The Population of the Galactic Center Filaments: Position Angle Distribution Reveals a Degree-scale Collimated Outflow from Sgr A* along the Galactic Plane

Figure 2 shows three different representations of color-coded identified filaments with different PA ranges with respect to Galactic north (PA = 0°), traced by different color tables. Figure 2(a) shows the full range of filament PAs between 0° and 180° (or equivalently, −90° to +90°) with different colors. Color tables are used in order to distinguish between the filaments with positive (red) and negative (blue) PAs. The distribution suggests that the filaments oriented perpendicular to the Galactic plane tend to be long. On the other hand, filaments with PAs running parallel to the Galactic plane are short. Figure 2(b) shows the distribution of short filaments, with PAs being preferred along the Galactic plane 60° < PA < 120°. This distribution is more obvious at negative longitudes where short filaments with PA ∼ 120° appear blue and point toward the GC. Figure 2 (bottom panel) displays the filaments with −60° < PA < 60°, indicating that vertical filaments are dominated by long filaments at positive and negative latitudes. An extension of the long filaments toward the Galactic plane do not converge toward a single origin such as Sgr A* but to the inner degree of the GC.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Another way to characterize the distribution of filament PAs is by restricting their lengths rather than their PAs, as was described above. There is generally a continuum of filament lengths, and 66'' was chosen as a lower limit that eliminated most filaments associated with known thermal sources such as H ii regions. An even cleaner selection at 132'' was discussed, but the sample size starts to become small (Yusef-Zadeh et al. 2022a). This convenient division was selected using 60 pixels with a pixel size of 11 in the original MeerKAT image with a resolution of 4'' (Yusef-Zadeh et al. 2022a).

Figure 3 (top panel) shows a histogram of long filaments with L > 66'' as a function of PA, emphasizing long nonthermal filaments, and largely excluding thermal features. A Gaussian fit to this histogram peaks at −32 with 1σ error of 129. Figure 3 (bottom panel) shows a histogram of long filaments with restricted PAs ranging between −60° and 60°. We note that long filaments have a negative spectral index α, the spectral index consistent with being nonthermal. There are also a number of long filaments with α ∼ −0.15. These also are most likely associated with a network of filaments associated with the Radio Arc near l ∼ 0.2, which are known to have a flat spectral index (Yusef-Zadeh et al. 2022a and references therein).

Zoom In Zoom Out Reset image size

Another view of the PA distribution of the filaments is presented by plotting them as a function of Galactic longitude. The PA distribution of long filaments as a function of Galactic longitude is displayed in Figure 4 (top panel), whereas the same distribution only for short filaments is shown in Figure 4 (bottom panel). Each black dot represents the PA of a filament. If the filament PAs were randomly distributed, there would not be any concentration. We note a high concentration of PAs in two clusters, one near PA ∼ 20°, l ∼ 02 and the other near PA ∼ −30°, l ∼ –025. A higher concentration of filaments at l > 0° is due to a larger number of filaments associated with the Radio Arc near l ∼ 02. An additional concentration of short filaments is noted near l ∼ 04 with PAs along the Galactic plane, as demonstrated in Figure 4 (bottom panel).

Zoom In Zoom Out Reset image size

We also note that, similar to Figure 2 (bottom pnael), the two concentrations in Figure 4 (top panel), show that the PA of the long filaments are tilted away from the Galactic north. We note that at l < 0°, the filament PAs of the concentrations are more widely distributed than those at l > 0°. Figure 4 (top panel) shows that within the range 05 < l < − 075, long filaments generally are found with PA ∼ 0°, indicating a vertical orientation. For l > 05, the long filaments have random orientations. For l ∼ − 075, long filaments are loosely clustered around PA = ±90°.

One of the most striking aspects of our PA study of the GC filaments is the evidence for a distinct population of filaments with PA close to the Galactic plane. This distribution is orthogonal to the PA distribution of long filaments that are aligned close to the Galactic north−south orientation. Figure 5 (top panel) shows the PA distribution of short (L < 66'') filaments. The short filament PAs are modulated by two peaks near 70° and −10°. The filaments in the peak at PAs ∼ 70° with a width of ∼40° are distributed mainly within ∼±20° of the the Galactic plane, whereas the filaments that peak close to PAs ∼ −10° are oriented closer to the Galactic north−south direction, having a distribution similar to that of the long filaments. In some cases the filaments can only be reliably traced over partial segments of the filament due to limited signal-to-noise and confusion.

Zoom In Zoom Out Reset image size

Another characteristic that distinguishes short from long filaments is their spectral index. Figure 5 (bottom panel) shows a histogram of the spectral indices of short filaments L < 66'', restricted to angles within ∼30° of the Galactic plane (60° < PA < 120°). The majority of short filaments have spectral indices consistent with thermal emission though we cannot exclude that they could be nonthermal filaments similar to the filaments of the Radio Arc near ∼02. This histogram also indicates that some short filaments are nonthermal with a steep spectrum.

The distribution of PAs is also examined as a function of the Galactic PA (GPA) defined as $\arctan (l/b)$. 0° < GPA < 90° corresponds to the NE quadrant of the GC, and subsequent 90° intervals are the SE, SW, and NW quadrants. The black points represent the PA of short- and long-identified filaments in Figures 6(a), (b), respectively. The dark diagonal line shows the expected PA distribution along a radial direction following a PA, which is equal to its GPA. Filaments along these lines are oriented radially with respect to the GC. However, we note a concentration of points close to GPA =270° (close to the Galactic plane and west of Sgr A*) along the Galactic plane. There is a tendency that filament PAs in the third GPA < 270° and fourth quadrant GPA >270° display horizontal components with PAs ∼ 70° and ∼110°, respectively. This range of filament PAs follow the diagonal line that points in the direction of Sgr A*. The width of the group of filaments following the diagonal line is about 20° above and below the Galactic plane. These trends indicate that filament PAs both north and south of the Galactic plane tend to be slightly more normal (rather the parallel) to the Galactic plane than would be expected for a purely radial distribution of filaments. Figure 6(c) schematically indicates that this trend could result from a strong radial trend (long magenta arrows). Deviations of filament PAs from the radial direction could result from the direction of orbital motion of individual sources when superimposed with radial direction of the outflow.

Zoom In Zoom Out Reset image size

3.3.1. Short Filaments PA in Sgr E

A more detailed representation of the radial distribution of short filament PAs pointing toward the GC can be viewed in the spatial distribution of filaments PA in the prominent Sgr E H ii complex G358.7-0.0, providing a vivid example of the nearly radial distribution of short filaments. This star-forming cloud contains a number of discrete H ii regions that have been imaged using the VLA, Green Bank Telescope (GBT), and MeerKAT (Gray et al. 1993; Gray 1994; Cram et al. 1996; Yusef-Zadeh et al. 2004; Law et al. 2008b). Figure 7(a) displays an unfiltered 20 cm continuum MeerKAT image of the Sgr E discrete H ii regions distributed around the southern half of a 04 ring-shaped structure (Yusef-Zadeh et al. 2004). The northern half of Sgr E is dominated by extended diffuse emission. Some filamentary features in Sgr E appear to be correlated with structure in IR emission There is no correlation in structure between traditional nonthermal radio filaments and IR emission. This is one reason that we argued that short filaments could be thermal.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 7(b) is similar to 7(a) except that a filtered image more clearly shows the large number of filamentary features. The white contour shows the outer boundary of ¹³CO (2-1) emission from the Sgr E cloud with a mass of 3 × 10⁵ M_⊙ (Anderson et al. 2020). The ¹³CO mean velocity is −170 km s⁻¹ in the SE, becoming more negative to –220 km s⁻¹ toward the NW (Anderson et al. 2020). The identification of specific filaments with measured PAs is indicated by red lines in Figure 7(c). The histogram of the PAs of these filaments in Sgr E, as shown in Figure 7(d), indicates that filament PAs peak near PA ∼ 923 within 1σ = 39°, as the Gaussian fit (red) shows. The PA distribution in Sgr E is consistent with larger-scale distribution of filament PAs at negative Galactic longitudes, thus implying that the same mechanism is responsible for the origin of the filaments on this smaller, sub-degree scale.

This pattern of radial structure close to the Galactic plane is seen on even small-scale observations of some of the compact molecular clouds in the region. Recent ALMA observations of CO associated with the Sgr E complex finds two molecular filaments running close to the Galactic plane within ALMA's 52'' field of view. These molecular filaments have an aspect ratio of ∼5 to 1 and show alignment within 2° of the Galactic plane (Wallace et al. 2022) and appear to align radially toward the GC. The origin of this filamentary structure is suggested to be due to stretching of the Sgr E cloud by the gravitational field of the Galactic bar potential. The PAs of the CO filaments resemble those found in our radio continuum study, so it is possible that they are produced by the mechanism responsible for aligning the radio filaments. MeerKAT radio image shows a short filament G358.712 + 0.027 with an extent of ∼1' and a PA of 106° (a deviation of 15° from a radial orientation), possibly associated with the molecular filament.

Figures 7(a), (b) also display a broadly triangular dark feature bounded by the lines of H ii regions on the SE and SW and also the CO contours on the SW and north. The mean brightness of 47 μJy beam⁻¹, corresponding to a brightness temperature of ∼22 K, is coincident with contours of molecular CO in the southern half of the ring-shaped structure. To examine if this dark feature is real, we examined radio continuum image of the GC at 8 GHz observed with the GBT (Law et al. 2008b). The dark feature is clearly seen toward the Sgr E cloud, and the estimated brightness temperature toward the cloud, ∼9.7 × 10⁻³ Jy beam⁻¹, is a factor of ∼2 lower than the background diffuse emission. The deficit of ∼10 mJy per 88'' GBT beam is equivalent to an emission measure drop of ${n}_{e}^{2}{L}_{c}\approx 560\,{\mathrm{cm}}^{-6}\,\mathrm{pc}$ for a line of sight through the radio dark cloud. Adopting a path length L_c = 14 pc through the cloud, equivalent to 01 at 8 kpc, we find an equivalent electron density deficit n_e ≈ 6 cm⁻³. Making the reasonable assumption that the scale of external ionized medium has similar path length, it provides a rough estimate of the density of the external ionized medium surrounding the cloud.

The dark feature in the continuum MeerKAT and GBT images is anticorrelated with the Sgr E CO cloud. This dark feature shows similar appearance to many GC molecular clouds that have been identified as radio dark clouds (Yusef-Zadeh 2012). We interpret that the dark feature in Sgr E surrounded by compact H ii regions is produced by a deficiency in radio continuum emission from molecular clouds that are embedded in a bath of UV radiation. The deficit of the continuum emission from the volume occupied by molecular gas results in dark features that trace embedded molecular clouds.

One population of filaments has orientations that are vertical to within ∼ ± 25° of the Galactic north. The negative and positive PAs tend to be found in positive and negative Galactic longitudes, respectively. There is a slightly reduced number of filaments at PAs ∼ 0°, giving roughly a bimodal distribution in the histogram. We also note that vertical filaments dominated by long filaments L > 66'' have nonthermal spectra. Considering global distribution of vertical filament PAs, there is no evidence that the extension of all filaments PA converge toward Sgr A* or any other compact sources. This suggests that the creation of roughly symmetric PA distribution is in the Galactic plane within a degree of Sgr A*. This is consistent with a trend noticed in the spectral index of nonthermal filaments as they become steeper with increasing absolute Galactic latitude (Yusef-Zadeh et al. 2022a). If synchrotron aging is responsible for the steepening of the spectrum, it implies that the vertical filaments are powered in the Galactic plane (Yusef-Zadeh et al. 2022a).

A scenario for the creation of nonthermal radio filaments uses the high cosmic-ray pressure in the GC estimated from ${{\rm{H}}}_{3}^{+}$ measurements (e.g., Geballe et al. 1999; Oka et al. 2005; Oka & Geballe 2020). This extreme pressure drives a large-scale wind away from the Galactic plane, creating bipolar X-ray and radio emission (Heywood et al. 2019; Ponti et al. 2019; Yusef-Zadeh & Wardle 2019). The nonthermal radio filaments result from the interaction of the large-scale wind and obstacles embedded within the flow, creating the filaments by wrapping the wind's magnetic field around the obstacles (Yusef-Zadeh & Wardle 2019). However, there are alternate models explaining the origin of the vertical filaments (see Yusef-Zadeh et al. 2022a; Sofue 2023 and references therein). This begs the question of the origin of the high cosmic-ray pressure, which is usually attributed to the explosive event a few million years ago that created Fermi bubbles and the bipolar X-ray and radio emission (Heywood et al. 2019; Ponti et al. 2019; Yusef-Zadeh & Wardle 2019).

The most interesting result of our study is the discovery of a new population of short filaments' PA that is found along the Galactic plane only at l < 0°. There is statistical evidence that short filaments PA point toward Sgr A* with the cone in which the filaments are found has FWHM ∼ 20°, but the deviations of filament directions from radial is much broader, over a large scale of ∼300 pc from Sgr A* at l < 0°. This large-scale anisotropy in the filaments PA is interpreted in the context of a large-scale collimated outflow from Sgr A* along the plane. A wind- or jet-driven outflow from Sgr A* has been suggested to explain evidence for outflow in the plane of the Galaxy within a few arcminutes of Sgr A*.

There are eight different spectroscopic, polarimetric, and broadband continuum measurements over a wide range of angular scales extending from ∼0.02 to ∼25 pc from Sgr A* that have inferred a jet-driven outflow along the Galactic plane (see Yusef-Zadeh et al. 2012, 2020 and references therein). In this picture the jet emerges perpendicular to the equatorial plane of the accretion flow and is aligned roughly east−west in Galactic coordinates. A schematic diagram in Figure 8 showing the interaction of the outflow from Sgr A* with filaments of ionized and molecular gas being aligned roughly along the Galactic flow. For example, blue- and redshifted features in radio recombination line observations on a scale within 14'' (0.56 pc) of Sgr A* invoke an interpretation in terms of the interaction of a collimated outflow driven outflow from Sgr A* along ∼ ± 30° of the Galactic plane (Royster et al. 2019). In this picture, the blueshifted arm of the jet emerges to the west of Sgr A* making an angle ∼45° to the line of sight.

Zoom In Zoom Out Reset image size

This small-scale geometry of the jet is consistent with our analysis of filaments PA if the two opposing cones of outflowing material are extrapolated to a scale of ∼300 pc. Thus, the blueshifted component of the jet projected along negative longitudes and is tilted toward us with respect to the line of sight, whereas the redshifted component is tilted away from the line of sight projected toward the positive longitudes. However, there is a lack of strong radial filament PAs at positive longitudes in the direction away from the line of sight. This could be explained by confusion due to crowding of radio continuum features in the inner Galaxy.

The jet picture quantified below can also be applied to molecular filaments running parallel to the Galactic plane in Sgr E (Wallace et al. 2022; see Figure 8). An alternative suggestion that has been put forth is that stretching of molecular gas in Sgr E is due to its motion along x1 orbits. The Sgr E complex lies at the intersection of CMZ and the Galactic dust lane, so it is possible that the gravitational potential of the bar is responsible for stretching the Sgr E cloud into filaments (Wallace et al. 2022). This suggestion is based on limited ALMA observations over a small region showing CO filaments on a length scale of 2 pc. It is not clear if this suggestion can explain the radial distribution of filament PAs pointing toward Sgr A* on a scale of 300 pc.

We consider a model in which Sgr A* is the source of an outflow that has sufficient ram pressure to distort and stretch thermal and nonthermal materials that are embedded within the outflow. In this picture, the outflow has a mass-loss rate $\dot{M}$, velocity u, and Lorentz factor γ and is assumed to be directed into opposing cones, each with half-opening angle θ (see Figure 8 for a schematic diagram). The outflow delivers momentum at rate $\dot{p}=\dot{M}\gamma u$ into solid angle ${\rm{\Omega }}=8\pi {\sin }^{2}(\theta /2)$. The ram pressure within an outflow cone at distance d from Sgr A* is ${P}_{\mathrm{ram}}=\dot{p}/({\rm{\Omega }}{d}^{2})$, i.e.,

$$\begin{eqnarray}&&{P}_{\mathrm{ram}}=\displaystyle \frac{\dot{M}\gamma u}{8\pi {d}^{2}{\sin }^{2}(\theta /2)}.\end{eqnarray} \tag{ 1 }$$

Adopting nominal values $\dot{M}\gamma u/c={10}^{-4}$ M_⊙ yr⁻¹ and θ = 20°, consistent with outflow parameters invoked on smaller scales (Yusef-Zadeh et al. 2012; Royster et al. 2019; Yusef-Zadeh et al. 2020), we find that the ram pressure within the cones is P_ram/k = 2 × 10⁶ cm⁻³ K at d = 300 pc. In order for the outflow to align nonthermal radio filaments with the outflow direction, the ram pressure must exceed the internal filament pressure. Assuming that the magnetic field B ∼ 86 μ G, the mass outflow rate has to exceed 10⁻⁴ M_⊙ yr⁻¹.

There are also ionized and molecular clouds, such as the Sgr E complex, which are embedded within the outflow. To estimate the velocity gradient that this outflow might induce in an embedded or partially embedded cloud, we note that cloud material with surface density Σ experiences acceleration a = P_ram/Σ. From a standing start, the velocity acquired after being accelerated through distance L is (2aL)^1/2. Clouds typically exhibit order-unity surface density variations, so the difference in the ram-pressure induced velocity across the cloud is of order v, and we write Δv ≈ (2aL)^1/2 (see Figure 8).

We now apply this scenario to the Sgr A E cloud, which exhibits Δv ≈ 20 km s⁻¹ over a scale of 05 (Anderson et al. 2020; see their Figure 10), i.e., L ≈ 70 pc at 8 kpc. We use the outflow parameters adopted above, 10⁻⁴ M_⊙ yr⁻¹. The mean integrated ¹³CO 2-1 line intensity is 17 K km s⁻¹(Anderson et al. 2020), and using X_CO = 10²¹ cm⁻² (K km s⁻¹)⁻¹ (Schuller et al. 2017), we obtain a hydrogen column N_H = 1.7 × 10²²cm⁻² and compute the cloud's surface density using Σ = 1.4m_H N_H. Then we find that the acceleration a ≈ 4 pc Myr⁻², yielding Δv ≈ 23 km s⁻¹, consistent with the observed velocity gradient. The timescale needed to accelerate the material in this scenario is (2L/a)^1/2 ≈ 6 Myr, placing a lower limit on the age of the outflow.

One advantage of this model is that the large-scale jet-driven outflow is consistent with a picture that has independently suggested to explain multiple observations at small scales from ∼0.5 to 25 pc from Sgr A* where a relativistic jet had previously been proposed (see Yusef-Zadeh et al. 2012; Royster et al. 2019; Yusef-Zadeh et al. 2020 and references therein).