Asking for an extra photon in Higgs production at the LHC and beyond

We study the inclusive production of a Higgs boson in association with a high-$p_T$ photon at the LHC, detailing the leading-order features of the main processes contributing to the $H\gamma$ final state. Requiring an extra hard photon in Higgs production upsets the cross-section hierarchy for the dominant channels. The $H\gamma$ inclusive production comes mainly from photons radiated in vector-boson fusion (VBF), which accounts for about 2/3 of the total rate, for $p_T^{\gamma,j}>30$ GeV, at leading order. On the other hand, radiating a high-$p_T$ photon in the main top-loop Higgs channel implies an extra parton in the final state, which suppresses the production rate by a further $\alpha_S$ power. As a result, the $H\gamma$ production via top loops at the LHC has rates comparable with the ones arising from either the $H t\bar t$ production or the $HW(Z)\gamma$ associated production. Then, in order of decreasing cross section, comes the single-top-plus-Higgs channel, followed in turn by the heavy-flavor fusion processes $b\bar b \to H\gamma$ and $c\bar c \to H\gamma$. The $H\gamma$ production via electroweak loops has just a minor role. At larger c.m. energies, the $H t\bar t\gamma$ channel surpasses the total contribution of top-loop processes. In particular, requiring $p_T^{\gamma,j}>30$ GeV at $\sqrt S \simeq 100$ TeV, $H t\bar t\gamma$ accounts for about $1/4$ of the inclusive $H\gamma$ production at leading order, about half of the total being due to VBF production.


Introduction
The observation of a Higgs boson signal at the LHC [1] opened up a new era for collider physics. On the one hand, a major task of the LHC and future high-energy colliders is now to verify with high accuracy the actual properties of the new state, in order to check whether the standard model (SM) really provides the complete description of the electroweak symmetry breaking (EWSB) through the Higgs mechanism [2] at the TeV energy scale, or some theory modification is needed. On the other hand, Higgs boson production in the SM can itself act as a background for new possible exotic states that might involve Higgs bosons in their production or decay channels. As a consequence, the most accurate predictions on both Higgs production mechanisms and Higgs decay characteristics in the SM are desirable.
In this paper, we discuss the Higgs production associated to a prompt high-p T photon at the LHC. The Hγ final state can be experimentally quite distinctive, and might probe Hγ interactions in a nontrivial way. After requiring an extra high-p T photon, the naive expectation is that the original Higgs production mechanisms should be suppressed by a few order of magnitudes, corresponding to an extra α factor in cross sections (where α = e 2 /4π), while maintaining their relative weight. Actually, the main Higgs production mechanisms react in different ways to the requirement of photon radiation. In particular, we will show here that the normal hierarchy in the Higgs production channels is upset by the requirement of an extra high p T photon. Higgs production at the LHC mainly proceeds, in order of decreasing rate, via gluongluon (gg) fusion (mostly through a top-quark loop), vector-boson fusion (VBF), associated V H production (where V is either a W or a Z boson), associated ttH and bbH production, and single-top tH production. Predictions for the corresponding cross sections have been worked out with good accuracy (including at least QCD NLO corrections for all processes [3]). At the LHC, there is then a substantial hierarchy in the corresponding cross sections, and the gluon-fusion production via a top-loop turns out to be by far the dominant contribution to the inclusive Higgs production, being an order of magnitude higher than the VBF process, as detailed in the following.
The request of an extra photon in the final state changes drastically the latter ordering. Indeed, the process gg → Hγ, occurring via a top-box diagram, is forbidden by Furry's theorem, and in general by C parity. Then, the lowest-order partonic processes proceeding via QCD interactions (and unsuppressed by small Yukawa couplings) are either the lightquark initiated processes gq(q) → Hγ q(q) and qq → Hγ g (both involving a top-loop ggH vertex) or gg → Hγ g (via a pentagon top loop) [ Figure 1]. The contribution of the gg → Hγ g amplitude to the Hγj rate has been recently evaluated in [4], where the latter process is claimed to be responsible for the dominant production of Hγj final states at the LHC, followed by the heavy-quark QQ, gQ scattering into Hγ.
In the present study, we will show that the gg → Hγ g channel actually contributes to the inclusive Hγj production to a much lesser extent than previously stated. Indeed, on the one hand, the ggH-loop-mediated production via the t-channels gq(q) → Hγ q(q) will be found to be about one order of magnitude larger than the one mediated by the top-pentagon amplitude gg → Hγ g at the LHC. On the other hand, the actual (by far) dominant production of Hγ final states accompanied by jets will turn out to proceed through an electroweak process, that is the VBF Higgs production qq → Hγ qq, where the high-p T photon radiation by the initial/final quarks, connected by W charged currents, is enhanced by the absence of suppressive QCD coherence effects [5,6].
We will also evaluate for the first time the contributions to the inclusive Hγ production arising from a hard photon radiated in the associated production of a Higgs plus either a topquark pair in Htt final states or a single top in the t-channel Ht (Ht) production. The Httγ will be found to contribute at the LHC at the same level as the gq(q) → Hγ q(q) channels. Remarkably, the relative Httγ weight increases at larger c.m. energies, approaching the relative (still dominant) contribution of the VBF component at √ S ∼ 100 TeV. The Htγ component is of course lower than the Httγ one, but less than naively expected, thanks to the enhancing mechanism related to the W -exchange in the t-channel Ht production, similar to the one acting in the VBF case [5,6].
The HV γ (with V = W, Z) associated production has been evaluated in [7,8] at the NLO in QCD. As we will see, at the LHC energies, it also contributes to the Hγ production in a comparable way to the t-channels gq(q) → Hγ q(q) (and to the Httγ process).
We then reconsider the total contribution to the Hγ final states of the heavy-flavour (bb and cc) scattering [9], that will be found comparable to the gg → Hγ g process at the LHC. We finally comment on the minor Hγ component due the qq → Hγ electroweak-loop process previously studied in [9].
We stress that the present study is not aimed to provide the most accurate estimate of the production cross sections for different Hγ channels, but rather to analyze the inclusive production of the Hγ system through its main components, discussing the corresponding relative weight. Such a discussion has non-trivial aspects that, to our knowledge, have not previously been detailed in the literature.
The plan of the paper is the following. In Section 2, we discuss the cross-section computation for the different channels contributing to the Hγ final state in proton-proton collisions. For some processes, a QCD next-to-leading-order (NLO) evaluation is available, for others, even the tree-level estimates are, to our knowledge, still missing in the literature, and we provide them here. In particular, we present LO Hγ cross sections relevant for LHC and future higher-energy pp colliders. In Section 3, we compare the different contributions, looking at both total rates and Higgs/photon kinematical distributions. Hierarchies of Hγ cross sections are then compared with the ordering of original Higgs production mechanisms. In Section 4, we present our conclusions and outlook. In the Appendix, we report the asymptotic behavior of the top pentagon amplitude for the gg → Hγg channel.

Processes contributing to the Hγ final state
In this section, we detail the present theoretical knowledge of the various channels contributing to the associated production of a Higgs boson and a high-p T photon at the LHC. For a few of them QCD NLO predictions are available, others have been computed only at leading order (LO), while some processes like the Higgs production in association with top quarks to our knowledge have not yet been considered in the literature. We will discuss the main processes in order of decreasing relevance of the corresponding channels with no photon emission, which are responsible for the dominant Higgs boson production at the LHC. Remarkably, we will see that the request of an extra high-p T photon in the basic Higgs production processes will have a strong impact on the relative weight of different channels.
When quoting the Hγ production rates in the present study, we will assume common sets of input parameters and kinematical cuts. The latter are needed in most of the channels considered, which are characterized by collinear-and soft-photon (and -parton) divergencies.
The setup applied in all cross-section computations (even in absence of divergencies) is where ∆R(a, b) = ∆φ(a, b) 2 + ∆η(a, b) 2 is the angular separation between a and b, and j i (i = 1, 2) is any parton in the final state. We then set the Higgs and the heavy quark masses as follows : where we assumed the running masses at the m H scale in the Yukawa couplings entering the bb, cc → Hγ cross sections.
In the present analysis, we are mainly interested in establishing the relative importance of the main processes giving rise to Hγ final states. Since, for most of the channels we will analyze in the following, QCD NLO cross sections have not yet been computed, in order to make a consistent comparison, we will always consider QCD LO rates (even when QCD NLO estimates are already available in the literature).
When considering the absolute rates that we will present below, one should then keep in mind the influence of the cut choice on the cross sections. In Section 3, we will show how relaxed requirements on kinematics can influence the relative weight of different channels.
To compute LO cross sections and distributions we use AlpGen [10] 1 , with the parton distribution set CTEQ5L [11], setting the factorization (µ F ) and renormalization (µ R ) scale at the common value µ = m H . We will estimate the LO cross-section uncertainties by varying In the next subsections, we will report cross sections for the different channels according to the parameters and settings described above, for proton collision c.m. energies of 14 TeV, 33 TeV and 100 TeV, the latter being relevant for Future Circular Collider (FCC) studies that are presently under way [12] 2 . Proton collisions at 33 TeV could be realized in an upgraded-energy program of the LHC (usually named HE-LHC [15]).

QCD production via top loops in gg, qg(qg),qq → Hγ j
When asking for an extra photon in the main Higgs production channel gg → H proceeding via gluon fusion into a top triangle loop, one is forced to pass to the next QCD order, and include an extra parton in the collision final state. Indeed, as already mentioned, Furry's theorem forbids the emission of a photon from the ggH top-quark loop, and the gg → Hγ amplitude vanishes 3 . Then, one can either require a further gluon emission in the latter process, and obtain a non-vanishing gg → Hγg amplitude via a top pentagon loop ( Figure 1), or ask for an extra photon radiation in the Higgs+jet production proceeding via the channels gq → Hq (gq → Hq), andqq → Hg by means of a top triangle loop ( Figure 2 and 3, respectively).
The production of Hγj final states from the gluon fusion gg → Hγg channel at hadron colliders has been recently studied in [4]. The gg → Hγg amplitude is gauge invariant and finite, and one can compute the gluon-fusion separate contribution to Hγ production. In our analysis, we have redone the evaluation of the top pentagon amplitude A Pent associated to the gg → Hγg channel. We detail our computation in the following.
A Pent is given by the sum of the 24 pentagon-like diagrams obtained by permuting in all possible ways the external vectors in Figure 1. Each diagram can be expressed in terms of a linear combination of one-loop scalar boxes, triangles, bubbles, massive tadpoles and rational terms. The coefficients of all scalar functions have been obtained numerically via the OPP approach [16], as implemented in CutTools [17], linked to the one-loop scalar functions in [18]. As each pentagon is separately ultraviolet (UV) convergent, no rational term of the R 2 kind is present [19]. Thus, the full rational part of A Pent is R 1 -like, and also numerically provided by CutTools.
The input needed by CutTools is the integrand of each diagram as a function of the integration momentum. In order to speed up the computation, we have used an in-house implementation of the massive helicity method [20] that expresses traces over gamma matrices in terms of scalar products in the spinor space. This gives a numerical stable answer for most of the phase-space points. In order to detect and rescue the remaining unstable configurations, we have used the built-in quadruple-precision facilities of CutTools. As a result, no randomly generated phase-space point is discarded during the Monte Carlo integration. As for the latter, the value of |A Pent | 2 computed by CutTools is plugged into a code based on AlpGen, which, besides integrating over the relevant phase-space, also takes care of the convolution with the gluon parton densities 4 .
In order to validate the correctness of our calculation, a numerical check of gauge invariance has been performed. In particular, we have numerically replaced polarization vectors by four-momenta, obtaining zero up to the machine precision. In addition, we have checked the numerical agreement between the result obtained when using a large input value for the top mass and the analytic asymptotic behavior of A Pent , as reported in the Appendix.
Finally, we have compared our outcome for the gg → Hγg channel with the results in [4], and found complete agreement on both numerical cross sections and kinematical distributions. We stress that there is no infrared divergence for either photons or gluons in the final state, and the maximum of the corresponding p T distributions is ruled by the top mass circulating in the pentagon loop. In particular, the photon and gluon p T distributions are both peaked at p max T ∼ 120 GeV at 14 TeV, while the Higgs distribution is maximal at p max T ∼ 80 GeV [4].
The gg → Hγg cross section corresponding to the setup in Eqs.
where the cross section central value assumes µ F = µ R = m H , and the upper (lower) variations correspond to µ F = µ R = 1 2 m H (µ F = µ R = 2 m H ). In [4], the gg → Hγg rate has been compared to the heavy-quark QQ + Qg → Hγj cross section (with Q = b, c), and claimed to provide the dominant contribution to the Hγj final state at hadron colliders. Here, we correct the latter statement, by including also the Hγj production initiated by light quarks, proceeding via the top triangle ggH vertex in either the t-channel (Figure 2) or the s-channel ( Figure 3). The corresponding LO cross section (summing up over the t and s channels, and assuming the same set of cuts and conventions as above) have been obtained by AlpGen, by a ggH effective vertex : Note that the s-channel qq → Hγg cross section provides a tiny component to the latter rates, amounting to about 2.8% of the total cross section at 14 TeV, and 1.9% at 100 TeV. As a result, at the 14-TeV LHC, the light-quark initiated contribution to the Hγj production turns out to be an order of magnitude larger than the pentagon gluon-fusion

Vector boson fusion
The Higgs-photon associated production in VBF is obtained by the emission of a highp T photon from either the initial/final quarks or the t-channel W propagators, as shown in Figure 4 for two representative diagrams out of the complete set. The qq → Hγ qq channel 5 has been studied at LO in [5], and at NLO in [6]. Apart from the setup detailed in Eqs. (2.1)-(2.2), we will assume a further cut on the quark-pair invariant mass, M j 1 ,j 2 > 100 GeV, hence depleting the contribution from the qq → HW/HZ associated production, which will be considered separately in the following.
Asking for an extra high-p T photon in VBF drastically increases the relative importance of the W W fusion component with respect to the ZZ one [5]. Indeed, in the ZZ fusion channel, destructive-interference effects occur between the photon radiation from initial and final quarks connected by a t-channel Z exchange. As a result, while asking for an extra central photon with p T > ∼ 20 GeV typically suppresses the W W -fusion cross section by two orders of magnitude, the corresponding decrease in the ZZ component is O (10 −3 ). This makes the ZZ fusion contribution to the total qq → Hγ qq cross section even smaller than naively expected, and almost negligible [5].
The total LO cross section for qq → Hγ qq, computed by AlpGen, is The VBF contribution is then found to be by far dominant over the top-loop mediated channels contributing to Hγ final states. In particular, σ(qq → Hγ qq) is almost an order of magnitute larger than σ(gq, gq, qq → Hγ q,q, g) at the LHC (cf. Eq. (2.6)), and six times higher at 100 TeV (cf. Eq. (2.8)). Note that here we are applying a p T > 30GeV requirement on forward jets, that is quite stricter than the 20-GeV cut usually applied in VBF studies at the LHC, hence considerably reducing the predicted cross sections. In Section 3, we will discuss the effect of relaxing the relevant cuts on transverse momenta. Contributions to the Hγjj production different from VBF can arise from the NLO treatment of the gg, qg(qg),qq → Hγ j channels analyzed in Section 2.1. These are expected to be quite depleted with respect to VBF [5], and will not be considered in this analysis.

Associated HW and HZ production
The Hγ final states arising from the associated production qq → HW , qq → HZ derive from the emission of a hard photon from the qq initial state, and, in the HW case, from either the W propagator or the final W (see Figure 5 for two representative diagrams out of the complete set). NLO predictions for qq → HγW and qq → HγZ have been presented in [7] and [8], respectively (see also [21]

Top-pair and single-top final states
The Higgs production via top-pair final states offers the unique opportunity to directly test the top Yukawa coupling. This channel is quite depleted with respect to the gg, VBF, and HV -associated production because of m t phase-space effects. Requiring an extra hard photon in the Htt final states affects this hierarchy, since high-p T photons are more naturally radiated in the production of (more spherical) massive charged systems (see Figure 6 for two representative diagrams out of the complete set). The Hγ tt cross section computed at The LHC Hγ tt cross section turns out to be in the same ballpark of the top-loop Hγj cross section, and also of the total HγV cross section (with V = W, Z). At larger √ S, the Hγ tt rate gets the upper hand, and approaches the VBF Hγ qq rate. At 100 TeV, the Hγ tt rate is just about half the VBF Hγ qq one, and we will see that the Hγ tt channel becomes the second most important production mechanism for Hγ final states. The Higgs production associated to a single top is an electroweak process that can proceed via three different channels at hadron colliders [22]. Here, we restrict to the tchannel bq → tHq which has the largest cross section. The t-channel rate is anyway quite small at the LHC. Nevertheless, its role has recently been emphasized for its sensitivity to a possible change in the relative sign of the ttH and W W H couplings [23].
In Figure 7, one can find two representative diagrams out of the complete set for bq → Hγtq. The W exchange in the t-channel gives rise to a radiative pattern similar to the one in W W fusion, where photon radiation from different quark legs does not interfere destructively. On the other hand, the photon radiation somehow weakens the original cancellation among the different components of the bq → tHq amplitude [22].
As a consequence, the requirement of an extra p T > 30GeV photon in the bq → tHq channel makes the cross section drop only by an amount O(10 −2 ). In particular, the AlpGen estimate for the LO bq → tHq cross section is where bq → Hγtq stands for a sum over the two charge-conjugated channels. 2)) is partly compensated by the larger c-parton distribution) is reduced by a factor (Q b /Q c ) 2 = 1/4 in the bb, cc → Hγ cross sections, as will be shown in Section 3. The bb → Hγ cross section has been evaluated in the SM in [9], and in supersymmetric extensions of the SM in [24].
Our AlpGen estimate in the five-flavor scheme (5FS), with running b and c masses evaluated at the m H scale in the Yukawa couplings, gives as a result  Table 1. Cross sections for Hγ associated production in pp collisions at 14 TeV, 33 TeV, and 100 TeV, for the dominant channels, for p γ,j T > 30 GeV. For comparison, also the cross sections for the corresponding channels without hard-photon radiation are reported, where the first row (gg, gq, qq) refers to the sum of the Higgs and Higgs-plus-one-jet contributions (see text). All cross sections are at LO, and computed via AlpGen. The complete set of selection cuts applied is described in the text.
Note that, although the gg, qq → H tt and bb → H cross sections are comparable at 14 TeV (cf. Table 1, next Section), the request of an extra photon depletes the bb → H with respect to not only the gg, qq → H tt channel, but also the single-top bq → tHq process (cf. Eqs. (2.18) and (2.21)).

Electroweakqq → Hγ production
A further channel mildly contributing to the Hγ associated production is qq → Hγ that occurs via light-quark annihilation, going through s-channel γ and Z exchange, involving a γγH and ZγH triangle loop of top quarks or W 's, and box diagrams with W 's and light quarks running in the loop. Figure 9 shows one diagram out of the complete set. The corresponding cross sections have been computed at the Tevatron and the LHC [9], and found to be quite smaller than the heavy-flavor tree-level QQ → Hγ contribution to the Hγ final state at the LHC. We will then neglect the corresponding rates in the present discussion.

Comparison of rates and distributions
In the previous section, we detailed the LO cross sections for the dominant Hγ production channels at different c.m. energies. We included a study of the scale dependence in order to get a flavor of NLO correction effects. We are now going to discuss how the LO central values (i.e., computed for µ = m H ) for cross sections of different processes compare, in order to pinpoint the main components of the Hγ inclusive production. We also confront the cross sections of various Hγ channels with the cross sections of the corresponding Higgs production channels where no high-p T photon is radiated. This will make manifest the fact that the presence of an extra photon in the final state deeply affects the hierarchy of importance for Higgs production channels.
In Table 1, we show, for √ S =14 TeV, 33 TeV, and 100 TeV, LO cross sections computed via AlpGen, with the parton distribution set CTEQ5L, and the factorization and renormalization scale at the common value µ = m H . We assume the setup defined by Eqs. (2.1) and (2.2), implying a cut p γ,j T > 30 GeV on the photon and (if present) final-jets transverse momenta. In Table 1, we alternate columns referring to cross sections for main Higgs production channels (with no final photon), named (H) √ S , with the corresponding ones where an extra photon is required, named (Hγ) √ S . Note that the first process considered (first row, named gg, qg,qq), corresponding to the original gluon-fusion Higgs production, includes in its photon-less (H) √ S component not only gg → H, but also the Higgs+jet channel proceeding at LO via the gg, qg(qg),qq → Hg, q(q), g scattering, mediated by an effective ggH vertex. This is to match the corresponding Hγ top-loop component, which requires at the lowest order an extra final parton in the processes gg, qg(qg),qq → Hγ g, q(q), g (as discussed in Section 2.1) 6 . Note also that, the corresponding (Hγ) √ S component gets only a minor contribution from the pentagon gg → Hγg process (see again Section 2.1). Table 1, it gets particularly clear that the requirement of an extra high-p T photon suppresses the original Higgs production rates by an amount that is widely dependent on the process. Top-loop production turns out to drop by a factor 10 −4 at all c.m. energies considered, and is the most suppressed process. Slightly less suppressed (by a factor ∼ 2.4 · 10 −4 ) is the bb → H rate. On the contrary, both VBF and single-top production loose just a factor 10 −2 when adding a photon, and present the least decreased rates. The rates for all other channels drop by a few 10 −3 , with a suppression factor increasing going from cc → H (10 −3 ), up to W H, ZH (∼ 1.3 · 10 −3 − 2 · 10 −3 ), and ttH (∼ 4 · 10 −3 ).

By looking at cross sections in
At the LHC, the most abundant Hγ production arises from VBF (22 fb), with oneorder-of-magnitude lower contributions from V H (3.2 fb), top-loop production (3.1 fb), and ttH (2.6 fb). At larger c.m. energies, VBF is still by far dominant, but the relative weight of top-loop and direct top production increases considerably. At √ S 100 TeV, VBF is about 0.31 pb (i.e., more than 50% of the total), ttH is about 0.16 pb, and all the remaining Hγ channels sum up to about 0.12 pb.
In Table 2 Table 3. Same as in Table 1 at √ S =14 TeV, and for p γ,j T > 20 GeV.   Tables 1 and 2. processes versus √ S are also plotted in Figure 10, which clearly shows the new hierarchy of different Higgs production channels.
We stress that all the rates (and corresponding hierarchies) presented in Tables 1 and  2 somewhat depend on the kinematical selection of the final state. On the one hand, all rates are affected by the choice of the photon p T cut (in general not in a universal way). On the other hand, the channels including jets among the final products are also sensitive to the jet selection. A different selection can hence affect in principle the relative weight of channels.
In Table 3, we present results at 14 TeV, when one relaxes the p γ,j T > 30 GeV cuts in Eq. (2.1) down to the less strict selection p γ,j T > 20 GeV, the latter being also quite realistic at the LHC energies. The most affected channels are VBF and the heavy-quark fusion channels. The former is quite dependent on both p γ T and p j T cuts, and as a consequence doubles its rate, the latter are, among the processes considered, the most sensitive to the p γ T cut. Indeed, the impact of a change in the kinematical selection can be guessed by looking at the various kinematical distributions for the different Hγ channels, which are shown in Figures 11, 12, and 13, for √ S = 14 TeV. All distributions are normalized to unity, after applying the kinematical cuts in Eq. (2.1). Distributions detailed by the lines named "Hjγ" refer to the gg, gq, qq channels mediated by the effective ggH vertex. One can see that the rate dependence on the photon, Higgs, and jet momenta of various processes can partly alter their relative weight when changing the kinematical selection. In Figure 11, the heavy-quark fusion bb, cc → Hγ channels present the steepest dependence on p γ T , while Hγ tt shows the mildest dependence among the channels considered. The p H T dependence in Figure 12 is somewhat more structured. All the processes but bb, cc → Hγ show the maximum of p H T distributions at a quite large p H T value. The VBFγ channel (where by VBFγ we name the Hγ production via VBF) shows the typical p H T ∼ M W maximum, which is also present in the basic VBF Higgs production. On the other hand, the Hjγ channel, mostly arising from qg → Hγq, has an even higher average p H T , since in this case the photon tends to be collinear with the initial/final quark, and the Higgs boson p T has to balance the p T of the jγ system, each component of which is required to have p T > 30 GeV. In Figure 13, we detail the jet distributions for the few processes where at least one jet is present in the final state.

Conclusions
We have made a general analysis of the processes giving rise to final states containing a Higgs boson and a high-p T photon in proton collisions at different c.m. energies, relevant at the LHC and future colliders. We showed that the request of an extra photon in Higgs production widely affects the normal hierarchy in the main Higgs production processes. In particular, most of the Hγ signal derives in general from the VBF production.
At the LHC, for p γ,j T > 20 GeV, VBF accounts for more than 70% of the Hγ final states in a LO analysis. The second most important process is the one mediated by the toploop effective ggH coupling which is responsible for about 9% of the production, although contributing only slightly more than the ttH direct top production, and a bit less than the total W H/ZH associated production. At larger c.m. energies (in particular at √ S > ∼ 33 TeV), ttH gets the upper hand, and becomes the second (following VBF) most relevant process.
As a result, asking for an extra photon in Higgs production reverses the order of importance of the gluon-fusion and VBF mechanisms. Remarkably, the emission of the photon highly suppress the ZZ fusion component with respect to the W W one in the VBF channel [5]. Hence, the requirement of an extra photon in inclusive Higgs production naturally selects samples with good purity of the W W VBF component, with a rate suppression factor of the order 10 −2 with respect to the main VBF Higgs channel.
In the present study, we have redone the computation of the gg → Hγg pentagon amplitude, confirming results recently appeared in the literature. On the other hand, we corrected the relative weight previously assigned to this channel in the production of Hγj final states, pointing out the dominant role of processes mediated by the ggH effective coupling. We also computed for the first time the associated Higgs and photon production in processes involving direct production of top pairs and single top.
Of course, the present analysis would be made more robust by a general NLO treatment of all processes. We anyway expect this refinement to keep the general features of the present discussion unchanged.
Accurate predictions for the associated production of a Higgs boson and a photon at the LHC will be crucial not only to test Hγ interactions, but also in probing new physics effects in the associated production of new scalar particles and photons [25], as well as in searching for resonant three-photon final states [26], [27]. For instance, in case of the production of a scalar φ with features departing from the Higgs-boson ones, the present cross-section hierarchy could be widely affected. Non-vanishing terms like ggφγ in the effective-Lagrangian interactions (violating C parity in the SM case) might resurrect the role of the gluon fusion process as a dominant production channel for the scalar-photon final state, and change the overall picture.