The quark–gluon plasma, or QGP, is a state of deconfined quarks and gluons produced in ultrarelativistic heavy-ion collisions. In the earliest instants of the collision, pairs of quarks and gluons undergo hard scatterings, producing high-momentum particles that fly outward. When the plasma is present, these energetic partons lose energy through collisional and radiative processes, a phenomenon known as jet quenching. Experiments at the Relativistic Heavy Ion Collider (RHIC) and the LHC have long observed a strong suppression of jets and high-momentum hadrons in central lead–lead and gold–gold collisions, and this suppression is one of the defining signatures of the plasma’s existence.

The challenge for experimentalists is establishing a reliable reference point. Jets are reconstructed from showers of particles, and the reconstruction itself can be biased by the very energy loss being studied. ALICE’s new measurement sidesteps this problem by exploiting electroweak particles, in this case direct-prompt photons, which are produced in the first instants of the collision and do not interact strongly with the QGP. At leading order in perturbative quantum chromodynamics, a prompt photon is produced back-to-back in azimuthal angle with a recoiling parton of similar transverse momentum. The photon therefore serves as an unmodified clock and energy scale for the hard scattering, while its partner parton plows through the plasma and emerges modified.

To isolate genuine prompt photons from a sea of background, ALICE applied an isolation criterion. Photons were required to carry transverse momenta between 18 and 40 GeV/c at midrapidity, and the sum of charged-particle transverse momenta within a small cone of angular radius 0.2 around the photon direction had to be below 1.5 GeV/c, after subtracting the estimated contribution of the underlying event. This tight selection suppresses fragmentation photons, bremsstrahlung photons, and photons from neutral-meson decays, which are typically accompanied by other particle fragments. The associated charged hadrons were measured with transverse momenta above 1.8 GeV/c using the Inner Tracking System and the Time Projection Chamber.

The analysis, based on lead–lead data collected in 2015 and 2018, examined three collision centrality classes: central (0–30%), semicentral (30–50%), and peripheral (50–90%). For each trigger photon, the researchers constructed the conditional yield D(zT) of associated hadrons, where zT is the ratio of the hadron’s transverse momentum to the photon’s. This distribution, extracted from the region of azimuthal angle opposite to the photon, acts as a proxy for the fragmentation function of the recoiling parton. Extracting it in the dense environment of a lead–lead collision required careful subtraction of the underlying event using mixed-event techniques and statistical removal of background triggers, dominated by decays of neutral pions, through a purity-based procedure.

Because the corresponding proton–proton measurement at the same photon momentum interval was statistically too limited, the collaboration used next-to-leading-order pQCD calculations without energy loss as the proton–proton reference, defining a ratio called IpQCD. These calculations have been shown to describe the published proton–proton data well, making them a sound stand-in. The result is striking: in central collisions, the ratio sits at roughly 0.5 across most of the zT range, meaning the yield of hadrons associated with the photon is cut roughly in half by the medium. In semicentral collisions the suppression is milder, and in peripheral collisions the ratio moves closer to unity, exactly the pattern expected if the quark–gluon plasma is responsible for the quenching.

The data were confronted with two state-of-the-art theoretical models that include energy loss. The first is an NLO pQCD calculation in the Higher-Twist formalism, in which the in-medium transport coefficient controls the energy loss and its parameters were extracted through a global Bayesian analysis of single-hadron, di-hadron, and photon–hadron data at several collision energies. The second is the CoLBT-hydro model, which couples the transport of jet shower partons to a (3+1)-dimensional hydrodynamic description of the expanding medium, allowing the jets themselves to excite the plasma. Both models agree with the measured D(zT) distributions and with the suppression ratios across all three centrality classes, although the current uncertainties do not yet allow the two descriptions to be distinguished from each other.

Crucially, the measurement rules out an alternative explanation. An NLO pQCD calculation that includes only cold nuclear matter effects, using nuclear parton distribution functions but no energy loss, predicts a ratio close to unity and fails to reproduce the magnitude of the suppression seen in central collisions. The suppression therefore cannot be attributed to initial-state effects alone; it demands genuine in-medium energy loss. The collaboration also computed a centrality ratio comparing central and semicentral yields to peripheral ones, finding flat distributions averaging around 0.5 and 0.75 respectively, again in agreement with the energy-loss model and confirming the expected centrality dependence of the quenching.

The measurement gains additional power from comparison with other experiments. The IpQCD results for central collisions are qualitatively compatible with CMS measurements of photon-tagged jets and of hadrons correlated with Z bosons in lead–lead collisions at the LHC, with the Z-boson comparison showing particularly good agreement given its similar centrality range and trigger momentum. They also align with direct photon–hadron correlation results from the STAR Collaboration at RHIC, which used a lower photon momentum range in more central gold–gold collisions. The PHENIX result, however, shows a significantly enhanced soft-hadron yield at low zT, likely because its lower trigger momentum means the recoiling partons lose a larger fraction of their energy, and its lower hadron threshold makes it more sensitive to soft enhancement.

By pushing the photon trigger down to 18 GeV/c, ALICE has opened a window onto a kinematic region where the largest nuclear effects are expected, probing lower momentum transfers than other LHC measurements and overlapping the momentum range explored at RHIC. This measurement serves as a benchmark for the collaboration’s Run 3 and Run 4 analyses, where vastly larger data samples will enable more precise correlations, finer centrality differentials, access to lower and higher photon momenta, and studies of how quenching depends on the orientation of the collision geometry. As the LHC continues to recreate and interrogate the hottest matter ever made in a laboratory, photons are proving to be the perfect witnesses: untouched by the inferno they illuminate, yet carrying the full story of what their partners endured on the way out.

Subject of Research: Jet quenching probed via isolated-prompt photon–hadron correlations in Pb–Pb collisions at the LHC

Article Title: Measurement of isolated-prompt photon–hadron correlations in Pb–Pb collisions at (\sqrt{{s}_{{\textrm{NN}}}} = 5.02) TeV

Article References: ALICE Collaboration, Abdallah, D. A. H., Abualrob, I. J., Acharya, S., Agarwal, K., Rinella, G. A., Aglietta, L., Agrawal, N., Ahammed, Z., Ahmad, S., Ahuja, I., Akbar, Z., Akishina, V., Al-Turany, M., Alessandro, B., Alfarasyi, A. R., Molina, R. A., Ali, B., Alici, A., … Zurlo, N. (2026). Measurement of isolated-prompt photon–hadron correlations in Pb–Pb collisions at $$\sqrt{{s}_{{\textrm{NN}}}} = 5.02$$ TeV. The European Physical Journal C, 86(9), Article 1126. https://doi.org/10.1140/epjc/s10052-026-16192-z

Image Credits: AI Generated

DOI: 10.1140/epjc/s10052-026-16192-z

Keywords: ALICE, LHC, quark–gluon plasma, jet quenching, photon–hadron correlations, Pb–Pb collisions, heavy-ion physics, fragmentation function, energy loss, pQCD, RHIC, CMS