WorksPhysics and Astronomy
Light (anti)nuclei production via coalescence in electron-proton collisions: from HERA to EIC
Abstract
Light (anti)nuclei production provides a sensitive probe of the particle-emitting source and of the coalescence mechanism in high-energy collisions. In this work, a coalescence model implemented as an afterburner to PYTHIA 8 is validated against pp data from ALICE and ep data from ZEUS. The validated framework is then applied to ep collisions at EIC energies to predict deuteron and antideuteron production within the ePIC acceptance. The results indicate sizeable yields and a clear matter-antimatter asymmetry in the proton-going direction, making this measurement a promising observable for studying coalescence and proton structure at the EIC.
Introduction
Light nuclei and antinuclei are produced in high-energy collisions through the coalescence of nucleons sufficiently close in phase space at kinetic freeze-out [1, 2, 3]. Their production depends on the nucleon-emitting source and the nuclear wave function, providing a sensitive probe of the hadronization environment. A proton and a neutron can coalesce into a deuteron, while the corresponding antinucleons can form an antideuteron. If a bound state is not formed, the relative momentum distribution of the outgoing nucleons retains information on the particle-emitting source and can be studied through femtoscopic correlations [4].
The coalescence probability depends on the nucleon phase-space distribution and, in particular, on the pair average transverse momentum , the source size, and the spatial extent of the nucleus [2, 5]. Increasing the source size reduces the probability of forming a bound state because the overlap with the nuclear wave function becomes smaller. This behaviour is supported by ALICE measurements showing a clear dependence of light-nucleus production on the emitting-system size [6].
Deep-inelastic electron–proton and electron–nucleus collisions provide a complementary environment to study this mechanism. Their cleaner and better-constrained initial state makes them suitable for testing the universality of coalescence across collision systems. Deuteron and antideuteron production in the proton-going region can also probe the matter–antimatter asymmetry of the nucleon source and may provide sensitivity to proton and nuclear parton distribution functions.
In this work, a coalescence model is validated using pp and ep measurements and applied to predict deuteron and antideuteron production at the future Electron–Ion Collider.
Simulation strategy and coalescence model
To simulate light (anti)nuclei production, Pythia8 is used to generate collision events, followed by a custom afterburner implementing the coalescence model [7]. The afterburner operates event-by-event on final-state nucleons and evaluates their probability to form a bound state.
The nucleon-emitting source is described by a Gaussian distribution multiplied by a plane wave, , where is the normalization amplitude, the relative nucleon distance, and the source radius. The pair relative momentum in its rest frame is , where and are the nucleon momenta in the pair rest frame.
The (anti)deuteron ground-state wave function is obtained by modelling the strong nucleon–nucleon interaction with a spherical square-well potential, , where is the (anti)deuteron radius and the square-well depth. For each , is fixed to reproduce the (anti)deuteron binding energy, .
The coalescence probability is obtained from the projection of the source wave function onto the (anti)deuteron ground-state wave function, .
The factor 3/8 accounts for the spin–isospin configuration required to form a spin-1 (anti)deuteron from two spin-1/2 nucleons. The free parameters are the source radius and the (anti)deuteron radius , fixed by tuning to experimental data.
Results
Validation
The model was first tuned using pp collision data and validated against ALICE measurements [8]. The observable is the coalescence parameter , corresponding to the case of the general parameter . Assuming identical proton and neutron momentum distributions,
For deuterons and antideuterons, , and provides an experimental measure of bound-state formation, allowing comparison across collision systems.
The validation uses pp collisions at TeV generated with Pythia8 Monash 2013 and processed with the coalescence afterburner. The source radius and the (anti)deuteron radius are treated as free parameters and fixed through tuning.
As shown in Figure 1, increases with . The model reproduces the ALICE measurements within experimental uncertainties over the full measured range, with the simulation-to-data ratio compatible with unity.

Figure 1. Coalescence parameter for antideuteron production in pp collisions at TeV as a function of . Simulations are obtained from pp collisions generated with Pythia8 Monash 2013 and processed with the coalescence afterburner, using fm and fm. Filled markers represent ALICE data [8], while open markers show the model results. The lower panel reports the simulation-to-data ratio.
The model was then tested against HERA ep data using ZEUS measurements [9]. The pp-tuned parameters, fm and fm, are used to evaluate for deuteron and antideuteron production at GeV.
As shown in Figure 2, the predictions are consistent with ZEUS data within experimental uncertainties and reproduce the overall magnitude of the measured values for deuterons and antideuterons. This supports the applicability of the coalescence framework to ep collisions.

Figure 2. Coalescence parameter for deuteron and antideuteron production in ep collisions at GeV as a function of . Filled markers represent ZEUS data [9], while open markers show predictions obtained with Pythia8 and the coalescence afterburner using fm and fm. The lower panel reports the simulation-to-data ratio.
The comparison with HERA is particularly relevant for testing the universality of coalescence across collision systems. Previous ALICE measurements compared the coalescence parameter across several systems, including pp, pA, p, and ep collisions [10]. HERA provides an important ep reference, although with limited statistical precision. The future Electron–Ion Collider (EIC), with in the first year of data taking, will allow a significantly more precise measurement.
Predictions
Building on the validated framework, the model is applied to ep collisions at EIC energies to predict deuteron and antideuteron production as a new observable for studying light-nuclei formation in deep-inelastic scattering.
Figure 4 shows the predicted deuteron and antideuteron yields as a function of pseudorapidity. A clear asymmetry is observed, with a larger deuteron yield in the proton-going direction. This behaviour originates from the valence-quark structure of the proton and makes the measurement potentially sensitive to proton parton distribution functions.

Figure 4. Predicted deuteron, antideuteron, and difference yields as a function of pseudorapidity in ep collisions at GeV for GeV. Dashed vertical lines indicate the ePIC dRICH acceptance, . Simulation: ep collisions at GeV and GeV generated with Pythia8 Monash and processed with the coalescence afterburner, using fm and fm.
The region , highlighted in Figure 4, corresponds to the ePIC dRICH acceptance. Within this region, the model predicts about deuterons and antideuterons in the first year of data taking. The forward region is therefore particularly interesting for the expected yields and the sensitivity of the nuclei–antinuclei asymmetry to proton structure.
Deuteron and antideuteron identification with the ePIC dRICH is illustrated in Figure 3. The detector covers and employs an aerogel radiator with refractive index –. For , the proton and deuteron Cherenkov momentum thresholds are and , while for they decrease to about and .

Figure 3. Transverse-momentum versus pseudorapidity distribution of deuterons produced via coalescence in ep collisions within the ePIC dRICH acceptance. The top and bottom panels correspond to aerogel refractive indices and , respectively. Dashed lines indicate the proton and deuteron Cherenkov thresholds. Above the deuteron threshold, identification is based on the Cherenkov angle; between the two thresholds, a hadron-veto technique can be used.
Two particle-identification regimes can be exploited. Above the deuteron Cherenkov threshold, deuterons and antideuterons can be identified through the Cherenkov angle. Between the proton and deuteron thresholds, identification can rely on a veto technique: Cherenkov emission indicates compatibility with a proton (or a lower mass particle), while its absence for a reconstructed charged track is consistent with a deuteron.
Conclusions
A coalescence model coupled to Pythia8 has been validated against ALICE pp measurements and ZEUS ep data, showing a consistent description of deuteron and antideuteron production across collision systems. At EIC energies, the model predicts large deuteron and antideuteron samples within the ePIC dRICH acceptance and a pronounced matter–antimatter asymmetry in the proton-going region. Light-nuclei production can therefore provide a new observable for testing coalescence and probing proton structure at the EIC.
References
- [1]J. I. Kapusta, Mechanisms for deuteron production in relativistic nuclear collisions, Phys. Rev. C 21 (1980) 1301.DOI
- [2]R. Scheibl and U. W. Heinz, Coalescence and flow in ultrarelativistic heavy ion collisions, Phys. Rev. C 59 (1999) 1585–1602.DOI
- [3]K. Blum, K. C. Y. Ng, R. Sato and M. Takimoto, Cosmic rays, antihelium, and an old navy spotlight, Phys. Rev. D 96 (2017) 103021 [arXiv:1704.05431].DOI
- [4]N. Ciavarelli, F. Ercolessi, F. Noferini and F. Bellini, Light antinuclei in pp collisions at the LHC: production by coalescence and interaction of antinucleons, Il Nuovo Cimento C 49 (2026) 157.DOI
- [5]F. Bellini and A. P. Kalweit, Testing production scenarios for (anti-)(hyper-)nuclei and exotica at energies available at the CERN Large Hadron Collider, Phys. Rev. C 99 (2019) 054905 [arXiv:1807.05894].DOI
- [6]S. Acharya et al. [ALICE Collaboration], Light (anti)nuclei production in Pb–Pb collisions at √sNN = 5.02 TeV, Phys. Rev. C 107 (2023) 064904 [arXiv:2211.14015].
- [7]M. Mahlein, L. Barioglio, F. Bellini, L. Fabietti, C. Pinto, B. Singh and S. Tripathy, A realistic coalescence model for deuteron production, Eur. Phys. J. C 83 (2023) 804 [arXiv:2302.12696].
- [8]S. Acharya et al. [ALICE Collaboration], (Anti-)deuteron production in pp collisions at √s = 13 TeV, Eur. Phys. J. C 80 (2020) 889 [arXiv:2003.03184].
- [9]S. Chekanov et al. [ZEUS Collaboration], Measurement of (anti)deuteron and (anti)proton production in DIS at HERA, Nucl. Phys. B 786 (2007) 181–205 [arXiv:0705.3770].DOI
- [10]S. Acharya et al. [ALICE Collaboration], Production of deuterons, tritons, ³He nuclei and their antinuclei in pp collisions at √s = 0.9, 2.76 and 7 TeV, Phys. Rev. C 97 (2018) 024615 [arXiv:1709.08522].