A parsec-scale radio jet launched by the central intermediate-mass black hole in the dwarf galaxy SDSS J090613.77+561015.2?

The population of intermediate-mass black holes (IMBHs) in nearby dwarf galaxies plays an important"ground truth"role in exploring black hole formation and growth in the early Universe. In the dwarf elliptical galaxy SDSS J090613.77+561015.2 (z=0.0465), an accreting IMBH has been revealed by optical and X-ray observations. Aiming to search for possible radio core and jet associated with the IMBH, we carried out very long baseline interferometry (VLBI) observations with the European VLBI Network (EVN) at 1.66 GHz. Our imaging results show that there are two 1-mJy components with a separation of about 52 mas (projected distance 47 pc) and the more compact component is located within the 1-sigma error circle of the optical centroid from available Gaia astrometry. Based on their positions, elongated structures and relatively high brightness temperatures, as well as the absence of star-forming activity in the host galaxy, we argue that the radio morphology originates from the jet activity powered by the central IMBH. The existence of the large-scale jet implies that violent jet activity might occur in the early epochs of black hole growth and thus help to regulate the co-evolution of black holes and galaxies.


INTRODUCTION
The population of low-mass black holes (BHs) in nearby dwarf galaxies, i.e. galaxies with M g ≤ 10 9.5 M ⊙ , plays a key role in shedding light on BH formation and growth in the early Universe. One of the reasons of such the role is related to the fact that the probability of undergoing merging events in the population of dwarf galaxies is lower than for their larger counterparts, and therefore the masses of their central BHs are likely to remain close to their "birth" values (e.g. Reines & Comastri 2016;Greene, Strader & Ho 2020). These BHs with masses of 10 2 M ⊙ ≤ M bh ≤ 10 6 M ⊙ are typically classified as intermediate-mass black holes (IMBHs). ⋆ E-mail: jun.yang@chalmers.se Finding and weighing them would enable us to distinguish betwen different BH seed formation mechanisms, those involving a direct collapse at the mass level of M bh ∼ 10 4 M ⊙ and population III star seeds with a typical mass of M bh ∼ 10 2 M ⊙ (e.g., the simulation by Volonteri 2010).
At present, there are several hundred accreting IMBH candidates in dwarf galaxies that exhibit optical spectroscopic or X-ray signatures (fraction < 1 per cent, Reines, Greene & Geha 2013;Pardo et al. 2016). Among them, only a small number of IMBHs (e.g., twelve in Schutte, Reines & Greene 2019) have been identified in dwarf galaxies hosting active galactic nuclei (AGN). About 0.3 per cent of dwarf galaxies have radio counterparts (Reines et al. 2020) in the FIRST (Faint Images of the Radio Sky at Twenty Centimeters, Becker, White & Helfand 1995) survey. High-resolution very long baseline interferometry (VLBI) observations of these radio counterparts provide direct insight in their nature and mechanism of emission, involving non-thermal radio jet/outflow activity. Both jets and wide opening angle winds are major ingredients in the feedback mechanisms which are reflected in the co-evolution between BHs and galaxy bulges, i.e. the M bh -σ correlation in which σ is the stellar velocity dispersion (e.g. Greene, Strader & Ho 2020;Kormendy & Ho 2013) and the M bh -M bulge correlation (e.g., Schutte, Reines & Greene 2019). Revealing radio AGN activity would enable us to probe the feedback of the IMBHs (e.g. Greene, Strader & Ho 2020;Manzano-King, Canalizo & Sales 2019). In addition, VLBI detections of compact radio cores would provide data points for filling the gap between supermassive and stellar-mass BHs for testing the mass-dependent relations (e.g. the fundamental plane relation, Yuan & Narayan 2014) and allow us to search for off-nuclear IMBHs (Reines et al. 2020) formed by galaxy mergers (e.g. Bellovary et al. 2019). To date, there are some high-resolution imaging observations of IMBHs, e.g. POX 52 (Thornton et al. 2008), Henize 2-10 (Reines & Deller 2012) and NGC 404 (Paragi et al. 2014). However, radio jets or steady radio-emitting polar outflows, compact on sub-pc scales, have been revealed in only one dwarf galaxy, NGC 4395 (Wrobel & Ho 2006).
The dwarf elliptical galaxy SDSS J090613.77+561015.2 at the redshift z = 0.0465 hosts an AGN and has a stellar mass of 2.3 × 10 9 M ⊙ (source ID: 9, Reines, Greene & Geha 2013). It displays not only some narrow-line AGN signatures but also a persistent broad Hα emission (source ID: RGG 9, Baldassare et al. 2016). Based on high spectral resolution observations, Baldassare et al. (2016) estimated the mass of its BH as M bh = 3.6 +5.9 −2.3 × 10 5 M ⊙ (including the systematic uncertainty of 0.42 dex). In the long-slit spectroscopy with the Keck I telescope, it shows some spatially extended ionized gas outflows that are most likely AGN-driven because their velocity (701±7 km s −1 ) exceeds the escape velocity (303±35 km s −1 ) of its halo (Manzano-King, Canalizo & Sales 2019). It is a slightly resolved point-like source with total the flux density of 22.4±4.1 mJy in the the GMRT (Giant Metrewave Radio Telescope) 150 MHz all-sky radio survey (Intema et al. 2017) and 4.7±0.2 mJy in the 1.4 GHz FIRST survey (Becker, White & Helfand 1995). It has a flux density of 0.93±0.05 mJy at 9.0 GHz and 0.78±0.04 mJy at 10.65 GHz in the Karl G. Jansky Very Large Array (VLA) observations (source ID: 26, Reines et al. 2020).
This Letter is composed as follows. In Section 2, we describe our European VLBI Network (EVN) observations and data reduction. In Section 3, we present the high-resolution imaging results. In Section 4, we discuss the physical nature of the detected components and the implication from our findings. Throughout the paper, a standard ΛCDM cosmological model with H 0 = 71 km s −1 Mpc −1 , Ω m = 0.27, Ω Λ = 0.73 is adopted; the VLBI images then have a scale of 0.9 pc mas −1 . Both experiments were performed in the phasereferencing mode to gain the calibration solutions and a reference position for our faint target. The bright source J0854+5757, about 2. • 4 apart from SDSS J090613.77+561015.2 and a key source in the International Celestial Reference Frame (Ma et al. 1998), was observed periodically as the phase-referencing calibrator. The calibrator has a J2000 poisition at RA = 08 h 54 m 41. s 996408 (σ ra = 0.2 mas), Dec. = 57 • 57 ′ 29. ′′ 93914 (σ dec = 0.1 mas) in the source catalogue of 2015a from the Goddard Space Flight Centre VLBI group 1 . The calibrator position has an offset of 0.7 mas with respect to the optical position in the second data release (DR2, Brown et al. 2018) of the Gaia mission (Prusti et al. 2016). The nodding observations used a dutycycle period of about five minutes (1 min for J0854+5757, 3 min for SDSS J090613.77+561015.2, 1 min for two scan gaps). During the observations, the sources was at an elevation of ≥18 • at all European telescopes.
The visibility data were calibrated using the National Radio Astronomy Observatory (NRAO) Astronomical Image Processing System (aips version 31DEC17, Greisen 2003) software package. We removed the visibility data of side channels because of their low correlation amplitude while loading the data into aips and then ran the task accor to re-normalize the correlation amplitude. A priori amplitude calibration was performed with the system temperatures and the antenna gain curves if provided by the telescopes. If these data were missing, the nominal values were used. The ionospheric dispersive delays were corrected according to the map of total electron content provided by the Global Positioning System (GPS) satellite observations. Phase errors due to the antenna parallactic angle variations were removed. After a manual phase calibration was carried out, the global fringe-fitting and bandpass calibration were performed.
The calibrator J0854+5757 was imaged using iterations of model fitting with a group of point sources (delta functions) and self-calibration (Stokes I ) in the software package difmap (version 2.5e, Shepherd, Pearson & Taylor 1994), fringe-fitting and self-calibration (Stokes RR and LL) in aips. The calibrator had a single-sided core-jet structure with a total flux density of 0.61±0.03 Jy in the first observing epoch. Due to a firmware bug in the European digital baseband converters, there were significant sensitivity losses in the 2nd epoch. According to the long-term light curve at 15 GHz observed by the 40-meter telescope at the Owens Valley Radio Observatory (Richards et al. 2011) and published online 2 , the calibrator had stable flux densities in 2017. Assuming that the calibrator had no flux density variation between the two epochs, we derived an amplitude correction factor of 1.63 in the imaging procedures. We used the jet base, the brightest component, as the reference point in the phase-referencing calibration. After about ten iterations, the deconvolved map of Stokes I using natural weighting reached an image noise level of 0.016 mJy beam −1 , as low as the map of zero-flux-density Stokes V. The core-jet structure in the phase-referencing calibrator J0854+5757 observed in the second epoch is shown in Fig. 1. In the final high dynamic range image, 82 positive point sources were used. Both the phase and amplitude self-calibration solutions were also transferred and applied to the target source. In the residual map of the target source, there are no clearly-seen systematic errors (noise peaks, strips and rings). This indicates that the phase-referencing calibration worked properly.

RADIO STRUCTURE IN SDSS
J090613.77+561015.2 The full-EVN image of SDSS J090613.77+561015.2 obtained on 2017 November 6 is shown in Fig. 2. There are two components detected and labelled as N and S in the clean map. The optical centroid, reported by the Gaia DR2 (Brown et al. 2018), is marked as a yellow cross (J2000, RA = 09 h 06 m 13. s 77047, Dec= 56 • 10 ′ 15. ′′ 1482, σ ra = σ dec = 0.8 mas, the astrometric excess noise to set the reduced χ 2 r = 1 is 7.0 mas). The large excess noise is most likely related to the extended optical morphology and a certain level of asymmetric brightness distribution in the bulge (Schutte, Reines & Greene 2019). The total error, added in quadrature, is shown as a dotted yellow circle. The two components are also detected in the first short e-EVN observations. Due to its relatively low image quality, that image is not shown here.
In order to determine the parameters of the obtained brightness distribution, we fitted two circular Gaussian components to the visibility data using difmap. The modelfitting results including the 1σ formal uncertainties at the reduced χ 2 r = 1 are listed in Table 2. The systematic error of S pk , S obs and L R are about ten per cent. Using J0854+5757 as reference, we estimate the coordinates of the component N as RA= 09 h 06 m 13. s 77005 and Dec= 56 • 10 ′ 15. ′′ 1429 with a total systematic error < 1 mas. The component N is partly resolved and has a faint (∼0.2 mJy) extension of about 4 mas toward south. The separation between the components N and S is 52.3±0.3 mas. With respect to the component N, the component S has a position angle of about −154 • .
All the available to date total flux density measurements of the source SDSS J090613.77+561015.2 are plotted in Fig. 3. Assuming no flux density variability, we fit the blue data points collected from the literature (Becker, White & Helfand 1995;Intema et al. 2017;Reines et al. 2020) to a power-law spectrum S ν = S 0 ν α and determine S 0 = 5.94±0.39 mJy, the spectral index α = −0.84± 0.04. According to this model, SDSS J090613.77+561015.2 has a total flux density of 3.9±0.3 mJy at 1.66 GHz. Compared to this estimate, the high-resolution EVN image recovers only about 50 per cent. We also tried to search for diffuse radio emission. On the shortest baseline Ef-Wb, there is a hint for a faint and diffuse structure connecting the two components and extending farther on both sides. However, because of the lack of the shorter baselines, it is hard to make a reliable image for the diffuse structure from the available data. The existence of the diffuse radio structure is also expected since the source is slightly resolved (deconvovled FWHM: 2. ′′ 1 × 1. ′′ 1 in the major axis position angle 40 • ) in the VLA FIRST image (Becker, White & Helfand 1995) and the elongation direction is roughly consistent with the overall EVN morphology extent.

Epoch
Name S pk S obs ∆α cos δ ∆δ  Figure 2. A two-component brightness distribution found by the EVN at 1.66 GHz in the dwarf galaxy SDSS J090613.77+561015.2 hosting an accreting IMBH. The yellow cross and circle mark the optical (Gaia DR2) centroid and the total 1σ error, respectively. The contours start at 0.048 mJy beam −1 (3σ) and increase by a factor of two. The peak brightness is 0.76 mJy beam −1 . The restoring beam is 5.81 mas × 4.05 mas (FWHM) at −3. • 03 position angle and plotted in the bottom-left corner.

1982)
, where S obs is the observed total flux density in mJy, ν obs is the observing frequency in GHz, θ size is the full width at half-maximum (FWHM) of the circular Gaussian model in mas, and z is the redshift. The components N and S have average brightness temperatures of 1.2×10 8 K and 7.1×10 6 K, respectively, at 1.66 GHz in the second-epoch 12-h full-EVN observations. Due to the very limited (u, v) coverage and the low image sensitivity, the component S has an underestimated θ size and thus an overestimated T b in the first-epoch 2-hour-long e-EVN observation.  The broad-band radio spectrum of SDSS J090613.77+561015.2. The blue points are from the earlier total flux density observations by the VLA (Becker, White & Helfand 1995;Reines et al. 2020) and the GMRT (Intema et al. 2017). The two red points are from our high-resolution EVN observations. The black line shows the best-fit power-law spectrum to the low-resolution total flux density measurements (blue points).

The nature of the components N and S
The only plausible explanation of the radio structure in SDSS J090613.77+561015.2 appears to be an AGN manifestation. The optical spectroscopic observations of SDSS J090613.77+561015.2 show that there are no signatures for on-going star-forming activity in the BPT (Baldwin, Phillips & Terlevich 1981) diagrams formed with some emission line ratios (Reines et al. 2020). The persistent broad Hα emission is consistent with an AGN origin (Baldassare et al. 2016). Therefore, we reject a possibility that the observed components represent a superposition of supernova remnants (SNRs) like in, e.g., Arp 220 (Varenius et al. 2019). Moreover, they cannot be explained as two young radio supernovae because of their radio structures resolved on the pc scales although their radio luminosities (L R ∼ 8 × 10 37 erg s −1 ) are in the luminosity range of young radio supernovae (maximum: L R ∼ 5 × 10 38 erg s −1 , Weiler et al. 2002).
The component N is either the radio core, i.e. the jet base, or a newly-emerging jet component. Its proximity (within the 1σ positional error) to the optical centroid of SDSS J090613.77+561015.2 is consistent with this hypothesis. There also exists a jet-like faint extension toward South in the component N when the image resolution in North-South is slightly improved to 4 mas. We can also fit the component N to two point sources with flux densities 0.73±0.02 and 0.19±0.02 mJy and a separation of 3.9±0.4 mas.
The component S is most likely an expanding ejecta. Because of its large distance (∼58.5 mas) to the Gaia centroid, it cannot be explained as the radio core. Compared to the component N, the component S has the more extended structure and the lower brightness temperature. Moreover, its position and elongation (position angle about 36 • ) are roughly aligned with the extension of the component N.
The Gaia positioning of the optical centroid close to the radio component N provides a strong indication on the nature of this component as the compact radio core. However, we cannot rule out that SDSS J090613.77+561015.2 is a young compact symmetric object (CSO, Wilkinson et al. 1994). In this scenario, the radio components N and S are a pair of moving-out radio ejectae (or mini-lobes) within its host galaxy, and the radio core is located somewhere inâȂŞbetween and undetected. The latter would be still consistent with the Gaia position within its 3σ error. Assuming a typical separation speed of 0.2c among CSOs (e.g. An et al. 2012), the pair of components would have a kinematic lifetime of ∼7.5×10 2 yr. This is a typical value in young extragalactic radio sources (e.g. An & Baan 2012). Steep spectra at > ∼ 1 GHz are not unknown among CSOs, e.g. J0132+5620 and J2203+1007 . Another example is the CSO PKS 1117+146, which has a spectral index of −0.7 (Torniainen et al. 2007) and a large angular separation between the opposite jet components,∼70 mas (Bondi et al. 1998), and thus is similar to SDSS J090613.77+561015.2. A conclusive test on possible CSO identification of the source will be provided by its future multi-frequency and multiepoch VLBI studies.

Implications of the presence of a radio jet associated with the IMBH
If the component N is the stationary radio core associated with the IMBH, SDSS J090613.77+561015.2 will have a relatively high radio luminosity. The source has an X-ray luminosity of L X = 4.5 × 10 40 erg s −1 in the 2-7 keV band (Baldassare et al. 2017). This leads to a ratio of L R L X = 1.8 × 10 −2 , much higher than the typical value ∼ 10 −5 in the radioquiet Palomar-Green quasar sample (Laor & Behar 2008). Due to the high ratio, it is hard to explain them as wideangle radio-emitting winds. At the low accretion rate state, there exists a correlation (e.g. Merloni, Heinz & Di Matteo 2003) between the radio core luminosity at 5 GHz, the Xray luminosity (L X ) in the 2-10 keV band, and the BH mass (M bh ): log L R = (0.60 +0.11 −0.11 ) log L X + (0.78 +0.11 −0.09 ) log M bh + 7.33 +4.05 −4.07 (2) According to Equation 2, we would expect L R = 10 36.1±0.9 erg s −1 . This estimate would be one order of magnitude lower while still in the acceptable range in view of the large scatter of the correlation. The radio jet is rarely-seen in dwarf galaxies. Compared to the first and only case, NGC 4395 (Wrobel & Ho 2006), SDSS J090613.77+561015.2 has a jet 160 times longer and 10 5 times more luminous. The finding of the large jet structure indicates that violent ejections might appear at the BH growth stage of M bh ∼10 5 M ⊙ in the early Universe. Most of these jets associated with low-mass BHs might be short-lived and sub-kpc objects because they have rather low radio luminosities (L R ≤ 10 41 erg s −1 at 1.4 GHz, Kunert-Bajraszewska et al. 2010;An & Baan 2012).
Manzano-King, Canalizo & Sales (2019) report some AGN-driven outflows in dwarf galaxies, indicating significant AGN impact on the large-scale kinematics and gas content. The kpc-scale high-velocity ionized gas outflows in SDSS J090613.77+561015.2 might be driven by the AGN jet activity, in particular the diffuse structure completely resolved out in our EVN image because of missing the shorter baselines. The unseen kpc-scale (relic) jet component has a flux density of 2.0±0.2 mJy. According the scaling relation P jet = 5.8 × 10 43 ( L R 10 40 ) 0.70 erg s −1 between jet kinematic power and the radio luminosity derived by Cavagnolo et al. (2010) using Chandra X-ray and VLA radio data of radio galaxies, the unseen relic jet has a power of P jet = 10 42.6±0.7 erg s −1 , reaching about ten percent of the Eddington luminosity L Edd = 10 43.6±0.4 erg s −1 of the IMBH. Thus, the AGN jet activity might have significant impact on the host galaxy.