A search for Higgs boson pair production in the $b \overline{b} γγ$ final state is performed. The proton-proton collision dataset in this analysis corresponds to an integrated luminosity of 308 fb$^{-1}$, consisting of two samples, 140 fb$^{-1}$ at a centre-of-mass energy of 13 TeV and 168 fb$^{-1}$ at 13.6 TeV, recorded between 2015 and 2024 by the ATLAS detector at the CERN Large Hadron Collider. In addition to a larger dataset, this analysis improves upon the previous search in the same final state through several methodological and technical developments. The Higgs boson pair production cross section divided by the Standard Model prediction is found to be $μ_{HH} = 0.9^{+1.4}_{-1.1}$ ($μ_{HH} = 1^{+1.3}_{-1.0}$ expected), which translates into a 95% confidence-level upper limit of $μ_{HH}<3.8$. At the same confidence level the Higgs self-coupling modifier is constrained to be in the range $-1.7 < κ_λ< 6.6$ ($-1.8 < κ_λ< 6.9$ expected).
Weighted di-photon invariant mass distribution summed over all categories and the two data-taking periods. The events in each category are weighted by $log(1+S_{SM}/B)$. $S_{SM}$ is the expected signal yield assuming $\mu_{HH}$=1, while B is the continuum background yield obtained from a fit to the sidebands plus the single Higgs boson background obtained from simulation, all in a ± 5 GeV window around the Higgs boson mass. The lines show the fit results for the continuum background only (light dotted), adding single Higgs boson backgrounds (black dotted) and the full fit (solid).
Weighted di-photon invariant mass distribution summed over all categories and the two data-taking periods. The events in each category are weighted by $log(1+S_{SM}/B)$. $S_{SM}$ is the expected signal yield assuming $\mu_{HH}$=1, while B is the continuum background yield obtained from a fit to the sidebands plus the single Higgs boson background obtained from simulation, all in a ± 5 GeV window around the Higgs boson mass. The lines show the fit results for the continuum background only (light dotted), adding single Higgs boson backgrounds (black dotted) and the full fit (solid).
The 95% CL upper limits on the signal strength, obtained with separate fits to Run-2 and Run-3 data as well as their combination. When computing the significance or upper limit for one data-taking period only, $\mu_{HH}$ of the other period is left free to vary. All other parameters of interest are fixed to their SM expectation.