Astrophysics and cosmology have established that about 80% of the mass in the universe consists of dark matter. Dark matter and normal matter interact gravitationally, and they may also interact weakly, raising the possibility that collisions at the LHC may produce pairs of dark-matter particles.

With low interaction strength, dark-matter particles would escape the LHC detectors unseen, accompanied by Standard Model particles. These particles, such as single jets, photons, or W, Z or Higgs bosons, could either be produced in the interaction with the dark matter or radiated from the colliding partons. One result would be "mono-X" signals, named because the Standard Model particle, X, would appear alone, without other visible particles balancing their momentum in the transverse plane of the detector.

During Run 1 of the LHC, ATLAS developed a broad programme of searches for mono-X signals. Now, new results from the ATLAS collaboration in the mono-jet and mono-photon channels are the first of these searches in the proton–proton collision data collected in 2015 after increasing the LHC collision energy to 13 TeV. With only 3.2 fb–1 of collisions, six times fewer than studied in Run 1, these first Run 2 results already achieve comparable sensitivity to beyond-the-Standard-Model phenomena. In each search, the data with large missing transverse momentum are compared with data-driven estimates of Standard Model backgrounds. As an example, the background to the mono-jet search is known to 4–12%, an estimate nearly as precise as that obtained in the final Run 1 analysis. ATLAS has also released preliminary Run 2 results in the mono-Z, mono-W and mono-H channels.

If dark-matter production is observed, ATLAS has the potential to characterise the interaction itself. To produce dark matter in LHC collisions, the interaction must involve the constituent partons within the proton. If the interaction is mediated by s-channel exchange of a new boson, a decay back to the Standard Model partons could also occur.

The ATLAS collaboration has also released new results from the dijet search channel, where new phenomena could modify the smooth dijet invariant mass distribution. With 3.6 fb–1 of data, the search already surpasses the sensitivity of Run 1 dijet searches for many kinds of signals. The dijet results are presented on a simplified model of dark-matter production, where the dark boson has axial-vector couplings to quarks and Dirac dark matter.

The results of the mono-photon, mono-jet and dijet searches are shown in figure 1, assuming a version of the axial-vector dark boson whose couplings to dark matter are four times stronger than those to Standard Model quarks. In this scenario, ATLAS dijet results exclude the existence of mediating particles with masses from about 600 GeV to 2 TeV. The mono-jet and mono-photon channels exclude the parameter space at lower mediator and dark-matter masses. For even larger ratios of the dark-matter-to-quark coupling values, dijet constraints quickly weaken, and mono-X searches play a more powerful role.

On the verge of new data-taking in 2016, with the LHC expected to deliver an order of magnitude more luminosity, mono-X and direct mediator searches at ATLAS are set to probe this and other models with unprecedented sensitivity.