Search for $CP$ violation in events with top quarks and Z bosons at $\sqrt{s}$ = 13 and 13.6 TeV

TL;DR

This paper searches for charge-parity violation in top quark and Z boson production using advanced machine learning and novel observables, setting limits on new physics effects with data from the CMS experiment at the LHC.

Contribution

It introduces the first use of CP-odd observables and physics-informed machine learning techniques in this context, providing new constraints on effective field theory operators.

Findings

Results are consistent with the Standard Model within two standard deviations.

Exclusion limits on Wilson coefficients are set at 95% confidence level.

A 2.5 sigma local significance hint is observed for one operator.

Abstract

A search for the violation of the charge-parity ($CP$) symmetry in the production of top quarks in association with Z bosons is presented, using events with at least three charged leptons and additional jets. The search is performed in a sample of proton-proton collision data collected by the CMS experiment at the CERN LHC in 2016$-$2018 at a center-of-mass energy of 13 TeV and in 2022 at 13.6 TeV, corresponding to a total integrated luminosity of 173 fb$^{-1}$. For the first time in this final state, observables that are odd under the $CP$ transformation are employed. Also for the first time, physics-informed machine-learning techniques are used to construct these observables. While for standard model (SM) processes the distributions of these observables are predicted to be symmetric around zero, $CP$-violating modifications of the SM would introduce asymmetries. Two $CP$-odd operators $\mathcal{O}_\text{tW}^\text{I}$ and $\mathcal{O}_\text{tZ}^\text{I}$ in the SM effective field theory are considered that may modify the interactions between top quarks and electroweak bosons. The obtained results are consistent with the SM prediction within two standard deviations, and exclusion limits on the associated Wilson coefficients of $-$2.7 $\lt$ $c_\text{tW}^\text{I}$ $\lt$ 2.5 and $-$0.2 $\lt$ $c_\text{tZ}^\text{I}$ $\lt$ 2.0 are set at 95% confidence level. The largest discrepancy is observed in $c_\text{tZ}^\text{I}$ where data is consistent with positive values, with an observed local significance with respect to the SM hypothesis of 2.5 standard deviations, when only linear terms are considered.