SLIDE 1
Searching for Sterile Neutrinos with MINOS
Ashley Timmons
- n behalf of the MINOS/MINOS+ collaboration
51th Rencontres de Moriond EW2016
1
2
Searching for Sterile Neutrinos with MINOS 51th Rencontres de - - PowerPoint PPT Presentation
Searching for Sterile Neutrinos with MINOS 51th Rencontres de - - PowerPoint PPT Presentation
Searching for Sterile Neutrinos with MINOS 51th Rencontres de Moriond EW2016 Ashley Timmons on behalf of the MINOS/MINOS+ collaboration 1 Intro - Neutrino Oscillations Neutrino oscillations arise from mixture of mass and flavour eigenstates X
SLIDE 2
SLIDE 3
Date
2005/05/02 2006/08/03 2007/11/05 2009/02/05 2010/05/10 2011/08/11 2012/11/12 2014/02/13 2015/05/18
)
18
10 × Protons per week (
2 4 6 8 10 12 14
)
20
10 × Total Protons (
2 4 6 8 10 12 14 16 18 20 22
Low Energy antineutrinos Low Energy neutrinos Higher Energies NOvA neutrinos
3
MINOS/MINOS+ Data Set MINOS MINOS+
MINOS+ is the continuation of MINOS with the NuMI beam in the medium energy configuration.
SLIDE 4
4
True Neutrino Energy (GeV)
5 10 15 20 25 30
(Arbitary Units)
CC
σ Flux x
0.02 0.04 0.06
CC
µ
ν CC
µ
ν
Near Detector Simulated Low Energy Beam Forward Horn Current
True Neutrino Energy (GeV)
5 10 15 20 25 30
(Arbitary Units)
CC
σ Flux x
0.02 0.04 0.06
CC
µ
ν CC
µ
ν
Near Detector Simulated Low Energy Beam Reverse Horn Current
The NuMI Beam
This analysis uses this beam mode
SLIDE 5
5
Far Near
- 23.7 ton fiducial mass
- 1.04 km downstream from target
- 4.2 kiloton fiducial mass
- Veto shield for cosmic suppression
- 705m underground
- Both detectors are magnetised tracking/sampling calorimeters, segmented into planes
composed of 2.54 cm-thick steel planes and 1 cm-thick scintillator strips
- Detectors designed to be functionality equivalent, cancels systematics uncertainties in
flux modelling and cross section to first order.
Detectors
SLIDE 6
6
Long-Baseline Sterile Search
MINOS was built for measurement of Δm232 by looking for νμ disappearance optimised for L/E = 500km/GeV Looking for perturbations from three-flavour disappearance (sensitive to θ24, θ23) This is at FD baseline
- f 735km
Neutral current interaction rate is the same for the three active flavours P(νμ -> νs ) depends on θ24, θ34, θ23,
SLIDE 7
7
Event Topologies CC NC
SLIDE 8
Reconstructed Neutrino Energy (GeV)
10 20 30 40
Events
100 200 300 400
running
µ
ν POT
20
10.56x10
MINOS Preliminary
) = 0.41
23
θ (
2
sin
2
eV
- 3
= 2.37 x10
32 2
m ∆ ) = 0.022
13
θ (
2
sin
2
eV
- 5
= 7.54 x10
21 2
m ∆
MINOS CC-like Far Detector data Three-flavour simulation Systematic uncertainty NC background
Reconstructed Visible Energy (GeV)
10 20 30 40
Events
50 100 150 200
running
µ
ν POT
20
10.56x10
MINOS Preliminary
) = 0.41
23
θ (
2
sin
2
eV
- 3
= 2.37 x10
32 2
m ∆ ) = 0.022
13
θ (
2
sin
2
eV
- 5
= 7.54 x10
21 2
m ∆
MINOS NC-like Far Detector data Three-flavour simulation Systematic uncertainty CC background
µ
ν background
e
ν Beam appearance
e
ν appearance
τ
ν
8
FD Spectra - Data
CC NC
FD spectra three-flavour oscillated with MINOS 2012 CC-analysis fit values This is NOT a fit 2563 CC-like events 1211 NC-like events R [0-3 GeV] = 1.10 +/- 0.06 +/- 0.07 R [0-40 GeV] = 1.05 +/- 0.04 +/- 0.10 If no NC disappearance R = 1
SLIDE 9
9
Oscillations
This is assuming a three-flavour model - no sterile neutrinos
SLIDE 10
10
Oscillations
Sterile oscillations in the FD only.
SLIDE 11
11
Oscillations
Fast oscillations in the FD, counting experiment
SLIDE 12
12
Oscillations
- scillations now occur at the ND. Typical extrapolation using ND data is no
longer possible!
SLIDE 13
13
Far Over Near Ratio
0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8
MINOS Preliminary
) = 0.41
23
θ (
2
sin
2
eV
- 3
= 2.37 x10
32 2
m ∆ ) = 0.022
13
θ (
2
sin
2
eV
- 5
= 7.54 x10
21 2
m ∆
1 2 3 4 5 1 2 3 4 5 5 10 15 20 5 10 15 20 20 30 20 30 30 40 30 40
CC selection
0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5 MINOS data Three-flavour simulation Systematic uncertainty
NC selection
1 2 3 4 5 1 2 3 4 5 5 10 15 20 5 10 15 20 20 30 20 30 30 40 30 40
)
- 3
Far / Near Ratio (x10 Reconstructed Energy (GeV) New analysis technique to probe many magnitudes of Δm241 Direct fit to F/N ratio for CC and NC events Assume 3+1 sterile model Set δ13, δ14, δ24 and θ14 to zero Parameters fit are: Δm232, Δm241, θ24, θ23, and θ34 Moved from likelihood method towards χ2 fit, containing covariance matrix with systematics
we assume no νe -> νs
SLIDE 14
14
Reconstructed Energy (GeV)
10 20 30 40
F/N Fractional Error
- 0.2
- 0.1
0.1 0.2 NC total systematic
MINOS Preliminary
Total Uncertainties
Reconstructed Energy (GeV)
10 20 30 40
F/N Fractional Error
- 0.2
- 0.1
0.1 0.2 CC total systematic
MINOS Preliminary
Including 26 systematics into the fit via covariance matrix, accounting for: Normalisation, Detector acceptance, NC selection, Hadron production, Beam focusing, Cross sections, Energy scale and background
SLIDE 15
15
Total Uncertainties
)
24
θ (
2
sin
0.02 0.04 0.06 0.08 0.1 0.12
)
2
(eV
41 2
m ∆
- 3
10
- 2
10
- 1
10 1 10
2
10
Statistics + Normalization + Acceptance + NC selection Total Systematics
running
µ
ν MINOS simulation, POT
20
10.56x10
MINOS Preliminary
Efgect on the sensitivity when including largest systematics added incrementally.
SLIDE 16
16
Disappearance Limit
)
24
θ (
2
sin
- 4
10
- 3
10
- 2
10
- 1
10 1
)
2
(eV
41 2
m ∆
- 4
10
- 3
10
- 2
10
- 1
10 1 10
2
10
running
µ
ν POT
20
10.56x10 MINOS data
MINOS Preliminary
MINOS data 90% C.L. MINOS data 95% C.L. Super-K 90% C.L. CDHS 90% C.L. CCFR 90% C.L. SciBooNE + MiniBooNE 90% C.L.
Feldman-Cousins procedure used for confidence limit
SLIDE 17
17
Combination with Bugey
2
|
4 µ
|U
2
|
e4
= 4|U
e µ
θ 2
2
sin
7 −
10
6 −
10
5 −
10
4 −
10
3 −
10
2 −
10
1 −
10 1
)
2
(eV
2
m ∆
4 −
10
3 −
10
2 −
10
1 −
10 1 10
2
10
mode
µ
ν LSND 90% CL ICARUS 90% CL OPERA 90% CL NOMAD 90% CL MiniBooNE 90% CL MINOS/Bugey* 90% CL
fluxes, courtesy of P. Huber * GLoBES 2012 fit with new reactor
POT
20
10 × MINOS data: 10.56
MINOS Preliminary MINOS disappearance search sensitive mainly to θ24 Bugey reactor experiment - electron anti-neutrino disappearance, θ14 Accelerator and reactor - largely uncorrelated systematic uncertainties
SLIDE 18
18
The Future
SLIDE 19
19
Summary
MINOS uses CC+NC sample to look for sterile neutrinos via deviations from the three-flavour model using a 3+1 model on the F/N Ratio Combination with reactor experiment Bugey allows for limit on same parameter space as LSND etc Use of covariance matrices for systematics, show power of two detector experiment with cancellation of large systematics Future with additional data from MINOS+ with higher stats at higher energies.
SLIDE 20
20
BACK UP
SLIDE 21
21
Sterile Neutrinos?
LSND MiniBooNE E ~ 20-200 MeV L ~ 30m E ~ 0.2 - 3 GeV L ~ 540m L/E ~ 1 km/GeV
Blue histogram includes
- scillations
for Δm2 ~ 1eV2
Looking for νe appearance in a νμ beam 3.8σ 3.4σ 2.8σ
anti-neutrinos
SLIDE 22
22
Sterile Neutrinos
An experiment with L/E ~ 1 km/GeV could only observe oscillations if Δm2 ~ 1eV2
U = Ue1 Ue2 Ue3 Ue4 Uµ1 Uµ2 Uµ3 Uµ4 Uτ1 Uτ2 Uτ3 Uτ4 Us1 Us2 Us3 Us4
experimentally we consider the model 3(active)+1(sterile) LEP measurements of the Z width show three active neutrinos. Therefore any additional ones must be sterile L/E ~ 1 km/GeV |Δm241| >> |Δm232|, |Δm221| amplitude
- f oscillations
∆41 = ∆m2
41L
4Eν
energy dependence
- f oscillations
Introduces: θ14, θ24, θ34, and Δm241
SLIDE 23
23
Sterile Neutrinos
2
|
4 µ
|U
2
|
e4
= 4|U
e µ
θ 2
2
sin
7 −
10
6 −
10
5 −
10
4 −
10
3 −
10
2 −
10
1 −
10 1
)
2
(eV
2
m ∆
2 −
10
1 −
10 1 10
2
10
mode
µ
ν LSND 90% CL LSND 99% CL KARMEN2 90% CL ICARUS 90% CL MiniBooNE 90% CL MiniBooNE 99% CL
SLIDE 24
24
The NuMI Beam
True Neutrino Energy (GeV)
20 40 60 80 100 120
# Events (A.U)
1 10
2
10
3
10
4
10
5
10
6
10
7
10
MINOS ND Simulation
Forward Horn Current µ
ν
µ
ν
e
ν
e
ν
True Neutrino Energy (GeV)
20 40 60 80 100 120
# Events (A.U)
10
2
10
3
10
4
10
5
10
6
10
7
10
MINOS ND Simulation
Forward Horn Current
+
π
- π
+
K
- K
+
µ
- µ
- Neutrino parents
- Mainly pions, significant koan
component at higher energies
- Anti-neutrino background, high
energies, hard to defocus
- Intrinsic electron neutrino
component.
SLIDE 25
25
Experimental Setup
Far Near
SLIDE 26
26
Long-Baseline Sterile Search
MINOS was built for measurement of Δm232 by looking for νμ disappearance optimised for L/E = 500km/GeV
P(νµ → νµ) ≈ 1 − sin2 2θ23 cos 2θ24 sin2 ∆31 − sin2 2θ24 sin2 ∆41
|Δm241| >> |Δm231| to first order the survival prob: MINOS sensitive to θ24 through muon disappearance Looking for perturbations from three-flavour disappearance. This is at FD baseline
- f 735km
SLIDE 27
27
Long-Baseline Sterile Search
Neutral current interaction rate is the same for the three active flavours νμ -> νs will reduce the NC rate
PNC = 1 - P(νμ -> νs )
P(νμ -> νs ) depends on θ24, θ34, θ23, The problem with NC events for sterile searches is the inability to estimate the true neutrino energy.
SLIDE 28
28
Poorly Reconstructed Events
the badly reconstructed events are unique to the near detector due to events overlapping in space and time “chunks” of activity split ofg from larger events which look the same as small hadronic showers - contamination in NC sample
Ereco Etrue < 0.3
Definition:
true visible energy expected for the neutrino interaction
SLIDE 29
29
Poorly Reconstructed Events
Max Consecutive Planes
10 20 30 40 50 60
Events
4
10
50 100 150
Near Detector Data Monte Carlo Prediction Poorly Reconstructed Background
Accepted
MINOS Preliminary
Slice Pulse-Height Fraction
0.2 0.4 0.6 0.8 1
Events
4
10
1 10
2
10
Near Detector Data Monte Carlo Prediction Poorly Reconstructed Background
Accepted MINOS Preliminary
Remaining data/MC disagreement is taken as a systematic uncertainty. Accept events with max planes > 3 Accept events with slice pulse height fraction > 0.5 When a shower develops longitudinally, it deposits energy in successive planes fraction of hits associated with event compared to all nearby hits (in space and time)
SLIDE 30
Track Extension (number of planes)
- 10
- 5
5 10 15 20
Events
50 100 150
Accepted
Far Detector Data Monte Carlo Prediction CC Background
MINOS Preliminary
30
NC Event Selection
Planes Crossed by Event
20 40 60 80 100
Events
100 200 300
Accepted
Far Detector Data Monte Carlo Prediction CC Background
MINOS Preliminary
Same cuts at the ND and FD track extension defined as the number of planes the track extends out of the reconstructed shower. relative size of track compared to shower NC events will have short tracks NC events don’t typically cross as many planes in the detector, compared to CC events that have long tracks.
SLIDE 31
31
CC Event Selection
CC/NC separation parameter 0.2 0.4 0.6 0.8 1 POT
16
Events / 10
- 2
10
- 1
10 1 10
Low Energy Beam Data MC expectation NC background
Four variable kNN algorithm used for CC selection
- # of scintillator planes in a track
- Ratio: energy deposited in track
and deposited energy of entire event
- Mean pulse height of all track
hits.
- Ratio of low pulse height to high
pulse height hits.
CC-like NC-like
accept
Next select all events that have a track
SLIDE 32
32
Selector Performance
Energy (GeV)
10 20 30 40
Purity , Efficiency
0.2 0.4 0.6 0.8 1
CC Selection Near Detector MC Efficiency Far Detector MC Efficiency Near Detector MC Purity Far Detector MC Purity
MINOS Preliminary
Energy (GeV)
10 20 30 40
Purity , Efficiency
0.2 0.4 0.6 0.8 1
NC Selection Near Detector MC Efficiency Far Detector MC Efficiency Near Detector MC Purity Far Detector MC Purity
MINOS Preliminary
Purity= (# selected true signal events) / (total # selected events) Efg = (# selected true signal events) / (total # selected signal events before selection) MINOS was optimised for identifying νμ CC events NC events it is more diffjcult: 86% efg, 61% Purity at the FD main bkg from inelastic νμ CC events CC ND efg is low due to events with tracks ending near the coil hole
SLIDE 33
33
ND Spectra
Reconstructed Neutrino Energy (GeV)
5 10 15 20
Events
4
10
20 40 60 80 100 120
MINOS CC-like Near Detector data Three-flavour simulation Systematic uncertainty NC background
- mode
µ
ν Low energy beam, POT
20
10 × 7.99
MINOS Preliminary
Reconstructed Visible Energy (GeV)
5 10 15 20
Events
4
10
20 40 60 80
MINOS NC-like Near Detector data Three-flavour simulation Systematic uncertainty CC background
µ
ν background
e
ν Beam
- mode
µ
ν Low energy beam, POT
20
10 × 7.99
MINOS Preliminary
The measured Near Detector energy spectrum is used to predict the Far Detector spectrum via the Far/Near Ratio method. This method relies on no parameterisation the ND data
CC NC
For each bin of energy correct FD MC by scale factor from ND data/MC
- discrepancies. Robust method - reduces systematic errors
SLIDE 34
34
FD Spectra
Reconstructed Neutrino Energy (GeV)
10 20 30 40
Events
100 200 300 400
running
µ
ν POT
20
10.56x10
MINOS Preliminary
) = 0.41
23
θ (
2
sin
2
eV
- 3
= 2.37 x10
32 2
m ∆ ) = 0.022
13
θ (
2
sin
2
eV
- 5
= 7.54 x10
21 2
m ∆
MINOS CC-like Far Detector data Three-flavour simulation Systematic uncertainty NC background
Reconstructed Visible Energy (GeV)
10 20 30 40
Events
50 100 150 200
running
µ
ν POT
20
10.56x10
MINOS Preliminary
) = 0.41
23
θ (
2
sin
2
eV
- 3
= 2.37 x10
32 2
m ∆ ) = 0.022
13
θ (
2
sin
2
eV
- 5
= 7.54 x10
21 2
m ∆
MINOS NC-like Far Detector data Three-flavour simulation Systematic uncertainty CC background
µ
ν background
e
ν Beam appearance
e
ν appearance
τ
ν
CC NC
FD spectra three-flavour oscillated with MINOS 2012 CC-analysis fit values 2563 CC-like events 1211 NC-like events Already we see no significant deviations from the three-flavour model
SLIDE 35
35
NC R Values
Use NC FD spectrum to look for a deficit of NC events, compared to that expect from 3-flavour. Not assuming a particular sterile neutrino model R [0-3 GeV] = 1.10 +/- 0.06 +/- 0.07 R [0-40 GeV] = 1.05 +/- 0.04 +/- 0.10 If no NC disappearance R = 1 Values seem consistent with seeing no deficit in NC event rate [stats] [sys]
SLIDE 36
36
Varying Baseline
Neutrino Distance Travelled to ND (km)
0.2 0.4 0.6 0.8 1
)
3
Neutrino Events (x10
2 4 6
+
π from
µ
ν
- π
from
µ
ν
+
from K
µ
ν
- from K
µ
ν
MINOS Preliminary
Because we now allow for short baseline oscillations, it is crucial that we account for the baseline varying due to the distribution
- f hadron decay points within the 675m decay pipe.
end of decay pipe target
SLIDE 37
37
Transverse Momentum [GeV]
0.2 0.4 0.6 0.8 ]
2
Invariant Differential Cross-section [mb/GeV 200 400 600
- NA49 Data
♦
FLUKA MC Param.
♦
FLUKA MC Spread σ
- Alt. Param. 1
♦
FLUKA MC Spread σ
- Alt. Param. 2
♦
FLUKA MC
= 0.05
F
X
+
π
MINOS Preliminary
Eur.Phys.J. C49 (2007)
- FLUKA 2008, CERN-2005-10 (2005)
♦
Hadron Production Uncertainty
Need to assign systematic to flux, can not use ND data Fit a beam simulation (FLUKA) of the NA49 target to the BMPT parametrization.
wide acceptance spectrometer for the study of hadron production
Vary fit parameters within their errors to create a collection of physically feasible alternate invariant difgerential cross-section parametrizations. Scale up the errors given by the fit until the collection of alternate parametrizations cover the difgerence between the FLUKA MC and NA49 data.
SLIDE 38
38
Hadron Production Uncertainty
Neutrino Energy [GeV]
10 20 30 40 Fractional Error on the MINOS Far/Near Ratio
- 0.4
- 0.2
0.2 0.4
- mode
µ
ν Low Energy Beam, MINOS Preliminary Charged Current Sample
Neutrino Energy [GeV]
10 20 30 40 Fractional Error at the MINOS Far Detector
- 0.4
- 0.2
0.2 0.4
- mode
µ
ν Low Energy Beam, MINOS Preliminary Charged Current Sample
Neutrino Energy [GeV]
10 20 30 40 Fractional Error at the MINOS Near Detector
- 0.4
- 0.2
0.2 0.4
- mode
µ
ν Low Energy Beam, MINOS Preliminary Charged Current Sample
ND CC FD CC F/N CC
SLIDE 39
Detector Acceptance Uncertainty
39
Acceptance uncertainties are determined by comparing the efgect of varying cuts on data/MC at the ND compared to the nominal cuts. Looked at:
- Varying fiducial volume
- Varying the containment criteria
- Excluding tracks ending near the join between the
- calorimeter and spectrometer.
- Varying how close tracks can come to the coil hole.
Together these have the largest efgect on the sensitivity
SLIDE 40
40
Degeneracies
)
24
θ (
2
sin
- 4
10
- 3
10
- 2
10
- 1
10 1
)
2
(eV
41 2
m ∆
- 4
10
- 3
10
- 2
10
- 1
10 1 10
2
10
running
µ
ν POT
20
10.56x10 MINOS data
MINOS Preliminary
MINOS data 90% C.L. MINOS data 95% C.L. Super-K 90% C.L. CDHS 90% C.L. CCFR 90% C.L. SciBooNE + MiniBooNE 90% C.L.
Non-negligible interference terms from atmospheric
- scillations cause degenerates (regions where signal = three flavour)
Δm241 = 2* Δm231
θ24 = π/4
Δm241 ~ Δm231 Δm241 << Δm231
when probing large Δm241 to avoid large values of Δm231 a constraint is implemented in the fit
sin2(θ24) = 0.5
SLIDE 41
41
Sensitivity
)
24
θ (
2
sin
- 4
10
- 3
10
- 2
10
- 1
10 1
)
2
(eV
41 2
m ∆
- 4
10
- 3
10
- 2
10
- 1
10 1 10
2
10
running
µ
ν POT
20
10.56x10
MINOS Preliminary
MINOS data 90% C.L. MINOS sensitivity 90% C.L.
SLIDE 42
42
CC and NC component
Power comes from the CC sample
)
24
θ (
2
sin
- 4
10
- 3
10
- 2
10
- 1
10 1
)
2
(eV
41 2
m ∆
- 4
10
- 3
10
- 2
10
- 1
10 1 10
2
10
running
µ
ν POT
20
10.56x10
MINOS Preliminary
No Feldman-Cousins correction MINOS data
CC and NC 90% C.L. CC 90% C.L. NC 90% C.L.
Data
SLIDE 43
)
34
θ (
2
sin
- 3
10
- 2
10
- 1
10 1
2
χ ∆
2 4 6
running
µ
ν POT MINOS data
20
10.56x10
MINOS Preliminary
2
= 0.5 eV
41 2
m ∆
90% C.L. 95% C.L.
Feldman-Cousins corrected limits
43
1D limits
1D C.L limits for when Δm241 = 0.5eV2
)
24
θ (
2
sin
- 3
10
- 2
10
- 1
10
2
χ ∆
2 4 6
running
µ
ν POT MINOS data
20
10.56x10
MINOS Preliminary
2
= 0.5 eV
41 2
m ∆
90% C.L. 95% C.L.
Feldman-Cousins corrected limits
SLIDE 44
44
CC and NC component
Power comes from the CC sample
Sensitivity
)
24
θ (
2
sin
- 4
10
- 3
10
- 2
10
- 1
10 1
)
2
(eV
41 2
m ∆
- 4
10
- 3
10
- 2
10
- 1
10 1 10
2
10
running
µ
ν POT
20
10.56x10
MINOS Preliminary
No Feldman-Cousins correction MINOS sensitivity
CC and NC 90% C.L. CC 90% C.L. NC 90% C.L.
SLIDE 45
45