The Second Galactic Center Black Hole? A Possible Detection of Ionized Gas Orbiting around an IMBH Embedded in the Galactic Center IRS13E Complex

We observed the continuum emission of Sgr A^* and the GCMS at 340 GHz as an ALMA Cy.3 program (2015.1.01080.S.; P.I.: M.Tsuboi). The observations were performed in three days (2016 April 23, August 30/31, and September 08). The observation in April was for the detection of extended emission. The field of view in FWHM (FOV) is 18'', which is centered at Sgr A^*:${\alpha }_{\mathrm{ICRS}}={17}^{{\rm{h}}}{45}^{{\rm{m}}}40\buildrel{\rm{s}}\over{.} 04$ and ${\delta }_{\mathrm{ICRS}}=-29^\circ 00^{\prime} 28\buildrel{\prime\prime}\over{.} 20$. J1717-3342 was used as a phase calibrator. The flux density scale was determined using Titan and J1733-1304. We used the self-calibration method in CASA to improve the dynamic range of the map. The angular resolutions using "natural weighting" and "uniform weighting" are $0\buildrel{\prime\prime}\over{.} 14\times 0\buildrel{\prime\prime}\over{.} 13,{PA}=82\buildrel{\circ}\over{.} 5$ and $0\buildrel{\prime\prime}\over{.} 101\times 0\buildrel{\prime\prime}\over{.} 090,{PA}\,=\,3\buildrel{\circ}\over{.} 8$, respectively. The sensitivities are $1\sigma =0.10$ and 0.18 mJy beam⁻¹ in the emission-free areas, respectively. The dynamic ranges are ∼28,000 and ∼16,000, respectively. These values are several times worse than the expected values generated by the ALMA sensitivity calculator. We will present the improvement of the analysis and the full results in another paper (M. Tsuboi et al. 2017, in preparation).

We analyzed the Director's Discretionary Time (DDT) observation toward the Galactic Center with ALMA, which was released in the autumn of 2016 (Project code: 2015.A.00021.S). The data set contains two-day observations (2016 July 12/13 and 18/19) of the Galactic Center in the H30α recombination line (${\nu }_{{rest}}=231.9$ GHz). The FOV is 25'', which is centered at Sgr A^*. Since the original observations aimed for the measurement of the time variation of Sgr A^*, the time spans of the observations are 22:40 UT–03:40 UT for the first day and 23:00 UT–05:45 UT for the second day, providing good coverages on the spatial frequency plane (uv plane) for imaging. J1744-3116 was used as a phase calibrator. The flux density scale was determined using Titan and J1733-1304. Before spectral line imaging, the continuum emissions of Sgr A^* and the GCMS were subtracted from the data using the CASA task mstransform (fitorder = 1) for each day of the observations. Imaging for the combined data for the two-day observations was done using CASA 5.0 and the tclean task that utilized a new auto-boxing capability. We used the multi-threshold auto-boxing algorithm (usemask = ''auto-multithresh'') to automatically identify the emission regions to be CLEANed using a threshold based on rms noise and sidelobe level, and updated the deconvolution iterations progress.⁷ For the weighting scheme, Briggs weighting with a robust parameter of 0.5 was used. As a result, we obtained an H30α recombination line data cube with high angular resolution ($0\buildrel{\prime\prime}\over{.} 41\times 0\buildrel{\prime\prime}\over{.} 30,{PA}=-77^\circ$) and high sensitivity (0.2 mJy beam⁻¹ at a line-free channel). The sensitivity is close to that expected from the ALMA sensitivity calculator.

The angular source size of the ionized gas toward IRS13E3 is derived to be ${\theta }_{\mathrm{maj}.\mathrm{obs}.}\times {\theta }_{\min .\mathrm{obs}.}=0\buildrel{\prime\prime}\over{.} 41\pm 0\buildrel{\prime\prime}\over{.} 03\times 0\buildrel{\prime\prime}\over{.} 31\,\pm 0\buildrel{\prime\prime}\over{.} 02$, ${PA}\sim 127^\circ$ by the 2D Gaussian fit to the total integrated velocity map (see Figure 1, lower panel). Because this is as large as the clean beam size of the H30α recombination line, the beam-deconvolved source size is estimated to be much less than the beam size. The peak position is also estimated to be ${\alpha }_{\mathrm{ICRS}}={17}^{{\rm{h}}}{45}^{{\rm{m}}}39\buildrel{\rm{s}}\over{.} 785\,\pm 0\buildrel{\rm{s}}\over{.} 001$, ${\delta }_{\mathrm{ICRS}}=-29^\circ 00^{\prime} 29\buildrel{\prime\prime}\over{.} 84\,\pm 0\buildrel{\prime\prime}\over{.} 01$. This position is almost equal to that of IRS13E3 in the 340 GHz continuum map within the FWHM beam size. The positional differences are ${\rm{\Delta }}\alpha \sim 0\buildrel{\prime\prime}\over{.} 005$ and ${\rm{\Delta }}\delta \sim 0\buildrel{\prime\prime}\over{.} 028$ according to the comparison between the 340 GHz continuum and H30α moment 0 positions of IRS16NE, of which images are compact in both of the maps (see Figure 1). The positional correspondence between both the observation is better than 10% of the FWHM beam size of the H30α moment 0 map ($\sim 0\buildrel{\prime\prime}\over{.} 4$).

Because IRS13E3 is emitting the 340 GHz continuum through thin free–free emission mechanism as mentioned above, IRS13E3 is emitting the H30α recombination line simultaneously. The area emitting the recombination line should be identical to IRS13E3 itself, as shown in the continuum map (see Figure 2). The size of the ionized gas would be ${\theta }_{\mathrm{maj}.\mathrm{obs}.}\times {\theta }_{\min .\mathrm{obs}.}=0\buildrel{\prime\prime}\over{.} 102\times 0\buildrel{\prime\prime}\over{.} 090$, ${PA}\sim 27^\circ$ or ${r}_{\mathrm{maj}}\times {r}_{\min .\mathrm{obs}.}=0.0020$ pc $\times 0.0018$ pc (400 au $\times \,350$ au). Although our estimated radius, r, is an upper limit, this radius is considered to be close to the real radius because the continuum observation of IRS13E3 with JVLA at 34 GHz shows a similar source size; $0\buildrel{\prime\prime}\over{.} 08\times 0\buildrel{\prime\prime}\over{.} 04$ (Yusef-Zadeh et al. 2014). The compactness and large velocity width suggest the presence of an IMBH in the IRS13E complex.

There is a bright ridge connecting IRS13E3 with the extended component around $\sim 2^{\prime\prime}$ south of IRS13E3 in Figure 4(a). This corresponds to the curved ridge from ${V}_{\mathrm{LSR}}\sim -300$ to +250 km s⁻¹ ("A") in the position–velocity diagram of Figure 4(b). In addition, there is also a weak component at ${V}_{\mathrm{LSR}}\sim -350$ km s⁻¹ at the offset of $\sim -0\buildrel{\prime\prime}\over{.} 5$ ("D"; white arrow) in the position–velocity diagram. The weak component is identified as a compact component at $0\buildrel{\prime\prime}\over{.} 8$ south of IRS13E3 in the channel map of ${V}_{{\rm{c}},\mathrm{LSR}}=-340$ km s⁻¹, as mentioned in the previous section (see Figure 3). These observed features suggest the presence of a Keplerian orbit with high eccentricity around IRS13E3. Such Keplerian orbits have a nearly linear part with a large velocity gradient, and double-curved ridges with a relatively small velocity gradient, in the position–velocity diagram (see Figure 12 in Tsuboi et al. 2017). In this case, the line intensity of the orbit decreases with increasing velocity, as shown in Figure 4(b), because the line intensity should be in inverse proportion to the orbital velocity. Note that the thermal velocity broadening for ionized Hydrogen gas of 10⁴ K is ${\rm{\Delta }}v\sim 20$ km s⁻¹ in FWHM, which is negligible compared with the rotation velocity.

Another hypothesis to explain the weak component "D" may be that the component is the counterpart in the He30α recombination line. However, the intensity ratio of the bright ridge and the weak component is up to ∼0.3. This is too high, as $\tfrac{I(\mathrm{He}30\alpha )}{I({\rm{H}}30\alpha )}$. In addition, the velocity shift between these components is ${\rm{\Delta }}v\sim -100$ km s⁻¹ which is less than the velocity difference between the H30α and He30α recombination lines, ${\rm{\Delta }}v\sim -122$ km s⁻¹, as mentioned previously. Therefore, the weak component would be a part of the Keplerian orbit with a high eccentricity around IRS13E3.

An example of the orbits describing well the observed features is shown in the position–velocity diagram (see the black broken line in Figure 4(b)). Note that it is difficult to determine exclusively the accurate orbital parameters by fitting in the position–velocity diagram, because the full orbit is not completely occupied by ionized gas and/or the observed features do not always belong to a single orbit. The apoastron of the orbit would be located in the bright extended ionized gas component seen $\sim 2^{\prime\prime}$ south of the IRS13E complex in projection. The angle between the direction of the semimajor axis and the line of sight is ${PA}\sim 60^\circ$. The observed major axis is $2a\sin 60^\circ \sim 2.5\times {10}^{17}$ cm. Then the semimajor axis of the orbit is estimated to be $a\sim 1.4\times {10}^{17}\,\mathrm{cm}=1\times {10}^{4}\,\mathrm{au}$. The inclination angle of the orbit would be $i\sim 0^\circ$ because of the elongated feature with a narrow width of orbiting ionized gas shown in Figure 4(a). Comparing the observed shape in the position–velocity diagram and the calculated Keplerian orbits, the eccentricity of the orbit is estimated to be $e\sim 0.97$.

The presence of an IMBH in the IRS13E complex is strongly supported by a large enclosed mass of IRS13E3. In the case of nearly edge-on view, the velocity width of the Keplerian orbit is estimated by

$$\begin{eqnarray}&&{\rm{\Delta }}V\sim \left(\sqrt{\displaystyle \frac{1+e}{1-e}}+\sqrt{\displaystyle \frac{1-e}{1+e}}\right)\sqrt{\displaystyle \frac{{GM}}{a}}.\end{eqnarray} \tag{ 1 }$$

Then the enclosed mass is estimated to be

$$\begin{eqnarray}&&M\sim a{\rm{\Delta }}{V}^{2}{\left(\sqrt{\displaystyle \frac{1+e}{1-e}}+\sqrt{\displaystyle \frac{1-e}{1+e}}\right)}^{-2}{G}^{-1}.\end{eqnarray} \tag{ 2 }$$

The observed velocity width is ${\rm{\Delta }}V\sim 600$ km s⁻¹, and the enclosed mass of the object is estimated to be $M\sim (4\mbox{--}7)\times {10}^{4}$ ${M}_{\odot }$, comparable to the mass of an IMBH, for $e=0.96\mbox{--}0.98$ and $a\sim 1.4\times {10}^{17}$ cm. Although the Keplerian orbit has large ambiguities in the orbit parameters, the mass would be consistent with M ∼ 10⁴ ${M}_{\odot }$ in the IRS13E complex (e.g., Maillard et al. 2004; Schödel et al. 2005; Paumard et al. 2006). Even if the compact ionized gas does not belong to the Keplerian orbit, the compactness and large velocity width would indicate ∼10⁴ ${M}_{\odot }$ in the IRS13E complex.

If there is an IMBH orbiting Sgr A^*, the position of Sgr A^* must be affected by it. The position of Sgr A^* on the celestial sphere has been monitored using VLBA for over 15 years (e.g., Reid & Brunthaler 2004). The positional shift along the Galactic plane is as large as −6.4 mas year⁻¹. However, one can not distinguish between the perturbation by the massive object and the proper motion by the Galactic rotation. On the other hand, the positional shift crossing the Galactic plane is found to be as small as −0.2 mas year⁻¹. This indicates that the upper limit mass of the second black hole is $M\lesssim {10}^{4}$ ${M}_{\odot }$ in the area of $r\sim {10}^{3}\mbox{--}{10}^{5}\,\mathrm{au}$ from Sgr A^* (Reid & Brunthaler 2004). Therefore, there is no conflict between our derived enclosed mass and the upper limit mass of the second black hole inferred from the VLBA observations.

The periastron distance is estimated to be $a(1-e)\,=4.3\times {10}^{15}$ cm and the projected distance from Sgr A^* is ${r}_{\mathrm{proj}.}\sim 4\times {10}^{17}$ cm. Because the mass of the IMBH is only 1% of the mass of Sgr A^*, the gravity of the IMBH associated with IRS13E3 is comparable to that of Sgr A^*, even around the periastron of the Keplerian orbit around IRS13E3. However, the flow of the ionized gas, which is extended to the south, seems to be described by a Keplerian orbit as mentioned above. This means that the gravity of the IMBH would be dominant there, and the IMBH might be located fairly far from Sgr A^* than the projected distance.

The ionized gas mass on the orbit is estimated to be $M\sim 6\times {10}^{-3}\,{M}_{\odot }$ based on this observation assuming the electron temperature of ${T}_{{\rm{e}}}\sim 1\times {10}^{4}$ K and the electron density of ${n}_{{\rm{e}}}\sim 1\times {10}^{6}$ cm⁻³ (e.g., Murchikova 2017; Tsuboi et al. 2017). When all of the ionized gas falls to the IMBH associated with IRS13E3 within one orbital period, $T=2\pi \sqrt{\tfrac{{a}^{3}}{{GM}}}\sim 4\times {10}^{3}\,\mathrm{years}$, the upper limit of the mass accretion rate is estimated to be $M/T\lesssim 1\times {10}^{-6}$ ${M}_{\odot }$ yr⁻¹. However, because both the gas approaching IRS13E3 (probably "D") and going away from it (probably "A") are observed on the orbit (see Oka et al. 2016), it would be overestimation to say that all of the gas falls into it within an orbital period. If the efficiency of the accretion is assumed to be 1% as a likely value, the mass accretion rate becomes comparable to the mass accretion rate of Sgr A^* (e.g., Quataert & Gruzinov 2000; Genzel et al. 2010).

Figure 2 also shows the position of a discrete X-ray source detected in the IRS13E complex, CXOGCJ174539.7-290029 (${\alpha }_{\mathrm{ICRS}}={17}^{{\rm{h}}}{45}^{{\rm{m}}}39\buildrel{\rm{s}}\over{.} 780$, ${\delta }_{\mathrm{ICRS}}=-29^\circ 00^{\prime} 29\buildrel{\prime\prime}\over{.} 59$; e.g., Baganoff et al. 2003; Muno et al. 2009). The statistical error and absolute uncertainty of the position are $0\buildrel{\prime\prime}\over{.} 16$ and $0\buildrel{\prime\prime}\over{.} 76$ in radius, respectively (Baganoff et al. 2003). The X-ray source is located near the 340 GHz continuum peak position of IRS13E3 with a positional difference of $\sim 0\buildrel{\prime\prime}\over{.} 2$ (see Table 1), which is smaller than the uncertainty of the X-ray observation. Such a small positional difference is generally caused by using different reference frames. Therefore, we consider that the two positions are coincident. In addition, in the case that the compact component is a part of the ionized gas orbiting around the periastron as shown previously, it is possible that the component and the X-ray source corresponding to the IMBH are observed to be up to a periastron distance apart ($\sim 0\buildrel{\prime\prime}\over{.} 3$).

The X-ray photon flux is ${F}_{2-8\mathrm{keV}}=3.092\times {10}^{-7}$ cm⁻² s⁻¹ at 1999.72 years, which is a half that of Sgr A^* (e.g., Muno et al. 2009). The spectrum of this bright X-ray source resembles that of the quiescent emission from the hot plasma around Sgr A^*, and the source has no long- and short-timescale variabilities (Muno et al. 2009). The X-ray emission should be originated in the hot plasma located at IRS13E3 through bremsstrahlung, rather than those from the usual X-ray binaries, cataclysmic variables, and so on. Thus we consider that this source is the X-ray counterpart of IRS13E3. The large X-ray photon flux may be consistent with the mass accretion rate estimated in the previous subsection.