Forward Neutrino Fluxes at the LHC

Felix Kling∗ Theory Group, SLAC National Accelerator Laboratory, Menlo Park, California 94025

With the upcoming Run 3 of the LHC, the FASERν and SND@LHC detectors will start a new era of neutrino physics using the far-forward high-energy neutrino beam produced in collisions at ATLAS. This emerging LHC neutrino physics program requires reliable estimates of the LHC’s forward neutrino fluxes and their uncertainties. In this paper we provide a new fast-neutrino flux simulation, implemented as a RIVET module, to address this issue. We present the expected energy distributions going through the FASERν and SND@LHC detectors based on various commonly used event generators, analyze the origin of those neutrinos, and present the expected neutrino event rates.

I. INTRODUCTION and present an alternative fast neutrino flux simulation implemented as a RIVET [14, 15] module. As the highest energy particle accelerator ever built, The rest of this paper is organized as follows. In Sec.II, the Large Hadron Collider (LHC) is also the source of the we provide a description of our fast neutrino flux simula- most energetic neutrinos created in a controlled labora- tion module. In Sec.III we present the neutrino energy tory environment. Proton-proton collisions at the LHC and rapidity spectra obtained from different event gener- typically lead to a large number of hadrons produced ators. In Sec.IV we present the expected neutrino inter- along the beam direction, which can inherit a signifi- actions rates in FASERν and SND@LHC and discuss the cant fraction of the protons’ energy. The decays of these effect of the beam crossing angle. We conclude in Sec.V. hadrons then lead to an intense and strongly collimated A comparison of forward hadron spectra obtained by dif- beam of highly energetic neutrinos of all three flavors in ferent MC generators can be found in AppendixA. the far-forward direction. Although the possibility of detecting and probing neu- trinos at the LHC was considered as early as 1984 [1– II. FAST NEUTRINO FLUX SIMULATION 6], no LHC neutrino has been detected for a long time. This situation changed very recently, when the FASER Neutrinos going through the far-forward LHC neutrino collaboration reported the observation of the first neu- experiments mainly originate from the weak decay of the trino interaction candidates at the LHC [7]. This sit- lightest mesons and baryons of a given flavor. At the uation will further improve during the third run of the LHC, this includes mainly pions, kaons, hyperons and LHC with upcoming FASERν [8,9] and SND@LHC de- D-mesons. tectors [10, 11]. Placed directly in the LHC’s forward While charmed hadrons decay approximately neutrino beam, both experiments are expected to detect promptly, light flavoured hadrons are long-lived thousands of neutrino interactions at TeV energies. This and decay downstream from the interaction point (IP). will open a new window to study neutrino interactions This requires us to model their propagation through the at high energies and therefore extend the LHC’s physics LHC’s beam pipe and magnetic fields. The geometrical program in a new direction. model and tracking algorithms used in this study are This emerging LHC neutrino physics program requires based on the one implemented in BDSIM [12] and reliable estimates of the LHC’s forward neutrino fluxes illustrated in Fig.1. The upper and lower panel show a and their uncertainties. These estimates are based on es- cross sectional view of the beam pipe geometry in the tablished Monte Carlo (MC) event generators to simulate horizontal and vertical plane, respectively, as indicated the production of hadrons in proton-proton interactions. by the solid black contour. Particles hitting this bound- arXiv:2105.08270v1 [hep-ph] 18 May 2021 For these hadrons we then have to simulate the propaga- ary of the beam pipe are assumed to be absorbed quickly tion through the LHC’s beam pipe and magnetic fields and are no longer tracked. Additionally, we show the as well as their decay into neutrinos. This can be done quadupole magnet and dipole magnet regions as light using dedicated simulation tools such as BDSIM [12] or and dark gray shaded area, respectively. FLUKA [13], which however tend to be rather compu- Located 19.1 m downstream from the IP is the TAS tationally expansive, time consuming, and often require front quadrupole absorber which absorbs hadrons with special expertise or code access that is not available to angles θ 0.9 mrad with respect to the beam axis. broad community. This makes the simulation of neutrino & The TAS is followed by a series of quadrupole magnets, fluxes with different generators, as for example needed to the so called inner triplet, and the D1 magnet. The D1 obtain flux uncertainties or for phenomenological studies, dipole separates the two protons beams and deflects most difficult to impossible. In this study we address this issue charged particles such that they collide with the beam pipe. Placed at 140 m downstream from the IP the TAN, which will absorb the forward going neutral particles. At ∗ [email protected] this location, the beam pipe splits into separate pipes 2

proton beam with y=150 rad, E=2TeV 0.10 + with y=150 rad, E=2TeV

0.05

0.00 x [m]

0.05

0.10

proton beam z [m] 0.10 with y=150 rad, E=2TeV + with y=150 rad, E=2TeV

0.05

0.00 y [m]

0.05

0.10 Q1-3 Quadrupole D1 Dipole D2 Dipole 0 25 50 75 100 125 150 175 200 z [m]

FIG. 1. Beam Pipe Geometry: The boundaries of the LHC’s beam pipe in the horizontal (top) and vertical (bottom) plane are shown as a black line. The shaded areas correspond to the quadrupole (light gray) and dipole (dark gray) magnets. The red dashed line shows the trajectory of the proton beam with initial half beam crossing angle of 150 µrad vertically upwards. The dotted blue and dot-dashed green curve show the trajectories of two oppositely charged 2 TeV energy pions with the same initial orientations.

for the two beams. Further downstream, at z = 153 m, the decay channels Λ → peν or Ds → τντ . To avoid is the outer beam separation dipole magnet D2, which this problem and allow for a fair comparison of different re-aligns the proton beams to be parallel. hadronic interaction models, we therefore simulate the Also shown in Fig.1 is the trajectory for the 6.5 TeV decays of hadrons into neutrinos ourselves. Each hadron proton beam with the nominal LHC Run 2 beam half- is first decayed in its rest frame, according to the their crossing angle 150 µrad vertically upwards, as well as two decay branching fractions and energy distributions ob- trajectories for oppositely charged 2 TeV energy pions. tained with Pythia 8 [16], and the resulting neutrino is We have validated these and other trajectories against then boosted into the laboratory frame. the full BDSIM prediction and found great agreement In addition to hadron decays inside the LHC’s beam of O(10 µm) and better. We can see that even charged pipe, neutrinos can also be produced inside hadronic particles with large energies of 2 TeV are deflected sig- showers resulting from collisions of primary hadrons with nificantly by the inner triplet quadrupole magnets, such the LHC infrastructure. However, as we will see in the that further downstream decays of these charged parti- next section, the corresponding contribution to the far- cles are not expected to contribute significantly to the froward neutrino flux is strongly suppressed. This is due far-forward neutrino flux. to the small probability of secondary hadrons to decay After obtaining the particle trajectory, we sample the in medium before interacting again and their typically decays of hadrons into neutrinos at multiple locations broad angular spread. Nevertheless, to get a first im- along their trajectory. Unfortunately, the hadron de- pression for the importance of this production mode, we cay branching fractions differ between different event include a simplified modeling of neutrinos produced in generators. Most notably, the decay tables for sev- the first stage of this shower. When a hadron collides eral dedicated simulators, including Sibyll 2.3c and with the LHC’s beam pipe material, we sample the result- DPMJET III.2017.1, are incomplete and do not contain ing neutrinos following their two-dimension energy-angle 3 distribution as obtained from Pythia 8. In the future, of 800 kg. this could be replaced with a more accurate description The baseline beam configuration assumed in this study taking into account the whole shower, for example based considers a 13 TeV LHC with a beam-half crossing an- on Geant 4 [17]. gle of 150 µrad upwards in vertical directions, as used All parts of the outlined simulation are included in during the end of LHC Run 2. The question on how the a RIVET module. This module i) reads the forward fluxes and event rates change for different beam angle hadron fluxes from HepMC [18] files produced by the MC configurations will be discussed in Sec.IVB. When pre- generator, ii) propagates the long-lived hadrons through senting event rates, we assume an integrated luminosity the LHC beam pipe and magnets iii) obtains the neutri- of 150 fb−1 for Run 3. nos from decays of hadrons at multiple locations along their trajectory or interactions of hadrons with the beam pipe material, and iv) stores the resulting neutrino fluxes B. Event Generation going through the forward LHC neutrino experiments FASERν and SND@LHC as histograms in the yoda file In this study we use and compare the neutrino flux format [14]. The run time of the RIVET module is com- from several commonly used MC event generators. For parable to that of event generation, meaning that neutri- light hadron production, we use the dedicated cos- nos from thousands of collisions can be simulated within mic ray and forward physics generators EPOSLHC [28], minutes on a normal computer. This provides a signifi- QGSJET II-04 [29], DPMJET III.2017.1 [30, 31], and cant improvement in computation time compared to the Sibyll 2.3c [32–35], as implemented in the CRMC sim- BDSIM or FLUKA simulations. ulation package [36]. In addition, we also simulate light hadrons with the multi-purpose event generator Pythia 8.2 [16, 37] using the Monash tune [38], but note III. FORWARD NEUTRINO FLUXES that these predictions have not yet been tuned to or val- idated with forward physics data. A. Experimental Setup To simulate the production of charmed hadrons, we use DPMJET III.2017.1, Sibyll 2.3c and Pythia 8.2. In the upcoming Run 3 of the LHC, which is scheduled Pythia 8.2 provides the option to simulate charm pro- from 2022 to 2024, two new LHC experiments, FASERν duction either as part of their SoftQCD process for min- and SND@LHC, will start their operation and probe neu- imum bias events, or directly as a HardQCD process, and trinos at the LHC for the first time. we will present results for both. We note that only The FASER experiment has been originally proposed Sibyll 2.3c has been tuned to charm production data. to search for light long-lived particles at the LHC [5, 19– In contrast, neither Pythia 8.2 nor DPMJET III.2017.1 27]. Placed at its front is a dedicated neutrino detector, have been validated for charm production and their pre- called FASERν, which consists of emulsion films inter- dictions should therefore not be trusted. We mainly in- leaved with tungsten plates of total mass 1.2 tons [8,9]. clude them for comparison, as they have been used in This setup allows to measure the neutrino energy and previous studies [8, 10]. Indeed, the prediction for for- identify the neutrino flavour based on their signature ward charm production differ significantly by orders of in the emulsion detector, and distinguish muon neu- magnitude between the different generators, as shown in trino and anti-neutrinos interactions in combination with AppendixA. the downstream FASER spectrometer. The experiment is located about 480 m downstream from the ATLAS IP in the previously unused side tunnel TI12. At this C. Energy Distribution location, a trench was dug, allowing the whole detec- tor to be centered on the beam collision axis. At the Let us now turn to the results of our simulation. In nominal location, the detector covers the rectangular re- Fig.2 we show the number of electron (top), muon (cen- gion |x|, |y| < 12.5 cm, corresponding the pseudorapidity ter) and tau (bottom) neutrinos going through the cross range η & 8.5. The FASERν detector has the possibil- sectional area of FASERν (left column) and SND@LHC ity of being moved in the vertical/horizontal direction to (right column) as function of the neutrino energy. correct for changing beam crossing angle orientations. Our simulation allows to differentiate the neutrino More recently, the SND@LHC collaboration proposed flux components by origin (in terms of parent parti- another neutrino detector to be placed in the tunnel cle species), as shown by the differently colored lines. TI18, which is also 480 m away from the ATLAS interac- Charged pion decays provide the dominant contribu- tion IP, but located on its opposite side [10, 11]. Notably, tion to the muon neutrino flux at lower energies, while the center of the detector would be displaced from the charged kaon decays dominate the muon neutrino flux beam collision axis. The detector covers the rectangular at higher energies above a few 100 GeV. Electron neu- region 8 cm < x < 47 cm and 15.5 cm < y < 54.5 cm, trinos are mainly produced in kaon decays. While the corresponding to the pseudorapidity range 7 . η . 8.5. largest contribution comes from decays of KL mesons, The detector target consists of tungsten and has a mass KS decays still provide a sizable contribution since their 4

FASER : e + e Shower SND: e + e DPMJET 3.2017 QGSJET II-04 K D c SIBYLL 2.3c Pythia8 (Hard) 1011 EPOSLHC Pythia8 (Soft)

1010

109 Neutrinos [1/bin]

108

FASER : + Shower SND: + DPMJET 3.2017 QGSJET II-04 K D c SIBYLL 2.3c Pythia8 (Hard) 1012 EPOSLHC Pythia8 (Soft)

1011

1010 Neutrinos [1/bin]

109

FASER : + D c SND: + DPMJET 3.2017 Pythia8 (Hard) SIBYLL 2.3c Pythia8 (Soft) 1010

109

108 Neutrinos [1/bin]

107

10 102 103 10 102 103 Neutrino Energy [GeV] Neutrino Energy [GeV]

FIG. 2. Neutrino Energy Distribution: The panels show the neutrino energy spectrum for electron (top), muon (center) and tau (bottom) neutrinos passing through FASERν (left) and SND@LHC (right). The vertical axis shows the number of neutrinos per energy bin that go through the detector’s cross sectional area for an integrated luminosity of 150 fb−1. We separate the different production modes: pion decays (red), kaon decays (orange), hyperon decays (magenta), charm decays (blue) and secondary hadronic showers (green). The different linestyles correspond to predictions obtained from SIBYLL 2.3c (solid), DPMJET III.2017.1 (short dashed), EPOSLHC (long dashed), QGSJET II-04 (dotted), and Pythia 8.2 using the SoftQCD processes (dot-dashed) and Pythia 8.2 with the HardQCD process for charm production (double-dot-dashed). smaller lifetime and hence larger decay-in-flight proba- branching fraction. Hyperon decays are generally sub- bility can compensate for the their small semi-leptonic leading, with the notable exception of anti-electron neu- 5

e + e K + + DPMJET 3.2017 10 10 D c SIBYLL 2.3c Shower EPOSLHC ] QGSJET II-04 2 Pythia8 (Hard) m

c 9 Pythia8 (Soft) / 10 n i b / 1 [ 8 x 10 u l F

o n i

r 7

t 10 u e N 106 SND@LHC FASERv SND@LHC FASERv SND@LHC FASERv 6 8 10 12 6 8 10 12 6 8 10 12 Neutrino Pseudorapidity Neutrino Pseudorapidity Neutrino Pseudorapidity

100 10 1 100 10 1 100 10 1 Displacement from LoS [cm] Displacement from LoS [cm] Displacement from LoS [cm]

FIG. 3. Neutrino Angular Distribution: The panels show the flux for electron (left), muon (center) and tau (right) neutrinos, in units of particles per area per bin, as function of pseudorapidity η, or equivalently the radial displacement from the line of sight (LoS) at z = 480 m. The flux components from light hadron decays, charmed hadron decays and downstream hadronic showers are shown in red, blue and green, respectively. The line-styles denote the different event generators. All energies Eν > 10 GeV are included. Shown at the bottom of each panel is the angular coverage of FASERν and SND@LHC. trino production via the decay Λ → pe−ν¯. High-energy ing [39], to provide more reliable predictions for forward forward Λ production in diffractive scattering is suffi- charm production. ciently enhanced, such that hyperon and kaon decays provide roughly equal contributions at FASERν. Decays of charmed hadrons, including both D-mesons D. Angular Distribution and Λc baryons, become the dominant production mode for electron and muon neutrinos at the highest energies. In addition to the energy spectrum, we can also con- Due to the high mass of the tau lepton, tau neutrinos sider the angular distribution of forward LHC neutrinos. are only produced in the decay of the Ds meson and the This is presented in Fig.3, where we show the number of subsequent τ decay. We can see a characteristic double neutrinos per unit area as a function of pseudorapidity peak signature in the energy spectrum, corresponding to for all three neutrino flavours. The pseudorapidity can tau neutrinos from direct Ds decay (lower energy) and τ also be translated into a displacement from the beam decays (higher energies). collisions axis (or Line of Sight) at the detector location, Neutrinos from the decay of secondary particles pro- which is shown as additional horizontal scale. At the bot- duced in downstream hadronic showers only provide a tom of each panel, we also illustrate the angular coverage subdominant contribution. These mainly originate from of both detectors. collisions of forward neutral hadrons or diffractive pro- For all neutrino flavours, the neutrino flux peaks tons with or around the TAN. From a physics point of around the beam collision axis and falls off when moving view, this is a very encouraging and interesting result: it away from it. This is both due to the angular spectrum implies that the neutrino flux can be used as an indirect of the parent hadrons as well as due to the LHC’s long probe of forward hadron production. but narrow beam pipe shape. The number of neutrino In Fig.2, we also show the predictions of different gen- events per detector mass can therefore be maximized for erators, as indicated by the different line styles. We can an experiment placed right on the beam collision axis, see that there are O(1) difference between the generator which is the case for FASERν. Here the neutrino flux is predictions for neutrinos originating from light hadrons. almost constant throughout the detector’s cross sectional In contrast, there are large differences of more than an area. In contrast, SND@LHC is displaced from the beam order of magnitude on the neutrinos flux from charmed collision axis and the neutrino flux drops considerably be- hadron decays. However, as mentioned before, DPMJET tween from the high and low pseudrorapidity ends of the III.2017.1 and Pythia 8.2 have not been tuned and detector. validated with charm production data such that their As before, the different colors corresponds to the differ- predictions should therefore be considered unreliable. ent production mechanisms: light hadron decays in red, Dedicated efforts are needed, and indeed already ongo- charmed hadron decays in blue and downstream hadronic 6

Generators FASERν SND@LHC

light hadrons heavy hadrons νe +ν ¯e νµ +ν ¯µ ντ +ν ¯τ νe +ν ¯e νµ +ν ¯µ ντ +ν ¯τ SIBYLL SIBYLL 1343 6072 21.2 184 965 10.1 DPMJET DPMJET 4614 9198 131 547 1345 22.4 EPOSLHC Pythia8 (Hard) 2109 7763 48.9 367 1459 16.1 QGSJET Pythia8 (Soft) 1437 7162 24.5 259 1328 10.7 +2238 +1649 +74.5 +208 +184 +7.5 Combination (all) 2376−1032 7549−1476 56.4−35.1 339−155 1274−308 14.8−4.7 +479 +763 +17.3 +96 +208 +3.8 Combination (w/o DPMJET) 1630−286 7000−926 31.5−10.3 270−85 1251−285 12.3−2.1

TABLE I. Expected number of neutrino interaction events occurring in FASERν and SND@LHC during LHC Run 3 with 150 fb−1 integrated luminosity. We provide predictions for SIBYLL 2.3c, DPMJET III.2017.1, EPOSLHC/Pythia 8.2 with HardQCD, and QGSJET II-04/Pythia 8.2 with SoftQCD. The two bottom rows provide a combined average, both including and excluding the DPMJET prediction, where the uncertainties correspond to the range of predictions obtained from different MC generators.

FASER SND@LHC 3 e + e e + e 10 + + + +

102

101

100 Interacting Neutrinos [1/bin] 10 1

10 102 103 10 102 103 Neutrino Energy [GeV] Neutrino Energy [GeV]

FIG. 4. Interacting Neutrino Energy Distribution: The panels show the number of neutrinos interacting with the FASERν (left) and SND@LHC (right) detectors during LHC Run 3 with 150 fb−1 integrated luminosity as function of the neutrino energy. The red solid, blue dashed and green dotted lines correspond to electron, muon and tau neutrinos, respectively. The thick line denotes to the average prediction of different generators, while the shaded band corresponds to the range of predictions obtained with different generators. showers in green. The line-styles denote to the different flux with the neutrino cross section presented in Ref [8]. event generators. Similar to what we have seen for the The resulting number of expected neutrino interac- energy spectrum, we observe O(1) differences between tions in FASERν and SND@LHC during LHC Run 3 the MC generator predictions for neutrinos from light with an integrated luminosity are presented in TableI. hadron decays, while the difference for neutrinos from Since not all generators are able to both simulate light charm decay are significantly larger. hadron and charm production, we group them together in four setups: i) SIBYLL 2.3c, ii) DPMJET III.2017.1, iii) EPOSLHC for light hadrons and Pythia 8.2 with IV. NEUTRINO EVENT RATES HardQCD for charmed hadrons, and iv) QGSJET II-04 for light hadrons and Pythia 8.2 with SoftQCD for charmed A. Interactions hadrons. As before, we observe sizable differences be- tween the different MC generator prediction, which are While we have so far concentrated on the number of mainly related to the neutrino flux from charmed hadron neutrinos passing though the detector, let us now discuss decays. The lowest and highest event rates are predicted the number of neutrinos interacting with each detector. by SIBYLL 2.3c and DPMJET III.2017.1, respectively. For this, we convolute the previously obtained neutrino Notably, the predictions for the tau neutrino event rate 7

Beam Configuration FASERν FASERν (adjusted) SND@LHC

Crossing Angle νe +ν ¯e νµ +ν ¯µ ντ +ν ¯τ νe +ν ¯e νµ +ν ¯µ ντ +ν ¯τ νe +ν ¯e νµ +ν ¯µ ντ +ν ¯τ +150µrad vertical 1343 6072 21.2 1442 6520 17.8 184 965 10.1 -150µrad vertical 1326 6111 22.3 1377 6491 21.5 137 677 8.5 +150µrad horizontal 1391 6440 27.8 1486 6597 24.5 176 906 10.0 -150µrad horizontal 1332 6196 23.1 1467 6585 18.1 147 746 9.5 no crossing angle 1443 6807 21.8 — — — 159 827 9.5

TABLE II. Expected number of neutrino interactions in FASERν and SND@LHC during LHC Run 3 with 150 fb−1 integrated luminosity for different orientations of the beam crossing angle. The simulations were performed using Sibyll 2.3c. For FASERν, we provide both predictions for its nominal position centered around x = y = 0 and an adjusted position moved 7.2 cm into the same direction as the beam. in FASERν differ by more than a factor six, ranging from At the detector locations 480 m downstream from the 21 (SIBYLL) to 131 (DPMJET). IP, a beam crossing angle of 150 µrad leads to a shift In the two bottom rows, we combine these four setups of the beam collision axis of 7.2 cm relative to its nomi- to an average prediction, where the uncertainty corre- nal position assuming no crossing angle. Different beam sponds to the range of predictions provided by the dif- crossing angle orientations can therefore shift the neu- ferent generators. Since DPMJET provides significantly trino beam and change the expected neutrino event rate. larger charm production predictions than the other gen- To analyze this effect, we run our simulation for four erators, we also provide a separate average excluding the different beam orientations (upward, downward, left and DPMJET prediction in the last row. We want to emphasize right) with a 150 µrad half crossing angle. For compari- again that neither DPMJET nor Pythia 8.2 have been val- son, we also consider the scenario with perfectly parallel idated or tuned with charm data, and their predictions beams and no crossing angle. can therefore not be considered trustworthy. Better esti- The resulting number of neutrino interactions with the mates for forward charm production are needed. In addi- detectors are shown in TableII, where we have used tion, the range of generator predictions are only a crude Sibyll 2.3c as event generator. We can see that the measure for the neutrino flux uncertainties and should event rates at FASERν, which is centered around the ideally be replaced by individual uncertainty prediction nominal beam collision axis at x = y = 0, are only provided by the generators. marginally effected as expected from its symmetric setup. In Fig.4, we present the energy spectrum of neutrinos In contrast, the event rates at SND@LHC, which is lo- interacting with the detectors. Compared to the energy cated in the x, y > 0 quadrant, sensitively depend on the spectrum presented in Fig.2, this spectrum is shifted to- beam orientation. In particular, when changing from a wards higher energy, since the neutrino interaction cross vertically upward to a downward orientation, the event section roughly increases linearly with the neutrino beam rates drop by a factor of about 2/3. energy. The shaded band corresponds to the range of The FASERν detector also has the option of being predictions obtained from the different MC event gener- moved around its nominal position, such that it can be ators, while the thick central line shows their average. As aligned with the center of the neutrino beam. As also expected, the uncertainties are largest at higher energies shown in TableII, such alignment would slightly increase and for tau neutrinos where the charm production mode the expected event rates. dominates. In Fig.5, we show the energy spectra of electron neu- trinos (dashed lines) and muon neutrinos (solid lines) passing through the detectors for different choices of B. Crossing Angles the crossing angles. The lower panels show the change with respect to the baseline orientation vertically up- To avoid parasitic collisions in the beam pipe away wards. For FASERν, we show results for beam orienta- from the IP, the LHC’s proton beams have a small beam tions +150 µrad vertically and horizontally for the nomi- crossing angle when they collide. So far, we have assumed nal detector location centered around x = y = 0, and the a beam configuration at the ATLAS IP similar to the adjusted location centered around x or y = 7.2 cm. A end of LHC Run 2 with a beam half-crossing angle of mildly increased flux of neutrino high energy neutrinos about 150 µrad vertically upwards. To distribute the with E & 1 TeV is observed for the adjusted detector collision debris more evenly and allow for a longer lifetime location compareds to the nominal one. For SND@LHC, of the LHC’s focusing magnets, changing orientations of we consider all four orientations of beam crossing angle. the beam crossing angle are considered for Run 3 of the A sizable reduction of the neutrino flux at high energies LHC. compared to the upward beam orientation is observed for 8

1012 FASER using SIBYLL 2.3c SND@LHC using SIBYLL 2.3c +150 rad v. +150 rad v. (moved) +150 rad v. -150 rad v. +150 rad h. +150 rad h. (moved) +150 rad h. -150 rad h.

1011

10 10 e Neutrinos [1/bin]

e

109 ) . v

d 1.5 a r 0

5 1.0 1 ( e

/ 0.5

e ) . v

d 1.5 a r 0

5 1.0 1 (

/ 0.5

10 102 103 10 102 103 Neutrino Energy [GeV] Neutrino Energy [GeV]

FIG. 5. Beam Configuration: We show the neutrino flux going through FASERν (left) and SND@LHC (right) for different orientations of the beam crossing angle considered for LHC Run 3. The simulations were performed using Sibyll 2.3c. The muon neutrino and electron neutrino fluxes are shown as solid and dashed lines, respectively. The lower panels show the change of the neutrino flux relative to the nominal setup with an angle of 150 µrad vertically upwards as shown in black. For FASERν, we also show results for a detector that was moved to be aligned with the center of the neutrino beam. horizontal or downward setups. However, while a change In this work, we presented an alternative fast neutrino of the beam orientation reduced the overall event rate, it flux simulation, implemented as a RIVET module, which also allows SND@LHC to probe the neutrino spectrum is accessible to the entire community and can be run on in a larger effective pseudorapidity range. a personal computer within minutes. We have used this simulation to obtain the neutrino flux going through both neutrino detectors for all flavours V. SUMMARY using six commonly used event generators. We have presented the energy and pseudorapidity spectrum for Starting in 2022, two dedicated experiments to de- the neutrino flux, and found sizable difference between tect neutrinos produced in LHC collisions, FASERν and the generator predictions, especially for neutrinos from SND@LHC, will start to take data in the far-forward re- charmed hadron decays. We note again that dedicated gion of the LHC. This emerging LHC neutrino physics efforts are needed, and have already started [39], to pro- program requires reliable estimates of the LHC’s for- vide more reliable predictions for this production mode. ward neutrino fluxes and their uncertainties. Such es- This module is also well suited for phenomenological timates currently rely on sophisticated but rather com- studies and applications. As an example, we have studied putationally expansive and time consuming computation the impact of different beam crossing angle orientations performed by dedicated groups using BDSIM or FLUKA. as well as detector locations for FASERν. Since the pre- 9 sented neutrino flux simulation is already implemented dated BDSIM configurations and his help to validate this as a RIVET module, it would also be ideally suited for simulation. We are also grateful to the authors and main- generator tuning applications in the future when data be- tainers of many open-source software packages, including comes available. As we have seen, the neutrino flux for CRMC [36], RIVET [14], scikit-hep [45], and uproot [46]. different flavours, at different energies and in different F.K. is supported by the Department of Energy under rapidity regions (or experiments) is sensitive to differ- Grant No. DE-AC02-76SF00515. ent flux components and hence provides information on the forward production for various parent hadron species. Given the high expected event rates, this would be valu- Appendix A: Event Generator Comparison for able data to constrain forward particle production, which Forward Particle Production plays an important role in astroparticle physics [40, 41]. For example, this data could help to understand the ob- In the main part of this paper, we have compared the served excess of muons in cosmic-ray air showers [42, 43] neutrino flux predictions provided by different MC event and to constrain the prompt atmospheric neutrino back- generators. In this appendix, we also compare generator ground at large-scale neutrino telescopes [44]. predictions for forward hadrons production rates directly. This is done in Fig.6 for three most important + + hadrons, π (top), KL (center) and D (bottom), in three different far-forward pseudorapidity bins. These results are consistent with our previous findings. The ACKNOWLEDGEMENTS generator predictions are consistent up to an O(1) fac- tor for light hadron production, although sizable shape We thank Akitaka Ariga, Tomoko Ariga, Tomohiro In- difference can be seen. In contrast, there are large dif- ada, Max Fieg and Holger Schulz for useful discussions. ferences of an order of magnitude and more for charm We particularly thank Laurie Nevay for providing the up- production at high pseudorapidities η > 8.

[1] A. De Rujula and R. Ruckl, “Neutrino and muon arXiv:2002.08722 [physics.ins-det]. physics in the collider mode of future accelerators,” in [11] C. Ahdida et al., “SND@LHC - Scattering and Neutrino SSC Workshop: Superconducting Super Collider Fixed Detector at the LHC,” tech. rep., CERN, Geneva, Jan, Target Physics, pp. 571–596. 5, 1984. 2021. https://cds.cern.ch/record/2750060. [2] F. Vannucci, “Neutrino physics at LHC/SSC,” Tech. [12] L. J. Nevay et al., “BDSIM: An accelerator tracking Rep. LPNHE-93-03, Paris 6. Lab. Phys. Nucl. Th´eor. code with particle–matter interactions,” Comput. Phys. Hautes En´erg.,Paris, Aug, 1993. Commun. 252 (2020) 107200, arXiv:1808.10745 https://cds.cern.ch/record/253670. [physics.comp-ph]. [3] A. De Rujula, E. Fernandez, and J. J. Gomez-Cadenas, [13] A. Ferrari, P. R. Sala, A. Fasso, and J. Ranft, “FLUKA: “Neutrino fluxes at future hadron colliders,” Nucl. A multi-particle transport code (Program version Phys. B405 (1993) 80–108. 2005),”. [4] H. Park, “The estimation of√ neutrino fluxes produced by [14] A. Buckley, J. Butterworth, D. Grellscheid, H. Hoeth, proton-proton collisions at s = 14 TeV of the LHC,” L. Lonnblad, J. Monk, H. Schulz, and F. Siegert, “Rivet JHEP 10 (2011) 092, arXiv:1110.1971 [hep-ex]. user manual,” Comput. Phys. Commun. 184 (2013) [5] J. L. Feng, I. Galon, F. Kling, and S. Trojanowski, 2803–2819, arXiv:1003.0694 [hep-ph]. “ForwArd Search ExpeRiment at the LHC,” Phys. Rev. [15] C. Bierlich et al., “Robust Independent Validation of D 97 (2018) no. 3, 035001, arXiv:1708.09389 Experiment and Theory: Rivet version 3,” SciPost [hep-ph]. Phys. 8 (2020) 026, arXiv:1912.05451 [hep-ph]. [6] S. Buontempo, G. M. Dallavalle, G. De Lellis, D. Lazic, [16] T. Sj¨ostrand,S. Ask, J. R. Christiansen, R. Corke, and F. L. Navarria, “CMS-XSEN: LHC Neutrinos at N. Desai, P. Ilten, S. Mrenna, S. Prestel, C. O. CMS. Experiment Feasibility Study,” Rasmussen, and P. Z. Skands, “An Introduction to arXiv:1804.04413 [physics.ins-det]. PYTHIA 8.2,” Comput. Phys. Commun. 191 (2015) [7] FASER Collaboration, H. Abreu et al., “First neutrino 159–177, arXiv:1410.3012 [hep-ph]. interaction candidates at the LHC,” arXiv:2105.06197 [17] GEANT4 Collaboration, S. Agostinelli et al., [hep-ex]. “GEANT4–a simulation toolkit,” Nucl. Instrum. Meth. [8] FASER Collaboration, H. Abreu et al., “Detecting and A 506 (2003) 250–303. Studying High-Energy Collider Neutrinos with FASER [18] M. Dobbs and J. B. Hansen, “The HepMC C++ Monte at the LHC,” Eur. Phys. J. C 80 (2020) no. 1, 61, Carlo event record for High Energy Physics,” Comput. arXiv:1908.02310 [hep-ex]. Phys. Commun. 134 (2001) 41–46. [9] FASER Collaboration, H. Abreu et al., “Technical [19] J. L. Feng, I. Galon, F. Kling, and S. Trojanowski, Proposal: FASERnu,” arXiv:2001.03073 “Dark Higgs bosons at the ForwArd Search [physics.ins-det]. ExpeRiment,” Phys. Rev. D97 (2018) no. 5, 055034, [10] SHiP Collaboration, C. Ahdida et al., “SND@LHC,” arXiv:1710.09387 [hep-ph]. 10

+ : 7 < < 8 + : 8 < < 9 + : 9 < < 10 SIBYLL 2.3c 10 EPOSLHC 10 QGSJET II-04 DPMJET 3.2017 Pythia8 (Soft)

109

108 Production Rate [pb/bin] 107

KL: 7 < < 8 KL: 8 < < 9 KL: 9 < < 10 SIBYLL 2.3c 9 EPOSLHC 10 QGSJET II-04 DPMJET 3.2017 Pythia8 (Soft)

108

107 Production Rate [pb/bin] 106

D + : 7 < < 8 D + : 8 < < 9 D + : 9 < < 10 SIBYLL 2.3c DPMJET 3.2017 Pythia8 (Soft) Pythia8 (Hard) 107

106 Production Rate [pb/bin] 105

102 103 102 103 102 103 Hadron Energy [GeV] Hadron Energy [GeV] Hadron Energy [GeV]

+ + FIG. 6. Predictions for the energy spectra of π (top row), KL (center row) and D (bottom row) in three pseudorapidity intervals using six commonly used MC generators.

[20] J. L. Feng, I. Galon, F. Kling, and S. Trojanowski, D99 (2019) no. 1, 015021, arXiv:1810.01879 “Axionlike particles at FASER: The LHC as a photon [hep-ph]. beam dump,” Phys. Rev. D98 (2018) no. 5, 055021, [23] FASER Collaboration, A. Ariga et al., “Letter of arXiv:1806.02348 [hep-ph]. Intent for FASER: ForwArd Search ExpeRiment at the [21] F. Kling and S. Trojanowski, “Heavy Neutral Leptons LHC,” arXiv:1811.10243 [physics.ins-det]. at FASER,” Phys. Rev. D97 (2018) no. 9, 095016, https://cds.cern.ch/record/2642351. Submitted to arXiv:1801.08947 [hep-ph]. the CERN LHCC on 18 July 2018. [22] A. Berlin and F. Kling, “Inelastic Dark Matter at the [24] FASER Collaboration, A. Ariga et al., “FASER’s LHC Lifetime Frontier: ATLAS, CMS, LHCb, physics reach for long-lived particles,” Phys. Rev. D99 CODEX-b, FASER, and MATHUSLA,” Phys. Rev. (2019) no. 9, 095011, arXiv:1811.12522 [hep-ph]. 11

[25] FASER Collaboration, A. Ariga et al., “Technical https://web.ikp.kit.edu/rulrich/crmc.html. Proposal for FASER: ForwArd Search ExpeRiment at [37] T. Sjostrand, S. Mrenna, and P. Z. Skands, “PYTHIA the LHC,” arXiv:1812.09139 [physics.ins-det]. 6.4 Physics and Manual,” JHEP 05 (2006) 026, [26] FASER Collaboration, A. Ariga et al., “FASER: arXiv:hep-ph/0603175 [hep-ph]. ForwArd Search ExpeRiment at the LHC,” [38] P. Skands, S. Carrazza, and J. Rojo, “Tuning PYTHIA arXiv:1901.04468 [hep-ex]. 8.1: the Monash 2013 Tune,” Eur. Phys. J. C74 (2014) [27] F. Kling and S. Trojanowski, “Looking forward to test no. 8, 3024, arXiv:1404.5630 [hep-ph]. the KOTO anomaly with FASER,” Phys. Rev. D 102 [39] W. Bai, M. Diwan, M. V. Garzelli, Y. S. Jeong, and (2020) no. 1, 015032, arXiv:2006.10630 [hep-ph]. M. H. Reno, “Far-forward neutrinos at the Large [28] T. Pierog, I. Karpenko, J. M. Katzy, E. Yatsenko, and Hadron Collider,” arXiv:2002.03012 [hep-ph]. K. Werner, “EPOS LHC: Test of collective [40] R. Engel, D. Heck, and T. Pierog, “Extensive air hadronization with data measured at the CERN Large showers and hadronic interactions at high energy,” Ann. Hadron Collider,” Phys. Rev. C92 (2015) 034906, Rev. Nucl. Part. Sci. 61 (2011) 467–489. arXiv:1306.0121 [hep-ph]. [41] L. A. (for the CF7 SNOWMASS Working Group), [29] S. Ostapchenko, “Monte Carlo treatment of hadronic Synergy of astro-particle physics and collider physics, interactions in enhanced Pomeron scheme: I. Aug., 2020. QGSJET-II model,” Phys. Rev. D83 (2011) 014018, https://doi.org/10.5281/zenodo.4009452. arXiv:1010.1869 [hep-ph]. [42] EAS-MSU, IceCube, KASCADE-Grande, [30] S. Roesler, R. Engel, and J. Ranft, “The Monte Carlo NEVOD-DECOR, Pierre Auger, SUGAR, event generator DPMJET-III,” in International Telescope Array, Yakutsk EAS Array Conference on Advanced Monte Carlo for Radiation Collaboration, H. Dembinski et al., “Report on Tests Physics, Particle Transport Simulation and Applications and Measurements of Hadronic Interaction Properties (MC 2000). 12, 2000. arXiv:hep-ph/0012252. with Air Showers,” EPJ Web Conf. 210 (2019) 02004, [31] A. Fedynitch, Cascade equations and hadronic arXiv:1902.08124 [astro-ph.HE]. interactions at very high energies. PhD thesis, KIT, [43] D. Soldin et al., “Studies of the Muon Excess in Cosmic Karlsruhe, Dept. Phys., 11, 2015. Ray Air Showers.” https://www.snowmass21.org/ [32] E.-J. Ahn, R. Engel, T. K. Gaisser, P. Lipari, and docs/files/summaries/CF/SNOWMASS21-CF7_CF0-EF6_ T. Stanev, “Cosmic ray interaction event generator EF7-AF4_AF0_Dennis_Soldin-083.pdf. SIBYLL 2.1,” Phys. Rev. D80 (2009) 094003, [44] M. Bustamante et al., “Cosmic Neutrino Probes of arXiv:0906.4113 [hep-ph]. Fundamental Physics.” https://www.snowmass21.org/ [33] F. Riehn, R. Engel, A. Fedynitch, T. K. Gaisser, and docs/files/summaries/CF/SNOWMASS21-CF7_CF1-NF4_ T. Stanev, “A new version of the event generator NF3-TF11_TF0_Mauricio_Bustamante-048.pdf. Sibyll,” PoS ICRC2015 (2016) 558, arXiv:1510.00568 [45] E. Rodrigues, “The Scikit-HEP Project,” in 23rd [hep-ph]. International Conference on Computing in High Energy [34] F. Riehn, H. P. Dembinski, R. Engel, A. Fedynitch, and Nuclear Physics (CHEP 2018) Sofia, Bulgaria, July T. K. Gaisser, and T. Stanev, “The hadronic interaction 9-13, 2018. 2019. arXiv:1905.00002 model SIBYLL 2.3c and Feynman scaling,” PoS [physics.comp-ph]. ICRC2017 (2018) 301, arXiv:1709.07227 [hep-ph]. [46] J. Pivarski, P. Das, D. Smirnov, C. Burr, M. Feickert, [35] A. Fedynitch, F. Riehn, R. Engel, T. K. Gaisser, and N. Biederbeck, J. Rembser, H. Schreiner, H. Dembinski, T. Stanev, “Hadronic interaction model sibyll 2.3c and benkrikler, M. R., D. Marley, R. Turra, N. Smith, inclusive lepton fluxes,” Phys. Rev. D100 (2019) no. 10, M. Peresano, and C. Gu, “scikit-hep/uproot: 3.7.2,” 103018, arXiv:1806.04140 [hep-ph]. June, 2019. [36] C. Baus, T. Pierog, and R. Ulrich, “Cosmic Ray Monte https://doi.org/10.5281/zenodo.3256257. Carlo (CRMC),”.