MAGNETICALLY CONFINED INTERSTELLAR HOT PLASMA IN THE NUCLEAR BULGE OF OUR GALAXY

A disk-like structure is seen in the stellar number density map (Figure 1). This is known as the nuclear bulge (NB), which consists of the nuclear stellar disk (NSD) and the nuclear stellar cluster (NSC; Launhardt et al. 2002). Here, our observations clearly reveal morphology of the NB on a large scale, with much higher spatial resolution than previous studies. The NB has a symmetric, disk-like structure with a scale height of 032  ±  002 (Figure 2), although several low-density regions are seen. Those are very dense molecular clouds in front of/inside the NB. A true stellar number density is difficult to be derived in these regions, and thus they are not used in the following analysis (see K. Yasui et al., in preparation).

Zoom In Zoom Out Reset image size

The longitudinal and latitudinal profiles of the 6.7 keV line emission measured by Suzaku (Koyama et al. 2007; Uchiyama et al. 2011) clearly show an excess at the NB region over the stellar number density profiles (Figure 3, top panels). The two profiles are overplotted and scaled to have the same values at 15  <  ∣l_*∣  <  28, i.e., in a region outside of the NB (l_* and b_* denote the angular distance from Sgr A* along the Galactic longitude and latitude, respectively, and (l_*, b_*) = (l + 0056, b + 0046)).

Zoom In Zoom Out Reset image size

Fitting the longitudinal profile in the range −07 ⩽ l_* ⩽ −01 with a power law of ∝θ^−α, where θ is angular offset from Sgr A*, gives α_star = 0.30 ± 0.03 for the stellar number density. This is different both from 0.44 ± 0.02 for the 6.7 keV profile in the same range, and from $0.60^{+0.02}_{-0.03}$ for the integrated emission of Fe 6.7 and 6.9 keV lines (Heard & Warwick 2013).

The majority of faint X-ray (L_2-10 keV < 10³⁰ erg s⁻¹) sources which have not been resolved but contribute to the GCDX are most likely to be old binary systems (Sazonov et al. 2006). Using the synthetic CMD computation (Aparicio & Gallart 2004), and a constant star formation history (SFH) during 13 Gyr for the NSD (Figer et al. 2004), we have confirmed that about 75% of the stars with K_{S, 0} < 10.5 are older than 1 Gyr. So the NIR map and profiles shown here trace the distribution of the old stars, and they are clearly different from those of the 6.7 keV emission.

A contribution from faint discrete sources to the GCDX has been claimed (Wang et al. 2002; Muno et al. 2004; Revnivtsev et al. 2007). To investigate the contribution of the point sources, especially of the old stellar population detectable in our observations, we construct longitudinal and latitudinal profiles for the ratio of the 6.7 keV emission to the stellar number density (Figure 3, bottom panels). When the profiles are scaled to be unity at 15  <  ∣l_*∣  <  28, the ratios are ∼1.5 and ∼3 in the NSD and NSC, respectively.

The NB and NSC have a different formation history from the Galactic bulge (GB), and have formed stars over their entire lifetime, indicating more bright stars in the NB and NSC. Here, we use the synthetic CMD computation (Aparicio & Gallart 2004) to estimate the fraction of the total number of stars formed in the computation, N_all, to the number of stars with K < 10.5, N_{K < 10.5}. This ratio, R ≡ N_all/N_{K < 10.5}, represents the ratio of the theoretically expected total number of stars to the number of stars detected in our observations. The SFHs used here are: a burst star formation from 10 to 13 Gyr ago for the GB (i.e., outside the NB; Zoccali et al. 2003), a constant star formation rate for 13 Gyr for the NSD (Figer et al. 2004), and the history derived by Pfuhl et al. (2011) for the NSC. When we scale the ratio R to be unity for the GB, we obtain R_GB: R_NSD: R_NSC ≈ 1: 0.8: 0.6. This result means that the densities of the old stellar population in the NSD and NSC are overpredicted by the bright stars. Taking into account this ratio, we have found that the contributions of the old stellar population to the GCDX are (1/1.5) × 0.8 ∼ 0.5 and (1/3) × 0.6 ∼ 0.2 for the NSD and NSC, respectively, in the assumption that the X-ray luminosity function is universal, and that the contribution of point sources to the GRXE is 100%. The contribution at the NSC is in good agreement with ∼1/6 derived by Koyama et al. (2009).

A larger X-ray emissivity per unit stellar mass for the GCDX than the GRXE has been claimed to explain the different distributions of stars and 6.7 keV emission (Revnivtsev et al. 2007; Heard & Warwick 2013). To change the emissivity, at least one of initial mass function (IMF), binary fraction (BF), or SFH is required to be different in the NB from the GB. In the preceding paragraphs, we have shown that different SFHs cannot explain the different spatial distributions of the stellar number density and 6.7 keV emission. Considering a universal IMF (Bastian et al. 2010) and a possible top-heavy IMF in the GC (Figer et al. 1999), the number of old, low-mass stars (i.e., CVs and ABs) per unit stellar mass never increases, it only decreases. Also, a higher stellar density tends to destroy binaries rather than to form them via a capture process, which seems to play a small role in binary formation (Tohline 2002). These imply a smaller X-ray emissivity per unit stellar mass by CVs and ABs, rather than a higher emissivity.

The most puzzling aspects of the GCDX is its high temperature. Since the kT ≈ 7 keV plasma is too hot to be gravitationally bound, it requires a huge energy source without the confinement of the plasma. One idea to address this energetics problem is the confinement of the plasma by MFs (Makishima 1994; Tanuma et al. 1999). If a large-scale toroidal MF is developed and sustained, and the MF is strong enough for nearly energy equipartition with the plasma, the GCDX could be almost confined within the NB. However, the large-scale MF configuration was thought to be predominantly vertical, suggesting that the magnetic confinement does not work well, although observations of the MF in the NB have been limited to the region in dense molecular clouds (Novak et al. 2000, 2003; Chuss et al. 2003) and very thin, non-thermal radio filaments (Tsuboi et al. 1986; Yusef-Zadeh et al. 1997; Lang et al. 1999). Recent observations have revealed that NIR and wide-field polarimetry offers a promising tool to trace a large-scale MF, and a toroidal configuration near the Galactic plane has been claimed in the GC (Nishiyama et al. 2010).

The obtained polarization map (Figure 4) suggests a large-scale toroidal MF configuration in the NB. The histogram (Figure 5) of the MF directions at ∣b∣ < 04 has a clear peak at 90° which is the direction parallel to the Galactic plane. On the other hand, at high Galactic latitude (|b| ≳ 04), the fields are nearly perpendicular to the plane, i.e., poloidal configuration. This suggests a transition of the large-scale configuration, and such a transition can be naturally explained by the time evolution of MFs. An initially predominantly poloidal, larger-scale MF is sheared out in the azimuthal direction by the differential rotation of an accreting gas disk (Uchida et al. 1985). The transition region, |b| ∼ 03–04, is in good agreement with the scale height of the 6.7 keV emission, 027 (Uchiyama et al. 2013).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The determination of the MF strength is still quite difficult in this region, but it seems to converge to the value of 50 ≲ B (μ G) ≲ 200 (e.g., Ferrière 2011; Crocker et al. 2011). At this field strength, the magnetic energy density (∼0.1–1 keV cm⁻³) reaches nearly equipartition with those of diffuse hot plasma (∼0.5 keV cm⁻³) and gas turbulence (see Figure 4 in Crocker et al. 2010). This suggests that MFs provide significant pressure support against the diffusion of the hot plasma.

If the plasma were not supported, it would be rushing out of the Galactic plane vertically as a galactic wind. The escape velocity, typically several hundred km s⁻¹, is smaller than the sound speed of the 7 keV hot plasma of ∼1400 km s⁻¹. Assuming the gas flows out from the X-ray-emitting region at the sound speed, the escape timescale is ∼4 × 10⁴ yr (Belmont et al. 2005). This requires a huge energy input to sustain the hot plasma; e.g., an unreasonably high supernova rate of ∼5 × 10⁻³ yr⁻¹ (Uchiyama et al. 2013; but note that kT ≈ 7 keV plasma is hotter than is observed in supernova remnants). If the plasma is magnetically confined, and there is no other cooling mechanism, the hot plasma only cools by radiation with a timescale of 10⁷–10⁸ yr (Muno et al. 2004), several orders of magnitude longer than the escape timescale. This would reduce the required energy input by several orders of magnitude and thus relax the energetics problem.

There is no widely accepted mechanism to heat the plasma to kT ≈ 7 keV. If past activities of the supermassive black hole (Koyama et al. 1996) heat ambient interstellar gas, created plasma could be confined magnetically. If a toroidal MF is sustained by a differential rotation of a gaseous disk, magnetic reconnection is a possible heating mechanism (Tanuma et al. 1999). Star formation activities and resultant supernovae are also implied to be the origin (Crocker 2012), while supernova-driven outflows advect materials out of the GC. However, outflows from the GC seems to be magnetized (Carretti et al. 2013), hence the existence of a large-scale toroidal field and outflows might not be contradictory.