Debosmita (Deb) Pathak

Debosmita (Deb) Pathak

they/them

  • NSF Graduate Research Fellow
  • Dean's Distinguished University Fellow, Ohio State
  • 2026 IPAC Visiting Graduate Student Research Fellow

Mid-Infrared Colors Vary with Galactic Environment: Contrasting Star-Forming Disks, Young Centers, and Quiescent Star-Formation Deserts

Published in The Astrophysical Journal, Volume 1008, Issue 1, id.35, 23 pp., 2026

Download Paper | See in ADS | See Figures

Abstract

We present 50─100 pc-resolution JWST/MIRI and NIRCam measurements of mid-infrared (mid-IR) color variations in the diffuse interstellar medium (ISM) of 71 nearby star-forming galaxies from the PHANGS-JWST survey. Mid-IR emission traces the dust column density, intensity (U) and hardness of the interstellar radiation field, and the physical state (charge, size) and abundance of polycyclic aromatic hydrocarbons (PAHs). Mid-IR colors that trace PAH band ratios remain fairly constant in the diffuse ISM of star-forming disks. However, they show stark variations in extreme environments: highly star-forming central molecular zones (CMZs) and star formation deserts/quiescent bulges. In CMZs, PAH-to-continuum (3.3/21, 7.7/21, and 11.3/21 μm) and the 10/21 μm continuum colors are 0.2─0.4 dex lower than in normal disks. We attribute this to higher U based on the far-infrared dust colors and the high 21 μm/ΣMol, which we suggest to be a good tracer of U outside star-forming regions. Meanwhile, star formation deserts show low 7.7 μm PAH emission, resulting in low 7.7/21 μm and 7.7/11.3 μm, while all other mid-IR colors remain typical. This suggests the presence of more neutral PAHs in star formation deserts, where low 7.7 μm likely reflects ISM conditions similar to early-type and elliptical galaxies. All environments form part of a continuous trend in 7.7/11.3 μm versus specific star formation rate.

Figures

Refer to caption
Refer to caption
Refer to caption
Refer to caption
Refer to caption
Refer to caption
Figure 1: Zoom-ins showing a few example galaxies from our sample in F300M, F770Wss, F2100Wss, and \( R_{\rm PAH}^{*} \), with Querejeta et al. [71] morphological environment masks for centers (blue) and bars (red) overlaid for reference.
Refer to caption
Figure 2: Correlating environment-integrated median \( \Sigma_{\rm SFR} \) (\( M_{\odot}\,\rm yr^{-1}\,kpc^{-1} \); ) and sSFR (\( \rm yr^{-1} \); ) measurements from MUSE with JWST F2100W intensities and F2100W/F300M, respectively, for the 42 galaxies with joint MUSE and JWST coverage. Environments (§II.3) are indicated with colors — old bulges (red), young star-forming centers (blue), normal centers (light blue), and normal disks (light yellow) for targets with full MIRI filter coverage (20/71; X’s) and partial filter coverage (51/71; circles). We include \( \log{-}\log \) fits to all environments (light grey); the corresponding correlation coefficient \( \rho \) and p-value; and the threshold in F2100W and F2100W/F300M used to classify young centers (dark blue) and old bulges (red) as in §II.3 are indicated as horizontal dashed lines.
Refer to caption
Figure 3: Left: Constructing radial profiles of \( \rm F770W_{ss}/F2100W_{ss} \) for NGC1300 . The solid red line shows our fiducial measurement constructed from the ratio of medians taken across the full data set. The other lines show the impact of varying the methodology. Green shows the profile without starlight subtraction for all pixels (solid green), for only diffuse emission (dot-dashed green; see §II.5.2), and constructed taking the median of pixel-wise ratios instead of the ratio of medians (dashed green). Other red lines show varying methodologies with star-subtraction applied: considering diffuse emission only (dot-dashed red), all pixels (solid red), or the median of pixel ratios (dashed red). Blue shows the ratio of median values for only nebular regions. The black dot-dashed line shows the typical value for the diffuse emission from Sutter et al. [84]. Middle: Testing agreement between ratio of medians vs median of ratios for all radial bins, including both diffuse and nebular pixels, across 71 galaxies with \( R_{\rm gal}\leq 3\ell_{*} \). Right: Testing agreement between ratio of medians when including all pixels vs including diffuse pixels only (masking nebular pixels before binning) for all radial bins at \( R_{\rm gal}\leq 3\ell_{*} \).
Refer to caption
Figure 4: The center of NGC7496, an example of galaxy centers removed from analysis due to contamination by prominent diffraction spikes in F2100W. We show maps of contaminated F2100W and resulting \( R_{\rm PAH}^{*} \), blue contours for the center mask from Querejeta et al. [71], and black contours for median \( R_{\rm PAH}^{*} \) in the center.
Refer to caption
Figure 5: From left to right, F300M, F770Wss, F2100Wss, \( R_{\rm PAH}^{*} \), and resulting masks for diffuse (blue) vs nebular emission (pink) in the center of NGC1512, with dark blue contours for the morphological environment mask for the center of NGC1512 from Querejeta et al. [71].
Refer to caption
Figure 6: Top: Radial profiles of F300M, F770Wss, F2100Wss intensity for all 71 galaxies; F1000Wss and F1130Wss for 20 galaxies, colored by galaxy stellar mass \( M_{*} \). Bottom: Median (solid lines) and \( 16^{\rm th}{-}84^{\rm th} \) percentile range (shaded region) of radial profiles of F300M, F770Wss, F2100Wss, F1000Wss and F1130Wss in six percentile bins of \( M_{*} \). All radial profiles shown as a function of (projected) galactocentric radius in units of galaxy exponential scale-length \( \ell_{*} \).
Refer to caption
Figure 7: Radial profiles of \( \log R_{\rm PAH}^{*} \) for individual galaxies colored by total \( M_{*} \) (left), and then median profiles for groups of galaxies sorted into percentile bins of \( M_{*} \) (right). All galactocentric distances for the radial profiles are shown in units of the exponential scale length \( \ell_{*} \). For each binned profile, the median profile within a bin (solid line) and \( 16^{\rm th}{-}84^{\rm th} \) percentile range (shaded area) are included. Diffuse ISM median for 19/71 of our targets from Sutter et al. [84] shown as horizontal black dot-dashed lines.
Refer to caption
Figure 8: Plateau value in radial profiles of \( \log R_{\rm PAH}^{*} \) for each galaxy, measured as the median of all \( 0.1\ell_{*} \) annular bins with \( \ell_{*}\leq R_{\rm gal}\leq 3\ell_{*} \), points colored by galaxy stellar mass, as a function of global sSFR. Median of 71 galaxies (orange solid line), \( 1\sigma \) scatter (orange hatched area), diffuse ISM median from Sutter et al. [84, horizontal black dot-dashed line], and Spearman rank correlation coefficient (printed) shown. See Table 1.
Refer to caption
Refer to caption
Refer to caption
Figure 9: \( \log R_{\rm PAH}^{*} \), \( \log\rm F770W_{ss}/F1130W_{ss} \), and \( \log\rm F1000W_{ss}/F2100W_{\rm ss} \) as a function of \( \log\rm\Sigma_{SFR}~(M_{\odot}\,yr^{-1}\,kpc^{-2};~top) \) and \( \log\rm sSFR=\log\Sigma_{SFR}/\Sigma_{*}~(yr^{-1};~middle) \) from MUSE, and \( \log\rm F2100W/F300M \) (bottom row), split by local environment—old stellar bulges (red), young CMZs (bright blue), normal centers (pale blue), and disks (yellow). The markers indicate medians within each environment, with corresponding Spearman rank correlation coefficients (\( \rho \)) printed. We indicate the 20 galaxies with coverage in all four MIRI filters as X’s, and the rest with only two MIRI filters as circles. For galaxy centers, we include medians within the full environment (X’s or circles) and selecting for only diffuse emission (blue triangles), and mark centers with known AGN [89, open squares;]. The median diffuse \( R_{\rm PAH}^{*} \) from Sutter et al. [84] is included for comparison (black dot-dashed line).
Figure 10: Mid-IR color-color variation by environment for 20/71 galaxies with full MIRI filter coverage. \( \rm F770W_{ss}/F2100W_{ss} \) (or \( R_{\rm PAH}^{*} \)) and \( \rm F770W_{ss}/F1130W_{ss} \) ratios vs \( \rm F1000W_{ss}/F2100W_{ss} \) for bulges (red), CMZs (dark blue), ‘normal’ centers (pale blue), and disks (yellow) shown.
Figure 11: Draine et al. (2021) dust model predictions for varying PAH charge and sizes (‘grids’)) assuming three different radiation field hardness models at \( \log U=1 \) – modified MMP (black), M31 bulge (red), and a 3 Myr-old starburst (blue); and a 3 Myr-old starburst at \( \log U=3 \). Arrows show the direction in which grids shift due to increasing PAH ionization, size, and \( \log U \). Modulo a 0.1 dex offset in F1000W between model grids and observations, changes in mid-IR colors for CMZs relative disks are consistent with increasing \( U \), while star-formation deserts or bulges lie off the standard grids. fig:mir-colors-with-grids
Refer to caption
Refer to caption
Refer to caption
Figure 12: Correlating far-IR dust temperature tracers with expected tracers of radiation field intensity: Herschel PACS far-IR ratios of specific intensities (\( I_{\nu} \)) for \( \sim 31 \) galaxies where PACS \( 70\,\mu \)m, \( 100\,\mu \)m, or \( 160\,\mu \)m data are available, with F2100W\( {}_{\rm ss}/I_{\rm CO(2{-}1)} \) (top row), F2100W\( {}_{\rm ss}/\Sigma_{\rm Mol} \) (middle row), and \( \Sigma_{\rm SFR} \) (bottom row), data points colored as in previous figures. \( \log-\log \) fits and corresponding Spearman rank correlation coefficients are included for comparison.
Refer to caption
Refer to caption
Figure 13: Top: \( R_{\rm PAH}^{*} \), \( \rm F770W_{ss}/F1130W_{ss} \), and \( \rm F1000W_{ss}/F2100W_{ss} \) as a function of F2100W\( {}_{\rm ss}/\Sigma_{\rm Mol} \), which correlates with \( T_{\rm dust} \) (and \( U \)). Points colored as in previous figures. Bottom: Predicted variation in \( R_{\rm PAH}^{*} \), \( \rm F770W_{ss}/F1130W_{ss} \), and \( \rm F1000W_{ss}/F2100W_{ss} \) with radiation intensity \( \log U \) for three different radiation hardness models from Draine et al. [30]—a 3 Myr-old starburst (blue), modified MMP or Milky Way average (yellow), and the stellar bulge of M31 (red). At each \( \log U \), we show the standard value (standard ionization and size; solid lines), maximum variation possible due to PAH ionization at standard grain size (hatches), maximum variation possible due to grain sizes at standard ionization (darker shading), and finally the maximum range of values possible at fixed \( U \) from changing both size and ionization in the models (light shading).
Refer to caption
Figure 14: PAH-to-continuum ratios for three PAH-dominated filters for the 19 galaxies from Cycle-1, points colored by environment as in previous figures: \( \rm F335M_{PAH} \) (from H. Koziol et al. submitted), \( \rm F770W_{ss} \), and \( \rm F1130W_{ss} \) relative to \( \rm F2100W_{ss} \). Dashed lines indicate the predicted variation in respective PAH-to-continuum ratios due to radiation hardness at fixed \( \log U=1 \) and standard grain size and ionization from Draine et al. [30].
Figure 15: Mid-IR and near-IR PAH color-color variation by environment. \( \rm F770W_{ss}/F1130W_{ss} \) and \( \rm F335M_{PAH}/F770W_{ss} \) (small, neutral PAH-to-small, ionized PAH) ratios vs \( \rm F335M_{PAH}/F1130W_{ss} \) (small, neutral PAH-to-larger, neutral PAH) ratios. D21 dust model grids at \( \log U=1 \) included for comparison, with direction of variation due to changing PAH charge and size indicated. No significant variation in dust grids is seen due to varying \( \log U \). The darker lines show the limiting case of 100% neutral PAHs (PAH\(^0\)) at \( \log U = 1 \) for the three radiation hardness models.
Refer to caption
Figure 16: MUSE sSFR vs median (markers) and \( 16^{\rm th}{-}84^{\rm th} \) percentile scatter (error bars) in Balmer decrement attenuation \( A_{\rm V} \) for bulges, centers, and disks in 42 galaxies. Medians and percentiles only measured where both \( \rm H\alpha \) and \( \rm H\beta \) are detected within each environment. While most low-to-moderate sSFR environments show low \( A_{\rm V}\approx 0.5 \), CMZs show higher \( A_{\rm V} \). The alternate \( y \)-axis shows the translation to \( \tau_{\rm 10\,\mu m} \), the optical depth at \( \rm 10\,\mu m \) due to silicate absorption [40, using], which in PHANGS CMZs results in \( <1\% \) attenuation.