Evidence of a Cloud–Cloud Collision from Overshooting Gas in the Galactic Center

We first estimate the amount of mass in G5 by using the Strong et al. (1988) CO-to-H₂ conversion factor of $2.3\times {10}^{20}\,{\mathrm{cm}}^{-2}\,{({\rm{K}}\,\mathrm{km}\,{{\rm{s}}}^{-1})}^{-1}$. This is the most commonly used Galactic CO-to-H₂ X-factor (X_CO). We use an integrated-intensity map of ¹²CO J = 2 → 1, multiply it by 0.8 to account for the difference in intensity between ¹²CO J = 1 → 0 and ¹²CO J = 2 → 1 (Leroy et al. 2009), and then multiply it by the X_CO to create a column density map of H₂ for Field 2. We use the molecular weight per hydrogen molecule ${\mu }_{{{\rm{H}}}_{2}}=2.8$ (Kauffmann et al. 2008) to calculate the mass in each pixel, and then summed over all of the pixels in Field 2 to give a mass estimate for the field. We report the estimated mass in Table 2 for this and all subsequent methods.

Table 2. Mass Estimate of Field 2

Method	Mass Estimate	Column Density (H2)	N(H2)/∫Iu (12CO) du	CMZ Inflow	Assumed Ratio
	(M⊙ × 105)	(cm−2 × 1022)	(${\mathrm{cm}}^{-2}\,{({\rm{K}}\,\mathrm{km}\,{{\rm{s}}}^{-1})}^{-1}$ × 1020)	(M⊙ yr−1)
X-factor a , b	2.23 ± 0.86	6.52 ± 2.53	2.3 ± 0.3	0.8 ± 0.6	⋯
SED fit	0.13 ± 0.02	0.43 ± 0.07	0.15 ± 0.07	0.05 ± 0.03	⋯
Max SED fit	0.32 ± 0.22	1.04 ± 0.72	0.37 ± 0.3	0.13 ± 0.11	⋯
PPMAP	0.28 ± 0.01	0.87 ± 0.04	0.31 ± 0.14	0.11 ± 0.05	⋯
LTE 12CO c	0.1 ± 0.04	0.3 ± 0.11	0.11 ± 0.06	0.04 ± 0.02	⋯
LTE 13CO	0.16 ± 0.08	0.48 ± 0.24	0.17 ± 0.11	0.06 ± 0.04	12C/13C = 25
LTE C18O	0.14 ± 0.08	0.43 ± 0.24	0.15 ± 0.11	0.05 ± 0.04	16O/18O = 250
LTE 12CO: τ correction	0.17 ± 0.06	0.51 ± 0.18	0.18 ± 0.1	0.06 ± 0.04	12C/13C = 25
LTE 12CO: τ correction	0.25 ± 0.09	0.77 ± 0.28	0.27 ± 0.16	0.1 ± 0.06	12C/13C = 40
LTE 12CO: τ correction	0.33 ± 0.12	1.01 ± 0.37	0.36 ± 0.21	0.12 ± 0.07	12C/13C = 53
LTE 12CO: τ correction	0.48 ± 0.17	1.45 ± 0.53	0.51 ± 0.3	0.18 ± 0.11	12C/13C = 77
LTE 12CO: τ correction	0.55 ± 0.2	1.67 ± 0.61	0.59 ± 0.34	0.21 ± 0.12	12C/13C = 89

Notes.

^a Assumed X_CO from Strong et al. (1988) ^b Mass inflow rate from Hatchfield et al. (2021) ^c All LTE masses assume CO/H₂ = 10⁻⁴

Download table as:  ASCIITypeset image

We next estimate the amount of mass by creating a dust emissivity spectral energy distribution (SED) of Field 2, and then we fit the SED with a modified blackbody. Appendix Figure 19 shows that the dust emission approximately matches the NH₃ gas emission contours. We used Herschel SPIRE and PACS data (Molinari et al. 2016) for wavelengths 70, 160, 250, 350, and 500 μm, ATLASGAL for 850 μm (Schuller et al. 2009), and BGPS for 1.1 mm (Ginsburg et al. 2013). We placed a rectangular aperture over Field 2 with a size of 126 by 27 for an area of 337 squared, placed a rectangular annulus around Field 2 with outer heights and widths twice that of Field 2, and subtracted the 10th percentile of the annulus from the aperture to remove background emission. We then summed over the annulus to find the background subtracted flux of Field 2. We found the errors of each measurement by taking the quadrature sum of the statistical uncertainty of the measurements and the inherent uncertainty due to flux calibration. The measured values are shown in Table 3. We then fit the values with a modified blackbody function. Figure 13 shows the dust SED of Field 2. The dust opacity is assumed to be defined by a continuous function of frequency defined by $\kappa ={\kappa }_{0}{\left(\tfrac{\nu }{{\nu }_{0}}\right)}^{\beta }$. The modified blackbody used a value of κ₀ = 4.0 cm² g⁻¹ at 505 GHz (Battersby et al. 2011) and a gas-to-dust ratio of 100. The fit for the modified blackbody resulted in a dust temperature of 18.1 K ± 1.2 K, β of 1.8 ± 0.3 , and H₂ column density of 4.34 × 10²¹ cm⁻² ± 0.68 × 10²¹ cm⁻². We expect that G5 has a dust temperature decoupled from the gas temperature, as the majority of the cloud has a lower volume density than the 10⁶ cm⁻³ needed for collisional equilibrium (Clark et al. 2013).

Zoom In Zoom Out Reset image size

We also report an alternate fit of the dust SED excluding the lowest wavelengths from the PACS survey. The modified blackbody fit of the remaining data resulted in a dust temperature of 11.9 K ± 3.4 K, β of 2.8 ± 0.8 , and H₂ column density of 1.04 × 10²² cm⁻² ± 7.2 × 10²¹ cm⁻². This alternate fit of the data resulted in a column density measurement closer to that using the Strong et al. (1988) X_CO measurement, but uses less of the available data.

We then estimated the mass of G5's Field 2 using PPMAP (Marsh et al. 2017). PPMAP uses Hi-GAL data, Herschel PACS, and SPIRE, to measure the dust column density by fitting a dust SED, using a factor of 100 dust-to-gas fraction. PPMAP assumes a dust opacity value of κ₀ = 0.1 cm² g⁻¹ at 300 μm, and β = 2.0. We made a cutout of Field 2 from the PPMAP column density map, multiplied each pixel by the pixel area, and then summed over the cutout to find the mass.

We estimated the total mass of G5 using PPMAP. We made a mask using the NH₃ (3,3) data from the Mopra HOPS Survey (Walsh et al. 2011; Purcell et al. 2012; Longmore et al. 2017) to select only areas of the map with NH₃ emission. We identify those boundaries as the extent of G5, as shown in Figure 3. We then applied the mask to PPMAP data, resulting in a total mass measurement of 3.87 × 10⁵ M_⊙ ± 0.45 × 10⁵ M_⊙. This mass measurement does not account for dust along the same line of sight, such as from the overlapping H ii region, and so is an upper estimate of the mass using this technique.

We estimated the mass of G5 using CO and its isotopologues by assuming they are in LTE. We note that if the CO lines are subthermally excited, then the real column density of CO found through this method is likely higher, but we find that this is unlikely. We used Equation (79) from Mangum & Shirley (2015)

$$\begin{eqnarray}\begin{array}{rcl}{N}_{\mathrm{tot}}^{\mathrm{thin}} & = & \left(\displaystyle \frac{3h}{8{\pi }^{3}S{\mu }^{2}{R}_{i}}\right)\left(\displaystyle \frac{{Q}_{\mathrm{rot}}}{{g}_{u}}\right)\displaystyle \frac{\exp \left(\tfrac{{E}_{u}}{{k}_{{\rm{B}}}{T}_{\mathrm{ex}}}\right)}{\exp \left(\tfrac{h\nu }{{k}_{{\rm{B}}}{T}_{\mathrm{ex}}}\right)-1}\\ & & \times \,\displaystyle \int \displaystyle \frac{{T}_{R}d\nu }{f\left({J}_{\nu }({T}_{\mathrm{ex}})-{J}_{\nu }({T}_{\mathrm{bg}})\right)}\end{array}\end{eqnarray} \tag{ 2 }$$

to calculate the column density where h is Planck's constant, Q_rot is the rotational partition function, g_u is the degeneracy, E_u is the energy of the upper energy level, k_B is the Boltzmann constant, ν is the frequency of the transition, the sum of relative intensities R_i = 1 for ΔJ = 1 transitions, f is the filling factor assumed to be 1, T_ex is the excitation temperature, ∫T_R d ν is the integrated intensity, T_bg is the cosmic microwave background, J_ν (T) is the Planck function, the line strength $S=\tfrac{{J}_{u}}{2{J}_{u}+1}$ for linear molecules where J_u is the upper energy level, and the value for the molecular electric dipole moment (μ) is from the Jet Propulsion Laboratory Molecular Spectroscopy database and spectral line catalog (Pickett et al. 1998). We calculated the column densities for ¹²CO, ¹³CO, and C¹⁸O. This equation assumes that the molecule being measured is optically thin.

While we can assume that ¹³CO and C¹⁸O are optically thin in G5, we cannot assume the same for ¹²CO. To remedy this, we estimate the optical depth of ¹²CO. We can estimate the optical depth of CO using Equation (3),

$$\begin{eqnarray}&&\displaystyle \frac{{I}_{\nu }{(}^{12}\mathrm{CO})}{{I}_{\nu }{(}^{13}\mathrm{CO})}=\displaystyle \frac{1-{e}^{-{\tau }_{12}}}{1-{e}^{-{\tau }_{12}/{R}_{C}}},{R}_{C}\,=\,\displaystyle \frac{{}^{12}{\rm{C}}}{{}^{13}{\rm{C}}},\end{eqnarray} \tag{ 3 }$$

where I_ν is the intensity of the lines and τ₁₂ is the opacity of ¹²CO. The isotope abundance ratio ¹²C/¹³C is lower in the Galactic center than in the disk and is thought to increase radially outward from the Galactic center (Langer & Penzias 1990) due to the Galaxy forming from the inside out (Chiappini et al. 2001; Pilkington et al. 2012). ¹²C is formed by He burning in massive stars on short timescales (Timmes et al. 1995), while ¹³C is formed from ¹²C seed nuclei in the CNO cycle of evolved low- and intermediate-mass stars (Henkel et al. 1994).

We estimate the opacity of ¹²CO and the mass of Field 2 using Galactic abundances of ¹²C/¹³C. We use a root-finding algorithm to solve to solve Equation (3) for the opacity of ¹²CO τ₁₂. We spectrally and spatially reproject the ¹²CO cube to the same spectral axis as the ¹³CO cube and then divide the two for a ratio cube of ¹²C/¹³C. We then use root finding to solve Equation (3) to estimate the optical depth of every voxel in the cube. We find separate optical depth cubes for the Galactic values of ¹²C/¹³C reported in Henkel et al. (1985), Wilson & Rood (1994), and Riquelme et al. (2010). Bars radially mix gas, so the ¹²C/¹³C ratio may not be much different from the Galactic center. Riquelme et al. (2010) measure conflicting ¹²C/¹³C ratios at G5, resulting in one measurement higher than 70 for an 87 km s⁻¹ component using H¹²CO⁺/H¹³CO⁺, and other measurements as low as 10–45 using other components and line ratios. We set the opacity values to 0 for each voxel with a value of ¹²CO or ¹³CO less than 1 standard deviation, and we set the opacity to 0 for ¹²CO/¹³CO ratios less than 1 or greater than the assumed ¹²C/¹³C. We show the distribution of calculated opacity values for each voxel in the opacity cube assuming ¹²C/¹³C = 40 in Figure 14.

Zoom In Zoom Out Reset image size

We then apply each estimated optical depth cube to the ¹²CO cube using the linear relationship between the integrated intensity and total column density using Equation (86) from Mangum & Shirley (2015),

$$\begin{eqnarray}&&{N}_{\mathrm{tot}}={N}_{\mathrm{tot}}^{\mathrm{thin}}\displaystyle \frac{\tau }{1-\exp (-\tau )},\end{eqnarray} \tag{ 4 }$$

where N_tot is the total column density taking into account optical depth, τ is the optical depth of ¹²CO, and N_tot ^thin is the estimated LTE column density assuming the line is optically thin from Equation (3). We then take the integrated intensity of the altered cubes and solve for the column density and mass of the field.

Finally, we estimate the mass of G5 using the optically thin CO isotopologues ¹³CO and C¹⁸O. The average column densities and the calculated X_CO for each above method are included in Table 2. The calculated X_CO uses the average velocity-integrated intensity of ¹²CO over Field 2 adjusted to account for the difference intensity between ¹²CO J = 1 → 0 and ¹²CO J = 2 → 1 (Leroy et al. 2009), 283.46 K km s⁻¹.

We begin to see evidence of a cloud–cloud collision in Figure 6, where integrating the intensity over two different ranges of velocities shows distinctly different structures. The velocity map in Figure 7 better displays the vastly different velocities of G5a and G5b, and the gradient between them that starts in the middle of Field 2 as red transitions to white where there is a peak in emission and then swaps fully to blue within a few arcminutes. This alone does not mean that there is a cloud–cloud collision. Molecular clouds may overlap along a line of sight. The velocity dispersion map in Figure 7 finds an extremely wide line width in the same region of the gradient in the velocity map. However, the elevated dispersion is due to overlapping spectral features, as shown in the second panel of Figure 4. Further evidence is needed to confirm the cloud–cloud collision. To that end, we view the region of the suspected collision in PV space.

Figure 8 shows a PV diagram of Field 2 where the offset is in Galactic longitude. The high-velocity component in the PV diagram is clearly the same high-velocity feature in the velocity field map in Figure 7 identified as G5a, and the low-velocity component is G5b, the low-velocity spectral feature in blue in the same map. Visible in PV space is a wide spectral feature stretching between G5a and G5b. This spectral feature is narrow in position space, as it is only about an arcminute across at its widest, but it has a length of ∼100 km s⁻¹. This feature is a velocity bridge.

A velocity bridge is a feature in a PV diagram that is wide in velocity space but relatively narrow in position space, and connects two features. A velocity bridge indicates that the two clouds are interacting, instead of being coincidentally along the same line of sight where they overlap (Haworth et al. 2015a, 2015b). As shown in Figure 8, a velocity bridge is clearly visible in the center of the PV diagram of Field 2. When two clouds collide, only a small amount of mass is involved in the collision at a time. Gas fills the entire space between the clouds, with different velocities in the bridge corresponding to different amounts of either cloud. Velocity bridges tend to last as long as the crossing time of the cloud, but for streams of gas flowing along bar lanes, the velocity bridge may remain for longer times. The simulation from Sormani et al. (2019) suggests that streams of gas overshooting the CMZ collide with material along bar lanes on the other side of the Galaxy certain at certain locations with vastly different line-of-sight velocities, a cartoon representation of which is shown in Figure 15. This simulation resembles the observed velocity bridge and the ∼100 km s⁻¹ velocity difference between G5a and G5b where they collide.

A cloud complex that can be compared to G5 is G1.3, which has a velocity bridge identified in CS J = 2 → 1 (Busch et al. 2022).

The gas temperatures of G5a and G5b shown in Figure 10 are comparable to typical Galactic center temperatures at an average of 60 K, but less than the more extreme temperatures (Ginsburg et al. 2016; Krieger et al. 2017). The temperatures of G5a and G5b are comparable to each other, but a K-S test of the data shows that the distribution of temperatures from the clouds is unlikely to be the same.

The gas temperature measured using H₂CO line ratios is much higher than the 18.1 K measured from the dust emissivity SED model of Field 2 in Figure 13. The gas and dust temperatures are decoupled, meaning that the gas is not efficiently cooled by interactions with the dust, likely due to G5 having a volume density too low to couple with dust (Clark et al. 2013; Ginsburg et al. 2016).

Akhter et al. (2021) found the gas rotational temperature of G5 using a transition line ratio of NH₃ (2,2)/(1,1), finding gas temperatures between 60 and 100 K. These temperatures are consistent with the range of gas temperatures found with H₂CO line ratios.

Zoom In Zoom Out Reset image size

Heating in molecular clouds has several different causes radiation, cosmic rays, and shocks.

First, we rule out radiation as a heating mechanism in G5. We looked at Herschel and Spitzer data sets of G5, the three color images shown in Figure 16. While Figure 16 shows the presence of a H ii region at the top of Field 1, that H ii region is not associated with G5. Wink et al. (1983) found the velocity of the H ii region using H₇₆ α, resulting in a measurement of 28.3 km s⁻¹ ± 0.9 km s⁻¹ with an FWHM of 20.8 km s⁻¹ ± 2.0 km s⁻¹. While CO emission from the H ii region is observed in Field 1, the emission is at a different velocity compared to the rest of the material associated with G5 nearby. There are no features that resemble stars interacting with gas in G5. There is no evidence of feedback from star formation, so we do not expect radiation to be the primary source of heating in G5.

Zoom In Zoom Out Reset image size

Cosmic rays are also unlikely sources of heating in G5. While the Galactic center has a higher cosmic-ray ionization rate (CRIR) than the Galactic disk (CRIR∼10⁻¹⁴ s⁻¹; Yusef-Zadeh et al. 2007; Indriolo et al. 2015; Oka et al. 2019), there is no evidence for such elevated CRIR in the Galactic bar. Unless there was some local source of cosmic rays, we would not expect cosmic rays to heat the molecular gas to Galactic center temperatures. We also do not see evidence of any such source, such as a supernova remnant, near G5.

We find that the most likely heating mechanism in G5 is shocks. Since the clouds are colliding, we expect shocks to be a significant heating mechanism in the molecular gas. Ginsburg et al. (2016) identify turbulent dissipation (shocks) as one of the key driving processes behind the high gas temperatures measured in the Galactic center. We measure very wide line widths in G5, as shown in Figure 7 and the PV diagrams in Figures 8 and 9. The collision between G5a and G5b causes an increase in the kinetic temperature as gas from vastly different velocities collides and mixes together. While the collision may explain the high temperatures found near the location of the cloud–cloud collision, the entirety of the clouds, not just the bridge feature, are warm. Sormani (2021) finds that a cloud–cloud collision such as the one found in G5 would not happen as one large collision of gas decelerating from 200 to 0 km s⁻¹, but as a series of much weaker shocks that allow the gas to cool as it is being shocked over time. Another form of shock heating is tidal shear stress heating, which would be experienced by material in bar lanes as it is stretched and manipulated by the gravity of the Galaxy's bar potential. Tidal stress may also cause the SiO J = 5 → 4 observed in G5.

We expect that shocks from tidal shear and the cloud collision cause turbulence within G5, warming it up to the temperatures we observe. G5 has likely been the location of many previous collisions, as evidenced by the spur feature in the PV diagrams (Appendix Figure 18), which would heat the cloud repeatedly over time. Ginsburg et al. (2016) show in their Appendix F that line widths of ∼20 km s⁻¹ and temperatures from 40–60 K match well with DESPOTIC models for turbulent heating at a volume density of 10⁵ cm⁻³. Shocks are able to explain the observed gas temperature.

The cloud–cloud collision in G5 happens between gas with a velocity difference of over 100 km s⁻¹ along the line of sight. As the clouds collide on an angle relative to the line of sight, we only see a component of the velocity of the collision, so the true velocity of the collision is even higher. A large amount of energy must be involved in this collision.

Using a lower mass estimate of 10⁴ M_⊙ from Section 4.5 for the mass involved in the collision, with a velocity difference of over 100 km s⁻¹, the collision would produce ∼10⁵¹ erg of kinetic energy, roughly equivalent to the mechanical energy from a supernova. We expect heating of up to 10⁵ K from the J shock due to the collision between the clouds, which would produce weak X-rays and ionize atoms in the region of the shock. Such a large collision would disassociate molecules, but we detect CO in the velocity bridge, the gas directly involved in the collision. There are a few options for where the energy of the collision is going. The first is that there is one big shock that is heating the gas up to 10⁵ K, but the molecules reform soon after the shock. The second is that the collision has only just begun and involves only the less dense, outer layers of the cloud. The third possibility is that molecules are not being destroyed. Instead of one big shock, the collision happens as a series of smaller shocks, which would heat up the gas much less (Sormani 2021).

Some of the energy goes into increasing the thermal energy of the clouds involved in the collision. In Section 5.1.2, we measure the kinetic temperature of the cloud's gas at approximately 60 K. The internal kinetic energy of the clouds can be estimated using the velocity dispersion, and is ∼10⁵⁰ erg. The internal energy of G5 is 10 times less than the collisional energy.

Much of the collision energy is radiated away by dust. The length of the collision is approximately the crossing time. Assuming the crossing length is 20 pc and the clouds collide with a constant velocity of 100 km s⁻¹, the crossing time is 200 kyr. The energy rate for ${M}_{\mathrm{cloud}}{v}_{\left(\mathrm{collision}\right)}^{2}\sim {10}^{51}\,\mathrm{erg}$ over 200 kyr is 10⁵ L_⊙. The dust luminosity can be estimated using the dust SED fit from Section 4.5.2. The luminosity of dust with a temperature of 18 K in an area of 16 pc² is 2 × 10⁵ L_⊙. We assume that the dust in G5 has a temperature in equilibrium, although the temperature we measure is higher than the solar neighborhood. We do not know what the interstellar radiation field or cosmic ray rate is at G5, but if they are stronger then we expect them to affect the equilibrium temperature of the dust to some extent. The energy generated by the collision is approximately the same as the amount of energy radiated by the dust.

The G5 cloud–cloud collision should produce a strong signal in shock tracers such as SiO. SiO is a strong shock tracer, often seen in strongly shocked regions of the ISM such as in molecular outflows (Schilke et al. 1997). We lack a detection of SiO J = 5 → 4 in the gas of the velocity bridge, which is directly involved in the collision. In Section 4.4, we simulate the line in non-LTE using DESPOTIC, finding that a low volume density in the velocity bridge material would prevent the detection of SiO J = 5 → 4 in the velocity bridge. The excitation we see in the main bodies of G5a and G5b is consistent with the presence of shocks in the clouds, but there is no evidence of a high-velocity, strong shock. A possible cause for the lack of evidence of a strong shock from the collision is that G5a and G5b are colliding not as monolithic masses, but as a series of weaker shocks, allowing the heat to dissipate more efficiently than two large masses colliding at once (Sormani 2021). These weak shocks would not destroy the dust grains as efficiently as a strong shock, resulting in a smaller SiO abundance than expected from the collision.

Initial estimates of the optical depth of G5 suggest that the gas flowing into the CMZ is not very optically thick. The Strong et al. (1988) X_CO relies on the CO being optically thick, but smaller optical depths measured in Figure 14 would result in an overestimation of the mass when using the X_CO. Comparing mass estimates using the X_CO and dust mass estimates from Table 2 suggests that the mass of G5 is ∼10–20× overestimated by the accepted X_CO. Liszt (2006) measured a mass of 6.1 × 10⁶ M_⊙ for G5 over an area of 180 × 60 pc using observed CO J = 1 → 0 and an ${{\rm{X}}}_{\mathrm{CO}}\,=2\times {10}^{20}\,{\mathrm{cm}}^{-2}\,{({\rm{K}}\,\mathrm{km}\,{{\rm{s}}}^{-1})}^{-1}$. We measured 3.87 × 10⁵ M_⊙ ± 0.45 × 10⁵ M_⊙ for the total mass of G5 using PPMAP column density measurements (Marsh et al. 2017), for a 16× difference between the two measurements. The consequences of this difference affect the results of Sormani & Barnes (2019) and other Galactic mass flow estimates, which overestimate the amount of the CMZ mass inflow rate.

We measure $1.5\times {10}^{19}\,{\text{}}N{({{\rm{H}}}_{2})\,{\mathrm{cm}}^{-2}({\rm{K}}\,\mathrm{km}\,{{\rm{s}}}^{-1})}^{-1}$ as the X_CO in G5. Ferrière et al. (2007) find that X_CO near the Galactic center is of the order of $\sim {10}^{19}\,{\text{}}N{({{\rm{H}}}_{2})\,{\mathrm{cm}}^{-2}({\rm{K}}\,\mathrm{km}\,{{\rm{s}}}^{-1})}^{-1}$. We calculated that G5 is 1.33 kpc distance from the Galactic center, meaning that the lower value of X_CO extends along the Galactic bar. The lower opacity of ¹²CO in G5 is likely due to the wide velocity dispersion of the cloud. If bar dynamics make the opacity and hence the X_CO different, then there is a significant impact on bar mass inflow rates calculated using the standard Galactic X_CO. A caveat of our estimate is that G5 might be special due to being the site of a collision, so it may not fully represent all of the gas on the bar lanes.

Sormani & Barnes (2019) find that the mass inflow rate along the nearside bar lane is 1.2 M_⊙ yr⁻¹, along the farside is 1.5 M_⊙ yr⁻¹, for a combined total of 2.7 M_⊙ yr⁻¹ flowing along bar lanes toward the CMZ. Updated estimates find that not all of the gas in the bar lane accretes onto the CMZ immediately, as only about 30% of the gas flowing along bar lanes accretes onto the CMZ, accreting with a rate of only 0.8 M_⊙ yr⁻¹ (Hatchfield et al. 2021). These mass inflow rates were found using the Strong et al. (1988) X_CO of $2.3\,\times {10}^{20}\,{\mathrm{cm}}^{-2}{({\rm{K}}\,\mathrm{km}\,{{\rm{s}}}^{-1})}^{-1}$ to estimate the amount of mass flowing into the CMZ, but we find that accepted X_CO overestimates the amount of mass in G5 in Section 4.5.

Mass inflow rates using the measured X_CO in G5 are lower than previously measured rates. Table 2 includes estimates of the CMZ mass inflow rate adapted from measurements by Hatchfield et al. (2021), using the X_CO calculated using estimates of the column density and 283.5 K km s⁻¹ for the average measured integrated intensity of ¹²CO 1 → 0. Our calculated mass inflow rates are consistently less than the current value of 0.8 M_⊙ yr⁻¹ (Hatchfield et al. 2021). The calculated mass inflow rates are close to the observed star formation rate of the CMZ. The rates are also less than the mass flowing out of the CMZ through Fermi bubbles and other outflows of 0.5 M_⊙ yr⁻¹ (Bordoloi et al. 2017; Di Teodoro et al. 2018, 2020).

The most likely cause of the lower X_CO observed in G5 is due to the low optical depth of the region. Figure 14 shows the opacity values calculated for Field 2 assuming ¹²C/¹³C = 40. The velocity bridge, shown in green, contains gas immediately involved in the observed cloud–cloud collision and seems optically thin as its opacity values drop off after 1. The two colliding clouds G5a and G5b are only moderately optically thick, with their opacity distributions peaking at around 2 and 3, respectively. The line-of-sight clouds, assumed to be somewhere else in the Milky Way but along the same line of sight as G5, show much higher opacity values, very few being under τ = 1. The Strong et al. (1988) X_CO assumes that CO is optically thick, so if all of the gas along the bar has an optical depth like G5, then the accepted measured mass flow into the CMZ is an overestimation.

Lower values of X_CO have been observed in the bars and bar lanes of other spiral galaxies (e.g., Meier & Turner 2001; Bolatto et al. 2013; Teng et al. 2022, 2023; Sormani et al. 2023). Our results are consistent with and offer additional explanations for these results.

Akhter et al. (2021) suggest that G5 and B1 are an antisymmetric pair, meaning that they have the same position in Galactic coordinates but with the signs flipped. They claim that the two clouds, due to their similar angular distances from the Galactic center and ammonia parameters, are related and trace identical features of the Galactic bar. They report that the clouds host hot gas and have wide emission lines due to shock heating. They offer two possibilities for the identity of the clouds: (1) the clouds are at the leading edges of the Galactic bar, having possibly passed through the bar lane shocks or (2) the clouds are on the innermost X₁ orbit, and collide with the gas where the orbits become self-intersecting.

However, in our model, the antisymmetric position of G5 and B1 with respect to the Galactic center is likely a coincidence. Figure 2 shows a model of the geometry of the Galactic bar, where the solid green lines point to the two ends of the bar and the dark-blue-dashed line points to the Galactic center. As shown in Figure 2, the angle between the line of sight to the near end of the bar and to the Galactic center is larger than the angle between the far end of the bar and the line of sight to the Galactic center. The green-dashed line has the same angle to the line of sight to the Galactic center as the line of sight to the near end of the bar. As shown in the figure, the green-dashed line does not point toward a symmetric feature on the far end of the bar. The foreshortening of the closer end of the Galactic bar means that it will appear at a larger angle from the Galactic center than the farther end of the bar. Therefore, while they may both be along the Galactic bar, G5 and B1 cannot both trace the ends of the bar nor be symmetric features of the Galaxy unless the bar was not tilted. As shown in Figure 15, G5 is likely somewhere along the length of the nearside bar lane. Assuming that the cloud is on the bar, B1's geometry places it further out from the Galactic center than G5, perhaps at the end of the 135 km s⁻¹ arm (Fux 1999).

An alternative explanation for the EVFs (Sormani et al. 2019; Liszt 2006) along the Galactic plane is the footpoints of magnetic loops. Fukui et al. (2006) identified two loop-like structures extending vertically out of the Galactic plane, which were concluded to be material lifted out of the Galactic plane by magnetic buoyancy, analogous to solar loops caused by the Parker instability.

According to the model, in a strong enough magnetic field, a small perturbation in the field causes the magnetic field lines to pinch and suddenly lift material out of the plane of the Galaxy. Gravity pulls the material back down along the slope of the magnetic perturbation toward the Galactic plane, creating a loop. Where that material intersects the Galactic plane is thought to be the footpoint of the magnetic loop. Fukui et al. (2006) associate these footpoints with areas of warm, dense gas with wide velocity dispersion, exactly like EVFs. One of the footpoints of the observed loops is B1 (Figure 1), which has been compared to G5 due to their axisymmetric Galactic coordinates (Akhter et al. 2021) and NH₃ (3,3) emission features. G5 is one of the other hypothetical magnetic footpoints, but on the other side of the Galaxy from B1.

We find that it is unlikely that G5 is evidence of the footpoint of a magnetic loop. The positions of the clouds involved with the cloud–cloud collision at G5 are displaced in Galactic longitude rather than in Galactic latitude, as predicted by the magnetic loop footpoint model. The loop footpoint model suggests that material falls down onto the plane of the Galaxy, so that the collision would have a major component in the z-direction, or Galactic latitude. As shown by observations from Enokiya et al. (2021) and Torii et al. (2010), the footpoints of magnetic loops have cloud–cloud collisions associated with velocity bridges identified with a cut along the Galactic latitude. The cloud–cloud collision at G5 is distinctly different from these observed loops, as it is identified with a cut along the Galactic longitude, as shown by Figure 8, showing that the collision is likely happening within the plane of the Galaxy. G5 is not likely to be the footpoint of a magnetic loop, as the collision does not involve gas that has risen out of the plane of the Galaxy.

Additionally, Sormani et al. (2019) show that G5 is connected to bar lanes. The velocity of the G5 EVF stops exactly at the velocity of the bar lane, and G5 is spatially associated with bar lane gas. Bar lanes have no role in the Parker instability model, so the association between G5 and the bar lanes is better explained by the explanation that G5 is due to overshooting gas.

We also find it unlikely that the magnetic field strength at G5 is strong enough for the Parker instability to lift material out of the Galactic plane. The Galactic center has an estimated magnetic field strength of 100 μm to 1 mG (Henshaw et al. 2023). The reported magnetic field strength to induce the Parker instability is 150 μG, but G5 is not in the Galactic center and is not likely to have the necessary magnetic field strength. According to the Suzuki et al. (2015) model, the magnetic field at the Galactic radius of G5 is only ∼10 μG, which is too low for the Parker instability.