints, Bottom: SCE SCE OF OFF sam sample 6/27/2018 11
An Analysis is methodology for or dEdx Cali alibration: (Si (Simil ilar Method use sed by y Mic icroBooNE) reference:MicroBooNE-NOTE-1048 1048-PUB(2018) Used Modified Box model to get dE/dx from calibrated dQ/dx: where, C=Calibration constant, to be determined, which converts dQ/dx from ADC/cm to number of electrons/cm ɛ=0.5kV/cm is the protoDUNE electric field W ion =23.6 x 10 -6 MeV/electron (work function of argon) ρ =1.38 g/cm3 (liquid argon density at a pressure 18.0 psia) α =0.93 β’= 0.212 (kV/cm)(g/cm2)/MeV where, the last two parameters were measured by ArgoNeuT experiment at 0.481kV/cm. 6/27/2018 12
dEdx cali alibration Analysis Methodology con ontinued: • Divided the last 200cm of residual range values into 40 bins of 5cm each • Calculated the most probable value of dE/dx for each bin ( for a track pitch = 0.5cm-0.7cm ) using Landau Convoluted Gaussian function fit for a given calibration constant . Used tracks with track length greater then 150cm . • Converted the residual range values into Kinetic Energy(KE) using the tabulated data set in the particle data group and plotted MPV of dE/dx for each bin vs KE. • We compared the curve generated in the previous step with the prediction made by the “Landau - Vavilov” function in the MIP region of the muons (250 MeV - 450 MeV) and χ 2 value is calculated using the Equation, where, σ 2 = δ 2 fit , δ fit is the uncertainty associated with the MPV dE /dx extracted by fitting a Landau convoluted Gaussian function to the energy distribution. • Repeated above steps for different calibration constant values. Finally, calibration constant C corresponding to minimum χ 2 gives the correct calibration constant. 6/27/2018 13
RESULTS: Calibration Constant = (4.926 ±0.0030)x10 -3 and χ 2 min /d.o.f=1.265 resultssults: -Expected (theory) most probable value plot is drawn using Landau-Vavilov function. 6/27/2018 14
We observed that in the the fi first bin bin there is a big gap between the expected and observed values as dQ/dx values fluctuated towards the end on the track. Here are some plots showing dQ/dx distribution for a few bins: dE/dx for bin-4 (res range 15-20 cm ) 6/27/2018 15 dE/dx for bin 28 (135-140cm)
We repeated the analysis using the truth information (selecting only those tracks whose end point lies within 5cm of True track ends and found: Calibration Constant = (4.911 ±0.0035)x10 -3 and χ 2 min /d.o.f=1.283 6/27/2018 16
Summary: dQdx calibration factor has been calculated and Jon Paley has added the corrections to the database. pandoraTrack Algorithm is found to give better reconstruction near the track end compared to pmtrack and so we used pandoraTrack for our dE/dx calibration analysis Calibration constant has been estimated using Monte Carlo MCC10 sample with 3ms lifetime and SCE ON. 6/27/2018 17
References: • MicroBooNE-NOTE-1048-PUB(2018) • pdg.lbl.gov/2014/reviews/rpp2014-rev-passage-particles-matter.pdf 6/27/2018 18
Recommend
More recommend
