A theoretical model predicts that the strongest charged-hadron production in proton-oxygen collisions shifts sides as the collisions become more peripheral. In the most central events, the maximum is predicted on the oxygen-going, or backward, side. In more peripheral events, it moves toward the proton-going, or forward, side.
The result comes from an analysis of p-O collisions at 9.618 TeV that compares earlier charged-hadron production predictions with preliminary Run 3 measurements from the ALICE experiment. The study adapts the model parameters to those data across eight centrality classes.
A three-part picture of the collision
The calculation treats particle production as the combined output of three sources: a central source linked to gluon-gluon interactions, plus two fragmentation sources associated with valence quarks and soft gluons moving in the proton-going and oxygen-going directions. The model also includes color-glass-condensate starting conditions and a relativistic-diffusion treatment of partial thermalization.
Where measurements were available, the charged-hadron yields in the two fragmentation sources were adjusted as free parameters using chi-square or least-square fits. The two fragmentation distributions were calculated with a C++ finite-element code that was later modified for the p-O prediction.
The centrality inputs change markedly across the analysis. The mean number of participants falls from 6.9 in the 0-5% class to 2.3 in the 60-90% class, while the saturation-scale parameter drops from 0.031 to 0.010 GeV2.
The fitted picture changes with centrality
Between the most central and most peripheral classes, the reported p-going transport parameter rises from 3.6 to 19.5. Over the same span, the central-to-fragmentation ratio falls from 4.98 to 1.66, the O-going-to-p-going ratio falls from 0.71 to 0.18, and the total charged-hadron yield falls from 468 to 81.
That pattern is reflected in the source amplitudes. In peripheral collisions, the p-going amplitude becomes larger than the O-going amplitude, while the central gluon-gluon source keeps a slight preference for the O-going direction. In the model, those source patterns are accompanied by a shift in the overall maximum toward the proton-going side.
The authors interpret the pattern in terms of the fields represented in the model. They associate the inversion with a stronger oxygen-going gluon field interacting with forward-proton valence quarks, compared with a weaker proton gluon field in the oxygen overlap zone. The paper presents this as a model-level interpretation, not an experimental demonstration of the mechanism.
A prediction still waiting for a fuller test
The comparison is not yet a final test. The ALICE inputs are preliminary, and the authors say they are avoiding precise quantitative conclusions because the results could change after publication. The earlier prediction also had a lower normalization: the preliminary ALICE data were 1.72 times higher, so the authors renormalized the distributions to the preliminary midrapidity data in all eight centrality classes.
The resulting distributions are anchored to the preliminary midrapidity data, so their absolute normalization is not an independent model prediction in this comparison.
The predicted inversion also needs measurements farther from midrapidity, where the directional difference can be assessed. The authors say the preferred coverage would extend to , where eta denotes pseudorapidity, the vertical bars mean its absolute value, and the less-than sign sets the upper limit at 5.
What would settle the prediction
The key test is whether final p-O measurements confirm the movement of the production maximum from the backward oxygen-going direction toward the forward proton-going direction. Broader pseudorapidity coverage is required to test that prediction.
The manuscript is a preprint, identified as arXiv version 2 dated 14 September 2026. It states that the preliminary ALICE data supporting the findings are available in its reference 10.
Paper data and sources
Original title: Particle production in $p$-O collisions at LHC energy
Authors: Chuan Li, Georg Wolschin
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text