CEERS Key Paper VII: JWST/MIRI Reveals a Faint Population of Galaxies at Cosmic Noon Unseen by Spitzer

Abstract: The Cosmic Evolution Early Release Science (CEERS) program observed the Extended Groth Strip with the Mid-Infrared Instrument (MIRI) on the James Webb Space Telescope (JWST) in 2022. In this paper, we discuss the four MIRI pointings that observed with longer wavelength filters, including F770W, F1000W, F1280W, F1500W, F1800W, and F2100W. We compare the MIRI galaxies with the Spitzer/MIPS 24$\mu$m population in the EGS field. We find that MIRI can observe an order of magnitude deeper than MIPS in significantly shorter integration times, attributable to JWST's much larger aperture and MIRI's improved sensitivity. MIRI is exceptionally good at finding faint ($L_{\rm IR}<10^{10} L_\odot$) galaxies at $z\sim1-2$. We find that a significant portion of MIRI galaxies are "mid-IR weak"--they have strong near-IR emission and relatively weaker mid-IR emission, and most of the star formation is unobscured. We present new IR templates that capture how the mid-IR to near-IR emission changes with increasing infrared luminosity. We present two color-color diagrams to separate mid-IR weak galaxies and active galactic nuclei (AGN) from dusty star-forming galaxies and find that these color diagrams are most effective when used in conjunction with each other. We present the first number counts of 10$\mu$m sources and find that there are $\lesssim10$ IR AGN per MIRI pointing, possibly due to the difficulty of distinguishing AGN from intrinsically mid-IR weak galaxies (due to low metallicities or low dust content). We conclude that MIRI is most effective at observing moderate luminosity ($L_{\rm IR}=10^9-10^{10}L_\odot$) galaxies at $z=1-2$, and that photometry alone is not effective at identifying AGN within this faint population.