Diphoton decays of stoponium at the Large Hadron Collider

Bottom line: the part of this I would actually cite is the Appendix. Martin recomputes the stoponium annihilation widths and finds that eqs. (A.1) and (A.2) of Drees–Nojiri [9] — the gg and γγ partial widths — are each too large by a factor of 2, with independent agreement claimed against [11] and [65]. He also generalizes the set to include sbottom mixing and CP phases. That is a real, checkable service, and it is the piece with a shelf life. The rest is a re-run of the Drees–Nojiri feasibility study with the top mass now known, CTEQ5L gluon densities, the Hagiwara et al. potential model for |R(0)|^2 and the binding energy, and model lines pinned to Omega_DM h² = 0.11 and the LEP Higgs bound. The targeting is the useful part: compressed SUSY with stop-mediated neutralino annihilation lands at m_eta_t around 400–600 GeV, and the baryogenesis window at 235–270 GeV. The conclusion is appropriately sober — 500 GeV gives S/sqrt(B) around 2 at 100 fb⁻¹, below 300 GeV maybe 10 fb⁻¹ — and Section V says so without dressing it up. On the stress test: it lands, but the author already says it. Section II states the NLO background is 'a factor of 2 or more' and declines to include it, and the figures carry no uncertainty band, so they read as calibrated when they are leading-order placeholders. Two things soften it. The signal is also leading order, and the excited-state feed-down he deliberately omits is quoted at roughly 1.5 in the signal's favor. And the sideband argument is right about measuring B, not about recovering significance. Still, a factor-2 band on Figs. 4, 7 and 9 would cost nothing and would keep the abstract's 'will be observable' from outrunning Section V. Two smaller items. Fig. 8's caption and the Section IV text swap the left and right panels (115 vs 120 GeV, 0.3 vs 0.5); the text's assignment is the physically sensible one, so the caption is the error. And the 120–135 GeV stop window leans on ref. [28], listed as 'to appear' — worth confirming it says what is claimed. Who should read it: anyone working on light stops with compressed spectra, and anyone who quotes stoponium widths. Send it out.

The central quantitative claim is the reach curves in Figs. 4, 7 and 9, which are S/sqrt(B) computed with a leading-order background (Fig. 2: qqbar -> gamma gamma plus gg -> gamma gamma). Section II states plainly: 'Higher-order corrections to the background can be quite important, increasing the cross-section after cuts by a factor of 2 or more [36]-[39].' Because S/sqrt(B) scales as 1/sqrt(B), a factor-2 background increase lowers every significance by sqrt(2) and raises every required luminosity in Figs. 4, 7 and 9 by roughly 2x. The paper declines to include these corrections ('beyond the scope of this paper') and attaches no uncertainty band, so the figures read as calibrated numbers when they are leading-order placeholders. The effect is directional and pessimistic: the 500 GeV point, already only S/sqrt(B) ~ 2 at 100 fb^-1, falls to ~1.4, i.e. no longer 'observable'; the m_eta_t <~ 300 GeV, 10 fb^-1 claim survives but with reduced margin. The paper's defense -- that the background will be measured from sidebands -- does not recover the lost significance; it only means B is measured rather than calculated. The missing signal higher-order corrections are unknown in sign and cannot be assumed to compensate. This is distinct from, though comparable in size to, the potential-model uncertainty the reader flags: the potential model moves |R(0)|^2/m^2 in either direction, whereas the NLO background moves the reach monotonically in the pessimistic direction. The abstract's unqualified 'will be observable over background' (abstract:s4) is therefore supported only in the lower-mass corner and only at leading order.

The paper is a careful, self-correcting feasibility study, and its core physics checks out. I verified the two widths that carry the correction: eq. (2.1) follows from eqs. (A.1)-(A.2) with the identical-particle factor 1/2, and eq. (2.2) follows from eqs. (A.1)-(A.3); the ratio Gamma(gamma gamma)/Gamma(gg) = (8/9)(alpha/alpha_S)^2 ~ 0.005 matches the text. The production formula eq. (2.5) is the standard narrow-resonance Drell-Yan expression, and the stated ratio to Higgs production is correct at leading order. The model lines are tied to external constraints (Omega_DM h^2 = 0.11 via micrOMEGAs, the LEP Higgs bound), and the limitations are stated candidly. So I do not think the central argument is broken; the concern is about calibration, not correctness. I partially agree with the reader. The potential-model extrapolation (eqs. 2.3-2.4, ref. [29]) is a genuine soft spot and the authors acknowledge it ('other potentials can give quite different results'). But I would weight the leading-order background treatment as at least as load-bearing, because the paper itself quantifies it ('a factor of 2 or more') and then does not propagate it into the reach figures, and because its effect is monotonic in the pessimistic direction rather than bidirectional. The reader mentions this in the rationale but does not make it the weakest assumption. The abstract/body tension the reader notes is real: abstract:s4 says the signal 'will be observable over background' and promises a 'uniquely precise measurement,' while Section V concedes the precision is 'quite sensitive to unknowns' and the 500 GeV point reaches only S/sqrt(B) ~ 2 at 100 fb^-1. That is an overstatement in the abstract, not an error in the analysis. My concern reinforces rather than overturns the reader's CONDITIONAL verdict: the channel is worth pursuing below about 500 GeV, but the numbers in Figs. 4, 7 and 9 should carry an explicit theory-uncertainty band and a look-elsewhere estimate before being treated as calibrated. I would keep the verdict at CONDITIONAL.

The paper is a careful, self-contained update of a long-standing proposal, and it earns credit on several counts. It corrects a factor-of-2 error in the published gg and gamma gamma widths of ref. [9] and verifies the corrected normalization against refs. [11, 65]; I checked eqs. (2.1)-(2.2) against the Appendix master formula (A.1)-(A.3) and they are mutually consistent. The Appendix gives explicit, reusable expressions for all relevant final states, so the calculation is reproducible. The spectra and relic density are obtained with standard public tools (SOFTSUSY, micrOMEGAs) with cross-checks against SuSpect, ISAJET and DarkSUSY, and the parton distributions are CTEQ5L. The reach estimates are deliberately conservative in places (no excited-state enhancement, a wider signal bin than ref. [9]). The result is a falsifiable prediction: a narrow diphoton peak at a mass tied to the stop mass, with quoted luminosity targets. These are genuine strengths and are appropriately reflected in the assessment below.

The paper asks whether the LHC can see stoponium: a bound state of the lighter top squark, which forms only when the squark has no allowed two-body decay into a chargino or neutralino. In that situation the squark lives long enough to bind, and the spin-0 ground state annihilates mainly into gluons, with a small but clean branch into two photons that would appear as a very narrow peak in the diphoton invariant mass spectrum. The author recomputes the production and decay rates, correcting factor-of-two errors in the earlier gluon and photon widths of the classic Drees–Nojiri analysis, and folds in the now-known top mass and a more conservative calorimeter resolution. He then evaluates the reach in two families of motivated models: compressed supersymmetry, where stop-mediated neutralino annihilation gives the observed dark matter density, and models with a first-order electroweak phase transition for baryogenesis. The claim is that in favourable regions of these models the peak is observable, and its position would read off a superpartner mass far more precisely than the usual kinematic edges of missing-energy events.

What carries the argument. The load-bearing object is the wavefunction at the origin, $|R(0)|^2/m^2_{eta_{tilde t}}$, taken from the non-relativistic potential model of Hagiwara et al. and parameterized in eqs. (2.3) and (2.4) together with the binding energy $2m_{tilde t_1} - m_{eta_{tilde t}}$. It fixes both the annihilation widths, $Gamma(eta_{tilde t} to gg) = frac{4}{3} alpha_S^2 |R(0)|^2/m^2$ and $Gamma(eta_{tilde t} to gamma gamma) = frac{32}{27} alpha^2 |R(0)|^2/m^2$, and the criterion for the bound state to form at all: the binding energy must exceed the total annihilation width, which it does comfortably since the gluon width is of order 2 MeV. The competing widths into $W^+W^-$, $ZZ$, $h^0h^0$, $t bar t$, $b bar b$ and neutralinos are collected in the Appendix, including interference terms and the possibility of resonant heavy-Higgs exchange, and these determine the branching ratios that rescale the idealized signal.

Core claim. The central claim is that $pp$ collisions at the LHC can produce the scalar stoponium ground state $p p to eta_{tilde t} to gamma gamma$, and that in realistic, dark-matter- and Higgs-consistent supersymmetric models this gives a detectable narrow resonance. The author writes the signal cross-section in a form that separates production from decay, $sigma = (pi^2/8 m^3) BR(eta_{tilde t} to gg) Gamma(eta_{tilde t} to gamma gamma) int dx tau x g(x) g(tau/x)$, so that any model's rate follows from the idealized case $BR(eta_{tilde t} to gg) approx 1$ by scaling with $1/[BR(eta_{tilde t} to gg)]^2$. In that idealized limit, the luminosity needed for $S/sqrt{B} = 2, 3, 4, 5$ in a bin of width $0.04 m_{eta_{tilde t}}$ is mapped over stoponium masses from 200 to 800 GeV; a 500 GeV resonance reaches only about $S/sqrt{B} = 2$ at $100$ fb$^{-1}$, while masses below roughly 300 GeV could be probed with about $10$ fb$^{-1}$. In the compressed-supersymmetry model lines the gluon branching ratio is typically 80% or more, and the photon branching ratio sits between 0.004 and 0.005, so the idealized reach largely survives, except where an $s$-channel heavy-Higgs resonance in the $b bar b$ or $t bar t$ channel suppresses the gluon width. In the baryogenesis model lines the stoponium is light enough that the reach is good even though $eta_{tilde t} to h^0 h^0$ takes a large share of the width.

The premise it rests on. Everything rests on a potential model extrapolated from charm and bottom quark bound states, and the paper itself notes that other potentials can give quite different wavefunctions and binding energies, which would move the luminosity targets by a comparable factor.

A diphoton peak at $M_{gamma gamma} approx 2m_{tilde t_1}$ minus a few GeV would give the lightest top-squark mass directly, complementing the mass differences obtainable from gluino and open-stop cascade endpoints.

In compressed supersymmetry the parameter region where stop-mediated neutralino annihilation yields $Omega_{DM} h^2 approx 0.11$ is the same region where the stoponium mass is 400–600 GeV, so the dark matter mechanism and the collider signal are tied to a common parameter choice.

Because the required luminosity scales as $1/[BR(eta_{tilde t} to gg)]^2$, a model with a suppressed gluon width — for instance through near-resonant $eta_{tilde t} to b bar b$ from heavy-Higgs exchange — can be pushed out of reach even at high luminosity.

In the baryogenesis models, where $235$ GeV $< m_{eta_{tilde t}} < 270$ GeV, a clear observation could come with less than $100$ fb$^{-1}$, and with less than $10$ fb$^{-1}$ if $eta_{tilde t} to h^0h^0$ is kinematically closed.

If the peak is found, the difference between twice the stop mass measured from cascade endpoints and the peak position would itself be a measurement of the stoponium binding energy, offering a rare handle on the strong-interaction potential between supersymmetric colour sources.

The same logic applies to any long-lived coloured scalar with closed two-body decays, so a null diphoton search at low mass would constrain not just stops but any such candidate whose width stays below its binding energy.

Since the background is best fixed from sidebands in the data, the practical sensitivity will hinge on the diphoton mass resolution and on how well photon isolation suppresses the $qg to gamma gamma q$ and jet-faking-photon contributions, which the paper deliberately leaves at leading order.

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review