The Instrumented Flux Return

1Daniel Azzopardi, December 1999Graduate Student Detector Seminar Series.

Instrumented Flux Return Instrumented Flux Return SeminarSeminar



OutlineOutline

• An Introduction to the IFR - role, status

• IFR Physics Background

• Physics Requirements for the IFR

• Hardware Description

• Reconstruction Software• Particle Identification



IntroductionIntroduction

• Instrumented Flux Return (IFR) magnet yoke for external flux path of 1.5 Tesla Super Conducting coil.

• Iron Structure, largest sub-detector

• Absorbs MIP (minimum ionizing particles - ) and , neutrals (Kl )

• Provides structural, seismic support– Allow access to interior sub-systems– Novel graded segmented design.– Uses established instrumentation technology RPCs ( Resistive

Plate Counters)– Novel Cylindrical RPC (iRPC)



• IFR is primary detector for muon identification and long-lived neutral kaon detection:

• Muon identification (~18% of decays)– tagging

• use above 1.4 GeV/c to tag flavour of parent B. • Muons contribute to 75% of dilepton measurements.

– leptonic analyses• EG: Inclusive analysis B Xl , |Vcb|, |Vub|

• Kl identification – CP violation in B0 J/ + K0

l (also use IFR for- J/ + - )

Physics Background Physics Background (1)(1)



Physics Background Physics Background (2)(2)• Require discrimination between , Hadrons ():

M 105.66 MeV/c2 M 139.57 MeV/c2

() 2.19 x 10-6 sec (658.6 m)

() 2.60 x 10-8 sec (7.8 m)

but … primary decay mode is:

+ + (branching ratio 99.99 %)» ~3.5 % of all pions decay before reaching IFR

also approximate multiplicities:

: 8 : 1



• Plots of typical muon and pion spectra from B mesons:

• From EvtGen - Official BaBar Event Generator


0.179/ Event

1.51/ Event



• Muons are minimum ionizing particles, minimal energy deposited in Crystal Calorimeter (some loss in coil also)

• Lose energy via ionization (dE/dx losses)


Critical energy (loss rates by radiation processes equal those of ionization process) depends on material traversed. For muons traversing a solid:

For Iron, Z=26

Ecrit = several

hundred GeV No loss by radiation processes.

876.02.05Z

GeV 6224

critE



• The (Famous) Bethe-Bloch Eqn

• Describes energy loss for heavy charged, moderately relativistic particles

• Basically a function of .

2

2ln

2

11 22

max222

22

I

Tcm

A

ZKz

dx

dE e

Mean rate of energy loss


Max. Ek that can be imparted

to a single electron

Mean excitation energy

Density effect correction



• Bethe-Bloch Curve:


Minimum Ionization



• Possible to integrate Bethe-Bloch to get Range• Important quantity to measure range (comparing , )

is the Nuclear Interaction Length (- length scale appropriate for hadronic cascades). It is denoted I

and defined:

I = , where

- depends on Energy and Material...


I = The Mean Free Path of a particle before undergoing an interaction that is neither elastic nor quasi-elastic (diffractive) in a given medium.

AIN

AdiffelasticTI



• Probability density function for distances between successive collisions is thus:

• Suitable approximation for a given material:

• Eg: For Iron, with A=56, I 133.8 g cm-2


dxx

dxx

exp

1

312-cm 35g A

MATERIAL Z A

NUCLEAR

INTERACTIONLENGTH I

(g/cm2)

DENSITY

(g/cm3)

NUCLEAR

INTERACTIONLENGTH I (cm)

Al 13 26.98 106.4 2.70 39.4

Fe 26 55.85 131.9 7.87 16.76Pb 82 207.2 194 11.35 17.09

Air - - 90.0 1.205 74.69

CsI - - ~167 4.53 36.87



• Fundamental idea then is to exploit difference in behavior between muons and hadrons in Iron:

– muons above a certain minimum energy will either lose all their energy in a certain depth of Iron, or they will range out; escaping detector volume.

– Pions may interacted strongly with Iron Nuclei, producing hadronic showers. May also punch through, mimicking muon, also decay to muon (directly or via hadronic intermediary)- can identify this sometimes (how?)

– Sample charged track or shower development in Iron (how?), discriminate between two cases above.




• Physics Goals of IFR subsystem:– Highest practical efficiency for detection of , misidentification

probability below 5%. Require high efficiency for as low a momentum limit as possible. However, below 500 MeV/c magnetic bending and energy losses mean some never reach IFR - can return no measurement). Lower limit on mis-identification comes from pion punch through and pion decay to muon before IFR. Not a “calorimeter” - input to physics analysis is ID likelihood.

– Kl identification - reconstruction of direction is sufficient for many analyses. Spectrum is much softer, 70% of Kl never reach IFR.

– Also possible to use IFR to veto events with missing hadronic energy.

• Practical Goals– Low-maintenance active detector choice– “Easy” accessibility to remainder of detector

Physics Physics RequirementsRequirements



• IFR Iron consists of three major components– Central Barrel, Forward and Backward Endcaps.

• Also iRPC inside coil.• Complex geometry

Hardware Description Hardware Description (1)(1)

Backward Endcap

Forward Endcap

Barrel

Coil (iRPC not shown)



• Hexagonal configuration - suited to role of structural support and for active detector choices

• Layers of Iron Absorber / Flux Return interspersed with active detector.

• Overall dimensions:– 635 cm length, 584 cm height, width of 675 cm

• Mass:– Barrel: 312 tonnes (excluding supports)– Endcaps: 225 tonnes (each, approximate value. Excludes supports)– Coil: 4.9 tonnes (excluding cryostat and associated systems)

Hardware Hardware Description (2)Description (2)




Photograph of detector, IFR for scale



• Question: Which type of Active Detector? Require low maintenance, high efficiency, and good reliability

• Resistive Plate Counters (Plastic Streamer Tubes were considered - see note 205, RPCs chosen as more durable solution)

– Basic idea: Instrument gaps between absorber sheets; charged component of hadronic shower will signal, as will passage of single charged MIP.

• Question: Is there a better segmentation option than “uniform”?




• From the middle outwards - the 2 mm gap is filled with an Argon-Freon-Isobutane based gas mixture (gas mixture - spark quenching, safety requirements)

• PVC spacers (0.8 cm2 area) placed in 10 cm-square grid ensure RPC planarity, & gas gap (hence also field) remains constant.

• Two Bakelite plates with bulk resistivity in range 8-800 giga ohm cm, (coated on outside with thin layer of graphite, surface resistivity around 10 kohm cm-2) and a 300 micron PVC insulating film, enclose this gap.

• A high (8 kV nominal) potential is applied between the graphite layers. (Al grounded)

Hardware Hardware Description (5)Description (5)• Resistive Plate Counters: (used at L3, and Belle)

• Total RPC areas:– Barrel: 1320 m2 – Endcaps: 1100 m2



Hardware Hardware Description (6)Description (6)• Resistive Plate Counters (cont.):

• Orthogonal aluminium readout strips affixed above and below PVC insulating film.

• Barrel:– Longitudinal strips (measuring z) have a pitch of 38.5 mm (2 mm of which is

space between strips)– Transverse strips (measure ) increase in pitch from 19.7 to 33.5 mm, depending

on layer.

• End Caps:– X strips: oriented horizontally - pitch 28.4 mm giving information on x

coordinates.– Y strips: oriented vertically (38 mm) - y coordinates.

• PVC spacers cause reduction (few %) of active detector area lowered efficiency



Hardware Hardware Description (7)Description (7)• Resistive Plate Counters (cont.):

• Charged particle traversing gap produces quenched spark– detected on external pickup electrodes

• The discharge (around 100 pC) is very fast– pulse rise time is 2ns and duration typically around 20-30ns

Can use IFR for triggering

• At 300 mV, signal pulse requires only very simple fixed threshold electronics for readout. Digital readout.

Measurement error (approx 1.1 cm).

• Can do better than this due to multiple (>2) strip firing - centroid position

determined about ~20% more acurately

• 16 strips are attached to one front end card (FEC) which signal TDC (Time to Digital Converter) circuits to deliver timing information for the active strips

12

w



• RPC Efficiencies• From Cosmic Rays test facility at Frascati,

• Currently ~80% efficiency in live RPC chambers in barrel (Cosmic run 11.29.99)


~120 RPCs Mean 97% Effect of spacers

Courtesy Fabio Anulli




RPC with a circular cut to match the Endcap geometry

Courtesy Francesca Pastore



• Graded Segmentation - why? note 194 describes optimization procedure:

– Simulate different configurations, use Monte Carlo Hadronic shower simulation to estimate achievable efficiency

– uniform segmentation: Muon identification and Kaon detection efficiency improve for a given amount of absorber as thickness of plate decreases; but effect is most important in first interaction length (first 17 cm). ~30% improvement in Kaon efficiency.

– Graded segmentation can decrease instrumentation in subsequent interaction lengths, allow thicker plates, less instrumentation




• The central (Barrel) section is ~375 cm in length. Constructed from (near identical) sextants, situated ~170 cm radially from beam-line, outside the coil. Each extends ~130 cm outwards.

• Innner edge of sextant ~190 cm, outer ~325 cm

• Inner volume segmented into 18 iron plates of increasing thickness in radial distance.

– 9 innermost plates are 2 cm thick (equivalent to one interaction length), followed by– 4 of 3 cm– 3 of 5 cm– 2 of 10 cm each, total thickness of 65 cm of Iron (approximately 4 interaction lengths at normal

incidence).

• Layers of RPCs between these plates fill remaining volume: – Gap of 3.2 cm between Iron plates for housing RPCs– Except last RPC layer (layer 19) on exterior of detector (reduced geometry)– And first 10 layers (between thinnest plates), gaps of 3.5 cm

• Iron plates and RPCs held together by 5 cm thick plates on outside of each sextant. Marginal impact on performance of detector as function of , but structural support role of IFR makes these mandatory

Hardware Description: Hardware Description: BarrelBarrel



Hardware Description: Hardware Description: BarrelBarrel

Barrel Sextant during installation




Hardware Description: Hardware Description: EndcapEndcap• The Endcaps extend solid angle coverage down to 300 mrad in

backward direction, 400 mrad in forward direction.

• Constructed flush with Barrel, however, layer and strip geometry orthogonal to that found in Barrel.

• In addition, approximately 15 cm of iron (providing needed flux return and extra support) between Barrel and both End Caps.

• In order to allow access to rest of the detector, each End Cap divided into two halves vertically.

• Each half also divided into three sections which are reinforced by spacers designed to withstand large magnetic forces on End Cap.

• End Cap has 18 RPC detector layers, starting behind the innermost iron plate.

• Total depth of iron in End Cap is 60 cm, slightly less than in barrel; 5 cm difference due to penultimate iron plate being only 5 cm thick rather than 10 cm as in Barrel.



Hardware Description: Hardware Description: EndcapEndcap




30º

Hardware Description: Hardware Description: iRPCiRPC• iRPC important for Kaon identification• Geometry divided into 4 quadrants (contrast remainder of IFR), each ¼

cylinder• Each quadrant has 2 layers (inner & outer) rotated relative to each other

by 30º so joins are mismatched.• Each layer consists of 4 modules (32 in all).• 147 cm in radius, 2.5 cm thick• Each layer has different view: u,v; , z• 128 strip readouts per view

• Strip pitch for u and v is 2.86 cm. is 1.623 cm, and z is 2.87cm

z

z

30ºu

v

z



Hardware Description: Hardware Description: iRPCiRPC

Inner RPC during insertion




• IFR Electronics and Readout:

– >40000 readouts• 16 channels per FEC > 2500 FECs (located adjacent to RPCs)

– ITB (IFR TDC Board) - identifies bunch crossing time of event using good time resolution of RPCs. Input signal is OR of FECs


• IFB (IFR FIFO board) - buffer between front end and DAQ

• FEE (Front-End Electronics) crate contains the IFB, ICB (IFR Crate Controller), ITB and the ICC (IFR Crate Controller). (8 - 4 for Barrel, 1 per Endcap door)

FEC Readout logical layout



• 3D imaging, assembles cluster candidates from more primitive candidates

• Neutral 3D clusters are formed in 3 stages:

Hits 1D clusters 2D clusters (2 views)

3D cluster (Composite Finder)

3D Composite Cluster

For charged hypotheses, track - 1D cluster association performed with a Swimmer (also used for PID)

Reconstruction (1)Reconstruction (1)



Strip 1D cluster

2D cluster

3D cluster


Luca Lista, with permission



• The Swimmer– Used for track cluster matching (relatively easy due to low

occupancy of IFR), Particle Identification, and alignment– Extrapolates DCH (charged) track through:

EMC, iRPC, Coil, IFR, exit– Also used for clustering using alternative reconstruction

technique– Operation:

• Iterative method (Geant like) using Runge-Kutta to update kinematic quantities

» B-Field strength, momentum, direction, position» Interaction lengths, radiation lengths

• Swimmer road width increases due to multiple scattering estimations» Need to swim with apriori particle hypothesis

• After track swam, calculate 2 for “near-by” 1D - clusters, and associate track to cluster for which this value minimum.




ClusterSwimmer

IN in Fe (Swimmer)

DOF

2 track-cluster match

Geometry

SW + CL

IN to IFR IN in last efficient layer

(“expected” IN )

Number of layers hitNumber of strips hit

IN in Fe (cluster)Last layer hit

Muon Identification Muon Identification (1)(1)



Muon Identification Muon Identification (2)(2) Number of layers hit Muons penetrate further, leaving signal in each

chamber. Pion induced shower may have neutralcomponent, so leave no signal in some chambers.

Number of strips hit Pions interact readily causing a broad shower,hence the mean number of strips hit per layer ishigher than for muons

Last Layer hit Muons reach the last layers of the IFR, below athreshold) equal energy pions don’t

Number of interaction lengthsexpected

Since tracks are swam under the muon hypothesis,- assume a high momentum muon and swim to lastlive layer

Number of interaction lengths tolast layer hit

This is compared with the above quantity – ifcomparable, than the candidate is likely a muon

Number of interaction lengths toIFR

This will be used for lower momentum candidates

Swimmer Track / Cluster match2 / degree of freedom

Pion Chi squared distribution wider due toincreased cluster width

Others…. To come:Kink finding in DchNumber of consecutive layers hit




• Discriminating variables from hit layers - length and extent of cluster




• Discriminating variables from hit layers and strips - measure of width of cluster

• Strip multiplicity must be determined from data




– Discriminating variables from the Swimmer, interaction lengths

– In principle, PID with swimmer can accommodate dead or inefficient RPC chambers




Muon identification using theta and momentum dependent cuts on difference between penetration depth and expected penetration

Muon Efficiency ~ 80%

Pion Misidentification < 5%



Muon Identification Muon Identification (7)(7)• J/ reconstruction from Muon Pairs

• = expected - measured interaction lengths

• Apply a tight cut on the track with the smaller Apply a loose cut on the other track

• Tight cut– < 0.9 for p 1 GeV/c

– < for p < 1 GeV/c

• Loose cut (parametric)– same as tight but substitute 0.9 with 0.9 × s where 2 < s < 5

0.90.5GeV1GeV

0.5GeVp

Guglielmo De Nardo, with permission




0.90.5GeV1GeV

0.5GeVp

Guglielmo De Nardo, with permission

Signal = 134Background = 42M=3.08 GeV = 17 MeV


Two different values of the scale paramenter s s=5 (left) and s=2 (right)





• IFR working OK, muon identification possible, robust, well characterized on real data early 2000

• Teething troubles addressed, or have already been addressed– Need to fully understand loss in RPC efficiencies and

rectify

• Work on Neutral Detection started relatively recently

ConclusionsConclusions

Documents

The Instrumented Flux Return