JWST Whirlpool Galaxy Treasury: Mid-Infrared Emission in M51 and its Relation to Gas Column and Star Formation

Abstract: Using JWST/MIRI imaging of M51 in eight broadband filters, we investigate correlations of mid-infrared emission from polycyclic aromatic hydrocarbons (PAHs) and dust continuum with molecular, atomic, and ionized gas traced by CO(1-0), HI, and Pa-alpha, respectively. In molecular gas-dominated regions, PAH-dominated filters (F560W, F770W, F1130W, F1280W) exhibit near-linear correlations with CO(1-0) at 40 pc scale, indicating that PAHs are well-mixed with gas and experience relatively constant radiation field intensities. The F1500W, F1800W, and F2100W dust continuum-dominated filters show shallower slopes with CO(1-0), reflecting contributions from star-forming regions with high radiation field intensities. This is reinforced by the near-linear scaling between F2100W and Pa-alpha. PAH-dominated bands do not show this linear trend with Pa-alpha, likely due to their destruction in ionized regions. F1000W behaves similarly to PAH bands in its correlations with CO(1-0) and Pa-alpha. Modeling mid-infrared emission with an empirical decomposition into gas- and star-formation-associated components shows that PAH-dominated filters receive comparable contributions from both, while the relative contribution associated with the Pa-alpha template increases toward longer wavelengths, reaching $\sim$75% in F2100W. These results demonstrate that mid-infrared simultaneously traces the gas column and star formation, but with a systematic wavelength-dependent shift in what drives the correlations: PAHs being more gas-tracing and dust-continuum reflecting star formation. Lastly, considering both HI and H$<em>2$ at 440 pc resolution, we find a tight, linear relation between $Σ</em>{HI+H_2}$ and PAH-dominated filters. Although most of our coverage is in H$_2$-dominated regions, we note similar observations with HI, suggesting that PAHs are also well-mixed with atomic gas.

• The paper demonstrates that PAH-dominated MIRI bands scale nearly linearly with molecular, atomic, and total gas, with slopes of roughly 0.8–1.05, supporting their use as gas-column tracers in near-solar-metallicity environments.
• The paper finds that PAH emission rises super-linearly with Paα, with slopes of about 1.4–1.6, while the F2100W band tracks ionized gas nearly linearly and therefore provides a stronger resolved star-formation indicator.
• The paper’s two-component model shows that mid-infrared emission reflects both diffuse gas heating and active star formation, assigning about 58–62% of PAH-band flux and 74% of F2100W flux to the Paα-associated component while retaining a significant cirrus contribution.

Overview and motivation

This paper uses JWST/MIRI imaging of M51 in eight broadband filters (F560W–F2100W) to characterize how mid-infrared (mid-IR) emission from polycyclic aromatic hydrocarbons (PAHs) and warm dust continuum relates to the molecular, atomic, and ionized gas phases of the interstellar medium (ISM). The analysis is built on a simple physical expectation: for stochastically heated small grains at near-solar metallicity, mid-IR intensity scales approximately as $$I_{\rm MIR} \propto (D/G)\,U\,q_{\rm PAH}\,\Sigma_{\rm gas}$$, where $D/G$ is the dust-to-gas ratio, $U$ the radiation field strength relative to the Solar neighborhood, $q_{\rm PAH}$ the PAH mass fraction, and $\Sigma_{\rm gas}$ the gas surface density. Because both $\Sigma_{\rm gas}$ and $U$ vary across a galactic disk, the central question is which of these drivers dominates in each MIRI band. The paper addresses two questions: (1) how do molecular, atomic, and ionized gas contribute to observed mid-IR emission, and (2) how do the relative roles of gas column density and local radiation field vary with wavelength and environment.

The dataset combines Cycle 1 (FEAST) F560W/F770W and Cycle 2 Treasury observations covering the central $\sim$6′×2′ of the M51 disk at an adopted distance of 7.59 Mpc (1″ ≈ 46 pc), reduced with {\tt pjpipe} and zero-point anchored to WISE Channel 3 via the F1130W band. Ancillary tracers include CO(1–0) from PAWS (PdBI), a continuum-subtracted Pa$\alpha$ map from NIRCam F187N, and H I 21-cm data combining THINGS and newer VLA configurations. The nuclear bar ($r \leq 0.63$ kpc) is masked to avoid AGN-related contamination. Correlations are quantified with Spearman rank coefficients ($D/G$0) and power-law fits using binned {\tt linmix} regression, with intrinsic scatter estimated by minimizing reduced $D/G$1 on unbinned residuals.

Molecular gas: near-linear scaling with PAH bands, sublinear with dust continuum

At $D/G$240 pc resolution, all eight MIRI bands correlate moderately with CO(1–0) intensity ($D/G$3–0.47), but the power-law slopes show a systematic wavelength dependence. The PAH-dominated filters — F560W$D/G$4, F770W$D/G$5, F1130W, and F1280W (starlight-subtracted where noted) — exhibit near-linear scaling with slopes $D/G$6–1.08 over roughly two orders of magnitude in intensity. Notably, F1000W also behaves like a PAH band despite being nominally continuum-dominated, consistent with its substantial contribution (~30%) from PAH feature wings. Slopes become progressively shallower toward longer wavelengths, reaching $D/G$7 for F2100W. Intrinsic scatter is similar across all bands ($D/G$8–0.40 dex).

Environmental analysis using the Colombo et al. mask shows that CO-to-mid-IR ratios for PAH-dominated bands vary by less than ~0.5 dex across the molecular ring, spiral arms, interarm, and diffuse regions, supporting the interpretation that PAHs are well mixed with the molecular gas reservoir. In contrast, the CO-to-F2100W ratio drops by ~0.4 dex in high-$D/G$9 environments (molecular ring, H II regions), demonstrating that long-wavelength continuum emission is preferentially weighted toward intensely heated star-forming regions. This environmental dependence directly explains the wavelength-dependent flattening of the CO–mid-IR power laws: at fixed gas column, dust continuum intensity rises with $U$0, producing sublinear slopes.

Ionized gas: super-linear scaling with PAH bands, near-linear with F2100W

Pa$U$1 correlates more strongly than CO with all mid-IR bands ($U$2–0.76 at $U$340 pc), though with comparable intrinsic scatter (~0.38–0.47 dex). The paper argues that Pa$U$4 requires no extinction correction here because young clusters in M51 typically have $U$5, and beam dilution at 40 pc reduces the impact of highly embedded sources; this assumption is stated explicitly rather than tested independently.

The slope behavior mirrors and complements the CO results. All filters except F2100W show super-linear Pa$U$6–mid-IR relations ($U$7–1.60), with the steepest slopes in the PAH-dominated bands (~1.4–1.6). F2100W alone shows a near-linear relation ($U$8). The interpretation is that PAH emission does not keep pace with rising ionizing photon rate — attributed to PAH destruction in H II regions and photodissociation region boundaries — while 21 μm continuum scales almost directly with the ionizing photon rate and therefore acts as a reliable resolved star formation tracer at these scales. Robustness tests (alternative binning, percentile cuts, common masks) change fitted slopes by $U$9, preserving the wavelength trend.

A caveat acknowledged in the paper is that roughly half of M51's ionizing photons reside in diffuse ionized gas outside H II regions, so the Pa$q_{\rm PAH}$0–mid-IR correlations partly reflect extended UV heating rather than compact star-forming regions alone.

Total gas and atomic gas correlations

At $q_{\rm PAH}$1440 pc resolution, constrained by the coarser H I map, all bands correlate strongly with total gas surface density $q_{\rm PAH}$2 (using $q_{\rm PAH}$3 M$q_{\rm PAH}$4 pc$q_{\rm PAH}$5/K km s$q_{\rm PAH}$6): $q_{\rm PAH}$7–0.90, rms scatter of only 0.11–0.13 dex, and slopes of 0.61–0.84. These slopes are ~10% shallower than the cloud-scale CO measurements, which the authors attribute to beam averaging of bright star-forming regions with diffuse emission. The same wavelength-dependent pattern persists: PAH bands are steepest, F2100W shallowest.

For atomic gas, restricting to low-H$q_{\rm PAH}$8 regions ($q_{\rm PAH}$9 erg s$\Sigma_{\rm gas}$0 cm$\Sigma_{\rm gas}$1 sr$\Sigma_{\rm gas}$2) to isolate the diffuse ISM yields $\Sigma_{\rm gas}$3–0.75 between H I and mid-IR, with near-linear slopes for PAH-dominated filters (0.79–1.01). At higher mid-IR intensities the H I distribution turns over and saturates, consistent with the atomic-to-molecular transition near $\Sigma_{\rm gas}$4 M$\Sigma_{\rm gas}$5 pc$\Sigma_{\rm gas}$6: once the ISM becomes H$\Sigma_{\rm gas}$7-dominated, H I no longer tracks the rising dust heating. The implication is that mid-IR emission traces atomic gas where it dominates the dust-bearing column, but cannot be driven by H I alone within the molecular disk.

Two-component decomposition

Extending the linear template model of Leroy et al., the authors fit each MIRI band as $\Sigma_{\rm gas}$8, allowing a non-linear response to the ionized-gas tracer. The best-fit exponent $\Sigma_{\rm gas}$9 is consistent with unity for all filters except F2100W ($\Sigma_{\rm gas}$0), indicating that over the range of $\Sigma_{\rm gas}$1 probed in M51's disk, the mid-IR response to intense heating remains approximately linear — regimes where dust models predict non-linearity ($\Sigma_{\rm gas}$2) are not strongly sampled. The model reproduces observed intensities to within mean absolute logarithmic residuals of 0.16–0.18 dex.

The decomposition attributes ~58–62% of modeled flux to the Pa$\Sigma_{\rm gas}$3-associated component in the PAH-dominated bands, rising to ~74% in F2100W (with the CO component contributing ~26%). Systematic residuals are physically informative: short-wavelength bands are over-predicted at high intensities, consistent with PAH suppression in H II regions, while F1500W–F2100W are under-predicted at the bright end, reflecting warm dust continuum from star-forming regions. Roughly half of the 5–13 μm emission therefore arises from dust mixed with molecular gas and heated by the diffuse ISRF, with the remainder tied to star-forming environments — a quantitative confirmation that mid-IR emission is never a pure tracer of either quantity.

Implications for gas and SFR calibrations

For gas tracing, the near-linear CO-to-PAH slopes in M51 agree with the homogeneous PHANGS-JWST measurements of Chown et al. (slopes 0.91 ± 0.07 for F770W and 1.02 ± 0.09 for F1130W against CO(2–1)), reinforcing PAH emission as a molecular gas proxy at near-solar metallicity. The similarity of PAH scaling with H I, CO, and $\Sigma_{\rm gas}$4 suggests PAHs are maintained throughout the neutral ISM by the growth–shattering balance predicted by dust evolution models, and may even probe CO-dark H$\Sigma_{\rm gas}$5. Metallicity remains the key limiting factor: in low-metallicity systems such as Sextans A ($\Sigma_{\rm gas}$6), PAH emission collapses into shielded clumps, invalidating the linear framework.

For star formation tracing, the near-linear F2100W–Pa$\Sigma_{\rm gas}$7 slope (1.01) supports the use of 21 μm continuum as a resolved SFR indicator, though the ~25% CO-tracing "cirrus" contribution means even continuum bands contain diffuse-ISRF heating unrelated to current star formation. This contrasts with radiative transfer modeling of M51 attributing ~94% of 24 μm emission to young stars, a tension the paper notes without resolving. The super-linear PAH–Pa$\Sigma_{\rm gas}$8 slopes (~1.4–1.6) quantify PAH destruction in irradiated environments and caution against using PAH luminosities as uncalibrated SFR measures at high intensities.

Limitations and open questions

Several limitations bear directly on the results. The analysis footprint is restricted to the CO(1–0) coverage and excludes the nuclear bar, so conclusions apply to the star-forming disk rather than the innermost high-$\Sigma_{\rm gas}$9 environments. The two-component model assumes CO traces all molecular gas and Pa$U$0 traces all local heating; CO-dark H$U$1 and diffuse ionized gas (~50% of ionizing photons) are folded into these templates implicitly. The distinction between PAH and continuum contributions within each filter rests on PAHFIT-style spectral decomposition, and the nature of the mid-IR continuum itself remains unsettled in dust models. Environmental subsamples span narrow intensity ranges, preventing independent slope fits per environment. Finally, whether the near-linear PAH–gas relation holds at lower metallicity or in galaxies with substantially different ISRF distributions remains open, as does the physical origin of the residual tension between the empirical cirrus fraction at 21 μm and radiative transfer estimates.

Conclusion

Using full MIRI coverage of M51 combined with CO(1–0), Pa$U$2, and H I mapping at 40–440 pc resolution, this work establishes that mid-IR emission simultaneously traces gas column density and recent star formation, with a systematic, wavelength-dependent shift in the dominant driver. PAH-dominated bands scale near-linearly with molecular, atomic, and total gas (slopes ~0.8–1.05) but super-linearly with Pa$U$3 (~1.4–1.6), reflecting well-mixed grains and destruction in intense radiation fields. Dust continuum bands show the opposite behavior, with F2100W scaling nearly linearly with the ionizing photon rate while carrying a non-negligible ~25% diffuse-heating contribution. The empirical two-component decomposition quantifies this duality across 5.6–21 μm, providing a practical basis for selecting mid-IR bands as gas or SFR tracers under stated assumptions about metallicity, radiation field, and PAH abundance.