CMB-S4: Forecasting Constraints on f_NL Through μ-distortion Anisotropy

Diffusion damping of the cosmic microwave background (CMB) power spectrum results from imperfect photon-baryon coupling in the pre-recombination plasma. At redshift $5 \times 10^4 < z < 2 \times 10^6$, the plasma acquires an effective chemical potential, and energy injections from acoustic damping in this era create $\mu$-type spectral distortions of the CMB. These $\mu$ distortions trace the underlying photon density fluctuations, probing the primordial power spectrum in short-wavelength modes $k_\mathrm{S}$ over the range $50 \ \mathrm{Mpc}^{-1} \lesssim k \lesssim 10^4 \ \mathrm{Mpc}^{-1}$. Small-scale power modulated by long-wavelength modes $k_\mathrm{L}$ from squeezed-limit non-Gaussianities introduces cross-correlations between CMB temperature anisotropies and $\mu$ distortions. Under single-field inflation models, $\mu \times T$ correlations measured from an observer in an inertial frame should vanish up to a factor of $(k_\mathrm{L}/k_\mathrm{S})^2 \ll 1$. Thus, any measurable correlation rules out single-field inflation models. We forecast how well the next-generation ground-based CMB experiment CMB-S4 will be able to constrain primordial squeezed-limit non-Gaussianity, parameterized by $f_\mathrm{NL}$, using measurements of $C_{\ell}^{\mu T}$ as well as $C_{\ell}^{\mu E}$ from CMB $E$ modes. Using current experimental specifications and foreground modeling, we expect $\sigma(f_\mathrm{NL}) \lesssim 1000$. This is roughly four times better than the current limit on $f_\mathrm{NL}$ using $\mu \times T$ and $\mu \times E$ correlations from Planck and is comparable to what is achievable with LiteBIRD, demonstrating the power of the CMB-S4 experiment. This measurement is at an effective scale of $k \simeq 740 \ \text{Mpc}^{-1}$ and is thus highly complementary to measurements at larger scales from primary CMB and large-scale structure.