Collisions between oxygen and neon nuclei at CERN’s Large Hadron Collider produce long-range patterns in the motion of particles, and the differences between the two systems become especially pronounced in the most central collisions. Measurements by the CMS Collaboration show that these patterns are sensitive to the initial geometry of the colliding nuclei, providing new constraints on models of nuclear structure and collective behavior in small collision systems.
Comparing oxygen and neon collisions
When atomic nuclei collide at extremely high energies, the particles produced in the collision do not simply fly away independently. Their directions can show correlations that extend across large ranges of pseudorapidity, a quantity used in particle physics to describe direction relative to the beam axis.
These correlations are described using flow harmonics. The second harmonic, called elliptic flow or v₂, measures an elliptical pattern in the particles’ azimuthal distribution. The third harmonic, v₃, describes a triangular pattern.
Such collective flow has been studied extensively in collisions involving large nuclei such as lead and gold. Hydrodynamic models can describe these systems by connecting the geometry of the initial collision to the collective response of the hot matter produced afterward.
Similar flowlike signals have also been observed in much smaller collision systems, including proton-proton and proton-nucleus collisions. But those systems are harder to interpret because the initial geometry is strongly affected by fluctuations and by the internal structure of the proton. Symmetric collisions between light nuclei offer a different experimental situation. Their initial geometry is more directly related to how nucleons are distributed inside the colliding nuclei.
That makes oxygen and neon particularly useful for comparison.
The nuclei used in this study were oxygen-16 (¹⁶O) and neon-20 (²⁰Ne). Their mass numbers are similar, so similar hydrodynamic evolution is expected. Differences in their measured flow can therefore provide information about differences in their initial collision geometry.
Modern ab initio nuclear-structure calculations predict different intrinsic spatial arrangements for the two nuclei. Oxygen-16 has been predicted to have a tetrahedral structure, while neon-20 has been predicted to have a more elongated, “bowling pin” shape. These calculations also allow for correlations in which nucleons group into alpha-like clusters containing two protons and two neutrons.
Earlier hydrodynamic calculations predicted that the ratio of elliptic flow in neon-neon collisions to that in oxygen-oxygen collisions would be particularly sensitive to the structure of neon-20. The new CMS measurement directly tests that expectation.
CMS measured flow in two ways
The CMS Collaboration analyzed oxygen-oxygen and neon-neon collisions recorded at the LHC in July 2025. The collisions had a center-of-mass energy per nucleon pair of 5.36 TeV.
The oxygen-oxygen sample corresponded to an integrated luminosity of 7 inverse nanobarns, while the neon-neon sample corresponded to 0.8 inverse nanobarns.
CMS measured correlations between charged particles over a wide pseudorapidity range. For the two-particle analysis, the selected charged particles had transverse momenta between 0.3 and 3 GeV and pseudorapidity within 2.4 of the collision axis.
The analysis focused on pairs separated by more than two units of pseudorapidity. This large separation suppresses short-range correlations, including effects associated with jet fragmentation and resonance decays. The remaining long-range correlations can then be used to extract the v₂ and v₃ harmonics.
CMS also measured v₂ using correlations among four particles rather than two. Four-particle correlations have a different sensitivity to event-by-event fluctuations in the collision geometry and suppress short-range two-particle effects from jets and resonance decays.
The researchers also accounted for residual correlations from back-to-back dijets. They estimated this contribution using peripheral collisions and found that the correction reduced the measured two-particle v₂ and v₃ values by about 4% to 15%, with the correction becoming smaller toward central collisions.
Both collision systems show elliptic and triangular flow
Significant v₂ and v₃ signals were observed in both oxygen-oxygen and neon-neon collisions. The behavior of the two harmonics with collision centrality was different.
The elliptic-flow coefficient v₂ increased from peripheral collisions toward midcentral collisions and then decreased in the most central events. The measured pattern is similar to that seen in lead-lead collisions and is consistent with the changing geometry of the overlapping region as the two nuclei collide with different degrees of overlap.
The triangular-flow coefficient v₃, in contrast, increased toward more central collisions. The study attributes this increase to a stronger hydrodynamic response as the volume of the collision overlap grows. The authors note that the behavior is consistent with peripheral lead-lead collisions where the system size is comparable.
The four-particle v₂ values were systematically smaller than the two-particle values across the measured centrality range. The study identifies the difference as an expected consequence of the different contributions from event-by-event geometry fluctuations to the two measurements.
The precision of these measurements also provides a way to test calculations of the initial collision geometry and the subsequent hydrodynamic response.
The neon-to-oxygen comparison reveals a strong central-collision effect
The most important comparison comes from taking the flow coefficients measured in neon-neon collisions and dividing them by the corresponding coefficients measured in oxygen-oxygen collisions.
This ratio helps focus attention on differences between the two nuclei rather than on effects common to both collision systems.
The v₂ ratio showed a significant increase toward the most central collisions in both the two-particle and four-particle measurements. The v₃ ratio behaved differently, decreasing toward the most central events.
The central rise in the v₂ ratio is consistent with the predicted difference in the intrinsic structures of oxygen-16 and neon-20. The study describes neon-20 as having stronger intrinsic spatial correlations and therefore a less spherical initial matter distribution in central collisions than oxygen-16.
The result does not by itself provide a quantitative measurement of the deformation of neon-20. The researchers explicitly state that such a determination would require further dedicated studies because the agreement between the measurements and the models is not uniform across the full centrality range.
The measured v₃ ratio presents a different challenge. None of the three model calculations examined in the study reproduced the measured v₃ ratio quantitatively. The authors identify this as evidence that improved treatment of initial-state fluctuations is needed in models of small collision systems.
Nuclear-structure models reproduce the main trends
CMS compared the measurements with hydrodynamic calculations that incorporate modern nuclear-structure inputs.
One framework, called Trajectum, used two different ab initio descriptions of the nucleon configurations in oxygen and neon. One was based on nuclear lattice effective field theory, while the other used the projected generator coordinate method. The initial energy densities were generated with the Trento model and then evolved using hydrodynamics.
For the individual v₂ measurements, the nuclear-lattice calculation reproduced the central-collision results better than the projected-generator-coordinate calculation, while both agreed with the data in more peripheral collisions. Both versions described the measured v₃ values across the centrality range.
A second calculation, based on the IP-Glasma framework and followed by MUSIC hydrodynamics and hadronic interactions, tended to predict larger v₂ and v₃ values than measured in central collisions. In more peripheral events, however, its results agreed with the measurements within uncertainties.
The comparison changes somewhat when the neon-to-oxygen ratios are considered.
For the v₂ ratio, the Trajectum calculation using the projected generator coordinate method gave a closer description of the measurements. The nuclear-lattice version tended to overestimate the ratio. Both calculations nevertheless reproduced the qualitative increase of the v₂ ratio toward central collisions.
The IP-Glasma calculation predicted a weaker centrality dependence of the v₂ ratio. The study notes that this may result from large multiplicity fluctuations that weaken the correlation between collision centrality and impact parameter.
For v₃, none of the three calculations reproduced the measured neon-to-oxygen ratio quantitatively.
The measurements constrain both collision geometry and collective response
The measurements provide two related pieces of information.
First, the individual v₂ and v₃ measurements show how collective flow changes with the geometry and size of the overlapping region in oxygen-oxygen and neon-neon collisions. Elliptic flow rises toward midcentral collisions and then falls in the most central events, while triangular flow rises toward central collisions.
Second, comparing neon with oxygen highlights differences associated with their initial nuclear structures. The strong increase in the v₂ ratio toward central collisions is qualitatively consistent with the predicted deformation of neon-20, while the decreasing v₃ ratio provides a separate test that current calculations do not yet reproduce quantitatively.
The measurements therefore place constraints on models that attempt to connect the spatial arrangement of nucleons inside a nucleus with the collective behavior produced after an ultrarelativistic collision. The study concludes that hydrodynamic models containing ab initio nuclear-structure inputs capture the qualitative trends of both the flow measurements and their neon-to-oxygen ratios, although they do not fully reproduce the measured flow magnitudes.
The study was published in Physical Review Letters.
