THE HIGH-DENSITY IONIZED GAS IN THE CENTRAL PARSEC OF THE GALAXY

Figures 1 and 2 show the H30α line data in the central parsec. Figure 3 shows a comparison between the SMA H30α and the VLA H92α line spectra in the direction of selected IR sources. For optically thin gas in local thermodynamic equilibrium (LTE) with no pressure broadening, the ratio of line peak intensity is $\displaystyle {R^{\rm H30\alpha }_{\rm H92\alpha }\equiv {S_{\rm H30\alpha }/ S_{\rm H92\alpha }} \approx {\nu _{\rm H30\alpha }/\nu _{\rm H92\alpha }}} \approx 28$. A detailed comparison of the H30α line spectra from the SMA and the H92α line spectra observed with the VLA yields line ratios in most of the regions that are close to the expected LTE ratio, suggesting that the ionized gas in the central parsec is optically thin and under LTE conditions. In some regions, however (e.g., IRS 1W, 10W, 16, and 33), the observed ratios show significant departure from the expected LTE ratio.

The equivalent electron temperature, T*_e, derived from the ratio of line-to-continuum emission under the assumption that the optically thin gas is in LTE, can be estimated from our measurements. In order to avoid contamination from both the dust emission, which becomes dominant at higher frequencies, and significant synchrotron emission at lower frequencies, we used the continuum data at 22 GHz, where the contributions from both the synchrotron and thermal dust emission are negligible relative to the free–free emission. Figures 4(a) and (b) show the integrated H30α line image and radio continuum image at 22 GHz, respectively. We assume that the continuum emission at 22 GHz is entirely due to optically thin free–free emission and scale its intensity to 231.9 GHz, according to the power law ν^−0.1, i.e., a factor of 0.79. Using Equation (1) below (Wilson et al. 2009), we determined the value of T*_e

$$\begin{eqnarray} {T_e^*\over K} &=&\Bigg [ {6985 \over a(\nu, T_e)} \left(\nu \over {\rm GHz}\right)^{1.1} {1\over {1+N(\mbox{He})/N(\mbox{H})}} \nonumber \\ & &\times \left(S_{C}\over {S_{L} \Delta V_{{\rm FWHM}}} \right) \Bigg ]^{0.87}, \end{eqnarray} \tag{ 1 }$$

where ν is the frequency, S_L is the line flux density, S_C is the continuum flux density, ΔV_FWHM is the line width (FWHM), and only Doppler broadening is involved, N(He)/N(H) is the number density ratio, which we take to be 0.08, and α(ν, T_e) ∼ 0.97 (at ν = 22 GHz and T_e, the electron temperature, ∼10⁴ K) is the correction to the Mezger & Henderson (1967) power-law approximation. The angular distribution of T*_e is shown in Figure 4(c). T*_e varies from 7000 K to 10,000 K in most regions of Sgr A West. However, in the Bar region (e.g., IRS 2L, IRS 12N, IRS 13E, and IRS 33), 3'' southeast of Sgr A*, T*_e increases to ∼15,000 K. The significant increase of T*_e in the Bar has also been derived from line observations of H92α (Roberts & Goss 1993) and H66α (Schwarz et al. 1989).

Zoom In Zoom Out Reset image size

Since the observed quantities of line and continuum intensities depend on the density, temperature, and path length of H ii gas, we can construct a model and fit the data to it over a wide frequency range in order to determine the parameters of the ionized gas. In particular, with the high-resolution data obtained from interferometer observations, uncertainties due to variations in structure and the physical parameters across the source are substantially reduced. Thus, the results from such an analysis become more reliable. Models for RRL emission from the nuclear region of external galaxies have been discussed by Puxley et al. (1991), Anantharamaiah et al. (1993), Anantharamaiah et al. (2000), Zhao et al. (1996), Zhao et al. (1997), Phookun et al. (1998), and Rodriguez-Rico et al. (2004). They found that the main constraints for the models are the integrated RRL strength at multiple frequencies and the observed radio continuum spectrum. The observed quantities (line and continuum intensities) depend in a nonlinear way on the distribution of T_e and n_e, the electron density, both along the line of sight and across the telescope beam. In the cases of the extragalactic nuclear regions, at a given frequency, the line emission often arises from H ii components under the physical conditions that are particularly favorable at this particular frequency. Usually, multiple collections of H ii regions with different physical parameters are required to fit the observations.

Both the SMA and VLA data achieve angular resolution (2'', ∼0.1 pc) adequate to resolve the ∼1 pc minispiral structure. The effects due to changes of physical parameters across the source are therefore less significant. However, the electron temperature and density could possibly change along the line of sight; T_e and n_e derived from the simple isothermal and homogeneous model correspond, therefore, to average physical properties of the H ii gas in the telescope beam. If the H ii gas in the beam along the path length is isothermal and homogeneous, then the line and continuum flux densities from a region subtending a solid angle Ω are given by (e.g., Shaver 1975)

$$\begin{eqnarray} S_L&=&{2k\nu ^2\over c^2} \Omega T_e \Bigg [\!\!\left(\tau _L/\beta _n +\tau _C \over \tau _L + \tau _C \right)(1-e^{-(\tau _L+\tau _C)}) -(1- e^{-\tau _C})\Bigg ],\nonumber\\ \end{eqnarray} \tag{ 2 }$$

and

$$\begin{equation} S_C = {2k\nu ^2\over c^2} \Omega T_e(1- e^{-\tau _C}), \end{equation} \tag{ 3 }$$

where k is Boltzmann's constant, c is the speed of light, and the line (τ_L) and continuum (τ_C) optical depths are given by

$$\begin{eqnarray} \displaystyle \tau _L &\approx & 575 b_n\beta _n \left(\nu \over {\rm GHz}\right)^{-1} \left(n_e\over {\rm cm}^{-3}\right)^2 \nonumber \\ & &\times \left(Lf_V \over {\rm pc}\right)\left(T_e\over {\rm K}\right)^{-5/2} \left(\Delta V_{D}\over {\rm km\, s}^{-1}\right)^{-1} \left(1+1.48{{\Delta V_P}\over {\Delta V_D}}\right)^{-1},\nonumber\\ \end{eqnarray} \tag{ 4 }$$

and

$$\begin{eqnarray} \displaystyle \tau _C &\approx & 0.08235 \left(n_e\over {\rm cm}^{-3}\right)^2 \left(Lf_V\over {\rm pc}\right) \left(\nu \over {\rm GHz}\right)^{-2.1} \nonumber \\ & &\times \left(T_e\over {\rm K}\right)^{-1.35} a(\nu,T_e), \end{eqnarray} \tag{ 5 }$$

respectively. b_n and β_n are the population departure coefficients, n_e is the electron density, ΔV_D and ΔV_P are the FWHM Doppler and pressure line widths in km s⁻¹, respectively, L is the path length, and f_V is the volume filling factor. Note that the last factor in Equation (4) provides a convenient approximation for the transition between the pure Doppler and pure pressure-broadened cases.

The Doppler broadening has a thermal and turbulent component and is given by

$$\begin{equation} \displaystyle \Delta V_D = \sqrt{{8{\rm ln(2)}kT_e\over M_{\rm H}}+ (\Delta V_t)^2}, \end{equation} \tag{ 6 }$$

where M_H is the mass of hydrogen and ΔV_t is the turbulent width, which here includes microscopic turbulence and dynamical velocity gradients. In the central parsec, the thermal width ($\Delta V_{\rm th}=\sqrt{{8{\rm ln(2)}kT_e\over M_{\rm H}}}$) is less than the turbulent width (ΔV_t). The line width due to pressure broadening, computed by Brocklehurst & Leeman (1971), is given by

$$\begin{equation} \displaystyle \Delta V_P = 1.7\times 10^{-18} {n_e\over T_e^{0.1}} N^{7.4}\, {\rm km\, s^{-1}}, \end{equation} \tag{ 7 }$$

where N is the principal quantum number of the transition. The total line width is approximately the quadratic sum of ΔV_D and ΔV_P

$$\begin{equation} \Delta V_{\rm FWHM} \cong \sqrt{(\Delta V_D)^2 + (\Delta V_P)^2}. \end{equation} \tag{ 8 }$$

ΔV_FWHM is an observable quantity. We solved for the parameters T_e, n_e, ΔV_t, and Lf_V for each point in the image in the following way. Since ΔV_P ≪ ΔV_th < ΔV_t for reasonable values of T_e and n_e, we assume initial values of T_e and n_e, and solve for ΔV_t via Equations (6)–(8). We computed the values for b_n and β_n for the assumed values of T_e and n_e using the NLTE code of Gordon & Sorochenko (2002), which followed the analysis of Walmsley (1990). We then formed the flux density ratios $\displaystyle {R^{\rm H30\alpha }_{\rm 22\; GHz}\equiv {S_{\rm H30\alpha }/ S_{\rm 22\; GHz}}}$, $\displaystyle {R^{\rm H92\alpha }_{\rm 22\; GHz}\equiv {S_{\rm H92\alpha }/ S_{\rm 22\; GHz}}}$, and $\displaystyle {R^{\rm H30\alpha }_{\rm 92\alpha }\equiv {S_{\rm H30\alpha }/ S_{\rm H92\alpha }}}$. The use of these ratios eliminates the solid angle Ω in Equations (2) and (3). At each point in the image, we used these ratios to solve for T_e, n_e, and Lf_V. We then iterated the process, i.e., solved for a new estimate of ΔV_t with these values of T_e and n_e. The process converged in a few iterations.

The parameters derived for the line of sight for each of the regions are summarized in Table 3. Instead of Lf_V, we report the EM (EM = n²_eLf_V). The ratio T_e/T*_e is given in Column  6, which mainly corrects for the effect due to underpopulated electrons at the level N = 30 as compared to that expected under the LTE; the ratio ΔV_t/ΔV_FWHM in Column 7; and ΔV^H92α_P/ΔV^TH_D, the ratio of the H92α line width of pressure broadening to that of thermal Doppler broadening, in Column  8. The pressure broadening term (ΔV^H30α_P) is negligible.

Table 3. Physical Conditions of Ionized Gas in IRS Regions

Region	Te	ne	EM	τC(22 GHz)	$\left(T_{\rm e}\over T_{\rm e}^*\right)$a	$\left(\Delta V_{t}\over \Delta V_{\rm FWHM}\right)$	$\left(\Delta V_{\rm P}^{\rm H92\alpha } \over \Delta V_{\rm D}^{\rm TH} \right)$b
	(103 K)	(104 cm−3)	(107 cm−6 pc)	(10−3)
IRS 1W	6.7 ± 0.5	6.0 ± 0.4	1.8 ± 0.1	14. ± 1.1	0.71	0.88	0.82
IRS 10W	6.8 ± 0.6	5.3 ± 0.4	1.1 ± 0.1	8.5 ± 1.0	0.71	0.85	0.72
IRS 5	5.0 ± 1.1	7.2 ± 1.2	0.4 ± 0.1	4.7 ± 1.3	0.68	0.78	1.2
IRS 16	6.7 ± 2.0	2.7 ± 0.7	0.8 ± 0.2	6.4 ± 2.7	0.68	0.99	0.37
IRS 6	5.0 ± 0.6	11. ± 1.1	1.6 ± 0.1	18. ± 2.3	0.71	0.96	1.8
IRS 13E	12. ± 2.5	21. ± 3.6	3.2 ± 0.5	12. ± 1.9	0.84	0.84	2.0
IRS 2L	8.5 ± 1.3	14. ± 1.4	2.7 ± 0.4	16. ± 2.9	0.78	0.80	1.7
IRS 12N	12. ± 4.7	2.8 ± 1.0	0.5 ± 0.2	1.8 ± 0.6	0.77	0.80	0.27
IRS 33	13. ± 3.3	10. ± 3.0	2.6 ± 0.3	8.5 ± 1.8	0.81	0.97	0.92
IRS 21	10. ± 1.8	8.5 ± 1.3	2.0 ± 0.3	9.1 ± 2.0	0.78	0.94	0.92
IRS 9W	11.−12.	0.8−18	1.3−1.3	5.5−4.9	0.74−0.83	0.76−0.93	1.7−0.08
IRS 9	5.7 ± 0.9	6.4 ± 0.8	1.0 ± 0.2	9.7 ± 1.8	0.69	0.97	0.97
IRS 4	5.8 ± 1.3	5.0 ± 1.0	0.5 ± 0.1	4.8 ± 1.4	0.68	0.91	0.75
IRS 28	5.6 ± 0.8	8.9 ± 1.0	0.7 ± 0.1	6.8 ± 1.1	0.71	0.86	1.4
IRS 7	7.0 ± 2.5	8.8 ± 2.7	0.2 ± 0.1	1.2 ± 0.5	0.74	0.64	1.2
Minicavity	9.3 ± 0.9	8.5 ± 0.7	2.0 ± 0.3	10. ± 2.0	0.77	0.78	0.96

Notes. ^a${T_{\rm e}\over T_{\rm e}^*} \approx [b_{30}(1-{1\over 2}\beta _{30}\tau _{\rm C}(232\;{\rm GHz})]^{0.87} \approx b_{30}^{0.87}$ for a very small value of τ_C(232 GHz) in a region at the Galactic center. ^b$\Delta V_{\rm D}^{\rm TH} = \sqrt{8 {\rm ln} (2) k T_e\over M_H}$ is the thermal Doppler line width.

Download table as:  ASCIITypeset image

The uncertainties in the physical parameters mainly come from the uncertainties in the measurements. Although the relation between the physical parameters and observable quantities is nonlinear, we assess the errors through linear approximation. We use the optically thin assumption, which appears to be valid at the frequencies used in this paper. The fractional uncertainty in T_e, which is proportional to the fractional uncertainties of the measured ratios, $\displaystyle {R^{\rm H30\alpha }_{\rm 22\; GHz}}$ and $\displaystyle R^{\rm H30\alpha }_{\rm H92\alpha }$, lies in the range from 7% for IRS 1W to 39% for IRS 12N. The fractional uncertainty in n_e is dominated by those of observed line ratios $\displaystyle R^{\rm H30\alpha }_{\rm H92\alpha }$ and ranges from 6% for IRS 1W to 36% for IRS 12N.

The variations of the H30α line intensities reflect the changes of excitation conditions in the region. Our model fitting suggests that, given similar solutions in n_e and T_e, the large difference in the H30α line intensity between the two extreme regions (IRS 1W and IRS 7) is mainly due to the difference of the EM or the path length along the line of sight (see Table 3).

In short, the line ratio of H30α to H92α varies from 15 ± 4 to 32 ± 4. The derived mean values of the line optical depth are $\overline{\tau }_L{\rm (H30\alpha)} =$ −0.001 and $\overline{\tau }_L{\rm (H92\alpha)} =$ −0.006 with standard deviations of $\sigma _{\tau _L{\rm (H30\alpha)}}=0.0004$ and $\sigma _{\tau _L{\rm (H92\alpha)}}=0.005$ from the corresponding emission regions. The free–free optical depth at the H30α line frequency τ_C(231.9 GHz) varies from 1×10⁻⁴ to 1×10⁻⁵ while τ_C(8.3 GHz) is in the range between 0.16 and 0.01. Because of the small optical depths in both line and continuum, the non-LTE effects in the minispiral, in general, are not critical.

There is a diffuse nonthermal synchrotron component in the central parsec region (Ekers et al. 1983). Based on the flux density of 22 Jy at 20 cm in the central 40'' × 70'' from Ekers et al. (1983), we estimated that the contribution to the continuum flux density in a 2'' beam is ∼10 mJy under the assumption that the nonthermal spectral index α ∼ −0.5 (S_ν ∝ ν^α) and the emission are evenly distributed. The optical depth of the H92α line is small (τ_L∼ −0.01). The contribution from the stimulated H92α emission of ∼0.1 mJy is insignificant in the regions discussed in this paper. The contribution from the stimulated line emission by the dust emission at 232 GHz is negligible because of the small optical depth of the H30α line.

The observed line widths of the H ii gas in the minispiral arms at the Galactic center are dominated by ΔV_t, which is consistent with velocity gradients due to gas motion around Sgr A* or the SMBH being primarily responsible for the broadening in the line profiles observed in the 2'' beam.

The ionized gas of the high-density components in the Northern and Eastern Arms has a mean kinetic temperature of $\overline{T}_e\approx 6\times 10^3$ K and an electron density of n_e in the range of 10^4–5 cm⁻³, which is consistent with the values (10⁴ cm⁻³) derived from Paα line emission at 1.87 μm (Scoville et al. 2003) for a given uncertainty in extinction corrections in the near-IR. The H ii gas properties in the Northern and Eastern Arms are similar to these in H ii regions around young and/or evolved massive stars.

We confirm that the ionized gas in the Bar (e.g., IRS 2L, IRS 12N, IRS 13E, and IRS 33 in Table 3) appears to have a higher kinetic temperature in the range of 8.5 × 10³ K ⩽T_e ⩽ 1.3 × 10⁴ K, as suggested by earlier RRL observations (Roberts & Goss 1993). The electron density shows a range from a few times 10⁴ to a few times 10⁵ cm⁻³.

The heating process may be complicated in the Bar because of complex gas interactions. There are several energy sources in the region. First, strong winds from the central stellar clusters such as IRS 16 and IRS 13 could provide kinetic energy to compress the ionized gas in this region. The heating due to the interaction between the winds and the ionized gas in the Bar likely occurs in the regions localized to the cluster sources.

In addition, the well-known minicavity (marked with the gray circle in Figures 4(a) and (b)), a region depressed of radio continuum, is located 35 southwest of Sgr A* (Yusef-Zadeh et al. 1989). This region is likely created by a fast wind outflow from the sources in the vicinity sweeping up the gas in the orbiting ionized streams, as suggested by the morphology of a bright [Fe iii] line in the infrared (Lutz et al. 1993). However, the minicavity is a few arcsec displaced from the high-kinetic-temperature region, which in fact is best seen as a depression associated with H30α line emission. The minicavity can also be explained by Melia et al. (1996) as a downstream, focused flow due to the accretion by the SMBH at Sgr A*.

Alternatively, the high temperature in the Bar could result from a collision between two ionized orbiting flows, as suggested in the dynamical model of Zhao et al. (2009). A computation using their Keplerian orbital model (see Section 3.2) shows that the specific kinetic energy reaches a maximum (see Figure 4(d)) in the Bar region where high kinetic temperatures (see Figure 4(c)) are observed. The distribution of specific kinetic energy (Figure 4(d)) shows a substantial increase from 2 × 10¹⁴ erg g⁻¹ at a region near IRS 5 in the Northern Arm to 12 × 10¹⁴ erg g⁻¹ at IRS 33 in the Bar. However, we note that a high kinetic energy of the orbiting flow does not have to correspond to a high kinetic temperature; the ionized flows can release their kinetic energy to shocks if the ionized flows collide with each other. If they collide with the strong stellar winds from the clusters of stars in the Bar area, then the gas could be heated by the shocks to higher temperatures. In addition, the model also shows that the ionized flows of the Northern and Eastern Arms do collide at the region with high kinetic temperatures (see Figure 21(d) of Zhao et al. (2009) and the three-dimensional structure of the minispiral arms in Figure 5). The observed high kinetic temperatures in the Bar suggest that the ionized gas could be further heated by the shocks generated from the collision.

Zoom In Zoom Out Reset image size

Using the Keplerian model described in Zhao et al. (2009) along with the orbital parameters derived from fitting the VLA H92α data and proper motion measurements, we recomputed the three-dimensional locations of the three ionized arms. Figure 5 shows a three-dimensional visualization of the three ionized flows in the Northern, Eastern, and Western Arms (Arc), which illustrates that the Northern Arm and Western Arc are nearly coplanar based on the similar longitudes of their ascending nodes (Ω) and inclination angles (i) derived from fitting the loci of the ionized flows. The mean values of the two orbital angles, $\overline{\Omega }=71\mbox{$^\circ $}\,\pm\,6$° and $\overline{i}=119\mbox{$^\circ $}\,\pm\,3$°, are inferred for the two coplanar orbits (Northern Arm and Western Arc), which form the ionized ring. The orbital plane of the ionized flows appears to be aligned with the plane of the circumnuclear disk (CND) derived by Güsten et al. (1987) from molecular line observations. If the fitted parameters from both the molecular lines (Güsten et al. 1987) and hydrogen recombination lines (Schwarz et al. 1989; Lacy et al. 1991; Roberts & Goss 1993; Vollmer & Duschl 2000; Liszt 2003; Zhao et al. 2009) are correct, the Keplerian motions dominate the kinematics in the central parsec, and the plane of the main nuclear gaseous disk or the CND can be described by the mean orbital angles inferred above. Therefore, the orbital plane of the Eastern Arm (i = 122° ±  5° and Ω = −42° ± 11°) is nearly perpendicular to that of the CND. Figure 5 shows that the Eastern Arm flow runs into the CND from the northeast and the near side with respect to Sgr A* and collides with the Northern Arm flow in the region located south and behind Sgr A*.

Figure 6 shows the spectra taken from 12 positions along each of the two ionized arms (shown in Figure 7). The observed spectra are compared with the radial velocities (thick green bars) derived from the Keplerian model. Note that we assume that the LSR velocity of the black hole, V_z, is 0 to within ± 5 km s⁻¹ (see discussion in Gillessen et al. 2009). In general, the kinematics of the ionized gas follow the Keplerian orbital motion expected for the gravity of the SMBH (4.2 × 10⁶ M_☉) at Sgr A*. Models of the enclosed mass (see Oh et al. 2009) show that the added enclosed mass starts to rise above the black hole's mass starting at a radius of 0.3 pc (7'') from Sgr A*. At a radius of 1 pc (24'') from Sgr A*, near the tip of the minispiral seen in projection, the enclosed mass has nearly doubled, to about 8 × 10⁶ M_☉. Thus, for a given circular orbit with a radius of 1 pc, the actual orbital velocity would be about 1.4 times higher than in a pure Keplerian approximation based on the mass of the black hole only. Apart from this effect, however, in several regions the radial velocities deviate even more significantly from those predicted from the Keplerian model.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In the northwest end of the Eastern Arm near the location E17 (Figure 7(a)), a feature with radial velocities blueshifted up to ∼−200 km s⁻¹ shows a large deviation from the value of ∼−100 km s⁻¹ predicted by the Keplerian model (Figure 6). From Paumard et al. (2006), we find that at least five massive stars—IRS 34W (Ofpe/WN9), IRS 34E (O9–9.5I), IRS 34NW (WN7), IRS 7SW (WN8), and IRS 3E (WC5/6)—are located in a region with a radius of 22 (indicated with a large black circle near region E16 in Figure 7(a)), referred to as IRS 34 hereafter. Except for IRS 3E, which has no proper motion measurements, the proper motion data of the remaining four stars (Paumard et al. 2006) show that the stars all move southwest toward the ionized flow with a mean transverse velocity of 185 km s⁻¹ (P.A. = –134°) (as shown by an arrow in Figure 7(a)). We compare the radial velocity of the stars with the spectrum (inset (c) in Figure 7(a)) of the H30α line emission integrated from the IRS 34 region. The four stars besides IRS 3E show negative radial velocities in the range –150 km s⁻¹ to –340 km s⁻¹, which is blueshifted with respect to the peak velocity (−146 ± 4 km s⁻¹) of the broad (ΔV_FWHM = 160 ± 8 km s⁻¹) spectral feature. The stellar winds from the Wolf-Rayet (WR) stars, evolved from early-type O-stars with mass-loss rate of 10⁻⁵M_☉ yr⁻¹, appear to play a considerable role in the local kinematics. The strong stellar winds with velocities up to a few times 10³ km s⁻¹ could sweep up the ionized flow of the Eastern Arm, causing the peculiar velocities and the velocity gradient across the Eastern Arm at the northwest end. In fact, the motions of the filaments associated with the minispiral found by Mužić et al. (2007) in infrared also suggest that a model of purely Keplerian motions of the gas is inadequate and the filaments could be due to the interaction of a fast wind with the minispiral.

In the tip region (Paumard et al. 2004) with peculiar radial velocity that deviates significantly from the Keplerian velocity by +150 km s⁻¹ (see panel (E12) in Figures 6 and 7(a) and (b)), there is also a WN8 star, IRS 9W. The radial velocity (+140 km s⁻¹) of IRS 9W (Paumard et al. 2004) is similar to that of ionized gas in the Eastern Arm. A comparison between the stellar transverse velocity of 215 km s⁻¹ (P.A. = 51°) and that of 304 km s⁻¹ (P.A. = –67°) predicted from the Keplerian model for the ionized flow at the location of IRS 9W suggests that the star does not follow the orbit of the Eastern Arm but may run across the Eastern Arm. Then, the high-radial-velocity tip is likely to be the part of the gas in the ionized flow that is compressed and accelerated by the suspected strong wind from the IRS 9W star. The velocity discontinuity at the tip may be evidence for the presence of shocks in the region.

In addition, deviations from Keplerian motions are also observed in the Bar region. The non-Keplerian kinematics could be attributed both to the interaction between the ionized flows in the Northern and Eastern Arms and to the interaction of the ionized flows with the strong winds from the massive star clusters (the IRS 16 and IRS 13 clusters, for example). Well-organized polarized emission from magnetically aligned dust grains in the central parsec has been observed (Aitken et al. 1991, 1998; Glasse et al. 2003), suggesting a magnetic field strength of ⩾2 mG. The corresponding energy densities in the magnetic fields are ∼1.6 × 10⁻⁷ erg cm⁻³, about twice larger than the thermal energy densities in the ionized gas (T_e ∼ 7 × 10³ K and n_e ∼ 6 × 10⁴ cm⁻³ at IRS 1W) but 2 orders of magnitude less than the kinetic energy densities of the orbital motions. The magnetic fields frozen in the ionized streamers cannot substantially alter the bulk motions of the gas but produce some cumulative magnetohydrodynamic (MHD) effects, such as orbital compression of the streamers as implied from the observed evidence for the convergence of field lines near Sgr A* (Glasse et al. 2003), and a helix of a natural morphology for a twisted magnetic field frozen into a plasma as observed in the Northern Arm (Yusef-Zadeh et al. 1989; Zhao et al. 2009). In the region close to Sgr A*, the magnetic field probably becomes stronger, and the MHD effects may also play a significant role in the kinematics of the ionized flows.

An interesting anomalous feature located ∼2'' northwest of Sgr A* can be seen in the integrated H30α line intensity image (Figure 1). The radial-velocity image (Figure 2) indicates a large velocity gradient across this feature. The H30α line spectrum (inset (d) in Figure 7) suggests that this feature is composed of at least three spectral components (⩾4σ) at radial velocities of 160 km s⁻¹, 39 km s⁻¹, and –250 km s⁻¹, with a line width of ∼50 km s⁻¹, a typical velocity width of the ionized gas in the minispiral arms. We note that there is a compact continuum source called ε (Yusef-Zadeh & Morris 1991), located 1'' south of this anomalous feature, corresponding to a hole in the H30α line images (e.g., Figures 1 and 7(a)). The source ε appears to be moving away from Sgr A* with a transverse velocity of 340 km s⁻¹ (Zhao et al. 2009), which has been interpreted as the confluence of gravitationally focused winds from the IRS 16 cluster (Wardle & Yusef-Zadeh 1992). The H30α line property of the anomalous feature appears to be consistent with the focused-wind model, suggesting that the velocity components in the anomalous feature correspond to ionized blobs of the minispiral arms that have been blown out of the region ε due to the high pressure of the confluent stellar winds. However, given the closeness to Sgr A*, the detection of this feature needs to be further confirmed with higher-sensitivity observations.

Finally, in the locations with large angular offsets from Sgr A*, we noticed that the deviations in radial velocity from the Keplerian motions become significant for the Northern Arm (>50 km s⁻¹), as shown in both Figures 6 and 7, while for the Eastern Arm, the deviations are less significant. The difference in the deviation of the observed radial velocities from the Keplerian velocities between the Northern and Eastern Arms suggests that the distribution of mass has a preference for the plane of the CND.