The Detection of Hot Cores and Complex Organic Molecules in the Large Magellanic Cloud

Complex organic molecules (COMs, $\geqslant 6$ atoms including carbon; e.g., CH₃OH, CH₃OCH₃, CH₃OCHO) are widespread in the Milky Way galaxy where they have primarily been found in the environments of young protostars–hot cores and hot corinos (e.g., Herbst & van Dishoeck 2009). Hot cores are compact ($D\lesssim 0.1$ pc), hot (${T}_{\mathrm{kin}}\gt 100$ K), dense (${n}_{{\rm{H}}}\,\gt {10}^{6\mbox{--}7}$ cm⁻³) sources (e.g., Kurtz et al. 2000) where ice mantles have recently been removed from dust grains, either by thermal evaporation or sputtering in shock waves. Interstellar COMs may be a chemical link to the prebiotic molecules that were involved in the processes leading to the origin of life (Ehrenfreund & Charnley 2000).

The observed hot core molecules may have their origin in the chemistry of the cold prestellar phase, either through formation in gas-phase reactions followed by freeze-out on the dust, or as products of reactions on the icy mantles of the dust grains (Brown et al. 1988). Grain-surface production could involve atom addition reactions on cold dust (e.g., Charnley & Rodgers 2008) or radical reactions on warm dust during the thermal warm up of the hot core (Garrod & Herbst 2006). Alternatively, COMs and other molecules may be formed in subsequent gas-phase reactions in the hot gas (e.g., Charnley et al. 1992).

In galaxies with subsolar metallicities, questions remain as to the formation efficiency of COMs (e.g., Acharyya & Herbst 2015; Shimonishi et al. 2016a). Apart from the lower elemental abundances of gaseous C, O, and N atoms, low metallicity leads to less shielding, greater penetration of UV photons into dense gas, and consequently warmer dust grains (e.g., van Loon et al. 2010; Oliveira et al. 2011). All of these factors inhibit the formation and survival of COMs.

The Large Magellanic Cloud (LMC) is a nearby dwarf galaxy (50.0 ± 1.1 kpc; Pietrzyński et al. 2013) that presents the opportunity to study COM chemistry in a low-metallicity environment (${Z}_{\mathrm{LMC}}\sim 0.3\mbox{--}0.5\,{Z}_{\odot }$; Westerlund 1997) with higher UV fluxes and lower cosmic-ray ionization rates than the Milky Way (e.g., Abdo et al. 2010). Shimonishi et al. (2016b) have claimed a hot core detection toward the LMC young stellar object (YSO) ST11 based on the derived physical conditions and the presence of simple molecules connected to hot gas chemistry (e.g., SO₂ and its isotopologues). Neither methanol or COMs were detected.

Nishimura et al. (2016a) observed seven molecular clouds in the LMC at low spatial resolution (∼8 pc) and also failed to detect methanol or any COMs. Several transitions of methanol have previously been reported in two star-forming regions in the LMC (N 159W and N 113; Heikkilä et al. 1999; Wang et al. 2009), both included in the Nishimura et al. (2016a) sample, and a few CH₃OH masers are reported in the literature (e.g., Sinclair et al. 1992; Green et al. 2008). The existing studies indicate that the abundance of CH₃OH in the LMC is very low, with potentially important general implications for COM formation in low-metallicity galaxies and at high redshift.

The sources observed by Nishimura et al. (2016a) contain large abundances of CCH, a good photodissociation region (PDR) tracer, with the highest CCH column density found in N 113. This suggests that the formation of COMs in the LMC and other low-metallicity galaxies (e.g., IC10; Nishimura et al. 2016b) could be inhibited by the presence of extensive PDRs (Hollenbach & Tielens 1999). Apart from destruction by photolysis, warm dust (≳20 K) leads to inefficient H atom sticking and CO hydrogenation to methanol on grain surfaces (e.g., Watanabe & Kouchi 2008).

In this Letter, we report the first extragalactic detection of the COMs dimethyl ether (CH₃OCH₃) and methyl formate (CH₃OCHO) in the N 113 region of the LMC. We present the Atacama Large Millimeter/submillimeter Array (ALMA) molecular emission maps and measured abundances of the COMs. The relationship to Galactic hot cores and the implications for extragalactic organic chemistry are briefly discussed.

LHA 120–N 113 (N 113; Henize 1956) is one of the most prominent star formation regions in the LMC. It contains one of the most massive (∼10⁵ ${M}_{\odot }$) and richest giant molecular clouds (GMCs) in the LMC. Its peak CO (1–0) brightness temperature of ∼8.1 K is the highest in the MAGMA survey with the Mopra telescope (half-power beam width, HPBW ∼ 45''; Wong et al. 2011).

The dense molecular gas in the N 113 GMC is clumpy with substructures that are directly revealed by the high volume density tracers HCO⁺ and HCN (e.g., Seale et al. 2012). N 113 hosts the largest number of H₂O and OH masers and the brightest H₂O maser in the entire LMC (e.g., Whiteoak & Gardner 1986; Green et al. 2008; Ellingsen et al. 2010), all located within 25'' (∼6 pc) from the GMC's CO (1–0) peak.

Current star formation activity in N 113 appears concentrated in the central part of the GMC as indicated by point-like mid-infrared emission, maser sources, and compact H ii regions, superposed on an extended emission component and aligned in a northwest–southeast direction (see Figure 1). The gas and dust in this central region are compressed by a complex structure of ionized gas bubbles (prominent in the Hα images) created by massive stars in several young clusters (Oliveira et al. 2006). Based on the Spitzer Space Telescope and the Herschel Space Observatory data, confirmed and candidate Stage 0–II YSOs down to $\sim 3\,{M}_{\odot }$ were identified (e.g., Sewiło et al. 2010; Carlson et al. 2012).

Zoom In Zoom Out Reset image size

At Spitzer and Herschel wavelengths (3.6–500 μm with 2''–38'' resolution), N 113 is dominated by emission from three massive (30–40 ${M}_{\odot }$; Ward et al. 2016) YSOs associated with high-density molecular clumps, radio continuum emission, and masers: YSO–1 (051317.69–692225.0; Gruendl & Chu 2009), YSO–3 (051325.09–692245.1), and YSO–4 (051321.43–692241.5); see Figure 1. Our Herschel spectroscopic data (e.g., [O i], [O iii], [C ii]) reveal distinct physical conditions around each of these nearby sources (J. M. Oliveira et al., in preparation). K-band observations obtained with VLT/SINFONI with $0\buildrel{\prime\prime}\over{.} 1$ resolution and a field of view of ∼3'' resolved YSO–3 and YSO–4 into a cluster of multiple components and revealed diverse characteristics of resolved sources (Ward et al. 2016). N 113 is a complex environment where we find clumps (∼1 pc) and cores (∼0.1–0.2 pc) in a range of environments (e.g., strongly influenced by the stellar winds in the west and more quiescent in the east) and at a range of evolutionary stages.

N 113 was observed with ALMA in Band 6 as part of project 2015.1.01388.S. The spectral setup observed ¹²CO (2–1), ¹³CO (2–1), and C¹⁸O (2–1) in narrow windows and additional 2 GHz windows centered on 231.7 and 216.9 GHz.

The data were obtained with the 12 m array on 2016 March 10 and June 16, for a total of 13.1 minutes per mosaic pointing over most of the map, on baselines from 15 to 704 m. It was also observed 19 times with the 7 m Atacama Compact Array (ACA) array between 2015 November 2 and December 17, for a total of 247 minutes per mosaic pointing. The data were calibrated with versions 4.7 through 5.0 of the ALMA pipeline in Common Astronomy Software Applications (CASA; McMullin et al. 2007). Amplitude was calibrated using Uranus, J0538–4405, J0519–4546, and Callisto (different calibrators on different dates). The bandpass was calibrated using J0538–4405, J0522–3627, J0854+2006, J0006–0623, and time-varying gain using J0529–7245 (12 m array) and J0440–6952 (7 m array).

Sensitivity of 0.1 mJy per 087 × 054 (∼0.21 × 0.13 pc) beam was achieved in the continuum. Continuum was subtracted in the uv domain from each line spectral window, and the 7 m and 12 m data were simultaneously imaged and deconvolved interactively. Sensitivity of 3.8 mJy per 098 × 062 (∼0.24 × 0.15 pc) beam was achieved in the 216.9 GHz cube with 1.09 MHz (1.5 km s⁻¹) channels, and 2.9 mJy per 088 × 058 (∼0.21 × 0.14 pc) beam in 2.0 MHz (2.6 km s⁻¹) channels at 231.7 GHz.

The distribution of our project's main targeted bright lines over the entire N 113 region will be discussed in a future publication, as the scope of this Letter is the complex organic molecules located in two small regions in the center of N 113 (Regions A and B in Figure 1).

4.1. 1.3 mm Continuum Emission

We have identified multiple 1.3 mm continuum sources in both Regions A and B (see Figure 2). We have assigned identification numbers to all of the continuum peaks that are associated with the molecular or H30α line emission in the order from the brightest to faintest. There is a number of the continuum peaks that are less significant and/or not associated with the molecular or H30α line emission that will be discussed in a follow-up paper. COMs have been detected toward sources A1 and B3. Both sources are associated with H₂O masers and A1 also with an OH maser (within the positional uncertainties). Only one SINFONI K-band source (N113–YSO01) has been detected at 1.3 mm (B2).

Zoom In Zoom Out Reset image size

There is some evidence for a small difference in the evolutionary stage between A1 and B3. B3 is brighter than A1 in DCN, which indicates that it is younger. The DCN/HCN ratio will drop rapidly with time in hot gas, irrespective of the origin of DCN in the gas phase or on ice. The abundance of all deuterated species decreases with time (Fontani et al. 2011); in hot cores, they are fossils of the earlier, colder evolutionary phases. The more mature nature of A1 is supported by an association with the OH maser.

We measured the continuum surface brightness for A1 and B3 in the continuum image centered at 224.6 GHz by fitting the sources with Gaussians, as well as measuring the value of the brightest pixel. The weighted average of the fitted and manually measured peaks (0.58 ± 0.09 K and 0.29 ± 0.02 K for A1 and B3, respectively) was used to calculate the dust column density. The dust optical depth was calculated by assuming that the dust temperature equals the fitted CH₃OH kinetic temperature (see Section 4.2). The largest source of uncertainty is the choice of dust emissivity used to calculate column density from optical depth. We use values calculated specifically for the LMC by Galliano et al. (2011), albeit on somewhat larger scales (≥10 pc): $N({{\rm{H}}}_{2})/{\tau }_{\mathrm{dust}}=1.8\,\times {10}^{26}$ cm⁻². If dust has coagulated on the small scales probed by ALMA, the dust emissivity could be higher, hydrogen column densities lower, and organic molecule abundances higher than we quote here. We determined $N({{\rm{H}}}_{2})$ of $(8.0\,\pm 1.2)\times {10}^{23}$ cm⁻² and $(7.0\pm 0.9)\times {10}^{23}$ cm⁻² for A1 and B3, respectively. The source diameters are about the beam size; thus, under the assumption of spherical symmetry, these column densities correspond to the number densities of ∼1.6 × 10⁶ cm⁻³ and ∼1.4 × 10⁶ cm⁻³, respectively.

4.2. Spectral Line Analysis

The spectra for A1 and B3 are shown in Figure 3. We detected COMs (CH₃OH, CH₃OCHO, and CH₃OCH₃), sulfur-bearing molecules (SO₂, H₂S, SiO, OCS, ¹³CS, and ³³SO), c-C₃H₂, and DCN (see Table 1). We have carried out the line identification with the madcubaij software (Martín et al. 2011; Rivilla et al. 2016), which uses the JPL¹⁴ and CDMS¹⁵ molecular line databases. This software provides theoretical synthetic spectra of the different molecules under local thermodynamic equilibrium (LTE) conditions, taking into account the individual opacity of each line. To identify a molecule, the detectable lines in the observed spectra predicted by the LTE analysis must be present, with the relative intensities of the different transitions consistent with the LTE analysis.

Zoom In Zoom Out Reset image size

Table 1.  Molecular Line Transitions Detected toward A1 and B3

Species	Transition	Frequency	${E}_{{\rm{u}}}$ a	Detectionb		${I}_{\mathrm{int}}$ (K km s−1)c		Column Densityd,e (cm−2)
		(GHz)	(K)	A1	B3	A1	B3	A1	B3
CH3OH vt = 0f	${5}_{\mathrm{1,4}}\mbox{--}{4}_{\mathrm{2,2}}$	216.94560	55.87102	+	+	13.8(0.5)	3.4(0.5)
CH3OH vt = 1	${6}_{\mathrm{1,5}}\mbox{--}{7}_{\mathrm{2,6}}$	217.299205	373.92517	+	+	8.2(0.2)	2.2(0.2)
CH3OH vt = 0	${20}_{\mathrm{1,19}}\mbox{--}{20}_{\mathrm{0,20}}$	217.88639	508.37582	+	−	4.5(0.3)	⋯	1.6(0.1) × 1016	6.4(0.8) × 1015
CH3OH vt = 0	${10}_{\mathrm{2,9}}\mbox{--}{9}_{\mathrm{3,6}}$	231.28110	165.34719	+	+	7.3(0.4)	1.8(0.4)
CH3OH vt = 0	${10}_{\mathrm{2,8}}\mbox{--}{9}_{\mathrm{3,7}}$	232.41859	165.40178	+	+	8.1(0.4)	3.7(0.4)
CH3OH vt = 0	${18}_{\mathrm{3,16}}\mbox{--}{17}_{\mathrm{4,13}}$	232.78350	446.53167	+	+	2.5(0.2)	2.7(0.2)
CH3OCHO v = 0	${18}_{\mathrm{2,16}}\mbox{--}{17}_{\mathrm{2,15}}$ E	216.83020	105.67781	+?	−	1.5(0.2)	⋯
CH3OCHO v = 0	${18}_{\mathrm{2,16}}\mbox{--}{17}_{\mathrm{2,15}}$ A	216.83889	105.66730	+?	−	2.2(0.2)	⋯
CH3OCHO v = 0	${20}_{\mathrm{1,20}}\mbox{--}{19}_{\mathrm{1,19}}$ E	216.96476	111.49762	+	+?			1.1(0.2) × 1015	<3.4 × 1014
CH3OCHO v = 0	${20}_{\mathrm{1,20}}\mbox{--}{19}_{\mathrm{1,19}}$ A	216.96590	111.48055	+	+?	6.2(0.3)	2.4(0.1)
CH3OCHO v = 0	${20}_{\mathrm{0,20}}\mbox{--}{19}_{\mathrm{0,19}}$ E	216.96625	111.49826	+	+?
CH3OCHO v = 0	${20}_{\mathrm{0,20}}\mbox{--}{19}_{\mathrm{0,19}}$ A	216.96742	111.48048	+	+?
CH3OCH3	${13}_{\mathrm{0,13}}\mbox{--}{12}_{\mathrm{1,12}}$ EE	231.98782	80.92308	+	+	6.0(0.6)	3.9(1.0)	1.8(0.5) × 1015	1.2(0.4) × 1015
c-C3H2 v = 0	${3}_{\mathrm{3,0}}\mbox{--}{2}_{\mathrm{2,1}}$	216.27876	19.47	+	+	0.5(0.1)	0.9(0.1)
c-C3H2 v = 0	${6}_{\mathrm{0,6}}\mbox{--}{5}_{\mathrm{1,5}}$	217.82215	38.61	+	+	2.0(0.3)	4.8(0.4)	3.7(0.7) × 1013	5.6(1.0) × 1013
c-C3H2 v = 0	${6}_{\mathrm{1,6}}\mbox{--}{5}_{\mathrm{0,5}}$	217.82215	38.61	+	+
c-C3H2 v = 0	${5}_{\mathrm{1,4}}\mbox{--}{4}_{\mathrm{2,3}}$	217.94005	35.42	+?	+	1.3(0.2)	2.9(0.3)
33SO	65–54, F = 9/2–7/2	217.82718	34.67135	+	+?
33SO	65–54, F = 11/2–9/2	217.82983	34.67292	+	+?	9.2(0.2)	2.9(0.3)	1.4(0.2) × 1014	⋯g
33SO	65–54, F = 13/2–11/2	217.83177	34.67502	+	+?
33SO	65–54, F = 15/2–13/2	217.83264	34.67766	+	+?
SO2 v = 0	${22}_{\mathrm{2,20}}\mbox{--}{22}_{\mathrm{1,21}}$	216.64330	248.44117	+	+	18.7(1.4)	3.6(0.6)	3.9(0.4) × 1015	9.4(1.7) × 1014
H2S	${2}_{\mathrm{2,0}}\mbox{--}{2}_{\mathrm{1,1}}$	216.71044	83.98035	+	+	11.9(0.8)	7.5(0.3)	1.2(0.1) × 1015	5.6(0.4) × 1014
SiO	5–4	217.10498	31.25889	+	+	29.7(1.1)	30.8(1.7)	6.9(0.3) × 1013	7.0(0.5) × 1013
DCN v = 0	3–2	217.238	20.85164	+	+	3.4(0.2)	7.6(0.2)	1.0(0.1) × 1013	1.8(0.1) × 1013
OCS v = 0	19–18	231.06098	110.89923	+	+	12.4(1.0)	4.5(0.3)	8.4(0.9) × 1014	4.1(0.3) × 1014
13CS	5–4	231.2210	33.29138	+	+	12.8(0.5)	7.9(0.6)	6.4(0.3) × 1013	4.1(0.4) × 1013

Notes.

^aE_u is the upper level energy of the transition. ^bThe symbols in these columns indicate a detection ("+"), a tentative detection ("+?"), or a non-detection ("−") of a given molecular line transition. ^c ${I}_{\mathrm{int}}$ is the integrated line intensity. ^dThe CH₃OH column densities and temperatures were derived based on modeling of multiple CH₃OH transitions (see the text). Column densities of other molecules were estimated assuming the same temperature as for methanol. ^eFor molecules with multiple transitions, all the transitions were used to estimate a total column density. ^fA typical line width for the strongest methanol line is ∼5 km s⁻¹ for A1 and ∼8 km s⁻¹ for B3. ^g33SO lines for B3 are too faint to get a good fit.

Download table as:  ASCIITypeset image

We used the results of the rotational diagram analysis (e.g., Goldsmith & Langer 1999) of CH₃OH as the initial parameters for the spectral line modeling using madcubaij. We utilized six CH₃OH transitions with a range of upper energy levels (${E}_{{\rm{u}}}$∼ 56–508 K; see Figure 4). This analysis assumes the gas is in LTE and the lines are optically thin. We obtained the rotational temperatures and total column densities (${T}_{\mathrm{rot}}$, ${N}_{\mathrm{rot}}$) of (134 ± 6 K, $1.6\pm 0.1\times {10}^{16}\,{\mathrm{cm}}^{-2}$) and (131 ± 15 K, $6.4\,\pm 0.8\times {10}^{15}\,{\mathrm{cm}}^{-2}$) for A1 and B3, respectively. The synthetic spectra corresponding to these parameters are shown in Figure 3. The uncertainties for the calculated temperature and column density values are the formal fit uncertainties from the least squares fits performed by madcubaij.

Zoom In Zoom Out Reset image size

Assuming that all molecular species observed toward A1/B3 are located in the same region as methanol, we estimate their column densities using the same ${T}_{\mathrm{rot}}$ as for methanol (see Table 1). To investigate how the results would change if we assumed a different value of ${T}_{\mathrm{rot}}$, we calculated column densities for ${T}_{\mathrm{rot}}$ of 100, 150, 200, and 250 K. In general, column densities increase gradually with increasing ${T}_{\mathrm{rot}}$, and they are not larger than a factor of ∼2 than those listed in Table 1 for any molecule at ${T}_{\mathrm{rot}}=250$ K. At 100 K, column densities are in general up to ∼20% lower.

A non-detection of CH₃OH in N 113 by Nishimura et al.'s (2016a) single-dish observations can be explained by the fact that they covered Region B only. Region B harbors the fainter and smaller of the two sources with CH₃OH detection, suffering larger effects from beam dilution. The single-dish observations of Wang et al. (2009) that detected methanol in N 113 had a similar sensitivity; however, they covered both Regions A and B. The CH₃OH column density estimated by Wang et al. (2009; ∼10¹³ cm⁻²) is lower than that based on the ALMA data, likely the effect of the beam dilution. Only with ALMA can the clumpy structure of methanol be revealed.

Both A1 and B3 resemble classic hot cores–dense condensations of molecular gas surrounding the massive star formation sites. They are compact ($D\sim 0.17$ pc) and hot (${T}_{\mathrm{rot}}\sim 130\,{\rm{K}}$), as evidenced by the high excitation lines of methanol with rotation temperatures similar to Galactic hot cores at the lower end of the ${T}_{\mathrm{rot}}$ range (e.g., Kurtz et al. 2000). A1 and B3 are slightly larger than "typical" Galactic hot cores, but their number densities and column densities are consistent with those of known hot cores, they are hot enough to release ice mantles, they show emission from COMs, and they are associated with masers; all of these properties support the classification of A1 and B3 as hot cores.

Some fainter methanol emission is detected in other locations in Region B (see Figure 2); however, it is unclear whether that traces physically distinct sources or shocked lobes of the outflows.

The detection of SiO demonstrates that shock sputtering of dust and ice may be as important as thermal evaporation for initiating the chemical evolution in these hot cores (e.g., Charnley & Kaufman 2000; Viti et al. 2001). A SiH₄ (i.e., non-shock) origin for SiO is possible (Mackay 1996), but it requires that hot cores contain abundances of O₂ that are significantly above the observed upper limits obtained with Herschel (e.g., Liseau et al. 2012; Yıldız et al. 2013).

The detection of CH₃OH indicates that the dust was once cold enough for CO hydrogenation to proceed and the observed DCN is probably a frozen gas-phase remnant of this cold cloud chemistry (Brown & Millar 1989).

Using the CH₃OH and H₂ column densities derived above, we calculate the fractional abundance of CH₃OH of $(2.0\pm 0.3)\times {10}^{-8}$ and $(9.1\pm 1.7)\times {10}^{-9}$ for A1 and B3, respectively. These CH₃OH fractional abundances are over an order of magnitude larger than an upper limit estimated for the candidate hot core ST11 by Shimonishi et al. (2016b). The (CH₃OCH₃, CH₃OCHO) fractional abundances are (2.2 ± 0.7, $1.4\pm 0.4)\times {10}^{-9}$ for A1 and (1.7 ± 0.7, $\lt 0.5)\times {10}^{-9}$ for B3. The (CH₃OCH₃, CH₃OCHO) abundances with respect to CH₃OH are (0.11 ± 0.03, 0.07 ± 0.01) and (0.19 ± 0.07, <0.05) for A1 and B3, respectively.

The abundances of COMs detected in N 113 are comparable to those found at the lower end of the range in Galactic hot cores (e.g., Herbst & van Dishoeck 2009; Taquet et al. 2016), when scaled by a factor of 2.5 to account for the lower metallicity in the LMC (assuming ${Z}_{\mathrm{LMC}}=0.4\,{Z}_{\odot }$). Thus, the chemistry of COMs detected in N 113 is similar to that of the Galaxy, indicating that regions where they exist are likely shielded from UV radiation (see Cuadrado et al. 2017). Both grain reactions on warm dust and post-desorption ion–molecule chemistry could form CH₃OCH₃ and CH₃OCHO (Garrod & Herbst 2006; Taquet et al. 2016).

We have also detected several S-bearing molecules in both cores. In A1 and B3, the H₂S/CH₃OH ratios are respectively 0.075 ± 0.008 and 0.088 ± 0.013, whereas the OCS/CH₃OH ratios are 0.053 ± 0.007 and 0.064 ± 0.009. These ratios are almost indistinguishable between A1 and B3, supporting the idea that these molecules were formed in ices during the N113 prestellar phase. OCS is known to be present in interstellar ices (Boogert et al. 2015) and, although H₂S has not yet been confirmed, cold grain-surface chemistry, involving reactions of S atoms with atomic hydrogen and CO (or oxidation of CS), is likely to be the origin of these molecules in the N113 cores.

SO and SO₂ could also originate from ice chemistry (they are detected in cometary ices; Calmonte et al. 2016), but they may also form in post-shock gas that is enriched in OH through neutral–neutral reactions (Hartquist et al. 1980). Alternatively, chemical models show that all the S-bearing molecules, including SO and SO₂, could be formed in post-evaporation gas-phase reactions starting from evaporated ices containing H₂S (Charnley 1997). In this case, comparison of calculated and observed SO/SO₂, H₂S/SO₂, and H₂S/SO ratios could be used as a chemical clock for the age of the core. We will explore this issue in a future publication.

Finally, c-C₃H₂ could either be formed by ion–molecule chemistry in the cool ambient medium or, like CCH, be in a PDR region, where it may be produced in the photodestruction of larger refractory organic (PAH) molecules (Guzmán et al. 2014).

We see the effective formation of COMs under subsolar metallicity conditions in two hot cores discovered in the LMC. The COMs observed in the N 113 hot cores could either originate from grain-surface chemistry or in post-desorption gas chemistry. Of the S-bearing molecules detected, it is likely that H₂S and OCS formed on grains; SO and SO₂ also could have originated from ice chemistry or formed in gaseous neutral reactions. The presence of SiO indicates that shock chemistry may also be playing a role. Future observations of LMC cores and detailed chemical modeling will be necessary to determine the relative contribution from each process. Studying the chemistry of the interstellar medium as a function of metallicity is important to understand the chemical evolution of the universe.

We thank the anonymous referee for insightful comments and suggestions that helped us improve the Letter. The work of M.S. was supported by an appointment to the NASA Postdoctoral Program at the Goddard Space Flight Center, administered by Universities Space Research Association under contract with NASA. S.C. acknowledges the support from the NASA's Emerging Worlds Program. S.Z. was supported by NAOJ ALMA Scientific Research grant number 2016-03B. J.L.W. acknowledges support from the German Research Council (SFB 881 "The Milky Way System", subproject P1). M.M. was supported by NSF grant 1312902. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. This Letter makes use of the following ALMA data: ADS/JAO.ALMA#2015.1.01388.S. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), NSC and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ.