results and prospects for sno low energy threshold analysis (leta) motivations analysis details...

Results and Prospects for SNO

• Low Energy Threshold Analysis (LETA)• Motivations• Analysis Details• Results

• Status of `three-phase’ Analysis• Summary and Other Recent Results

Josh Klein, for the SNO CollaborationUniversity of Pennsylvania

15 June 2010

Sudbury Neutrino Observatoryneutrino reactions on deuterons

Neutrino-Electron Scattering (ES)

Neutral Current (NC)

Charged Current (CC)

Signal rates determined by statistical fit

National Geographic

Phase I: Just D2O • Simple detector configuration, clean measurement• Low neutron sensitivity• Poor discrimination between neutrons and electrons

Phase II: D2O + NaCl• Very good neutron sensitivity• Better neutron electron separation

Phase III: D2O + 3He Proportional Counters• Good neutron sensitivity• Great neutron/electron separation

Three Phases of SNO

Low Energy Threshold Analysis

En=6 MeV En=6 MeV

Motivations: ne Statistics

CC ES

Night

Day

Low Energy Threshold Analysis Motivations: NC Precision

Phase I (D2O) NC

+74%

+68%

Phase II (D2O+NaCl)

NC

nx (NC) Statistics

“Beam On”

“Beam Off”

Breaking NC/CC Covariance

I

Low n capture eff.

High n capture eff.

II

Low Energy Threshold Analysis Overview

1. Joint-Phase (I+II) fit for all signals and remaining bkds2. Reduction of Backgrounds3. Reduction of Systematic Uncertainties4. `Float’ Dominant Uncertainties in Fit

Key components:

Results: 8B flux measured by NC rates Bin-by-bin electron energy spectrum using CC & ES Parameterized Psurv(En) (New) Two-flavor and three-flavor extraction of mixing params.

Needed to rework SNO’s entire analysis chain and simulation, from measurement of charge pedestals to final fit methods.

Low Energy Threshold AnalysisSignal Extraction Fit (Signal PDFs)

Not used

1-D projections of 3-D and 4-D PDFS

Teff (MeV) cosqsu

n

(R/RAV)3 Isotropy =

Monte Carlo

(unconstrained in fit)

Low Energy Threshold AnalysisCosmic rays < 3/hour Low Energy Backgrounds

D2OAcrylic VesselH2O

}×{+

PMT 208Tl

++

Acrylic Vessel Surface Neutrons [(α,n) reactions]

214Bi (U, Rn)208Tl (Th)

24Na (neutron activation of salt)

= 12 external bkds + 5 internal bkds

Teff>3.5 MeV

All events ( but only ~5000 ns)

For each phase

(most backgrounds constrained by ex-situ radioassays)

3 neutrino signals+ 17 backgrounds

Kinetic Energy Spectrum

Low Energy Threshold Analysis

New Threshold = 3.5 MeV

MCPMT -b gs

internal (D2O)

external (AV + H2O)

NC+CC+ES (Phase II)

Old

thre

shold

Low Energy Backgrounds

Low Energy Threshold AnalysisSignal Extraction Fit (3 out of 17(x2) Background PDFs)

Teff (MeV) cosqsu

n

(R/RAV)3 Isotropy =

1-D projections of 3-D and 4-D PDFS

Monte Carlo

Low Energy Threshold Analysis Background Reduction: Energy Resolution

`Prompt’ (direct) light easy to model: we know the path traveled

Using all hits increased hit statistics by ~12%->6% reduction in resolution~60% reduction internal bkds

Time Residual (ns)

Prompt Timing Cut

Late Timing Cut

Rayleigh Scatter

t t0 t pmt d

c

(used in prior analyses)

Low Energy Threshold Analysis

Only information is PMT charges, times, and hit patterns

• 4 KS tests of PMT pattern against single Cherenkov e-

• 1 KS test of PMT times against Cherenkov e-

• 3 cuts on various isotropy parameters• 2 cuts on energy reconstruction uncertainty• In-time ratio vs. Nhit to remove misreconstructed events

Background Reduction: New Cuts

Low Energy Threshold Analysis

Fiducial Volume

βγβ

High charge early in time

• `Early’ Charge to cut PMT b-gs

Note: This would have been impossible if we hadn’t fixed `little’ things like charge pedestals

Background Reduction: New Cuts

Low Energy Threshold Analysis

PassFail

FailPass

FailFail

PassPass

NPF = e1(1-e2)Nb

NFP = (1-e1)e2Nb

NFF = (1-e1)(1-e2)Nb

NPP = e1e2Nb + Ns

NPMT= NPP – Ns = NFP * NPF /

NFF

Special Case: PMT b-g PDFs

Not enough CPUs to simulate sample of events Use data instead

In-time ratio In-time ratio

Earl

y c

harg

e p

rob

abili

ty

Earl

y c

harg

e p

rob

abili

ty`Bifurcated’ analysis

Low Energy Threshold AnalysisSystematic Uncertainties: Brief Summary

b14

(isotropy)

0% 1% 3% 4%2%

n capture

Teff scale

Fiducial volume

I

II

LETA I

LETA II

N/A

I=D2OII=D2O+Salt

Low Energy Threshold AnalysisSystematic Uncertainties

Comparison of 208Tl calibration source data to MC

Run near the AV(to model AV 208Tl events)

Tests of PDF shapes

Low Energy Threshold Analysis

Tests of PDF shapes Distributed Rn Spike

Low Energy Threshold Analysis

Fit to spike energy spectrum allowing Teff scale to float: shift is 0±0.6%

1. Maximum likelihood with binned pdfs: Manual scan of likelihood space

• Data helps constrain systematics• `human intensive’

2. Kernel estimation---ML with unbinned pdfs:

Low Energy Threshold AnalysisSignal Extraction Fit

(3 signals+17 backgrounds)x2, and pdfs are multidimensional:ES, CC

NC, backgrounds

Two distinct methods:

• Allows full `floating’ of systematics, incl. resolutions

• CPU intensive---use graphics card!

Low Energy Threshold AnalysisFit Results: Binned fit, 1D Projections

LETA A LETA B

Low Energy Threshold Analysis8B Flux Results with `unconstrained’ CC spectrum

Low Energy Threshold Analysis `Unconstrained’ CC Electron Spectrum

Flat:2 = 21.52/15 d.o.f.

Low Energy Threshold Analysis `Unconstrained’ CC Electron Spectrum

PeeDAY(E) = c0 + c1 (E - 10 MeV)

+ c2 (E - 10 MeV)2

PeeASYM(E) = a0 + a1 (E - 10 MeV)

PeeNIGHT(E) = Pee

DAY(E) x [1 + (1/2)*PeeASYM(E)]

[1 – (1/2)*PeeASYM(E)]

Parameterize distortion to ne spectrum with quadraticPsurv is independent of any flux model:

CC and ES rates constrained to be less than NC

Note: Fit is now in En, not Teff

Low Energy Threshold Analysis Direct fit to data for Psurv(En)

This helps separate signals and backgrounds: PDFs are now 4D

Direct Fit for Energy-Dependent Survival

Probability

No distortion, no D/N:2 = 1.94 / 4 d.o.f.LMA-prediction:2 = 3.90 / 4 d.o.f.

Previous global best-fit LMA point: tan212 = 0.468, m2 = 7.59x10-5 eV2

DAY

NIGHT ASYM

8B = 5.046 +3.8 -

3.9 %

Borexino

SNO Day

Night

Comparisons of 8B SpectraJ.L. Raaf, Boston University

arXiv:0808.2868v2

Oscillation Analyses: SNO Only

Best-fit point:

tan212=0.437±0.058

m2=1.15x10-7 +0.438-0.18

eV2

LETA paper 2009:LETA joint-phase fit+ Phase III (3He)

(LOW)

SNO Collaboration, Phys. Rev C81, 55504

Solar + KamLAND 2-flavor Overlay

KamLAND Collab, Phys.Rev.Lett.90:021802,2003.

Brief History

Solar + KamLAND 2-flavor Overlay

KamLAND collaboration

Brief History

Solar + KamLAND 2-flavor OverlayBrief History

S. Abe et al. (KamLAND Collaboration), PRL 100, 221803 (2008)

Solar + KamLAND 2-flavor OverlayBrief History

LETA paper 2009:LETA joint-phase fit+ Phase III+ all solar expts+ KamLAND

LETA paper 2009:LETA joint-phase fit+ Phase III+ all solar expts+ KamLAND

2-flavor overlay

2 model

Solar + KamLAND 2-flavor Overlay

Oscillation Analyses: Solar + KamLANDLETA paper 2009:LETA joint-phase fit+ Phase III+ all solar expts+ KamLAND

Best-fit LMA point:

tan212 = 0.457 +0.040-0.029

(q12=34.06+1.16-0.84 deg)

sin2q12-1/3=-0.02+0.016-

0.018

m2 = 7.59x10-5 eV2 (+0.20 -0.21)8B uncert = +2.38

-2.95 %

2 model

LETA paper 2009:LETA joint-phase fit+ Phase III+ all solar expts+ KamLAND

3-flavor fit/overlay->Pointed out by many authors

3 model

Solar + KamLAND 3-flavor Overlay

Best-fit:

sin213=2.00 +2.09-1.63 x10-2

sin213 < 0.057 (95%

C.L.)

• Combine LETA+Phase III (3He) in single fit

• Pulse Shape Analysis to separate 3He signal from background

• Constrain 3-phase fit using 3He neutron count• Output is 8B flux using NC + Psurv(En)

``Three-Phase’’ Analysis

+

``Three-Phase’’ AnalysisPulse Shape Analysis

Hypoth

esi

s Te

st 1

Hypothesis Test 2

Two 2-D Cuts:

Fit to counter pulse energy spectrum used to constrain number of neutrons in full fit

See poster by R. Martin, N. Oblath, N. Tolich

``Three-Phase’’ AnalysisPulse Shape Analysis

All phases combined with Psurv(En) fit

See poster by P-L. Drouin, C. Howard, N. Barros

Also: expect to bring limits on hep down by x2

Expected Dm2 improvement

Other SNO Results

High frequency periodicity search

See poster by A. Anthony,ApJ. 710:540-548

Low-multiplicity burst search

Expected Sensitivity

Neutrons and spallation products

See poster by J. Loach

• LETA analysis improved precision on NC by more than factor of 2.

• Lowest analysis threshold yet achieved by water Cherenkov technique

• Low E spectrum (still) consistent with no distortion

• First model-independent fit for solar ne survival probability

• 3-flavor analysis shows non-zero q13 but consistent with q13=0:

• Expect further improvement with 3-phase analysis

• Just a few other things left to do…

Summary

sin213=2.00 +2.09-1.63 x10-2sin213 < 0.057 (95% C.L.)

Central runs remove source positioning offsets,

MC upgrades reduce shifts

Fiducial volume uncertainties (> factor of 3 improvement: Old: Phase I ~ ±3% Phase II ~ ±3% New: Phase I ~ ±1% Phase II ~ ±0.6%

Systematic UncertaintiesPosition

Tested with: neutron captures, 8Li, outside-signal-box ns

Old New

Systematic UncertaintiesIsotropy (b14)

MC simulation upgrades provide biggest source of improvementTests with muon `followers’, Am-Be source, Rn spike

b14 Scale uncertainties (factor of 2 improvement): Old: Phase I --- , Phase II = ±0.85% electrons, ±0.48% neutrons New: Phase I ±0.42%, Phase II =±0.24% electrons,+0.38%

-0.22% neutrons

8B Flux Result

NC = 5.140 +4.0 -3.8 %

J. N. Bahcall, A. M. Serenelli, and S. Basu, AstroPhys. J. 621, L85 (2005)

Calibrations

Parameters for simulation measured and tested with sources

• Laser source (optics/timing)• 16N 6.13 MeV ’s• Radon `spikes’• Neutrons 6.25 MeV ’s• pT 19.8 MeV ’s• 8Li ’s, E<14 MeV• Encapsulated U and Th

sources

Monte Carlo Upgrades

Volume-weighted uncertainties: Old: Phase I = ±1.2% Phase II = ±1.1% New: Phase I = ±0.6% Phase II = ±0.5% (about half Phase-correlated)

Systematic UncertaintiesEnergy Scale

No correction With correction

16N calibration source6.13 MeV gs

Tested with: Independent 16N data, n capture events, Rn `spike’ events…

New Cuts Summary

~80% reduction in external bkds

Direct Fit for Energy-Dependent Survival Probability

Previous global best-fit LMA point: tan212 = 0.468, m2 = 7.59x10-5 eV2

DAYNIGHT

Survival Probability

DAY

NIGHT

Survival Probability

DAY

NIGHT

Survival Probability

DAY

NIGHT

Survival Probability

DAY

NIGHT

Oscillation Analyses: Global SolarLETA paper 2009:LETA joint-phase fit+ Phase III+ all solar expts

Best-fit LMA point:

tan212 = 0.457 (+0.038 -0.041)

m2 = 5.89x10-5 eV2 (+2.13 -2.16)