X-Ray Observation of a Magnetized Hot Gas Outflow in the Galactic Center Region

Figure 1 shows 2° × 2° XIS mosaic images of the northern GC region in the (a) 1.78–1.94 keV, (b) 0.7–1.0 keV, and (c) 5.0–7.0 keV bands. The first energy band includes Si xiii Heα, thus it highlights emissions from thin thermal hot gases. The second shows foreground emissions because of strong interstellar absorption toward the GC. The third traces the diffuse X-ray background spreading over the entire GC region (Uchiyama et al. 2013; Ponti et al. 2015). Differences in exposures for each observation and the vignetting effect are corrected after subtracting the NXB, and images of all the XIS sensors are coadded. To clarify the diffuse emissions, we use an adaptive binning algorithm developed by Diehl & Statler (2006).

Zoom In Zoom Out Reset image size

Figure 1(a) reveals an elongated structure at (l, b) ∼ (02, 06) with an angular size of ∼1°. This structure seems weakly connected with the intense diffuse emission at the Sgr A region, named G0.1−0.1 (Ponti et al. 2015). Hereafter, we refer to this structure as the "X-ray plume." No corresponding structures can be seen in Figure 1(b) or (c).

The diffuse X-ray background in the GC region, which is detected even at $| b| \gt 1^\circ$, is crucial to the spectral analysis of the X-ray plume. Since the background intensity and interstellar absorption column significantly vary depending on lines of sight (Uchiyama et al. 2013; Yamauchi et al. 2016), simply subtracting an off-source spectrum from the source spectrum could bias the result. Alternatively, we construct a background spectral model using off-source regions and include it with parameter adjustments in the spectral model of the plume.

We use the four off-source regions, labeled B1–B4 in Figure 1(a). The XIS spectra of those regions, in which all the FI CCD data are coadded, are shown in Figure 2. They exhibit emission lines from highly ionized Mg, Si, S, Ar, Ca, and Fe, suggesting a composite of multi-temperature hot gases.

Zoom In Zoom Out Reset image size

The conventional spectral model of the GC diffuse background above ∼2 keV consists of two-temperature (kT ∼ 1 and 7 keV) plasma models in CIE plus a Gaussian for Fe i Kα (Koyama et al. 1996; Muno et al. 2004; Uchiyama et al. 2013; Ponti et al. 2015). However, we find that an additional low-temperature component is necessary to reproduce the <2 keV spectrum, especially the Mg and Si emission lines. Therefore, we use three temperature CIE plasma models with the APEC atomic code (apec in Xspec; Foster et al. 2012). The free parameters are the temperatures (kT), metal abundances (Z), and emission measures (EMs). Those for the low-, middle-, and high-temperature components are identified by subscripts L, M, and H, respectively. The metal abundances are reported with respect to the protosolar composition (Z_⊙) of Lodders et al. (2009). The energy centroid and width of the Gaussian are 6.4 keV and 0 eV, respectively, while its flux is allowed to vary. The GC background emission is subject to interstellar absorption represented by the TBabs code version 2.3 (Wilms et al. 2000), where hydrogen column density (N_H,bgd) with the solar metal composition is a free parameter.

The spectra below ∼1.5 keV, especially in regions B3 and B4, are likely dominated by foreground emission because of heavy interstellar absorption toward the GC. Following Uchiyama et al. (2013), we model the foreground emission by a single absorbed CIE plasma with an absorption column density of 5.6 × 10²¹ cm⁻², temperature of 0.59 keV, metal abundance of 0.05 Z_⊙, and emission measure of 0.1 cm⁻⁶ pc. Another lower temperature component (kT = 0.09 keV) shown in Uchiyama et al. (2013) is not included in our model because it does not affect the spectrum above 1 keV.

We add the cosmic X-ray background (CXB) component, even though its contribution is minor. The CXB spectrum is represented by a single absorbed power-law function with a photon index of 1.45 and normalization at 1 keV of 10.91 keV cm⁻² s⁻¹ sr⁻¹ keV⁻¹ (Cappelluti et al. 2017). The chosen absorption column density is two times higher than N_H,bgd because the CXB comes from extragalactic sources through the Galactic disk.

The XIS spectra are binned such that each bin contains at least 20 events. The B1–B4 spectra are simultaneously fitted with the above model to minimize the χ² statistic. Response matrix files and ancillary response files are generated with the xisrmfgen and xisximarfgen tools and are used to convolve the models. We assume that parameters of the plasma component and of Fe i Kα are common between regions, except for the entire normalization (f_bgd). The normalization for B1 is fixed to 1 as a reference, while those for the other regions are allowed to vary. The parameters of N_H,bgd for each region vary independently.

The best-fit parameters are tabulated in Table 1, and the model curves are overlaid on the data in Figure 2. Our model reproduces the observed spectra with ${\chi }_{\nu }^{2}=1.07$. If we add a model systematic error of 4% to represent the instrumental calibration uncertainties, ${\chi }_{\nu }^{2}$ decreases to 1.03 and the fit is formally acceptable with a null hypothesis probability of >5%. The temperatures and metal abundances of the middle- and high-temperature components roughly agree with those derived from Uchiyama et al. (2013). The low-temperature component explains most of the Mg xi Heα flux and a part of the Si xiii Heα flux. N_H,bgd and f_bgd for each region increase with decreasing distance from the Galactic plane as expected.

Table 1.  Background Model Parameters

Parameters (Units)	B1	B2	B3	B4
NH,bgd (1022 cm−2)	${1.47}_{-0.03}^{+0.04}$	${2.09}_{-0.04}^{+0.05}$	${4.66}_{-0.08}^{+0.08}$	${6.48}_{-0.10}^{+0.10}$
fbgd	1 (fixed)	${1.15}_{-0.02}^{+0.02}$	${2.36}_{-0.05}^{+0.05}$	${2.76}_{-0.05}^{+0.05}$
kTL (keV)	⋯	${0.43}_{-0.03}^{+0.01}$		⋯
kTM (keV)	⋯	${1.04}_{-0.03}^{+0.04}$		⋯
kTH (keV)	⋯	${7.45}_{-0.27}^{+0.24}$		⋯
ZL (Z⊙)	⋯	${0.18}_{-0.03}^{+0.04}$		⋯
ZM (Z⊙)	⋯	${0.77}_{-0.13}^{+0.22}$		⋯
ZH (Z⊙)	⋯	${1.00}_{-0.06}^{+0.06}$		⋯
EML (cm−6 pc)	⋯	${0.59}_{-0.11}^{+0.15}$		⋯
EMM (cm−6 pc)	⋯	${0.10}_{-0.02}^{+0.02}$		⋯
EMH (cm−6 pc)	⋯	${0.07}_{-0.00}^{+0.00}$		⋯
Fe i Kα (10−6 ph cm−2 s−1)	⋯	${7.5}_{-0.7}^{+0.8}$		⋯
${\chi }^{2}/\nu$	5092.6/4769

Note. All the parameters but N_H,bgd and f_bgd are common to all the regions.

Download table as:  ASCIITypeset image

We also fit this model to spectra of other regions in Figure 1 allowing only N_H,bgd and f_bgd to vary. The spectra of all regions at $| b| \gt 0\buildrel{\circ}\over{.} 3$ other than the X-ray plume region are reproduced by this background model, indicating that only the X-ray plume significantly exceeds the background emission at $| b| \gt 0\buildrel{\circ}\over{.} 3$. Moreover, the temperatures of the background model (kT_L, kT_M, and kT_H) are different from that of the X-ray plume (see the next section). This suggests that the origin of the background, which is beyond the scope of this paper, is different from that of the X-ray plume.

Since the absorption column density strongly depends on the Galactic latitude as shown in the background modeling (Table 1), we divide the X-ray plume into five sub-regions labeled S1–S5 in Figure 1(a). The derived spectra are shown in Figure 3. The background spectral model from the previous section does not reproduce the spectra, leaving large residuals, especially at Si and S emission lines.

Zoom In Zoom Out Reset image size

We represent the X-ray plume spectra by adding another apec model to the background model. The free parameters of the additional apec are the temperature and emission measure. We fix the metal abundance to Z_⊙, because it is not constrained with a large statistical error if it is allowed to vary. The additional apec is subject to interstellar absorption (TBabs) whose column density (N_H,src) is independent of N_H,bgd. The background absorption column (N_H,bgd) and normalization (f_bgd) are also allowed to vary.

The fitting procedure is the same as in the background modeling but each sub-region is independently fitted. The best-fit parameters are summarized in Table 2, and the model curves are shown in Figure 3. All spectra are reproduced by the model with ${\chi }_{\nu }^{2}\sim 1.1$. If we add a model systematic error of 4% to mimic instrumental calibration uncertainty, the fits are formally acceptable with a null hypothesis probability of >5%. We also try to fit non-equilibrium ionization plasma models (nei and rnei in Xspec), but they give no significant improvements.

Table 2.  X-Ray Plume Model Parameters

	${N}_{{\rm{H}},\mathrm{bgd}}$	${N}_{{\rm{H}},\mathrm{src}}$	kT	EM	fbgd	${\chi }^{2}/\nu$	Unabsorbed Surface Brightness in 1.5–3.0 keV
	(1022 cm−2)	(1022 cm−2)	(keV)	(cm−6 pc)			(erg cm−2 s−1 arcmin−1)
S1	${2.17}_{-0.09}^{+0.09}$	${1.92}_{-0.08}^{+0.13}$	${0.74}_{-0.02}^{+0.02}$	${0.16}_{-0.01}^{+0.01}$	${1.42}_{-0.02}^{+0.02}$	1973.6/2021	$0.71\times {10}^{-14}$
S2	${2.83}_{-0.11}^{+0.11}$	${2.93}_{-0.06}^{+0.08}$	${0.62}_{-0.01}^{+0.01}$	${0.76}_{-0.03}^{+0.03}$	${1.52}_{-0.02}^{+0.02}$	2551.2/2449	$2.47\times {10}^{-14}$
S3	${3.43}_{-0.14}^{+0.15}$	${3.41}_{-0.08}^{+0.11}$	${0.73}_{-0.01}^{+0.01}$	${0.76}_{-0.03}^{+0.03}$	${1.92}_{-0.02}^{+0.02}$	2664.5/2410	$3.27\times {10}^{-14}$
S4	${5.06}_{-0.40}^{+0.46}$	${3.65}_{-0.09}^{+0.10}$	${0.85}_{-0.03}^{+0.03}$	${0.82}_{-0.08}^{+0.09}$	${2.63}_{-0.07}^{+0.07}$	824.2/765	$4.37\times {10}^{-14}$
S5	${7.80}_{-0.44}^{+0.52}$	${4.26}_{-0.11}^{+0.11}$	${0.76}_{-0.03}^{+0.03}$	${0.76}_{-0.07}^{+0.08}$	${5.22}_{-0.08}^{+0.09}$	1845.2/1747	$3.51\times {10}^{-14}$

Note. Metallicity is fixed to the solar value.

Download table as:  ASCIITypeset image

The measured absorption column densities for the X-ray plume (N_H,src) are consistent with that of the background emission (N_H,bgd) in region S1, S2, and S3, indicating that the X-ray plume is not a foreground source but is in the GC region. In the lower latitude S4 and S5 regions, N_H,src are still as large as a typical column density toward the GC region (several 10²² cm⁻²), but are smaller than N_H,bgd by a factor of 1.4–1.9. This can be explained by the inhomogeneous distribution of dense molecular clouds at $| b| \lesssim 0\buildrel{\circ}\over{.} 5$, the so-called Central Molecular Zone (CMZ; Morris & Serabyn 1996). Even though the line-of-sight distribution of the CMZ at this latitude is uncertain, our observations suggest that the molecular clouds could be located slightly to the far side of the GC and/or the X-ray plume could be located slightly to the near side of the GC.

Assuming a distance of 8 kpc to the X-ray plume, the projected physical size is estimated to be ∼100 pc × 50 pc. The line-of-sight depth is uncertain, but we assume 50 pc hereafter. Accordingly, the hot gas density can be estimated from the observed EM to be 0.06–0.13 cm⁻³. The thermal pressure of the gas is in the range (0.7–1.8) × 10⁻¹⁰ erg cm⁻³. The total mass is ∼600 M_⊙ and the total thermal energy reaches 7 × 10⁵⁰ erg. Since the sound velocity of a kT = 0.7 keV gas is ∼400 km s⁻¹, the sound crossing time of the X-ray plume is 2 × 10⁵ yr, which gives a rough estimate of its age, assuming expansion of the hot gas.

Figure 1(d) is a composite image of the radio continuum (red), polarized radio intensity (green), and ¹²CO clouds (blue). The X-ray plume is to the east of the GC radio lobe (Sofue & Handa 1984), which is considered to be a relic of a mass outflow from the GC ≈10 Myr ago. No significant X-ray excesses over the diffuse background are found inside or to the west of the radio lobe. The polarized radio plume was found on the eastern edge of the radio lobe (Seiradakis et al. 1985; Tsuboi et al. 1986) and apparently connects the X-ray plume with the intense X-ray emission in the GC. The tower-like molecular cloud extending from b ∼ 01 to b ∼ 05 is a molecular counterpart to the double helix nebula discovered by Enokiya et al. (2014). The cloud is located at the western edge of the polarized radio lobe.

The base of the X-ray plume, corresponding to region S5 in Figure 1(a), was reported by Ponti et al. (2015) and Yamauchi et al. (2018), who named the structure the eastern edge and NE, respectively. The physical properties they reported roughly agree with ours. They also reported an X-ray excess on the western side, (l, b) ∼ (3597, 02), which is referred to as the western edge or NW, respectively. They discussed the origin as multiple supernovae in the GC and/or past flares of Sgr A* in relation to the almost symmetric X-ray morphology of the east and west sides at 01 < b < 03. However, our extended survey reveals a highly asymmetric morphology at 03 < b < 10, which raises an issue with this scenario; an outflow on the western side should be significantly suppressed if explosive events at the GC are the origin. Given the existing outflow feature on the western side, i.e., the radio lobe (see Figure 1(d)), there is no strong evidence that the hot gas outflow is suppressed in this region.

Since the total thermal energy of the X-ray plume can be explained by a single supernova explosion, it might be a supernovae remnant (SNR) at high latitude. Indeed, the temperature and abundance of the plume are similar to those of middle-aged Galactic SNRs (e.g., Kawasaki et al. 2005). However, its physical size is significantly larger than that of a typical SNR. In addition, no corresponding radio emitting shell has been found (LaRosa et al. 2000; Law et al. 2008). Therefore, an in situ supernova is unlikely to be the origin of the X-ray plume, even though it cannot be completely ruled out.

As pointed out by Ponti et al. (2015), the most characteristic feature of the X-ray plume is its apparent association with the polarized radio plume. Therefore, the magnetic flux tube at the polarized radio plume likely plays a crucial role in forming the X-ray plume; if the hot gas has a source at the footpoint of the radio plume, it would blow out along the magnetic flux tube. Indeed, the magnetic field at the radio plume is estimated to be several tens of μG (Tsuboi et al. 1986), giving a magnetic pressure comparable to the thermal pressure of the X-ray plume, B²/8π ≈ 10⁻¹⁰ erg cm⁻³. At the northern end of the radio plume, the polarized radio intensity (and hence the magnetic field) becomes one order of magnitude lower than the peak value, and the X-ray-emitting hot gas is no longer confined by the magnetic flux tube.

Additionally, an average pressure of the cold interstellar medium at b > 03 is estimated to be ≲10⁻¹¹ erg cm⁻³ according to its spatial distribution (Ferrière et al. 2007), and is significantly lower than that of the X-ray plume. This is because the observed X-rays extend to the north and east (even southeast) of the radio plume. Even though the X-ray emission does not extend to the west of the radio plume, it can still be explained by external pressure owing to the warm ionized gas at the edge of the radio lobe (∼5 × 10⁻¹⁰ erg cm⁻³; Law et al. 2009) and/or the tower cloud, as shown in Figure 1(d).

If the above magnetized outflow scenario is the case, what is the source of the hot gas? The current X-ray observations are inconclusive, but magnetic reconnection at the radio arc (Serabyn & Morris 1994) or stellar winds from the Arches cluster (Tsujimoto et al. 2007) are viable candidates because of their proximity to the polarized radio lobe. The arc bubble (Heard & Warwick 2013; Ponti et al. 2015) is another possible source but is rather young (several 10⁴ yr) to provide hot gas to the X-ray plume, whose sound crossing timescale is 0.2 Myr.

Finally, we comment on the relationship of the X-ray plume with the elongated hot gas at (l ∼ 0, −18 < b < −10) found by Nakashima et al. (2013). This southern hot gas might also originate from the magnetized outflow, because the southern part of the polarized radio plume reaches b ∼ −09 (Tsuboi et al. 1986) and is connected with the hot gas. The timescale of the southern hot gas (∼0.1 Myr) is consistent with that of the northern X-ray plume. However, its temperature is lower and its ionization state is different from that of the X-ray plume. Determining the physical association between the northern and southern X-ray-emitting hot gases will require further investigations, especially at a low-latitude region (−10 < b < −02).