New OPERA results: the atmospheric muon charge ratio
description
Transcript of New OPERA results: the atmospheric muon charge ratio
New OPERA results:the atmospheric muon charge
ratioNicoletta Mauri
Università di Bologna and INFN
on behalf of the OPERA Collaboration
UZH Seminar on Particle and AstrophysicsUniversität Zürich, April 30th, 2014
Outline
The OPERA experiment detector and physics case neutrino oscillation results
New results on non-oscillation physics atmospheric muon charge ratio (submitted
to EPJC)• OPERA as a cosmic ray detector• Data Analysis and Combination• Inference on primary composition and forward
physics
Conclusions
N. Mauri, UZH Seminar30 April 2014 2
Introduction
3N. Mauri, UZH Seminar30 April 2014
• In the last decades several experiments provided evidence for neutrino oscillations: conversion in-flight of lepton flavor– Quantum-mechanical interference phenomenon– Solar, atmospheric, artificial beams: all
disappearance experiments (or indirect appearance)
• The OPERA experiment was mainly designed to unambiguously prove the oscillation phenomenon through direct nt appearance
– Definitely close of the discovery phase of neutrino oscillations
• The detector - although optimized for beam neutrino detection - can also be exploited for non-oscillation studies– Cosmic ray physics at the Gran Sasso Lab
The OPERA experiment
Full coverage of the parameter space for the atmospheric neutrino sector
Main physics goal: prove nm nt oscillations in appearance mode
• Long baseline neutrino oscillation experiment located in the CNGS (CERN Neutrinos to Gran Sasso) nm beam• Direct search for nm nt oscillations detecting the t lepton produced in n t
CC interactions (appearance mode)4N. Mauri, UZH Seminar30 April 2014
Appearance detection
5N. Mauri, UZH Seminar30 April 2014
Direct observation of nm nt oscillation
Large mass ~O(kton) signal selection and background rejection High granularity ~1mm resolution
dEEEmEPEMNN CCDA )()(),()( 2
Emulsion Cloud Chamber
30 April 2014 N. Mauri, UZH Seminar 6
Neutrino interaction detector (ECC)• Target basic unit: brick of 57 nuclear emulsions interleaved by lead plates + 2 interface emulsions (CS) high resolution and large mass in a modular way• unambiguous measurement of the kink
• “stand-alone” detector
30 April 2014 N. Mauri, UZH Seminar 7
20m
10m
10m
SM1SM2
Total target mass = 1.25 ktons
Target section:27 brick walls (75000 bricks)31 Target Tracker walls (TT)
Magnetic spectrometer:22 RPC planes 6 drift tube layers (PT stations)
Brick selection
m ID, charge, momentum
Calorimetry
Neutrino interaction trigger
n
B = 1.53 T
Brick: ECC target basic unit (57 nuclear emulsion films + 56 lead plates)
OPERA general structure
30 April 2014 N. Mauri, UZH Seminar 8
OPERA: an hybrid detector
N. Mauri, UZH Seminar 930 April 2014
n m n t search: signal candidates
t (r p0 p-) candidate
First n t candidate (Phys. Lett. B 691
(2010) 138)
t 3 h candidate
Second n t candidate (JHEP 11
(2013) 036)
t m candidate
Third n t candidate (Phys. Rev. D 89
(2014) 051102)
10
Neutrino oscillation results in progress
With a simple counting method and
conservative background
evaluation: 4.2s significance
30 April 2014 N. Mauri, UZH Seminar
Fourth n t candidate very recently presented by the OPERA Spokesperson at a Seminar at LNGS (25 March 2014)
t h candidate
30 April 2014 N. Mauri, UZH Seminar 11
OPERA as a cosmic ray detector
1400
m
CNGS beam events are identified through a timing coincidence with CERN possibility to collect cosmic events during the
physics run
Gran Sasso underground lab: 1400 m of rock (3800 m.w.e) shielding, cosmic ray flux reduced by a factor 106 w.r.t. surface, very reduced environmental radioactivity
OPERA vs previous and current underground experiments:a deep underground detector with charge and momentum reconstruction and excellent timing capabilities (~10 ns)
Atmospheric muon charge ratio
30 April 2014 N. Mauri, UZH Seminar 12
The atmospheric muon charge ratio
• The atmospheric muon charge ratio Rm ≡ Nm+/Nm-
is being studied and measured since many decades– Depends on the chemical composition and energy spectrum
of the primary cosmic rays– Depends on the hadronic interaction features– At high energy, depends on the prompt component
• It provides the possibility to check HE hadronic interaction models (E>1TeV) in the fragmentation region, where no data exists
• Since atmospheric muons are kinematically related to atmospheric neutrinos (same sources), Rm provides a benchmark for atmospheric n flux computations (e.g. background for neutrino telescopes)
13
m
m
(ordinary) meson decay: dNm/d cosq ~ 1/ cosq
p
K
hadronic interaction: multiparticle production s(A,E), dN/dx(A,E) extensive air shower
m
short-lifetimemeson production and prompt decay
(e.g. charmed mesons)
Isotropic angulardistribution
detection: Nm(A,E), dNm/dr
transverse size of bundle PT(A,E)
Primary C.R. proton/nucleus: A, E, isotropic
TeV muon propagation in the rock: radiative processes andfluctuations
The physics of cosmic ray TeV muons
30 April 2014 N. Mauri, UZH Seminar
30 April 2014 N. Mauri, UZH Seminar 14
Analytic predictions
p
p (primary) airnucleus
• Assume only primary protons with a spectrum dN/dE = N0E-(1+g)
• Assume only pions and neglect muon decays (HE limit)• Consider the inclusive cross-section for pions
The pion spectrum is then
Ed
dEEEf p
inelpp
pp
),(
),()(
)( )1(pp
E
EEfdEEE
constE
Naïve prediction
Assuming Feynman scaling
)(),(~
xfEd
dEEEf pE
p
inelpp
pp
Feynmanscaling
30 April 2014 N. Mauri, UZH Seminar 15
Analytic predictions
R (E )
(E ) (E )
(E )Z
p
Zp
Finally, the muon charge ratio prediction:
Interpretation of the prominent features:• The result is valid only in the fragmentation region, since in the central
region Feynman scaling is strongly violated• But the steeply falling primary spectrum ( ~ 1.7)g in the SWM suppresses
the contribution of the central region in the secondary production scaling holds (at least for E < 1 TeV);In other words: each pion is likely to have an energy close to the one of the projectile (primary CR proton) and comes from its fragmentation (valence quarks) positive charge
• R does not depend on E (or E ) nor on the target nature• R depends on the primary spectrum g
Zp f p
~
(x)x 1dx0
1Spectrum weighted moments
pZEconstE )1()()(
The pion spectum becomes:
30 April 2014 N. Mauri, UZH Seminar 16
Neutrons in the primary flux
• Elaborating the minimal model:– Introducing the neutron component in the primary flux
(in heavy nuclei) and considering the isospin symmetries:
one obtains:
p0, n0 = proton and neutrons in the primary flux
npnp
ZZZZ ,
82.0)np/()np( 00000
R 101 0
1.22
where
(Zp Z p ) /(Z
p Z p )
(1 Z pp Z pn ) /(Z pp Z pn )
30 April 2014 N. Mauri, UZH Seminar 17
Kaon contribution
• At higher energy (>100 GeV) the contribution of K becomes important
• In general, the contribution of eachcomponent to the muon flux Npar = (p, K, charmed, etc.) depends on the relative contributionof decays and interaction probabilities:
where
parN
i ii
Nii
NN
N
b
Za
Z 1 )(/11
)(
E
E
)1)(1(
)()1()(
1
i
iii r
iBrraa
)/log)(1)(1(
))(1)(2()(
2
1
Niii
Niiii r
rbb
2)/( ii mmr
ei = ei(q) is the “critical energy”, i.e. the energy above which interactions dominate over decays. Along the vertical (q = 0o) ei(0)= mich/ti (h = 6.5 km)
ep = 115 GeV eK = 850 GeVeX > 107 GeV
q = 0o
q = 60o
ep eK
30 April 2014 N. Mauri, UZH Seminar 18
Kaon contribution to Rm
• For kaons:
because the reaction
pp K+ L N + anything
is favoured.
Conservation of strangeness requires production of K+ and K0 in association with a Λ or S (associate production).
On the other hand, the production of K−,K0 requires the creation of an associated baryon and an additional strange meson.
This leads to a larger Rm ratio at high energy
pnpZZZ
30 April 2014 N. Mauri, UZH Seminar 19
Parameterization of the charge ratio
• Let us consider again the general form for the muon flux
where we have explicited the ei(q) dependence on q
where q* is the zenith angle at the production point• The correct variable to describe the evolution of Rm is
therefore Emcosq*
• The Rm evolution as a function of Emcosq* spans over the different sources
Rm = wpRm p + wKRm
K + wcharmRmcharm +…
par
Ni
N
i ii
i
NN
N
b
Za
Z 1* )0(/cos11
)(
E
E
*cos/)0()( ii Earth
q*q
POWERFUL HANDLE TODISCRIMINATE MODELS
OPERA: Emcosq* ≈ 2000 GeV The (magnetized) experiment withthe largest Emcosq*
30 April 2014 N. Mauri, UZH Seminar 20
Rm measurements with Emcosq* > 1 TeVExperiments with magnetic field:• Utah:
G. K. Ashley et al., Phys. Rev. D12 (1975) 20– Underground at Utah University, flat surface above ~1400 m.w.e., magnetic
spectrometer (1.63 T) + spark chambers, six bins with 46 < q < 78
• MINOS:P. Adamson et al., Phys. Rev. D76 (2007) 052003 + Phys. Rev. D 83 , 032011 (2011)– Underground at Soudan, magnetized steel, flat surface above ~2000 m.w.e.,
0 < q < 90• OPERA:
N. Agafonova et al., Eur. Phys. J. C67 (2010) 25 + arXiv:1403.0244 (submitted to EPJC)– Underground magnetic spectrometer (1.53 T) at LNGS, average overburden
~3800 m.w.e., drift tubes + RPC + scintillators, 0 < q < 90
Experiments without magnetic field:• Kamiokande-II
M. Yamada et al., Phys. Rev. D44 (1991) 617– Underground Cherenkov detector at Kamioka ~2700 m.w.e., delayed events on stopping muons, one
bin with 0 < < 90 • LVD:
N. Agafonova et al., Proc. 31th ICRC, ŁÓDZ 2009 + arXiv:1311.6995– Underground at LNGS, average overburden ~3800 m.w.e., scintillators, delayed events on stopping
muons, one bin with < 15
30 April 2014 N. Mauri, UZH Seminar 21
Cosmic event reconstruction in OPERA
• Cosmic-event tagging:– Outside the CNGS spill window
• Dedicated pattern recognition and track finding/fitting:– Cosmic events are passing-through– Cosmic events comes from all directions– Cosmic events may have multiple parallel
tracks (in OPERA 5% of the events are muon bundles)
– Different reconstruction philosophy
30 April 2014 N. Mauri, UZH Seminar 22
Event reconstruction: pattern recognition
Real double muon event
Dedicated pattern recognition for cosmic events ( outside CNGS spill window): takes into account event topology and multiple events (m bundles):
hybrid strategy: global method (Hough Transform) + local method (pivot points)
r
y
30 April 2014 N. Mauri, UZH Seminar 23
PT track reconstruction3D tracks are used to “guide”track finding and fitting in the PTsystem: Find the line tangent to the drift circles with the best c2
250 mm position resolution 0.15 (1) mrad angular resolution for doublets (singlets) for f = 0 (improve for f > 0)
Residuals ~250 mm
Track reconstructed in 1 PT station (in a pair) singletTrack reconstructed in 2 PT station (in a pair) doublet
30 April 2014 N. Mauri, UZH Seminar 24
Good overall angular resolution“resolutions” < 1 deg both for zenith and azimuth direction reconstruction
Event reconstruction performance
q f
Multiple muon events well reconstructed
High angular resolution in the PT system
30 April 2014 N. Mauri, UZH Seminar 25
Charge and momentum reconstruction
• In each side of the magnet arm we can reconstruct an independent angle fj, j=1,...,6.
• Each fj can be reconstructed with one station (singlets) or two stations (doublets)
• We compute Dfk = fi – fj , k=1,...,4, angle differences between adjacent station-pairs
2
2
11
] / ) / ( exp[ 1
) / (
xz
yzk
s
s
Be dx dE
dx dE lp
B ≡ Bd/l = effective magnetic field
Top view of the OPERA detector
• Charge is reconstructed according to the Df sign• If each track contributes to multiple Df angles, a weighted average is computed
[weights = experimental errors on ]Df
Df bending angle
30 April 2014 N. Mauri, UZH Seminar 26
Momentum reconstruction performance
Momentum resolution
30 April 2014 N. Mauri, UZH Seminar 27
Monte Carlo simulationsTwo Monte Carlo simulations for different purposes: 1) a fast tool for the cross-check with experimental data, for the validation of the analysis software and for unfolding
a parameterized generator (MC1) with a primary chemical composition based on the MACRO fit model [Phys. Rev. D56 (1997) 1418]
2) a reliable tool for surface muon energy estimation and a link between underground variables and primary cosmic ray parameters
a FLUKA-based full Monte Carlo simulation (MC2) with primary composition model from Astropart. Phys. 19 (2003) 193
Muon flux
Aknee
A2
Aknee
A1
EE,EKdEdN
EE,EKdEdN
A2
A1
EEcut
g~2.7÷3Ecut~3000 TeV
30 April 2014 N. Mauri, UZH Seminar 28
Charge misidentification
• h ≡ fraction of tracks reconstructed with wrong charge sign
• Used to correct data (unfold)
• In the LE limit (only MCS) the estimate is ~10-3
• Real h is larger due to spurious effects:– Secondary particles– Timing errors
)1(
)1(
meas
measunf
R
RR
2
2222
)]1([
)()1()()21(
meas
measmeas
unf
R
RRR
30 April 2014 N. Mauri, UZH Seminar 29
Data analysis• Data taken during the CNGS Physics Runs
in Standard Magnet Polarity (SP): 2008 (from Jun 18th until Nov 3rd)2009 (from Jun 1st until Nov 23rd) 2010 (from Apr 29th until Nov 22nd)2011 (from Mar 18th until Nov 17th)
in Inverted Magnet Polarity (IP): 2012 (from Mar 22nd until Dec 3rd)
• Data Pre-Selection: on the basis of data quality and global variable distributions (based on TT, RPC, PT digits)
Live time Standard Polarity (SP): 625.0 days Live time Inverted Polarity (IP): 234.8 days
Total number of events: ~2.22 M (20082011) + ~823 k (2012)
30 April 2014 N. Mauri, UZH Seminar 30
Data reduction
A set of progressive cuts applied in order to isolate a clean data sample:a) at least one reconstructed Df angle
(acceptance cut)b) Remove events with large number of PT hits
(clean PT cut)c) Remove events with bendings smaller than
the experimental resolution (deflection cut)c’) Remove events with very large bendings
(effective for pm<5 GeV/c)
30 April 2014 N. Mauri, UZH Seminar 31
Quality cuts IClean PT Cut: number of digits allowed from geometrical considerations:
Ndata(re-scaled) = Ndata – NMC(geometrical)(f)
Geometrical MC dependence on the fangle
cut at 3s
Remove events with a large number of PT tubes fired (d-rays, secondary interactions, electronic noise etc) can induce wrong matching
M = M(f)N’ = N - M(f)
Re-scaled distribution of the experimental number of PT digits
30 April 2014 N. Mauri, UZH Seminar 32
Quality cuts II
misidentification probability
Deflection Cut: Cut on bending angle information below the PT resolution
before Deflection cut
after Deflection cut
• /Df sDf > 3
R saturates above 3s; as expected, for lower number of s, R 1 (charge misidentification ~50%, total randomization of charge reconstruction)
We also require Df < 100 mrad, to reject isolated secondary particles produced at large angles by high energy muons
MC distributions are split in the 2 regions m+
true and m-
true
Exp. data
MC data
m+ m+m- m-
03.0total
wrong
N
N
w/o deflection cut
w/ deflection cut
30 April 2014 N. Mauri, UZH Seminar 33
Alignment of the PT systemThe comparison between the two data sets with opposite magnet polarities allowed checking the systematics related to the alignment of each magnet arm
Comparing the Df distributions for the two magnet polarities:
(compatible with the systematic uncertainty quoted in the previous paper, EPJC 67(2010) 25)
|dfsyst|≈ 0.1 mrad
IP SP
4th magnet arm
30 April 2014 N. Mauri, UZH Seminar 34
Combination of data setswith opposite magnet polarities
Using the two data sets with opposite magnet polarities allows disposing of misalignment systematics.
Assuming the symmetric behaviour of the bin-to-bin migration matrix in the opposite polarities, the proper combination of the two data sets is given by the ratio of the sums (m+
SP + m+
IP) and (m-SP + m-
IP), where the numbers are normalized by their polarity live time
Rmeas
(iSP / lSP
i
iIP / lIP )
(iSP / lSP
i
iIP / lIP )
where lSP and lIP
are the respective polarity live time
Finally unfolding the result with the charge misidentification:
)1(
)1(
meas
measunf
R
RR
30 April 2014 N. Mauri, UZH Seminar 35
New systematic uncertainty on Rm
Misalignment: combination procedure• in principle all the systematic contributions due to
misalignment cancel with the SP and IP data combination
• Estimate of the residual systematic uncertainty related to the combination procedure: difference between the charge ratio Rm for muons coming from opposite directions: dRm = |Rm (up)- Rm (down)|
Two main sources of systematic uncertainties: misalignment and charge
misidentification
m
(up)m
(down)
Reversing the muon orientation, the misalignment bias flips the sign
For single m: dRmunf(syst) = ±0.001
For multiple m: dRmunf(syst) = ±0.013
30 April 2014 N. Mauri, UZH Seminar 36
Systematic uncertainty on Rm
Charge misidentification h from experimental data• estimate dh = hdata – hMC for a subsample of events
crossing both arms of a spectrometer: computation of the probability p of reconstructing opposite charges [sign(Df1) = -sign(Df2)]
• the probability that Df1 and Df2 have opposite sign is related to hthrough a function h = h(p), computed via MC
dh = 0.007 dRm = 0.007 (one-sided: hreal ≥ hMC)Total systematic uncertainty for single m: dRm
unf(syst) = +0.007, -0.001
Total systematic uncertainty for multiple m: dRmunf(syst) = +0.015, -0.013
Quadratic sum of the two contributions (misalignment and misidentification):
30 April 2014 N. Mauri, UZH Seminar 37
Results: underground muon charge ratio
OK,nm=3
Rm computed separately for single and multiple muon events– Multiple muons: compute R m when the 3D
multiplicity is > 1, independently on the number of measured charges in the event
Nm ‹A› ‹E/A›primary
[TeV]H fraction Np/Nn Rm
unf
= 1 3.35 ± 0.09
19.4 ± 0.1 0.667 ± 0.007
4.99 ± 0.05
1.377 ± 0.006
> 1 8.5 ± 0.3 77 ± 1 0.352 ± 0.012
2.09 ± 0.07
1.098 ± 0.023
“dilution” of Rm when proton-Air and neutron-Air interactions change their relative contributions
primary features extracted from MC2
New OPERA data (full statistics)
30 April 2014 N. Mauri, UZH Seminar 38
Charge ratio of multiple muon events
• The smaller value of the charge ratio of multiple muons is due to the convolution of two effects:larger n/p ratio in the all-nucleon spectrum different xF region
Feynman x: xF ≅ Esecondary/Eprimary
n/p ratio in primary cosmic rays
Multiple muon sample: higher E/nucleon, higher average A
Multiple muon sample: smaller xF, towards the central region
30 April 2014 N. Mauri, UZH Seminar 39
Rm as a function of pm
• Rm (single muons) was computed in bins of the underground momentum pm and unfolded
• Evolution with p m is compatible both with a constant and with a logarithmic energy increase, with a 2.4 spreference for the latter
Rm = a0 + a1 log10 p m
a0 = 1.322 ± 0.023a1 = 0.030 ± 0.012(c2/dof = 14.99/16)
Rm = c0
c0 = 1.377 ± 0.006(c2/dof = 20.86/17)
Dc2/dof = 5.87/1 (~2.4 sigma)
30 April 2014 N. Mauri, UZH Seminar 40
The resolution on E m is dominated by fluctuations in the stochastic term b of the energy loss
Best performance is achieved using the “crude” MC2 Build a table EmMPV = f(h,pm): take the MPV of the Landau distribution better resolution and residuals well centered
h (bin 2)
pm (bin 3)
Emsurface
MPV
/)/( heEE
a= ( )a E = ( )b b E
EEEdh
dE)()(
Rm as a function of E m cos q*
30 April 2014 N. Mauri, UZH Seminar 41
Fixing RKp = 0.127 (weighted average of experimental values, Ref. Grashorn et al.):fp+ = 0.5512 ± 0.0014fK+ = 0.705 ± 0.014
only single muons
Rm as a function of E m cos q*
Compute and unfold the charge ratio in each bin
Fit with the function
30 April 2014 N. Mauri, UZH Seminar 42
Rm as a function of E m cos q* and d0
Taking into account an explicit dependence on d0 = (p - n)/(p + n): (Gaisser, Astropart. Phys. 35 (2012) 801)
d0 depends on Eprimary/nucleon ≈ 10 Em (not on E m cos q*!)
Different dependencies: fit in 2-dimensions (Em, cos q*)
20 bins: 5 energy bins × 4 angular bins
Fixed parameters (see table) Inferred parameters: ZpK+ and d0
30 April 2014 N. Mauri, UZH Seminar 43
Rm as a function of Em
d0 (EN ≈ 20 TeV/n)= 0.61 ± 0.02ZpK+ = 0.0086 ± 0.0004
Fit result:
Projecting the fit result on the average OPERA zenith <cos q*> ≅ 0.7:
Rm as a function of the surface muon energyonly single muons
30 April 2014 N. Mauri, UZH Seminar 44
Conclusions• The OPERA detector, designed to prove n n oscillations in
appearance mode, was exploited for the measurement of the atmospheric muon charge ratio R
– New data with the full OPERA statistics (2008-2012) were presented: the atmospheric muon charge ratio was measured in the highest energy region
– The combination of the two data sets with opposite magnet polarities allowed minimizing systematic uncertainties
– Rm was measured separately for single and for multiple muon events We found a strong reduction of the charge ratio for multiple muon events The integral value and the energy dependence of the charge ratio for single
muons are compatible with the expectation from a simple p-K model No significant contribution of the prompt component up to E m cos q*
~ 10 TeV
We extracted relevant parameters on the primary composition (d0) and the associated kaon production in the forward fragmentation region (ZpK+ moment)
The Rm behaviour as a function of E m supports the validity of Feynman scaling in the fragmentation region up to E m ~ 20 TeV, corresponding to primary energy/nucleon E N ~ 200 TeV