From DeepCore to PINGU: Measuring atmospheric neutrino oscillations at the South Pole

0.29 MB·

by  J. P. Yañez

EPJWebofConferenceswillbesetbythepublisher
DOI:willbesetbythepublisher
(cid:13)c Ownedbytheauthors,publishedbyEDPSciences,2016
6
1
0
From DeepCore to PINGU
2

n
Measuring atmospheric neutrino oscillations at the South Pole
a
J
0 J.P.Yáñez1,a fortheIceCube-Gen2Collaboration
2
1DESY,D-15735Zeuthen,Germany

]
x
e Abstract.Verylargevolumeneutrinotelescopes(VLVNTs)observeatmosphericneutri-
- nos overa wide energyrange (GeV toTeV), after theytravel distances aslarge as the
p
Earth’sdiameter. DeepCore, thelowenergyextension ofIceCube, hasstartedmaking
e
h meaningful measurements of the neutrino oscillation parameters θ23 and |∆m232| by an-
[ alyzing the atmospheric flux at energies above 10GeV. PINGU, a proposed project to

lower DeepCore’s energy threshold, aims to use the same flux to further increase the
1
precisionwithwhichtheseparametersareknown,andeventuallydeterminethesignof
v
∆m2 . ThelatestresultsfromDeepCore,andtheplannedtransitiontoPINGU,aredis-
5 32
4 cussedhere.
2
5
0
.
1 1 Introduction
0
6
Evidencethatneutrinoschangeflavor(oscillate)astheytravel,andarethusmassive,hasaccumulated
1
: in the last decades (see [1] and references therein). Although the phenomenon has been confirmed
v
from a variety of neutrino sources, atmospheric (anti-)neutrinos, ν and ν , remain a very powerful
i µ e
X tooltostudyit.
r Theamplitudeofoscillationprobabilitiesdependsonthemixingofflavorandmasseigenstates,
a encodedin mixingangles θ . The oscillationphase isproportionalto ∆m2L/E, where L isthe dis-
ij ij
tance traveled by the neutrino, E is its energy and ∆m2 is the squared mass difference of the mass
ij
eigenstates. AtmosphericneutrinosareproducedwithenergiesfromMeVtotheTeV-scaleandcan
traveldistancesaslargeastheEarth’sdiameterbeforedetection. Thismeansthatexperimentsmea-
suringtheseneutrinosaresensitivetothemass-splitting∆m2 andthemixinganglesθ andθ over
32 23 13
awideL/Erange[2].
Current VLVNTs can measure atmospheric neutrinos starting at approximately 10GeV [3, 4],
and future proposed telescopes aim to lower the threshold to a few GeV [5, 6]. Neutrinos crossing
theEarthattheseenergiesundergoneutrino-electroncoherentforwardscattering,whichaffectshow
theyoscillate[7]. BetweenE ≈ 1−12GeV,thisleadstolargedeviationsfromvacuumoscillations
ν
forneutrinosorantineutrinos,dependingontheorderingoftheneutrinomasses[8–10]. Abovethis
energy range saturation occurs and, while oscillation probabilities still differ from those in vacuum,
theyarethesameforneutrinosandantineutrinosregardlessoftheneutrinomassordering.
ae-mail:[email protected]
EPJWebofConferences
IceCube’sDeepCore[4]detectsneutrinosinthesaturationregime. Theexperimentisthussuited
formeasuringtheatmosphericparametersθ and|∆m2 |[11,12]. ThePrecisionIceCubeNextGen-
23 32
erationUpgrade(PINGU)[5],aproposedsuccessortoDeepCore,willlowertheenergythresholdto
accesstheresonantregime,identifythesignof∆m2 and,withthat,willplacethefinalpieceofthe
32
neutrinomassorderingpuzzle.
2 Detector performance of DeepCore and PINGU
IceCubeisaniceCherenkovneutrinodetectorthatconsistsof5160digitalopticalmodules(DOMs)
deployedoveravolumeof1km3,atdepthsbetween1450and2500matthegeographicSouthPole
[13]. The DOMs are arranged in 86 strings, with each string holding 60 of them. Most strings are
situatedoveraquasi-hexagonalgrid,withaninter-stringseparationof125m,andaverticalspacing
betweenDOMsof17m. Towardsthemiddleofthedetector,eightstringsarelocatedinbetweenthe
regular grid, reducing the inter-string spacing to 40−70m. The DOMs on these strings have 35%
higherquantumefficiencythanthestandardDOMsandareonly7mapart. Inthisvolume,knownas
DeepCore,itispossibletodetectneutrinoswithenergiesaslowas10GeV[4].
PINGUwillconsistof40additionalstringsdeployedinsidetheDeepCorevolumewithaninter-
stringspacingofapproximately20m[5,14]. Eachstringwillhost96highquantumefficiencyDOMs
3m apart. The additional optical modules result in a factor of 10 increase in photocathode area
density with respect to DeepCore. The closer spacing between the DOMs in PINGU lowers the
energythresholdfordetectiontoafewGeV,andimprovesthereconstructionoftheenergyandarrival
zenithangleoftheneutrinosinteractinginitsvolume.
2.1 Signalandbackground
Thereareeightdistinctneutrino-nucleoninteractionsthatDeepCorecan, andPINGUwill, observe:
chargedcurrent(CC)ν , ν andν , andneutralcurrent(NC)interactionsofallflavors, bothofneu-
e µ τ
trinos and antineutrinos. Only two of them, CC ν and ν¯ , are expected to produce muons with a
µ µ
range similar to the DOM spacing1. All of these interactions below E ≤ 100GeV constitute the
ν
signal for oscillation measurements in DeepCore and PINGU. Muon neutrino disappearance is the
strongesteffectexpected, andsincemostmuonneutrinososcillateintotauneutrinos, ν appearance
τ
is also testable. Transitions involving electron neutrinos are relevant in the resonance region. The
identificationandreconstructionoftheseeventsareexplainedinthefollowingsections.
Atmosphericmuons, producedinthesamecosmicrayshowersastheneutrinos, areadominant
sourceofbackgroundevents. InDeepCoredataanalyses,atmosphericmuonsarerejectedbyavari-
ety of veto algorithms that try to correlate the signal observed in DeepCore with the arrival time of
photonsobservedoutsidetheDeepCorevolume. Thesamealgorithmsarealsousedtoextractapure
atmosphericmuonsamplefromdata. Thesedataareusedasabackgroundtemplateinthefinalfitof
thesimulation. A similarstrategyisbeing implementedforPINGU. Aftertheveto informationhas
been exhausted, the remaining atmospheric muon background is rejected by requiring the events to
haveaminimumof5directphotons(seebelow)andbyapplyingcutsonthereconstructionquality.
BackgroundeventsarealsointroducedbyPMTthermalnoiseandradioactivedecaysintheglass
oftheDOM,whichbecomerelevantwhenlookingforfaintevents. Eventsproducedbythisnoiseare
alsoremovedbyqualitycutsonthereconstruction. Amoreimportanteffectofthisbackgroundisthat
itcanbiastheenergyreconstructionandaffecttheefficiencyofvetoalgorithms. Inordertomodelit
correctly,noisesimulationistunedtothedata[15].
1While18%ofCCντinteractionswillalsoproducemuonsfromtaudecay,theexpectednumberofeventsofthistypeat
energiesrelevantforatmosphericneutrinooscillationsisnegligible.
7thVeryLargeVolumeNeutrinoTelescopeWorkshop
2.2 Reconstructionofneutrinoevents
All neutrino events are reconstructed using the general hypothesis that at the point of interaction a
hadronic shower is produced together with a collinear muon. A muon hypothesis is fit to the event
withtracklengthsrangingfromzero(i.e. ultimatelyacascade-likeevent)tothedetector’ssize. Only
ahandfulofthephotonsemittedinlowenergyneutrinointeractionsareobservedbyDeepCore,and
thesephotonsmayscattermultipletimesbeforebeingdetected. ThelatestDeepCoreresults[12]re-
liedonfindingunscatteredphotonstoaidtheeventselectionandeventreconstruction.Unscattered,or
direct,photonsareidentifiedbyrestrictingthepatternsformedbytheirarrivaltimetoastring. While
requiring5ormoredirectphotonsresultsina30%signalefficiency,thiscriterionkeepsthesubsetof
thedetectedneutrinoswhosedirectioncanbeeasilyreconstructed. Themuontrackdirectionisfitus-
ingthesephotons,following[16]. Sincethelightproducedbyacascadeisalsopreferentiallyemitted
intheCherenkovangleeventswithoutamuonarealsofit,althoughtheresolutionobtainedissignifi-
cantlyworse. Theneutrinointeractionposition,theenergyofthehadroniccascadeattheinteraction
point,andtherangeofthemuonproducedarefitinasecondstepusingallphotonsrecordedbythe
detectorwhilethedirectionalongwhichtheseparametersarevariediskeptfixed. At E = 15GeV,
ν
the resulting median zenith angle resolution is 10◦ for ν and 25◦ for ν , while the median energy
µ e
resolutionis25%forboth.
Anewmethodforreconstructingneutrinointeractionshasbeensuccessfulydevelopedinthelast
fewyears. Itusesamaximumlikelihoodestimatorwheretheexpectedtimeofarrivalofphotonsto
aDOMgivenaparticlehypothesisisfinelybinned. TheeffectsoftheopticalpropertiesoftheSouth
Pole ice are taken into account. In contrast to the method described above, this approach has the
advantagesofbeingabletoreconstructnearlyallneutrinoevents(nodirectphotonsarerequired)and
ofproducingasignificantlybetterdirectionalreconstructionforeventswithoutamuonpresent.Other
resolutionsobtainedarecomparable. Themaximumlikelihoodreconstructionisappliedtostudythe
performance of PINGU, where the energy and zenith angle resolutions are expected to improve by
approximately30%.
2.3 Particleidentification
MuonswithlongtracksareusedtoidentifyCCν interactions. InthelatestresultsfromDeepCore,
µ
neutrino-inducedmuonsareidentifiedbycomparingthequalityofthemuontrackdirectionalrecon-
structiontoacascadefitthatassumesthelightobservedhasbeenrandomizedbyscatteringandtravels
sphericallyoutwardsfromitsemissionpoint. TheclassificationstabilizesatE ≥30GeV,andabove
ν
this energy it correctly identifies 60% of CC ν , while misidentifying 30% of cascades. A similar
µ
strategyisbeingdevelopedforthemaximumlikelihoodreconstruction.
TheparticleidentificationinPINGUismorecomplex,asν andν areexpectedtocontributeto
e µ
thesensitivitytodiscriminatetheneutrinomassordering[5]. Amultivariatemethodhasbeenputin
place to classify the events according to their topology taking advantage of the reduced inter-string
spacing. The classifier shows a stable behavior at E ≥ 15GeV, correctly identifying about 80% of
ν
CCν andmisidentifyinglessthan20%ofcascade-likeevents.
µ
3 Analysis methods and systematic uncertainties
CurrentandfutureanalysisofDeepCoredataareaimedatmeasuringtheatmosphericneutrinoflux
and extracting the oscillation parameters θ and |∆m2 |. The simulation is fit to the data under the
23 32
assumption of normal and inverse mass ordering, and both results are reported. The CP-violating
phase δ is held to zero. Only marginal sensitivity to the neutrino mass ordering is expected from
EPJWebofConferences
Table1.ListofthesourcesoferrorincludedinneutrinooscillationanalysesinthelatestDeepCoreresult(DC),
ongoingstudiesofDeepCoredata(DC+)andPINGU(P).DISstandsfordeepinelasticscattering.BYrefersto
thesixparametersusedintheBodek-YangmodelimplementedinGENIE.
Sourceoferror Nominalvalue Uncertainty DC DC+ P
Totalcrosssectionscaling Free x x x
Linearenergydependence E±0.03 x x x
Neutrino DISlowQ2tuning BY±[25,40]% x x
GENIEmodel[18]
interactions NCscaling ±10% x x
Quasi-elasticaxialmass −15% +25% x x x
Resonanceaxialmass ±20% x x x
Overallscaling Free x x x
Spectralindex E±0.04 x x x
Atmospheric Honda2015[19]
Fluxratioν/ν¯ ±20% x x
νandµflux
ν /ν relativescaling ±3% x x x
e µ
Atm.µcontamination Fromdata Free x x
DOMoverallefficiency Muonsandflashers ±10% x x
Detection DOMangularacceptance Flashers,laser Upto50% x x
process Bulkicemodel Flashers Modelsin[20,21] x x
Hadronicenergyscaling Geant4[22] ±5% x x
DeepCore data. A main goal of PINGU analyses, on the other hand, is to discriminate between the
twopossiblemassorderings. Inthismeasurement,theatmsophericoscillationparameterswouldalso
bemeasuredwhilethesolarparametersandθ wouldbetakenfromglobalanalysesofallavailable
13
oscillationdata[17].
AlikelihoodratiomethodisusedbothforDeepCoreandPINGUtofindthebestfitbetweendata
andsimulation,andtoestimatetheerrorofthemeasurement. InDeepCorethemethodreliesonthe
productionofseveralhighstatisticssimulationsetsthatspanthespaceofthesystematicuncertainties.
Sources of these systematic uncertainties are implemented as nuisance parameters and the data is
left to select the set of parameters in the simulation that describe it best. The uncertainties on the
oscillationparametersareestimatedfromscanningthelikelihoodlandscapearoundthebestfitpoint
and using a χ2 approximation. The same method is used for studies which aim at improving the
precisionofcurrentDeepCoreresults.
PINGU uses two methods for estimating its sensitivity to the neutrino mass ordering. The like-
lihood ratio approach is implemented as outlined in [23]. In this method, pseudo-experiments are
drawnfromtwosimulationtemplates,eachproducedunderadifferentmassordering. Theoscillation
parameters requried for the two templates are selected so that they result in the case of “maximum
confusion”betweenhierarchies. Eachpseudo-experimentisfitundertheassumptionofbothhierar-
chiesandthelikelihoodratioiscalculated. Theseparationbetweentheresultingdistributionsserves
as an estimate of the confidence with which the two hypotheses can be discriminated. The second
methodusedinPINGUisasimplified∆χ2 approach. ThisisafastparametricanalysisoftheFisher
informationmatrix[24]andreliesonthepartialderivativesoftheeventcountsineachbinwithrespect
totheparametersunderstudy. Theresultsobtainedfromthesimplified∆χ2 andthelikelihoodratio
approacheshavebeenfoundtobeinagreement,andbothareusedtoreporttheprojectedsensitivity
ofPINGU.
Inallmethodsdiscussed,sourcesofuncertaintyareimplementedasadditionalparametersinthe
fit to the data. Table 1 contains a list of the uncertainties considered for the latest DeepCore result,
those being studied for future DeepCore measurements, and those included in the PINGU studies.
7thVeryLargeVolumeNeutrinoTelescopeWorkshop
Anadditionalfine-grainedstudyoftheimpactoftheuncertaintiesoftheneutrinofluxpredicionsfor
PINGU analyses, not listed in Table 1, can be found in [25]. We note that, directly impacting the
measurementoftheoscillationparameters, simulationtestsindicatethatthedetectorenergyscaling
is strongly correlated with |∆m2 |. The value of sin2θ shows a dependence on the assumed angu-
32 23
lar acceptance of the DOM and also on the parameters that modify the angular flux of atmospheric
neutrinos. Thesensitivitytotheneutrinomassorderingdependsstronglyonthetruevalueofsin2θ .
23
4 Results and projections for DeepCore and PINGU
The latest published results of DeepCore are a muon disappearance analysis of three years of data
consideringonlytrack-likeeventswithcosθ < 0and E = [6,56]GeV.Some5174eventsare
reco reco
foundinthatrange. Thebestfitpointisfoundatsin2θ =0.53+0.09and∆m2 =2.72+0.19×10−3eV2.
23 −0.12 32 −0.20
Figure 1 shows the best fit point and the 90% confidence regions as a function of the atmospheric
oscillationparameters.
The sensitivity of an eventual re-analysis of the three years data set has been investigated in a
simulationstudy. Thisstudyincludesalessrestrictiveeventselection(includingcascade-likeevents),
events from the complete zenith range, and the maximum likelihood event reconstruction described
inSec.2.2. Thesenstivitycalculationusingthesemodificationsindicatesthepossibilitytoreducethe
uncertaintyon|∆m2 |tohalfofthecurrentlyreportedvalue,andthattheuncertaintyonsin2θ may
32 23
bedecreasedby20%.
IceCube 2014 T2K 2014 [NH]
3.8 MINOS w/atm [NH] SK IV [NH]
3.6
90% CL contours
)3.4
2V Figure1.Comparisonofbestfitandconfidenceintervals
e3.2
3− ofatmosphericoscillationparametersobtainedwith
2m(1032|23..80 DeepCore[12]withthoseofSuperKamoikande[26],T2K
∆ [27]andMINOS[28].
|2.6
2.4
2.2
2.0
0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70
sin2(θ23)
ThepredictedsensitivityofPINGUtotheneutrinomassorderingasafunctionoftimeisshownin
Fig.2(left),wheretheaccumulatedimpactofthesystematicuncertaintiesconsideredisalsoshown.
ThesestudiesindicatethatPINGUwillbeabletoidentifytheneutrinomassorderingwithasignifi-
canceof3σafter3−3.5yearsofoperation. Thesensitivitydependenceonthetruevalueofsin2θ at
23
the3−yearsbenchmarkisshowninFig.2(right).Afavorablecombinationoftruephysicsparameters
resultsinafasteridentificationofthemassordering.
5 Outlook
Atmospheric neutrinos are a valuable tool to study neutrino oscillations. IceCube’s DeepCore has
demonstratedthatverylargevolumeneutrinotelescopescanuseatmosphericneutrinostomakemean-
ingful contributions to the field. Current results have started to approach the precision of dedicated
experiments. Studies indicate that the full potential has not been reached and the addition of the
PINGUarraycangrealtyimproveonthesemeasurementsbyaccumulatinglargesamplesofverywell
reconstructedGeV-scaleneutrinosandultimatelyidentifytheneutrinomassordering.
EPJWebofConferences
5.0 3 year significance vs.θ23
IMH
4.5 NMH
PRELIMINARY
4.0
r)
3y3.5
(
σ
3.0
2.5
2.0
0.40 0.45 0.50 0.55 0.60
sin2θ
23
Figure2. PINGUprojections. Left:Expectedsignificancefordiscriminatingthemassorderingasafunctionof
time.Right:Significanceasafunctionofthetruevalueofsin2θ atthethreeyearbenchmark.
23
References
[1] K.Oliveetal.(ParticleDataGroup),Chin.Phys.C38,090001(2014)
[2] J.P.Yañez,A.Kouchner,Adv.HighEnergyPhys.2015,271968(2015),1509.08404
[3] S.Adrian-Martinezetal.(ANTARESCollaboration),Phys.Lett.B714,224(2012),1206.0645
[4] R.Abbasietal.(IceCubeCollaboration),Astropart.Phys.35,615(2012),1109.6096
[5] M.Aartsenetal.(IceCube-PINGUCollaboration)(2014),1401.2046
[6] J.Brunner(KM3NeTCollaboration),inICRC(2015)
[7] L.Wolfenstein,Phys.Rev.D17,2369(1978)
[8] S.P.Mikheev,A.Yu.Smirnov,Sov.J.Nucl.Phys.42,913(1985),[Yad.Fiz.42,1441(1985)]
[9] V.Ermilovaetal.,ShortNoticesoftheLebedevInstitute5,26(1986)
[10] E.K.Akhmedov,Sov.J.Nucl.Phys47,301(1988)
[11] M.Aartsenetal.(IceCubeCollaboration),Phys.Rev.Lett.111,081801(2013),1305.3909
[12] M.Aartsenetal.(IceCubeCollaboration),Phys.Rev.D91,072004(2015),1410.7227
[13] A.Achterbergetal.(IceCubeCollaboration),Astropart.Phys.26,155(2006)
[14] K.Hanson(IceCube-Gen2Collaboration),inVLVnTWorkshop,Rome(2015)
[15] M.Larson,Master’sthesis,UniversityofAlabama(2013)
[16] J.A.Aguilaretal.(ANTARESCollaboration),Astropart.Phys.34,652(2011),1105.4116
[17] M.Gonzalez-Garcia,M.Maltoni,T.Schwetz,JHEP1411,052(2014),1409.5439
[18] C.Andreopoulosetal.,Nucl.Instrum.MethodsA614,87(2010),0905.2517
[19] M.Hondaetal.,Phys.Rev.D92,023004(2015),1502.03916
[20] M.G.Aartsenetal.(IceCubeCollaboration),Nucl.Instrum.MethodsA711,73(2013)
[21] D.Chirkin(IceCubeCollaboration),inICRC(2013),1309.7010
[22] S.Agostinellietal.,Nucl.Instrum.MethodsA506,250(2003)
[23] D.Francoetal.,JHEP1304,008(2013),1301.4332
[24] R.A.Fisher,Phil.Trans.Roy.Soc.Lond.A222,309(1922)
[25] J.Sandroos(IceCube-Gen2Collaboration),inVLVnTWorkshop,Rome(2015)
[26] M. Nakahata (Super-Kamiokande Collaboration), in XVI International Workshop on Neutrino
Telescopes,Venice(2015)
[27] K.Abeetal.(T2KCollaboration),Phys.Rev.D91,072010(2015),1502.01550
[28] P.Adamsonetal.(MINOSCollaboration),Phys.Rev.Lett.110,251801(2013),1304.6335