Beauty Production at HERAB

1
Beauty Production at HERA-B

• Performance of the Second Level Trigger
• And bb Cross Section in 920 GeV pN Collisions

Ph.D. thesis by Brian Aagaard Petersen
June 4, 2002
2
Outline

• Introduction
• The HERA-B experiment
• The trigger system
• The trigger performance
• Beauty production cross section
• Conclusions

3
Standard Model
4
b Hadron Decays
b hadrons decay weakly
s
c
Hundreds of different final states available
W
b
c
B0
d
d
5
CP Violation

• Main goal of b physics is to measure CP violation

What is CP?
CP is a symmetry in QED and QCD
CP violation seen in weak decays of s quarks
6
CP Violation in Beauty
CP violation predicted in many channels
Two decay paths
7
New Beauty Experiments
Three experiments built to measure CP in b system

• BaBar and Belle
• Located at SLAC, USA and KEK, Japan
• Asymmetric ee- colliders on ?(4s) resonance
• ?(4s)?B0B0 clean event topology
• Measured sin2ß0.78?0.08
• HERA-B
• Fixed target exp. at DESY, Germany
• Use 920 GeV proton beam on different nuclei
• Only 1 in 106 interactions contain b hadrons!
• Golden decays in 10-11 events. Need high IA rate

8
The HERA-B Experiment
9
The HERA-B Experiment

• Vertex detector
• 8 superlayers of silicon microstrip detectors
  (50 mm pitch)
• Provides high resolution vertexing (500 mm in z,
  50 mm in x/y)

10
The HERA-B Experiment

• Magnet
• Conventional warm dipole magnet with field
  integral of 2.2 Tm
• Equipped with several tracking chambers (Removed
  in 2002!)

11
The HERA-B Experiment

• Main tracking system
• Divided into two systems with different
  segmentation
• MSGC and Honeycomb drift chambers (90 and 300
  mm resol.)

12
The HERA-B Experiment

• Hadron identification
• Ring Imaging Cherenkov detector (C4F10 radiator
  gas)
• Provides p, K and P separation for 3 GeV/c lt p lt
  60 GeV/c

13
The HERA-B Experiment

• Charged lepton identification
• Electrons Electromagnetic calorimeter
• Muons 4 tracking stations between hadron
  absorbers

14
A Golden Decay at HERA-B
Topology for a golden event

• Unfortunately B hadrons are not produced alone

15
Typical HERA-B Event

• At design rate average of five interactions per
  event

Need a trigger system to filter out interesting
events
16
The Trigger System

• Requirements
• Handle a new event every 96 ns
• Reduce event rate with O(105)
• Have high efficiency for signal events (J/y
  events)

• Solution
• Four levels of triggers
• Buffer events during processing
• Events are processed in parallel
• Main idea is to track leptons with high momentum
  through the complete detector
• Tracking only done in Region of Interest (RoI)

17
The HERA-B Trigger System

• Pretriggers
• Searches for electron and muon candidates
• Used to define RoIs for the first level trigger

18
The HERA-B Trigger System

• First Level Trigger (FLT)
• Searches for hits in RoIs after the magnet (
  Kalman Filter)
• Can apply mass and pt cuts on pairs of found
  tracks

19
The HERA-B Trigger System

• Second Level Trigger (SLT)
• Confirms and refines the FLT tracks (RoI based)
• Prolongs tracks through the magnet and the VDS
  to the target

20
The SLT Algorithm

• Slicer
• Requires hits in RoI behind magnet (ghost
  removal)
• RefitX/RefitY
• Refits track parameters from FLT (Kalman Filter)
• L2Magnet
• Parameterizes bending in magnet.
• L2Sili
• Finds tracks and calculates parameters in the VDS
• L2Vertex
• Requires track pairs to come from a common vertex

21
Trigger Hardware

• Pretriggers and FLT
• Uses custom-made hardware
• Events pipelined on detector
• Specialized data transmission
• SLT
• Processing on PC farm
• Buffer and switch made from DSP processors
• TLT runs on the same farm
• Fourth level trigger (4LT)
• Uses a second PC farm
• Does full event reconstruction
• Does not reject any events

22
SLT Performance

• The key performance parameters
• Efficiency How many signal events is written to
  tape?
• Trigger rates Do we get enough suppression?
• Timing Is the algorithm fast enough?
• Ghost rate How many of the tracks are real
  leptons?
• Track precision How well do reconstruct tracks?

Evaluated for 2000 and 2002 trigger configurations
23
The 2000 Trigger Config.

• The FLT was not ready for the 2000 run
• Use pretriggers to seed SLT directly instead
• Reduce trigger rate by hard ET cuts and
  downscaling
• Interaction rate was adjusted to about 5 MHz
• SLT first emulates part of the FLT (muon
  tracking)
• The SLT tracking algorithms has to use large RoIs
• No common vertex requirement is used

• Reduced detector performance
• ITR not available for the SLT (30 less
  acceptance)
• OTR, ECAL and MUON systems plagued by dead and
  noisy channels. Efficiencies not very well-known

24
J/y Trigger Efficiency

• The main losses are due to detector performance

25
Muon Tracking Efficiency

• The muon tracking requires hits in all 4
  chambers

Pretriggers
SLT
Pretrigger handles last two layers SLT handles
first two layers
p
m
The problem is the first layer! The SLT here
requires a double hit
Chambers
Absorber
m
Single cell efficiency 85 Double hit
efficiency (85)272 Two muons (72)252
m
26
J/y Rates in Data and MC
Run 16665

• MC underestimates signal!
• Muon detector efficiency is very uncertain

N(J/y)192
In electron channel, MC signal is 50 too high!
27
SLT 2000 Trigger Rates

• The lack of FLT gives more suppression in the SLT
• Suppression more than sufficient SLT limited at
  input
• MC statistics insufficient to evaluate full chain!

28
SLT 2000 Timing

• ee- channel needs 60 nodes
• mm- channel needs 40 nodes

Only 100 nodes busy
29
The 2002 Configuration

• Detector performance have been improved
• FLT and ITR are now available
• FLT pair mode efficiency very uncertain

• The 2002 baseline trigger
• FLT only required to find one lepton
• SLT finds second lepton based on pretriggers
  (as in 2000 configuration)
• Interaction rate of 5-10 MHz
• FLT output rate adjusted with pt cut

30
SLT Upgrades

• 300-450 MHz CPUs upgraded to 1200 MHz
• ITR included in Slicer/Refit algorithms
• Muon tracking rewritten for higher efficiency
• Refit code speeded up
• L2Magnet now use hits in a magnet chamber
• L2Sili rewritten for speed and precision
• L2Vertex is now used

Algorithm has not been tested on 2002
data Performance evaluation based on 1 million MC
events
31
FLT Trigger Rates
Rates as function of pt
SLT/DAQ is limited to 30 KHz input rate

• Muon channel OK with
• Ptgt0.7 GeV at 5 MHz
• Ptgt1.4 GeV at 10 MHz

Electron rate is too high! Require bremsstrahlung
32
Bremsstrahlung

• Electrons emit bremsstrahlung in material

e?
Dx
e?
g
Magnet
ECAL
Bremsstrahlung visible as second cluster Very
clean signature of electrons! ECAL pretrigger can
detect these FLT requires the electrons to have
bremsstrahlung Lowers FLT trigger rate by 2
orders of magnitude
33
SLT Trigger Rates
Maximum reconstruction rate in 4LT 50 Hz
Can reduce rate by 50-75 with a 2.0 GeV mass cut
Output rate looks OK, but at the limit!
34
SLT Timing for muons
Only 40-65 nodes needed
More than sufficient CPU power available!
35
J/y Trigger Efficiency
Electron efficiency low due to bremsstrahlung
All SLT components at 90 or above Very
difficult to improve up on
36
Efficiency Dependency on pt
10 MHz cut
5 MHz cut

• At 10 MHz (ptgt1.4 GeV)
• J/y?mm- efficiency drops 30
• J/y?ee- efficiency drops 50

37
Predicted 2002 J/y Rates

• Expected rates per hour

After 2000 hours of running at 10 MHz
s
Still have to correct for detector performance
38
Physics Program 2002
J/y yield too low for CP violation measurement

• New HERA-B physics program
• A dependence of charmonium production
• Measure production rate of J/y, y, cc and ?
• as a function of xF and target material
• Beauty production cross section
• Make a precise measurement of the cross section
• using beauty hadron decays into J/y mesons

39

• The bb Cross Section
• Using the 2000 data

40
The bb Cross Section

• Beauty production a test of QCD

Large theory uncertainties Contradictory
experiments
More measurements needed!
41
The Method
Use b ? J/yX decays with J/y?mm- and J/y?ee-
Luminosity and efficiency poorly known
Measure relative to prompt J/y signal

• Procedure
• Count J/y events ( prompt J/y events, nP)
• Select detached J/y events to count b events (nB)
• Calculate s(bb) from the ratio nB/nP

42
J/y?mm- Data Sample

• The full muon triggered sample (100 hours)

43
Improved J/y Sample

• Reduce background
• Muon ID
• RICH ID (not K)
• Common vertex

S/B factor 4 better
nP2880 events
44
Detached Event Selection

• Use the long lifetime of b hadrons

Vertex at primary
45
The Detached Events

• A few detached J/y events observed downstream

Use unbinned likelihood fit of mass spectrum Free
exponential for bkg and fixed Gaussian for signal
Dzlt0
2.2
Fit result nB1.9 events
-1.5
Background mainly from semileptonic b and c decays
46
A Candidate Event
J/y candidate with two extra tracks
Primary vertex
B candidate is not fully reconstructed!
47
The J/y?ee- Channel

• More bkg in ee- channel
• Use hard PID to see J/y
• Not needed for detached!

Estimate without PID cuts nP7750520STAT280SYS
48
Detached J/y?ee- Selection

• Dzgt 0.5 cm
• e impact parameter to wire Iw gt 200 mm
• OR
• Closest track at wire more than 250 mm away

49
Combined Likelihood Fit

• Simultaneous fit of s(bb) in ee- and mm-
  channels

mm-
Combined fit
ee-
50
Systematics
Not reducible (14)
statistics limited
51
Result and Comparisons
HERA-B result s(bb)32 8sys nb
14
-12
HERA-B result is fully compatible with theory and
E771
Statistical precision still low
Need 2002 data
52
Summary

• HERA-B is a high rate, low S/B experiment
• SLT is flexible, but subject to detector perf.
• SLT algorithm has been improved for 2002 run
• Much higher J/y yields are now expected
• s(bb) has been measured from 2000 data

53
The Golden Decay
Golden event rate at HERA-B O(10-10-10-11)
54
The HERA-B Experiment

• Other systems
• Special tracking system for triggering on
  high-pt hadrons
• Transition radiation detector for e-p separation

55
Design Values