Opera neuvel de-cern

Measurement of the neutrino velocity with
the OPERA detector in the CNGS beam

Dario Autiero
IPN Lyon

On behalf of the OPERA Collaboration

D. Autiero - CERN - 23 September 2011 1

The OPERA Collaboration
160 physicists, 30 institutions, 11 countries

Italy Korea
Belgium Jinju
IIHE-ULB Brussels LNGS Assergi
Bari
Bologna Russia
Croatia
IRB Zagreb LNF Frascati INR RAS Moscow
L’Aquila LPI RAS Moscow
Naples ITEP Moscow
France Padova SINP MSU Moscow
LAPP Annecy Rome JINR Dubna
IPNL Lyon
Salerno
IPHC Strasbourg
Switzerland
Japan Bern
Germany Aichi ETH Zurich
Hamburg Toho
Kobe
Israel Nagoya Turkey
Technion Haifa Utsunomiya METU Ankara

http://operaweb.lngs.infn.it/scientists/?lang=en

We profited from the collaboration of individuals and groups that worked
with us for the various metrology measurements reported here:

CERN: CNGS, Survey, Timing and PS groups

The geodesy group of the Università Sapienza of Rome

The Swiss Institute of Metrology (METAS)

The German Institute of Metrology (PTB)


Principle of the neutrino velocity measurement

Definition of neutrino velocity:
Definition of neutrino velocity:

ratio of precisely measured baseline and time of flight
ratio of precisely measured baseline and time of flight

Time of flight measurement:

tagging of neutrino production time
tagging of neutrino interaction time by a far detector

accurate determination of the baseline (geodesy)

expected small effects: long baseline required

blind analysis: “box” opened after adequate level of systematic
errors was reached


Past experimental results

FNAL experiment (Phys. Rev. Lett. 43 (1979) 1361)

high energy (Eν > 30 GeV) short baseline experiment. Tested deviations down
to |v-c|/c ≤ 4×10-5 (comparison of muon-neutrino and muon velocities).

SN1987A (see e.g. Phys. Lett. B 201 (1988) 353)

electron (anti) neutrinos, 10 MeV range, 168’000 light years baseline.
|v-c|/c ≤ 2×10-9.
Performed with observation of neutrino and light arrival time.

MINOS (Phys. Rev. D 76 072005 2007)

muon neutrinos, 730 km baseline, Eν peaking at ~3 GeV with a tail extending
above 100 GeV.
(v-c)/c = 5.1 ± 2.9×10-5 (1.8 σ).


THE DESIGN OF THE OPERA EXPERIMENT
ECC BRICKS + ELECTRONIC DETECTORS FOR νµ ντ OSCILLATION STUDIES

νµ beam

ECC brick
1 mm
electronics
trackers
τ

ντ

Pb

emulsion layers

interface films (CS)


THE IMPLEMENTATION OF THE PRINCIPLE
SM1 SM2

Target area Muon spectrometer


The Target Tracker (TT)
pre-location of neutrino interactions and event timing

- Extruded plastic scintillator strips (2.6 cm width)
- Light collections with WLS fibres
- Fibres read out at either side with multi-anode 64 pixels PMTs (H7546)

1.7 m 64 strips

TT module 6.9m

Wave length shifting fibers
H7546
Read out by 1 Front-End DAQ board per side

OPERA readout scheme
General Server ORACLE D.B. Server & replicant

Internet GPS
antenna
firewall
CISCO 3750
DHCP
DNS
External Manager
lab Optical link

Undergound
lab

TT RPC PT TT RPC PT GPS
DAQ 1 DAQ 1 DAQ 1 DAQ 2 DAQ 2 DAQ 2 clock
PC

CISCO 3750
CISCO 2950
CISCO 2950

×31 ×22 ×2 ×6 ×31 ×22 ×2 ×6
Veto
trigger ×3 ×3
2 16 2 8 16 2 8
PT PT
RPC XPC RPC XPC
trigger trigger
trigger trigger trigger trigger
E/O E/O

Veto T.T. RPC XPC PT T.T. RPC XPC PT

Trigger-less, asynchronous Front-End nodes (1200); Gigabit Ethernet network

Clock distribution system (10 ns UTC event time-stamp granularity)
Front-end Target Tracker
board 31 planes (SM1)

Plane i

Master board

OPERA
master clock

OPERA master clock
1 PPms local oscillator VECTRON
OC-050 (stability 10-12/s)

Mezzanine DAQ card common to all sub-detectors Front End nodes:
CPU (embedded LINUX), Memory, FPGA, clock receiver and ethernet

“INTERNAL” and “EXTERNAL” OPERA EVENTS

νµ CC

µ from external interaction

NC


The LNGS underground physics laboratory

CERN
73
0k
m

LNGS

1400 m

CNGS

OPERA


THE CNGS neutrino beam

• SPS protons: 400 GeV/c
• Cycle length: 6 s
• Two 10.5 µs extractions (by kicker magnet) separated by 50 ms
• Beam intensity: 2.4 1013 proton/extraction
• ~ pure muon neutrino beam (<E> = 17 GeV) travelling through
the Earth’s crust

CNGS events selection

GPS

TSPS TOPERA

Offline coincidence of SPS proton extractions (kicker time-tag) and OPERA events
|TOPERA – (TKicker + TOFc)| < 20 µs

Synchronisation with standard GPS systems ~100 ns (inadequate for our purposes)
Real time detection of neutrino interactions in target and in the rock surrounding OPERA


CNGS events selection

cosmics

D. Autiero - CERN - 23 September 2011

OPERA data: narrow peaks of the order of the spill width (10.5 µs)

Negligible cosmic-ray background: O(10-4)

Selection procedure kept unchanged since first events in 2006


From CNGS event selection to neutrino velocity measurement

Extraction 1

ns

Extraction 2

ns

Typical neutrino event time distributions in 2008 w.r.t kicker magnet trigger pulse:

1) Not flat
2) Different timing for first and second extraction

Need to precisely measure the protons spills
Need to precisely measure the protons spills

Clock 1
GPS clocks at LNGS w.r.t.
60 ns Cs clock:

(ns)
1) Large oscillations
hours 2) Uncertainties on CERN-OPERA
Clock 2 synchronisation

hours

Need accurate time synchronisation system

Collaboration with CERN timing team since 2003
Collaboration with CERN timing team since 2003

Major upgrade in 2008
Major upgrade in 2008


OPERA sensitivity

• High neutrino energy - high statistics ~16000 events

• Sophisticated timing system: ~1 ns CNGS-OPERA synchronisation

• Accurate calibrations of CNGS and OPERA timing chains: ~ 1 ns level

• Precise measurement of neutrino time distribution at CERN through proton
waveforms

• Measurement of baseline by global geodesy: 20 cm accuracy over 730 km

Result: ~10 ns overall accuracy on TOF with similar stat. and sys. errors


CNGS-OPERA synchronization

Time-transfer Time-transfer
equipment equipment


Standard GPS receivers ~100 ns accuracy:
CERN Symmetricom XLi (source of General Machine Timing)
LNGS: ESAT 2000

2008: installation of a twin high accuracy system calibrated by METAS (Swiss metrology
institute) Septentrio GPS PolaRx2e + Symmetricom Cs-4000

PolaRx2e:

• frequency reference from Cs clock
• internal time tagging of 1PPS with respect to individual satellite observations
• offline common-view analysis in CGGTTS format
• use ionosphere free P3 code

Standard technique for high accuracy time transfer

Permanent time link (~1 ns) between reference points at CERN and OPERA


LNGS

ESAT 2000

PolaRx2e

PolaRx2e

Cs 4000

Cs 4000

CERN
CERN 21

GPS common-view mode

Standard GPS operation:
resolves x, y, z, t with ≥ 4 satellite observations

Common-view mode (the same satellite for the two
sites, for each comparison):
x, y, z known from former dedicated measurements:
determine time differences of local clocks (both
sites) w.r.t. the satellite, by offline data exchange

730 km << 20000 km (satellite height) similar
paths in ionosphere


Result: TOF time-link correction (event by event)


CERN-OPERA inter-calibration cross-check
Independent twin-system calibration by the Physikalisch-Technische Bundesanstalt

High accuracy/stability portable time-
transfer setup @ CERN and LNGS

GTR50 GPS receiver, thermalised, external
Cs frequency source, embedded Time
Interval Counter

Correction to the time-link:

tCERN - tOPERA= (2.3 ± 0.9) ns


Proton timing by
Beam Current
SPS Transformer

Fast BCT 400344
(~ 400 MHz)

CNGS

Proton pulse digitization:

• Acqiris DP110 1GS/s waveform digitizer (WFD)

• WFD triggered by a replica of the kicker signal

• Waveforms UTC-stamped and stored in CNGS
database for offline analysis
2010 calibration with Cs clock

Proton spill shape

Reminiscence of the Continuous Turn extraction from PS (5 turns)

SPS circumference = 11 x PS circumference: SPS ring filled at 10/11

Shapes varying with time and both extractions

Precise accounting with WFD waveforms:

more accurate than: e.g. average neutrino distribution in a near detector
26

Typical waveform (2011)

(zoom)

(zoom)
Fourier analysis
5 ns

200 MHz


Neutrino event-time distribution PDF

• Each event is associated to its proton spill waveform
• The “parent” proton is unknown within the 10.5 µs extraction time

normalized waveform sum: PDF of predicted time distribution of neutrino events
compare to OPERA detected neutrino events

Extraction 1 Extraction 2
2009- 2011 2009- 2011
PDF PDF

(ns) (ns)

different timing w.r.t. kicker magnet signal

Neutrino production point

Unknown neutrino production point:

1)accurate UTC time-stamp of protons
2)relativistic parent mesons (full FLUKA simulation)

TOFc = assuming c from BCT to OPERA (2439280.9 ns)
TOFtrue = accounting for speed of mesons down to decay point
∆t = TOFtrue -TOFc
〈∆t〉 = 1.4 x 10-2 ns

Summary of the principle for the TOF measurement

y x

z

Measure δt =TOFc - TOFν

Geodesy at LNGS

Dedicated measurements
at LNGS: July-Sept. 2010
(Rome Sapienza
Geodesy group)

2 new GPS benchmarks
on each side of the 10 km
GPS highway tunnel

GPS measurements
ported underground to
OPERA

GPS

Combination with CERN geodesy

CERN –LNGS measurements (different periods) combined in the
ETRF2000 European Global system, accounting for earth dynamics
(collaboration with CERN survey group)

LNGS benchmarks
In ETRF2000

Cross-check: simultaneous CERN-LNGS measurement of GPS
benchmarks, June 2011

Resulting distance (BCT – OPERA reference frame)
(731278.0 ± 0.2) m


LNGS position monitoring

Monitor continent drift and important geological events (e.g. 2009 earthquake)

Time calibration techniques
Delay TA ?
Start End
• Portable Cs-4000:
Comparison: time-tags vs 1PPS signal (Cs clock)
at the start- and end-point of a timing chain

Start End

CS 1PPS CS 1PPS

• Double path fibers measurement:
by swapping Tx and Rx component of the opto-chain
TA ?
Start End
Optical fiber
TB ? Measure TA-TB

Measure TA+TB TA,TB
TA ?
Start End
Optical fiber
TB ?


BCT calibration (1)

Dedicated beam experiment:

BCT plus two pick-ups (~1 ns) with LHC beam (12 bunches, 50 ns spacing)

∆tBCT = t4 - t3 = (580 ± 5) ns
t4

WFD

t1 t2 t3

BPK1 BPK2 BCT

t3 : derived by t1 - t2 measurement and survey


BCT calibration (2)

ZOOM

LHC beam

result: signals comparison after
∆BCT compensation


(+) delays increasing δt
(-) delays decreasing δt


TT time response measurement

FPGA
PMT

Scintillator, WLS fibers, PMT, analog FE chip (ROC)
up to FPGA trigger input

UV laser excitation:
delay from photo-cathode to FPGA input: 50.2 ± 2.3 ns

Average event time response: 59.6 ± 3.8 ns (sys)

(including position and p.h. dependence, ROC time-walk,
DAQ quantization effects accounted by simulations)


Delay calibrations summary
Item Result Method

CERN UTC distribution (GMT) 10085 ± 2 ns • Portable Cs
• Two-ways
WFD trigger 30 ± 1 ns Scope

BTC delay 580 ± 5 ns • Portable Cs
• Dedicated beam experiment
LNGS UTC distribution (fibers) 40996 ± 1 ns • Two-ways
• Portable Cs
OPERA master clock distribution 4262.9 ± 1 ns • Two-ways
• Portable Cs
FPGA latency, quantization curve 24.5 ± 1 ns Scope vs DAQ delay scan
(0.5 ns steps)
Target Tracker delay 50.2 ± 2.3 ns UV picosecond laser
(Photocathode to FPGA)
Target Tracker response 9.4 ± 3 ns UV laser, time walk and photon
(Scintillator-Photocathode, arrival time parametrizations, full
trigger time-walk, quantisation) detector simulation

CERN-LNGS intercalibration 2.3 ± 1.7 ns • METAS PolaRx calibration
• PTB direct measurement


Continuous two-way measurement of UTC delay at CERN
(variations w.r.t. nominal)

± 0.4 ns

July 2011
Technical stop
(no data)


Event selection (earliest TT hit of the event as “stop”)
Statistics: 2009-2010-2011 CNGS runs (~1020 pot)

Internal events:
Same selection procedure as for oscillation searches: 7586 events

External events:
Rock interaction require muon 3D track: 8525 events
(Timing checked with full simulation, 2 ns systematic uncertainty by adding external
events)

Data/MC agree for 1st hit timing (within systematics)

Event time corrections

Time-link correction (blue points)

origin
14 m
Correction due to the earliest
hit position

average correction: 140 cm
(4.7 ns)


Analysis method
For each neutrino event in OPERA proton extraction waveform

Sum up and normalise: PDF w(t) separate likelihood for each extraction

Maximised versus δt:

δt= TOFc- TOFν

Positive (negative) δt neutrinos
arrive earlier (later) than light

statistical error evaluated from log
likelihood curves


Blind analysis

Analysis deliberately conducted by referring to the obsolete timing of
2006:

1) Wrong baseline, referred to an upstream BCT in the SPS, ignoring accurate
geodesy
2) Ignoring TT and DAQ time response in OPERA
3) Using old GPS inter-calibration prior to the time-link
4) Ignoring the BCT and WFD delays
5) Ignoring UTC calibrations at CERN

Resulting δt by construction much larger than individual
Resulting δt by construction much larger than individual
calibration contributions ~ 1000 ns
calibration contributions ~ 1000 ns
“Box” opened once all correction contributions reached
“Box” opened once all correction contributions reached
satisfactory accuracy
satisfactory accuracy


Data vs PDF: before and after likelihood result
First extraction Second extraction
δt = 0 ns δt = 0 ns

(BLIND) δt = TOFc-TOFν =

(1048.5 ± 6.9) ns (stat)

First extraction Second extraction
δt = 1048.5 ns δt = 1048.5 ns χ2 / ndof :

first extraction: 1.06
second extraction: 1.12


Zoom on the extractions leading and trailing edges


Analysis cross-checks

1) Coherence among
CNGS
runs/extractions
2) No hint for e.g. day-
night or seasonal effects:

|d-n|: (17.1 ± 15.5) ns

|(spring+fall) – summer|:
(11.3 ± 14.3) ns

3) Internal vs external events:

All events: δt (blind) = TOFc -TOFν = (1048.5 ± 6.9 (stat.)) ns

Internal events only: (1047.4 ± 11.2 (stat.)) ns


Opening the box

timing and baseline corrections systematic uncertainties

Systematic uncertainties ns
Baseline (20 cm) 0.67
Decay point 0.2
Interaction point 2
UTC delay 2
LNGS fibres 1
DAQ clock transmission 1
FPGA calibration 1
FWD trigger delay 1
CNGS-OPERA GPS synchronization 1.7
MC simulation (TT timing) 3
TT time response 2.3
BCT calibration 5

Total uncertainty (in quadrature) 7.4


Results

For CNGS νµ beam, <E> = 17 GeV:

δt = TOFc-TOFν=

(1048.5 ± 6.9 (stat.)) ns – 987.8 ns = (60.7 ± 6.9 (stat.) ± 7.4 (sys.)) ns

relative difference of neutrino velocity w.r.t. c:

(v-c)/c = δt /(TOFc - δt) = (2.49 ± 0.28 (stat.) ± 0.30 (sys.)) ×10-5

(730085 m used as neutrino baseline from parent mesons average decay point)

6.0 σ significance


Study of the energy dependence

<E>=28.1 GeV

• Only internal muon-neutrino CC events used for energy measurement (5489
events)
(E = Eµ + Ehad)

• Full MC simulation: no energy bias in detector time response (<1 ns)
systematic errors cancel out

δt = TOFc-TOFν= (60.3 ± 13.1 (stat.) ± 7.4 (sys.)) ns for <Eν> = 28.1 GeV
(result limited to events with measured energy)


No clues for energy dependence within the present sensitivity in
the energy domain explored by the measurement


Conclusions (1)

• The OPERA detector at LNGS in the CERN CNGS muon neutrino beam has allowed the
most sensitive terrestrial measurement of the neutrino velocity over a baseline of about 730
km.

• The measurement profited of the large statistics accumulated by OPERA (~16000 events), of
a dedicated upgrade of the CNGS and OPERA timing systems, of an accurate geodesy
campaign and of a series of calibration measurements conducted with different and
complementary techniques.

• The analysis of data from the 2009, 2010 and 2011 CNGS runs was carried out to measure
the neutrino time of flight. For CNGS muon neutrinos travelling through the Earth’s crust with
an average energy of 17 GeV the results of the analysis indicate an early neutrino arrival time
with respect to the one computed by assuming the speed of light:

δt = TOFc-TOFν= (60.7 ± 6.9 (stat.) ± 7.4 (sys.)) ns

• We cannot explain the observed effect in terms of known systematic uncertainties. Therefore,
the measurement indicates a neutrino velocity higher than the speed of light:

(v-c)/c = δt /(TOFc - δt) = (2.48 ± 0.28 (stat.) ± 0.30 (sys.)) ×10-5

with an overall significance of 6.0 σ.


Conclusions (2)

• A possible δt energy dependence was also investigated. In the energy domain
covered by the CNGS beam and within the statistical accuracy of the measurement
we do not observe any significant effect.

• Despite the large significance of the measurement reported here and the stability of
the analysis, the potentially great impact of the result motivates the continuation of
our studies in order to identify any still unknown systematic effect.

• We do not attempt any theoretical or phenomenological interpretation of the results.


Thank you for your attention
Thank you for your attention


Opera neuvel de-cern

More Related Content

What's hot (18)

Viewers also liked (15)

Similar to Opera neuvel de-cern (20)

More from orengomoises (20)

Recently uploaded (20)

Opera neuvel de-cern