Isolating a light Higgs boson from the di-photon background at the LHC

TL;DR

This paper delivers a next-to-leading order QCD analysis of the irreducible di-photon background to Higgs searches in the H→γγ channel at the LHC by computing NLO corrections to the gluon-fusion subprocess gg→γγ and combining them with quark-annihilation and fragmentation contributions. It provides a detailed treatment using dipole subtraction to handle infrared divergences, analyzes scale dependence and photon isolation criteria, and studies jet veto strategies. The study finds that NLO gluon-fusion corrections modestly affect the total background and reduce theoretical uncertainties, while photon isolation has only a mild effect on the Higgs significance; angular distributions offer a small but useful enhancement. The results guide optimization strategies for Higgs searches in the diphoton channel and outline future improvements, including completing remaining αs^2 contributions and incorporating detector effects and resummation.

Abstract

We compute the QCD corrections to the gluon fusion subprocess gg to gamma gamma, which forms an important component of the background to the search for a light Higgs boson at the LHC. We study the dependence of the improved pp to gamma gamma X background calculation on the factorization and renormalization scales, on various choices for photon isolation cuts, and on the rapidities of the photons. We also investigate ways to enhance the statistical significance of the Higgs signal in the di-photon channel.

Figures (11)

• Figure 1: Sample quark loop diagrams contributing to $pp\rightarrow\gamma\gamma X$, which are computed in this paper: (a) the leading order gluon fusion subprocess $gg \rightarrow \gamma\gamma$, (b) the virtual correction to this subprocess, and (c) the radiative process $gg \rightarrow \gamma\gamma g$.
• Figure 2: Sample diagrams for contributions to $pp\rightarrow\gamma\gamma X$ which are not treated in this paper, although they are of order $\alpha_s^2$: (a) the tree-level subprocess $gg \rightarrow \gamma\gamma q{\bar{q}}$, (b) doubly virtual correction to $q{\bar{q}}\rightarrow\gamma\gamma$, (c) doubly real correction to $q{\bar{q}}\rightarrow\gamma\gamma$, of the form $qg\rightarrow \gamma\gamma gq$, and (d) the process $qg \rightarrow \gamma\gamma q$via a quark box.
• Figure 3: Illustration of two possible ways to strengthen photon isolation, beyond the "standard" isolation represented by the inner cone. One can either increase the cone radius to the large outer one shown, or one can veto on jets within such a radius.
• Figure 4: (a) Contribution of the gluon fusion subprocess to $pp\rightarrow\gamma\gamma X$ at the LHC, at leading order (lower four), and at NLO (upper three), using various parton distributions. (b) Total $pp \rightarrow \gamma\gamma X$ production at NLO, including NLO $q{\bar{q}}\rightarrow\gamma\gamma$ and fragmentation contributions, with the gluon fusion subprocess treated at LO (dashed) and at NLO (solid). MRST99 set 2 partons are used in (b). Contributions not involving gluon fusion into photons are obtained from DIPHOX. Both sets of plots are for $\mu_R = \mu_F = 0.5 M_{\gamma\gamma}$, and a standard photon isolation criterion with $R=0.4$, $E_{\rm T\,max} = 15$ GeV. Statistical errors from numerical integration are shown.
• Figure 5: Scale dependence of (a) the gluon fusion subprocess contribution to $pp \rightarrow \gamma\gamma X$, and (b) the total $pp\rightarrow \gamma\gamma X$ production cross section, for standard photon isolation with $R=0.4$, $E_{\rm T\,max} = 15$ GeV. In both plots, the bands represent the result of varying $\mu_R$ and $\mu_F$ over the square region $0.5 M_{\gamma\gamma} < \mu_R , \mu_F < 2 M_{\gamma\gamma}$. The dashed (solid) hatched band corresponds to including the gluon fusion subprocess at LO (NLO). For the leading order band in (a) only, the LO CTEQ5L parton distributions were used; otherwise the NLO MRST99 set 2 distributions were employed.