Atmospheric Electron Neutrinos in the MINOS Far Detector

1
Atmospheric Electron Neutrinos in the MINOS Far
Detector

• Ben Speakman, Unversity of Minnesota
• Thesis Defense Talk
• May 3, 2007

2
Talk Outline

• A Brief History of Atmospheric Neutrinos
• The MINOS Far Detector
• Analysis Motivation
• Showering and Track-like Event Selection
• Double Ratio Measurement
• Neutrino Oscillation Analysis
• Atmospheric Neutrino Flux Measurement

3
The Neutrinos Anomalous Timeline
60 years after the trouble started, along come a
new batch of anomalies

• 1914 Beta-decay spectrum anomaly
• 1930 Pauli and Fermi posit the neutrino
• 1956- Reines and Cowan observe the neutrino
• 1962 AGS Neutrino experiment observes 2
  neutrino flavors
• 1991- LEP Accelerator determines there are
  exactly 3 neutrino flavors

• 1970s Solar Neutrino Anomaly
• 1980s Atmospheric Neutrino Anomaly
• In both cases, anomalous deficits from expectation

? Birth of Neutrino Oscillation
4
What are Atmospheric Neutrinos?

• Cosmic rays hit Earths upper atmosphere
• Interactions create hadrons (p, K, etc.)
• hadrons decay to m n
• p or K ? m nm(nm)
• m ? e ne(ne) nm(nm)
• Expect nm 2 ? ne
• Expect nm-up nm-down

5
Atmospheric Neutrino Anomaly
Super-K (50 kton H20)

• 70s P-decay Experiments
• P decay experimental requirements
• deep underground
• lots of protons
• Atm.-n interactions present background to P
  decay
• Anomalous Atm.-n Fluxes
• nm lt 2 x ne
• nup lt ndown
• Oscillation of massive neutrinos explanation

Soudan2 (963 tons steel)
6
Explaining the Atmospheric Neutrino Deficit
Oscillated Neutrino Spectrum sin2(2q) 1.0, Dm2
0.0027eV2 P 1- sin2(2q) sin2(1.27 Dm2L/E)

• Oscillation of nm is the cause of anomaly
• Osc. disappearance experiment
• Experiments with E and q resolution can measure a
  distribution in L/E
• L/E distribution offers oscillation likelihood
• L/E distribution in also decay dependent.

Decayed Neutrino Spectrum L/E Characteristic C
4000 km/GeV P e C(L/E)
7
Understanding the Atmospheric Neutrino Flux

• Flux Model Presented by Barr et. al.
• Primary Cosmic Flux
• Solar Cycle
• Geomagnetic fields
• Hadronization
• Advanced Model uses 3D Hadronization
• Uncertainties

• Flux Model Uncertainties
• Total flux 20
• R (nm / ne) 1
• R (nm-up / nm-down) 1
• Flux uncertainty could account for deficit.
• Oscillation analysis requires a flux model
  normalization
• Double flavor ratio negates the total flux
  uncertainty

3D Enhancement of flux at horizon
8
MINOS Far Detector

• Soudan Underground Lab
• 27th level 2314 feet
• Sampling Hadron Calorimeter
• 486 steel plates (2.54 cm x 8m)
• 484 x 192 active scintillator strips
• PMT photon measurement
• Current coil produces 1.5 T magnetic field.
• Cosmic-ray veto shield actively reduces the
  background.

9
Hodoscopic Scintillator Planes
10
Active Shielding

• Rejection of Cosmic Muons
• Four Total Sections
• Double layer on top
• 30 modules (600 strips)
• Single layer on sides
• 6 modules (120 strips)
• High gain on PMT to increase CR rejection
  efficiency
• Tag events as vetoed if vertex and shield hits
  are space and time matched

11
Neutrino Interactionsin the Far Detector
Reconstruction
12
Event Reconstruction

• Calibrated Digit Response
• Demultiplexing
• Resolve the 8 fold ambiguity
• Remove hits tagged as cross talk
• Find Tracks and Showers
• Identify vertex ( end) location and direction
• Calculate Track Energy
• Range and Curvature
• Find Shower Energy
• Summed Pulse Height

13
Analysis Motivation Strategy

• Isolate contained vertex (CV) atmospheric
  ne CC n NC rich Showering events and
  nm CC rich Track-like events.
• Evaluate oscillation with the flavor double ratio
  
• R ( Trk/ Shw)Data/MC (nm / ne)Data/MC
• Measure the atmospheric neutrino flux with the
  combination of Trk and Shw.
• Use data set from construction completion
  (8/2003) until beam running (2/2005), total of
  418.5 live days.

14
Atmospheric n Selection and Cosmic Ray Reduction
How does one go about reducing the cosmic-ray
background?

• Bury the detector, ½ mile should do it.
• Select contained vertex (CV) events.
• Remove steep and shallow events.
• Observe n-like event topologies.
• Use the cosmic-ray veto shield.

15
Energy and Vertex Containment
Fiducial Mass 3.94 kton

• Energy Containment low-level filter
• 3-Dim Position of Hits
• 30 cm to Outer Edge
• 5 Planes to SM Edge
• Defines events as PC, FC, or through-going
• Vertex Containment event-level filter
• Shower/Track Vertex
• 50 cm to Outer Edge
• 5 Planes to SM Edge
• 40 cm from Center, or 100cm from Center on outer
  planes

16
CV Event Selection Strategy

• Showering Events
• 1 Shower ( 5 Planes)
• Clean Event Shower
• Shower Length Cut Optimization
• Short Shower ? 8 Planes
• Long Shower gt 8 Planes
• Trace Z Selection
• Shower Topology
• Principal Axes Moments
• Energy Deposition Profile
• Veto Shield Tagging

• Track-like Events
• 1 Track ( 8 Planes)
• Clean Event Track
• FC PC Down Tracks
• Trace Z Selection
• Vertex Hits Topology
• Veto Shield Tagging
• PC Up Tracks
• Timing Quality

17
TraceZ Enhanced Containment

• Trace the distance back the nearest edge.
• Project Trace on to the Z-axis ? TraceZ
• Tracks TraceZ gt 50 cm
• Showers Use ?TraceZ
• ?TraceZ gt 60 cm (Long Shw)
• ?TraceZ gt 80 cm (Short Shw)

18
Shower Topology Selection

• Moment about Principal Axis

• Shower Energy Deposition Profile
• Shower energy deposition in a plane number of
  strips
• RMSStpPln lt(strips/plane)2gt½

19
FC / PCDN Track Topology Selection
Topology of hits in vertex planes are examined
for two pathologies.

• DT Distance from vertex in each view, use mean
  and RMS.
• QMax Maximum Vertex Plane Charge, for steep
  events

20
PCUP Track Timing Selection

• Track direction is decided with timing fit

• Quality of Upward vs Downward Fit verifies
  direction

• One direction fits better, and track sides
  labeled vertex and end

21
Veto Shield Tagging

• Tagging Efficiencies
• Cosmic-ray eff. (e)
• Atmos n eff (h)
• Data used to measure Efficiencies.
• D (data) S (Signal) V(Vetoed)
• D Nn Nm
• V hNn eNm
• S (1-h)Nn (1-e)Nm

• Coincident hits in the shield will tag Shower or
  FC/PCDN Track as vetoed.

Vetoed Track
Shield Hits
22
Shower Selection Results
eshw 0.976 0.002 hshw 0.0261 0.0011
Scale MC n, N by (e - h)/e to match shield Scale
Vetoed by (1-h)/e to measure CR m
23
Selected Shower Events
24
FC / PCDN Track Selection Results
etrk 0.973 0.004 htrk 0.0255 0.0011
Scale MC n, N by (e - h)/e to match shield Scale
Vetoed by (1-h)/e to measure CR m
25
Selected FC Track Events
26
Selected PCDN Track Events
27
PCUP Track Selection
28
Atmospheric Neutrino Flavor Double Ratio

• Double Ratio ( Tracks/ Showers)(Data/MC)
• Observe 89 Showers and 112 Tracks
• Use the shield efficiencies to adjust FC/PCDN
  Track and Shower expectations.
• ExpShw 88.8 0.9 (MC Statistical)
• ExpTrk 149.8 0.9
• Coverages found with Monte Carlo, also find that
  observed disfavours null oscillation hypothesis
  to 98.7 single-sided confidence limit.

29
Double Ratio Systematic Errors

• Use the following systematic variances to
  observe
• D( Tracks), D( Showers), and DR
• Tracking Energy Cutoff 100keV vs.10keV
• Neutrino Flux Normalization 20
• Quasi-Elastic cross-section 10
• Neutral-Current cross-section 25
• Neutron Flux Normalization 20

30
Double Ratio Systematic Errors
Cumulative Systematic Error DR 5.93
31
Double Ratio Results

• Statistically disfavors null oscillation with
  98.7 single-sided confidence limit
• Accounting for systematic error,
• disfavors null oscillation with 98.0 SSCL

32
Atmospheric Neutrino Flux Measurement

• In order to measure the flux, oscillations must
  be taken into account.
• Flux measurement is expressed as a normalization
  term to a flux model (Satm).
• Used to isolated methods to measure the flux
  normalization
• Frequentist Fit of Double Ratio, combined with a
  showering event count normalization.
• Likelihood Fit to Shw Trk, with the flux
  scale as a free parameter.

33
Maximum Likelihood Method

• Minimize the negative-log likelihood.
• -2lnL 2SEi Oiln(Ei) (a/s)2
• Three parameter fit Satm, sin2(2q), Dm2.
• Two bins shower and track count.
• Penalize the scale factor a Satm 1.0
• Use s 2.0 to penalize weakly, permitting the
  flux normalization to float freely and account
  for oscillation

34
Weakly Penalized Scale Factor
Combine D 2 ln (L) With Best Fit Satm To
measure Satm
35
Flux Scale Result
Cumulative Systematic Error 8.92 Satm 1.07
0.12(stat.) 0.09(syst.) Model Gives Satm
1.0 0.2
36
Analysis Results

• Double Ratio
• 98.0 Rejection of Null Oscillation
• Barr Flux Model Normalizaton
• Satm 1.07 0.12(stat.) 0.09(syst.)
• Compare to Model Satm 1.0 0.2

37
Projected Double Ratio Sensitivity
Project the double ratio and compare to SK R
(0.63 0.03 0.05) for 46 kton-yr
38
Conclusion

• Atmospheric flavor double ratio suggests neutrino
  oscillation, disfavors null oscillation with high
  confidence.
• Understanding of the atmospheric neutrino flux
  model has been improved with the measurement of a
  normalization for the Barr flux model.
• Future possibilities for this analysis
• Improve selection and statistics.
• Sub-dominant oscillation studies might be
  performed with enhanced shower reconstruction.

39
Backup Slides
40
Optical Multiplexing

• 16-pixel Hamamatsu PMT (e 13 )
• 8 fibers optically summed per pixel
• Double-ended readout assist demultiplexing
• Readout by VA electronics.
• 2 Trigger system
• 2/36 PMT
• 4/5 Planes

41
Far Detector Magnet

• Each SM has its own coil
• Coil Wrapped, total current 15 kAturns and
  BField 1.5T
• PC muon momentum with curvature
• Additional measurement to the range momentum
• Enhances containment

42
Selected PCUP Track Events
43
Neutrino Oscillation

• If the neutrino has mass, leptonic neutrino may
  be different from massive neutrino
• Osc Prob. Magnitude of inner product
• Simplification 2 neutrinos and a mixing angle
  (q)
• In two neutrino framework, experiments look for
  apperance P(nl?nh) and disappearance P(nl?nl)

44
Frequentist Double Ratio Fit

• An oscillation hypothesis (null or otherwise)
  posits an expected double ratio
• The expected double ratio is fluctuated to
  estimate the confidence limit for rejecting the
  measured double ratio.
• Found expected shower count for each oscillation
  hypothesis
• Weight expected shower event count by (1-
  rejection CL).
• The distribution of (1-CL) weighted shower count
  is centered at the shower count to use, and the
  width is a systematic error due to oscillation
  uncertainty

45
DR Oscillation Fit Normalize
Find range of confidence limits from DR
frequentist fit Apply the range of CL values to
shower count
46
Flux ScaleMeasurement and Systematics

• Obtain Flux Scale (Sn) from expected and observed
  shower counts.
• Sn Obsshw / Expshw
• Investigate the following systematic variances
• Tracking Energy Cutoff 100keV vs.10keV
• Quasi-Elastic cross-section 10
• Neutral-Current cross-section 25
• Neutron Flux Normalization 20
• Oscillation (1-CL) Weight Shower Count RMS

47
Flux Scale Method 1 Result
Cumulative Systematic Error 9.03 Sn 1.06
0.12(stat.) 0.09(syst.)
48
Flux Scale Method Comparison

• Double Ratio / Shower Count Method
• Sn 1.06 0.12(stat.) 0.09(syst.)
• Statistically Consistent with Sn1.0 to 58.1
• Stat Syst Consistent with Sn1.0 to 66.2
• Likelihood Track Shower Count Method
• Sn 1.07 0.12(stat.) 0.09(syst.)
• Consistent with first method.

49
Comparison of Oscillation Slices
2.43x10-4 eV2 ltDm2 lt2.02x10-2 eV2
2.28x10-4 eV2 ltDm2 lt1.78x10-2 eV2

• Take 1D Slices from 2D Osc grid and compare.
• Frequentist fit deals with 1 constraint and 2
  parameters.
• Likelihood fit deals with 2 constraints and 3
  parameters.
• Both fits are under-constrained, but differ in
  shape.
• 68 CL in red boxes, compare nicely between
  methods.

sin2(2q) gt 0.611
sin2(2q) gt 0.592