⚛️ nuclear experiments

Evidence for sequential $\Upsilon$ (nS) suppression in light ion collisions

This paper presents the first measurements of $\Upsilon$ (1S), $\Upsilon$ (2S), and $\Upsilon$ (3S) production in oxygen-oxygen and neon-neon collisions at $\sqrt{s_\mathrm{NN}}$ = 5.36 TeV, revealing a significant sequential suppression pattern where excited states are more strongly suppressed than the ground state, providing evidence for quark-gluon plasma formation in light ion collisions.

Original authors: CMS Collaboration

Published 2026-07-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: CMS Collaboration

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Cosmic Soup and the Melting Ice Cubes

Imagine the universe just a fraction of a second after the Big Bang. It wasn't the cold, empty space we see today; it was a seething, super-hot soup of the most fundamental building blocks of matter: quarks and gluons. Scientists call this state of matter the Quark-Gluon Plasma (QGP). In our everyday world, quarks are glued together tightly to form particles like protons and neutrons, but in this primordial soup, they are free to roam. To understand how this soup behaves, physicists need a way to "taste" it without eating the whole bowl. They use special probes called quarkonia.

Think of quarkonia as heavy-duty "ice cubes" made of a heavy quark and its anti-quark partner. Just like regular ice cubes, they come in different sizes and strengths. Some are tightly bound and sturdy (the "ground states"), while others are loosely held together and wobbly (the "excited states"). When you drop a wobbly ice cube into a hot drink, it melts instantly. A sturdy one might survive for a moment. By watching which ice cubes melt and which survive when dropped into the cosmic soup, scientists can figure out how hot and dense that soup is. This paper is about dropping these specific "ice cubes" into a new, smaller version of the soup to see if they melt the same way they do in the giant, heavy versions.

The Experiment: Dropping Ice Cubes into a Smaller Pot

For years, scientists at the Large Hadron Collider (LHC) have been smashing giant lead or gold nuclei together to create massive, long-lasting droplets of this Quark-Gluon Plasma. They found a clear pattern: the wobbly, excited "ice cubes" (called $\Upsilon(2S)$ and $\Upsilon(3S)$ ) melted away almost completely, while the sturdy ground-state ones ( $\Upsilon(1S)$ ) survived. This is called sequential suppression. It's like a heat wave that is strong enough to melt the weak ice but not the strong ice.

But a big question remained: Does this melting happen only in the giant, heavy collisions, or does it happen even in smaller, lighter collisions? To find out, the CMS Collaboration (a massive team of scientists) decided to run a new experiment. Instead of smashing heavy lead atoms, they smashed oxygen atoms together (Oxygen-Oxygen or OO) and neon atoms together (Neon-Neon or NeNe). These are much lighter and create a much smaller, shorter-lived "soup" than the lead collisions. They did this in 2025 at an energy of 5.36 TeV per nucleon pair.

To make sure they knew what "normal" looked like, they also compared their results to collisions of single protons (proton-proton or pp) at the exact same energy. If the "ice cubes" behaved normally, the ratio of melted ice to surviving ice in the light-ion collisions should look the same as in the proton collisions. If the "soup" was hot enough to melt them, the ratios would change.

The Findings: Even the Small Soup Melts the Ice

The results were clear and exciting. When the scientists looked at the oxygen-oxygen and neon-neon collisions, they found that the wobbly ice cubes were indeed melting.

The $\Upsilon(2S)$ vs. $\Upsilon(1S)$ Ratio: In the oxygen collisions, the ratio of the excited state to the ground state was 0.664 (meaning the excited state was significantly suppressed compared to the reference). In neon-neon collisions, this ratio dropped even further to 0.27. Both of these numbers are much lower than what is seen in proton-proton collisions, and the difference is so large that it is statistically significant by more than five standard deviations. In the world of science, this is a very loud "yes."
The $\Upsilon(3S)$ vs. $\Upsilon(1S)$ Ratio: The most fragile ice cube, the $\Upsilon(3S)$ , showed an even bigger drop. In oxygen collisions, the ratio was 0.392. This is a massive reduction compared to the proton baseline.
The "Smoking Gun": The most crucial finding was comparing the $\Upsilon(3S)$ directly to the $\Upsilon(2S)$ . The paper reports that the suppression of the $\Upsilon(3S)$ relative to the $\Upsilon(2S)$ exceeds three standard deviations. This is the first time scientists have seen this specific "sequential" pattern (where the weakest melts first, then the next weakest) in light ion collisions.

The data shows that even in these tiny, short-lived collisions of oxygen and neon, a hot, dense medium is formed that is capable of melting the weakly bound quarkonia. The effect gets stronger as the system gets bigger (from neon to oxygen to lead), but it is clearly present even in the smallest systems.

What This Means (and What It Doesn't)

This discovery is a big deal because it challenges the idea that you need a massive, long-lasting collision to create a "soup" hot enough to melt these particles. The results suggest that the "melting" effect (final-state interactions) can emerge in systems that are much smaller and shorter-lived than previously thought.

The paper also compared their findings to a theoretical model called SHINCHON. The model tried to predict what would happen, but it significantly underestimated the amount of melting the scientists actually saw. This means our current theories might be missing some extra "heat" or some other dynamic effect that makes the soup even more effective at breaking up these particles than we thought.

In short, the CMS team has provided strong evidence that the "sequential suppression" of these heavy particles is a universal feature of these high-energy collisions, appearing even in the smallest, lightest systems. The "soup" is smaller, but it's still hot enough to melt the ice.

Technical Summary: Evidence for Sequential $\Upsilon(nS)$ Suppression in Light Ion Collisions

Problem Statement
Quarkonia, bound states of heavy quark-antiquark pairs, serve as sensitive probes for the temperature and space-time evolution of the quark-gluon plasma (QGP). In heavy-ion collisions (e.g., PbPb, AuAu), a characteristic pattern of "sequential suppression" has been observed: weakly bound excited states (like $\Upsilon(2S)$ and $\Upsilon(3S)$ ) are more strongly suppressed than the tightly bound ground state ( $\Upsilon(1S)$ ) due to color screening and dissociation in the QGP. However, disentangling these final-state QGP effects from cold nuclear matter (CNM) effects and initial-state interactions remains challenging, particularly in smaller collision systems like proton-lead (pPb). While sequential suppression has been established in large systems, it is unclear whether the same mechanisms operate in smaller, symmetric light-ion systems where the lifetime and transverse area of the medium are significantly reduced. This paper addresses the need to study systems of varying sizes with different initial conditions to determine if sequential suppression persists in light ion collisions, thereby testing the universality of QGP formation and properties across system sizes.

Methodology
The study utilizes data collected by the CMS detector at the LHC during 2025, focusing on oxygen-oxygen (OO) and neon-neon (NeNe) collisions at a nucleon-nucleon center-of-mass energy of $\sqrt{s_{NN}} = 5.36$ TeV. These measurements are compared against a reference proton-proton (pp) dataset collected at the same energy in 2024.

Data Samples: The analysis uses integrated luminosities of $8.4 \pm 0.3$ nb $^{-1}$ for OO, $0.81 \pm 0.03$ nb $^{-1}$ for NeNe, and $456 \pm 14$ pb $^{-1}$ for the pp reference.
Event Selection: $\Upsilon$ states are reconstructed via their prompt decay into two oppositely charged muons. Events are selected requiring at least one (for AA) or two (for pp) muons within a single bunch crossing, a reconstructed primary vertex, and "high-purity" track criteria to suppress beam-gas backgrounds. Muons are required to have $p_T > 4$ GeV and $|\eta| < 2.4$ .
Yield Extraction: The invariant mass spectra of muon pairs are fitted in the range 8–14 GeV using extended unbinned maximum likelihood fits. The signal model comprises three peaks (for $\Upsilon(1S)$ , $\Upsilon(2S)$ , and $\Upsilon(3S)$ ), each described by a sum of two Crystal Ball functions to account for $\eta$ -dependent resolution. The peak positions and mass resolutions for excited states are constrained relative to the $\Upsilon(1S)$ using world-average mass ratios. A background component is modeled as an error function multiplied by an exponential.
Quantification: Suppression is quantified using double ratios ( $D_{nm}$ ) of excited-to-reference-state yields in AA collisions relative to pp collisions:
$D_{nm} = \frac{(N_{\Upsilon(nS)} / N_{\Upsilon(mS)})_{AA}}{(N_{\Upsilon(nS)} / N_{\Upsilon(mS)})_{pp}}$
where $(n, m)$ correspond to $(2, 1)$ , $(3, 1)$ , and $(3, 2)$ . This construction cancels common experimental uncertainties and initial-state nuclear effects that modify all $\Upsilon(nS)$ states similarly.
Systematic Uncertainties: Uncertainties are evaluated by varying analysis procedures, including alternative signal and background parameterizations, MC $p_T$ weighting variations, and tag-and-probe corrections.
Theoretical Comparison: Results are compared with predictions from the SHINCHON framework, which combines event-by-event hydrodynamics with bottomonium dissociation calculations based on gluo-dissociation and inelastic parton scattering.

Key Contributions and Results
This work presents the first measurements of the $\Upsilon(1S)$ , $\Upsilon(2S)$ , and $\Upsilon(3S)$ states in OO collisions, and the $\Upsilon(1S)$ and $\Upsilon(2S)$ states in NeNe collisions.

Sequential Suppression Evidence:
- In OO collisions, the double ratios are measured as $D_{21} = 0.664 \pm 0.055 \text{ (stat)} \pm 0.007 \text{ (syst)}$ , $D_{31} = 0.392 \pm 0.077 \text{ (stat)} \pm 0.023 \text{ (syst)}$ , and $D_{32} = 0.59 \pm 0.12 \text{ (stat)} \pm 0.04 \text{ (syst)}$ .
- In NeNe collisions, $D_{21} = 0.27 \pm 0.12 \text{ (stat)} \pm 0.01 \text{ (syst)}$ .
- All double ratios are significantly below unity. The suppression of $\Upsilon(2S)$ and $\Upsilon(3S)$ relative to $\Upsilon(1S)$ exceeds five standard deviations in OO collisions.
- Crucially, the relative suppression of $\Upsilon(3S)$ to $\Upsilon(2S)$ ( $D_{32}$ ) in OO collisions is below unity by 3.2 standard deviations. This constitutes the first evidence for sequential $\Upsilon(nS)$ suppression in light ion collisions.
System Size Dependence: The results extend the observed hierarchy of excited-state suppression to systems with an average number of participating nucleons ( $\langle N_{part} \rangle$ ) of approximately 10. The suppression pattern in OO and NeNe collisions shows a clear evolution with system size, consistent with trends observed in pPb and PbPb collisions.
Kinematic Dependence: Measurements as a function of charged-particle multiplicity ( $N_{corr}^{trk}$ ) and transverse momentum ( $p_T$ ) show no significant $p_T$ dependence within uncertainties. The trend of increasing suppression with event activity extends from pp to PbPb, though the paper notes that event activity alone does not fully determine the suppression magnitude.
Theoretical Comparison: The SHINCHON framework predictions significantly underestimate the measured suppression, particularly in the $p_T$ -differential measurements. This suggests that current models may require additional medium dynamics or stronger dissociation effects to describe the data.

Significance
The paper claims that these results provide evidence that the sequential suppression of $\Upsilon$ mesons, a hallmark of QGP formation, is not exclusive to large heavy-ion systems but also emerges in light ion collisions (OO and NeNe). The observation that state-dependent suppression persists in systems with smaller lifetimes and transverse areas than PbPb collisions implies that final-state effects, such as deconfinement, can manifest in these smaller systems. The compatibility of the $\Upsilon(3S)/\Upsilon(2S)$ ratio's magnitude and $p_T$ dependence between OO and PbPb collisions supports a common interpretation of sequential suppression across different collision systems. These findings reinforce the view that quarkonium suppression does not strictly require large volumes of matter and provide new constraints for theoretical models aiming to describe bottomonium production from pp to heavy-ion collisions. The paper modestly concludes that while the data supports a common interpretation, the tension with theoretical models indicates a need for further refinement in understanding the underlying production mechanisms and medium dynamics.

The Cosmic Soup and the Melting Ice Cubes

The Experiment: Dropping Ice Cubes into a Smaller Pot

The Findings: Even the Small Soup Melts the Ice

What This Means (and What It Doesn't)

More like this