First-hand information for everyone
Scientists working with the CMS experiment at the Large Hadron Collider measured how pairs of particle jets scatter in proton–proton collisions at a collision energy of 13 TeV. They used the full 2016–2018 data set, which corresponds to an integrated luminosity of 138 inverse femtobarns (a measure of the total data collected). The measured angular patterns of the two highest-energy jets generally agree with the predictions of the standard theory of the strong force, but the researchers see small shape differences in two dijet-mass ranges: 2.4–4.8 TeV and above 6 TeV. They used these measurements to search for many proposed types of new physics and set the strongest limits yet for several scenarios.
The analysis looks at “dijet” events, where two energetic sprays of particles (jets) come from collisions of quarks and gluons inside the protons. To focus on the scattering dynamics and reduce sensitivity to uncertainties in the proton structure, the team used a normalized angular variable called χ_dijet = exp(|y1 − y2|), where y1 and y2 are the jets’ rapidities (a way to describe angle and speed along the beam). Small values of χ_dijet correspond to more central, head-on scattering, which is where many new-physics effects would show up. The study concentrates on events with large dijet mass (Mjj) and with the dijet system not too boosted along the beam direction (|yboost| < 1.11). The leading and subleading jets were required to have transverse momenta above 500 and 200 GeV, respectively. The largest dijet mass observed in the data is 8.3 TeV.
A key technical step was correcting the measured distributions for detector effects so they can be compared to theoretical predictions at the level of particles leaving the collisions. The team used a response matrix to model how true particle-level events appear in the detector and performed unfolding to recover the particle-level distributions. For the first time in this kind of CMS measurement, those unfolded angular distributions were compared directly with state-of-the-art theory from perturbative quantum chromodynamics (QCD) computed at next-to-next-to-leading order (NNLO). NNLO means the calculation includes more terms in the series expansion used by theorists and is therefore more precise than older predictions. The comparison also includes next-to-leading-order (NLO) corrections from the electroweak force.
The angular distributions were tested for a range of beyond-the-standard-model (BSM) ideas. These include quark compositeness or contact interactions (the idea quarks might not be point-like), extra spatial dimensions, microscopic quantum black holes, mediators of dark matter, axion-like particles that couple to gluons, and anomalous three-gluon interactions described in an effective field theory. The analysis finds no clear sign of new particles or forces. It sets new limits at 95% confidence level. For one benchmark quark contact interaction model that involves only left-handed quarks, contact interactions are excluded up to scales of 17 TeV for destructive interference and 37 TeV for constructive interference. The coupling of axion-like particles to gluons, written as c_g/f_a, is constrained to be below 0.42 TeV^−1. An anomalous triple-gluon coupling parameter C_G/Λ^2 is limited to be less than 0.0076 TeV^−2.
Important caveats apply. The small differences in the shape of the normalized distributions are noted but are not presented as discoveries; overall the data are largely consistent with the standard model. The exclusion limits depend on specific benchmark assumptions in each model (for example, the quark-contact limit quoted assumes only left-handed quarks participate and that the theory is valid to NLO in QCD). Theoretical uncertainties remain, for example from choices of calculation scales and from the parton distribution functions (PDFs, which describe how momentum is shared among the proton’s constituents). Experimental uncertainties from jet measurements and the unfolding procedure also affect the results. The authors provide detailed numbers in a public data record for others to study.