About this Article

Background

Over a galaxy's lifetime, metals are produced in stars and deposited into the interstellar medium (ISM). These metals cycle between various phases of the ISM: some remain in the gas at different temperatures and pressures; others are locked into dust; and others are ejected through galactic winds into the circumgalactic medium—the region surrounding a galaxy within its virial radius but outside its disk—where they can rain back down into the ISM. This incessant cycling of material between stars, interstellar gas and dust, and galaxy halos—the baryon cycle—drives galaxy evolution. A critical, yet poorly understood, aspect of this baryon cycle is the depletion of metals from the gas to the dust phase via dust growth in the ISM, and vice versa, the return of heavy elements from the dust to the gas phase via dust destruction. Two key parameters in the lifecycle of metals in the neutral ISM are the dust-to-metal mass ratio (D/M) and the dust-to-gas mass ratio (D/G = D/M  Z, where Z is the metallicity).

The dust-to-metal and dust-to-gas ratios are set by the equilibrium between dust formation, dust growth, and dust destruction. Dust formation occurs in the winds of evolved stars and in supernova remnants, dust growth occurs primarily in the ISM, and dust destruction is due to supernova-driven shocks and harsh radiation from young massive stars. Additionally, inflows of pristine, low metallicity, dust-poor gas fueling star formation dilutes the dust and metal content of galaxies, while galaxy-scale outflows driven by star-formation eject the chemically enriched gas and dust into their halos and into the intergalactic medium (Feldmann 2015).

At high metallicity, abundant heavy elements in the gas-phase collide with dust grains at a higher rate than in low metallicity galaxies, where heavy elements are rare. In other words, timescales for dust growth are much shorter in high metallicity galaxies than metal-poor ones (Asano et al. 2013). As a result, there exists a "critical" metallicity above which interstellar dust growth is efficient and the dust-to-metal is high (most metals are depleted onto dust grains), and below which the dust-to-metal ratio is low, with the dominant dust input being winds from evolved stars and supernova remnants. Thus, the dust-to-metal ratio, gas-to-dust ratio, and more generally dust properties are expected to vary significantly between galaxies, particularly with metallicity, and within galaxies, depending on the local density of the ISM.

While the Spitzer and Herschel space telescopes have mapped the dust content and properties of galaxies both locally (e.g., Galliano et al. and references therein) and at high redshift (e.g., Rowlands et al. 2012), thereby providing the overall view of the distribution dust in galaxies, it takes the UV spectroscopic capabilities of the Hubble Space Telescope to perform a detailed census of metals in neutral gas and dust required to understand how heavy elements deplete from the gas into the dust phase, and thus how the fraction of metals in dust varies with environment. Indeed, the far-infrared (FIR) opacity of dust, used to convert the dust emission to a dust mass, is poorly constrained, leading to large systematics in dust-mass estimates from the FIR. Our large HST program, METAL (Metal Evolution, Transport, and Abundance in the Large Magellanic Cloud), was designed to perform an accurate and precise census of the dust and metal content in the LMC at half solar metallicity, by obtaining medium-resolution UV spectroscopy of the intervening interstellar medium toward background massive stars.

While the metals locked in dust cannot be observed directly, gas-phase abundances can be measured accurately for the key components of dust (Fe, Si, Mg, Ni) and other metals (Zn, S, Cu) using the gas absorption lines in the UV spectra. The photospheric abundances of young stars recently formed out of the ISM (O, B, and A stars), derived from UV-optical spectroscopy, can then be used as a proxy for the total (gas + dust) neutral ISM abundances. Once the total (gas + dust) and gas-phase abundances in the neutral ISM are known, the fraction of each metal in the gas (i.e., interstellar depletions) and in the dust, and therefore D/M, can be derived, along with abundance ratios tracing the level of depletion.

The METAL observations (Roman-Duval et al. 2019; Roman-Duval et al. 2021, ApJ, in press), combined with similar data in the Milky Way (Jenkins 2009) and SMC (Tchernyshyov et al. 2015; Jenkins & Wallerstein 2017) have shown that the depletions of different elements are tightly correlated, indicating a common physical origin. Hydrogen density appears to be the main driver for variations of the fraction of metals in the dust phase—the dust-to-metal ratio—within each of the Milky Way, LMC, and SMC. Summing up all the heavy-element abundances in the dust phase, we derive the dust-to-gas ratio (i.e., the abundance of dust) as a function of hydrogen column density (Fig. 1). The increase of the dust abundance by a factor 4–5 from the diffuse (N(H) ~ 10²⁰ cm^–2) to the dense (N(H) ~ 10²² cm^–2) phase confirms that dust grows faster and is destroyed at a slower rate in the dense ISM.

Additionally, the dust-to-metal ratio traced by the depletions decreases significantly at low metallicity, resulting in non-linear relation between metallicity and dust abundance (Fig. 2). This finding is consistent with the theoretical expectation that dust growth in low metallicity environments is not efficient enough to counteract dust dilution and destruction processes, leaving evolved stars as the dominant albeit weak dust source.

Variations of the dust properties and abundance, and how well such variations can be observationally constrained, have important implications for our understanding of the build-up of galaxies over cosmic times. For example, gas masses are often estimated based on far-infrared dust emission in both nearby (Bolatto et al. 2011; Schruba et al. 2012) and distant (Rowlands et al. 2012) galaxies, because 21 cm H Ⅰ emission cannot be observed beyond redshift z ~ 1 (Chowdhury et al. 2020) and the CO-to-H₂ calibration for tracing molecular gas is highly uncertain, particularly at low metallicity (e.g., Bolatto et al. 2013). Such estimates rely on the assumption of a dust-to-gas ratio, which as we show can vary significantly with density and metallicity. Our results provide some important constraints on these variations and can therefore potentially improve the accuracy of dust-based gas mass estimates in galaxies.

The differences in dust-to-metal ratio between the Milky Way, LMC, and SMC (left panel of Fig. 3) offset almost exactly the total (gas + dust) metallicity differences between these galaxies. This leads to the surprising finding that, for a given hydrogen column density, gas in the Milky Way, LMC, and SMC have the same gas-phase metallicities, despite the large differences in stellar mass and total metallicities for these galaxies (Fig. 3, right panel). This has important implications for our understanding of the chemical enrichment of the universe obtained through spectroscopic studies of damped Lyman‑α systems (DLAs; e.g., Rafelski et al. 2012; Quiret et al. 2016). Indeed, our results show that it would be impossible to distinguish between SMC-like and MW-like systems in DLA samples based on gas-phase metallicity alone. Our observations and analysis, which will ultimately include the ongoing METAL‑Z (Metal Evolution, Transport, and Abundance at Low‑Z (metallicity)) large HST program obtaining similar depletion measurements in IC1613 (15% solar metallicity) and Sextans A (8% solar metallicity), will provide the calibrations of depletions—the corrections needed to infer the chemical enrichment of the Universe through DLA metallicity measurements—as a function of abundance ratios and hydrogen column density.

References

Asano, R. S., et al. 2013, Earth, Planets and Space, 65, 213

Bigiel, F., et al. 2008, AJ, 136, 2846

Bolatto, A. D., et al. 2011, ApJ, 741, 12

Bolatto, A. D., et al. 2013, ARA&A, 51, 207

Chowdhury, A., et al. 2020, Nature, 586, 7829, 369

Feldmann, R. 2015, MNRAS, 449, 3274

Galliano, F., et al. 2018, ARA&A, 56, 673

Jenkins, E. B. 2009, ApJ, 700, 1299

Jenkins, E. B. & Wallerstein, G., 2017, ApJ, 838, 85

Quiret, S., et al. 2016, MNRAS, 458, 4074

Rafelski, M., et al. 2012, ApJ, 755, 89

Roman-Duval, J., et al. 2017, ApJ, 841, 72

Roman-Duval, J., et al. 2019, ApJ, 871, 151

Roman-Duval, J., et al. 2021, ApJ, in press

Rowlands, K., et al. 2012, MNRAS, 419, 2545

Schruba, A., et al. 2012, AJ, 143, 148

Tchernyshyov, K., et al. 2015, ApJ, 811, 78