ALMA Unveils Widespread Molecular Gas Clumps in the Ram Pressure Stripped Tail of the Norma Jellyfish Galaxy

The galaxy ESO 137-001 in the Norma cluster (A3627; M_dyn ∼ 1 × 10¹⁵ M_⊙, σ = 925 km s⁻¹) is one of the nearest "jellyfish galaxies" (z = 0.0163), with a long X-ray, Hα, warm H₂, and CO tail, including young stars (Sun et al. 2006, 2010, 2007; Sivanandam et al. 2010; Fumagalli et al. 2014; Jáchym et al. 2014; Fossati et al. 2016). The presence of the one-sided gas tail, a strong H i deficiency of the main galaxy (Vollmer et al. 2001b), and an undisturbed stellar disk are clear characteristics of a ram pressure stripping interaction. The galaxy is projected at only ∼145 = 270 kpc from the cluster center (assuming ${H}_{0}=71$ km s⁻¹ Mpc⁻¹, Ω_M = 0.27, 1'' ≈ 0.327 kpc). The tail is observed in its full extent thanks to a favorable projection on the sky; the galaxy's radial velocity relative to the cluster mean is only about −200 km s⁻¹, so its motion is mostly in the plane of the sky.

The multiphase ESO 137-001 tail is also a multicomponent system with a complex morphology; see the composite X-ray/Hα image in Figure 1. The central tail component that is brightest and longest in X-ray and Hα emission connects to the main body of the galaxy. It is rather narrow, with a width of ∼15'' = 5 kpc and a length of ∼210'' = 70 kpc and a slightly bent shape. In addition, there is also emission laterally beyond this central tail component. To the south, there is a secondary X-ray and Hα tail, which appears to originate outside the main body of the galaxy, above the stellar disk. To the north, there are some compact Hα sources ("orphan H ii regions" with no associated diffuse X-ray emission) and patches of X-ray and Hα emission. This part of the tail shows no gaseous emission connecting it to the stellar disk, and there is also less diffuse (Hα and X-ray) emission between the sources than in the southern tail component. Since RPS generally acts from the outside in (see Section 3.4), the different parts are likely at different evolutionary stages of stripping.

Zoom In Zoom Out Reset image size

We used ALMA to observe all three tail components and nearly the full length of the (X-ray) tail; see the maps in Figure 2. In the present paper, we first describe the ALMA observations in Section 2 and compare them with CO(2−1) single-dish observations. We present in Section 3 maps of CO emission in ESO 137-001 and their comparison with ancillary data. We identify signatures of direct stripping and those indicating in situ formation in Section 4 and describe "fireball" features that are distinctive of ram pressure stripping in Section 5. Our conclusions are given in Section 6.

Zoom In Zoom Out Reset image size

We use our previously obtained single-dish observations of ESO 137-001 with APEX (Jáchym et al. 2014) to make a direct comparison of the CO fluxes detected with APEX and ALMA in a number of common apertures. While in the main body of the galaxy, the APEX and ALMA fluxes are consistent (flux ratio of ∼1.3), in the tail regions, the APEX-to-ALMA ratio increases to values of 3–4. This suggests that while in the main body of the galaxy, both telescopes reveal about the same amount of CO-emitting gas, in the tail regions, the ALMA observations (which included ACA) resolve out ∼67%–75% of the CO emission. This is mainly due to missing total power observations and short spacings leading to a loss of extended, low surface brightness emission in the ALMA data. The RPS tail thus may be dominated by a diffuse, possibly warmer, subthermally excited molecular component (e.g., Pety et al. 2013) distributed on spatial scales larger than ∼18'' ≈ 6 kpc, which is the largest recoverable scale in our ALMA observations. Moreover, the ratio of the APEX-to-ALMA fluxes increases with distance along the tail, implying that the fraction due to an extended component becomes larger further out in the tail.

We assume that for the compact features visible in the ALMA map and discussed further in this paper, the standard (Galactic) CO-to-H₂ conversion factor is applicable, while in the extended diffuse CO-emitting component, it may be a factor of a few lower (e.g., Bolatto et al. 2013). The uncertainty in the value of the conversion factor does not affect our conclusions. Assuming that the extended CO emission is distributed smoothly along the tail, it is likely that locally, the fraction of retrieved (compact) emission by ALMA is much larger, possibly near 90%. The Spitzer IRS showed warm H₂ in the inner tail (Sivanandam et al. 2010). The origin of the diffuse molecular component and why it is supposedly stable against (gravitational) instabilities and thus does not form stars will be a matter for future studies.

In Figure 2, we show the total CO(2−1) intensity and velocity maps of ESO 137-001 and its tail, and in Figure 3, we show the CO intensity overlaid on Hubble Space Telescope (HST) and Hα images. In Figure 2, the maps are rotated (see the arrows indicating N and E). The RPS tail extends toward the right-hand side of the mosaic in a NW direction on the sky. Besides the bright CO-emitting molecular gas in the main body of the galaxy, the map reveals a rich distribution of mostly compact CO emission regions detected in the tail out to nearly 60 kpc distance from the disk and over a lateral extent of about 24 kpc. The CO clumps are associated with all three tail components: the northern "orphaned" regions, central tail, and secondary southern tail.

Zoom In Zoom Out Reset image size

The CO emission in the main body of the galaxy is confined to the central region out to a radius of about 4''–5'' ∼ 1.3–1.6 kpc (S–N). This is the region of the galaxy disk where stripping is currently active. The stellar disk beyond this radius has already been stripped (except in the north, where there is a CO clump at ∼7'' from the disk center). The amount of CO flux in the disk area (within a 4'' × 10'' area centered along the disk) is ∼43 ± 5 Jy km s⁻¹. The integrated flux corresponds to an H₂ mass¹¹ of ∼6.9 × 10⁸ M_⊙ and an average face-on gas surface density within the central r = 1.5 kpc of about 150 M_⊙ pc⁻². Such a surface density is typical for the center of gas-rich spirals and high enough to make the gas resistant to rapid ram pressure stripping.

Outside the main body, the brightest part of the tail is the central component, with a total CO integrated flux of 57 ± 10 Jy km s⁻¹, which corresponds to M(H₂) ∼ 9.1 × 10⁸ M_⊙. It extends to ∼58 kpc from the disk. It is brightest in its innermost parts; within a mere 2.5 kpc adjacent to the disk, about 40% of the total CO flux is encompassed. By contrast, the lateral components of the tail contain much less molecular gas: the southern tail registers about 9 ± 2 Jy km s⁻¹, which is ∼1.4 × 10⁸ M_⊙ of H₂, whereas the northern "orphaned" regions contain ∼6 ± 2 Jy km s⁻¹, i.e., M(H₂) ∼ 9.6 × 10⁷ M_⊙. Thus, about 80% of the detected total tail CO flux is in the central component, while small fractions of ∼10% are in the southern and northern regions. This indicates that the most important contribution to the molecular tail component comes from the relatively dense ISM most recently stripped from the central disk regions.

The total mass of the detected dense molecular gas in the tail is comparable to the total estimated X-ray gas mass of about ∼10⁹ M_⊙ but larger than the upper limit estimate for neutral atomic gas of ∼5 × 10⁸ M_⊙ (Jáchym et al. 2014; Sun et al. 2006; Vollmer et al. 2001b).

The velocity structure shown in the bottom panel of Figure 2 is dominated by the galaxy's rotation that is imprinted into the stripped gas: most of the northern emission regions are blueshifted relative to the tail centroid, while those in the south are mostly redshifted. Due to the galaxy's orbital motion that is mostly in the plane of the sky, a velocity gradient along the tail is absent. A similar velocity structure was also reported from the early Hα observations (Sun et al. 2010) and later confirmed from the detailed MUSE Hα map (Fumagalli et al. 2014). The CO velocity dispersion is typically ∼5–15 km s⁻¹ in most clumps but larger (∼15–25 km s⁻¹) in the galaxy and the brightest regions of the inner tail.

The spatial resolution of the present ALMA observations of ∼350 pc does not allow us to resolve individual giant molecular clouds (GMCs) that in galaxies have typical sizes of the order of ∼1–100 pc. Besides some unresolved clumps, most of the detected CO emission regions in the tail of ESO 137-001 are larger structures, some with a clear substructure. Typical CO clumps have masses characteristic of GMCs or giant molecular associations (GMAs) of ∼10⁶–10⁷ M_⊙, but their sizes (and also velocity dispersion) are larger (∼500–800 pc). This suggests that the clumps are not gravitationally bound and will likely disperse with time. For the selection of the CO clumps described in the present work, we can estimate their virial parameter α. Following Bertoldi & McKee (1992),

$$\begin{eqnarray}&&\alpha =5{\sigma }^{2}R/(G\,M),\end{eqnarray} \tag{ 1 }$$

where σ, R, and M are a clump's velocity dispersion, radius, and mass, respectively (see the CO spectra in Figure 5). The parameter α spans a range of ∼2–12 that indicates subcritical structures (nonmagnetized clouds are expected to have critical virial parameters of ∼2; e.g., Kauffmann et al. 2013). This supports the abovementioned suggestion that molecular clumps will dissolve with time; however, the uncertainty of these estimates may be large. A detailed study of the molecular clump population detected in the ALMA data in the tail and its relationship to the star-forming regions will be the subject of a future paper. Also in the narrow RPS tail of the Coma galaxy D100, stellar sources were measured to be much larger than single gravitationally bound star clusters, indicating that they are likely to be unbound and disperse (Cramer et al. 2019).

In Figure 3, the CO(2−1) intensity map (red) is overlaid with Hα (green) and HST UV–visible–near-IR composite images covering about the inner half of the length of the (X-ray) tail. Contours of X-ray emission are added into the overlay (Sun et al. 2010). The image indicates various levels of correlation between the CO, Hα, and young stars. It also clearly shows that the narrow central tail is connected to the CO-emitting molecular gas still in the central parts of the main body of the galaxy. The CO clumps detected outside the main body of the galaxy are associated with all tail components, but in comparison to the more extended Hα distribution, the CO emission comes from more compact regions. Given that the CO emission traces a relatively dense gas phase (see Section 2.1), the ALMA CO observations reveal for the first time the dense "skeleton" of the multiphase gaseous tail, as well as the component that forms the stars that are visible in the HST image. The overlay image in Figure 3 offers the most complete view of the spectacular RPS tail up to now, with the CO, Hα, and X-ray-emitting gas components, as well as the regions of young star formation.

The overall molecular gas distribution in the tail roughly follows the distribution of the other gas phases and reflects the stripping history of the disk. The lateral extent of CO in the tail is much broader than the lateral extent of CO in the disk. This is because RPS generally acts from the outside in, and outer disk gas is generally stripped earlier and forms the broadest parts of the tail. The broadest parts of the tail (the northern and southern components) originated from outer disk regions that were fully stripped some time ago. The kinematics support this with red- and blueshifted velocities in the parts of the tail furthest from the tail centroid (Figure 2).

The complex morphology of the multicomponent tail of ESO 137-001 is not consistent with the simple outside-in stripping. The continuous central long component of the tail that is also bright in CO and connects to the CO in the main body of the galaxy extends much further out than the innermost northern tail component, and it is also spatially separated from both the northern and southern components. This indicates that the central few kpc of the galaxy have been experiencing stripping at the same time as the outer disk. In a galaxy with ISM substructure, the ICM wind may penetrate through low-density parts of the disk, and multiple radii may be stripped simultaneously (e.g., Quilis et al. 2000). This is what we think is likely happening in ESO 137-001.

The timescale of gas stripping depends on the galaxy's orbit in the cluster, as well as on the gas column density. Assuming a cluster model with a virial mass and radius consistent with the measurements of Woudt et al. (2008) and the ICM distribution β-model fit to the ROSAT X-ray data of Boehringer et al. (1996), semi-analytic modeling indicates that ESO 137-001 is currently only ∼100 Myr before pericenter along a highly radial orbit with a large relative velocity >3000 km s⁻¹ and the corresponding peaked ram pressure time profile having a FWHM ∼ 200 Myr and current value ${P}_{\mathrm{ram}}={\rho }_{\mathrm{ICM}}\,{v}_{\mathrm{orb}}^{2}\approx 1.5\times {10}^{-10}$ dyne cm⁻² (Jáchym et al. 2014). For the range of column densities corresponding to the mean values of the typical CO clumps detected in the tail, we can calculate the distances along the tail to which stripped disk gas parcels have been pushed by the (face-on) ram pressure and their corresponding travel times. From, e.g., 2 kpc radius in the disk, a gas clump with a column density of 10 (20 or 30) M_⊙ pc⁻² was pushed to 80 (20 or 7) kpc distance along the tail, with the corresponding travel time of 200 (90 or 50) Myr since the gas parcel crossed 1 kpc height above the disk midplane. From the outer disk radii, stripping has started much earlier along the orbit, and the stripped gas reaches much larger distances. These calculations neglect all hydrodynamic effects, as well as the internal structure of gas clouds.

The new observations also show the asymmetry between the northern and southern tail components. There are prominent isolated CO clumps in the northern tail occurring to nearly 40 kpc from the main galaxy, whereas in the southern tail, clumps are found only out to ∼20 kpc (with some marginal detections at larger distances). The asymmetry of the tail is likely caused by the effects of disk inclination with respect to the ram pressure direction. The interaction with the ICM wind is not face-on but tilted¹² by ∼47°, which means that the rotation of the galaxy has affected the trajectories of the stripped gas in an asymmetric manner: the gas on the northern disk side has a rotation velocity component pointing against the ICM wind, while the gas on the southern disk side rotates with the wind.

Thus, different parts of the tail are at different evolutionary stages of stripping: the central component is still being fed by the galaxy, while the outer, northern, and southern components represent later evolutionary stages, since these parts of the tail are no longer being fed from the galaxy. The northern tail component appears to be at a later evolutionary stage than the southern component, as it has less diffuse Hα and X-ray emission with isolated ("orphaned") dense CO and Hα clumps apparently left behind the more diffuse gas. Despite their different ages, all three tail components are found to host molecular gas and young stars.

At about 9.5 kpc from the galaxy, we find an extended bright CO feature that is tilted ∼60° from the tail axis (see the CO velocity field and integrated CO spectrum in Figure 4, top panels), nearly parallel to the main disk. It is about 10'' long and 35 wide (≈3.3 kpc × 1.1 kpc). Its total integrated flux is 8 ± 1 Jy km s⁻¹, corresponding to ∼1.3 × 10⁸ M_⊙. The velocity of the structure ranges from ∼−50 to 20 km s⁻¹, with a gradient along its length in the N–S direction.

Zoom In Zoom Out Reset image size

The feature may be part of a longer structure that extends through the inner tail and bends upward (see dotted lines in Figure 4). The fact that it is tilted at a large angle with respect to the main tail direction and is possibly connected to the gas in the galaxy strongly suggests it originated from a large dense gas complex in the galaxy, such as a spiral arm, that was gradually removed from the disk by ram pressure. The reason is that the flow in the downstream direction naturally makes elongated gas features in the tail direction but does not naturally create structures oriented at an angle to the tail direction. To directly entrain dense, large clumps, dynamical effects can help. In a tilted ram pressure orientation, some larger structures may have a rotation velocity component in the direction of the wind, and in combination with a locally enhanced gravitational potential (e.g., by overdensities in spiral arms), they may be pushed gradually by the wind instead of being disrupted.

The 60° tilted feature is an example of a dense structure of the RPS tail that remained hidden in the previous observations; neither Hα nor X-rays show an associated coherent structure, and there is also no excess of young stars. In Section 3.4 we estimated that a gas clump with the column density of 30 M_⊙ pc⁻² at 2 kpc disk radius has been pushed to ∼7 kpc from the disk in the last 50 Myr. Despite the simplicity of the model, it matches well the mean column density (∼35 M_⊙ pc⁻²) and location of the 60° tilted feature and thus indicates that the (stripping) age of the feature is ∼50 Myr. This time also corresponds to (less than) half of the galaxy rotation period (for a galaxy rotation velocity of ∼110 km s⁻¹; Jáchym et al. 2014). The observed shape of the CO structure connecting the 60° tilted feature to the main galaxy that is not twisted but forms a single arc is consistent with this evolutionary path.

The main issue for the direct stripping scenario is the survival of molecular clouds in the presence of the disruptive effects of the hot ICM wind, notably to the Kelvin–Helmholtz (KH) and Rayleigh–Taylor instabilities. The crushing time of a cloud with a radius ${r}_{\mathrm{cloud}}$ in a wind with a speed v_wind can be approximated by the time it takes for a shock to propagate through the cloud (Klein et al. 1994),

$$\begin{eqnarray}&&{t}_{\mathrm{cc}}={\chi }^{1/2}\displaystyle \frac{2\,{r}_{\mathrm{cloud}}}{{v}_{\mathrm{wind}}},\end{eqnarray} \tag{ 2 }$$

where χ is the density contrast between the cloud and the wind. For clouds with parameters of typical GMCs, the crushing time due to an ∼1000 km s⁻¹ wind is rather short, ∼10⁷ yr. For larger clumps, it is longer, >10⁸ yr, which is comparable to their travel (drag) time to the inner regions of the tail by ram pressure. Some physical processes may protect the clouds and prolong their lifetime, such as magnetic fields that reduce the growth rate of the KH instability (e.g., Kauffmann et al. 2013). Also, radiative cooling effectively extends the lifetime of dense clouds by replenishing the cold gas and preventing complete mixing with the hot wind (e.g., Cooper et al. 2009). Since the cooling time is shortest for the densest gas, it will be the densest stripped gas clouds or clouds compressed by RP that will form most of the extraplanar molecular clouds in the inner parts of the tail, such as the 60° tilted feature.

It is thus possible that direct stripping is a viable scenario for the origin of the CO-bright large features observed in the inner parts of the tail, close to the main galaxy. This is consistent with the continuous connection of the CO emission in the base of the tail to the galaxy. Dense molecular clouds may be embedded in large reservoirs of more diffuse molecular ISM (Pety et al. 2013), which further protect them against the disruptive effects of ram pressure and may itself be directly stripped (see Section 2.1).

Another large CO feature in the central tail component is a linear structure located at nearly 30 kpc from the galaxy, somewhat south of the central tail (Figure 4, bottom panels). It is longer and thinner, has more substructure than the 60° tilted feature, and, most substantially, is parallel to the main tail direction. Along its length of ∼18'' ≈ 6 kpc, at least eight clumps are identified. Some of the clumps are resolved, but the width of the feature is mostly unresolved at ∼1''. The total CO flux of nearly 2 Jy km s⁻¹ corresponds to ∼3.2 × 10⁷ M_⊙. The masses of individual clumps are typically ∼2.5–5 × 10⁶ M_⊙. Their characteristic separation in the filament is ∼2'' ≈ 0.5–0.6 kpc. The radial velocity range of the structure is ∼0–40 km s⁻¹ with no apparent velocity gradient along its length (although the inner half is at a somewhat lower velocity than the outer half).

While there is diffuse Hα observed along the entire feature, there are virtually no young stars or associated compact Hα peaks. This implies that the CO filament is relatively young, unevolved with respect to star formation. Moreover, it is located far from the galaxy and oriented parallel to the main tail. These characteristics make it consistent with a dense gas filament formed recently in situ in the tail from a less dense gas. It then fragments via a gravitational instability, forming a regular substructure of molecular clumps, which may eventually form new stars. Thanks to the offset location from the main part of the central tail, the CO filament is a particularly clear example of a possible in situ formation.

From the theory of fragmentation of self-gravitating isothermal, hydrostatic cylinders, following the analysis in, e.g., Mattern et al. (2018), we can estimate the most unstable scale,

$$\begin{eqnarray}&&{\lambda }_{\max }=\displaystyle \frac{22\,{c}_{s}}{\sqrt{4\pi G{\rho }_{c}}},\end{eqnarray} \tag{ 3 }$$

where ρ_c is the central density of a filament in virial equilibrium, ${c}_{s}=\sqrt{{k}_{B}T/\mu {m}_{{\rm{H}}}}$ is the sound speed, and T, μ, and m_H represent the isothermal temperature, mean molecular weight, and mass of the hydrogen atom, respectively. For the mean density calculated for individual clumps from their masses (∼2.5 × 10⁶ M_⊙) and sizes (∼350 pc), and assuming T = 15 K to calculate the sound speed, the corresponding fragmentation length scale is ∼250 pc, which is within a factor of ∼2–3 consistent with the observed clump separation in the filament. The predicted value would be higher and thus better match the observed separation when a turbulent medium is assumed, in which the velocity dispersion is higher than the sound speed. While these estimates are strongly simplified by neglecting the presence of an external pressure, rotation, magnetic fields, and by using mean densities, they suggest that fragmentation due to self-gravity is taking place in the stripped gas.

In addition to the central filamentary structure, the most distant CO emission region detected at about 58 kpc from the galaxy also has a linear morphology. It is shorter; composed of at least three clumps with a total integrated flux of ∼1 Jy km s⁻¹, thus about 1.6 × 10⁷ M_⊙; and is oriented roughly along the tail direction (see Figure 2). The flow of matter in the downstream direction naturally creates elongated gas features in the tail direction that may eventually cool down, condense, and fragment. The two CO filamentary structures may provide the first direct evidence for molecular gas formation in situ in RPS tails.

The importance of the scenario of molecular gas formation in the tail is also supported by the large total amount of molecular gas revealed in the tail in the present observations (plus taking into account the more extended component; see Section 2.1). Following the scaling relations of, e.g., Catinella et al. (2018), for an unperturbed galaxy with a stellar mass ∼10¹⁰ M_⊙, the expected total gas fraction is ∼0.3, and the molecular-to-atomic gas mass ratio is ∼0.2; thus, the estimated pre-stripping H₂ content of ESO 137-001 is ∼5 × 10⁸ M_⊙. While such estimates may vary by almost an order of magnitude at a fixed stellar mass, this value is well consistent with the detected amount of ∼7 × 10⁸ M_⊙. This suggests that the galaxy is not H₂-deficient. But it is strongly H i-deficient, following an expected pre-stripping H i content of ∼2.5 × 10⁹ M_⊙ and a weak upper limit obtained from Australia Telescope Compact Array observations of ∼5 × 10⁸ M_⊙ (Vollmer et al. 2001b). Thus, there was, in principle, enough stripped H i to account for the observed H₂ content.

Recent simulations of the evolution of cold (∼10⁴ K) gas in a hot galactic wind have shown that if the cooling is fast (t_cool ≪ t_cc) and the clouds are sufficiently large, they can withstand longer mixing times due to the KH instability, and new cold gas may form far downstream from the clouds from a "warm" mixture of the cold gas and hot wind (Gronke & Oh 2018). Moreover, in the simulations, the amount of newly formed cold gas may exceed the original cold gas in the clouds. Assuming that the ∼10⁴ K gas in the simulations can cool further down to reach the typical temperatures of molecular gas, this process could help to understand the origin of the large total H₂ content of ESO 137-001.

In order to turn the stripped cold gas (H i) into a molecular phase, the presence of dust is crucial. Imaging of the galaxy with HST clearly shows prominent dust features on the west side of the disk and the inner tail. In addition, Herschel SPIRE imaging (S. Sivanandam et al. 2019, in preparation) reveals a dust trail emanating from the galaxy in the direction of the gas tail. Other optical observations (e.g., HST) of RPS galaxies clearly show dust being stripped from the galaxies along with the gas (Elmegreen et al. 2000; Abramson et al. 2016; Cramer et al. 2019).

Interestingly, the physical scenarios of direct stripping and in situ formation are also relevant for galactic winds, which can have a cold gas component associated with the hot gas in outflows (see, e.g., Banda-Barragán et al. 2019), and for brightest cluster galaxies, which can have molecular gas filaments extending radially from the galaxies (e.g., Vantyghem et al. 2019).

In all three tail components, one finds CO clumps with offset small streams of young stars extending toward the galaxy. Four examples of fireballs are shown in Figure 5. The stellar streams have lengths of ∼15–45, i.e., ∼0.5–1.5 kpc, and are all oriented roughly parallel to the tail direction. The CO clumps at the heads of the fireballs are associated with compact Hα (H ii) regions. Some components of fireballs (Hα and young stars) have previously been observed, but this is the first time the important molecular component has been detected.

Zoom In Zoom Out Reset image size

Figure 5 also shows integrated CO spectra of the clumps at the heads of fireballs. The integrated CO fluxes are typically ∼0.5 Jy km s⁻¹, thus about 8 × 10⁶ M_⊙, but some are fainter, with an integrated CO flux of ∼0.16 Jy km s⁻¹, i.e., 2 × 10⁶ M_⊙. The typical sizes of the CO clumps are ∼2''–25 ≈ 650–800 pc, but some are smaller, ∼500 pc. Line widths are 10–15 km s⁻¹. Many other fireballs in the tail have compact Hα clumps at their heads but no detected CO emission, similar to fireballs observed previously in other galaxies. This may indicate a different evolutionary stage in which molecular gas has already been consumed by star formation.

While we cannot measure the velocity structure of the fireballs due to the orientation of the orbit of ESO 137-001 in the plane of the sky, it is clear that their kinematics evolves with time. It depends on the timescales of ram pressure acceleration, gravitational acceleration from the galaxy, cloud condensation, and star formation. The newly formed stars gradually decouple from the gas clump and may eventually start falling back toward the galaxy if they have not reached the escape speed. Simple modeling with a realistic time-varying ram pressure and a galactic potential (adopted from Jáchym et al. 2014) shows that a separation of ∼1 kpc between the parent gas cloud and a decoupled star (modeled as a high column density parcel) typically establishes within (a few) 10 Myr. Thus, the observed fireballs (including those shown in Figure 5) are most likely dynamically young structures with molecular clumps at their head that have been actively forming new stars over the past ∼10 Myr.

In the northern component of the tail, close to the disk, we find possible examples of much larger fireballs. In this part of the tail, which likely is in a later evolutionary stage of stripping, there is no diffuse X-ray or Hα emission but distinct streams of young stars, H ii regions (compact Hα sources), and bright CO clouds. Most of the young stars seen in the HST image in this region fall within two linear streams ∼15''–25'' ≈ 5–8 kpc in length, parallel to the main tail (see a closeup in Figure 6).

Zoom In Zoom Out Reset image size

Each of these streams have CO clouds coincident with or somewhat beyond the outermost stars (in the downstream direction). This is consistent with the fireball scenario. The CO clouds have diameters of ∼2'' and 4'', i.e., ∼650 and 1300 pc (but with possible substructure); their line widths are ∼22 and 30 km s⁻¹ and H₂ masses ∼6.4 × 10⁶ and 1.9 × 10⁷ M_⊙, respectively, for the E and W structures. The long streams do not have young stars extending continuously to the CO cloud or along the length of the streams. This indicates that molecular clouds in tails may survive after a period of active star formation ends and possibly go through several more phases of star formation activity. This is consistent with the subcritical values of the virial parameter α of the detected CO clumps (see Section 3.2) if the clumps have denser substructures, unresolved by our present observations. Then the smaller fragments of otherwise dispersing clumps could survive, and some of them could eventually condense into star-forming regions. For example, the western fireball shows an Hα compact feature (with no detected associated CO emission) in its inner half. If not simply due to the effects of projection, this could correspond to a previous episode of condensation when denser (pre-stellar) fragments decoupled from a parental clump. The latter continued to be accelerated further by ram pressure in the downstream direction and condensed again only recently. The multiple features are then parts of one large-scale structure. In this picture, small-scale fireballs can, in fact, be part of large-scale fireballs. We note that the locations of the two CO clumps at the heads of the large fireballs lay outside the MUSE area, but previous Southern Observatory for Astrophysical Research (SOAR) observations show bright compact Hα emission associated with at least the western CO clump (Sun et al. 2007).

Superfireballs, due to their lengths, are expected to be dynamically older structures than the small-scale fireballs. This is consistent with their location in the northern outer tail component. Our modeling (described in Section 5.1) shows that ∼50 Myr is needed to develop a separation of ∼10 kpc between the parent cloud and decoupled stars. We may also be seeing superfireballs in the process of formation in the southern tail component. There, the sensitive MUSE Hα observations reveal many elongated structures (see Figure 3), some of which show a head–tail morphology, with diffuse tails extending away from the galaxy and compact clumps at the heads situated closer to the galaxy (sometimes associated with CO emission). Some are aligned into long coherent structures encompassing other Hα compact clumps. It is possible that with ongoing ram pressure, the diffuse component will be cleared out of the structures and moved downstream along the tail, leaving behind "nude" structures of denser clumps and stellar streams, i.e., superfireballs.

Some of the Hα head–tail features in the tail are identified to be interconnected with fireballs. In addition to the Kenney et al. (2014) model, the tail of ESO 137-001 indicates that the fireballs may be double-tail features, with a second, gaseous tail of diffuse (Hα) gas that is ablated from the dense star-forming clump that extends in the opposite direction to the stellar tail, away from the galaxy. The central dense clump of a fireball is bright in Hα and possibly also in CO emission. A cartoon of the model of the double-sided fireballs is shown in Figure 7. Similar features of considerable lengths were previously found in the RPS tail of the Coma galaxy RB199 (Yoshida et al. 2008), although without any associated molecular component yet observed.

Zoom In Zoom Out Reset image size

An example of a double-sided fireball is given in Figure 8, which shows a compact H ii region associated with a (slightly offset) weak CO clump at the head of a small fireball with a diffuse tail extending along the direction of the wind. It is located in the northern component of the tail, close to the main galaxy. The associated CO emission region has an H₂ mass of ∼2.4 × 10⁶ M_⊙. More similar structures (though less clear and without detected associated CO clumps) may be found mainly in the southern tail component.

Zoom In Zoom Out Reset image size