Search for Proton Decay in Super-Kamiokande

Yusuke SudaMini-Workshop for High Energy Gamma Ray Astrophysics

Max-Planck-Institute for Physics, Nov. 24, 2017

Who am I• I am Yusuke Suda from Univ. of Tokyo, Japan

• Master of Science (Mar. 2014)

• “Research and Development of Large-Aperture Hybrid Photo-Detectors for Hyper-Kamiokande”

• Doctor of Philosophy (Sep. 2017)

• “Search for Proton Decay Using an Improved Event Reconstruction Algorithm in Super-Kamiokande”

• Postdoc at Center for High Energy gEophysics Research (CHEER), Earthquake Research Institute (ERI), UTokyo

• Muon radiography (Muography) for geoscience

2

First

What is Proton Decay• Phenomena predicted by Grand Unified Theories (GUTs) of

elementary particles

• Unification of three fundamental forces (EM, Weak, Strong)

• Very high energy scale ( ≧1015 GeV)

• Solve electromagnetic charge quantization, etc.

• Mixture between quarks and leptons → Proton decay

• Proton decay is the best tool to test GUTs!

3

Accelerators

Prot

on

⇡0

↵↵

Predicted Proton lifetime:1032 - 1039 years

⌧ ⇠ M4X

↵2m5p

Year1940 1960 1980 2000 2020 2040

Life

time

Lim

it (y

ears

)

2110

2510

2910

3310

3710 0π+ e→Minimal SU(5), p0π+ e→Minimal SO(10), p

Goldhaberet alFlerov

Evans & SteinbergFiremanBennett

(1954)et alReines (1958)et alReines

et alBackenstoss Giamati & ReinesKropp & Reines

et alGurr Bergamasco & PicchiReines & Crouch

et alLearned (NUSEX)et alBattistoni

(KGF)et alKrishnaswamy (Soudan)et alBartelt (Frejus)et alBerger (Homestake)et alCherry

(Kamiokande)et alHirata (IMB)et alMcGrew

(Super-K I)et alShiozawa (Super-K I-II)et alNishino

(Super-K I-IV)et alAbe

Proton Decay So Far

4

SO(10) GUT

SU(5) GUT

Water Cherenkov

Iron Calorimeter

Liquid Scintillator

Geochemical Radiochemical

(1954)

• Long history of over 60 years. No significant candidate

• Super-Kamiokande (SK) is the top and still in predictions

• Proton decay may happen at any time

• Let’s search proton decay in SK!

Baryon number conservation

5

Super-KamiokaNDE• 50kton water Cherenkov detector

• 1000m underground, Kamioka, JPN

• Record Č light (Q・T) by 20-inch PMTs

• PID by Č ring pattern

SK1 SK2 SK3 SK4

Period 1996-2001 2002-2005 2006-2008 2008-

Live time 1489.2 798.6 518.1 2650.4(~Sep. 2016)

Photocoverage(# of PMTs)

40%(11,146)

19%(5,182)

40%(11,129)

40%(11,129)

Outer Detector

(veto)

InnerDetector

50kton Tank

ucleonecay

xperiment

39.3 m

41.4

m

Half of the all data

e, γ

Fuzzy

μ

Sharp

6

7

1033 protons in inner detector

10 years null observation of proton decay→ Proton lifetime > 1034 years

8

My target decay mode: p→e+π0 ・Most dominant in non-SUSY GUTs ・Most sensitive for SK

Decay time of π0: 8×10-17 s Gamma conversion length: ~40cm

9

Back-to-back topology→ Easily distinguish from

atmospheric neutrinos

e+

γ

γ

Proton Decay Simulation

Elec.

Cable hole μ

Invisible μ

Stopping μ

Through-going μ

Radioactive BG

FC neutrino

VETO

Rock

PC neutrino

UPMU neutrino

Flasher

Proton decay

Michel e

Michel e

Data Flow

10

106 events/day

8 events/day

Reconstruction

Proton decay search

Fully contained reduction(no OD activity, no flasher

PMT etc.)

Mostly atmospheric neutrino events

(BG for proton decay)

Vertex, #rings, PID, momentum etc.

Cut-based analysis

I was an expert of fully contained reduction(trouble shooting, preparation for data set etc.)

Reconstruction Algorithm• Conventional: APfit

• Determine reconstruction params. step-by-step (vertex → #rings → PID → momentum)

• Use charge&time information of “hit” PMT only

• Momentum determination by using observed charge inside Č-cone w/ a half angle of 70° → Bias

• Developed 20 years ago and written by Fortran (hard to maintain)

• New: fiTQun

• Determine all params simultaneously by a maximum likelihood method

• Use not only hit PMT information but also “unhit” PMT information

• Initial development by the T2K experiment, written by C++ 11

12

fiTQun

Charge PDF Time PDFUnhit PDF

Distance btw. true/recon. vertex (cm)0 10 20 30 40 50 60 70 80 90 100

Num

ber o

f Eve

nts

0

500

1000

1500

2000

2500

3000

3500

4000

): σResolution (1: 19.6 cm (Dwall>1.5m)fiTQun: 19.5 cm (Dwall>2.0m)fiTQun

: 29.6 cm (Dwall>2.0m)APfit

True Momentum (MeV/c)0 200 400 600 800 1000 1200 1400

Mom

entu

m R

esol

utio

n (%

)

0123456789

10

(Dwall>1.5m)fiTQun (Dwall>2.0m)fiTQun

(Dwall>2.0m)APfit

Hit PDF

Vertex Resolution Momentum ResolutionEx) Atm.-νe CCQE MC

(Poisson w/ correction)

• Fit params. x = {x, y, z, t, θ, φ, p} × (# of rings)• Construct likelihood function for given ring(s) hypotheses• Maximize likelihood for each event

(Poisson) (Gaussian)

Observed Q&T

)µLe/Lln(-3000 -2000 -1000 0 1000 2000 3000

Num

ber o

f Eve

nts

0

100

200

300

400

500

600

700

800 SK-IV DataπCC w/o eν

oth.0πCC w/ 0eν 0πCC w/ eνπCC w/o µν

oth.0πCC w/ 0µν0πCC w/ µν

NC

Validation of fiTQun• Performance in MC was well examined, but treatment of real

data was not enough to physics analysis

• I did

• Check in data/MC likelihood distributions

• Correction of time dependent detector params.

• Estimation of energy scale uncertainty

• Uncertainty 2.1%, same as APfit

• Tuning of ring counting parameter

• Finally, fiTQun has been verified for SK-IV data/MC

• Then, T2K experiment employed fiTQun for their analysis and announced a new hint for leptonic CP violation in summer 2017

13

PID μ←→e

Elapsed Years0 1 2 3 4 5 6 7 8

PMT

Gai

n (a

.u.)

0.95

1

1.05

1.1

1.15

1.2

Calibration of fiTQun

14

Date1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018

Wat

er A

ttenu

atio

n Le

ngth

(m)

50

60

70

80

90

100

110

120

130

PMT gain corr. Water Atten. corr.w/o corr.

Mom

entu

m/R

ange

(M

eV/c

/cm

)

Ex) SK1

RMS/Mean=2.2% RMS/Mean=1.0% RMS/Mean=0.6%

SK1SK2SK3SK4

Year

+2%/year

• PMT gain and water attenuation length vary with time

PMT Gain Water Attenuation Length

Search for Proton Decay• First application of fiTQun to proton decay

• First time of changing SK proton decay analysis framework

• Target decay mode: p→e+π0

• Hybrid search: APfit (SK1-3, null observation) + fiTQun (SK4)

• Selection criteria

15

1. Fiducial volume

2. Number of rings (2 or 3)

3. PID (all shower ring, no Micheal-e)

4. π0 mass (for 3-ring events)

5. No gamma-ray from neutron capture

6. Total invariant mass and momentum cut

Distance from the wall to the true vertex (cm)0 200 400 600 800 1000 1200 1400 1600 1800

Effic

ienc

y in

eac

h bi

n

0

0.2

0.4

0.6

0.8

1

1.2

16

• By changing selection criteria for fiTQun, I did

• Expand fiducial volume by 10%

• Reduce # of BG events by arrpox. 30%

• Tighter total invariant mass cut is applied

• While keeping similar level of signal efficiency as APfit

fiTQun free-pAPfit free-p

fiTQun all-pAPfit all-p

APfitfiTQun

Chance of discovery!(22.5 kton → 24.7kton)

17

fiTQun APfitLower box Upper box Lower box Upper box

Eff. (all) (%) 20.0±0.3 18.1±0.3 19.0±0.3 19.0±0.3BG (/Mt/yr) 0.028±0.019 0.778±0.102 0.030±0.021 1.116±0.132

＊stat. error only

Bound-pFree-p

Signal MC Atm-ν MC (500 years)

• To enhance sensitivity, signal region is divided by two

• Lower box: less BGs & systematics error (Fermi motion)

• #total BG in SK4: ~0.14 (0.19) events for fiTQun (APfit)

Upper boxLower box

Result

18

)2Total Invariant Mass (MeV/c0 200 400 600 800 1000 1200

Tota

l Mom

entu

m (M

eV/c

)

0

100

200

300

400

500

600

700

800

900

1000

)2Total Invariant Mass (MeV/c0 200 400 600 800 1000 1200

Tota

l Mom

entu

m (M

eV/c

)

0

100

200

300

400

500

600

700

800

900

1000

APfit (163.0 kton*years)fiTQun (179.5 kton*years)

SK-IV SK-IV

• No candidate was found in SK-IV data for both fitters

• Combine with the other SK data (APfit, no candidates) and calculate lower lifetime limit by Bayes’ theorem

• Lifetime limit: τ/B(p→e+π0) > 1.88×1034 years @ 90% C.L.

• 5% improvement from APfit-based analysis

Not official yetNot official yet

Signal box

Summary• Proton decay is a smoking gun for GUTs

• Developed and validated the new event reconstruction algorithm, fiTQun

• Improved search for p→e+π0 with fiTQun was conducted

• Same efficiency but fiducial volume +10%, #BGs -30%

• No candidate was found

• World leading proton lifetime limit: 1.88×1034 years @ 90% C.L.

• Most stringent constraint for non-SUSY GUT

19

Detail information can be found in Suda’s thesis http://www-sk.icrr.u-tokyo.ac.jp/sk/publications/index-e.html#doctor

20Let’s go as much as we can!

74 m

60 m

Think Bigger

• Hyper-Kamiokande project (~10 times bigger than SK)

• Search region of proton decay will reach 1035 years

• Many rich physics: CP violation, ν mass hierarchy, SN relic etc.

• FiTQun is compatible with HK

• Photosensor is a key to success21

Hyper-K

Super-K

KAMIOKANDE 1983

1996

2026 (expected)

/ ndf 2χ 473.8 / 161Prob 1.465e-32p0 6.4± 375.9 p1 0.057± 9.577 p2 0.064± 2.133 p3 7.0± 587.1 p4 0.1± 18.8 p5 0.122± 3.304 p6 14.7± 652.2 p7 0.1± 27.8 p8 0.149± 3.082 p9 39.8± 510.5 p10 0.20± 36.37 p11 0.285± 3.413 p12 8.2± 389.9 p13 0.45± 46.07 p14 0.634± 5.003

Charge [pC]0 20 40 60 80 1000

100

200

300

400

500

600

700 / ndf 2χ 473.8 / 161

Prob 1.465e-32p0 6.4± 375.9 p1 0.057± 9.577 p2 0.064± 2.133 p3 7.0± 587.1 p4 0.1± 18.8 p5 0.122± 3.304 p6 14.7± 652.2 p7 0.1± 27.8 p8 0.149± 3.082 p9 39.8± 510.5 p10 0.20± 36.37 p11 0.285± 3.413 p12 8.2± 389.9 p13 0.45± 46.07 p14 0.634± 5.003

Multi-photoelectron distribution

Avalanche Diode

8-inch

200ton tank

Cosmic-ray μInstall

R&D of Hybrid Photo-Detector• One of candidates. Better performance than SK PMT

• Hope to use in HK (20-inch HPD is testing in water tank)

22

SudaHPD

Photon counting!

1PE

Photon counting!

2PE 3PE4PE

5PE

Current Research

23

↑ Satsuma-Iwojima Volcano, Japan

H. Tanaka, 2009

2m

3m2mScintillation detector →

μ

Magma

• Muon radiography (Muography) @ Earthquake research institute, Japan

• Explore inner structure of volcanos and active fault by measuring cosmic-ray muon flux through target

• Commit to geophysics and disaster prevention

• One of the few applications of high energy physics

Active Fault

24

• Muography of Atotsugawa fault (near Super-K)

• Estimation of 3D density profile of the fault by measuring muon flux (θ, φ) at each depth (Data taking is ongoing)

• I will evaluate expected muon flux at detector by GEANT4

• Working hard to publish the world’s first result

Borehole

Fractured zone(low density)

Rock

Detector

μTerrain @ GEANT4

Scintillation muon detector

Transition of My Research

25

Past

Now

Future?

Expected Future Work

26

• Gamma-ray burst search @ CTA LST

• GRB is the brightest explosion in the universe

• Explore mechanisms of jet formation and particle acceleration especially for long GRBs with LST’s high statistics data

• I would like to contribute to

• Low energy threshold (20GeV or less) in order to detect GRBs

• PMT calibration and analysis tool development utilizing my experience (fiTQun, HPD, etc.)

• Pointing calibration in order to not miss GRBs

• Eager to accomplish the first measurement of a GRB by the ground based telescope