Calibrating Mid-Infrared Emission Features As Diagnostics of Star Formation in Infrared-Luminous Galaxies via Radiative Transfer Modeling

TL;DR

This paper investigates how mid-infrared emission features, specifically the 6.2 and 11.2 μm PAH bands and the [Ne II] and [Ne III] lines, trace star formation in local ULIRGs and whether their calibrations depend on bolometric luminosity or AGN activity. Using archival Spitzer data for 42 ULIRGs and radiative transfer modeling (CYGNUS) to decompose IR luminosities into starburst, host, and AGN components, the authors derive log-linear calibrations linking MIR feature luminosities to starburst luminosity and star formation rate, and examine dependencies on starburst age and AGN strength. They find PAH and [Ne II] predominantly arise from star-forming regions, while [Ne III] has mixed origins; ULIRGs show deviations from calibrations derived in lower-luminosity systems, with starburst age significantly affecting slopes. The study further shows AGN continuum can suppress PAH detectability (EW), but does not erase the underlying star-formation–driven Neon emission, providing refined diagnostics for JWST/MIRI analyses of dusty, luminous galaxies up to redshifts near 3. The results highlight the need to use luminosity- and age-appropriate calibrations to avoid underestimating SFRs in extreme systems and offer practical relations for disentangling bolometric components from MIR spectra.

Abstract

Luminous infrared galaxies are key sites of obscured stellar mass assembly at z > 0.5. Their star formation rates (SFRs) are often estimated using the luminosities of the 6.2 micron and 11.2 micron polycyclic aromatic hydrocarbon (PAH) features, or those of the [Ne II] and [Ne III] fine-structure lines, as they are minimally affected by obscuration. It is uncertain whether the calibration of these features as SFR tracers depends on the starburst bolometric luminosity or the level of Active Galactic Nucleus (AGN) activity. We here investigate the relationship between the luminosities of PAH and Neon lines with star formation rate for highly luminous objects using radiative transfer modeling and archival observations of 42 local Ultraluminous (>= 10^12 L_sun) Infrared Galaxies (ULIRGs). We find that PAH and [Ne II] features arise mainly in star-forming regions, with small contributions from the AGN or host, but that the [Ne III] line has a mixed contribution from both star formation and AGN activity. We present relations between L_PAH and L_NeII, and both starburst luminosity and SFR. We find relations for lower luminosity (L_IR ~= 10^10-10^12 L_sun) systems underestimate the SFRs in local ULIRGs by up to ~1 dex. The 6.2 micron and 11.2 micron PAH features, and the [Ne II] line, are thus good tracers of SFR in ULIRGs. We do not find that a more luminous AGN affects the relationship between SFR and PAH or Neon luminosity, but that it can make PAH emission harder to discern. Our results and derived relations are relevant to studies of star-forming and composite galaxies at z < 3 with the James Webb Space Telescope.

Figures (7)

• Figure 1: Starburst luminosity versus 6.2$\mu$m (left) and 11.2$\mu$m (right) PAH luminosity, color-coded by SFR (left) and redshift (right). Models derived above (Equations \ref{['eq: 62sb']} and \ref{['eq: 11sb']}) are plotted atop the data.
• Figure 2: 11.2$\mu$m PAH Luminosity versus $L_{\rm Sb}$, color-coded by starburst age. The dashed and dotted lines represent models derived for the young (Equation \ref{['eq: young11_lsb']}) and old (Equation \ref{['eq: old11_lsb']}) subgroups, respectively. An analogous plot of 6.2$\mu$m PAH Luminosity versus $L_{\rm Sb}$ is similar.
• Figure 3: 6.2$\mu$m PAH Luminosity versus $L_{\rm Sb} + L_{\rm Host}$, including models developed by Cortsen2019 (Equations \ref{['eq:cortsf']} and \ref{['eq:cortall']}, shaded to -1$\sigma$) and our own model (Equation \ref{['eq:comb62']}). We note that Cortsen2019 plots are against $L_{\rm Tot}$. However, their sample contained very few AGN sources, so $L_{\rm Sb} + L_{\rm Host}$ is the closest comparison.
• Figure 4: SFR versus both 6.2 (left) and 11.2$\mu$m PAH luminosity (right). Models derived by Shipley2016 (dotted line), Xie_Ho_2019 (dash line), and Mordini2021 (dash-dot line) are plotted atop the data, along with our own models (Equations \ref{['eq: sfr62_alt']} and \ref{['eq: sfr11_alt']}; black, solid line). We also include the model derived converting $L_{\rm Sb}$ to SFR (Equations \ref{['eq: sfr62_kennicut']} and \ref{['eq: sfr11_kennicut']}; long dash line).
• Figure 5: [Ne2] (purple triangles) and [Ne3] (light purple circles) luminosity versus starburst luminosity (left) and SFR (right) with their corresponding models (Equations \ref{['eq: neii_lsb']} and \ref{['eq: neii_sfr']}) as solid lines.