Feature Intensity Mapping: Polycyclic Aromatic Hydrocarbon Emission from All Galaxies Across Cosmic Time

TL;DR

Feature intensity mapping (FIM) generalizes line intensity mapping (LIM) to broad spectral features by convolving intrinsic feature SEDs $S_i(\lambda_ ext{rf})$ with instrument responses and constructing a clustering power-spectrum matrix across spectral channels to infer the redshift evolution $M_i(z)=b_i(z)rac{dL_i}{dV}(z)$. The framework uses a nearly assumption-free set of redshift anchors for $M_i(z)$ and a Fisher-information approach to forecast constraints, accounting for inter-channel correlations to mitigate interloper contamination. Applying FIM to SPHEREx and PRIMA forecasts, multiple PAH features can be detected with $S/N\gtrsim10$ across relevant redshift ranges ($z<0.5$ for SPHEREx; $1<z<5$ for PRIMA), with the 3.3 µm PAH feature being especially strong for SPHEREx and several mid-IR PAHs driving the high significance for PRIMA. This approach enables 3D maps of the aggregate PAH background, offering new probes of star formation, dust content, and metallicity evolution across cosmic time, and complements JWST and other LIM studies by accessing faint, unresolved galaxy populations.

Abstract

Line intensity mapping (LIM) is an emerging technique for probing the aggregate emission of a spectral line from all sources, without requiring individual detections. Through the wavelength-redshift relation, one can map the line-of-sight evolution of the line emission that traces the underlying large-scale structure in a spectral-imaging survey. In this work, we present a new technique -- feature intensity mapping -- as an extension of the LIM formalism to map broad spectral features in 3D, rather than the narrow emission lines typically targeted by LIM. By accounting for the convolution of spectral features with the instrument's spectral response across redshift, our technique enables simultaneous constraints on the redshift-dependent emission from multiple features. This approach enables 3D intensity mapping with some of the brightest features in the infrared spectra of galaxies: the polycyclic aromatic hydrocarbon (PAH) emission bands. We forecast the detectability of PAH signals using feature intensity mapping with the ongoing SPHEREx mission in the near-infrared and the proposed PRIMA mission in the far-infrared. We find that $S/N$ of $\gtrsim 10$ per redshift bin of widths $Δz = 0.1$ and $0.5$ can be achieved at $z < 0.5$ and $1 < z < 5$ with SPHEREx and PRIMA, respectively, for multiple PAH features, suggesting a promising prospect for mapping the aggregate PAH emission at cosmological distances with upcoming datasets.

Figures (13)

• Figure 1: Top: SPHEREx spectral resolution of each channel. Bottom: SPHEREx surface brightness sensitivity per spectral channel in a $6".2$ sky pixel. The brown and gray points represent the SPHEREx all-sky and deep-field sensitivity, respectively.
• Figure 2: PRIMA surface brightness sensitivity per spectral channel in a wavelength-dependent beam ($5".1/7".6$ to $10".4/22".9$ from short to long wavelengths for PHI/FIRESS) with spectral resolution $R=10/100$ for PHI/FIRESS. The brown and gray points represent the PHI and FIRESS sensitivity, respectively, assuming a fiducial survey of 1000 hr of integration covering an area of $1/0.1$ deg$^2$ for PHI/FIRESS.
• Figure 3: Normalized observed spectral profiles ($\nu I_\nu$) as a function of observed wavelength for various spectral features considered in this work. The relative luminosity and spectral shape of each feature are described in Section \ref{['S:feature_models']}. The left, middle, and right columns show the cases for SPHEREx, PHI, and FIRESS, respectively, with three example redshifts in each row ($z=0$, 0.2, and 0.5 for SPHEREx; $z=1$, 3, and 5 for PHI and FIRESS). Solid lines represent the intrinsic spectral shapes of the features, and dots indicate the intensities that will be observed by the spectral channels of the respective survey. The central wavelength of each feature is marked with a dashed vertical line for visual clarity. The inset panels show a zoomed-in view of the same spectrum on a log scale to better visualize the faint features.
• Figure 4: Top: Bias-weighted luminosity density as a function of redshift for the four spectral features considered in the SPHEREx case from our model (Section \ref{['S:signal_modeling']}). Bottom: Bias-weighted intensity as a function of observed wavelength, derived from the bias-weighted luminosity density shown in the top panel using Equation \ref{['E:dLdV_to_nuInu']}.
• Figure 5: Top: Bias-weighted luminosity density as a function of redshift for the four spectral features considered in the PRIMA case from our model (Section \ref{['S:signal_modeling']}). Bottom: Bias-weighted intensity as a function of observed wavelength, derived from the bias-weighted luminosity density shown in the top panel using Equation \ref{['E:dLdV_to_nuInu']}.