Free Access
Volume 630, October 2019
Article Number L5
Number of page(s) 4
Section Letters to the Editor
Published online 23 September 2019

© ESO 2019

1. Introduction

According to currently accepted cosmological and structure formation models, galaxies grow through frequent mergers (e.g. Volonteri et al. 2003). During these events, supermassive black holes (SMBHs) residing in merging galaxies shrink to the central region of ∼1 kpc losing energy by dynamical friction. There, a bound SMBH pair may form, which eventually will be able to emit gravitational waves before their final coalescence (Begelman et al. 1980). Detection of an SMBH is relatively straightforward if it is actively accreting matter from its surrounding as in an active galactic nucleus (AGN). According to numerical simulations (e.g. Capelo et al. 2015; Blecha et al. 2018), the simultaneous activity in the nuclei of merging galaxies is expected at a separation of ≲10 kpc.

Despite various efforts, efficient observational selection of multiple AGN candidates has so far been elusive. Satyapal et al. (2017) proposed a new selection method to find possible dual AGNs. These authors selected galaxies from the Sloan Digital Sky Survey (SDSS) experiencing interactions or mergers according to their optical images. This sample is cross-correlated with the infrared AllWISE catalogue1 (Cutri et al. 2012) to identify those sources in which AGNs can be present. The authors employ a colour cut using the two shortest wavelength observing bands, W1 − W2 >  0.5, where W1 and W2 are the 3.4 μm and 4.6 μm observing bands, respectively, to ensure the selection of AGN candidates. This colour-cut has been shown to be the most effective at finding AGNs in late stage mergers based on hydrodynamic simulations (Blecha et al. 2018). Satyapal et al. (2017) conducted X-ray follow-up observations of selected sources from this sample, for which the separation of the assumed AGN can be resolved spatially by the Chandra X-ray satellite. In their pilot study, four sources from the observed six showed two distinct X-ray components, indicating that they possibly host a dual AGN at separations < 10 kpc. In a follow-up study, Pfeifle et al. (2019a) found that 8 out of 15 merging galaxy systems had multiple nuclear X-ray sources suggestive of a dual AGN.

One of those sources, J0849+1114, is a system of three interacting galaxies within ∼5″ projected separation (Fig. 1), all of which exhibit nuclear X-ray sources with 2 − 10 keV luminosities in excess of the expected star formation and show optical line ratios consistent with AGNs photo-ionisation processes (Pfeifle et al. 2019a,b). In the 12th data release of the SDSS (SDSS DR12; Alam et al. 2015) Galaxy 1 is listed as SDSS J084905.51+111447.2 and Galaxy 3 is listed as SDSS J084905.43+111450.9.

thumbnail Fig. 1.

Archival Hubble Space Telescope WFC3 F105W image of J0849+1114 from Pfeifle et al. (2019b). The three interacting galaxies are indicated.

Open with DEXTER

Recently, Liu et al. (2019) also reported on the existence of three AGNs in J0849+1114. According to the authors, all three stellar nuclei can be classified as Seyfert 2 galaxies. They detected radio emission features at 9 GHz with the Karl G. Jansky Very Large Array (VLA) at the positions of Galaxy 1 and Galaxy 3 (Fig. 1).

We report on the results of the milliarcsec scale resolution very long baseline interferometry (VLBI) observation of J0849+1114 performed by the European VLBI Network (EVN). In the following, we define the radio spectral index, α as S ∝ να, where S is the flux density and ν is the observing frequency. We assume a flat ΛCDM cosmological model with H0 = 70 km s−1 Mpc−1, Ωm = 0.27, and ΩΛ = 0.73. At the redshift of J0849+1114, z = 0.077, 1 mas angular separation corresponds to 1.46 pc projected linear distance (Wright 2006).

2. Observation and data reduction

An exploratory EVN observation of J0849+1114 was conducted on 2019 Jan. 22 (project code: RS09b) at ν = 1.7 GHz. The names, diameters, and observing times of the participating antennas are given in Table 1. At the end, data from all antennas except Medicina and Irbene were used to create the final map of the source. Eight intermediate frequency channels (IFs) each with 16 MHz bandwidth were used in both left and right circular polarisation. Each IF was divided into 32 spectral channels. The observation was carried out in e-VLBI mode (Szomoru 2008). The data from the antenna sites were transferred through optical fibre connection to the Joint Institute for VLBI European Research Infrastructure Consortium, Dwingeloo (in the Netherlands), where they were correlated with an integration time of 2 s.

Table 1.

Radio telescopes of the EVN that participated in the observation.

The observation lasted for ∼2 h, and the on-source integration time was 55 min. The observation was conducted in phase-reference mode (Beasley & Conway 1995), using J0851+0845 as the phase-reference calibrator (target separation: ). An additional phase-reference candidate source, J0850+1108 (target separation: ) was also included in the observation. For the target source, we used the coordinates from the Faint Images of the Radio Sky at Twenty-centimeters survey (FIRST; Helfand et al. 2015) as phase centre, right ascension , and declination 11 ° 14′48″.

We followed standard procedures (Diamond 1995) using the National Radio Astronomy Observatory (NRAO) Astronomical Image Processing System (AIPS; Greisen 2003) for the calibration. The amplitudes of the interferometric visibilities were calibrated using the antenna gain curves and the system temperatures measured at the telescope sites2. Fringe-fitting was performed for the phase calibrator. The fringe-fitted data of the phase calibrator source were imaged using the Difmap software package (Shepherd et al. 1994) following standard hybrid mapping procedure, which included several cycles of CLEANing (Högbom 1974), phase-only self-calibration, and finally amplitude self-calibration. The antenna-based gain correction factors determined in Difmap were applied to the data in AIPS3. A new fringe-fit of the calibrator was performed, now taking into account the CLEAN model components describing its brightness distribution, to reduce the small residual phase errors due to its structure. The obtained solutions were applied to the target source J0849+1114 and to the candidate secondary phase-reference source, J0850+1108. Then both of these were imaged in Difmap.

To obtain the lowest noise-level image of J0849+1114, natural weighting was used and instead of several point-like CLEAN components, a single circular Gaussian brightness distribution model was fitted to the visibility data. The value of the reduced χ2, 1.4, indicates an adequate fit to the data. The resulting image of J0849+1114 is shown in Fig. 2.

thumbnail Fig. 2.

Image of J0849+1114 taken on 2019 Jan. 22 at 1.7 GHz with the EVN. The peak corresponds to the location of Galaxy 1. The peak brightness is 0.6 mJy beam−1; the contours are drawn at ( − 2.4, 2.4, 5, 7.2)×64 μJy beam−1 (1σ image noise level). The restoring beam is 4 mas × 2.6 mas (FWHM) with a major axis position angle as shown in the lower left corner of the image.

Open with DEXTER

Because of the faintness of the target source, no self-calibration was attempted. The imperfect phase correction can result in a coherence loss, which can lead to a flux density loss in the detected source. To estimate the amount of coherence loss, we turned to the candidate phase calibrator source, J0850+1108. We mapped the dataset obtained after fringe-fitting of J0850+1108 and that obtained by phase solution transfer from the phase calibrator. In the latter case, the recovered flux density was ∼25% lower compared to fringe-fitting. This is in agreement with typical coherence loss values (e.g. Mosoni et al. 2006).

3. Results

We detected a radio feature slightly extended to the north in J0849+1114. Using the AIPS task MAXFIT, we derived the position of the brightness peak, right ascension , and declination . The positional uncertainty of the phase calibrator source is 0.27 mas4. By far the dominant contribution to the positional uncertainty (∼3 mas) is caused by the angular separation between the target and the phase-reference source (e.g. Chatterjee et al. 2004; Rioja et al. 2017). Thus, the derived coordinates are accurate within ∼3 mas.

The integrated flux density of the fitted Gaussian brightness distribution is 3.7 ± 0.1 mJy. After accounting for the coherence loss, the flux density of the radio feature is S = 5.0 ± 0.1 mJy. In the following, we use this flux density value. The full width at half maximum (FWHM) size of the Gaussian model component, θ = 8.3 ± 0.2 mas (corresponding to ∼12 pc projected size), is larger than the smallest resolvable size of this observation (1.3 mas; Kovalev et al. 2005), therefore it can be used to derive the brightness temperature of the milliarcsec-scale radio emitting feature as follows:


where S is measured in jansky, θ in milliarcsec, ν is the observing frequency in gigahertz. The resulting brightness temperature, 3.3 × 107 K, is two orders of magnitude higher than the brightness temperature limit measured above 1 GHz for star-forming galaxies (105 K; Condon 1992).

Assuming a flat radio spectrum (α = 0), the 1.7 GHz radio power is 6.7 × 1022 W Hz−1. Owing to the small redshift of the source, the spectral index assumption does not significantly influence the obtained radio power. According to Kewley et al. (2000) and Middelberg et al. (2011), the radio power of an AGN core exceeds 2 × 1021 W Hz−1. Few violent star-forming galaxies are known to have supernova complexes exceeding this value; two examples of these are Arp 299-A and Arp-220, which have a 1.7 GHz power of 4.5 × 1021 W Hz−1 and 1.6 × 1022 W Hz−1, respectively (Alexandroff et al. 2012, and references therein). However, the radio power of the detected feature J0849+1114 surpasses these values as well.

We did not detect any other milliarcsec-scale radio feature down to a 7σ noise level of 0.45 mJy beam−1 within a region of 8″ × 8″ around the source that covers the cores of all galaxies in the interacting system. According to the EVN sensitivity calculator5, bandwidth smearing and time-average smearing (Wrobel 1995) limit the undistorted field of view of our EVN observation to an area with radii and , respectively. Thus, all three galaxy cores are within the observed field, where the smearing effects would not reduce the response to a point source significantly; the expected flux density loss is < 10%.

4. Discussion

The detected milliarcsec-scale radio feature (Fig. 2) is positionally coincident with Galaxy 1 of Pfeifle et al. (2019a). Its high brightness temperature and 1.7 GHz radio power indicate that the radio emission originates from an AGN. Thus, we can confirm there is a radio-emitting AGN in Galaxy 1.

Using the fundamental plane of black hole activity (Gültekin et al. 2019), the black hole mass can be estimated from the 2 − 10 keV X-ray (LX) and the 5 GHz radio luminosity. Pfeifle et al. (2019b) gave LX = (2.37 ± 0.17)×1041 erg s−1 for the core of Galaxy 1 using a combination of Chandra observations conducted in 2013 and 2016. To estimate the 5 GHz luminosity of the milliarcsec-scale radio core, we used the spectral index range found by Giroletti & Panessa (2009), −0.7 ≤ α ≤ 0.1, for the milliarcsec-scale cores of faint Seyfert galaxies. The obtained black hole mass estimates range between ∼4.6 × 108M (for α = −0.7) and ∼2.9 × 109M (for α = 0.1). Pfeifle et al. (2019b) reported the detection of broad Paα emission lines in Galaxies 1 and 3. According to the relationship between the emission line and the black hole mass (Landt et al. 2013), the measured value implies a black hole mass of 3.2 × 107M for Galaxy 1. Since the internal scatter of both relationships are 1 dex, and because of the non-simultaneous X-ray and radio observations the estimate from the fundamental plane can also be affected by brightness variability, we can regard the obtained black hole masses as not incompatible. These estimates all indicate that the measured X-ray, radio, and Paα emission line characteristics can be explained by an SMBH residing in the core of Galaxy 1.

The recovered flux density in our EVN observation, S = 5 mJy is much less than that measured at a close frequency, i.e. 1.4 GHz within the FIRST survey (Helfand et al. 2015), SFIRST = 35.4 ± 0.1 mJy. While it could also be related to source variability, it is most likely that the missing ∼30 mJy flux density is in the more extended structure for which our EVN observation is not sensitive. The largest recoverable feature in an interferometric observation is defined by the shortest baseline in the array (Wrobel 1995). In our case, this is the Effelsberg−Westerbork baseline with ∼266 km. The corresponding largest recoverable size at 1.7 GHz is ∼70 mas, thus significant radio emission may be contained in features extending above this limit.

The low resolution of FIRST (∼5″) does not allow us to relate the radio emission to any of the three interacting galaxies. The 1.4 GHz radio power corresponding to the flux density not detected by our EVN observation (∼30 mJy) is ∼4 × 1023 W Hz−1. If we assume that this is caused only by star formation in the galaxies, it implies a summed star formation rate of ∼222 M yr−1 (Hopkins et al. 2003). Izotov et al. (2014), who used multi-wavelength data to derive the spectral energy distribution and star formation rate for a large sample of galaxies, obtained a star formation rate of 1.1 M yr−1 for the galaxy group J0849+1114. Pfeifle et al. (2019b) estimated an upper limit for the star formation rate assuming no AGN contribution in the system. The summed value for the three galaxies is ∼15 M yr−1. These numbers indicate that star formation alone cannot be responsible for this 30 mJy flux density, and AGN-related processes must dominate the 1.4 GHz radio emission. The rms noise level of our map implies that there is no other compact radio-emitting AGN with 1.7 GHz radio power larger than ∼6 × 1021 W Hz−1 in a region encompassing the two other galaxies. However, the existence of an extremely low-power radio AGN cannot be ruled out. Even so, most of the AGN-related emission must be in large, i.e. at least 100 milliarcsec-scale lobe(s).

Liu et al. (2019) reported the results of a VLA A configuration observation of J0849+1114 at 9 GHz. They detected radio emission at the position of Galaxy 1 with a flux density of 18.65 ± 0.57 mJy and peak intensity of 5.7 mJy beam−1, and at the position of Galaxy 3 with a flux density of 2.28 ± 0.07 mJy and peak intensity of 0.5 mJy beam−1. Assuming α = 0 for the compact, milliarcsec-scale core for the AGN in Galaxy 1, and subtracting 5 mJy from the flux density measured at 9 GHz, the emission at arcsec scales is expected to be ∼16 mJy. Then the spectral index of the arcsec-scale structure between 1.7 and 9 GHz can be estimated, α ∼ −0.4. This value is close to the average spectral index, −0.5 derived for a sample of Seyfert galaxies using VLA observations by Ho & Peng (2001). Thus, the radio flux densities measured at high and low resolutions and at 1.7 and 9 GHz can be explained with a compact core and extended AGN-related emission region. However, whether the radio emission detected at the position of Galaxy 3 by Liu et al. (2019) is related to the AGN in that Seyfert galaxy, or if it is an extended jet-like feature of the radio AGN in Galaxy 1 cannot be ascertained. More sensitive VLBI, to reveal a faint radio-emitting AGN in Galaxy 3 if it exists, and/or intermediate-resolution multi-frequency radio interferometric observations, to map the structure and spectral shape of the radio emission between Galaxy 1 and 3, could answer this question.

5. Summary

We conducted 1.7 GHz high-resolution VLBI observation of the merging triple galaxy system, J0849+1114, which shows convincing multi-wavelength evidence for a triple AGN system (Pfeifle et al. 2019a,b; Liu et al. 2019). We detected a high brightness temperature, compact radio feature in Galaxy 1, thus we can confirm the existence of a radio AGN there. Compared to lower resolution radio data, a significant amount of flux density remained undetected in our EVN observation. The amount of star formation derived for the system from independent methods cannot be responsible for most of this flux density. Thus, AGN-related emission, probably from 100 milliarcsec-scale lobe-like feature(s) must contribute to this flux density. It is unclear whether the large-scale emission is related to the EVN-detected compact AGN in Galaxy 1, or it is related to the AGN in Galaxy 3 that remains undetected with the EVN. Nevertheless, the latter scenario is suggested by the 9 GHz radio observation of Liu et al. (2019) which showed arcsec-scale radio emission at the position of Galaxy 3.


Only nominal system temperature values were available for the Jodrell Bank Mk2 and the Toruń antennas.


The flux density of the phase calibrator source before and after applying the gain scales was 124 mJy and 108 mJy, respectively.


Global VLBI solution rfc_2019a catalogue, available at the website, maintained by L. Petrov.


K. É. G. was supported by the János Bolyai Research Scholarship of the Hungarian Academy of Sciences and by the Ministry of Human Capacities within the framework of the ÚNKP (New National Excellence Program). The European VLBI Network is a joint facility of independent European, African, Asian, and North American radio astronomy institutes. Scientific results from data presented in this publication are derived from the following EVN project code: RS09b. e-VLBI research infrastructure in Europe is supported by the European Union’s Seventh Framework Programme (FP7/2007-2013) under grant agreement number RI-261525 NEXPReS. This publication has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 730562 [RadioNet].


All Tables

Table 1.

Radio telescopes of the EVN that participated in the observation.

All Figures

thumbnail Fig. 1.

Archival Hubble Space Telescope WFC3 F105W image of J0849+1114 from Pfeifle et al. (2019b). The three interacting galaxies are indicated.

Open with DEXTER
In the text
thumbnail Fig. 2.

Image of J0849+1114 taken on 2019 Jan. 22 at 1.7 GHz with the EVN. The peak corresponds to the location of Galaxy 1. The peak brightness is 0.6 mJy beam−1; the contours are drawn at ( − 2.4, 2.4, 5, 7.2)×64 μJy beam−1 (1σ image noise level). The restoring beam is 4 mas × 2.6 mas (FWHM) with a major axis position angle as shown in the lower left corner of the image.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.