$^{13}$CO/C$^{18}$O Gradients Across the Disks of Nearby Spiral Galaxies

TL;DR

This study measures 13CO/C18O ratios across nine nearby spiral galaxies, revealing gradients that inform on chemical abundance variations and stellar nucleosynthesis effects within galaxy disks.

Contribution

It provides high-resolution measurements of isotopologue ratios across galaxy disks, which are rare and enhance understanding of chemical processes in galaxies.

Findings

Mean 13CO/C18O ratio of 6.0 in galaxy centers

Ratios increase with galactocentric radius

Ratios decrease with star formation rate surface density

Abstract

We use the IRAM Large Program EMPIRE and new high-resolution ALMA data to measure 13CO(1-0)/C18O(1-0) intensity ratios across nine nearby spiral galaxies. These isotopologues of CO are typically optically thin across most of the area in galaxy disks, and this ratio allows us to gauge their relative abundance due to chemistry or stellar nucleosynthesis effects. Resolved 13CO/C18O gradients across normal galaxies have been rare due to the faintness of these lines. We find a mean 13CO/C18O ratio of 6.0$\pm$0.9 for the central regions of our galaxies. This agrees well with results in the Milky Way, but differs from results for starburst galaxies (3.4$\pm$0.9) and ultraluminous infrared galaxies (1.1$\pm$0.4). In our sample, the 13CO/C18O ratio consistently increases with increasing galactocentric radius and decreases with increasing star formation rate surface density. These trends qualitatively agree with expectations for carbon and oxygen isotopic abundance variations due to stellar nucleosynthesis, with a possible effect of fractionation.

Tables

Table 1. Table 1 Adopted Properties of Observed Galaxies

Source	Distance	Incl.	P.A.	$r_{25}$	Type	${\Sigma }_{\text{SFR}}$
	(Mpc)	(∘)	(∘)	(kpc)		(ddAverage SFR surface density inside 0.75$r_{25}$, units of $\times {10}^{-3}{M}_{\odot }$ yr-1 kpc-2. From Leroy et al. (2013).)
NGC 628aaObserved with only IRAM.	9.0	7	20	13.1	SAc	4.0
NGC 2903aaObserved with only IRAM.	8.7	65	204	15.2	SABbc	5.7
NGC 3184aaObserved with only IRAM.	12.0	16	179	13.0	SABcd	2.8
NGC 3351bbObserved with only ALMA.	10.1	11	73	10.6	SBb	5.2
NGC 3627ccObserved with both IRAM and ALMA.	9.8	62	173	14.6	SABb	7.7
NGC 4254ccObserved with both IRAM and ALMA.	15.6	32	55	11.4	SAc	18
NGC 4321ccObserved with both IRAM and ALMA.	16.4	30	153	14.3	SABbc	9.0
NGC 5055aaObserved with only IRAM.	8.2	59	102	14.0	SABbc	4.1
NGC 6946aaObserved with only IRAM.	5.6	33	243	9.35	SABcd	21

Table 2. Table 2 Measured 13 CO/C 18 O line intensity ratios for our sample and literature values

Source	Total disk	Central 30”
	(EMPIRE)	(ALMA & EMPIRE)
EMPIRE &ALMA
NGC 628	2.4$\pm$0.8	$>$ 2.5
NGC 2903	$>$ 8.0	$>$ 7.0
NGC 3184	$>$ 8.2	$>$ 6.2
NGC 3351**Only observed with ALMA, measurements for NGC 3351 cover the central $\sim$1 kpc ($\sim$20”).	–	5.4$\pm$0.8
NGC 3627	9.0$\pm$1.1	$>$ 3.2
NGC 4254	8.5$\pm$0.6	6.2$\pm$0.7
NGC 4321	9.8$\pm$0.5	5.4$\pm$0.7
NGC 5055	9.9$\pm$0.8	8.7$\pm$2.5
NGC 6946	7.6$\pm$0.5	3.8$\pm$0.3
LITERATURE
Milky WayaaWouterloot et al. (2008)	8.27$\pm$0.2	7.1$\pm$0.2
Solar SystembbWilson and Rood (1994)	5.5	–
LMCccHeikkila et al. (1998)	30$\pm$5	–
NGC 5194	2.6$\pm$1.7ddTan et al. (2011)	3.6$\pm$0.3eeVila-Vilaró (2008)
NGC 6946ffMeier and Turner (2004)	–	2.3$\pm$0.2
M82ddTan et al. (2011) – Starburst	–	2.7$\pm$0.9
NGC 253ggHenkel et al. (2014) – Starburst	–	3.60$\pm$0.04
Maffei2hhMeier et al. (2008) – Starburst	–	3.8$\pm$0.8
NGC 1068iiAladro et al. (2013) – Starburst	–	3.4$\pm$0.9
Arp 220jjGreve et al. (2009) – ULIRG	–	1.0$\pm$0.4
Mrk 231ggHenkel et al. (2014) – ULIRG	–	1.3$\pm$0.4
SMM J2135kkDanielson et al. (2013) – High-$z$	–	1.0$\pm$0.3

Equations

$${}^{13}{\text{C}}^{+}{+}^{12}\text{CO}{\to }^{12}{\text{C}}^{+}{+}^{13}\text{CO}+\Delta \text{E}$$

Full text

††slugcomment: Accepted for publication in the Astrophysical Journal Letters

1. Introduction

Rotational transitions of the isotopologues of 12CO are observable as discrete lines, with the strongest being 13CO and C18O. Although these lines are fainter than 12CO, they can be useful diagnostic tools. In contrast to the low-$J$ 12CO lines, 13CO and C18O transitions often remain optically thin over large areas in galaxies. Contrasting them with the thick 12CO lines can constrain the optical depth of both lines and illuminate the underlying molecular gas column and volume densities (e.g. Young and Scoville, 1982; Pineda et al., 2008; Wilson et al., 2009). Comparing transitions of different optically thin isotopologues, e.g., 13CO/C18O, offers the chance to trace abundance variations across the disks of galaxies. Such variations may be due to chemistry and/or stellar nucleosynthesis.

The main obstacle to studying the rare CO isotopologues is the faintness of their emission lines, which is driven by their lower abundance. Typically 13CO and C18O are $\sim 50$ and $\sim 500$ times less abundant than 12CO. As a consequence, these isotopologues have been mainly studied in bright, actively star forming systems such as (ultra)luminous infrared galaxies (U/LIRGs) and starburst galaxy centers (e.g. Meier and Turner, 2004; Costagliola et al., 2011; Aladro et al., 2013). Most of our knowledge about the relative variation of the 13CO and C18O lines across the disk of a normal star-forming galaxy comes from the Milky Way (Langer and Penzias, 1990; Wilson and Rood, 1994). Better knowledge of this ratio across the disks of normal galaxies will inform our interpretation of 12CO emission, cloud chemistry, and the influence of recent nucleosynthesis.

In this Letter, we report measurements of the 13CO (1-0)-to-C18O (1-0) ratio across wide areas in the disks of nine nearby galaxies. These are targets of the IRAM large program EMPIRE (EMIR Multiline Probe of the ISM Regulating Galaxy Evolution; Bigiel et al. (2016)) and a related ALMA program (Gallagher et al., in prep., Jiménez-Donaire et al. (2016)).

2. Observations

EMPIRE (Bigiel et al., 2016) is mapping a suite of molecular lines across the whole area of nine nearby galaxy disks using the IRAM 30-m telescope. This includes 13CO (1-0) and C18O (1-0), with rest frequencies of 110.20 and 109.78 GHz. The observations began in December of 2014 and are still underway. Science-ready 13CO and C18O data now exist for the nine galaxies listed in Table 1.

Bigiel et al. (2016) describe the basic EMPIRE observation and reduction strategy. In brief, we observed each galaxy in on-the-fly mode using the 3 mm band of the dual-polarization EMIR receiver (Carter et al., 2012) and the Fourier transform spectrometers. We reduce the data using the GILDAS/CLASS software, subtracting a low order baseline and flagging data with rms noise that deviates strongly from that expected based on the radiometer equation. Finally the data for each spectral line were gridded into a cube, and then we removed low-order baselines from the gridded cube. The final cubes were convolved to a common resolution of $33\arcsec$ and corrected for the main beam efficiency. At $33\arcsec$ resolution with $4$ km s*-1* wide channels, the typical rms noise is $\sim 2.0$ mK ($T_{mb}$). Line calibrators observed as part of EMPIRE varied by $\sim 5{-}8$% from observation to observation. We expect the 13CO/C18O ratio to be even more robust because both lines were observed simultaneously and any gain variations will apply to both. Statistical errors and baseline errors induced by receiver instabilities therefore dominate the uncertainties in this paper.

We also observed four targets with ALMA, using a spectral setup that covered 13CO and C18O. Three of the four ALMA observed systems overlap with the EMPIRE sample (NGC 3627, 4254, 4321). For this letter, we use versions of these cubes convolved to 8$\arcsec$. Those that overlap EMPIRE have been corrected to account for short and zero spacing information using the CASA task feather; for more details, see Gallagher et al. (in prep.). The close frequency of the two lines ($109.8$ and $110.2$ GHz), convolution to a common resolution, and simultaneous observation of 13CO and C18O (so that they have nearly matched $u-v$ coverage) should make the internal line ratios from ALMA robust.

Both 13CO and C18O become faint at large galactocentric radii. In order to measure their intensities at large $r_{\rm gal}$, we employ spectral stacking. To do this, we divide the galaxies into zones of fixed radius or IR surface brightness. Then, we follow a procedure similar to that of Schruba et al. (2011) and Caldú-Primo et al. (2013): before averaging the spectra, we estimate the local mean velocity of gas from bright 12CO (2-1) emission from HERACLES (Leroy et al., 2009). We recenter the 13CO and C18O spectra about this local mean 12CO velocity. Then, we average all spectra in each zone to construct a single, high signal-to-noise spectrum. After stacking, baseline issues due to receiver and weather instabilities sometimes emerge. When needed, we correct for these with an additional low-order polynomial fit.

We use this spectrum to measure the integrated intensity of the line, summing over the channels that show bright 12CO(2-1) emission. We apply a similar procedure to the ALMA data, but using the brighter 13CO line to stack the C18O. Thus the ALMA measurements are restricted to the regions of bright 13CO emission. We calculated the uncertainty in the integrated intensity based on the noise estimated from the signal-free part of the stacked spectrum and the width of the integration window (usually $\sim$60 km s*-1*).

We use the total infrared (TIR) surface brightness as a proxy for the local surface density of star formation. To estimate this, we combine $\lambda=70$, 160, and 250$\mu$m maps from Herschel (KINGFISH Kennicutt et al., 2011). We convolve these to match the $33\arcsec$ beam of our line data (Aniano et al., 2011), calculate the TIR surface brightness following Galametz et al. (2013), and then convert to star formation rate surface density using the prescription of Murphy et al. (2011). NGC 2903 lacks Herschel data, therefore we use Spitzer 24$\mu$m (from LVL, Dale et al., 2009) to estimate the IR surface brightness.

3. Results

Figure 1 shows spectra of both lines averaged over the whole EMPIRE or ALMA area for each target. We observe widespread 13CO(1-0) from each galaxy while C18O(1-0) is fainter, requiring substantial averaging to achieve good signal to noise. After averaging, we detect C18O at good significance for six targets in EMPIRE. The other two, NGC 2903 and NGC 3184, show only weak emission and we place lower limits on 13CO/C18O in these systems. We recover the C18O line in all four ALMA targets.

Table 2 reports our measured 13CO/C18O ratios of line-integrated intensities on the $T_{\text{mb}}$ scale. For each target, we quote the average ratio over the whole galaxy (from EMPIRE) and the mean ratio within the inner $30\arcsec$ radius from ALMA and EMPIRE combined (except for NGC 3351, see Section 2). In cases without clear detections, we report $3\sigma$ upper limits. Treating all galaxies equally, but neglecting upper limits, we find a mean ratio of 7.9$\pm$0.8 for the whole galaxy disks. Except in NGC 628, this galaxy-averaged ratio does not vary much from galaxy to galaxy. This agrees with results by Davis (2014), who compiled observations of a large set of Seyfert, starburst, and normal star-forming nuclei and found no clear trend relating the 13CO/C18O ratio to galaxy type.

The average 13CO/C18O ratio for the inner $30\arcsec$ region of our targets is 6.0$\pm$0.9, slightly lower than the disk-averaged value. Our mean values for both the disk and the nuclear regions of our targets are consistent with early work on the Milky Way, which found a 13CO/C18O abundance ratio of 5-10 (Langer and Penzias, 1990), and more recent results from Wouterloot et al. (2008).

Our mean values for local star forming galaxies differ from those found for starburst and ULIRGs. Table 2 includes a compilation of literature measurements targeting the central regions of starburst galaxies and ULIRGs. There the 13CO/C18O ratio tends to be lower than what we find, with the lines nearly equal in strength. Our mean value for disk galaxies differs from that found for starburst galaxies by a factor of $\sim$2 (e.g. Tan et al., 2011; Henkel et al., 2014) and from that found for ULIRGs by a factor of $\sim$6 (e.g. Greve et al., 2009; Danielson et al., 2013).

Averaging over whole galaxies may obscure variations in the line ratio by blending together many different physical conditions. In Figure 2, we break apart the galaxies where we securely detect C18O at multiple radii (this removes NGC 3627 from the sample, but we show its profile using ALMA data below). We plot 13CO/C18O as a function of galactocentric radius and star formation rate surface density.

In the left panel of Figure 2, we see 13CO/C18O increase with increasing radius for NGC 4254, NGC 4321, NGC 5055 and NGC 6946. Again NGC 628 represents the outlier, with a low ratio and perhaps a dropping rather than rising profile. Wouterloot et al. (2008) showed a similar trend for 13CO/C18O to increase with increasing radius in the Milky Way (see their Figure 3a), consistent with the EMPIRE observations of normal-disk galaxies. These rising profiles imply high 13CO/C18O at the largest radius where we can still achieve a detection, typically 13CO/C18O $\sim$16 at $\sim$ 0.4$\times r_{25}$. At this radius, the ISM transitions from H2 to H I dominate (see Schruba et al., 2011). To our knowledge, such high ratios have not been reported before for normal star-forming disk galaxies.

Using ALMA, we construct more detailed profiles of the inner part of four of our targets. We show these in Figure 3. The inner, molecule-rich parts of these galaxies do not yet show the rising profiles seen at larger radii in Figure 2. However, these “zoomed in” profiles show several features not immediately visible at lower resolution. Three of our targets, NGC 3351, NGC 3627, and NGC 4321 host bright inner regions ($\sim$500 pc, $\sim 0.05\,r_{25}$) likely fed by gas flows along the bars in these galaxies. No such features are apparent in NGC 4254, which lacks a strong bar. At least in the barred galaxies we might expect less isotopic abundance variations due to efficient mixing of the gas in their inner parts, similar to what is observed in the Milky Way.

The right panel in Figure 2 shows the 13CO-to-C18O ratio as a function of the surface density of star formation, $\Sigma_{\rm SFR}$. Again, NGC 628 appears as an outlier, while the other targets with resolved gradients show decreasing 13CO/C18O with increasing $\Sigma_{\rm SFR}$. As discussed below, the sense of this trend is what is expected if massive star nucleosynthesis alters the isotopic abundances, but it might also reflect optical depth effects. In either case, as we sort our targets by $\Sigma_{\rm SFR}$, we observe a clear trend from “Milky Way” type values in low $\Sigma_{\rm SFR}$ regions towards “Starburst” type values in high $\Sigma_{\rm SFR}$ regions, though our targets do not approach the ULIRG regime. We discuss explanations for these trend in the next section.

4. Discussion

Leaving aside NGC 628, our targets tend to show rising 13CO/C18O with increasing galactocentric radius and falling 13CO/C18O with increasing $\Sigma_{\rm SFR}$. What drives these variations? We consider two scenarios that could explain our observations: (1) changes in the abundances of the observed molecules, and (2) changes in the optical depths of the observed lines.

4.1. Variations in the Molecular Abundances

The observed ratio 13CO/C18O will vary if the relative abundance of the two species varies. We highlight selective photodissociation, isotope dependent fractionation, and selective enrichment by massive star nucleosynthesis as possible explanations.

Ultraviolet radiation dissociates the molecules in molecular clouds, and preferential dissociation of one species or another could change the 13CO/C18O ratio. Selective photodissociation is possible because 13CO is more effective at self-shielding against UV photons than C18O due to its higher abundance (Bally and Langer, 1982). However, even for the far more abundant isotopolgue 12CO, Safranek-Shrader et al. (2017) find that self-shielding is only comparable in importance to shielding by dust and cross-shielding by the Lyman-Werner damping wings of H2. For the much less abundant 13CO and C18O, self-shielding must therefore be unimportant compared to dust- and H2-shielding, which are non-selective. Indeed, using DESPOTIC (Krumholz, 2014), we find that these mechanisms together are sufficient to reduce the dissociation rate for 12CO and its isotopologues to near zero for columns of $N_{\text{H}_{2}}\sim 10^{22}$ cm*-2*, even if there is no self-shielding. We can therefore discard the possibility of selective photodissociation as an explanation for our results.

Alternatively, isotope dependent fractionation may lead to different abundances. In cold regions, where ion-molecule chemistry dominates, there would be preferential formation of 13CO by chemical fractionation:

[TABLE]

If, as expected, cloud temperature scales with $\Sigma_{\rm SFR}$, and the C18O abundance is not affected by fractionation processes, then the 13CO/C18O ratio will decrease with $\Sigma_{\rm SFR}$ and will increase with the distance to the center of the galaxy. This could explain the increasing radial profiles for the sample if the gas is cold enough further out in their disks, as well as the increasing profiles seen in Figure 2 as a function of $\Sigma_{\rm SFR}$. It would not explain the decrement in the 13CO/C18O ratio of NGC 628 as a function of $\Sigma_{\rm SFR}$.

Very massive stars produce 12C at the end of their lives. The 18O is also produced by high-mass stars, but the yield depends on the amount of 12C and 16O that is available (e.g. Henkel and Mauersberger, 1993; Meier and Turner, 2004; Nomoto et al., 2013, and references within). On the other hand, 13C is only produced as a result from the CN cycle of Helium Burning in intermediate-mass stars, or as a secondary product in low and high mass ($>$10 M⊙) stars (Sage et al., 1991). Thus, one would expect recent star formation to increase the 18O and 12C abundances on a short timescale, as massive stars explode as supernovae and enrich the ISM. Meanwhile, 13C comes primarily from the red giant phase of intermediate-mass stars and requires more time to enrich the ISM. C18O could thus be expected to become overabundant relative to 13CO in regions where enrichment from young stars sets the isotopic abundances. In reality, the abundances in the molecular gas will reflect a complex combination of pre-existing enrichment, enrichment from recent star formation, and removal of gas by feedback. But the basic explanation is that gas preferentially enriched by massive stars should show low 13CO/C18O.

Abundance variations due to selective nucleosynthesis offer a plausible mechanism that explains the increasing radial profiles of our sample. However, the observed 12C/13C abundance ratio in the Milky Way shows the opposite trend (Langer and Penzias, 1990; Milam et al., 2005), implying that the 16O/18O abundance ratio decreases with galactocentric radius as well. This argues against nucleosynthesis as a general explanation. In NGC 628 where the radial profiles are flat, there could be older populations where the 13C has been created more efficiently and well mixed back into the ISM afterwards. This galaxy could also present strong fractionation effects; however more lines are needed to investigate it in detail.

4.2. Changes in the Optical Depth

Our observations could also be explained if the optical depth of 13CO changes across a galaxy. We expect that 12CO remains optically thick over most areas in our targets. Variations in 13CO optical depth could result from changing line widths or changes in excitation due to temperature or density variations (e.g., see Meier and Turner, 2004).

We gauge the importance of the optical depth effects on our 13CO measurements by comparing to 12CO measurements (Jiménez-Donaire et al. in prep., Cormier et al., in prep.). Assuming that the 12CO is optically thick and that both species have a common excitation temperature and beam filling factor (our observations show that 13CO and 12CO have similar large structure, similarly to Heyer et al. (2009)) the ratio allows us to estimate the optical depth of the 13CO transition, $\tau_{13CO}$. Applying these simple assumptions to our observations, and assuming that the CO (2-1) measured by HERACLES is approximately thermalized with the CO (1-0), we find $0.05<\tau_{\rm 13CO}<0.25$. These values imply optically thin 13CO across our sample at the factor of $\sim 2$ level, even if our simplifying assumptions break down.

In other words, based on the observed 13CO/12CO line ratios, which are substantially lower than $\sim 1$, the 13CO emission that we observe appears mostly optically thin. In order to explain our observed line ratio variations, the 13CO line would need to become optically thick over some regions. Thus, we do not expect that the high 13CO/C18O ratios that we observe originate from changes in $\tau_{13CO}$. To illustrate this, the right panel in Figure 3 shows the expected dependence of the isotopologue ratio on the optical depth of 13CO over the range of modest $\tau_{\rm 13CO}$ that we find. The effect of changing optical depth in this regime is negligible. We instead prefer one of the abundance-based explanations described above.

5. Conclusions

We report observations of the 1-0 transitions of 13CO and C18O across the disks of nine nearby galaxies from the IRAM 30m program EMPIRE and ALMA. Using a spectral stacking approach, we measured the radial variation of this ratio and its dependence on the local surface density of star formation. Because of the faintness of the C18O line, this is the first significant sample of 13CO/C18O radial profiles for nearby galaxy disks.

Averaging over whole galaxy disks, we find a mean value of 13CO/C18O ratio of $7.9\pm 0.8$ with no clear variations from galaxy to galaxy. In NGC 4254, NGC 4321, NGC 5055 and NGC 6946 we do observe systematic internal variations. Here 13CO/C18O increases by $\sim$40% with increasing radius. We also find a decreasing 13CO/C18O ratio with increasing surface density of star formation. As a result, the central regions of our targets have a mean 13CO/C18O of $6.0\pm 0.9$, somewhat lower than the whole disk averages. This central value resembles that found in the Milky Way (Wouterloot et al., 2008), but differs from values for starburst galaxies (3.4$\pm$0.9) and ULIRGs (1.1$\pm$0.4).

We argue that 13CO optical depth is unlikely to drive our observed trends. Instead, we suggest that the variations in the observed ratios reflect real changes in the abundance of the two species, driven by either selective enrichment of the ISM in active regions by massive star nucleosynthesis, or a decrease in the amount of 13CO due to suppression of 13CO in hotter, higher column density environments.

FB, MJ and DC acknowledge support from DFG grant BI 1546/1-1. AH acknowledges support from the Centre National d’Etudes Spatiales (CNES). AU acknowledges support from Spanish MINECO grants FIS2012-32096 and ESP2015-68964. The work of AKL is partially supported by the National Science Foundation under Grants No. 1615105 and 1615109. MRK acknowledges support from ARC DP160100695. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. This paper makes use of the following ALMA data: ADS/JAO.ALMA#2011.0.00004.SV. 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.