Constraining $CP$-violating Higgs sectors at the LHC using gluon fusion

We investigate the constraints that the LHC can set on a 126 GeV Higgs boson that is an admixture of $CP$ eigenstates. Traditional analyses rely on Higgs couplings to massive vector bosons, which are suppressed for $CP$-odd couplings, so that these analyses have limited sensitivity. Instead we focus on Higgs production in gluon fusion, which occurs at the same order in ${\alpha }_S$ for both $CP$-even and -odd Higgs couplings to top quarks. We study the Higgs plus two jet final state followed by Higgs decay into a pair of tau leptons. We show that using the 8 TeV data set it is possible to rule out the pure $CP$-odd hypothesis in this channel alone at nearly 95% C.L, assuming that the Higgs is $CP$-even. We also provide projected limits for the 14 TeV LHC run.

Figure 1

Observable distributions for the signal and background. From the top left and proceeding clockwise: $m_{jj}$, $\mathrm{\Delta }{\varphi }_{jj}$, sin($|\mathrm{\Delta }{\varphi }_{jj}|/2$), and $\mathrm{\Delta }{\eta }_{jj}$. For each figure the yields are normalized to the expected yields at 8 TeV for the gluon fusion channel at $20{\mathrm{fb}}^{-1}$ with $\alpha =0$. Samples have been passed through the detector pseudosimulation and subjected to the full selection on all channels. The loose WBF selection and the additional category selections are applied in all cases.

Figure 2

Efficiency curves for our boosted decision tree for 8 (left) and 14 TeV (right). The red curves are for $\alpha =1.5$ and the blue curves for $\alpha =0.6$. The dashed curves show the results only including the $\sin (|\mathrm{\Delta }{\varphi }_{jj}|)$ variable and the solid curves those for the full BDT with all 18 observables included, as described in the text.

Figure 3

Signal distributions for $\mathrm{\Delta }{\varphi }_{jj}$ with inclusive (non-WBF-like) cuts at $\sqrt{s}=8\mathrm{TeV}$: $p_{Tj}>30\mathrm{GeV}$, $\mathrm{\Delta }{R}_{jj}>0.4$. In particular there are no cuts on $m_{jj}$ or $p_{T,H}$. These are generator-level distributions, with each channel separately normalized by its cross section, and the whole plot is normalized to $1/\pi$. Note that by assumption the shape of the WBF distribution does not vary with $\alpha$.

Figure 4

Expected limits that can be achieved with our analysis using the $20{\mathrm{fb}}^{-1}$ 8 TeV data set (left) and using a $50{\mathrm{fb}}^{-1}$ data set at 14 TeV (right). The dashed curves show the estimated significance of the total signal over the Standard Model backgrounds, and the solid curves show the exclusion significance computed using the ${\mathrm{CL}}_s$ method relative to the $\alpha =0$ case. See the text for details.

Figure 5

Left: The projected 95% exclusion confidence limit on the mixing angle $\alpha$ that can be set as a function of the integrated luminosity at the 14 TeV run of the LHC (assuming $\alpha =0$ as the null hypothesis). Right: As in the left-hand plot, but comparing with an inflated theoretical error of 25%.