Centrality evolution of the charged-particle pseudorapidity density over a broad pseudorapidity range in Pb-Pb collisions at $\sqrt{s_{\rm NN}}$ = 2.76 TeV

The centrality dependence of the charged-particle pseudorapidity density measured with ALICE in Pb-Pb collisions at $\sqrt{s_{\rm NN}}$ over a broad pseudorapidity range is presented. This Letter extends the previous results reported by ALICE to more peripheral collisions. No strong change of the charged-particle pseudorapidity density distributions with centrality is observed, and when normalised to the number of participating nucleons in the collisions, the evolution over pseudorapidity with centrality is likewise small. The broad pseudorapidity range allows precise estimates of the total number of produced charged particles which we find to range from $162\pm22$ (syst.) to $17170\pm770$ (syst.) in 80-90% and 0-5 central collisions, respectively. The total charged-particle multiplicity is seen to approximately scale with the number of participating nucleons in the collision. This suggests that hard contributions to the charged-particle multiplicity are limited. The results are compared to models which describe $\mbox{d}N_{\mbox{ch}}/\mbox{d}\eta$ at mid-rapidity in the most central Pb-Pb collisions and it is found that these models do not capture all features of the distributions.


Introduction
The measurement of the charged-particle pseudorapidity (η) density distribution in heavy-ion collisions provides insight into the dominant particle production mechanisms, such as parton fragmentation [1] and the observed phenomenon of limiting fragmentation [2].The unique capability of ALICE to perform such measurements from large to small overlaps of the colliding nuclei over a broad pseudorapidity range allows for significant additional information to be extracted e.g., the total number of charged particles and the evolution of the distributions with centrality.
The charged-particle pseudorapidity density (dN ch /dη) per se does not provide immediate understanding of the particle production mechanism, but as a benchmark tool for comparing models it is indispensable.Various models [3][4][5] make different assumptions on how particles are produced in heavy-ion collisions resulting in very different charged-particle pseudorapidity density distributions -both in terms of scale and shape.Models may, for example, incorporate different schemes for the hadronisation of the produced quarks and gluons which leads to very different pseudorapidity distributions of the charged particles.
The ALICE collaboration has previously reported results on the charged-particle pseudorapidity density in the 0-30% most central Pb-Pb collisions at √ s NN = 2.76 TeV over a wide pseudorapidity range [6], and in the 80% most central collisions at mid-rapidity (η ≈ 0) only [7].The ATLAS collaboration has reported on the charged-particle pseudorapidity density in the 80% most central events in a limited pseudorapidity range of |η| < 2 [8].Similarly, the CMS collaboration has reported on the same measurements in the 90% most central events at η ≈ 0, and for selected centralities up to |η| < 2 [9].
In this Letter we present the primary charged-particle pseudorapidity density dependence on the event centrality from mid-central (30-40%) to peripheral (80-90%) collisions over a broad pseudorapidity range to complement results previously reported by ALICE in the 0-30% centrality range.Unlike previous [6], in the forward regions where the signal is dominated by secondary particles produced in the surrounding material, we use a data-driven correction to extract the primary charged-particle density.
Primary charged particles are defined as prompt charged particles produced in the collision, including their decay products, but excluding products of weak decays of muons and light flavour hadrons.Secondary charged particles are all other particles observed in the experiment e.g., particles produced through interactions with material and products of weak decays.
In the following section, the experimental set-up will be briefly described.Section 3 outlines analysis procedures and describes a data-driven method to isolate the number of primary charged particles from the secondary particle background at large pseudorapidity.Systematic uncertainties are discussed in Sect. 4. In Sect.5, the resultant charged-particle pseudorapidity density distributions are presented along with their evolution with centrality.Furthermore we extract from the measured dN ch /dη distributions the total number of charged particles as a function of the number of participating nucleons.We finally compare the measured charged-particle pseudorapidity density to a number of model predictions before concluding in Sect.6.

Experimental setup
A detailed description of ALICE can be found elsewhere [10,11].In the following we briefly describe the detectors relevant to this analysis.
The Silicon Pixel Detector (SPD) is the inner-most detector of ALICE.The SPD consists of two cylindrical layers of 9.8 × 10 6 silicon-pixels possessing binary read-out.It provides a measurement of charged particles over |η| < 2 using so-called tracklets -a combination of hits on each of the two layers (1 and 2) consistent with a track originating from the interaction point.Possible combinations of hits not consistent with primary particles can be removed from the analysis, with only a small (a few %) residual correction for secondary particles derived from simulations.The SPD also provides a measurement, by combining hits on its two layers, of the offset with respect to the interaction point, where the collisions occurred.IP = (0, 0, 0) is at the centre of the ALICE coordinate system, and IP z is the offset along the beam axis.Finally, a hardware logical or of hits in each of the two layers provides a trigger for ALICE.
The Forward Multiplicity Detector (FMD) is a silicon strip detector with 51 200 individual read-out channels recording the energy deposited by particles traversing the detector.It consists of three sub-detectors FMD1, 2, and 3, placed approximately 320 cm, 79 cm and −69 cm along the beam line, respectively.FMD1 consists of one inner type ring (1i), while both FMD2 and 3 consist of inner (2i,3i) and outer type rings (2o,3o).The rings have almost full coverage in azimuth (ϕ), and high granularity in the radial (η) direction (see Tab. 1).
The V0 is the most forward of the three detectors used in this analysis.It consists of two sub-detectors: V0-A and V0-C placed at approximately 333 cm and −90 cm along the beam line, respectively.Each of the sub-detectors are made up of scintillator tiles with a high timing resolution.While the V0 provides pulse-height measurements, the energy-loss resolution is not fine enough to do an independent charged particle measurement.In previous measurements, using so-called satellite-main collisions (see Sect. 3), one could match the V0 amplitude to the SPD measurements to obtain a relative measurement of the number of charged particles.However, for collisions at |IP z | < 15 cm no such matching is possible, and the V0 is therefore not used to provide a measurement of the number of charged particles in this analysis.The detector is used, in an inclusive logical or with the SPD, for triggering ALICE and to provide a measure of the event centrality [7].
Details on the coverage, resolution, and segmentation of the three used detectors are given in Tab. 1.  1: Overview of the resolution (δ ), segmentation (∆), and coverage of the detectors used in the analysis.The 'A' side corresponds to z > 0, while the 'C' side corresponds to z < 0. The η range is specified for collisions with IP z = 0.

Data sample and analysis method
The results presented in this paper are based on Pb-Pb collision data at √ s NN = 2.76 TeV taken by ALICE in 2010.About 100 000 events with a minimum bias trigger requirement [7] were analysed in the centrality range from 0% to 90%.The data was collected over roughly 30 minutes where the experimental conditions did not change.
The standard ALICE event selection [12] and centrality estimator based on the V0-amplitude are used in this analysis [13].We include here the 80-90% centrality class which was not present in the previous results [7].As discussed elsewhere [13], the 90-100% centrality class has substantial contributions from QED processes and is therefore not included in this Letter.
Results in the mid-rapidity region (|η| < 2) are obtained from a tracklet analysis using the two layers of the SPD as mentioned in Sect. 2. The analysis method and data used are identical to what has previously been presented [6,7].
The measurements in the forward region (|η| > 2) are provided by the FMD.The FMD records the full energy deposition of charged particles that impinge on the detector.Since all charged particles that hit the FMD are boosted in the laboratory frame, the detection efficiency is close to 100% for all momenta.As reported earlier [6], the main challenge in measuring the number of charged primary particles in this region, is the large background of secondary particles produced in the surrounding material.Due to the complexity and the limited knowledge of the material distribution of support structures away from the central barrel, it has not been possible to adequately describe (on the few %-level) the generation of secondary particles in the forward directions within the precision of the current simulation of the ALICE apparatus.
A suitable means to extract the number of primary charged particles was found by utilising collisions between so-called 'satellite' bunches and main bunches offset in intervals of 37.5 cm along the beamline.Satellite bunches are caused by the so-called debunching effect [14].A small fraction of the beam can be captured in unwanted RF buckets, due the way beams are injected into the accelerator, and create these satellite bunches spaced by 2.5 ns.Collisions between satellite and main bunches can cause instabilities in the beam, and the LHC has taken steps to reduce the number of these kinds of collisions.ALICE has therefore not recorded collisions between satellite and main bunches before or after the Pb-Pb run of 2010.In satellite-main collisions the background of secondary particles was much smaller and much better understood since significantly less detector material shadows the forward detectors [6].
A study utilising these satellite-main collisions led to the publication of the measurement of the chargedparticle pseudorapidity density in the 30% most central events over |η| < 5 [6].The study was limited in centrality reach by the need to use the Zero-Degree Calorimeter (ZDC) for the centrality estimation for collisions between satellite and main bunches.The ZDC measures the energy of spectator (noninteracting) nucleons with two components: one measures protons and the other measures neutrons.The ZDC was located at about 114 m from the interaction point on either side of the experiment [10], and was therefore ideally suited for that study.The centrality determination capability of the ZDC is however limited to the 30% most central collisions [13].
For centralities larger than 30% the V0 amplitude is used as the centrality estimator, which is available only for collisions at |IP z | < 15 cm -the so-called nominal interaction point corresponding to main bunches of one beam colliding with main bunches of the other beam.
To extend the centrality reach of the dN ch /dη measurement, a data-driven correction for the number of secondaries impinging on the FMD has been implemented.For each centrality class C, we form the ratio That is, the ratio of the measured inclusive charged-particle density from main-main collisions (|IP z | < 10 cm) provided by the FMD to the primary charged-particle density from satellite-main collisions [6].
Here, 'inclusive' denotes primary and secondary charged particles i.e., no correction was applied to account for secondary particles impinging on the FMD.
Note, that the correction is formed bin-by-bin in pseudorapidity, so that the pseudorapidity is the same for both the numerator and denominator.However, the numerator and denominator differ in the offset along the beam line of origin of the measured particles: For the numerator the origin lies within the nominal interaction region, while for the denominator the origin was offset by multiples of 37.5 cm.
This ratio is obtained separately for all previously published centrality classes: 0-5%, 5-10%, 10-20% and 20-30%.The variation of E c for different centralities is small (< 1%, much smaller than the precision Centrality evolution of the charged-particle pseudorapidity density in Pb-Pb ALICE Collaboration of the measurements).The weighted average is used as a global correction to obtain the primary charged-particle pseudorapidity density where X stands for an event selection e.g., a centrality range.
The simulation-based correction S(η) for secondary particles to the charged-particle pseudorapidity density in the forward directions is given by where N inclusive,FMD is the number of primary and secondary particles impinging on the FMD -as given by the track propagation of the simulation, and N primary,generated is the number of generated primary particles at a given pseudorapidity.Complete detector-simulation studies show that three effects can contribute to the generation of secondaries, and hence the value of S(η).These three effects are: material in which secondaries are produced, the transverse momentum (p T ) distribution and particle composition of the generated particles, and lastly the total number of produced particles.Of these three the material is by far the dominant effect, while the p T and particle composition only effects S(η) on the few percent level.The total number of generated particles has a negligible effect on S(η).That is, the material surrounding the detectors amplifies the primary-particle signal through particle production by a constant factor that first and foremost depends on the amount of material itself, and only secondarily on the p T and particle composition of the generated primary particles.
To estimate how much E C (η) itself would have changed if another system or centrality range was used to calculate the correction, S(η) is analysed from simulations with various collision systems and energies.We find that, even for large variations in particle composition and p T distributions, S(η) only varies by up to 5%.Re-weighting the particle composition and p T distributions from the various systems to match produces consistent values of S(η) ensuring that the 5% variations found were only due to particle composition and p T distributions differences.This uncertainty is applied to E(η) to account for all reasonable variations of the particle composition and p T distributions, which cannot be measured in the forward regions of ALICE.
Figure 1 shows the comparison of the data driven correction E(η) to the simulation-based correction S(η) from PYTHIA [15] (pp) and a parameterisation of the available ALICE results [16,17] for Pb-Pb collisions.The simulated collisions are for two distinct systems and span over almost an order of magnitude in collision energy.The total number of produced particles in these simulations span five orders of magnitude, and no dependence of S(η) on charged-particle multiplicity is observed.
By comparing E(η) to S(η) from simulations, one finds a good correspondence between the two corrections except in regions where the material description in the simulations is known to be inadequate.This, together with the fact that the numerator and denominator of Eq. 1 measure the same physical process, but differ foremost in the material traversed by the primary particles, and hence the number of secondary particles observed, implies that the correction E(η) is universal.That is, Eq. 3 is applicable for any event selection X in any collision system or at any collision energy, where the produced multiplicity, p T distributions, and particle composition is close to the range of the simulated systems used to study S(η).
Note, for the previously published results [6], which used satellite-main collisions, the simulation-based approach for correcting for secondary particles i.e., applying S(η) directly, was valid.As mentioned Centrality evolution of the charged-particle pseudorapidity density in Pb-Pb above, in satellite-main collisions, the particles that impinge on the FMD traverse far less and better described material in the simulation of the ALICE apparatus.The use of a simulation-based correction for secondary particles was in that analysis cross-checked by comparing to and combining with measurements from the V0 and SPD [6].Despite concerted efforts to improve the simulations by the Collaboration it has not been possible to achieve the same accuracy in S(η) for main-main collisions.
Finally, the effect of variation of the location of the primary interaction point on E(η) was studied.It was found, that the effect is negligible, given that the distribution of IP z are similar between the numerator of Eq. 1 and right-hand side of Eq. 3, as was the case in this analysis.
The method used in this analysis to extract the inclusive number of charged particles from the FMD is the same as for previous published results [6], except that the data-driven correction E(η) -rather than a simulation-based one S(η) -is used to correct for secondary particles.

Systematic uncertainties
Table 2 summarises the systematic uncertainties of this analysis.The common systematic uncertainty from the centrality selection is correlated across η and detailed elsewhere [13].
For the SPD measurements, the systematic uncertainties are the same as for the previously published mid-rapidity result [7], except for a contribution from the correction due to the larger acceptance used in this analysis.This uncertainty stems from the range of IP z used in the analysis (here |IP z | < 15 cm).At larger absolute values of IP z the acceptance correction for the SPD tracklets grows, and the uncertainty with it, being therefore η-dependent and largest at |η| ≈ 2.
The various sources of systematic uncertainties for the FMD measurements are detailed elsewhere [6], but will be expanded upon in the following since some values have changed due to better understanding Centrality evolution of the charged-particle pseudorapidity density in Pb-Pb ALICE Collaboration of the detector response.
In the analysis, three η-dependent thresholds are used.The values for these thresholds are obtained by fitting a convoluted Landau-Gauss distribution [18] to the energy loss spectrum measured by the FMD in a given η range.The uncertainties associated with these thresholds are detailed below.
A charged particle traversing the FMD can deposit energy in more than one element i.e., strip, of the detector.Therefore it is necessary to re-combine two signals to get the single charged-particle energy loss in those cases.This recombination depends on a lower threshold for accepting a signal, and an upper threshold to consider a signal as isolated i.e., all energy is deposited in a single strip.The systematic uncertainties from the recombination of signals are found by varying the lower and upper threshold values within bounds of the energy loss fits and by simulation studies.
To calculate the inclusive number of charged particles, a statistical approach is used [6].The strips of the FMD are divided into regions, and the number of empty strips is compared to the total number of strips in a given region.Strips with a signal below a given threshold are considered empty.The threshold was varied within bounds of the energy loss fits and investigated in simulation studies to obtain the systematic uncertainty.
The data-driven correction for secondary particles defined in Eq. 2 is derived from the previously published results, and as such contains contributions from the systematic uncertainties of those results [6].
Factoring out common correlated uncertainties e.g., the contribution from the centrality determination, we find a contribution of 4.7% from the previously published results.By studying the variation of the numerator of Eq. 1 under different experimental conditions e.g., different data-taking periods, and adding the variance in quadrature, the uncorrelated, total uncertainty on E(η) is found to be 6.1%.Systematic uncertainties can in general not be cancelled between the numerator and denominator of Eq. 1, since the same η regions are probed by different detector elements in each.
Note, that the previously published result [6] used in Eq. 1 already carries a 2% systematic uncertainty from the particle composition and p T distribution [6].This contribution is contained in the 4.7% quoted above, and is propagated to the final 6.1% systematic uncertainty on E(η).
Finally, it was found through simulations that the acceptance region of FMD1 is particularly affected by the variations in the number of secondary particles stemming from variations in the particle composition and p T distribution, and gives rise to an additional 2% systematic uncertainty, which is added in quadrature to the rest of the systematic uncertainties, but only for η > 3.7.Previously published results for 0-30% over the full pseudorapidity range (diamonds) [6] and for 0-80% at mid-rapidity (stars) [7] are also shown.
The combined distributions in Fig. 3 are calculated as the average of the individual measurements from the FMD and SPD, weighted by statistical errors and systematic uncertainties, omitting those which are common such as that from the centrality determination.The distributions are then symmetrised around η = 0 by taking the weighted average of ±η points.Points at 3.5 < η < 5 are reflected on to −5 < η < −3.5 to provide the dN ch /dη distributions in a range comparable to the previously published results [6].
The lines in Fig. 3 are fits of to the measured distributions.The function f GG is the difference of two Gaussian distributions centred at η = 0 with amplitudes A 1 , A 2 , and widths σ 1 , σ 2 .The function describes the data well within the measured region with a reduced χ 2 smaller than 1.We find values of A 2 /A 1 for all centralities, from 0.20 to 0.31 but are consistent within fit uncertainties, with a constant value of 0.23 ± 0.02.Likewise values of σ 2 /σ 1 for all centralities, ranges from 0.28 to 0.36 and are consistent with a constant value of 0.31 ± 0.02.
Qualitatively the shape of the charged-particle pseudorapidity density distributions broadens only slightly toward more peripheral events, consistent with the above observation.Indeed, the full-width halfmaximum (FWHM) shown in Fig. 4 versus the number of participating nucleons N part -calculated using a Glauber model [13] -increase sharply only in the very most peripheral collisions.The dN ch /dη distributions does not extend far enough to calculate reliable values for FWHM directly from the data.Previously published results for 0-30% over the full pseudorapidity range (diamonds) [6] are also shown.The lines correspond to fits of Eq. 5 to the data.
Instead f GG (η) − max( f GG )/2 = 0 was numerically solved, and the uncertainties evaluated as the error of f GG at the roots, divided by the slope at those roots.The width of the dN ch /dη distributions follows the same trend, in the region of 0-50%, as was seen in lower energy results from PHOBOS reproduced in Figure 4 for comparison [2].
Figure 5 presents the charged-particle pseudorapidity density per average number of participating nucleon pairs N part /2 as a function of the average number of participants N part .Although there is a slight increase in the ratio to the central pseudorapidity density distribution at low N part (see lower part of Fig. 5), the uncertainties are large and no strong evolution of the shape of the pseudorapidity density distribution over pseudorapidity with respect to centrality is observed.The ratio at 3.5 < |η| < 4.5 does deviate somewhat in peripheral collisions, which is attributed to the general broadening of the pseudorapidity density distributions in those collisions.
To extract the total number of charged particles produced in Pb-Pb collisions at various centralities, a number of functions, including Eq. 5, is fitted to the dN ch /dη distributions.A trapezoid was successfully used by PHOBOS to describe limiting fragmentation [2].Here, [−M, M] is the range in which the function is constant, and A is the amplitude.The parameterisation    [6,17].The uncertainties on N part from the Glauber calculations are only included on the points at midrapidity.Thus, the uncertainty band around the mid-rapidity points reflect both the measurement uncertainties and the uncertainty on N part , while other η ranges only show the measurement uncertainties.The lower part shows the ratio of each distribution to the previously published distributions for |η| < 0.5.
Centrality evolution of the charged-particle pseudorapidity density in Pb-Pb ALICE Collaboration as suggested by PHOBOS, is likewise fitted to the dN ch /dη distributions.The parameter a expresses the width of the distribution, and α and β , and expresses the width and depth of the dip at η ≈ 0, respectively.
A is an overall scale parameter.Finally, to remedy some of the obvious defects of the trapezoid i.e., a non-continuous first derivative at η = M, we use a Bjorken-inspired function [6] which has plateau at A for |η| < µ connected to Gaussian fall-off beyond ±µ.The fitted functions are integrated over η up to the beam rapidity ±y beam = ±7.99.Although the dN ch /dη distributions in principle continue to infinity, there is no significant loss in generality or precision by cutting the integral at η = ±y beam since the distributions rapidly approach zero.Notice that all parameters of the functions are left free in the fitting procedure.All functions give reasonable fits (with a reduced χ 2 smaller than 1), though the trapezoid and Bjorken-inspired ansatz are too flat at the mid-rapidity.The calculation of the central values and uncertainties are done as for previous results [6]: The central value is calculated from the integral of the trapezoid fit to compare directly to previous results; the spread between the integrals and the central value is evaluated to obtain the uncertainty on the total N ch .
The extrapolated total N ch versus N part is shown in Fig. 6, and compared to lower energy results from PHOBOS [19].At LHC energies the particle production as a function of N part shows a similar behaviour to the lower energy results, and the factorisation [2] in centrality and energy seems to hold (see fit in Fig. 6).[13].The uncertainty on the extrapolation is smaller than the size of the markers.The four right-most points are the previously published results [6].A function inspired by factorisation [2] is fitted to the data, and the best fit yield a = 35.8± 4.2, b = 0.22 ± 0.05 with a reduced χ 2 of 0.18.Also shown is the PHOBOS result at lower energy result [19] scaled to the ALICE total number of charged particles per participant at N part = 180.
In Fig. 7 we show comparisons of various model calculations to the measured charged-particle pseudorapidity density as a function of centrality.The centrality class for a given model-generated event was Centrality evolution of the charged-particle pseudorapidity density in Pb-Pb ALICE Collaboration determined by sharp cuts in the impact parameter b and a Glauber calculation [13].
The HIJING model [3] (version 1.383, with jet-quenching disabled, shadowing enabled, and a hard p T cut-off of 2.3 GeV) is seen to overshoot the data for all centralities.In addition, the distributions at all centralities decrease with increasing |η| faster than the data would suggest.
AMPT [4] without string melting reproduces the data fairly well at central pseudorapidity for the most central events -exactly in the region it was tuned to, but it fails to describe the charged-particle pseudorapidity density for more peripheral events.Also, AMPT without string melting would suggest a wider central region than supported by data, and similarly to HIJING decreases faster than the data.AMPT with string melting -which essentially implements quark coalescence, and therefore a more predominant parton phase -is seen to be very flat at mid-rapidity and underestimates the yield, except for peripheral collisions.
Finally, EPOS-LHC [5] reproduces the shape fairly well, but underestimates the data by 10 to 30%.

Conclusions
The charged-particle pseudorapidity density has been measured in Pb-Pb collisions at √ s NN = 2.76 TeV over a broad pseudorapidity range, extending previous published results by ALICE to more peripheral collisions.In the mid-rapidity region the well-established tracklet procedure was used.In the forward regions, a new data-driven procedure to correct for the large background due to secondary particles was used.The results presented here are consistent with the behaviour previously seen in more central collisions and in a limited pseudorapidity range.No strong evolution of the overall shape of the charged-particle pseudorapidity density distributions as a function of collision centrality is observed.When normalised to the number of participating nucleons in the collision, the centrality evolution is small over the pseudorapidity range.Since the measurement was performed over a large pseudorapidity range (−3.5 < η < 5), it allows for an estimate of the total number of charged particles produced in Pb-Pb collisions at √ s NN = 2.76 TeV.The total charged-particle multiplicity is found to scale approximately with the number of participating nucleons.This would suggest that hard contributions to the total charged-particle multiplicity are small.From peripheral to central collisions we observe an increase of two orders of magnitude in the number of produced charge particles.A comparison of the data to the different available predictions from HIJING, AMPT, and EPOS-LHC show that none of these models captures both the shape and level of the measured distributions.AMPT however comes close in limited ranges of centrality.The exact centrality ranges that AMPT describes depend strongly on whether string melting is used in the model or not.EPOS-LHC -although systematically low -shows a reasonable agreement with the shape of the measured charged-particle pseudorapidity density distribution over a wider pseudorapidity range.

Figure 1 :
Figure 1: [Colour online] Comparison of data-driven to simulation-based corrections for secondary particles impinging on the FMD.Different markers correspond to different collision systems and energies, and the colours indicate the five FMD rings.S(η)is shown for 0 cm < IP z < 2 cm as an example, while E(η) is independent of IP z (see also text).PYTHIA was used for pp collisions, and the Pb-Pb points are from simulation with a parameterisation which include the available ALICE data on particle composition and p T distributions.Black circles correspond to E(η).

Figure 2 :
Figure 2: [Colour online] Measurement of dN ch /dη per centrality from SPD (squares) and FMD (circles) separately.Error bars reflect the total uncorrelated systematic uncertainty and statistical error on each point.Error bars on the left and right reflect the correlated systematic uncertainties on the SPD and FMD points, respectively.Previously published results for 0-30% over the full pseudorapidity range (diamonds)[6] and for 0-80% at mid-rapidity (stars)[7] are also shown.

Figure 3 :
Figure 3: [Colour online] Measurement of dN ch /dη for all centralities and a broad η range.Combined and symmetrised dN ch /dη over 30-90% centrality from both SPD and FMD (circles).Open boxes reflect the total uncorrelated systematic uncertainties and statistical errors, while the filled boxes on the right reflect the correlated systematic uncertainty.Also shown, is the reflection of the 3.5 < η < 5 values around η = 0 (open circles).Previously published results for 0-30% over the full pseudorapidity range (diamonds)[6] are also shown.The lines correspond to fits of Eq. 5 to the data.

Figure 4 :|η| < 0. 5 − 5 2. 5 < 5 ALICE
Figure4: [Colour online] Full-width half-maximum of the charged-particle pseudorapidity distributions versus the average number of participants.The uncertainties on the ALICE measurements are from the fit of f GG only and evaluated at 95% confidence level.Also shown are lower energy results from PHOBOS[2].

Figure 5 :
Figure 5: [Colour online] The charged-particle pseudorapidity density distributions scaled by the average number of participants in various pseudorapidity intervals as a function of the number of participants.The four right-most points (open symbols) in each η range, as well as the mid-rapidity points (circles) are from previously published results[6,17].The uncertainties on N part from the Glauber calculations are only included on the points at midrapidity.Thus, the uncertainty band around the mid-rapidity points reflect both the measurement uncertainties and the uncertainty on N part , while other η ranges only show the measurement uncertainties.The lower part shows the ratio of each distribution to the previously published distributions for |η| < 0.5.

Figure 6 :
Figure 6: [Colour online] Extrapolation to the total number of charged particles as a function of the number of participating nucleons[13].The uncertainty on the extrapolation is smaller than the size of the markers.The four right-most points are the previously published results[6].A function inspired by factorisation[2] is fitted to the data, and the best fit yield a = 35.8± 4.2, b = 0.22 ± 0.05 with a reduced χ 2 of 0.18.Also shown is the PHOBOS result at lower energy result[19] scaled to the ALICE total number of charged particles per participant at N part = 180.

Figure 7 :
Figure 7: [Colour online] Comparison of dN ch /dη per centrality class from HIJING, AMPT (with and without string melting), and EPOS-LHC model calculations to the measured distributions. Table

Table 2 :
Summary of systematic uncertainties: the common systematic uncertainties shared by both the SPD and the FMD, and the uncertainties particular to the detectors.