Collider fingerprints of freeze-in dark matter produced during the fast expansion phase of Universe

TL;DR

This paper investigates a non-standard early universe expansion scenario affecting freeze-in dark matter production, proposing a novel LHC search strategy using jet substructure and machine learning to probe the expanded parameter space.

Contribution

It introduces a new approach to detect freeze-in dark matter produced during rapid expansion phases, utilizing boosted jets and advanced analysis techniques at the LHC.

Findings

Enhanced sensitivity to non-standard cosmology scenarios.

Inclusion of one-loop QCD corrections improves prediction accuracy.

Machine learning techniques effectively discriminate signal from background.

Abstract

We examine a simple dark sector extension where the observed dark matter (DM) abundance arises from a freeze-in process through the decay of heavy vector-like quarks into a scalar dark matter candidate. The detection prospects of such DM are challenging due to the feeble nature of the interactions, but these vector-like quarks can be produced copiously at the LHC, where they decay to Standard Model quarks along with DM. Depending on the decay rate, this scenario is typically probed through long-lived particle or displaced vertex signatures, assuming a radiation-dominated background. An alternative hypothesis suggests that the Universe may have experienced a rapid expansion phase instead of the standard radiation-dominated one during freeze-in. This would significantly alter the dark matter phenomenology, requiring a substantial increase in the interaction rate to match the observed relic density, resulting in the rapid decay of the parent particle. As a result, much of the parameter space for this scenario is beyond the reach of traditional long-lived particle and displaced vertex searches. Due to this non-standard cosmic evolution, existing constraints do not cover the expanded dark matter parameter space. We propose a complementary search strategy to explore this scenario, offering additional limits alongside searches for long-lived particles and displaced vertices. In our search, we investigate the FIMP dark matter model at the LHC using boosted fatjets and significant missing transverse momentum. To improve precision, we include one-loop QCD corrections for LHC production processes and employ a boosted decision tree multivariate analysis, leveraging jet substructure variables to explore a vast parameter space for this minimally extended FIMP dark matter model at the 14 TeV LHC.