FIRST DETECTION OF HCO⁺ ABSORPTION IN THE MAGELLANIC SYSTEM

To search for absorption signatures in the observed HCO⁺ spectra, we used the Bayesian analysis technique developed, implemented, and described by Allison et al. (2012). Given a likelihood function and prior distribution, this procedure searches all parameter space using the MULTINEST Monte Carlo sampling algorithm (Feroz & Hobson 2008) and calculates the resulting posterior probability distribution and evidence statistic. For this study, we tested for the presence of a single Gaussian function in each unsmoothed HCO⁺ absorption spectrum given the following priors: (1) the line width is not less than the channel spacing ($0.2\;\mathrm{km}\;{{\rm{s}}}^{-1}$) or greater than the bandwidth ($180\;\mathrm{km}\;{{\rm{s}}}^{-1}$); (2) the spectral noise is equal to the standard deviation in off-line channels; (3) the spectral channels are independent. We considered an absorption line to be "detected" if the Bayesian evidence statistic of the Gaussian model is greater than the evidence statistic for a model containing no spectral lines.

After analyzing all spectra, we detected one absorption line in the spectrum toward J0454–810. We display the J0454–810 spectrum with the Gaussian model in the top panel of Figure 1. The detection significance is described by R = 7, where R is the natural logarithm of the Bayesian evidence statistic for the Gaussian model divided by evidence statistic for a model containing only noise. The line is located at $v\;=214.0\pm 0.4\;\mathrm{km}\;{{\rm{s}}}^{-1}$, has an FWHM of ${\rm{\Delta }}v\;=4.5\pm 1.0\mathrm{km}\;{{\rm{s}}}^{-1}$, and an optical depth of $\tau ({\mathrm{HCO}}^{+})=0.10\pm 0.02$. The integrated optical depth of the line derived from the model is, $\int \tau ({\mathrm{HCO}}^{+}){dv}=0.44\pm 0.08$, which indicates a $\sim 5\sigma$ detection significance and is consistent with the quoted R value.

Zoom In Zoom Out Reset image size

As a sanity check, we note that the detected HCO⁺ absorption line is present in the J0454–810 spectrum for: (1) each of the four individual days of observations, (2) both 0.2 and $0.8\;\mathrm{km}\;{{\rm{s}}}^{-1}$ velocity resolutions, (3) data reduced using different subsets of the available ATCA antennas, and (4) both linear polarizations. We are therefore confident that the feature is not spurious.

Assuming excitation in equilibrium with the cosmic microwave background (Lucas & Liszt 1996), we calculated the HCO⁺ column density, N(HCO⁺), from the J = 1–0 optical depth ($\tau ({\mathrm{HCO}}^{+})$) for an HCO⁺ dipole moment of 3.92 Debye (Mount et al. 2012), so that,

$$\begin{eqnarray}&&N({\mathrm{HCO}}^{+})=1.107\times {10}^{12}\;{\mathrm{cm}}^{-2}\int \tau ({\mathrm{HCO}}^{+})(v){dv},\end{eqnarray} \tag{ 1 }$$

where the line integral is expressed in $\mathrm{km}\;{{\rm{s}}}^{-1}$.

For the J0454–810 detection, we integrated the Gaussian model to compute N(HCO⁺) using Equation (1). For the non-detection sight lines, we computed a limit to N(HCO⁺) using a $3\sigma$ upper limit to $\tau ({\mathrm{HCO}}^{+})$ (with ${\sigma }_{\tau ({\mathrm{HCO}}^{+})}$ in Table 1) and assumed a line width of $4.5\;\mathrm{km}\;{{\rm{s}}}^{-1}$ based on the J0454–810 detection.

From MW studies, there is a correlation between N(HCO⁺) and N(H₂), given by X(HCO⁺) = N(HCO⁺)/N(H₂) ∼ $2\times {10}^{-9}$ (e.g., Liszt et al. 2010). Although the correlation may be different in the lower-metallicity Magellanic System, we applied this factor to N(HCO⁺) to estimate the Magellanic H₂ column density, N(H₂)${}_{\mathrm{Mag}}$.

In addition, we extracted H i emission spectra from the stray radiation-corrected Galactic All Sky Survey (GASS; McClure-Griffiths et al. 2009; Kalberla et al. 2010) for each line of sight. From these data, we computed N(H i)${}_{\mathrm{Mag}}$, assuming that the H i is optically thin:

$$\begin{eqnarray}&&N{({\rm{H}}\;{\rm{I}})}_{\mathrm{Mag}}={C}_{0}\;{\displaystyle \int }_{180}^{360}{T}_{{\rm{B}}}(v){dv},\end{eqnarray} \tag{ 2 }$$

where ${C}_{0}=1.823\times {10}^{18}\;{\mathrm{cm}}^{-2}\;{{\rm{K}}}^{-1}{(\mathrm{km}\;{{\rm{s}}}^{-1})}^{-1}$ and ${T}_{{\rm{B}}}(v)$ (K) is the H i brightness temperature integrated between radial velocities: $180\leqslant v\leqslant 360\;\mathrm{km}\;{{\rm{s}}}^{-1}$. All column densities described above are included in Table 2.

Table 2.  Column Densities and Limits

Source	N(H i)${}_{\mathrm{Mag}}$	N(HCO+)	N(H2)${}_{\mathrm{Mag}}$
	(1020 ${\mathrm{cm}}^{-2}$)	(1012 ${\mathrm{cm}}^{-2}$)a	(1020 ${\mathrm{cm}}^{-2}$)b
J0056–572	0.02 ± 0.01	$\leqslant 0.88$	$\leqslant 4.4$
J0102–7546	0.21 ± 0.03	$\leqslant 1.34$	$\leqslant 6.7$
J0208–512	0.05 ± 0.02	$\leqslant 0.76$	$\leqslant 3.8$
J0311–7651	0.92 ± 0.02	$\leqslant 0.81$	$\leqslant 4.1$
J0440–6952	1.00 ± 0.02	$\leqslant 0.68$	$\leqslant 3.4$
J0454–810	0.42 ± 0.03	0.49 ± 0.09	2.45 ± 0.45
J0506–6109	0.01 ± 0.02	$\leqslant 1.34$	$\leqslant 6.7$
J0530–727	7.17 ± 0.03	$\leqslant 0.28$	$\leqslant 1.4$
J0637–752	0.13 ± 0.03	$\leqslant 0.15$	$\leqslant 0.8$

Notes.

^aAssuming $4.5\;\mathrm{km}\;{{\rm{s}}}^{-1}$ line width. ^bBased on N(HCO⁺) and X(HCO⁺)$\;=\;2\times {10}^{-9}$.

Download table as:  ASCIITypeset image

Using available Planck maps (version 1.20) and H i emission from GASS, we can estimate limits to the Magellanic dust reddening, H₂ column density, and CO brightness along each line of sight.

To correct for the Galactic contribution, we computed the total Galactic H i column density, $N{({\rm{H}}\;{\rm{I}})}_{\mathrm{MW}}$, by integrating the GASS ${T}_{{\rm{B}}}(v)$ profile in Equation (2) for all $v\leqslant 180\;\mathrm{km}\;{{\rm{s}}}^{-1}$. We then converted this column density to an estimate of the reddening due to the MW, $E{(B-V)}_{\mathrm{MW}}$, using the results from Liszt (2014): $E{(B-V)}_{\mathrm{MW}}=N{({\rm{H}}\;{\rm{I}})}_{\mathrm{MW}}/8.3\times {10}^{21}\;{\mathrm{cm}}^{-2}\;{\mathrm{mag}}^{-1}$, which holds for $E(B-V)\;\lt 0.3\;\mathrm{mag}$. To find the total reddening along the line of sight, $E{(B-V)}_{\mathrm{tot}}$, we used the Planck model of dust radiance, or total dust emission integrated over frequency, ${\mathcal{R}}$. The Planck collaboration showed that ${\mathcal{R}}$ is a better tracer of the total column density than $353\;\mathrm{GHz}$ dust optical depth in diffuse, high-latitude regions, and exhibits the strong correlation: $E(B-V)/{\mathcal{R}}=(5.40\pm 0.09)\times {10}^{5}$ (Planck Collaboration et al. 2014). We used this correlation to compute $E{(B-V)}_{\mathrm{tot}}$, and then subtracted $E{(B-V)}_{\mathrm{MW}}$ from $E{(B-V)}_{\mathrm{tot}}$ to estimate the total Magellanic contribution, $E{(B-V)}_{\mathrm{Mag}}$. For all lines of sight, we find $E{(B-V)}_{\mathrm{Mag}}\;\sim 0.001-0.01\;\mathrm{mag}$.

We note that the above calculations require the assumption that the dust-to-gas ratio is constant across the sampled volume. Using results from FUSE observations of the SMC and LMC, we converted $E{(B-V)}_{\mathrm{Mag}}$ to another estimate of $N{({{\rm{H}}}_{2})}_{\mathrm{Mag}}$ given $N({{\rm{H}}}_{2})/{E}_{B-V}=6\times {10}^{20}\;{\mathrm{cm}}^{-2}\;{\mathrm{mag}}^{-1}$ (Welty et al. 2012). For all lines of sight, we find $N{({{\rm{H}}}_{2})}_{\mathrm{Mag}}\sim {10}^{17}-{10}^{18}\;{\mathrm{cm}}^{-2}$. Finally, we estimated the Magellanic CO brightness, ${W}_{\mathrm{CO},\mathrm{Mag}}=N{({{\rm{H}}}_{2})}_{\mathrm{Mag}}/{X}_{\mathrm{CO},\mathrm{Mag}}$, by assuming (from observations of the SMC): ${X}_{\mathrm{CO},\mathrm{Mag}}=4\times {10}^{21}$ $\;{\mathrm{cm}}^{-2}{({\rm{K}}\;\mathrm{km}\;{{\rm{s}}}^{-1})}^{-1}$ (e.g., Bolatto et al. 2013). For all lines of sight, we find ${W}_{\mathrm{CO},\mathrm{Mag}}\sim {10}^{-3}-{10}^{-4}\;{\rm{K}}\;\mathrm{km}\;{{\rm{s}}}^{-1}$.

The HCO⁺ absorption detection toward J0454–810 is the first of its kind in the Magellanic System, and suggests the presence of molecular gas far outside of the SMC and LMC. Of the eight non-detections, only two were more sensitive and six were up to three times less sensitive due to compromised weather conditions. Overall, our ATCA observations are more sensitive than previous searches for HCO⁺ absorption in diffuse, high-velocity clouds. These previous studies had typical $3\sigma$ limits to an HCO⁺ optical depth of $\sim 0.1-0.9$ per $0.6\;\mathrm{km}\;{{\rm{s}}}^{-1}$ channels and returned no detections (Akeson & Blitz 1999) and one tentative and non-confirmed detection (Combes & Charmandaris 2000) in clouds with N(H i)$\lt {10}^{20}\;{\mathrm{cm}}^{-2}$.

Our detected absorption line is wider (${\rm{\Delta }}v\;=4.5\pm 1.0\;\mathrm{km}\;{{\rm{s}}}^{-1}$) than individual HCO⁺ absorption lines detected in the MW (${\rm{\Delta }}v\sim 1-2\;\mathrm{km}\;{{\rm{s}}}^{-1}$; Lucas & Liszt 1996), and the only previously reported high-velocity cloud detection (${\rm{\Delta }}v=1.1\;\mathrm{km}\;{{\rm{s}}}^{-1}$; Combes & Charmandaris 2000). Although our measured Magellanic column density, $N({\mathrm{HCO}}^{+})=(0.49\pm 0.09)\times {10}^{12}\;{\mathrm{cm}}^{-2}$, is slightly smaller than the typical column densities observed in the MW ($\sim 1-10\times {10}^{12}\;{\mathrm{cm}}^{-2}$; Lucas & Liszt 1996), it is similar to those observed in diffuse regions. For example, for the only sight line with N(H i) $\sim \;{10}^{20}\;{\mathrm{cm}}^{-2}$, Liszt et al. (2010) measure $N({\mathrm{HCO}}^{+})=(0.25\pm 0.08)\times {10}^{12}\;{\mathrm{cm}}^{-2}$.

In Figure 2, we overlay the observed HCO⁺ absorption coordinates on an N(H i) image of the Magellanic System from the Parkes radio telescope (Putman et al. 2003). The detection is located at the leading edge of the Magellanic Bridge. Nearby regions contain other evidence for cold atomic and molecular gas, including H i detected in absorption with temperatures $\sim 20-70\;{\rm{K}}$ (Kobulnicky & Dickey 1999; Matthews et al. 2009), and CO emission close to the SMC (Muller et al. 2003; Mizuno et al. 2006). However, the detected HCO⁺ absorption probes molecular gas farther away from the LMC and SMC (about 10 kpc from the SMC's center) than any previous radio or millimeter-wave studies.

Zoom In Zoom Out Reset image size

Our estimates of N(H₂)${}_{\mathrm{Mag}}$ in Table 2 ($\sim {10}^{20}\;{\mathrm{cm}}^{-2}$) are orders of magnitude higher than both the Planck based estimates ($\sim {10}^{17-18}\;{\mathrm{cm}}^{-2}$) and those from FUSE UV-absorption in the Stream and Bridge. Toward Fairall 9, located at the beginning of the Stream in a region of enhanced metallicity likely stripped from the LMC, Richter et al. (2013b) measured $\mathrm{log}(N{({{\rm{H}}}_{2})}_{\mathrm{Mag}})={17.93}_{-0.16}^{+0.19}\;{\mathrm{cm}}^{-2}$. Toward an early-type star in the Bridge, Lehner (2002) measured $\mathrm{log}(N{({{\rm{H}}}_{2})}_{\mathrm{Mag}})\approx 15.4\;{\mathrm{cm}}^{-2}$. The discrepancies between these values and our Table 2 estimates suggest that the that the MW-based conversion factor $X({\mathrm{HCO}}^{+})$ may not be applicable in the lower-metallicity Magellanic regime. Given that the $X(\mathrm{CO})$ factor is expected to be up to a factor of 100 higher in low-metallicity regimes (e.g., Bolatto et al. 2013), it is likely that $X({\mathrm{HCO}}^{+})$ will be similarly different in the Magellanic system. Additional measurements of HCO⁺ and H₂ column densities are required to constrain this value further.

The origin and survival of the detected HCO⁺ in the highly diffuse ($N({\rm{H}}\;{\rm{I}})=(0.42\pm 0.03)\times {10}^{20}\;{\mathrm{cm}}^{-2}$) region of the Bridge is puzzling. The material either formed within the Bridge or it originated from a stripping interaction with the LMC or, more likely given its proximity, the SMC. The age of the Bridge is constrained to 100–300 Myr by the oldest detected stellar populations known to have formed in situ (e.g., Harris 2007). Within the Bridge, low metallicities (∼0.05 Solar; Lehner et al. 2008) indicate reduced dust grain formation rates, and young stellar populations (e.g., Demers & Battinelli 1998) indicate potentially strong UV radiation fields (10–100 × MW value). These two properties, as discussed by Lehner (2002), imply that molecule formation and destruction equilibrium cannot be reached within the Bridge's lifetime for reasonable hydrogen volume densities (e.g., ${n}_{{\rm{H}}}\leqslant 100\;{\mathrm{cm}}^{-3}$; Lehner 2002). Therefore, extremely high densities (${n}_{{\rm{H}}}\gg 100\;{\mathrm{cm}}^{-3}$) are likely required in the Bridge for equilibrium molecule formation to occur. This was also discussed by Kobulnicky & Dickey (1999) in the context of cold atomic gas formation in the Bridge.

However, non-equilibrium molecule formation may be responsible. Simulations show that ram-pressure effects during the SMC–LMC interaction which created the Bridge (required to explain the Bridge's observed in situ star formation; e.g., Connors et al. 2006; Besla et al. 2012) can produce high-density, star formation-ready peaks within the diffuse medium (e.g., Mastropietro et al. 2009). Further simulations of ram pressure-stripping of the ISM from galaxies have shown that stars can form within diffuse, gaseous tails when ablated gas cools and condenses in the turbulent wake of the stripping interaction (e.g., Tonnesen & Bryan 2012). In the bottom panel of Figure 1, we display the H i brightness temperature spectrum from GASS (McClure-Griffiths et al. 2009). There is an offset between the two peaks in H i and the location of HCO⁺ absorption, which may be indicative of a shock driven by dynamical interactions and support for molecule formation in the post-shock material.

In terms of molecule survival in this environment, dust is very important, as it provides shielding against the UV radiation field and acts as a catalyst for grain formation. Our Planck-based estimates indicate very little dust and CO along the lines of sight probed by this study. A lack of shielding by dust or sufficiently large N(H i) means that the molecular material identified by the HCO⁺ detection must disperse quickly and we may be observing the material in the process of being disrupted.

Alternatively, the detection presented here may not be tracing dense molecular gas. Comprehensive studies using the Plateau de Bure Interferometer (e.g., Lucas & Liszt 1996; Liszt et al. 2010) have shown that strong absorption by HCO⁺, ¹²CO, CN, HCN, HNC, and C₂H can be found in surprisingly diffuse MW regions (${A}_{V}\lt 1\;\mathrm{mag}$) where fully developed molecular chemistry is not expected to exist. Recent Herschel Space Observatory observations have detected similar molecular species to HCO⁺, including OH⁺, in extremely diffuse gas where the fraction of H₂ is $\lt 5\%$ (Indriolo et al. 2015). These studies indicate that many diatomic and polyatomic molecules can form and reach significant abundances even when they are not well-shielded.

In support of this scenario, there is a growing supply of evidence for the existence of molecules and star formation in H i-diffuse, dust-poor structures. For example, low-N(H i) intermediate-velocity clouds in the MW halo exhibit ubiquitous H₂ absorption (Richter et al. 2003b, 2003a). Widespread CO emission has been detected in the ram pressure-stripped tail of the Norma cluster galaxy ESO 137–001, $40\;\mathrm{kpc}$ away from the disk (Jáchym et al. 2014). In addition, young stellar associations in low-N(H i) regions ($\sim {10}^{20}\;{\mathrm{cm}}^{-2}$) with high H i velocity dispersions ($\sim 30\;\mathrm{km}\;{{\rm{s}}}^{-1}$) have been detected in stripped tails up to $30\;\mathrm{kpc}$ away from NGC 1533 (Werk et al. 2008).

Overall, additional observations of molecular gas tracers at high sensitivity and in a wide range of galactic environments are needed to understand the formation and evolution of molecular gas at these extremes. HCO⁺ absorption is a promising tracer of this material, and future observations with ALMA will be able to confirm the detection presented here and extend the sample of available sources to probe the full extent of the Magellanic System and the outskirts of galaxies at higher redshifts.