Zelimir Djurcic

1
Measurement of the off-axis NuMI beam with
MiniBooNE

• Zelimir Djurcic
• Columbia University

• Outline of this Presentation
• Off-axis Neutrino Beam
• NuMI flux at MiniBooNE
• MiniBooNE Detector and Reconstruction
• CC nm Sample
• NC p0 Sample
• CC ne Sample

Brookhaven National Lab Seminar, 1/17/08
2
Fermilab Neutrino Beams
3
MiniBooNE Motivated the LSND Experiment Result
LSND observed a (3.8?) excess of anti-?e events
in a pure?anti-?? beam 87.9 22.4 6.0 events
In SM there are only 3 neutrinos

? Models developed with 2 sterile ?s
4
Result of the MiniBooNE Oscillation Search
reconstructed
neutrino energy bin (MeV)
200-300 300-475
475-1250 data 375
369 380 total
background?????284 25 274 21
358 35 Excess 91 30 95
27 22 40
No Oscillation Signal Found! MiniBooNE Data Shows
Low Energy Excess!
5
NuMI Beam
MINOS Experiment L700 km E2-5GeV

• 120 GeV protons 3xE13/pulse.
• Primarily for the MINOS long baseline
  experiment.

However, the NuMI beam points in the direction of
MiniBooNE as well.
6
Off-axis Beam

• On-axis, neutrino energy more tightly
  related to hadron energy.
• Off-axis, neutrino spectrum is narrow-band
  and softened.
• Easier to estimate flux correctly all
  mesons decay to same energy ?.

? Detector
First Proposed by BNL-E889
Target
?
Decay Pipe
Horns
7
Future off-axis Neutrino Experiments
On-axis beam
Off-axis beam
Use off-axis trick for optimized ??-gt?e search.

• NOvA
• NuMI off-axis beam
• 810km baseline
• 14.5mrad E?2GeV

• T2K
• J-PARC 50GeV proton beam
• Use SK as Far detector 295km away
• 35 mrad E?0.6GeV

8
NuMI beam and MiniBooNE detector
NuMI events (for MINOS) detected in MiniBooNE
detector!
MiniBooNE
q
p, K
p beam
Decay Pipe
MiniBooNE detector is 745 meters downstream of
NuMI target. MiniBooNE detector is 110 mrad
off-axis from the target along NuMI decay
pipe.
9
Analysis Motivation
Observation and analysis of an off-axis beam.
Measurement of ?/K components of the NuMI
beam.The NuMI beam provides MiniBooNE with an
independent set of neutrino interactions.Enables
a comparison of the Booster Neutrino Beam (BNB)
with the NuMI neutrino beam (off axis)-Similar
energy spectrum.-Proton target is further away
(746 m vs. 550 m)-Very different background
composition.-Rich in ?e flux ?can study ?e
reactions in greater detail.
10
Joint collaboration between MiniBooNE and NuMI

• Beam Information and Neutrino Fluxes at MiniBooNE
  are provided by the MINOS collaboration (BNL, U
  Texas).
• Analysis of MiniBooNE data performed by the
  MiniBooNE collaboration.

11
NuMI off-axis beam at MiniBooNE detector

• Opportunity to demonstrate off-axis technique.
• Known spectral features from ?, K decays.
• Expected energy spectra is within MiniBooNE
  energy acceptance.

12
NuMI as a ?e Source

• NuMI off-axis beam produces strong fluxes
  in both ?µ and ?e flavors.
• The ?e s are helpful to study the MiniBooNE
  detector.
• Provide a new setting for oscillation
  studies.
• Rates
• NuMI off-axis(at MB) ?e 6
• NuMI on-axis ?e 1
• BNB on-axis ?e 0.5

stopped K
K
m
13
NuMI Neutrino Spectrum is Calibrated

• Extensive experience with MINOS data.
• MINOS acquired data sets in variety of NuMI
  configurations.
• Tuned kaon and pion production (xF,pT) to MINOS
  data.

MINOS nm
MINOS nm

• Same parent hadrons produce neutrinos seen by
  MiniBooNE
• Flux at MiniBooNE should be well-described by
  NuMI beam MC?

D.G. Michael et al, Phys. Rev. Lett. 97191801
(2006) D.G. Michael et al, arXiv0708.1495 (2007)
14
Two Views of the Hadron Decays

• Decays of hadrons produce neutrinos that
  strike both MINOS and MiniBooNE.
• Parent hadrons sculpted by the two
  detectors acceptances.
• Plotted are pT and p of hadrons which
  contribute neutrinos to MINOS (contours) or
  MiniBooNE (color scale).

MiniBooNE
MiniBooNE
MINOS
MINOS
15
Neutrino Origin Along NuMI Beam Line
ne

• Higher energy neutrinos mostly from
  particles created in target.
• Interactions in shielding and beam absorber
  contributes in lowest energy bins.
• Plots show where the parent was created.

nm
MiniBooNE
diagram not to scale!
16
Neutrino sources along NuMI beamline
Higher energy neutrinos mostly from particles
created in target. Interactions in shielding
and beam absorber contributes in lowest energy
bins.
17
Are the neutrinos coming from the target?
18
Flux Uncertainties

• Focusing uncertainties are negligible.
• Uncertainty is dominated by production of
  hadrons.
• off the target (estimated from MINOS tuning).
• in the shielding (estimated in gfluka/gcalor).
• in beam absorber (estimated in gfluka, 50 error
  assigned).

MINOS Tuning
stopped mesons excluded in this plot
19
MiniBooNE
(Booster Neutrino Experiment)
becomes
An off axis neutrino experiment using Main
Injector
20
NuMI Beam and MiniBooNE Detector
NuMI events (for MINOS) detected in MiniBooNE
detector!
MiniBooNE
q
p beam
p, K
Decay Pipe
MiniBooNE Detector 12m diameter sphere 950000
liters of oil(CH2) 1280 inner PMTs 240 veto PMTs
Main trigger is an accelerator signal indicating
a beam spill. Information is read out in 19.2 ?s
interval covering arrival of beam.
21
Detector Operation and Event reconstruction
No high level analysis needed to see neutrino
events
Events in DAQ windowno cuts
Removed cosmic ray muons PMT veto hits lt 6
Removed cosmic ray muons and ?-decay
electrons PMT veto hits lt 6 and PMT tank hits gt
200
6-batch structure of MI about 10 ?s
duration reproduced.
Backgrounds cosmic muons and decay electrons
-gtSimple cuts reduce non-beam backgrounds to 10-5
22
Detector Operation and Event reconstruction
The rate of neutrino candidates was constant
0.51 x 10-15 /P.O.T. Neutrino candidates counted
withPMT veto hits lt 6 and PMT tank hits gt 200
The data set analyzed here 1.42 x 1020 P.O.T.
We have a factor two more data to analyze!
23
Particle Identification
Cerenkov rings provide primary means of
identifying products of ? interactions in
the detector
m candidate
nm n ? m- p
electron candidate
ne n ? e- p
p0 candidate
nm p ? nm p p0
n n
p0 ? gg
24
Events from NuMI detected at MiniBooNE
Flux
Event rates
NuMI event composition at MB ??-81,
?e-5,???-13,??e-1
Neutrino interactions at carbon simulated by
NUANCE event generator neutrino flux converted
into event rates.
Event rates
CCQE 39 CC ? 26 NC ?0 9
25
Analysis Algorithm
26
Event Reconstruction
The tools used in the analysis here are developed
and verified in MiniBooNE oscillation analysis of
events from Booster beam.
Details
Phys. Rev. Lett. 98, 231801 (2007),
arXiv0704.1500 hep-ex
arXiv0706.0926 hep-ex Accepted for publ. by
Phys.Rev.Lett.
and
Event selection very similar to what was used in
MiniBooNE analyses.
To reconstruct an event -Separate hits in beam
window by time into sub-events of related
hits. -Reconstruction package maximizes
likelihood of observed charge and time
distribution of PMT hits to find track
position, direction and energy (from the charge
in the cone) for each sub-event.
27
Analysis Method
Uses detailed, direct reconstruction of particle
tracks, and ratio of fit likelihoods to identify
particles.
Apply likelihood fits to three hypotheses -single
electron track -single muon track -two
electron-like rings (?0 event hypothesis )
Compare observed light distribution to fit
prediction Does the track actually look like an
electron?
Form likelihood differences using minimized logL
quantities log(Le/L?) and log(Le/L?)
log(Le/L?)lt0 ?-like events
log(Le/L?)
log(Le/L?)gt0 e-like events Example from
MiniBooNE Oscillation Analysis.
e
?
28
?? CCQE Analysis
29
Analysis of the ?? CCQE events from NuMI beam
?? CCQE (?n ? ?p) has a two subevent
structure (with the second subevent from stopped
?? ???e e)
Tank Hits
Cerenkov 1
e
?
12C
nm
Cerenkov 2
n
Scintillation
p
Event Selection
Subevent 1 Thitsgt200, Vhitslt6 Rlt500 cm
Le/L? lt 0.02
Subevent 2 Thitslt200, Vetolt6
30
Visible E of ? final state interactions in ??
CCQE sample
Log(Le/L?)lt 0.02
Total MC nN?nXp0 Beam ne nmp? m-D nmn? m-p nmn?
m-np Other nm Events Events Events
Data
CCQE
Monte Carlo
other
CC?
CCQE
CC?
PRELIMINARY
Visible energy in tank GeV
Data (stat errors only) compared to MC
prediction for visible energy in the tank.
This sample contains 18000 events of which 70
are CCQEs.
31
Compare ?? CCQE MC to DataParent Components
Beam MC tuned with MINOS near detector
data. Cross-section Monte Carlo tuned with
MB measurement of CCQE pars MA and ?.
p
K
PRELIMINARY
arXiv0706.0926 hep-ex
Visible energy in tank GeV
MC is normalized to data POT number with no
further corrections!
32
Compare ?? CCQE MC to DataParent Components
p
K
PRELIMINARY
Visible energy in tank GeV
Predicted Kaons are matching the data out of box!
33
Systematic Uncertainties in ?? CCQE analysis
To evaluate Monte Carlo agreement with the data
need estimate of systematics from three sources
-Beam modeling flux uncertainties. -Cross-section
model neutrino cross-section uncertainties.
-Detector Modeldescribes how the light emits,
propagates, and absorbs in the detector (how
detected particle looks like?).
Detector Model
Cross-section
Visible energy GeV
Visible energy GeV
Total
PRELIMINARY
Beam
PRELIMINARY
Visible energy GeV
Visible energy GeV
34
Add Systematic uncertainty to ?? CCQE Monte Carlo
p
Predicted Pions are matching the data within
systematics!
K
? visible energy distribution
PRELIMINARY
Visible energy in tank GeV
K
p
Outgoing ?angular distribution
PRELIMINARY
Information about incoming ?wrt NuMI target
direction.
cos ??
35
?? CCQE sample Reconstructed energy E? of
incoming ?
Reconstructed E? QEfrom Elepton (visible
energy) and lepton angle wrt neutrino direction
p
K
PRELIMINARY
Understanding of the beam demonstrated MC is
normalized to data POT number !
36
Conclusion from ?? CCQE analysis section
This is the first demonstration of the off-axis
principle. There is very good agreement between
data and Monte Carlothe MC need not be tuned.
Because of the good data/MC agreement in ?? flux
and because the ??? and ?e ?share same parents
the beam MC can now be used to predict ?e rate,
and mis-id backgrounds for a ?e analysis.
37
?e CCQE Analysis
38
Backgrounds to ?e CCQE sample
?e CCQE (?n ? ep)
When we try to isolate a sample of ?e
candidates we find background contribution to
it -?0 (?0???) and radiative ? (??N?) events,
and -dirt events.
Therefore, before analyzing ?e CCQE we constrain
the backgrounds by measurement in our own data.
39
Among the e-like mis-ids, ?0 decays which are
boosted, producing 1 weak ring and 1 strong
ring is largest source.
Analysis of ?0 events from NuMI beam
g
p0
g
?
g
?
StrategyDont try to predict the ?0 mis-id
rate, measure it! Measured rates of reconstructed
?0 tie down the rate of mis-ids
g
?
?0
p
p
? decays to a single photon with 0.56
probability
What is applied to select ?0s Event
pre-selection 1 subevent Thitsgt200,
Vhitslt600 Rlt500 cm
log(Le/L?)gt0.05 (e-like) log(Le/L?)lt0 (?0-like)
40
Analysis of ?0 events from NuMI beam ?0 mass
The peak is 135 MeV/c2
Data
Monte Carlo
?0
?e
??
?e appear to be well modelled.
This sample contains 4900 events of which 81 are
?0 events world second largest ?0 sample!
41
Analysis of ?0 events from NuMI beam ?0 mass
The peak is 135 MeV/c2
Data
Monte Carlo
?0
?e
??
The ?0 events are well modeled with no
corrections to the Monte Carlo!
42
Analysis of ?0 events from NuMI beam ?0 momentum
Data
Monte Carlo
??
?0
?e
PRELIMINARY
We declare good MC/Data agreement for ?0 sample
going down to low mass region where ?e candidates
are showing up!
Further Cross Check!
43
Analysis of dirt events from NuMI beam

• - Dirt background is due to ? interactions
• outside detector. Final states
• (mostly neutral current interactions)
• enter the detector.
• - Measured in dirt-enhanced samples
• - we tune MC to the data selecting a sample
• dominated by these events.
• -Dirt events coming from outside deposit only a
  fraction
• of original energy closer to the inner tank
  walls.
• -Shape of visible energy and event vertex
  distance-to-wall distributions are
  well-described by MC good quantities to measure
  this background
• component.

44
Selecting the dirt events
log(Le/L?)gt0.05 (e-like) Ee lt550 MeV
Distance-to-wall lt250 cm m?lt70 MeV/c2 (not
?0-like)
Event pre-selection 1 subevent Thitsgt200,
Vhitslt600 Rlt500 cm
Fits to dirt enhanced sample
Uncertainty in the dirt rate is less than 20.
Dirt sample

• interactions
• in the tank

Events/bin
Events/bin
PRELIMINARY
We declare good MC/Data agreement for the dirt
sample.
Dist-to-wall of tank along track m
Visible energy GeV
45
Analysis of the ?e CCQE events from NuMI beam
?e CCQE (?n ? ep)
1 Subevent Thitsgt200, Vhitslt6 Rlt500 cm, Eegt200MeV
Likelihood cuts as the as shown below

Eegt200MeV cut is appropriate to remove ?e
contribution from the dump that is hard to model.
Likelihood e/? cut
Likelihood e/? cut
Mass(?0) cut
Signal region
Signal region
Cut region
Cut region
MC example plots here come from Booster beam MC
Cut region
Signal region
Visible energy MeV
Visible energy MeV
Visible energy MeV
Analysis of ?e events do we see data/MC
agreement?
46
Visible energy of ?e CCQE events
Data
Monte Carlo
?e
Other ??
?0
dirt
PRELIMINARY
?
Visible energy in tank GeV
Data 783 events. Monte Carlo prediction 662
events.
Before we further characterize data/MC agreement
we have to account for the systematic
uncertainties.
47
Systematic Uncertainties in ?e CCQE analysis
Detector Model
Cross-section
PRELIMINARY
Beam
Visible energy GeV
Visible energy GeV
PRELIMINARY
Total
Visible energy GeV
Visible energy GeV
dirt component of Xsec 20 error ?0 component
of Xsec 25 error
48
?e CCQE events e visible energy and angular
distribution
K
e visible energy distribution
KL
?
PRELIMINARY
p
Visible energy in tank GeV
All ??
All ??
Outgoing e angular distribution
PRELIMINARY
cos ?e
49
Outgoing electron angular distribution
?e CCQE sample Reconstructed energy E? of
incoming ?
All ?e
PRELIMINARY
All ??
50
Summary of estimated backgrounds vs data ??e CCQE
sample
Looking quantitative into low energy and high
energy region
E? QE MeV 200-900 900-3000
total background 40166
26150 ?e intrinsic
311 231 ?? induced
90 30
NC ?0
30 25
NC ??N? 14 1
Dirt 35
1 other
11 3
Data 49822
28517 Data-MC 97?70
24?53 Significance 1.40
? 0.45 ?
At this point systematic errors are large we
cannot saymuch about the difference between low
and high-E regions.In the future we will reduce
?e CCQE sample systematics constraining it with
our large statistics ?? CCQE sample.
51
Summary and Future Steps
52
We performed analyses of neutrinos from NuMI beam
observed with MiniBooNE detector. The sample
analyzed here corresponds to 1.42x 1020 protons
on NuMI target. We observed good description of
the data by Monte Carlo with both ?? CCQE and ?e
CCQE sample successful demonstration of an
off-axis beam at 110 mrad. ?? CCQE sample
demonstrated proper understanding of the Pion and
Kaon contribution to neutrino beam.
PRELIMINARY
53
In the future we will reduce ?e CCQE sample
systematics constraining it with our large
statistics ?? CCQE sample. The ?e CCQE
sample will be compared to what we observed with
Booster beam. We are currently reprocessing and
collecting more data (expect about 3 x 1020
P.O.T. collected by now.)
These errors will be reduced
PRELIMINARY