Probing the Higgs boson CP properties in vector-boson fusion production in the $H\rightarrowτ^+τ^-$ channel with the ATLAS detector

TL;DR

This study probes CP violation in the Higgs–vector-boson interaction via VBF production with H→ττ decays using the ATLAS Run 2 dataset of 140 fb$^{-1}$ at $\sqrt{s}=13$ TeV. The analysis hinges on CP-odd observables, notably the Optimal Observable, to capture interference between SM and CP-odd EFT contributions within two widely used EFT bases (HISZ and Warsaw). Through a data-driven embedding strategy for the dominant $Z\to\tau\tau$ background, extensive MC-based modelling, and a binned likelihood fit across multiple signal/control regions, the paper derives stringent 95% CL limits on the CP-odd EFT parameters $\tilde{d}$ and $c_{H\tilde{W}}$ (with $\ ilde{d}=-0.012$ to $0.044$ and $c_{H\tilde{W}}=-0.24$ to $0.83$ for $\Lambda=1\ \mathrm{TeV}$) while finding results fully compatible with the SM. The work demonstrates that the Optimal Observable approach yields improved sensitivity relative to other CP-odd observables and provides robust, combinable EFT constraints for future global analyses. The findings reinforce the SM CP structure in HVV couplings and set the stage for tighter SMEFT tests with upcoming data.

Abstract

The CP properties of the Higgs boson are studied in the vector-boson fusion production mode. The analysis exploits the decay mode of the Higgs boson into two $τ$-leptons using 140 fb$^{-1}$ of proton-proton collision data at $\sqrt{s}=13$ TeV collected by the ATLAS experiment at the Large Hadron Collider. Results are obtained using the Optimal Observable method. CP-violating interactions between the Higgs boson and electroweak gauge bosons are considered in the effective field theory framework, with the interaction strength described in the HISZ basis by $\tilde{d}$, and in the Warsaw basis by $c_{H\tilde{W}}$, $c_{H\tilde{B}}$, and $c_{H\tilde{W}B}$. No deviations relative to the Standard Model are observed, and limits are obtained on the strength parameters. The $\tilde{d}$ parameter is constrained to the interval [$-0.012, 0.044$] at the 95% confidence level while $c_{H\tilde{W}}$ is constrained to [$-0.24, 0.83$], when considering both linear and quadratic effects of physics beyond the Standard Model.

Figures (7)

• Figure 1: Distributions for SM VBF Higgs boson production and two VBF Higgs boson samples with $\tilde{d}\xspace\neq 0$, for (a) the Optimal Observable, $\mathcal{OO}$, (b) $\Delta\phi_{jj}^{\textrm{sign}}$, and (c) $p_{\text{T}+}p_{\text{T}-}\sin(\Delta\phi_{jj}^{\mathrm{sign}} \xspace)\xspace$, shown for final states where one $\tau$-lepton decays hadronically, and the other one, leptonically. The MC simulations used are those described in Section \ref{['sec:samples']}.
• Figure 2: Post-fit distribution of the neural network signal score for the three channels, (a) $\Pgt_{\text{lep}}$$\Pgt_{\text{lep}}$, (b) $\Pgt_{\text{lep}}$$\Pgt_{\text{had}}$, and (c) $\Pgt_{\text{had}}$$\Pgt_{\text{had}}$. The ratio of the SM predictions to the observed data is shown in the bottom panel. The "Other bkg" includes diboson, $Z\rightarrow\ell\ell$, and $W\rightarrow\tau\nu+$jets, and other non-VBF-Higgs processes. The hatched area represents the impact of the experimental, theoretical and statistical uncertainties in the Standard Model predictions. The dashed lines represent examples of signal samples, for different values of $\tilde{d}$.
• Figure 3: Distributions of the $\mathcal{OO}$ in the VR for the three channels: (a) $\Pgt_{\text{lep}}$$\Pgt_{\text{lep}}$, (b) $\Pgt_{\text{lep}}$$\Pgt_{\text{had}}$, (c) $\Pgt_{\text{had}}$$\Pgt_{\text{had}}$, after a background-only fit. The value of the parameter of interest, the signal strength, the normalisation of backgrounds, and the systematic uncertainties are set to the value from the best fit result. In the bottom panel, the hatched area represent the total post-fit systematic uncertainty in the final SM prediction.
• Figure 4: Post-fit distributions of the $\mathcal{OO}$ in the (a,b,c) $\mathrm{SR}^{\mathrm{High NN}}$ , (d,e,f) $\mathrm{SR}^{\mathrm{Low NN}}$ and for the three channels: (a,d) $\Pgt_{\text{lep}}$$\Pgt_{\text{lep}}$, (b,e) $\Pgt_{\text{lep}}$$\Pgt_{\text{had}}$, (c,f) $\Pgt_{\text{had}}$$\Pgt_{\text{had}}$. The value of the parameter of interest, the signal strength, the normalisation of backgrounds and the systematic uncertainties are set to the value from the best fit result. The dashed line in the bottom panel shows the total pre-fit SM predictions (with $\tilde{d}\xspace=0$). The dashed area represent the total post-fit systematic uncertainty in the final SM prediction.
• Figure 5: Expected and observed $\Delta$NLL distributions for the combined fit as a function of the CP-violating strength parameters (a) $\tilde{d}$, and (b) $c_{H\tilde{W}}$. The dashed lines show the observed $\Delta$NLL curves for the fit to the single channels. The expected curve is obtained assuming the SM predictions. The horizontal lines correspond to the $\Delta$NLL values used to determine the 68% and 95% confidence intervals.