Jet quenching seen in RHIC oxygen collisions, a sign of quark–gluon plasma

In the busiest 10% of oxygen–oxygen collisions recorded at Brookhaven in 2021, STAR counted about 16% fewer hadrons and 19% fewer jets recoiling from a high-momentum particle than in quieter collisions, a gap the team attributes to energy loss in quark–gluon plasma.

Peer reviewed · Collider measurement · Single experiment

Physicists with the STAR experiment at Brookhaven National Laboratory’s Relativistic Heavy Ion Collider (RHIC) report jet quenching in collisions of oxygen nuclei at 200 GeV per nucleon pair. Jets and particles that had to cross the collision zone came out about 20% less often in the busiest, most head-on collisions than in less busy ones. Jet quenching is a hallmark of quark–gluon plasma, so the result suggests this hot matter forms even when two 16-nucleon nuclei meet at about 1/27 of the LHC’s oxygen energy. How strong the evidence is depends on the baseline chosen.

Evidence at a glance

Is this a big deal?

For nuclear physicists, a solid step. Small systems had shown one plasma signature, collective flow, without the other, jet quenching. In oxygen at RHIC energy, both now appear. It does not settle where plasma formation switches off.

What scientists found

Quark–gluon plasma is matter so hot and dense that quarks and gluons, the particles inside protons and neutrons, roam free. The paper notes it filled the universe a few microseconds after the Big Bang. Colliders make a speck of it that lasts less than a billionth of a billionth of a second, Brookhaven says.

When two quarks or gluons collide hard, they fly apart back to back as jets, narrow sprays of particles. A jet that crosses plasma loses energy on the way. That is jet quenching.

STAR picked one fast “trigger” particle and counted what came out opposite it. In the busiest 10% of collisions, recoiling hadrons of 3 to 7 GeV/c were at 0.836 ± 0.021 (stat.) ± 0.032 (syst.) of their rate in the 40–60% reference group. Recoil jets of 8 to 19 GeV/c, measured with radius R = 0.5, were at 0.806 ± 0.037 ± 0.025.

How they found it

The data come from RHIC’s 2021 oxygen run, about 500 million collisions. STAR’s tracker recorded charged particles, while a separate forward detector measured how busy each collision was, so jets could not make a collision look busy.

Comparing busy with quiet collisions was deliberate. The usual measure, R_AA, compares nucleus collisions with proton collisions and needs a model of how many nucleon pairs collided. In small systems that estimate is very uncertain, the paper says. Counting recoils per trigger cancels it. This ratio is called I_CP.

The trigger side is the control. Particles travelling with the trigger were not suppressed, which the paper reads as the trigger’s jet losing negligible energy. Li Yi of Shandong University, who oversaw the analysis, compared the far side with “shooting a bullet through, say, water” in Brookhaven’s release. The analogy has a limit: jet particles keep moving at nearly light speed. What changes is that energy is knocked out of, or spread beyond, the cone counted as the jet. STAR puts this at an effective shift of 0.70 ± 0.15 (stat.) ± 0.10 (syst.) GeV/c for R = 0.5 jets, roughly 4 to 9% of a jet’s momentum in this range.

Why it matters to you

It won’t change anything you do. It is about origins: how small a speck of matter can still behave like the early universe’s quark soup. For the force that normally keeps quarks confined, see the strongest force in the universe.

What the evidence does not show

Not all of the drop has to be plasma. Quarks and gluons are distributed differently inside nuclei than inside free protons, which can shift jet rates without any plasma. Three theory calculations of that effect all predicted a few percent more recoils, not fewer. But they compare all oxygen collisions with proton collisions, which the authors flag as not precisely comparable.

The strength of the claim moves with the baseline. Against no change, the hadron and R = 0.5 jet drops are 4.3 standard deviations, meaning 4.3 times the combined uncertainty. Against the trigger-side control, the hadron result is 3.3. Across all baselines it is 2.7 to 5.2, and the authors write “strong evidence”. It is one experiment, and its I_CP cannot be lined up directly with LHC R_AA values.

How this fits what we already knew

RHIC has seen jet quenching in gold collisions for years. Gold nuclei hold 197 nucleons, about 12 times oxygen’s 16. Flow patterns later turned up in far smaller systems, down to proton–proton collisions, but the paper lists no earlier evidence of jet quenching in them.

The LHC first collided oxygen in 2025, at 5.36 TeV per nucleon pair. CMS reported an R_AA dipping to 0.69 ± 0.04 near 6 GeV, and CERN says ATLAS, ALICE and LHCb also see energy-loss signs. In the release, Yi notes that higher energy should make a hotter, denser plasma and a bigger signal. STAR’s result indicates the signal survives at 27 times lower energy.

Independent perspective

SciQuest found no comment from a physicist outside STAR as of 21 October 2026, 07:40 UTC, and has not yet interviewed one. Brookhaven’s Rongrong Ma, a STAR member, described the earlier state of play in the release: “We haven’t seen any unambiguous signal, so it’s still kind of under debate in these small systems.”

How other headlines framed it

Only Brookhaven’s release, “Oxygen Collisions Quench Jets’ Energy at RHIC“, has covered it, reposted verbatim by Phys.org and Newswise. It presents a single 20% figure as a clear plasma sign. The paper gives separate hadron and jet values, a significance range and a caveat on its theory baselines.

What happens next

RHIC stopped running in early 2026. Its final run ended with more oxygen collisions on 6 February 2026, recorded by STAR and the newer sPHENIX detector, and Brookhaven says more results are to come. Those data could test whether the suppression holds and how it depends on a jet’s path through the plasma.

In one sentence

In head-on oxygen collisions at RHIC, jets and particles that cross the collision zone come out up to about a fifth less often, a sign that quark–gluon plasma forms even in this small system at low energy, though how much of the drop is plasma depends on the baseline.

Read next

• What is the strongest force in the universe? The force that binds quarks inside protons and neutrons, until a collider briefly frees them.
• What are the parts of an atom? Nucleons, explained.
• Why is the Big Bang theory the most accepted? The evidence for the hot early universe this plasma recreates.

Sources and disclosure

1. STAR Collaboration (B. E. Aboona et al.). “Measurement of Jet Quenching in O+O Collisions at √s_NN = 200 GeV by the STAR Experiment at RHIC.” Physical Review Letters 137, 172301, published online 20 October 2026. doi:10.1103/9w2y-bz9n. Version read: arXiv:2604.13935v1 (15 April 2026), the only arXiv version.
2. Brookhaven National Laboratory. “Oxygen Collisions Quench Jets’ Energy at RHIC.” News release, 20 October 2026 (source of all quotes).
3. CMS Collaboration. “Observation of Suppressed Charged-Particle Production in Ultrarelativistic Oxygen-Oxygen Collisions.” Physical Review Letters 136, 162301 (2026). doi:10.1103/89sf-9t1x; abstract read via arXiv:2510.09864.
4. CERN. “Oxygen collisions at the LHC show new indications of extreme state of matter.” 24 July 2026.

Disclosure: This article was drafted with AI assistance and checked against the sources above, using the open arXiv version of the paper rather than the final journal text. The research was funded by the US Department of Energy Office of Science, the US National Science Foundation and agencies in other countries. Calculations by SciQuest: 1 − 0.836 = 16.4% and 1 − 0.806 = 19.4%; combined uncertainties √(0.021² + 0.032²) ≈ 0.038 and √(0.037² + 0.025²) ≈ 0.045; 5,360 GeV ÷ 200 GeV ≈ 27; 197 ÷ 16 ≈ 12; 0.70 ÷ 19 ≈ 4% and 0.70 ÷ 8 ≈ 9%. The graphic is a SciQuest original drawn from the paper’s values. Spotted an error? See our corrections policy.