Latest LHCf physics results

The LHCf experiment is dedicated to the measurement of very forward particle production in the high energy hadron-hadron collisions at LHC, with the aim of improving the cosmic-ray air shower developments models. The detector has taken data in p-p collisions at different center of mass energies, from 900 GeV up to 7 TeV, and in p-Pb collisions at √ s = 5.02 TeV. The results of forward production spectra of neutrons in p-p collisions and π0 in p-Pb collisions, compared with the models most widely used in the High Energy Cosmic Ray physics, are presented in this paper.


Brief description of the experiment
The LHCf experiment is composed by two independent detectors (Arm1 and Arm2), each made by two sampling and imaging calorimeter towers, composed of 16 tungsten layers, 16 plastic scintillator layers for energy measurement and four position sensitive layers for impact position determination. The position sensitive layers of Arm1 and Arm2 are X-Y scintillating fiber (SciFi) hodoscopes and X-Y silicon strip detectors, respectively. The transverse cross sections of calorimeter towers are 20×20 mm 2 and 40×40 mm 2 in Arm1 and 25×25 mm 2 and 32×32mm 2 in Arm2. The detectors are installed 140 m away from the IP1 (Atlas) LHC interaction point, inside the zero-degree neutral absorber (Target Neutral Absorber, TAN). Charged particles from the IP are swept away by the inner beam separation dipole before reaching the TAN, so that only photons mainly from π 0 decays, neutrons and neutral kaons reach the LHCf calorimeters. The LHCf detectors are hence able to measure neutral particles in the pseudo-rapidity range from η = 8.4 to infinite (zero degree). The energy resolution of the detectors for photons and neutrons is 5% and 35%, respectively, while the position resolution in the neutral par-ticle's reconstruction is better than 200µm for photons and a few mm for neutrons. More details of the detector performance are reported in [1,2].
The experiment has taken data during different periods, between 2009 and 2013, both in p-p and p-Pb collisions. Here we will report on the latest analysis results, the neutron inclusive energy spectra in 7 TeV p-p collisions and the P T vs η π 0 spectra in p-Pb collisions at √ s = 5.02 TeV.

Neutron spectra in p-p collisions
In this section we will present some preliminary results for the inclusive neutron spectra measured by the LHCf experiment in p-p collisions at √ s = 7 TeV. The data used in this analysis have been taken at the beginning of the LHC run, in 2010, in low luminosity conditions, that are optimal for the LHCf experiment.
The performance of the LHCf detectors for hadron measurements have been carefuly studied by Monte Carlo (MC) simulations and confirmed by using 350 GeV proton beams at CERN-SPS. Details can be found in [3].
The analysis workflow for the neutron energy spectra measurement is similar to the workflow used for the γ analysis, described in [4]. The luminosity has been measured with the help of the LHCf front counter rates, properly normalized with the Van der Meer LHC scan. Particle identification has been carried out by looking at the longitudinal shower development, using two dimensional cuts in the L20 % and L90 % plane, where L20 % and L90 % are the longitudinal depths containing 20% and 90% of the total deposited energy, respectively. Hit position has been evaluated by using the transverse shower distribution, measured with the position sensitive layers, to optimally correct the energy measured for the leakage effects.
Due to the limited energy resolution for the neutron sample ( 30-40%), the measured spectra should be properly unfolded to get the realistic neutron spectra. We used a multidimensional-spectra unfolding method based on Bayesian theory, relying on the two measured variables Energy and P T . Figure 1 shows the preliminary energy spectra of forward neutrons measured by the Arm1 detector compared with the MC predictions; the three panels correspond to the spectra measured in different regions on the small tower (η > 10.76 in the first plot) and on the large tower (8.99 < η < 9.22 in the second plot and 8.81 < η < 8.99 in the thrid plot).
The effect of the limited energy resolution for hadrons is evident from these plots. The preliminary neutron physics spectra obtained after the unfolding procedure are presented in Figure 2. These spectra are obtained from the combination of the Arm1 and Arm2 spectra in the three above defined rapidity regions.
The most striking feature of these spectra is the pronounced high energy neutron peak in the very forward region (η > 10.76), that is predicted only by the QGSJET-II model. The presence of this peak is a clear indication of small inelasticity in the very forward p-p collision. We can hence expect that the baryons produced in the first interaction of a primary cosmic ray with the atmosphere will carry on a large fraction of the primary energy, leading to deeply penetrating VHECR induced atmospheric showers.
3. π 0 spectra in p-Pb collision at √ s = 5.02 TeV At the beginning of 2013, LHCf took data with the Arm2 detector in p-Pb collisions at the center-of-mass collision energy per nucleon of √ s NN = 5.02 TeV. In most of the operation time, the detector was located on the p-remnant side. The spectra measured in the proton remnant side of the high energy p-Pb collisions provide new information on the effect of the nuclear matter in the forward particle production; this kinematical region is particularly important for HECR Physics since the interaction of the primary cosmic rays in the atmosphere occur between proton and nucleus or between nucleus and nucleus, where one of the nucleus is light component [5]. The gathered data have been analysed to extract the π 0 P T spectra in different rapidity regions, and the measured spectra have been compared to the predictions of the various hadronic interaction models. We present here the results that have been published in [6]. Figure 3 shows the invariant mass spectrum of the photons pairs reconstructed in the two calorimetric towers, demonstrating the excellent performances of the detector in the reconstruction of a clean π 0 sample.
Once the π 0 has been clearly identified, we can determine it's transverse momentum and rapidity, allowing the measurement of the P T vs η spectra.
We can expect thet approximately half of the forward produced π 0 arise from Ultra-Peripheral collisions (UPC) of the proton in the nucleus electromagnetic field. To obtain the QCD induced spectra, the effect of the UPC should be subtracted. We have used the Weizsacher-Williams approximation to estimate the energy distribution of the virtual photons emitted by the fully ionized Pb nucleus, and we simulated the electromagnetic interactions between the virtual photons and  cation factor is dened as Ed 3 σ pPb /dp 3 Ed 3 σ pp /dp 3 , where Ed 3 σ pPb /dp 3 and Ed 3 σ pp /dp 3 are the inclusive cross sections of π 0 production in pPb and pp collisions at 5.02 TeV respectively. The UPC component is already subtracted in σ pPb . The uncertainty on σ pPb inel is estimated to be ±5 % by comparing the σ pPb inel value with other calculations and experimental results presented in [35,36]. The average number of binary nucleonnucleon collisions in a pPb collision, ⟨N coll ⟩ = 6.9, is obtained from MC simulations using the Glauber model [33]. The uncertainty in σ pp inel /⟨N coll ⟩ is estimated by varying the parameters in the calculation with the Glauber model [5,37] (where the cancellation of the uncertainties in σ pp inel and ⟨N coll ⟩ is taken into account) and is of the order of ±3.5 % . Finally the quadratic sum of the uncertainties in σ pPb inel and σ pp inel /⟨N coll ⟩ is added to R pPb . Since there is no data at √ s = 5.02 TeV for pp collisions Ed 3 σ pp /dp 3 is derived by scaling the p T spectra taken in pp collisions at 7 TeV and 2.76 TeV. The derivation follows three steps. First (I) the ⟨p T ⟩ values at 5.02 TeV are estimated by interpolating the ⟨p T ⟩ values at 7 TeV and 2.76 TeV, assuming that the Feynman scal-ing of ⟨p T ⟩ is only a function of rapidity. Then (II) the absolute normalizations of the p T spectra at 5.02 TeV are determined by applying the measured absolute normalizations at 7 TeV directly to those at 5.02 TeV. Finally (III) the p T spectra at 5.02 TeV are derived assuming that the p T spectra follow a Gaussian distribution with width 2⟨p T ⟩/ √ π (obtained in step (I)) and using the normalizations obtained in step (II). The rapidity shift −0.465 explained in Sec. III is also taken into account in the p T spectrum at 5.02 TeV. The details of the procedure are discussed in Appendix B. Figure 6 shows the nuclear modication factors R pPb from the LHCf measurements and the predictions by hadronic interaction models dpmjet 3.04 (red solid line), qgsjet II-03 (blue dashed line), and epos 1.99 (magenta dotted line). The LHCf measurements, although with a large uncertainty which increases with p T (mainly due to systematic uncertainties in pPb collisions at 5.02 TeV), show a strong suppression with R pPb equal 0.1 at p T ≈ 0.1 GeV rising to 0.3 at p T ≈ 0.6 GeV. All hadronic interaction models predict small values of R pPb ≈ 0.1, and they show an overall good agreement with the LHCf measurements within the uncertainty. Clearly other analyses which are more sensitive to exclusive π 0 signals are needed, for example diractive dissociation, to investigate the reason for this strong suppression. However the the proton with the SOPHIA model [7]. The spectra obtained in this way, after the UPC subtraction, have been compared with the spectra in p-p collisions at 5.02 TeV, obtained by interpolationg the LHCf results at 2.76 TeV and 7 TeV [6] (see Figure 4). Results are also compared with the prediction of EPOS 1.99, DPMJET 3.04 and QGSJET II-03 models.
From the comparison of p-p and p-Pb results we have derived the nuclear modification factor, R pPb , that is used to quantify the effect of the nuclear medium on the production of forward energetic particles. Figure 5 shows R pPb measured by LHCf compared with the predictions by DPMJET 3.04, QGSJET II-03, and EPOS 1.99. The LHCf measurements, although with a large uncertainty, show a strong suppression: R pPb varies from 0.1 at p T ≈ 0.1 GeV to 0.3 at p T ≈ 0.6 GeV. All hadronic interaction models predict small values of R pPb ≈ 0.1, and they show an overall good agreement with the LHCf measurements within the uncertainty. The inclusive production rates of π 0 s measured by LHCf after the subtraction of the UPC component are summarized in Tables IV IX. Appendix B Derivation of the p6 spectra in pp collisions at √ s = 5.02 TeV To investigate the nuclear eects involved in the nuclear target it is essential to compare the p T spectra measured in pPb collisions at a given collision energy to the reference p T spectra in pp collisions at the same collision energy. In this analysis, since a measurement in pp collisions at √ s = 5.02 TeV is not available, the reference p T spectra are made by scaling the p T spectra measured in the pp collisions at √ s = 7 TeV and 2.76 TeV. First the average p T at 5.02 TeV is estimated by scaling the average p T obtained at 7 TeV and 2.76 TeV. Figure 7 shows the average p T in pp collisions at √ s = 7 TeV (lled circles) and 2.76 TeV (open circles) as a function of rapidity loss ∆y ≡ y beam − y cms , where y beam is beam rapidity and y cms is the rapidity of the center-of-mass frame. For the proton beam with E = 3.5 TeV and 1.38 TeV, y beam gives 8.917 and 7.987, respectively. In the following we assume y cms is positive.
According to the scaling law proposed by several authors [3840] (Feynman scaling), the average p T as a function of ∆y should be independent of the center-ofmass energy in the projectile fragmentation region. Thus the average p T can be directly compared among dierent collision energies. The values of the average p T at 7 TeV are taken from measurements by LHCf [29] in which the associated ∆y points are modied to take into account event population for each rapidity bin. These weighted bin centers are estimated using the MC simulation by epos 1.99. The values of the average p T at 2.76 TeV are obtained by a similar analysis on the data that was taken in pp collisions at √ s = 2.76 TeV on February 13, 2013. These data were taken with essentially the same data acquisition conguration as at 5.02 TeV.
Although the two measurements in Fig. 7 have limited overlap on the ∆y range owing to the smaller collision energy at 2.76 TeV, the ⟨p T ⟩ spectra at 7 TeV and 2.76 TeV follow mostly a common line. A linear function t is then made to these measurements. The solid line and shaded area in Fig. 7 show the best-t linear function and the 1 standard deviation uncertainty obtained by a chi-square