ALMA Images of the Host Cloud of the Intermediate-mass Black Hole Candidate CO−0.40–0.22*: No Evidence for Cloud–Black Hole Interaction, but Evidence for a Cloud–Cloud Collision

Figure 1 shows the integrated intensity maps and the PV diagrams for the HCN J = 3–2 and CO J = 3–2 lines. The PV diagrams are plotted along the right ascension (R.A.) axis and are averaged over the full decl. range. Multiple velocity components exist along the line of sight toward the target cloud. The emission from CO−0.4 extends from −110 to −20 km s⁻¹ in the velocity of the local standard of rest (LSR) (v_LSR). The ¹³CO J = 2–1 image is self-absorbed at v_LSR = −50, −25, and −5 km s⁻¹ by foreground clouds in the Galactic disk, whereas the HCN J = 3–2 data are almost free of self-absorption dips for the velocity range of CO−0.4. The narrow line emission at v_LSR = 20 km s⁻¹ in the CO J = 2–1 PV diagram is the foreground infrared dark cloud G359.62−0.24 (or the "Freccia Rossa" cloud; Bally et al. 2010; Ravi et al. 2017), which is seen as an absorption dip in the HCN J = 3–2 image. Other positive-velocity emissions are from Galactic center (GC) clouds unrelated to CO−0.4.

Zoom In Zoom Out Reset image size

The CO integrated intensity map illustrates the highly complicated cloud structure, which consists of many small, entangled filamentary features. In contrast, most of the spatially extended emissions are transparent in the HCN J = 3–2 image, in which the central high-density portion of the cloud is captured owing to its high critical density (n_crit). In Section 3.2., we mainly use the HCN J = 3–2 data to analyze the cloud structure.

Figure 2 shows the continuum maps at 230 and 265 GHz. We identified two bright point-like sources that are present in both frequency bands, which we labeled MM1 and MM2. Source MM1 corresponds to the IMBH candidate CO−0.4* (Oka et al. 2017); it is not an isolated point source in our images, but is located in a ridge-like extended background emission 7'' long.

Zoom In Zoom Out Reset image size

Figure 3 shows the spectra of the three spectral windows averaged over the five positions indicated in Figure 4. In addition to the CO and HCN lines, more than 20 lines from CH₃OH, CH₂NH, NO, ${\mathrm{SO}}_{2}$, and c-HCCCH are detected. These emissions are mostly concentrated in a filament running through the central region of CO−0.4 in an approximately southeast–northwest direction, as represented by the CH₃OH 11₀–10₁ map shown in Figure 4. This filament has an approximately constant velocity of approximately −80 km s⁻¹ throughout the full length. A corresponding filamentary structure is also seen in the HCN map in the −90 to −70 km s⁻¹ velocity range. We refer to this structure henceforth as the −80 km s⁻¹ filament. The brightness of the filament in optically thin methanol and other rare molecular lines indicates that the filament is the densest part of CO−0.4.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The −80 km s⁻¹ filament has a few knots with bright methanol emissions. The two brightest knots are labeled as knots A and B in Figure 2. Both knots are also bright in the HCN J = 3–2 image, and the filament is wider at knot B than at other portions along its length. Knot B is located at the position where Oka et al. (2017) reported the gravitationally kicked cloud.

3.2.1. Three Velocity Components

In this section, we analyze the internal structure and velocity field of CO−0.4 using the HCN J = 3–2 data cube. First, we extract the primary structure of the cloud in position–position–velocity (PPV) space by means of the dendrogram method (Rosolowsky et al. 2008); this step is necessary since the HCN data cube exhibits highly complicated PPV structure, in which almost every line-of-sight contains multiple velocity components including those physically unrelated to the main body of the cloud. We identified six main branches in the dendrogram created from the HCN data cube, and reconstructed the image of the primary PPV structure by pruning other minor structures and components unrelated to the cloud. The details of the dendrogram analysis are described in the Appendix.

Figure 5 shows the averaged spectra for the six main branches. The figure indicates that the cloud consists of three major velocity components: two narrow line components at −40 and −80 km s⁻¹, and a moderately broad component at −60 km s⁻¹. The FWZI velocity widths are 10 km s⁻¹ for the −40 and −80 km s⁻¹ components, and 50 km s⁻¹ for the −60 km s⁻¹ component. The single-dish HCN J = 4–3 spectrum averaged over the entire cloud (Tanaka et al. 2014) is also presented in the figure for comparison, showing that the broad profile of the averaged spectrum is the superposition of the three velocity components at −40, −60, and −80 km s⁻¹, with ≲50 km s⁻¹ widths, but none of the six main branches alone can account for the ∼90 km s⁻¹ velocity width of the averaged spectrum.

Zoom In Zoom Out Reset image size

3.2.2. Moderately Broad-line Emissions

Figure 6 shows the 2D maps of the zeroth moment, first moment (moment-0 and moment-1), and velocity width (Δv), which are created from the reconstructed image of the main structure of the cloud. The quantity Δv is defined by the moment-0 values divided by the spectral-peak flux densities pixel-by-pixel. We also show the moment-1 map of the HCN map smoothed to a 20'' resolution, i.e., the typical resolution of the previous single-dish observations (Tanaka et al. 2014; Oka et al. 2016) in Figure 6(d).

Zoom In Zoom Out Reset image size

The Δv map (Figure 6(c)) shows that individual spectra consisting of CO−0.4 are predominantly narrow; approximately 75% of the pixels have Δv narrower than 15 km s⁻¹, and the most frequent value is 10 km s⁻¹. These velocity widths are smaller than the variation in the moment-1 velocities over the entire cloud, ∼60 km s⁻¹. Pixels with Δv larger than 25 km s⁻¹ are 5% of the entire map, and their spatial distribution is concentrated to a few narrow areas near the −80 km s⁻¹ filament, in particular, to knots A and B. This confirms that the 90 km s⁻¹ width broad emission in the single-dish image consists of a superposition of multiple velocity components with narrow velocity widths.

Figure 7 is a close-up view of knots A and B that presents the distributions of the three velocity components and the moderately broad emission with Δv > 25 km s⁻¹. The figure shows that the −40 and −80 km s⁻¹ components are loosely anti-correlated with each other on the plane of the sky. In addition, the spatial distribution of the −60 km s⁻¹ component is correlated well with that of the moderately broad emission, and both extend along the narrow boundary regions between the −80 and −40 km s⁻¹ components. This indicates that the moderately broad emissions and the −60 km s⁻¹ component are the same physical entity, and they represent the transition region of the other two components in PPV space; this structure can be interpreted such that the line-broadening at knots A and B is caused by a physical interaction between −80 and −40 km s⁻¹ components.

Zoom In Zoom Out Reset image size

3.2.3. V-shaped Velocity Gradient across the −80 km s⁻¹ Filament

Figure 8 presents PV diagrams of the original HCN J = 3–2 data for two cuts crossing the −80 km s⁻¹ filament at the locations of knots A and B. Along both cuts, the PV diagrams show characteristic V-shapes; the −80 km s⁻¹ filament and the −40 km s⁻¹ component are smoothly bridged by a pair of bright emission features with steep velocity gradients. These bridging emissions correspond to the moderately broad −60 km s⁻¹ component identified with the dendrogram analysis. We note that this V-shape is not a PV pattern that appears only along these two particular cuts, but it persists over a wide spatial range along the filament. The moment-1 map (Figure 6(b)) shows that the northern part of the filament from knot A to knot B is interposed by a pair of bands with −60 to −40 km s⁻¹ velocities that are running parallel to the filament; this configuration is observed as the V-shaped pattern across cuts perpendicular to the filament. Formation of this V-shaped PV pattern is readily explained by assuming physical interaction between the −80 km s⁻¹ filament and clouds at −40 km s⁻¹. We will discuss this in more detail in a later section (Section 4.2).

Zoom In Zoom Out Reset image size

We have investigated the physical conditions in the −80 km s⁻¹ filament by means of a population diagram. Figure 9(a) shows the rotational population diagrams for the a-CH₃OH and CH₂NH lines, constructed from the peak-intensity values of the spectra averaged over the five positions along the filament indicated in Figure 4. We ignored the c-HCCCH line that blended with the 251.517 GHz CH₃OH line, assuming that the former is significantly weaker than the latter. The error bars include both the rms noise measured for a blank-sky region in the data cube and conservatively assumed 20% relative errors. The rate coefficients and partition functions are taken from the Leiden Atomic and Molecular Database (LAMDA; Schöier et al. 2005), the Cologne database for molecular spectroscopy (CDMA; Muller et al. 2001), and Motoki et al. (2014).

Zoom In Zoom Out Reset image size

Each of the a-CH₃OH and CH₂NH level population diagrams is consistently fitted by a single-component, local thermodynamic equilibrium (LTE) curve except for the 250.507 GHz line of methanol, which is a predicted class-I maser line (Voronkov et al. 2012). The diagram indicates two temperature components, corresponding to cold gas with a CH₂NH rotational temperature T_rot ∼ 24 K and warm gas with a CH₃OH rotational temperature T_rot ∼ 70 K. Although accurate values of n_crit for the CH₂NH lines are unknown, the similar A coefficients for the CH₃OH and CH₂NH lines (∼10⁻⁴ s⁻¹) allow us to assume that they have similar values of n_crit and hence that the different T_rot values reflect differences in the physical conditions of the emitting regions.

We performed a non-LTE analysis for the methanol lines using collisional excitation coefficients taken from LAMDA, where the gas kinetic temperature T_kin, column density N, and hydrogen volume density ${n}_{{{\rm{H}}}_{2}}$, are taken as free parameters. Figure 9(b) shows the simultaneous credible intervals for T_kin and ${n}_{{{\rm{H}}}_{2}}$, calculated from a Bayesian analysis assuming a prior function that is flat for 1 < T_kin/K < 1000 and 1 < ${n}_{{{\rm{H}}}_{2}}$/cm⁻³ < 10⁷ and zero otherwise. Although the degeneracy between T_kin and ${n}_{{{\rm{H}}}_{2}}$ is not resolved, this result indicates that the filament has highly elevated T_kin and/or ${n}_{{{\rm{H}}}_{2}}$ compared with the physical conditions of standard GC clouds, for which T_kin = 30–100 K and ${n}_{{{\rm{H}}}_{2}}$ = 10⁴ cm⁻³ (e.g., Martin et al. 2004; Nagai et al. 2007; Arai et al. 2016; Ginsburg et al. 2016). In particular, this analysis indicates that ${n}_{{{\rm{H}}}_{2}}$ ≳ 10⁶ cm⁻³, unless an unrealistically high T_kin of > 10³ K is assumed. This result is also consistent with that from a non-LTE analysis using HCN lines (Tanaka et al. 2014).

It is worthwhile to note that T_rot of 70 K for the warm component is among the highest of those measured for the CMZ clouds; it is comparable to that measured toward the Sgr B2 hot cores. In contrast, typical CMZ clouds have 10–20 K CH₃OH rotational temperatures (Requena-Torres et al. 2006). The coexistence of this exceptionally high-temperature gas and the cold component indicates the existence of some heating mechanism working near or inside the −80 km s⁻¹ filament.

The column densities of CH₂NH and CH₃OH obtained from the non-LTE analysis are not particularly large compared with those in typical GC clouds or hot cores. We calculated the fractional abundances of CH₂NH and CH₃OH to be 2 × 10⁻⁸ and 1 × 10⁻⁷, respectively, using hydrogen column densities obtained from the 265 GHz dust flux assuming a dust temperature of 20 K and dust-emissivity index of 1.5. These abundances are typical values for hot cores or GC clouds (Requena-Torres et al. 2006; Suzuki et al. 2016).

The observational bases of the gravitational-kick scenario presented by Oka et al. (2016, 2017) can be summarized as follows:

• 1.  
  The low-resolution images obtained with the NRO 45 m and ASTE 10 m telescopes show a slightly elongated elliptical shape. Oka et al. (2016) detected a steep velocity gradient from −110 to −20 km s⁻¹ across the major axis, and interpreted it as a beam-smeared image of a gas stream formed thorough IMBH–cloud interaction. Although the stream is not fully resolved in the single-dish images, which have ∼20'' angular resolutions, Oka et al. (2016) predicted its PV structure on the basis of numerical simulations.
• 2.  
  Oka et al. (2017) identified a compact clump with an extremely broad emission of a 110 km s⁻¹ width near the predicted position of the hypothesized IMBH, using the high-resolution image obtained with the ALMA observation. The authors consider that this extremely broad emission clump had been scattered by tidal interactions with the IMBH. The PV position of the clump is approximately consistent with the trajectory of the stream predicted by Oka et al. (2016).
• 3.  
  Oka et al. (2017) argued that the compact continuum source MM1/CO−0.4* is self-absorbed synchrotron emission from an IMBH, on the basis of the shallow spectral index of α = 1.18 ± 0.65 measured from the continuum fluxes at 230 and 265 GHz. The position of the source is consistent with the locus of the gravitational source in the model in Oka et al. (2016).

In this section, we examine these arguments using our reanalyzed ALMA data.

4.1.1. Stream

4.1.2. The 110 km s⁻¹ Wide Emission

4.1.3. Continuum Source MM1/CO−0.4*

The shallow spectral index α ∼ 1 measured for the narrow band of the ALMA data (Oka et al. 2017) contradicts the reported nondetection at 34.25 GHz above the 3σ upper limit of 0.285 mJy (Ravi et al. 2017), which requires α ≳ 2. The reason for this inconsistency is likely to be the omission of beam-size corrections by Oka et al. (2017). Although the synthesized beam areas of their continuum images are 1.74 arcsec² and 0.74 arcsec² at 230 GHz and 265 GHz, respectively, they measured the fluxes by summing the pixel values inside the beam of the 265 GHz data for both frequencies; hence, the effective beam area for the flux measurement at 230 GHz is approximately twice that at 265 GHz. This leads to an underestimation of α since MM1 is not an ideal point source but includes non-negligible background emission.

We obtained a steeper spectral index from the same data as that used by Oka et al. (2017) by applying a beam-size correction. Figure 11 shows the 230 and 265 GHz continuum images of MM1, which we constructed using the uv data within the spatial frequency range of 20–1500 kλ to correct for the differences in uv coverage. The synthesized beam sizes are 144 × 116 and 157 × 100 at 230 GHz and 265 GHz, respectively. We further smoothed the images so that they both have an effective beam size of 2'' FWHM, since the center positions of the source at the two frequencies differ by approximately half the original beam. The corrected flux densities are 8.6 ± 0.2 mJy and 13.2 ± 0.4 mJy at 230 GHz and 265 GHz, respectively, which gives a spectral index of α = 3.0 ± 0.3. Although our measurement is contaminated by the background emission, the net spectral index of MM1 should not differ significantly from that obtained here, since measurement of the extended emission (indicated by the broken-lined ellipse in Figure 11) gives a similar value of α = 2.5 ± 0.4.

Zoom In Zoom Out Reset image size

Although it is difficult to identify the nature of MM1 from the ALMA data alone, the steep value of α ∼ 3 is typical for thermal dust emission. This supports the argument of Ravi et al. (2017) that the source could be emission from warm dust in circumstellar material, whereas interpretations as optically thin synchrotron emission from a relativistic jet or wind (Ravi et al. 2017) are not favored. The hypothesis of synchrotron emission from an IMBH (Oka et al. 2017) might be marginally consistent with our result, as self-absorbed synchrotron emission can have values of α up to 2.5, but assuming such an exotic object would not be reasonable when the value of α is common for thermal dust emission.

Zoom In Zoom Out Reset image size

The thermal origin for MM1 is also suggested by the detection of a molecular-line counterpart. Figure 11(c) presents the HCN J = 3–2 integrated intensity for the v_LSR range from 10 to 20 km s⁻¹, which shows a compact source at the position of MM1. As other velocity channels lack intensity peaks at this position, this 15 km s⁻¹ source is most likely a molecular-line counterpart to MM1. The systemic LSR velocity of 15 km s⁻¹ suggests that the source is associated with the foreground cloud G359.62−0.24, for which the distance has been measured to be ${3.1}_{-0.9}^{+2.2}$ kpc through parallax measurements using a water maser source (Iwata et al. 2017).

To conclude, the signature of a gravitational interaction with an IMBH is not confirmed in the reanalyzed data. The features previously reported as evidence for the IMBH scenario are most likely due to severe beam smearing in the single-dish images, blurring of the spectra introduced by errors in the data reduction, and insufficient correction for the difference in the beam sizes between the two measurement frequency bands.

The dendrogram analysis has shown that the characteristic broad-lined emission from CO−0.4 does not originate from one coherent kinematical feature, but instead is a superposition of multiple velocity components. The −40 and −80 km s⁻¹ components have narrow velocity widths, Δv = 10 km s⁻¹, whereas the −60 km s⁻¹ feature consists of moderately broad emission with Δv = 25–50 km s⁻¹. The moment analysis described in Sections 3.2.2 and 3.2.3 has revealed that these components are smoothly connected at knots A and B on the −80 km s⁻¹ filament in PPV space; the small-scale distributions of the −80 and −40 km s⁻¹ components inside the knots exhibit a spatial anti-correlation on the plane of the sky, but these two primary velocity components are bridged by the moderately broad emissions that extend along the boundary between them. Such a configuration, consisting of two narrow line components and a broad emission bridging them, is a signpost of a CCC system (Haworth et al. 2015b; Torii et al. 2017). In this CCC hypothesis, Knots A and B are interpreted as the positions at which the −80 km s⁻¹ filament collides with the −40 km s⁻¹ cloud.

The bridging emission can be seen more clearly by comparing the ALMA HCN data with the single-dish ¹³CO J = 3–2 map (Tanaka et al. 2014), in which spatially extended or low-density components unobservable with ALMA are visible. Figure 12 shows the position–velocity diagrams of the single-dish ¹³CO J = 3–2 and the ALMA HCN J = 3–2 data along a cut approximately parallel to the −80 km s⁻¹ filament, as indicated in Figure 12(a). The ¹³CO map has counterparts to the −40 and −80 km s⁻¹ components of the ALMA data, but lacks the −60 km s⁻¹ component. The ¹³CO line is generally optically thin and primarily traces the mass distribution of the gas, whereas the HCN J = 3–2 is sensitive to gas with a low column density, but higher temperature and gas volume density are required to excite it. Therefore, the absence of bridging emission in the ¹³CO map indicates that CO−0.4 is separated into two distinct dynamical components represented by the −40 and −80 km s⁻¹ components. Superposition of the single-dish and ALMA data presented in Figure 12(b) shows clearly that the broad-lined HCN emission bridges the velocity gaps between the two primary components. This bridging emission is minor in mass, but has a highly elevated temperature and density; it can be interpreted as a thin layer of turbulent gas created by the collision of the two clouds.

Zoom In Zoom Out Reset image size

This CCC hypothesis is strengthened by the detection of the V-shaped PV pattern across the width of the −80 km s⁻¹ filament. Theoretical studies of CCC systems of two nonidentical clouds (Habe & Ohta 1992; Anathpindika 2010; Takahira et al. 2014) predict the formation of bow shocks inside the larger cloud. If the same model is applicable to a collision between a narrow filament and an extended cloud, the shocked region should have a V-shaped PV pattern when the collision interface is observed in the face-on geometry. We show a schematic illustration of such a collision of the −80 km s⁻¹ filament and the −40 km s⁻¹ cloud in Figure 13. The central portion of the −40 km s⁻¹ cloud is dragged along by the plunging −80 km s⁻¹ filament and the bow shock, while the motion of the outer part of the −40 km s⁻¹ cloud remains less affected. As the intrinsic velocity widths in the pre-collision clouds are smaller than the line-of-sight collision velocity, this gas kinematics is expected to be observed as a V-shaped velocity pattern when viewed along the collision axis (Takahira et al. 2014; Haworth et al. 2015b; Torii et al. 2017). Similar models have been proposed for the formation of half cavities in the Brick cloud, the Sgr B2 complex, and the 50 km s⁻¹ cloud (Higuchi et al. 2014; Tsuboi et al. 2015a, 2015b).

Zoom In Zoom Out Reset image size

The CCC model also explains the mechanism responsible for heating the −80 km s⁻¹ filament to the observed high temperature. By balancing the heating produced by dissipation of the velocity dispersion of 10 km s⁻¹ in the knots against the radiative cooling by gas and dust (Ao et al. 2013), we obtain 220 K for the gas temperature, where we used velocity gradient dv/dr = 2 × 10² km s⁻¹ pc⁻¹, filament width L = 0.1 pc, and ${n}_{{{\rm{H}}}_{2}}$ = 10⁶ cm⁻³ for the calculation. This temperature is within the parameter range obtained from the non-LTE analysis of the multitransition methanol lines.

Ravi et al. (2017) proposed a CCC hypothesis that is different from ours, in which the infrared dark cloud G359.62−0.24 is assumed to have fallen from the Galactic halo and collided with CO−0.4. However, PV patterns indicative of physical interactions between G359.62−0.24 and CO−0.4 are not identified in the ALMA HCN J = 3–2 data (Figure 1(a)) nor in the single-dish ¹³CO data (Figure 12). The parallax measurement toward a water maser source in G359.62−0.24 (Iwata et al. 2017) suggests that the source is most likely in the foreground spiral arm region.

Another hypothesis by Yalinewich & Beniamini (2017), assuming SN–cloud interactions as the origin of the HVCCs, does not conflict with our CCC scenario. As we have shown in a previous paper (Tanaka et al. 2014), CO−0.4 is located on the rim of an expanding-shell-like structure of dense molecular gas with diffuse radio continuum emission, which could be understood as an SN-driven shell. This expanding shell is a plausible candidate of the trigger of the CCC event at CO−0.4.

The large-scale CO surveys toward the CMZ have detected approximately 100 HVCCs and HVCC-like broad-line features (Oka et al. 2012; Tokuyama et al. 2017). Signatures of CCC similar to those found in CO−0.4 have been detected in another archetypical HVCC, CO−0.30–0.07 (Tanaka et al. 2015). If a significant fraction of the HVCCs in the CMZ are CCC sites similar to those archetypical HVCCs, it would indicate a high CCC frequency in the CMZ, since HVCCs are rather short-lived features. By naively assuming that every HVCC indicates one CCC event, the CCC frequency can be estimated from the number of HVCCs, and the duration of the HVCC phase, τ_HVCC. On ignoring radiative disruption induced by star formation (SF) activity (Haworth et al. 2015a), we obtain τ_HVCC ∼ r/σ_v = 7 × 10⁴ year, where r = 1 pc and σ_v = 15 km s⁻¹ are typical size and velocity dispersion for an HVCC, respectively. Given that a typical cloud mass in the CMZ is ∼10⁴ M_⊙ (Miyazaki & Tsuboi 2000) and that the total cloud mass of the CMZ is 3 × 10⁷ M_⊙ (Molinari et al. 2011), the CCC frequency is estimated to be 3 per cloud in one orbital period (6 Myr; Sofue 2013). The actual CCC frequency could be somewhat higher, since whether a CCC system is detectable as a broad emission feature may depend on the viewing angle; Haworth et al. (2015b) estimate the detection efficiency to be 20%–30% due to this effect. In total, the CCC rate is roughly estimated to be an order of 1–10 per cloud per orbital period, if CCC is the primary origin of the HVCCs. This is in good agreement with the theoretically predicted CCC frequency: a few to 15 times per orbital period (Tan 2000; Tasker & Tan 2009; Dobbs et al. 2015). Although these theoretical calculations are not necessarily tuned for the GC environment (Dobbs et al. 2015), it is unlikely that the CCC frequency in the CMZ is lower than in the Galactic disk. The high volume filling factor of dense gas, fast orbital velocities, and frequent SN explosions may further increase the CCC frequency.

CCC-triggered SF is among the popular theories for the formation of the massive clusters in the CMZ (Hasegawa et al. 1994; Stolte et al. 2008). Our analysis has shown that ${n}_{{{\rm{H}}}_{2}}$ in the post-shock region is ≳10⁶ cm⁻³, which is approximately two orders of magnitude higher than the average ${n}_{{{\rm{H}}}_{2}}$ in CMZ clouds. This indicates that CCC is actually able to compress the gas to densities typical of star-forming cores during the dynamical time of a collision. However, direct evidence has not been obtained for current SF in CO−0.4; the compact continuum source MM1 is likely to be in the foreground Galactic disk cloud and not associated with CO−0.4. Radio continuum sources indicative of H ii regions are not detected in existing centimeter-wavelength images (e.g., Liszt & Spiker 1995). Although the bright methanol spots on the −80 km s⁻¹ filament are similar to hot cores, their high temperature and rich abundances of complex organic molecules (COMs) could also be an immediate result of a shock passage. Indeed, the fractional methanol abundance of 10⁻⁷ is approximately equal to the average value for non-star-forming clouds in the CMZ (Requena-Torres et al. 2006), where the high COM abundance is thought to be maintained by mechanical sputtering of dust mantles, not by thermal desorption from hot cores.

One explanation for the lack of SF signatures may be that the cloud is at an early phase in a CCC-triggered SF process, in which SF has not yet been activated. Alternatively, it is possible that the collision is nonproductive. CCCs may stabilize clouds by enhancing turbulent pressure, or they may even destroy the clouds if the ram pressure of the collision exceeds the gravitational binding pressure (Habe & Ohta 1992; Tan 2000; Dobbs et al. 2011; Johnston et al. 2014; Kruijssen et al. 2014; Rathborne et al. 2014). The general absence of active SF in the HVCCs and the CCC candidate regions—except for the Sgr B2 complex (e.g., Oka et al. 2012; Rathborne et al. 2014; Higuchi et al. 2014; Tanaka et al.2015)—may indicate that CCCs in the CMZ are predominantly in such a nonproductive regime. Considering that every CMZ cloud may experience more than one CCC during its lifetime, nonproductive CCCs may be a non-negligible mechanism for suppressing the SF efficiency in the GC region (Kauffmann et al. 2017).