Search for low mass vector resonances decaying into quark-antiquark pairs in proton-proton collisions at $\sqrt{s} =$ 13 TeV

The Big Picture

Finding a new particle at the Large Hadron Collider is a bit like hearing a whisper in a sold-out stadium. The LHC smashes protons together billions of times per second, generating a flood of debris. Somewhere in that noise, exotic particles might briefly flicker into existence before decaying into ordinary matter.

The Standard Model of particle physics has been wildly successful, but it leaves big questions on the table. Why do fundamental particles have the masses they do? Why is there more matter than antimatter? Many physicists suspect the answers involve new, undiscovered particles.

One class of candidates: new force-carrying particles, cousins of the photon but heavier, that couple to quarks (the constituents of protons and neutrons) while ignoring electrons and their relatives. These are called leptophobic vector resonances. If they exist at low masses, below 450 GeV or roughly 450 times the proton mass, they’re especially hard to spot. At those energies the LHC’s detectors are swamped by ordinary collisions, and a subtle new signal becomes nearly impossible to pick out with conventional strategies.

The CMS Collaboration devised a clever workaround. They applied it to 77 inverse femtobarns of collision data, the largest dataset used for this type of search. No new particles turned up, but the team set the most stringent limits ever placed on how strongly these hypothetical resonances could couple to quarks across a broad mass range.

Key Insight: By exploiting a trick where new particles recoil against radiation before decaying, CMS pushed the hunt for hidden new physics into mass territory where conventional searches go blind, ruling out a large swath of theoretical possibilities.

How It Works

The core problem with hunting for low-mass new particles is the trigger: the electronic gatekeeper that decides which collisions get saved. At low energies, the rate of ordinary quark-gluon collisions governed by quantum chromodynamics (QCD) is so enormous it saturates the trigger bandwidth. Any new-particle signal at, say, 100 GeV gets drowned out before the data even reaches storage.

CMS found a workaround. They looked for events where the new resonance was produced alongside initial state radiation (ISR), a gluon or quark radiated off an incoming proton just before the collision. This ISR jet carries enough energy to pass the trigger, effectively using ordinary background physics as a lever to lift the low-mass signal above the noise floor.

When a hypothetical Z’ boson (the stand-in model for these resonances) decays into a quark-antiquark pair at high momentum, the two daughter quarks get swept together by the boost. Instead of flying apart into two well-separated jets, they merge into a single fat jet with distinctive two-pronged substructure: a snowball that, on closer inspection, turns out to be two smaller snowballs stuck together.

Two wide-jet algorithms caught these merged decays across the full mass range:

• AK8 jets (anti-kT algorithm, radius R = 0.8): efficient for resonance masses below ~200 GeV, where decay products are tightly collimated
• CA15 jets (Cambridge-Aachen algorithm, radius R = 1.5): captures wider angular spreads for heavier resonances between 200 and 450 GeV

To separate signal from QCD background, the team applied a jet substructure tagger called N-subjettiness. This variable quantifies how well a jet’s radiation pattern matches two prongs versus one. Signal jets from Z’ decay score high on “two-pronginess,” while ordinary QCD jets tend to look more diffuse.

After selecting substructure-tagged candidates, the analysis looked for a narrow bump in the invariant jet mass spectrum, the distribution of masses reconstructed from all particles inside the fat jet. QCD background falls smoothly as a function of mass. A new resonance would appear as a localized excess sitting on top of that smooth curve.

The search covered masses from 50 to 450 GeV. The 2017 dataset extended earlier work into the previously unexplored 300–450 GeV range.

No bump showed up. The collaboration used this absence to set upper limits at 95% confidence level on the coupling strength of such resonances to quarks, drawing a line on how strongly these hypothetical particles could interact and still have escaped detection. For masses between 50 and 300 GeV, these are the most sensitive constraints ever placed on leptophobic vector resonances. The 300–450 GeV window is entirely new territory for jet substructure-based searches.

Why It Matters

Null results are the workhorses of particle physics. Every ruled-out region carves the map of possible new physics more precisely. Models predicting leptophobic Z’ bosons aren’t abstract curiosities: they’re motivated by dark matter, the matter-antimatter asymmetry, and unexplained patterns in quark masses. Closing off mass ranges forces theorists to refine their models or abandon dead ends.

The jet substructure techniques refined here carry well beyond this particular search. Fat jets, substructure variables, ISR-boosted topologies: this toolkit is now standard at the LHC for hunting new physics at low and intermediate masses. Future runs with more collisions will tighten these limits, and machine learning is already being folded in to squeeze more sensitivity from jet substructure.

Open questions remain. Can similar techniques probe below 50 GeV? What about broader resonances? Could AI-driven anomaly detection find signals that template-based searches miss?

Bottom Line: CMS found no new particles between 50 and 450 GeV using an ISR-boosted jet substructure strategy, but set the world’s most stringent limits on hidden quark-coupling resonances across most of that range. For the 300–450 GeV window, this is the first time jet substructure methods have been applied at all.