Cristiano Bozza – The OPERA Collaboration – Oct 2007, Villa Olmo (Como – Italy) 1/21 Nuclear...

Cristiano Bozza – The OPERA Collaboration – Oct 2007, Villa Olmo (Como – Italy)

1/21

Nuclear Emulsion Scanning in OPERA: Methods and Results

The ECC technique in OPERA

The European Scanning System

The Japanese Scanning System

Data quality and analysis capabilities


2/21


Emulsion Cloud Chamber:a stack of dense material (e.g. lead) interleavedwith thin tracking planes of nuclear emulsions

Ionizing particles sensitize atomsUpon development, metallic Ag growsin grains 0.5÷1 μm

AgBr in gel matrix – continuous sensitivity since production

Opera ECC cell: 1 lead plate (1 mm)1 emulsion plate (291μm)in contact

Emulsion layers (43 μm)

Transparent plastic support (205 μm)


3/21


Opera: ECC bricks in a beam (CNGS), searching for appearancedue to neutrino flavour oscillation

Average residuals of grains from microtrack fit: 0.066 μm

Topological analysis of interactions is possible to search for nonambiguous appearance signature

X

X

X

CNGS

microtrack = sequence of aligned grainsin a single emulsion layer


4/21


Evolution of automatic scanning systems

OPERA (CERN-LNGS)

TS (Track Selector)Nagoya

NTS (New Track Selector) Nagoya

UTS (Ultra Track Selector) Nagoya

S-UTS (Super UTS) Nagoya

SySal (System of Salerno)Salerno

ESS (European ScanningSystem)European labs

CERN-NIKHEF system

1990 1996 1998 2003 2005

Speed: 1 cm2/h/side Speed: 2050 cm2/h/side

CHORUS (WA95-CERN)


5/21


CMOS camera1280×1024 pixel256 gray levels376 frames/sec(Mikrotron MC1310)

XY stage (Micos)0.1 μm nominalprecision

Emulsion Plate

Z stage (Micos)0.05 μm nominalprecision

Illumination system, objective (Oil 50× NA 0.85) and optical tube (Nikon)


6/21


The ESS is mostly SW-based and very modular

New, better components are quickly integratedand performances increase

2000 2001 2003 2004 2007

0

2

4

6

8

10

12

14

16

18

20

Sp

eed

(cm

2/h

)

Year

DA

LSA

1M

60

C

am

era

,

3

×Genesi

sPhoto

nfo

cus

cam

era

,M

atr

ox O

dyss

ey

Mik

rotr

on

cam

era

SW

im

pro

vem

ents

Hit

ach

i K

PF1

10

ca

mera

Scanning speed (area scan of 43 m thick emulsion)


7/21

280×365μm2

Tomographic sequencesZ axis moving, 2D imagesspanning emulsion thickness

Move XYZ to next view

Process/save data

Next field of view,Z at top, new cycle


DAQ cycle(185 ms)

Camera

2D Images(peak 452 MB/s,avg. 97 MB/s)

Vision Processor(Matrox Odyssey)

Binarized2D Images

Host PC(Dual PentiumWorkstation)

RunningWinXP 3D

microtracks

XYZ MotionCommands

Motion Controller(National Instruments

FlexMotion)

Motors(VEXTA

Nanostep)Power

Functional blocks


8/21


15 images10 grains signal/image,

3000 grains background+noise, shadows, scratches, spots

2D FIR Filter+Equalization+Threshold

Grain recognition (Host PC, multithreadedAssembler code)

300 ÷ 3000 microtracks / view

3D microtrack reconstruction(Host PC, multithreaded C++ code)


9/21


Data sent to temporary file serversDistributed computing based on .NET

Batch Manager coordinatesjobs and directs Scan Servers

Data Processing Serverstake care of heavy computingfor data postprocessing

Data ready for physics analysisare written to Oracle DB servers(also used to share data)

ScanServers:PC + microscopes

File servers + BatchManagers

Private network hub

Data processing servers

Oracle servers

External network gateway

Web server (I IS)

ScanServers:PC + microscopes

File servers + BatchManagers

Private network hub

Data processing servers

Oracle servers

External network gateway

Web server (I IS)


10/21

A track is the result of the connection of two

microtracks on opposite sides of a

plate


Performances of the European Scanning System

Scanning speed: 20 cm2/h/side (40 GB/day/microscope of raw data) Purity: 10 fake tracks / cm2 (slope < 0.5) Efficiency: up to 95% using tracks, ~100% using microtracks (however, background is larger with microtracks) 0.3÷0.7μm precision for reconstructed tracks


11/21


Custom CMOS camera512×512 pixel3000 frames/sec

Piezoelectric fine drivefor Z motion of lens

X axis is drivenwith continuous motion

Oil objective 35× NA 0.85

Mechanics based on Nikonmicroscope stages X/Y/Z nominal precision = 0.1m

Super-UTS microscope (Japanese scanning system)


12/21


The Super-UTS is mostly hardware based, with many custom components

High speed achieved mostly by high repetition rate of DAQ cycle

Piezoelectric motion for lens allows high accelerations with minimum vibrations

X axis is operated in continuous motion on a line; Y axis moves to switch to next line X

Y

Lens motion is synchronised with X motion ofstage to follow the field of view


13/21


The Super-UTS is mostly hardware based, with many custom components

Camera

2D Images(3000/s)

Super-UTS

Image processing,largely based on

FPGA

Motion control

3D microtracks Host PC

3Dmicrotracks

Start/Stop,Control

Motors, Piezo drive

Power

Functional blocks

Image processing and motion control are combined in a single piece of HW:synchronisation must be perfect!


14/21


Principle of track recognition in SUTS: pile up and sum pulse height

Grains of a microtrackin the opticaltomography(after noisefiltering)

Shift images(many trials using predefined slopes)Sum images and get a peak over a background


15/21


Data flow in the SUTS-based scanning station

Data reduction is applied at the level of thelocal storage server by deleting tracks thatcannot be connected across neighbour plates(“linklet” concept)

SUTS Host PCsLocal storage server

OPERA Central DB(Oracle)


16/21

A track is the result of the connection of two

microtracks on opposite sides of a

plate


Performances of the Super-UTS

Scanning speed: 50 cm2/h/side average (72 cm2/h/side peak)

Purity: 104 fake tracks / cm2 (slope < 0.4)

Efficiency: 95% using tracks

104 base trk/cm2/(±0.4)2

Different Ph sum cut

Efficiency vs Number of Fake

12

13

141

516


17/21


Residuals of tracks w.r.t. multi-plate fit of 8 GeV

t, l

Slope=0.2t~0.3m

Top viewSide view

Slope=0.2l~0.5m

Slope=0.7t~0.7m

tl

plate-to-plate alignment by high-energy passing-through-tracks


18/21


Interaction/decay vertex reconstruction

interaction vertex from test exposure at NuMi beam (FERMILAB): not a simulation!Distance of extrapolations at vertex point (in lead): 6.2 m

0 2 4 6 8 10 12 14 16 18 200

2

4

6

8

Co

un

ts

Average distance of tracks at vertex point (m)

Average = 5.9 m

Sample of multi-prong

interactionvertices fromNuMi beam


19/21


Momentum measurement through Multiple Coulomb Scattering

Particle momentum is related to the fluctuation of track slope due to scattering (mostly in lead)

Pb

j

i

ij

Em

Good agreement between reconstructedand simulated momenta26% momentum resolution obtained at 8 GeV/c


20/21


Particle identification through neural network analysis

Input variables : • track and film number• longitudinal and transversal profile

Network output


21/21

Conclusions

The two automated scanning systems used in OPERA have been presented:• the ESS has a SW-oriented approach• the SUTS has a HW-oriented approach

Unprecedented scanning speed is coupled with precision, efficiency and purity20÷50 cm2/h/side on 43m emulsion layers0.3÷0.7 m precision90%÷95% track finding efficiency10÷104 fake tracks / cm2 (slope < 0.5)

Automatic measurements with micrometric precision allow large scale sample for several analysis modes:• Topological event reconstruction (unique feature of nuclear emulsion)• Momentum measurement (26% resolution at 8 GeV/c)• Particle identification (95% e- separation with 1% contamination above 2 GeV/c)

Cristiano Bozza – The OPERA Collaboration – Oct 2007, Villa Olmo (Como – Italy) 1/21 Nuclear...

Documents

Transcript of Cristiano Bozza – The OPERA Collaboration – Oct 2007, Villa Olmo (Como – Italy) 1/21 Nuclear...