A survey of molecular clouds in the Galactic center's outflow

TL;DR

This study provides the first systematic census of molecular gas in the Milky Way’s nuclear wind using CO(2-1) observations with APEX across 19 high-velocity HI clouds. A total of 207 CO clumps within 16 HI clouds are identified, with radii 1–3 pc and σ_v up to 6 km s⁻¹, indicating turbulence-dominated, pressure-supported structures that are mostly gravitationally unbound (α_vir ≈ 10–100). Radiative-transfer modelling with DESPOTIC yields X_CO(2-1) in the range $6$–$20 imes 10^{20}$ cm⁻² (K km s⁻¹)⁻¹, implying molecular masses that are substantial but highly uncertain due to non-equilibrium conditions in the wind. The clumps reside on the standard M_mol–R relation like disk clouds but exhibit elevated linewidths and turbulent pressures; molecular mass fractions are typically 0.6–0.9 with no clear latitude trend, and CO is absent above ~1 kpc, suggesting rapid molecular dissociation in the wind. The results support a scenario where a hot wind entrains disk gas, progressively dissociating CO-at-risk molecular gas via stripping and radiation, offering key constraints for simulations of cloud survival, confinement, and dissociation in galactic winds.

Abstract

The nucleus of the Milky Way is known to drive a large-scale, multiphase galactic outflow, with gas phases ranging from the hot highly-ionized to the cold molecular component. In this work, we present the first systematic search for molecules in the Milky Way wind. We use the Atacama Pathfinder EXperiment (APEX) to observe the 12CO(2-1) emission line in 19 fields centered on previously known high-velocity atomic hydrogen (HI) clouds associated with the outflow. Over 200 CO clumps are detected within 16 different HI clouds. These clumps have typical radii of 1 - 3 parsec, high velocity dispersions of 1 - 6 km/s and molecular gas masses ranging from a few to several hundred solar masses. Molecular clumps in the wind sit on the low-mass end of the mass - size relation of regular molecular clouds, but are far displaced from the mass (or size) - linewidth relation, being generally more turbulent and showing high internal pressures. Nearly 90% of the clumps are gravitationally unbound with virial parameters >> 10 - 100, indicating that these structures are either being disrupted or they must be confined by external pressure from the surrounding hot medium. While the observed properties of CO clumps do not seem to evolve clearly with latitude, we find that molecular gas is not detected in any of the 6 HI clouds with projected distances over 1 kpc from the Galactic Center, suggesting the existence of a maximum timescale of ~ 3 Myr for the dissociation of molecular gas within the wind. Overall, current observations in the Galactic center support a scenario in which a hot wind entrains cold gas clouds from the disk, driving their progressive transformation from molecular to atomic and ultimately ionized gas through stripping, turbulence, and dissociation.

Figures (12)

• Figure 1: The sample of high-velocity clouds studied in this work. The grey-scale map represents the MW disk Lockman+16, while the blue colorscale map the outflowing H i high-velocity clouds DiT+18. Red circles mark the clouds followed-up in the $^{12}$CO(2$\rightarrow$1) and $^{12}$CO(4$\rightarrow$3) lines with APEX and analysed in this work, numbered as in \ref{['tab:sample']}. Dotted gray circles denote clouds not detected in CO emission.
• Figure 2: Moment maps and example spectra for two clouds (MW-C10 and MW-C13). For each cloud, we show the H i column density map (left, blue colorscale) from GBT data, the CO total intensity map (center, orange colorscale) and the CO velocity map (right, spectral colorscale). The red squares denote the field observed in CO with APEX. Beam sizes for the GBT and APEX data are indicated as grey circles in the bottom-right corner of the H i and CO maps, respectively. Spectra are extracted from the positions indicated on the CO total maps, with radius equal to the APEX beam size. Blue shaded bands in the spectra denote the velocity range of the detection. Contour levels in the H i and CO maps are at S/N levels of 2.5, 5, 10, 20 and 40.
• Figure 3: Cloud segmentation for three clouds in our sample: MW-C7_2 (left), MW-C1 (center) and MW-C10 (right). The blue-green colormaps show the masked $^{12}$CO(2$\rightarrow$1) integrated intensity maps. The black dashed contours denote the projected boundaries of each clump. The red crosses and ellipses show the emission centroid and an elliptical approximation of the emission in each clump. These ellipses represent a rough estimate of the size and orientation of the molecular clumps obtained through beam deconvolution. Clumps that could not be deconvolved are indicated as a filled dot. These clumps are not included in the analysis.
• Figure 4: DESPOTIC results for a model with $\sigma_\mathrm{turb} = 3 \, \,{\rm km\,s}^{-1}$, $\chi=\chi_0$, $\zeta=10^{-16}$ s$^{-1}$ and solar dust abundance. Top panels, from the left to the right: predicted H$_2$ column density, H i column density, H$_2$ and $\rm HI{H{\sc i}}$ abundances per H nucleus. Bottom panels, from the left to the right: gas temperature, CO abundance per H nucleus, integrated brightness temperature and CO-to-H$_2$ conversion factor for the $^{12}$CO(2$\rightarrow$1) emission line. Blank pixels denote regions of the grid where the models did not converge. The black contours on the $X_\mathrm{CO}$ plot represent an example of the constraint on the parameter space given by our data for MW-C1. Similar radiative-transfer models are used to calculate fiducial conversion factors within each H i cloud.
• Figure 5: Histogram distributions of the observed properties of the molecular clumps detected in the APEX data. From the left to the right: cloud radius (assuming a distance of 8.2 kpc), velocity dispersion and $^{12}$CO(2$\rightarrow$1) luminosity (Equation \ref{['eq:luminosity']}).