Neutrino Oscillation Results from MINOS and MiniBooNE Tobias Raufer Rutherford Appleton Laboratory for the MINOS collaboration FPCP08, Taipei, 5-9 May, 2008
Jan 17, 2016
Neutrino Oscillation Results from MINOS and MiniBooNE
Tobias RauferRutherford Appleton
Laboratory
for the MINOS collaboration
FPCP08, Taipei, 5-9 May, 2008
Introduction to Neutrino Oscillations• Neutrino masses and mixing• Sterile neutrinos • Current world knowledge on neutrino oscillation parameters
Results from MINOS• The NuMI beam and MINOS detectors• Charged-current disappearance analysis• Neutral current analysis
• νe appearance status
Results from MiniBooNE• The MiniBooNE Beam and Detector• Results
Outlook
Overview
2
Introduction to Neutrino Oscillations
Neutrinos mix
Two conditions necessary for neutrino oscillations:– Neutrinos mix– Neutrinos are massive
Flavour eigenstates
Mass eigenstates
flavour eigenstates govern interactions
mass eigenstates propagate
4
Talk by Reyco Henning
Atmospheric+LBL Chooz Solar+KamLAND Majorana
Neutrinos are massive
There are only 3 light neutrinos coupling to the Z0 …
5
Neutrinos are massive
There are only 3 light neutrinos coupling to the Z0 …
6
Neutrinos are massive
… but there could be sterile neutrinos!
There are only 3 light neutrinos coupling to the Z0 …
7
Neutrinos oscillate
If mass and weak eigenstates are different:
– Neutrino is produced in weak eigenstate.
– It travels a distance L as a (superposition of) mass
eigenstate(s).
– It is detected as a (possibly different) weak eigenstate.
8
Neutrino oscillations: disappearance
• Experiments in Homestake, Kamioka
and Sudbury established deficit of νe from
the sun
• Total solar flux measured by SNO
agrees with prediction
• Super-K, K2K and MINOS have
measured L/E form of νμ disappearance
rate
• KamLAND have measured L/E form of
reactor anti-νe disapperance rate
•CHOOZ have placed a limit on θ13 using
non-disappearance of reactor anti-νe
arXiv:0801.4589 [hep-ex]
9
Super-K Zenith Angle
Upward stopping Sub GeV Multi ring ()
Multi GeV Multi ring ()
Upward through going
Honda Best fitSub GeV 1ring e-like
Sub GeV 1ring -like
Multi GeV 1ring e-like
Multi-GeV 1ring -like + Partially Contained
Reactor Experiments
νe
νe
νe
νe
νe
νe
Distance
Pro
babi
lity
νe
1.0
Well understood, isotropic Well understood, isotropic source of electron anti-source of electron anti-neutrinosneutrinos
Oscillations observed as Oscillations observed as disappearance of disappearance of ννee
sinsin2222θθ
1313
Survival ProbabilitySurvival Probability
+ O(m122 / m13
2)
P(e e)1 sin2 213 sin2(1.27m13
2 L /E )
Neutrino oscillations: appearance
• Another signature of neutrino oscillation is appearance of ''wrong'' flavour neutrinos• LNSD observes excess ofanti-νe
in anti-νμ beam:
87.9 ± 22.4 ± 6.0 (3.8σ)
LSND signal
12
Neutrino oscillation parameters
MINOS result later in the talk!
13
The MINOS experiment
The MINOS experiment
MINOS (Main Injector Neutrino Oscillation Search)– Long-baseline neutrino
oscillation experiment– Neutrino beam provided by 120
GeV protons from the Fermilab Main Injector
Basic concept– Measure energy spectrum at the
Near Detector, at Fermilab– Measure energy spectrum at the
Far Detector, 735 km away, deep underground in the Soudan Mine
– Compare Near and Far measurements to study neutrino oscillations 15
Producing Neutrinos
• Neutrinos from the Main Injector (NuMI)
• 10 μs spill of 120 GeV protons every 2.4 s
• 180 kW typical beam power
• 2.5 1013 protons per pulse
• Neutrino spectrum changes
with target position
Tobias Raufer 1616
MINOS Detectors
Detectors magnetised to ~1.3 T
GPS time-stamping to synch FD data to ND/Beam
Flexible software triggering in DAQ PCs: FD triggers from FNAL over IP
Coil
Veto Shield Far
Near
Plane installation fully completed on Aug 11, 2004
5.4 kt mass, 8830m 484 steel/scintillator planes Divided into 2 super modules
M64 multi-anode PMTs
1 kt mass, 3.84.815m282 steel and 153 scintillator
planesFront 120 planes Calorimeter
Remaining planes SpectrometerM16 multi-anode PMTs
Tobias Raufer 1717
νμ CC Event νe CC EventNC Eventνμ CC Event NC Event νe CC Event
Event Topologies
18
νμ CC Event νe CC EventNC Eventνμ CC EventUZ
VZ
long μ track & hadronic activity at vertex
3.5m
Monte CarloNC Event νe CC Event
Event Topologies
19
νμ CC Event νe CC EventNC Eventνμ CC EventUZ
VZ
long μ track & hadronic activity at vertex
3.5m
Monte CarloNC Event
short event, often diffuse
1.8m
νe CC Event
Event Topologies
20
νμ CC Event νe CC EventNC Eventνμ CC EventUZ
VZ
long μ track & hadronic activity at vertex
3.5m
Monte CarloNC Event
short event, often diffuse
1.8m
νe CC Event
short, with typical EM shower profile
2.3m
Event Topologies
21
Event Topologiesνμ CC Event νe CC EventNC Eventνμ CC Event
UZ
VZ
long μ track & hadronic activity at vertex
3.5m
Monte CarloNC Event
short event, often diffuse
1.8m
νe CC Event
short, with typical EM shower profile
2.3m
Energy resolution
π±: 55%/E(GeV)
μ±: 6% range, 10% curvature
22
CC disappearance measurement
CC disappearance
Selecting charged-current events:• Reconstructed track with θ<53° w.r.t. beam direction
• Vertex in fiducial volume
• In time with beam spill
• Reject NC background using a likelihood ratio discriminant
constructed from 6 variables, e.g.
24
LE-10/170kA LE-10/185kA
pME/200kA
Horn off
LE-10/200kA
pHE/200kA
Hadron Production Tuning
Use tuning of hadron production in CC events to provide flux corrections for Monte Carlo
Parameterize Fluka2005 prediction as a function of xF and pT
Perform fit which reweights neutrino parent pion xF and pT to improve data/MC agreement
25
Hadron production in the NuMI target has large uncertainties! uncertain beam flux
• directly use Near Detector data to perform extrapolation between Near and Far
• use Monte Carlo to provide necessary corrections due to energy smearing and acceptance.
• use our knowledge of pion decay kinematics and the geometry of our beamline to predict the FD energy spectrum from the measured ND spectrum
Predicting the FD spectrum
FD
Decay Pipe
π+Target
ND
p
26
CC disappearance Result
m322 2.38 0.16
0.20(stat syst) 10 3eV2
sin2(223) 1.00 0.08(stat syst)
27
NC measurement
Neutral Current Analysis
Why look at NC events?• If oscillations only involve active neutrinos, NC events
are unaffected.• Oscillations into sterile neutrinos cause energy
dependent deficit in neutral current energy spectrum.
Toy Simulation
No νs
With νs mixing
29
NC Event pre-selection
• FD: remove cosmics, detector noise and split events– Fiducial volume and cleaning cuts
• ND: many interactions in one beam spill, both inside the detector and in the surrounding rock– Separate events based on topology and timing– Tight fiducial volume
Calorimeter Spectrometer
30
NC Event selection
• Event classified as NC-like if:
− event length < 60 planes
− has no reconstructed track or
− has one reconstructed track that does not protrude more than 5 planes beyond the shower
• Final neutral current event selection proceeds via cuts on three variables• Error envelopes shown reflect systematic uncertainties due to cross-section modeling
and beam modeling
Excluded Excluded
Excluded
31
NC Energy Spectrum
• Far Detector reconstructed energy spectrum for NC-like events.
• Oscillation parameters are fixed. MC predictions with Θ13=0 and Θ13 at the CHOOZ limit are shown.
32
Is there a deficit?
• Comparisons between observed Data and MC Prediction
• Significance is given by
• For the 0-3 GeV reconstructed energy range, a 1.15σ
difference between Data and Monte Carlo is observed in the case where Θ13 = 0.
33
4-flavour Model
• Introduce one additional, sterile neutrino
• Assume Δm241= 0
• Oscillation at single mass scale • Oscillation probabilities simplify to:
• Fit for Δm231, |Uμ3|2 and |Us3|2
• Joint fit of NC and CC spectra
• Fix |Ue3|2 = 0 and 0.04 (CHOOZ limit) 34
4-flavour Result
• 90% C.L. contour for the fit to |Us3|2 and |Uμ3|2
• Showing the limiting cases: |Ue3|2=0 and |Ue3|2=0.04
35
Outlook
• 90% C.L. sensitivity curves for different NuMI beam exposures
• Input values of oscillation parameters– |U3|2 = 0.5, |Us3|2 = 0.1, Δm2
32 = 2.38 x 10-3 eV2,|Ue3|2 = 0
• Only MC events are used
36
νe appearance search
MINOS νe appearance search
• Search for far detector νe appearance in initially 99% νμ beam• Select νe with neural net based algorithm• Selected near detector events are mostly CCνμ and NC• Selection depends on details of hadronic simulation• Solution: use two independent data driven methods to estimate NC and CCνμ backgrounds
selected νe sample
38
MINOS νe sensitivity
• Projected limits shown with current and expected MINOS exposure
• At CHOOZ limit expect 12 νe signal events and 42 background events with 3.25x1020 protons
• Use sidebands to study predicted far detector backgrounds
• Expect first result later this year
39
The MiniBooNE experiment
MiniBooNE searches for νμ→νe
• 8GeV/c protons hit beryllium target
• 4x1012 protons/spill with up to 4Hz rate
• 174kA pulsed magnetic horn focuses positively
charged hadrons: x6 flux gain
• Detector is 800 tons of mineral oil placed with L/E
similar to LSND:
LSND: 0.03km/0.05GeV ~ 0.60 km/GeV
MiniBooNE: 0.50km/0.80GeV ~ 0.63 km/GeV
41
MiniBooNE flux prediction
• HARP pion data were fit with
Sanford-Wang parametrization:
~17% uncertainty
• Fit world kaon data in 10-24GeV
using Feynman scaling:
~30% uncertainty
• Kaon flux is checked with off-axis
''Little Muon Counter'' and high
energy events
• Geant4 simulation of beamline:
target, horn, decay volume and
absorber
K
e e
K e e
e/= 0.5%
42
veto cutremoves cosmic μ
CC1π3 subevents
CCQE2 subevents
• PMTs collect scintillation and Cherenkov light
• Subvent is set of PMT hits close in time
• Select subevents within beam spill window
• Number of veto hits < 6
• Number of tank hits > 200
• Fiducial R < 500cm
• Only 1 subevent in spill window -
removes CC1π and CCQE events
MiniBooNE beam eventsveto hits<6 veto hits<6
tank hits>200
removes electrons from cosmic μdecays
43
MiniBooNE νe selection: step 1
Signal
Background
Background
,ΦE
t,x,y,z
light
• Maximize likelihood of observed hits under 2 hypotheses: 1) track is an electron 2) track is a muon• Vary 7 parameters
Select electron tracks using likelihood ratio
44
MiniBooNE νe selection: step 2
• Maximize likelihood of observed hits under 2 hypotheses with 10 parameters: 1) event is electron 2) event is π0→γγ
• Select electron events using likelihood ratios
E1,1,Φ1
t,x,y,z
lights1
s2E2,2Φ2
blin
d
π0 mass≈135Mev
45
MiniBooNE νe signal prediction
• Blind analysis: signal box was opened in steps by gradually revealing more details
about data events
• Event selection tuning and background estimation used events outside signal
region
• ''Bad'' initial χ2 for data and predicted visible energy (null and best fit): move energy
threshold to 475MeV
46
• No excess of νe events in expected
signal region from LSND
• Significance (stat + syst error):475-1250 MeV: 22 ± 40300- 475 MeV: 95 ± 28200- 300 MeV: 91 ± 31
What does it all mean?
LSND was wrong? Difference between νe and anti-νe? New physics that doesn't scale with
L/E? Detector or flux effects?
MiniBooNE Result
47
Summary & Outlook• MINOS
– MINOS is steadily accumulating data– CC disappearance result for 2.5 x 1020 POT:
– NC result for 2.5 x 1020 POT:• 3-flavour analysis: 1.15σ deficit for E < 3GeV
consistent with no sterile admixture
• 4-flavour analysis:
In the near future:– Updated CC result with more data and an improved analysis– First MINOS electron-neutrino appearance result
• MiniBooNE– No excess observed – Incompatible with LSND at 98% C.L. (two-flavour approx.)
m322 2.38 0.16
0.20(stat syst) 10 3eV2, sin2(223) 1.00 0.08(stat syst)
Us3
2 0.14 0.130.18, for Ue3
2 0 (no e admixture)
Us3
2 0.21 0.120.20, for Ue3
2 0.04 ( e admixture)
48
Back-up slides
CC future sensitivity
50
Accumulated Beam Data
RUN I - 1.24x1020 POT (LE)(PRL Publication
disappearance)
Higher energy beam
RUN IIa1.22x1020 (LE) POT(new)
RUN IIb~0.75x1020 POT(not included)This Analysis: Run I + Run IIa => 2.46x1020 POT
(LE Beam only)
RUN III>0.9x1020 POT(current)
51
Predicting the FD spectrum
x =
- Beam Matrix
- F/N
- NDFit
- 2DFit
52
This method is known as the Beam Matrix method.
Far/Near Ratio Method• An approach that uses the ND data in a non-parameterized way is
provided by the F/N ratio method:
• For every event that passes FD NC selection, a reconstructed energy vs true energy 2D histogram is created– Oscillation weights are calculated for bins of true Energy
– For each bin of true energy, the reconstructed energy projection is multiplied by the corresponding oscillation weight
– Prediction is obtained by multiplying each bin by NiData/Ni
MC
• Simple, makes no assumptions about ND Data parameterization, robust to systematic errors
FDipredicted
FDiMC
NDiMCNDi
Data
53/5
• Normalization: 4% – POT counting, Near/Far reconstruction efficiency, fiducial mass
• Relative Hadronic Calibration: 3%– Inter-Detector calibration uncertainty
• Absolute Hadronic Calibration: 11%– Hadronic Shower Energy Scale(6%), Intranuclear rescattering(10%)
• Muon energy scale: 2%– Uncertainty in dE/dX in MC
• CC Contamination of NC-like sample: 15%• NC contamination of CC-like sample: 25%• Cross-section uncertainties:
– mA (qe) and mA (res): 15%– KNO scaling: 33%
• Poorly reconstructed events: 10% • Near Detector NC Selection: 8% in 0-1 GeV bin• Far Detector NC Selection: 4% if E < 1 GeV, <1.6% if E > 1 GeV • Beam uncertainty: 1 error band around beam fit results
Systematic Errors
54
• Systematic errors studied using simulated Far Detector data histograms with oscillation parameters m2 = 2.38 x10-3 eV2, sin2223=1
• Left plot displays magnitude of shift in FD simulated data compared to nominal• Ratio plots show shifted/nominal ratio for FD simulated data, overlaid with
shifted/nominal MC FD prediction – Displays ability of F/N extrapolation method to reproduce systematic shift
• relative, where shifts only applied to one detector
Systematic Errors
Simulated Data
55
Systematic Errors
56
MiniBooNE cross-section model
• Nuance is tuned to world ν data
• Same approach for Neugen/MINOS
• Monte-Carlo event rates are
adjusted using MiniBooNE data
outside signal region
Nuance ν generatorD. CasperNPS 112 (2002) 161
PRL 100, 032301 (2008).
arXiv:0803.3423 [hep-ex]
57