Interstellar Gas-phase Element Depletions in the Small Magellanic Cloud: A Guide to Correcting for Dust in QSO Absorption Line Systems

TL;DR

This study measures and interprets the gas-phase element depletions in the Small Magellanic Cloud, providing insights into dust grain formation and offering a model to correct for dust effects in distant quasar absorption systems.

Contribution

It presents new measurements of element depletions in the SMC and compares them to Galactic patterns, proposing a model applicable to low-metallicity absorption systems.

Findings

Elements Si, S, Cr, Fe, Ni, and Zn show similar depletion patterns to the Milky Way.

Mg and Ti deplete less rapidly in the SMC than in the Milky Way.

Depletion sequences in the SMC can improve interpretation of distant quasar absorption lines.

Abstract

We present data on the gas phase abundances for 9 different elements in the interstellar medium of the Small Magellanic Cloud (SMC), based on the strengths of ultraviolet absorption features over relevant velocities in the spectra of 18 stars within the SMC. From this information and the total abundances defined by the element fractions in young stars in the SMC, we construct a general interpretation on how these elements condense into solid form onto dust grains. As a group, the elements Si, S, Cr, Fe, Ni, and Zn exhibit depletion sequences similar to those in the local part of our Galaxy defined by Jenkins (2009). The elements Mg and Ti deplete less rapidly in the SMC than in the Milky Way, and Mn depletes more rapidly. We speculate that these differences might be explained by the different chemical affinities to different existing grain substrates. For instance, there is evidence that the mass fractions of polycyclic aromatic hydrocarbons (PAHs) in the SMC are significantly lower than those in the Milky Way. We propose that the depletion sequences that we observed for the SMC may provide a better model for interpreting the element abundances in low metallicity Damped Lyman Alpha (DLA) and sub-DLA absorption systems that are recorded in the spectra of distant quasars and gamma ray burst afterglows.

Tables

Table 1. Table 1: SMC Target Stars a a Data and their sources are taken from Table 1 of Welty et al. (2012).

Star	AzVbbNumber in the catalog of Azzopardi & Vigneau (1982).	R.A. (J2000)	Dec. (J2000)	$V$	$\left(B-V\right)$	$E\left(B-V\right)$	Spectral	HST Obs.
		(${}^{hms}$)	(° ′ ″)	(mag)		Tot/SMC	Type	Pgm(s).ccObserving program numbers and principal investigators: (1) 13778 (E. Jenkins), (2) 9383 (K. Gordon), (3) 9116 (D. Lennon), (4) 12978 (D. Welty), (5) 7437 (D. Lennon), (6) 7480 (G. Koenigsberger), (7) 9094 (G. Koenigsberger), (8) 11623 (G. Koenigsberger), (9) 13373 (G. Koenigsberger), (10) 4048 (J. Hutchings), (11) 5608 (D. Welty).
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Sk 13	18	0 47 12.2	$-73$ 06 33	12.44	+0.03	0.20/0.16	B2 Ia	1, 2, 3
Sk 18	26	0 47 50.0	$-73$ 08 21	12.46	$-0.17$	0.15/0.11	O7 III	1, 4
$\cdots$	47	0 48 51.5	$-73$ 25 59	13.44	$-0.18$	0.13/0.09	O8 III	1, 5
HD 5045 (Sk 40)	78	0 50 38.4	$-73$ 28 18	11.05	$-0.05$	0.14/0.10	B1 Ia+	1
$\cdots$	80	0 50 43.8	$-72$ 47 42	13.32	$-0.13$	0.19/0.15	O4-6n(f)p	1, 5
$\cdots$	95	0 51 21.6	$-72$ 44 15	13.78	$-0.18$	0.14/0.10	O7 III	1, 5
$\cdots$	104	0 51 38.4	$-72$ 48 06	13.17	$-0.16$	0.06/0.02	B0.5 Ia	3
$\cdots$	207	0 58 33.2	$-71$ 55 47	14.25	$-0.20$	0.12/0.09	O7.5 V((f))	1
$\cdots$	216	0 58 59.1	$-72$ 44 34	14.22	$-0.13$	0.13/0.09	B1 III	3
HD 5980 (Sk 78)	229	0 59 26.6	$-72$ 09 54	11.85	$-0.23$	0.07/0.04	WN6h	6, 7, 8, 9ddWhile a detailed analysis of interstellar lines toward this star was undertaken by Koenigsberger et al. (2001), their column densities for the SMC features were derived using the formula for optically thin absorption, which we feel is inappropriate. Therefore, we derived all of the column densities independently.
Sk 85	242	1 00 06.9	$-72$ 13 57	12.07	$-0.12$	0.07/0.04	B1 Ia	1
$\cdots$	321	1 02 57.1	$-72$ 08 09	13.76	$-0.19$	0.12/0.09	O9 IInp	1
Sk 108	332	1 03 25.2	$-72$ 06 44	12.40	$-0.24$	$\cdots$	WN3+O6.5(n)	10, 11eeColumn densities were not derived here, but instead were taken from Welty et al. (1997), Mallouris (2003) and Sofia et al. (2006). When appropriate, we adjusted column densities to reflect changes in some $f$-values.
$\cdots$	388	1 05 39.5	$-72$ 29 27	14.09	$-0.21$	0.11/0.08	O4 V	1
Sk 143	456	1 10 55.8	$-72$ 42 56	12.83	+0.10	0.36/0.33	O9.7 Ib	1, 2
$\cdots$	476	1 13 42.5	$-73$ 17 30	13.52	$-0.09$	0.23/0.20	O6.5 V	1
Sk 190	$\cdots$	1 31 28.0	$-73$ 22 14	13.54	$-0.18$	0.11/0.07	O8 Iaf	1
Sk 191	$\cdots$	1 41 42.1	$-73$ 50 38	11.85	$-0.04$	0.14/0.10	B1.5 Ia	1, 3

Equations

$$D(X)=\log N(X)-\log N({\mathrm{H}}_{\mathrm{tot}})-\log (X/\mathrm{H}{)}_{\mathrm{stellar}},$$

$$t=r[(1-r^2)/(n-2){]}^{-1/2},$$

$$[X_{\mathrm{gas}}/\mathrm{H}]=B_X+A_X(F_{\ast }-z_X),$$

$$F_{\ast }(\mathrm{SMC})=\frac{D(\mathrm{Fe})-B_{\mathrm{Fe}}}{A_{\mathrm{Fe}}}+z_{\mathrm{Fe}},$$

$$(X_{\mathrm{dust}}/\mathrm{H})=(X/\mathrm{H}{)}_{\mathrm{stellar}}(1-10^{D_X}),$$

$$\frac{d(X_{\mathrm{dust}}/\mathrm{H})}{d{F}_{\ast }(\mathrm{SMC})}$$

$$A_X=1.13A_X(T15)$$

$$B_X=A_X(T15)(1.13z_X+0.334)+\delta (X{)}_0(T15).$$

$$A_X=1.016B_{2,X}$$

$$B_X=B_{2,X}(1.016z_X+0.744)+A_{2,X}.$$

Full text

dust, extinction – ISM: abundances – galaxies: ISM – galaxies: individual (SMC) – quasars: absorption lines – ultraviolet: ISM

††software: FITEXY (Press et al. 2007)

1 INTRODUCTION

When combined with theories of stellar evolution and nucleosynthesis, measurements of the abundances of elements in different parts of any galactic system reveal much about how it formed and evolved, how and when it exchanged gas with the intergalactic medium (both infall and outflow), and how its stellar populations changed with time. Stars of different masses, initial compositions, and age produce their own distinct imprint of element production (Wheeler et al. 1989 ; Timmes et al. 1995 ; McWilliam 1997 ; Chiappini et al. 1999 ; Matteucci 2003 ; Kobayashi et al. 2006). For our own Galaxy, objects are close enough that we can examine in great detail how different element groups, such as $\alpha$, Fe-peak, neutron-capture or cosmic ray spallation products, change from one location (or star) to the next. From theories of nucleosynthesis, we know that even-Z elements of intermediate mass, such as Mg, S, Si, Ca, and probably Ti arise primarily from fundamental reactions in massive stars and core-collapse SNe, while the odd-Z elements such as P, Na, Al, and K depend on having a neutron excess and thus are driven by the initial metallicities of the stars (Suess & Urey 1956 ; Burbidge et al. 1957 ; Cameron 1957). Type 1a supernovae are mostly responsible for the production of the Fe-group elements (V, Cr, Mn, Fe, Co, and Ni). In the studies of individual stars, we are able to trace the mix of these groups as a function of stellar ages (conventionally traced by their relative abundances of iron [Fe/H]), or by their memberships in dynamically distinct populations (thin disk, thick disk, bulge, etc.).

For other galaxies, we do not have access to the tremendous level of detail that we have from nearby. Nevertheless, the global values of some element abundances and their gradients across the surfaces of the galaxies can still be studied from the spectroscopy of emission lines from H II regions (Thuan et al. 1995 ; Izotov & Thuan 1999 ; Chen et al. 2005 ; Christensen et al. 2005 ; Ellison et al. 2005 ; Schulte-Ladbeck et al. 2005 ; Péroux et al. 2011 ; 2012 ; 2014). This information is useful for gaining a better understanding of the changes caused by internal processes, mass loss, and mergers from one galaxy to the next. However, to go to even greater distances and explore the elemental makeup of galaxies during early times in the history of our universe, i.e., at redshifts $z\gtrsim 2.5$, it becomes difficult to obtain much spectroscopic detail from the light that is emitted. Virtually all of our knowledge on element abundances arises from studies of UV absorption lines, redshifted to visible wavelengths, that are seen in the spectra of background quasars or the afterglows of gamma ray bursts.

From UV absorption-line studies of interstellar material in our own Galaxy, we know that the gas-phase abundances of different elements are depleted by condensation into solid form within interstellar dust grains.333Contrary to some early misconceptions, the elements O, S, and Zn are not* undepleted. Only N seems to be mostly undepleted.* We have known for some time that the strengths of these depletions vary strongly from one element to the next and from one sight line to another (Habing 1969 ; Wallerstein & Goldsmith 1974 ; Morton 1975 ; Jenkins et al. 1986 ; Savage & Sembach 1996 ; Dwek 2016). Researchers who study quasar absorption-line systems, such as Damped Lyman Alpha systems (DLAs having $\log N({\rm H~{}I})/{\rm cm}^{-2}\gtrsim 20.3$) or sub-DLAs ($18\lesssim\log N({\rm H~{}I})/{\rm cm}^{-2}\lesssim 20.3$) have had to use what we have learned from the depletion patterns in the Milky Way to make corrections that will help to define the pattern of the true, intrinsic abundances of any system (Pettini et al. 1994 ; Lu et al. 1996 ; Kulkarni et al. 1997 ; Pettini et al. 1997 ; Pettini et al. 1999 ; Prochaska & Wolfe 1999 ; Pettini et al. 2000 ; Hou et al. 2001 ; Ledoux et al. 2002 ; Prochaska & Wolfe 2002 ; Calura et al. 2003 ; Prochaska et al. 2003 ; Dessauges-Zavadsky et al. 2004 ; Vladilo 2004 ; Lopez et al. 2005 ; Dessauges- Zavadsky et al. 2006 ; Rodríguez et al. 2006 ; Levshakov et al. 2009 ; Meiring et al. 2009 ; Cooke et al. 2011 ; Rafelski et al. 2012 ; Som et al. 2013 ; Fox et al. 2014 ; Kulkarni et al. 2015 ; Prochaska et al. 2015 ; Som et al. 2015 ; Guber & Richter 2016 ; Morrison et al. 2016 ; Quiret et al. 2016 ; Wiseman et al. 2016). Some early attempts to characterize the depletions of certain elements in order to correct for them were devised by Savaglio (2001), Vladilo (2002) and Prochaska & Wolfe (2002).

In a comprehensive review of the gas-phase abundances of 17 different elements in the interstellar medium in our local region of the Milky Way, Jenkins (2009) presented an analysis that showed how strongly each element depletes into solid form (dust) as the overall levels of depletions for the other elements change from one sight line to the next. A remarkable finding from this investigation was that the logarithms of the depletion factors of different elements tracked each other in linear fashions, but at different rates. This unified picture of depletion greatly simplified our understanding of how elements disappear from the gas phase and bind into solid form. Going further, we were able to learn from these results that the proportions of different the atomic constituents of grains change as the overall severity of depletions changed (Jenkins 2009 ; 2013).

2 MOTIVATION

In principle, one might suppose that it should be a simple matter to use our knowledge of how depletions behave in our Galaxy to make corrections for such processes in other environments. Indeed, from the lack of any better choice many investigations invoked this method to determine the total element abundances in distant systems, most of which had metallicities of order 1/300 to 1/3 solar (Dessauges-Zavadsky et al. 2006 ; Dessauges-Zavadsky et al. 2007 ; Rafelski et al. 2014 ; Quiret et al. 2016). However, evidence has accumulated that shows that the depletion patterns probably change when the basic element abundances or production sequences are different. For example, in the Milky Way Si and Fe deplete at about the same rate, and the gas-phase logarithmic abundance ratio relative to the solar one [Si/Fe]${}_{\rm gas}\approx 0.8$. However, Wolfe et al. (2005) showed that without corrections for depletion [Si/Fe]gas starts at about 0.3 for DLAs with $[{\rm Si/H}]_{\rm gas}\lesssim-1.0$, and increases only slightly as the metallicities approach that of our Galaxy. After we correct for depletion, we would say that the intrinsic (i.e., gas plus dust) $[{\rm Si/Fe}]\approx-0.5$ for these systems, a trend that is contrary to $[{\rm Si/Fe}]\approx+0.5$ for stars with $[{\rm Fe/H]}<-1.5$ in the Milky Way (Timmes et al. 1995). Likewise, Ledoux et al. (2002) stated that “The correlation between [Mn/Fe] and [Zn/Fe] … cannot be accounted for by any dust depletion sequence: it implies either variations of the intrinsic Mn abundance relative to Fe from $-0.3$ to +0.1 dex and/or a relation between depletion level and metallicity.” They also stated that “The variations of [Ti/Fe] vs. [Zn/Fe] cannot be fitted by a single dust depletion sequence either.” These abnormalities may be explained by chemical considerations in pre- existing solids: for instance, Lodders (2003) presented examples where the condensation of the elements Ni and Ge depend on the presence of a host element Fe to create an alloy. Likewise, the formation of refractory compounds that contain Zn and Mn, such as ${\rm Zn_{2}SiO_{4}}$, ${\rm ZnSiO_{3}}$, or ${\rm Mn_{2}SiO_{4}}$, are aided by host minerals such as forsterite and enstatite. In some galaxy environments where the ratio of $\alpha$-group to Fe-peak elements differs from that of our own, or where the previous buildup of some element groups diverged from that of the Milky Way, certain elements may have had more or less than their respective host compounds, thus altering the depletion rates in a way that is difficult to predict.

From the preceding discussion, we can see a clear need to investigate the depletion sequences in a low-metallicity system, much as Jenkins (2009) had done with solar metallicity gas in the local part of our Galaxy. The Small Magellanic Cloud (SMC) is a nearby dwarf irregular galaxy that presents an excellent opportunity to perform such a definition. It has a considerably lower metallicity than that of the Milky Way, and its chemical evolution history might possibly present a better match to more distant galaxies with metallicities $[{\rm M/H}]\sim-1$ that are typically represented by DLAs at redshifts $z\lesssim 3.5$ (Rafelski et al. 2012). Moreover, most of its stars indicate foreground extinction curves that differ from those of stars in the Milky Way (Hutchings 1982 ; Bromage & Nandy 1983 ; Prevot et al. 1984 ; Gordon & Clayton 1998 ; Gordon et al. 2003 ; Cartledge et al. 2005 ; Maíz Apellániz & Rubio 2012 ; Hagen et al. 2016), which suggests deviations in the distributions of dust grain sizes and/or compositions (Draine & Lee 1984 ; Boulanger et al. 1994 ; Weingartner & Draine 2001 ; Zubko et al. 2004 ; Zonca et al. 2015). This difference could be relevant to studies of abundances in DLAs, since their extinction curves are similar to that of the SMC (Murphy & Bernet 2016). As with our Galaxy, we can measure the abundances of elements in young stars, although with less accuracy. These stellar abundances can serve as a standard for the combined element abundances in both gas and dust.

There have been a number of studies of ISM abundances in the SMC that have already been carried out using data from spectrographs on the Hubble Space Telescope (HST) and the Far Ultraviolet Spectroscopic Explorer (FUSE) (Roth & Blades 1997 ; Welty et al. 1997 ; Koenigsberger et al. 2001 ; Mallouris et al. 2001 ; Welty et al. 2001 ; Sofia et al. 2006). A dispute on Si abundances between Sofia et al. (2006) and Welty et al. (2001) for a single sight line, amounting to 0.19 dex, highlights the difficulties that have been encountered previously. Tchernyshyov et al. (2015) performed a more comprehensive survey that derived abundances of the elements Si, P, Cr, and Fe for many stars in the SMC using spectra recorded by the Cosmic Origins Spectrograph (COS) on the HST. (We compare our results to theirs in Section 9.2.) Our approach differs from previous ones by obtaining medium resolution STIS echelle data with broad wavelength coverages that enabled us to cover many transitions of differing strengths for a significant collection of sight lines.

3 SELECTION OF TARGET STARS

Our investigation of element depletions made use of 18 sight lines toward stars in the SMC. Table 4 lists the stars used in this study. Fourteen of the target stars were observed in our Cycle 22 observing program (46 orbits, Program ID = 13778, E. Jenkins, PI) on the Hubble Space Telescope (HST). In selecting which stars to observe, we made use of the SMC interstellar titanium abundances reported by Welty & Crowther (2010) as a guide for sampling a wide selection of relative depletions. As a supplement to our observations, an additional four stars were observed for other programs, which produced suitable spectra that were publicly available in the Mikulski Archive for Space Telescopes (MAST) maintained by the Space Telescope Science Institute. Nearly all of the stars had measurements of atomic and molecular hydrogen column densities reported by Welty et al. (2012) for gas only within the SMC. For most stars, the combined column densities of hydrogen in both atomic and molecular form exceeded $10^{21}{\rm cm}^{-2}$; only 3 stars had lower values (see Table 5 in Section 6). For such high column densities, corrections for unseen ionization stages should be negligible (Vladilo et al. 2001). Another important criterion in selecting stars was to insure that the projected rotational velocities $v\sin i>50\,\rm{km\,s^{-1}}$, so that stellar features would not create confusing continuum levels.

4 OBSERVING STRATEGY AND DATA

ANALYSIS

While it is usually desirable to obtain spectra at the highest possible wavelength resolution for analyzing interstellar absorptions, especially if the features are strong, we decided that a broad coverage in wavelength would register many different transitions of different strengths, which outweighed the importance of fully resolving the velocity structures within the lines. For this reason, we constructed our observing program to record spectra of the SMC stars using the medium resolution ($\lambda/\Delta\lambda=30,000-45,800$) echelle modes (E140M and E230M) of the Space Telescope Imaging Spectrograph (STIS) on the HST. Typical signal-to-noise ratios per resolution element ranged from about 10 at 1800 Å, to 30 at 1300 Å, and 40 at 2300 Å.