Dust Attenuation Curves in the Local Universe: Demographics and New Laws for Star-forming Galaxies and High-redshift Analogs

TL;DR

The paper addresses the diversity of dust attenuation curves in the local universe and their dependence on galaxy properties. It introduces an IR-luminosity constrained SED fitting approach (SED+LIR) to derive individual attenuation curves for a vast, multiwavelength galaxy sample, enabling high-resolution constraint of slope and UV bump. Key findings reveal a wide range of slopes, with local star-forming galaxies on average exhibiting steep, SMC-like curves and modest UV bumps, while high-redshift analogs tend to have similarly steep or steeper curves; the slope correlates strongly with optical opacity $A_V$ and stellar mass mainly through this opacity, while the UV bump shows only modest, mass- and SFR-related variation and little metallicity dependence. The work provides practical functional forms and a public GSWLC-2 catalog for use in high-redshift studies and IRX-$\beta$ diagnostics, advancing dust corrections in galaxy SED analyses and informing dust radiative-transfer models.

Abstract

We study dust attenuation curves of 230,000 individual galaxies in the local universe, ranging from quiescent to intensely star-forming systems, using GALEX, SDSS, and WISE photometry calibrated on Herschel-ATLAS. We use a new method of constraining SED fits with infrared luminosity (SED+LIR fitting), and parameterized attenuation curves determined with the CIGALE SED fitting code. Attenuation curve slopes and UV bump strengths are reasonably well constrained independently from one another. We find that $A_λ/A_V$ attenuation curves exhibit a very wide range of slopes that are on average as steep as the SMC curve slope. The slope is a strong function of optical opacity. Opaque galaxies have shallower curves - in agreement with recent radiate transfer models. The dependence of slopes on the opacity produces an apparent dependence on stellar mass: more massive galaxies having shallower slopes. Attenuation curves exhibit a wide range of UV bump amplitudes, from none to MW-like; with an average strength 1/3 of the MW bump. Notably, local analogs of high-redshift galaxies have an average curve that is somewhat steeper than the SMC curve, with a modest UV bump that can be to first order ignored, as its effect on the near-UV magnitude is 0.1 mag. Neither the slopes nor the strengths of the UV bump depend on gas-phase metallicity. Functional forms for attenuation laws are presented for normal star-forming galaxies, high-z analogs and quiescent galaxies. We release the catalog of associated SFRs and stellar masses (GSWLC-2).

Figures (13)

• Figure 1: Comparison between the IR luminosity derived from just the mid-IR flux point (22 ${\rm m}$, WISE channel W4), using the new calibration described in the text, and the IR luminosity derived from the full IR SED from Herschel (PACS and SPIRE data in combination with W4). Our procedure is based on the use of luminosity-dependent templates plus the application of corrections, and results in estimates that have a small dispersion and no offset with respect to the full IR SED values. Green solid line is the 1:1 line, whereas the dashed blue line is the robust linear fit. Values of the standard deviation and the correlation coefficient are given in the plot.
• Figure 2: Parameterization of dust attenuation curves. Following noll09, we define the dust attenuation curve as a two-parameter modification of the Calzetti curve (black solid line), shown in absolute formulation ($A_{ }/A_V$). Parameter $$ modifies the power-law slope of the curve (with negative values making it steeper in the UV/optical region), while $B$ specifies the amplitude of the 2175Å bump in the total formulation of the curve ($A_{\rm bump}/E(B-V)$). Fixed $B$ corresponds to roughly similar level of contribution of the attenuation due to the bump ($A_{\rm bump}$) to the total attenuation at 2175 Å ($A_{2175}$).
• Figure 3: Contour maps of dust attenuation curve parameters (average slope (left panels) and average UV bump strength (right panels)) as a function of the specific SFR and stellar mass, for all galaxies (upper panels) and galaxies classified as star-forming (lower panels). White line represents the median sSFR for star-forming galaxies (the "main sequence" of SF). The slope is expressed as the exponent of the power-law deviation ($$) with respect to the Calzetti curve, the latter being represented by the lightest contour ($=0$). There is a wide range of curve slopes, with lower-mass galaxies having steeper slopes. Bump strengths (amplitudes in units of $A_{\rm bump}/E(B-V)$, which for the Milky Way extinction curve has a typical value of 3) are on average weaker than the MW extinction curve bump, especially for star-forming galaxies of higher mass. Bins with 10 or more galaxies are shown.
• Figure 4: Contour maps of dust attenuation curve parameters as a function of stellar mass and the level of dust attenuation, in $V$ (upper panels) and FUV (lower panels). White lines are median trends. Attenuations are effective. The slope is strongly correlated with $A_V$, being steeper for low $A_V$ (and consequently higher $A_{\rm FUV}$). The UV bump is stronger for galaxies with lower optical opacity.
• Figure 5: Correlation between the dust attenuation slope and effective optical opacity. Attenuation curves of galaxies with higher optical opacity are systematically shallower. Slope is expressed as the exponent of the power-law deviation ($$) with respect to the Calzetti curve ($=0$). The average trend and the $1$ range around it are shown as green curves. The relation is in qualitative agreement with chevallard13 modelling results (red curve). Mild discretization of model grid parameters is visible.