Study of Low Mass Dimuon Production in 400 GeV Proton-Nucleus Collisions at the CERN-SPS with the NA60 Apparatus

2 1. Physics Motivation
1.1 Main Results found by NA60 in In-In Colli-
sions
Low Mass Region
The NA60 experiment has accumulated in the dimuon low mass region (below
∼ 1.2 GeV/c
2
) a statistics ∼ 1000 times larger than the one collected by the ex-
periment CERES having measured di-electrons. This allowed, for the first time, a
very accurate study of the properties of the ρ meson in nuclear collisions. Two main
models were developed in the 90s, predicting the behaviour of the ρ in the hot and
dense matter produced in heavy-ion collisions. The first model directly links the
mass of the ρ to the restoring of chiral symmetry, predicting a lowering of the mass
pole by more than 100 MeV in the hot and dense medium [Bro91]. The second was
a hadronic inspired model which predicted a strong broadening due to the coupling
of the ρ to baryons and due to in-medium modifications of pions, but essentially no
mass shift [Rap00]. The NA60 data clearly ruled out the scaling scenario [NA60].
Intermediate Mass Region
The dimuon mass spectrum between the φ and J/ψ resonances is dominated by
Drell-Yan and simultaneous semi-leptonic decays of D mesons. The superposition
of these two sources describes the measurements done in proton-nucleus collisions,
while in nucleus-nucleus collisions the dimuon mass spectrum shows an excess which
increases with the number of nucleons participating in the interaction [NA50]. Two
interpretations of this excess have been considered: it can be due to an unexpected
enhancement of open charm production or to thermal dimuons emitted from the
deconfined phase. The NA60 In-In data clearly showed that open charm is not
enhanced. The observed excess, whose yield is compatible with the NA50 measure-
ment, is prompt in nature [Sha06]. Further analysis of the p
T
spectra has shown
that the excess could be the first direct evidence of thermal radiation produced by
a partonic phase [NA60].
High Mass Region
The NA38 and NA50 experiments studied J/ψ suppression, as a signature of the
formation of a deconfined state. In Pb-Pb collisions the J/ψ production pattern, as
a function of the collision centrality, shows that above a certain centrality threshold
the J/ψ yield is considerably lower than expected from the “nuclear absorption”
curve, derived from proton-nucleus and light-ion collisions [NA50]. One of the cur-
rent interpretations of this result is that the dense and hot medium formed in the
collisions dissolves the χ
c
resonance, leading to the disappearance of the fraction
(∼ 30%) of J/ψ mesons that would otherwise originate from χ
c
decays. The NA60
data have shown that the J/ψ is suppressed also in In-In collisions with a pattern
1.2. Motivations and Expectations for the NA60 Proton-Nucleus Data 3
compatible with the NA50 observations. Furthermore, the accuracy of the measure-
ment allowed to falsify several models which were tuned on the NA50 Pb-Pb data
[NA60].
1.2 Motivations and Expectations for the NA60
Proton-Nucleus Data
Besides In-In data, NA60 also collected a large data sample in proton-nucleus colli-
sions. One important motivation is the need to establish a robust reference baseline
for measurements in heavy ion collisions by studying systems where no new physics
is expected. Several examples can be provided; before claiming that J/ψ suppression
is linked to new physics, for instance, one should make sure that other conventional
mechanisms, at work also in light systems, are accounted for.
Nevertheless, there is also an independent physics interest in in proton-nucleus
collisions. In the low mass region, measurements performed in p-A interactions
claimed that the properties of the ρ meson cold be modified also in “cold” nuclear
matter [Oza01]. Other experiments found no evidence of such effects; it is hence
important to accurately determine the line-shape in this environment. Another
interesting effect involving the ρmeson is the ρ-ω interference, observed, for instance,
in e
+
e
−
collisions [Lan85].
Another key study is the production of strangeness. In heavy ion collisions an
enhancement of strangeness production was considered for long time a signature
for the production of a deconfined state. For this reason, it is important to assess
whether strangeness is enhanced already in proton-nucleus collisions.
This thesis reports on the data collected during three days within the 2004 proton
run (amounting to ∼ 3.8 × 10
6
dimuon triggers), with a 400 GeV proton beam
incident on several nuclear targets. From the physics point of view, the work was
focused on the production of the φ meson, and, in particular, on its transverse
momentum distribution.
The document is organized in the following way. After a description of the NA60
apparatus (Chapter 2), the run and burst selection will be considered (Chapter 3),
followed by a detailed description of the event analysis (Chapter 4) which allowed
us to get a “final sample” on which the physics studies could be based. After
presenting a short description of the expected dimuon sources in the low mass region
(Chapter 5) we will study the nuclear dependence of the p
T
spectrum of the φmeson,
combining the real data measured by the experiment with the results of the MC
simulations.

2The NA60 Apparatus
2.1 General Overview
The purpose of the NA60 apparatus is to accurately study dimuon production in
proton-nucleus and heavy ion collisions. The produced dimuons are identified by
the muon spectrometer, composed of a hadron absorber that lets only muons pass.
The stopping of hadrons (provided by more than 20 interactions lengths) allows us
to select the highly rare dimuon events. In this way, and by implementing a high
selective trigger, we can run at very high luminosities (roughly speaking, only one out
of 100 thousand collisions are recorded). On the other hand, the material which stops
the hadrons also induces multiple scattering and energy loss on the muons, degrading
the mass resolution of the measurement made in the spectrometer. To overcome this
problem, NA60 already measures the muons before the absorber. This requires that
the muon tracks are found among the many other charged particle tracks, by a
correct matching with the reconstructed tracks in the muon spectrometer. For this
reason, the angles and momenta of charged particles must be known in the vertex
region with sufficient accuracy, requiring a magnetic field in the target region. The
particle tracking in the vertex region is performed using a tracking telescope, made
essentially of pixel and a few microstrip silicon detectors.
Figure 2.1 shows an overall representation of the NA60 apparatus. Note that
the term “muon spectrometer” designates the detector system after the hadron
absorber, including the trigger hodoscopes, while we speak of the “vertex tracking
telescope” when we mean the tracking elements in the vertex region. In the following
we will use the expressions “Jura” and “Sale`ve” sides, referring to the left and the
right sides of the experiment with respect to the central vertical plane, when seen
from the beam line looking downstream.
6 2. The NA60 Apparatus
The following sections describe the motivation, the design and performance of the
detector components which are important for the analysis presented in this thesis.
After a short overview of the beam, we will start with the muon spectrometer, the
most important detector in this experiment. The subsequent sections then describe
each detector in the sequence in which they are assembled along the beam direction.
MWPCs
MWPCs
trigger hodoscopes 
trigger hodoscopes
  R1 
  P1 
  R2 
toroidal magnet ACM
hadron absorber
PT7 magnet
beam tracker
+ targets 
1 m
 beam 
vertex
spectrometer
10 cm
fake ZDC +
beam dump 
 beam 
dipole field    2.5 T
 beam tracker 
 targets 
 VT tracker 
pre-
 absorber 
R3
R4
P2 
iron
 wall 
Figure 2.1: Overall representation of the NA60 apparatus (top) and detail of the vertex
region (bottom).
2.2 The Beam
NA60 collected data with proton and ion beams provided by the CERN SPS acceler-
ator. Ion beams were used in 2002 (test run with a Pb beam of 30 GeV per nucleon
followed by 5 days with 20 GeV per nucleon) and in 2003 (In beam of 158 GeV per
nucleon incident on In targets), while a 400 GeV proton beam was used both in
2002 (incident on Be, In and Pb targets) and − together with a 158 GeV proton
beam − in 2004 (incident on Be, Al, Cu, In, W, Pb and U targets).
At the CERN SPS particles do not circulate in continuous mode, but rather
they are delivered during well defined time intervals called “bursts”, separated by
“interburts”. In the 2002 and 2004 proton runs, the bursts were ∼ 4.8 s long,
separated by interburts lasting ∼ 12 s [Mem05]. The intensity of the proton beam
was different in 2002 and 2004 proton runs: in the former the beam intensity was
2.3. The Muon Spectrometer and Trigger System 7
 2 ·10
8
protons per burst, while in the latter the beam intensity was approximately
ten times higher, amounting to  2 ·10
9
protons per burst, in order to gain as much
statistics as possible to allow a study of the rare χ
c
meson.
2.3 The Muon Spectrometer and Trigger System
The purpose of the muon spectrometer is to identify, trigger on, and reconstruct the
muon pairs produced as a result of the beam-target interactions. In order to achieve
this, it consists of four main elements [And84]: the hadron absorber, eight tracking
multi-wire proportional chambers (MWPC), four trigger scintillator hodoscopes (R1-
R4), the last one placed behind a 120 cm thick iron wall, and an Air Core toroidal
Magnet (ACM) for the momentum measurement of the muons.
These components can be moved along the z-axis to ease maintenance and to
keep the angular muon acceptance around mid-rapidity, in spite of changes in the
beam energy. In the current setup, the angular muon acceptance lies approximately
in the range 35–120 mrad (from 2
◦
to 7
◦
), as imposed by the magnet aperture. For
beam energies of 400 GeV this corresponds roughly to one unit of rapidity at mid-
rapidity, where particle production is most copious. The muon’s acceptance also
depends on the magnetic field settings and on the effective number of interaction
lengths of the absorber; it is also affected by the trigger conditions.
2.3.1 The Hadron Absorber
The muons are filtered out among the many other produced particles by the hadron
absorber. This is a simple, but effective, “particle identification” system: particles
that hit the R4 trigger hodoscope (see Figure 2.1), by definition, are muons. Indeed,
apart from neutrinos, only muons are penetrating enough to cross all the matter of
which the absorber is made up:
• the pre-absorber, made of 41 cm BeO and 25.4 cm Al
2
O
3
• the main absorber, made of 400 cm of graphite followed by 80 cm of iron
1
• the 120 cm thick iron wall, placed after the muon chambers so as to not degrade
the tracking accuracy through multiple scattering, while ensuring a very clean
muon trigger
The hadron absorber starts as close as possible to the target region, immediately
after the vertex telescope, in order to stop a large fraction of pions and kaons from
decaying into muons and becoming a source of background.
1
The composition of the main absorber sligthly changed during the 2004 proton run, see
Chapter 3; here, we refer to the composition in the period of interest for this thesis.
8 2. The NA60 Apparatus
The main absorber, placed between the target and the muon chambers, is made
of materials with a low atomic number Z, in order to minimize the multiple scattering
induced on the traversing muons, and with the highest available densities, so as to
stop the hadrons in a relatively small thickness. The non-interacting protons are
stopped in an “Uranium Plug”, placed inside the absorber, aligned with the beam
axis outside of the muon’s acceptance window. The carbon blocks are surrounded by
cast iron and concrete. Table 2.1 gives an overview of the position and thicknesses
of the elements of the absorber for the data taking period relevant for this thesis
(see Chapter 3).
Placing the last part of the absorber − the 120 cm thick iron wall − after the
tracking stations and before the last trigger hodoscope, ensures that no energetic
punch-through hadrons give rise to a fake trigger, without contributing to the degra-
dation of the tracks measured in the chambers.
Material ρ [g/cm
3
] z
in
[cm] ∆z [cm] λ
I
[cm] ∆z/λ
I
BeO 2.81 43.6 41.0 35.85 1.14
Al
2
O
3
3.52 84.6 25.4 32.65 0.78
C 1.93 110.0 400.0 44.70 8.95
Fe 7.87 510.0 80.0 16.76 4.77
Total number of interaction lengths in main absorber 15.64
Fe (iron wall) 7.87 1676.3 120.0 16.76 7.16
Total number of interaction lengths 22.8
Table 2.1: Composition of the hadron absorber for the data taking period relevant for
this thesis (see Chapter 3) during the 2004 proton run. λ
I
is the nuclear interaction
length.
2.3.2 The Multi-Wire Proportional Chambers
The muons which have crossed the main absorber are tracked in eight multi-wire
proportional chambers, separated into two sets of 4 chambers by the toroidal magnet
ACM (Air Core Magnet).
Each muon chamber consists of three independent tracking planes, interspaced
by 2.2 cm, rotated by 60
◦
with respect to each other, to allow a good measurement
of one space point (see Figure 2.2).
The sensing elements of these chambers are gold-plated tungsten anode wires
with a diameter of ∼ 20 µm, inter-spaced by 3 mm and sandwiched between two
graphited Mylar cathode planes, 6 mm far away. The overall read-out gate is ∼ 80 ns
2.3. The Muon Spectrometer and Trigger System 9
Figure 2.2: Each MWPC consists of three independent tracking planes. For visibility
purposes the individual planes are shown well separated from one another.
[And84]. The chambers have hexagonal shape and their transverse size increases
with increasing distance from the target to cover the angular acceptance, defined
by the aperture of the ACM magnet. The z-positions with respect to the target
and the transverse size of all components of the muon spectrometer can be found in
Table 2.2.
The tracking volume of the chambers is filled with a gas mixture, consisting of
∼ 80% of Argon (of which 50% is flushed through 0.8% of Isopropyl alcohol), of
18% Isobutane iC
4
H
10
used as a quencher, and of 2% Tetrafluorethane used as a
“cleaning” gas. A gas mixer rack measures and controls the flow of the gas compo-
nents. Due to the fact that the gas mixture had to be changed in 2002 with respect
to the previously used “magic gas”, the chambers showed visible “ageing effects”;
in particular, 5 planes out of 24 were not working during the data taking period rel-
evant to this thesis [Mem05], see Chapter 3: the track reconstruction could still be
performed using the remaining working planes, with a sligthly degraded efficiency.
It is worth noting that − apart from chamber 1, which was completely rebuilt in
1994 − all MWPCs work since 1980, when NA10 (the predecessor experiment of
NA38, NA50 and NA60) began to operate.
2.3.3 The Trigger Hodoscopes
The trigger system consists of four “R” hodoscopes, two before and two after the
ACM magnet, made of scintillator slabs. Like all other components of the muon
spectrometer, the hodoscopes have hexagonal shape, given by the geometry of the
10 2. The NA60 Apparatus
ACM magnet. All R-hodoscopes are designed in a similar way. The scintillator slabs
of each sextant, oriented parallel to the outer edges, become longer with increasing
distance from the beam axis (see Figure 2.3).
sextant 1
sextant 2
s
e
x
t
a
n
t
5
s
e
x
t
a
n
t
6
s
e
x
t
a
n
t
3
s
e
x
t
a
n
t
4
sextant 1
sextant 2
s
e
x
t
a
n
t
4
s
e
x
t
a
n
t
3
s
e
x
t
a
n
t
6
s
e
x
t
a
n
t
5
Saleve Jura
Figure 2.3: The segmented slabs of the R3 (left) and P (right) hodoscopes.
The width of the slabs of the R1 and R2 hodoscopes increases with the distance
from the beam line so that a muon produced in the target which passes through slab
i in R1 will also hit slab i in R2. In order to accomodate the longitudinal spatial
extent of the target region (of the order of ∼ 1 m in the NA10 experiment, which
designated the trigger logic) and to allow for multiple scattering for low energetic
muons, the combination of hitting slab i in R1 and i − 1 in R2 is also allowed.
This “R1-R2 coincidence” for each muon is combined with the information from R3
and R4. The last trigger hodoscope, R4, is placed behind the 120 cm thick iron
wall, which absorbs remaining hadrons thereby ensuring a clean (di)muon trigger;
however, this implies that the muons must have a minimal momentum of ∼ 7 GeV/c
to give a signal in R4. The dimuon trigger also requires that the two muons pass
through two different sextants: this requirement reduces the fraction of low mass
muon pairs that give rise to a trigger, in order to not saturate the bandwidth of the
data acquisition system. The overall read-out gate of the R-hodoscopes is 20 ns.
The R3 hodoscope has a small inactive zone on the “Jura” side, called “Beam-
Killer”, as can be seen in Figure 2.3. Although the other trigger hodoscopes have
no such dead areas, the muon acceptance is affected by this cut in sextants 4 and 5,
since the trigger requires a hit in all trigger stations.
The P-hodoscopes
Furthermore, NA60 uses two so-called “P-hodoscopes”, P1 and P2. The former
is placed before the ACM magnet, the latter after the iron wall. They are used
in special runs to measure the efficiency of the R1-R4 system. Their geometry is
different from the R-hodoscopes; the scintillator slabs of each sextant are oriented
2.3. The Muon Spectrometer and Trigger System 11
radially, so that their width increases with increasing distance from the beam axis
(see Figure 2.3).
In the 2004 proton run an additional feature was implemented: the measure of
the arrival time of each muon at the P2 hodoscope; this (see Chapter 4) improves our
ability of rejecting trigger coincidences of 2 muons that were produced in different
events. This possibility improves significantly our event selection, needed because
of the very high luminosities.
Element z [cm] Main Characteristics
MWPC 1 615.8 448 wires per plane, ∅ = 134 cm
Hod. R1 629.6 6× 30 scintillator slabs of 1.05–3.45 cm width
MWPC 2 684.1 512 wires per plane, ∅ = 153 cm
Hod. P1 712.0 6× 8 scintillator elements; width: 2.73–13.65 cm
MWPC 3 748.7 576 wires per plane, ∅ = 172 cm
Hod. R2 761.0 6× 30 scintillator slabs of 1.25–3.35 cm width
MWPC 4 818.2 640 wires per plane, ∅ = 192 cm
ACM 828.7 – 1311.7
MWPC 5 1347.2 1024 wires per plane, ∅ = 306 cm
Hod. R3 1390.2 6× 23 scintillator slabs of 5.5 cm width
MWPC 6 1445.6 1088 wires per plane, ∅ = 326 cm
MWPC 7 1544.1 1152 wires per plane, ∅ = 345 cm
MWPC 8 1642.1 1216 wires per plane, ∅ = 364 cm
Iron wall 1676.3 – 1796.3
Hod. R4 1800.7 6× 32 scintillator slabs of 5.5 cm width
Hod. P2 1820.7 6× 8 scintillator elements; width: 8.10–47.50 cm
Table 2.2: Detector components of the muon spectrometer. Except for the ACM and
iron wall, the z values given refer to the centre of the respective device.
2.3.4 The Toroidal Magnet ACM
The magnetic field of the toroidal magnet ACM is produced between 6 radial iron
poles, which are 4 m long and cover 18
◦
in azimuth. The magnet’s air gap starts at
a radius of 29.5 cm, while the outer radius is 154 cm. These two values are the ones
determining the detector’s rapidity acceptance. Events with muons which cross one