Charm and intermediate mass dimuons in In+In collisions R. Shahoyan, IST (Lisbon) on behalf of the NA60 collaboration Quark Matter 2005, Budapest Motivation (NA38/NA50 results) NA60 concept and data analysis Intermediate mass region (IMR) analysis (preliminary results) Summary and outlook
32
Embed
Charm and intermediate mass dimuons in In+In collisions R. Shahoyan, IST (Lisbon) on behalf of the NA60 collaboration Quark Matter 2005, Budapest Motivation.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Charm and intermediate mass dimuons in In+In collisions
R. Shahoyan, IST (Lisbon)on behalf of the NA60 collaboration
Quark Matter 2005, Budapest
Motivation (NA38/NA50 results)
NA60 concept and data analysis
Intermediate mass region (IMR) analysis
(preliminary results)
Summary and outlook
2
NA38/NA50 was able to describe the IMR dimuon spectra in p-A collisions as the sum of Drell-Yan and Open Charm contributions
However, the yield needed to describe the NA38/NA50 spectra (with PYTHIA’s kinematical distributions, after B.R., acceptances, in the window
The yield of intermediate mass dimuons seen in heavy-ion collisions (S-U, Pb-Pb)exceeds the sum of DY and Open Charm decays, extrapolated from the p-A data
peripheralcollisions
centralcollisions
IMR dimuons in heavy-ion collisions: the excess
M (GeV/c2)M (GeV/c2)
4
The intermediate mass dimuon yields in heavy-ion collisions can be reproduced: by scaling up the Open Charm contribution by up to a factor of 3 (!) or by adding thermal radiation
Thermal dimuon production or charm enhancement?
The data collected by NA38/NA50 cannot distinguish among these two alternatives. We need to measure secondary vertices with ~ 50 m precision to separate prompt dimuons from D meson decays
5
hadron absorber
and trackingmuon trigger
magnetic field
iron w
all
muonother
Concept of NA60
targets
Concept of NA60: place a silicon tracking telescope in the vertex region to measure the muons before they suffer multiple scattering in the absorberand match them to the muon measured in the spectrometer
Improved kinematics (~20 MeV/c2 at instead of 80 MeV/c2 in NA50)Origin of muons can be accurately determined
2.5 T dipole magnet
beam tracker vertex tracker
6Muon Matching
Muons from the Muon Spectrometer are matched to the Vertex Telescope tracks by comparing the slopes and momenta.
Each candidate passing a matching 2 cut is refitted using both track and muon measurements, to improve kinematics.
Most background muons from and K decays are rejected in this matching step… but
a muon might be matched to an alien track (or to its proper track which picked too many wrong clusters) Fake matches, additional source of background
By varying the cut on the matching 2 we can improve the signal/background ratio
7Vertex resolution (along the beam axis)
Beam Trackersensors
windows
Good target identification even for the most peripheral collisions ( 4 tracks)
The interaction vertex is identified with better than 200 m accuracy along the beam axis
8Vertex resolution (in the transverse plane)
The interaction vertex is identified witha resolution of 10–20 m accuracy in the transverse plane
Dispersion between beam track andVT vertex
Vertex resolution (assuming BT=20 m)
10
20
30
0
(
m)
Number of tracks
Beam Tracker measurement vs. vertex reconstructed with Vertex Telescope
BTBT
The BT measurement (with 20 m resolution at the target) allows us to control the vertexing resolution and systematics
9
J/
Using the muons from J/ decays (no background, from the interaction point) we determine the resolution of the impact parameter of the track at the vertex (offset) : 40–50 m
The non-Gaussian tails are caused by imperfect alignment (to be improved)
Offset resolution
10
Good enough to separate prompt dimuons from Open Charm off-target decays !
vertex impact < c (D+ : 312 m, Do : 123 m)
To eliminate the momentum dependence of the offset resolution, we use the offset
weighted by the error matrix of the fit:
for single muons
and for dimuons
2/)2( 11212
xyyyxx VyxVyVx
2/)( 22
21
Offset resolution
J/
Weighted Offset () 100
Offs
et r
esol
utio
n (
m)
11Background Subtraction: method
Our measured dimuon spectra consist of:
correctly matched signal signal muons from the spectrometer are associated with their tracks in the Ver.Tel.
wrongly matched signal (fakes) at least one of the muons is matched to an alien track
correctly matched combinatorial pairs muons from ,K decays are associated with their tracks or with the tracks of their parent mesons
association between the ,K decay muon and an alien track
All these types of backgroundare subtracted by
Event Mixingin narrow bins in centrality, for each target,
and magnetic field polarities (~6000 samples)
wrongly matched combinatorials (fakes)
12Background Subtraction: method
Combinatorial Background (mainly from uncorrelated and K decays)
Subtracted by building a sample of pairs using muons from different Like Sign events.
Mixing procedure accounts for correlations in the data due to the dimuon trigger.
CBmixing
13Background Subtraction: method
CBmixing
Subtracting the Mixed CB from the data we obtain the Signal (correct and fake) in +-
sample and zero (or residual background) in the Like Sign dimuons sample.
14Background Subtraction: method
CBmixing
The Fake Matches Background is subtracted by Monte Carlo (used for the Low Mass Analysis) or by matching the muons from one event to tracks from another one; a special weighting procedure is used to account for the mixed fake matches…
Fakesmixing
15Background Subtraction: method
CBmixing
Fakesmixing
In order to extract from the fake matches the signal contribution we repeatthe Combinatorial Mixing procedure for the generated fakes sample, obtaining the combinatorial fake matches
FakesCB
mixing
16Background Subtraction: method
CBmixing
Fakesmixing Fakes
CBmixing
Subtracting the combinatorial fakes from all fakes we obtain the fake signal
17Background Subtraction: method
CBmixing
Fakesmixing Fakes
CBmixing
Subtracting the fake signal from the total matched signal leads to the correct signal spectrum
18
The “mixed” background sample (fake matches and combinatorial) must reproduce the offsets of the measured events: therefore, the offsets of the single muons (from different events) selected for mixing must be replicated in the “mixed” event.
mixed eventevent 1
event 2
Background Subtraction: method (offsets)
(All masses)
19
The quality of combinatorial subtraction can be controlled by comparing the built mixed event Like Sign dimuon spectra to the corresponding measured data.
Background Subtraction: accuracy
20
The Like-Sign spectra should be similar to the background on the OS dimuon spectrum use residual LS background as an estimate of the unsubtracted OS background
It is accounted as a systematical error: the errors on the background are globally scaled upto guarantee that the residual LS background is zero within 3 standard deviations
Because of the high background level, a ~1% error in the background estimate leads to ~10% systematical error on the extracted signal
Accounting for residual background
21Background Subtraction: resulting mass distribution
Data integrated over centrality
Matching 2 < 1.5
SignalTotal Background
22Background Subtraction: resulting offset distribution
Signal Fake Matches
Dimuon weighted offsets
1.2 < M < 2.7 GeV/c2
0 < yCM < 1|cos| < 0.5
Kinematical domain where the analysis is performed:
23Offset distributions of the expected sources
To account for residual misalignments in the real data, the offsets of the reconstructed MC muons were smeared until they reproduce the weighted offsets measured for J/ muons.
Prompt contribution: use an average of the J/ and measured offset distributions
Open Charm contribution: use the MC distribution, after smearing
Dimuon weighted offsets
24NA60 Signal analysis: simulated sources
Charm and Drell-Yan contributions are calculated by overlaying Pythia events on real data
(using CTEQ6M PDFs with EKS98 nuclear modifications and mc=1.3 GeV/c2)
The fake matches in the MC events are subtracted as in the real data
Absolute normalization:
The expected DY contribution, as a function of the collision centrality, is obtained from the number of observed J/ events and the suppression pattern; see talk by Roberta Arnaldi
A 10% systematical error is assigned to this normalization
Relative normalizations:
for DY: K-factor of 1.8; to reproduce DY cross-sections of NA3 and NA50
for charm: we use two options for the expected cross-section:
a) 6.3 b/nucleon: suggested by a “world average” of direct charm measurements
b) a factor 2 higher: needed to reproduce the NA50 p-A dimuon data 450 GeV
The fits to mass and weighted offset spectra are reported in terms ofthe DY and Open Charm scaling factors needed to describe the data
25Is there an excess in In-In collisions?
Fix the Charm and DY contributions to the expected yields,and see if their Sum describes the measured Data
Answer: Yes, an excess is clearly present !
(Even if we use the higher charm yield)
“world average” “NA50 p-A ”
26Is it compatible with the NA50 observation?
Can we describe the measured mass spectrum by leavingthe Charm normalization as a free parameter?
NA50 could, with up to a factor of 3 Charm enhancement in central Pb-Pb collisions…
~ 2 in terms of NA50p-A normalization
Answer: Yes, leaving the Charm contribution free describes the In-In data, with a “charm enhancement” factor around 2 in “NA50 units”
(but with a poor 2)
27Is this validated by the offsets information?
Fix the prompt contribution to the expected DY,and see if we can describe the offset distribution with an enhanced Charm yield
Dimuon weighted offsets
Answer: No, the fit fails: Charm is too flat to describe the remaining spectrum…
we need more prompts
28How many more prompts do we need?
Dimuon weighted offsets
Leave both contributions free,and see if we can describe the offset distribution
Answer: Two times more prompts than the expected Drell-Yan provides a good fit
(and the Charm yield is as expected from the NA50 p-A dimuon data)
29Is the prompt yield sensitive to the Charm level?
Fix the Charm contribution to either of the two references,and see how the level of prompts changes
Answer: No, both options require two times more prompts than the expected Drell-Yan !
(the Charm contribution is too small to make a difference)
Dimuon weighted offsets
“world average” “NA50 p-A ”
30Mass shape of the excess with respect to DY (or Charm)
The mass spectrum of the excess dimuons is steeper than DY(and flatter than Open Charm)
Fix the DY and Charm contributions to expected yields
31
Relative excess:(Data – Sources) / Sources
(Data – Sources) / Nparticipants
Faster than linear increase with Nparticipants
Centrality dependence of the Excess = [Data - Sources]
very
preliminary
1.There is an excess of intermediate mass dimuons in Indium-Indium collisions
2.The offset distribution requires a factor 2 more prompts than expected from DY The excess is not due to open charm enhancement
3.The excess grows faster than linearly with the number of participants
Results are very robust with respect to variations of the matching 2 cut: changing the Signal / Background ratio by a factor of 2 changes the results by less than 10% The excess cannot be due to a bias in the background subtraction
For the moment, our offset distribution cannot discriminate between the two expected charm yields (which differ by a factor of two)
Reprocess already analyzed data after improving the detector’s alignment
Explore full Indium-Indium statistics (~ 50% of the data not yet reconstructed)