Cosmological Collider Searches beyond the Hubble Scale with Planck Data

Abstract

Searches for primordial non-Gaussianity (NG) has the potential to not only reveal the physics of cosmic inflation, but also the structure of fundamental interactions at the highest energies. The cosmological collider (CC) physics program exemplifies this possibility and demonstrates how searches for oscillatory NG can lead to mass-spin spectroscopy of extremely heavy states. Adopting an effective field theory approach, we find the class of Feynman diagrams that can give the largest NG mediated by a heavy scalar particle with mass $M\sim H$, the inflationary Hubble scale. We compute the full shape of the NG and perform the first search for this shape using Planck data, finding no evidence for NG. This search loses its sensitivity as $M\gg H$ since quantum vacuum fluctuations cannot efficiently produce such heavier particles. We then focus on a mechanism where a chemical potential excites on-shell scalar particles with mass $M\gg H$. Computing the full shapes, we perform the first CC search for particles parametrically heavier than $H$ using Planck data. For a range of chemical potential $ω$ and $M$ satisfying $ω-M \simeq 3H$, we find a global $1.7σ$ evidence for non-zero NG, after taking into account the look-elsewhere effect.

Figures (3)

• Figure 1: Results of the search performed in this work with Planck data. Top: Estimated strength of the bispectrum ($f_{\rm NL}$) mediated by a heavy scalar with mass $M$ via the 'triple-exchange' diagram. This search is consistent with $f_{\rm NL}=0$ at 95% CL. Bottom: Estimated $f_{\rm NL}$ in the 'scalar chemical potential' Bodas:2020yho scenario for two different shape functions originating from temporal and covariant derivative couplings. For the latter and for a chemical potential $\omega=10H$ and $M=6.7H$, we find $f_{\rm NL} = -203 \pm 82$ at 68 % CL, with a $2.5\sigma$ local, corresponding to $1.7 \sigma$ global significance for $f_{\rm NL}\neq 0$. Similar significances are obtained for other combinations satisfying $\omega-M \simeq 3H$. The error bands represent 68% CL. See the main text for further details.
• Figure 2: Constraints on the QSFI model from the Planck search for bispectrum mediated by the triple-exchange diagrams. The bispectra search probes heavy scalars for $1.5H\leq M \leq 2H$, while for larger $M$ the bispectrum constraint weakens, and are not shown. The partial wave unitarity bounds $\kappa \leq 4\sqrt{\pi} M$ are violated above the color-coded horizontal lines.
• Figure 3: Comparison of ${\cal{S}}^{\rm cov}(\omega=10H, M=6.7H)$ ('Signal') with the local (top left), equilateral (top right), orthogonal (bottom left), and ${\cal{S}}^{\rm triple}(M=6.7H)$ (bottom right) shapes. The cosine correlation Sohn:2024xzd between the 'Signal' and the four shapes calculated by CMB-BEST are 5%, 54%, 22%, and 20% respectively.