© The Authors 2024

1. Introduction

Observed for the first time in 1969 (Cheung et al. 1969), the most common maser emission line arises from the H₂O roto-transitional levels 6₁₆ and 5₂₃ and is emitted at 22 GHz in the radio domain. Among the extragalactic H₂O maser sources, those with an isotropic luminosity of L_iso > 10 L_⊙ are traditionally defined as megamasers; however, this threshold should be used with caution (see Sect. 4.2 in Tarchi et al. 2011). These sources are generally found in active galactic nuclei (AGNs). The majority of H₂O masers have so far been found to be associated with radio-quiet AGNs in the local Universe (z ≤ 0.05) and are classified as Seyfert 2 (Sy2) or low ionisation nuclear emission-line regions (LINERs), although there are some exceptions (e.g., Tarchi 2012; Braatz et al. 2018).

The activity of H₂O maser emission in AGNs has been associated with three main AGN components: (i) disc; (ii) jet; and (iii) nuclear outflows. Jet and outflow masers arise from the interaction between the jet(s) or outflow(s) and the encroaching molecular clouds, or because of an accidental overlap along the line of sight between a warm dense molecular cloud and the radio continuum of the jet(s) or outflow(s). Jet maser sources provide important information about the evolution of jets and their hotspots (some examples are NGC 1068 and Mrk 348; Gallimore et al. 2001; Peck et al. 2003); while nuclear outflow masers trace the velocity and geometry of nuclear winds at a few parsecs from the nucleus (e.g. Circinus; Greenhill et al. 2003). All these studies are possible thanks to high-angular-resolution measurements made with the very long baselines interferometry (VLBI) technique. Indeed, VLBI observations allow us to determine the distribution of the maser emission and the absolute position of the maser spots at milliarcsecond (mas) resolution and consequently enable us to study the geometry of the nuclear regions. In particular, by analysing the disc masers, which are associated with the nuclear accretion disc, it is possible to estimate the rotation velocity and enclosed nuclear mass, namely the mass of the supermassive black hole (SMBH; e.g., Gao et al. 2017; Pesce et al. 2020a); to obtain distances to the host galaxies (e.g., Braatz et al. 2013; Reid et al. 2013); and to provide a direct estimate of the Hubble constant, H₀ (Reid et al. 2009; Pesce et al. 2020b). The tracing of the accretion disc at subparsec scales allows us to study the co-evolution of SMBH and galaxies in the lower mass regime (Greene et al. 2016) and to constrain the spin of obscured AGNs, as recently proposed by Masini et al. (2022). The maser spectra associated with discs are identified by the presence of a three-peaked pattern, that is, of three distinct groups of features: one around the systemic velocity of the galaxy and the other two at blueshifted and redshifted velocities (in general, of the order of hundreds of km s⁻¹; e.g., Tarchi 2012). However, galaxies with only two maser line groups have been reported (e.g., Mrk 1210, Mrk1 and NGC 5728; Zhao et al. 2018; Kuo et al. 2020). Recently, Darling (2017) discussed the fact that maser radiation could also be detected via gravitational lensing or deflection by massive black holes in inclined accretion discs (more than 10 deg from edge-on). Based on this, this author suggests the existence of a new type of maser: the ‘inclined maser disc’ and one of the candidates for this class is the galaxy IC 485 (see Sect. 2). The observational signature of an inclined maser disc would be a narrow line, or line complex, at the systemic velocity of the galaxy and at the apparent location of the black hole. However, these peculiarities might also be due to other mechanisms and only VLBI observations can solve the ambiguity.

In order to determine the nature of the H₂O maser emission in the galaxy IC 485, that is, by providing the position and distribution of the maser spots at milliarcsecond resolution, we performed multi-frequency and multi-epoch Very Long Baseline Array (VLBA) and European VLBI Network (EVN) observations. Their details are reported in Sect. 3, while the results are presented in Sect. 4. We discuss our results and draw conclusions in Sects. 5 and 6, respectively. We mention here that throughout the manuscript, we adopt a cosmology with Ω_matter = 0.27, Ω_vacuum = 0.73, and H₀ = 70 km s⁻¹ Mpc⁻¹. Unless stated otherwise, the quoted velocities are calculated using the optical velocity definition in the heliocentric frame.

2. IC 485

IC 485 has been optically classified as an Sa spiral galaxy, and is located at a distance of (122.0 ± 8.5) Mpc (Kamali et al. 2017). The spectroscopic classification of the galaxy is uncertain. Darling (2017) classify IC 485 as a LINER, while Kamali et al. (2017) classify it as a Seyfert 2. Darling (2017), with the VLA, detected a broad multi-component H₂O maser emission with a peak flux of around 80 mJy, which corresponds to an isotropic luminosity of L_iso = (868 ± 46) L_⊙. Furthermore, Darling (2017) also reported the detection of an unresolved (with an angular resolution of around 90 mas ≃ 50 pc) and faint continuum radio source of I_ν = (77 ± 15) μJy beam at 20 GHz. IC 485 was also part of the FIRST¹ (S_peak = 3.0 mJy) and NVSS (1.4 GHz = 4.4 mJy; Condon et al. 2002) surveys and although the presence of an AGN cannot be excluded, its dominant energy source is believed to be star formation (see Darling 2017, and references therein). At 33 GHz, a tentative (3.5σ) radio source of 66 μJy was reported by Kamali et al. (2017). Furthermore, the galaxy was sampled in a high-sensitivity single-dish analysis in Pesce et al. (2015) and more recently in a survey with the purpose of observing the water maser transition at 183 GHz in Pesce et al. (2023).

3. Observations and data reduction

In the following, we describe our observations and calibration, and our analysis of the data at three different bands acquired in six different VLBI epochs. The main details of the observations are summarised in Table 1. We reduced and analysed the data of all epochs with the NRAO Astronomical Image Processing System (AIPS²) software, following standard procedures. Briefly, regarding the names used by the authors for the different epochs, these are preceded by the array used to observe, and so, for example, the epoch ‘VLBA 2018.16’ refers to epoch 2018.16 observed with VLBA. We introduce, here, our definition of ‘maser component’: a group of one or more features that can be fit with a single Gaussian in a certain range of velocity. This definition is adopted throughout the paper.

3.1. K Band: Spectral line observations

3.1.1. VLBA observations and data reduction

We observed the 6₁₆ → 5₂₃ H₂O maser transition towards IC 485 with eight and ten antennas of the VLBA on Febraury 26, 2018 (epoch 2018.16), and on October 30, 2018 (epoch 2018.83) under project codes BT142 and BT145 (P.I.: A. Tarchi), respectively. In epoch 2018.16, we observed with one intermediate frequency (IF) of 64 MHz (870 km s⁻¹) centred at the systemic velocity of the galaxy (V_sys = 8338 km s⁻¹), while in epoch 2018.83, we observed with two slightly overlapping IFs of 64 MHz each, placed in such a way as to span a total velocity range of ∼1400 km s⁻¹, which is enough to cover the systemic and the redshifted maser components reported by Pesce et al. (2015). The IFs of both epochs were correlated with 4096 spectral channels (resolution of 15.6 kHz, corresponding to 0.2 km s⁻¹) using the DiFX correlator (Deller et al. 2011). We observed in phase-referencing mode, namely with cycles of scans between the phase-reference and the target lasting 45 s each. In this way, it is possible to correct the atmosphere phase variation and to measure absolute positions of the maser. We used J0802+2509 (α_J2000 = 08^h02^m41.58742257^s; δ_J2000 = +25° 09′10$\stackrel{″}{.}$8982794³) at ∼1.7° from the target as a phase calibrator. The bandpass and the amplitude corrections were made using the fringe-finder calibrators DA193 and 4C39.25. The total observing time was 6 and 8 h in epochs 2018.16 and 2018.83, respectively. We then self-calibrated the phase reference source J0802 + 2509  and the solutions obtained were transferred to the target. Finally, in both epochs, we used the AIPS task IMAGR to produce the total intensity image cube and a continuum map by averaging the line-free channels. In epoch 2018.83, the image cube was produced after gluing together the two IFs using the AIPS task UVGLU.

To identify the maser features, we analysed the map obtained by summing the channels that should contain the maser emission with the task SQASH. The channel range was selected considering the single-dish spectra reported in Pesce et al. (2015) and Braatz et al. (2020)⁴. We visualised this map and identified bright spots by eye. The spectra were then produced using the task ISPEC, setting a window of 3 × 3 pixels (mainly covering the beam size and where 1 pixel = 0.1 mas) around the spots identified in the sqashed map (obtained from the cube map where the planes of the velocities were collapsed by summing each contribution of the latter). The absolute position was estimated by fitting the brightest maser spot of each maser feature with a two-dimensional Gaussian fit with the task JMFIT. We extracted the spectrum of each maser feature in an ASCII file using the task ISPEC. These files were imported in the GILDAS software called CLASS⁵, where we analysed the line properties (e.g., peak flux densities and peak velocity). The signal-to-noise ratio (S/N) threshold at and above which a maser detection was considered to be real was 5σ. However, we also considered any emission found in the velocity range of the features observed and reported in the single-dish spectrum (Pesce et al. 2015; Braatz et al. 2020). The continuum sources were identified in the continuum map considering only the emissions with S/N ≥ 5σ, which were Gaussian fitted with the AIPS task JMFIT.

3.1.2. Absolute positions uncertainties

To measure the uncertainty of the absolute position (named Δα and Δδ for α_J2000 and δ_J2000, respectively) of the maser features and of the continuum sources in IC 485, we considered the position errors of the phase calibrator⁶ J0802 + 2509 and the position errors due to the thermal noise (ε_rms) of the maps, which is evaluated following Eq. (13) of Reid & Honma (2014), that is, ε_rms ≈ (0.5 θ_beam)/(S/N), where ε_rms is the thermal noise and θ_beam is the dimension of the beam. These values are reported in Table 2.

3.1.3. Imaging of the data sets

To realise the continuum and cube maps, the data were Fourier-transformed using natural weighting and were deconvolved using the CLEAN algorithm (Högbom 1974). To map the continuum, we covered a field of 0.2 × 0.2 arcsec² (corresponding to ∼120 × 120 pc²). Meanwhile, to make the cube map, we mapped a field of 0.04 × 0.04 arcsec² (corresponding to ∼24 × 24 pc²). Both maps were centred at the position of the main maser feature, namely M1. Details of the maps produced are reported in Table 1.

3.2. L and C Band: Continuum observations

3.2.1. VLBA observations

We observed IC 485 with the VLBA⁷ of the NRAO at L-band (1.6 GHz) on February 11 and August 5, 2018, in two runs of 4.5 h each (project codes BT142 and BT145 and P.I.: A. Tarchi). The data were taken with four 64 MHz IFs in dual circular polarisation and recorded at a rate of 2048 Mbps. The correlation of the data was performed using the DiFX software correlator (Deller et al. 2011) with 128 channels per IF. Similarly to the spectral observations (see Sect. 3.1.1), we observed in phase-referencing mode (with cycles of scans phase-calibrator – target of 1 min–3 min each, respectively), where the phase-calibrator is J0802+2509. We also observed the strong compact quasars DA193 and 4C39.25 at the beginning and at the end of the observing run, using these as fringe finders and bandpass calibrators in order to correct instrumental delays.

After some editing, where we flagged ∼23%–30% of the visibilities of IC 485, we applied the ionospheric corrections and the latest value of the Earth’s orientation parameters. We then corrected the delays and phases for the effect of diurnal feed rotation (parallactic angle) and the amplitudes for the digital sampler voltage offsets. We removed the instrumental delays caused by the passage of the signal through the electronics using the phase-calibration measurements associated with the data (in the PC table). At this point, we calibrated the bandpass using DA193 and 4C39.25, and the amplitude using the measured antenna gains and system temperatures. Finally, we self-calibrated the phase reference source J0802+2509 and the solutions were applied to the target source IC 485.

3.2.2. EVN observations

We observed the nucleus of IC 485 with the EVN⁸ at L- (central frequency = 1.7 GHz) and C-band (5.0 GHz) in May 2018 (project codes: ET038; P.I.: A. Tarchi). The data were recorded at 1024 Mbps, with 8 × 16 MHz IFs and dual circular polarisation. The correlation of the data was performed using the EVN software correlator (SFXC; Keimpema et al. 2015) at the Joint Institute for VLBI ERIC (JIVE), using 32 channels per IF. We observed in phase-referencing mode (with cycles of scans phase-calibrator – target of 3 min–7 min and 3 min–5 min at L- and C-band, respectively) to correct phase variation caused by the atmosphere and to obtain information about absolute position. The chosen phase calibrator is the same as that used for the VLBA observations, namely J0802+2509, and the strong compact sources J0854+2006 and DA193 were observed as fringe finders. The total observing time was 4 and 2.5 h at L- and C-band, respectively.

Before calibrating, we inspected the data to look for radio frequency interference (RFI) and the so-called ‘bad points’ or time ranges. We flagged ∼34%–38% of the visibilities of the target source at both bands. As opposed to VLBA data, initial a priori amplitude and parallactic angle calibrations were carried out using the standard EVN pipeline. We then calibrated the bandpass using all the fringe finders. Subsequently, we removed the instrumental delays by fringe fitting the calibrators J0854+2006 and DA193. In order to solve for atmospheric phase variations, we self calibrated the phase reference source J0802+2509. Finally, similarly to the VLBA calibration, we interpolated and applied the solutions of J0802+2509 to our target source IC 485. We followed the same calibration steps for both frequency bands; however, for the L-band dataset, after the a priori calibration, we performed the ionospheric corrections as a first step of the calibration.

3.2.3. Imaging of the data sets

We used the following procedure to image both the EVN and VLBA data sets. The data were Fourier-transformed using natural weighting and deconvolved using the CLEAN algorithm (Högbom 1974). We mapped a field of 4 × 4 arcsec² and 0.6 × 0.6 arcsec² (corresponding to ∼2.4 × 2.4 kpc² and ∼360 × 360 pc²) at L- and C-band, respectively, centred at the position of the main maser feature, namely M1. In order to suppress the small-scale side-lobes and increase the beamwidth, we also produced a set of EVN maps, down-weighting the data at the outer edge of the (u, v) coverage using a taper of 40 and 60 Mλ at L- and C-band, respectively. Details of the best maps produced are reported in Table 1.

4. Results

In this section, we report the outcomes from our data reduction. The first subsection is dedicated to K-band where we describe spectral and continuum results. Meanwhile, in the second subsection, we expose the results of L- and C-band.

4.1. K band

4.1.1. Maser

In the two VLBA epochs, we detected one H₂O maser component⁹ close to the systemic velocity of IC 485 (v_sys = 8338 km s⁻¹) and one component shifted towards higher velocities. In particular, the component at the systemic velocity was observed in both epochs (named M1 and M1^* in epochs 2018.16 and 2018.83, respectively) and the redshifted component was only observed in epoch 2018.83 (named M2). Their parameters are reported in Table 3. The component M1 was fit using three Gaussian curves (see Fig. 1): a broad feature with a full width at half maximum (FWHM) of 38.3 km s⁻¹ and two narrower Gaussian features with a FWHM of 4.1 and 5.3 km s⁻¹, respectively. The estimated total isotropic luminosity of the three features is 526 L_⊙. The two components identified in VLBA 2018.83, M1^* and M2, are displayed in Fig. 2. The component M1^* is composed of only one Gaussian feature and shows a flux density of 19.7 mJy and a line width 35 km s⁻¹, which implies an isotropic luminosity of (239  ±  28) L_⊙. The component M2 is also composed of one Gaussian feature and its flux density and line width are 4 mJy and 18 km s⁻¹, respectively, with an isotropic luminosity of 24 L_⊙.

Fig. 1. Spectrum of the component M1 detected towards IC 485 with the VLBA in epoch 2018.16. The thin red line is obtained by summing together the three Gaussian features. The three different Gaussian features are also reported in different colours (see Table 3): in violet, green, and blue are features 1, 2, and 3, respectively.

Fig. 2. Spectra of the systemic component M1* (top panel) and of the redshifted component (bottom panel) as detected in epoch VLBA 2018.83. Top panel: spectrum covering the velocity range 8100–8600 km s−1. Bottom panel: spectrum covering the velocity range 8700–9000 km s−1.

4.1.2. Coincidence between M1 and M1^*

We compared the absolute positions of M1 and M1^*, which are detected close to the systemic velocity of IC 485 in order to assess the coincidence of the two maser components. In particular, we summed the spectral channels within the line emission of M1 and M1^*, separately. Afterwards, we imaged their contour maps and estimated their absolute positions by performing a 2D Gaussian fit using the AIPS task JMFIT. The results and the sqashed map are shown in Fig. 3. We also estimated the angular distance between M1 and M1^* according to the expression for small angular distance, $\vartheta \approx \sqrt{{[({\alpha }_{M1\ast }-{\alpha }_{M1})cos{\delta }_{M1\ast }]}^2+{({\delta }_{M1\ast }-{\delta }_{M1})}^2}$, where α_M1*, δ_M1* and α_M1, δ_M1 are the right ascension and the declination of M1^* and M1, respectively. We obtained an angular distance of ϑ ≈ (0.27 ± 0.24) mas. The error of the angular distance was calculated considering the values of Table 2. As a matter of fact, the two positions are consistent within the error. We conclude that the two components M1 and M1^* are, in reality, the same component, and therefore in the following we refer to a single source, namely M1.

Fig. 3. Contours of the sqash map of the feature maser M1 in the epoch VLBA 2018.16. The cross indicates the position of the peak M1* in VLBA 2018.83. The plus symbol (+) is proportional to the relative error between the two epochs. In the bottom left panel, the clean beam of VLBA 2018.16 is reported: (0.78 × 0.44) mas.

4.1.3. Continuum

In Table 4, we report the absolute positions, the peak intensity, the integrated flux, and the S/N of the identified continuum sources at K-band in the two VLBA epochs. The sources are labelled with a ‘C’ followed by a number in ascending order according to the right ascension. In the VLBA 2018.16 epoch, three continuum sources (from C1 to C3) were detected above the threshold (S/N ≥ 5σ). Six sources (from C4 to C9) were instead detected in the VLBA 2018.83 epoch. All sources detected were unresolved and their positions relative to the maser emission are shown in Fig. 4. Moreover, no spatial correspondence among the sources was observed in the two epochs. Furthermore, none of these nine sources coincide with the tentative ones detected at L- and C-bands (see Sect. 4.2).

Fig. 4. Comparison of the absolute positions of the identified radio continuum compact sources (C1–9, see Table 4) and those of the water maser components M1 and M2 (see Table 3). The plus symbols “+” indicate the compact sources while the star symbol is centred at the position of M1, (the dimensions of the map do not allow us to discern M1 from M2, and for this reason the point is denoted M). The sizes of the symbols do not correspond to the absolute position uncertainties.

4.2. L and C Band

No continuum source was detected above the 5σ noise level, either at 1.4 or at 5.0 GHz, in a region of 100 mas (∼600 pc) radius centred on the maser position (see Sect. 4.1.1), although there are a number of tentative sources between 3σ and 5σ. Nevertheless, it is worth mentioning that a tentative source is visible in the most sensitive EVN map at L-band, with a peak flux density of 68 μJy (3.8σ), the position of which (α = 08^h00^m19$\stackrel{\text{s}}{.}$7522; δ = 26°42′05″.051) coincides, within the errors, with the VLA source detected at 20 GHz by Darling (2017). However, there is no hint of this source in the most sensitive VLBA image, not even at the 2σ level. With the aim of confirming the presence of this feature, we combined the EVN and the first VLBA data sets in the (u, v) plane and then imaged the resulting data set using natural weighting. The noise of the cleaned image is 24 μJy beam⁻¹. Unfortunately, the source was not detected in the combined map. This suggests that this source is also an artefact.

5. Discussion

5.1. The nature of maser emission in IC 485

Water maser emission was reported by Darling (2017) with the VLA array in A-configuration with a resolution of around 100 mas. Our VLBI measurements suggest that the bulk of the emission is coincident with the position reported by Darling (2017), improving the absolute positional accuracy (around 1 mas) by two orders of magnitude.

We analysed the main water maser emission using the moment maps shown in Fig. 5. From our analysis of the maps, we note that the emission is concentrated in the systemic velocity range of 8350–8360 km s⁻¹ and the masing gas shows a compact and uniform distribution in the region of emission. From the zeroth moment, we notice that the bulk of the emission coincides with the position of the maser feature M1. Our analysis of the first moment suggests the absence of a velocity gradient. The maps were made considering the emission above 3σ. A second map was produced with a cutoff at 2σ, where the components M1B in epoch 2018.83 and M1D in epoch 2018.16 were observed. Given the low significance level of these findings, they are treated as tentative and discussed only in Appendix A.

Fig. 5. Maps of the zeroth, first, and second moments. Left panel: moment-zero map (colour scale) of the water maser emission in IC 485 superimposed on the sqashed image (contours). Contour levels are: (−3, 3, 6, 9, 12, 15, 24, 26) × 64 μJy beam−1 (1σ rms = 64 μJy beam−1). Centre panel: mean velocity (first moment) map. Right panel: velocity dispersion (second moment) map. These figures were obtained from the epoch VLBA 2018.16. The beam size of (1.16 × 1.02) mas is reported in the bottom left corner of each panel.

The detection of the redshifted component M2, albeit below 3σ, is considered real. Indeed, its presence is strongly supported by the detection mentioned in Pesce et al. (2015). Indeed, these latter authors reported a multi-epoch averaged and strongly sensitive spectrum (rms = 0.74 mJy) for IC 485. This spectrum shows the redshifted and blueshifted components, though the latter is weak (2–3 mJy). In 2015, the redshifted component showed a flux density of 3–4 mJy, which is confirmed by our observations (see Table 3). Regarding the blueshifted component, Pesce et al. (2015) reported it as a tentative detection. In the present work, it was not possible to observe this component because of the bandwidth and the arrangement of the two IFs selected during the VLBA observations (see Sect. 3.1.1). Indeed, the IFs, whose number and size was dictated by the VLBA capabilities at the time of the observations, did not allow us to cover the velocity range of the blueshifted component (see Fig. 6).

Fig. 6. Spectrum of the water maser detected by Pesce et al. (2015) with the Green Bank Telescope, where we report the velocity coverage (1400 km s−1) of the two IFs used during the VLBA observations in epoch 2018.83. An overlap of about 100 km s−1 is visible. The green and red arrows represent the velocities of the water maser components M1 and M2, respectively, as detected in the present work. The (tentative) blueshifted component reported in Pesce et al. (2015) is highlighted with the blue question mark.

The presence of these features in the spectrum observed by Pesce et al. (2015), the position of M1 at the systemic velocity, and the redshifted component M2 observed in this work, together with their linear distribution, lead us to the conclusion that the maser is in the form of a disc. In addition, a recent survey made by Pesce et al. (2023) of 183 GHz H₂O maser emission from AGNs known to host a 22 GHz megamaser showed that a significant fraction of the sample – including many known disc masers and among which IC 485 – hosts emission from both transitions. Additionally, some of the targets also have triple-peaked spectra at higher frequency. In IC 485, on the other hand, 183 GHz emission is detected only close to the systemic velocity, which is likely due to the expected weakness of the satellite lines. The detection of the two water-maser transitions in IC 485 may still support and encourage our hypothesis of a disc maser.

Analysing the position of the systemic feature and that of the redshifted components, we assume that they are two of the three typical components present in the full spectrum of a water maser associated with a disc. A simple portrayal of our idea is presented in Fig. 7. Here, we show the components M1 and M2, and we estimate the position of the blueshifted component (referred to here as M3) assuming that this is symmetrically opposite to M2. Furthermore, we also sketch the black hole and its accretion disc. According to this scenario, the disc would be edge on with north–south orientation and rotating clockwise. The angular dimension of the disc is 0.8 mas, which corresponds to a linear dimension of 0.47 pc at the distance of IC 485. This value is consistent with that of the accretion disc in the galaxy NGC 4258 (R_in = 0.1 pc; Herrnstein et al. 1997) – which is considered the prototype in studies of maser discs –, with that in NGC 1068 (R_in = 0.6 pc; Morishima et al. 2023), with that in Mrk 1419 (R_in =   0.13–0.43 pc; Henkel et al. 2002), and with the values measured in other galaxies hosting H₂O megamaser discs of which the radii are estimated to fall in the range of 0.03–1.3 pc (e.g., Gao et al. 2017; Kuo et al. 2011). Assuming Keplerian rotation, the relation v² = GMR⁻¹ gives the black hole mass at the centre of the nuclear region. According to this relation we have:

$$\begin{array}{c}\hfill \frac{M_{\mathrm{BH}}}{M_{\odot }}=1.12\times {\left(\frac{v_{\mathrm{r}}}{[{\mathrm{km}\mathrm{s}}^{-1}]}\right)}^2\times \left(\frac{\theta }{[\mathrm{mas}]}\right)\times \left(\frac{D}{[\mathrm{Mpc}]}\right),\end{array}$$(1)

Fig. 7. Portrayal of the disc geometry based on the water maser components. The disc (in cyan), assumed to be in Keplerian rotation, is edge-on and oriented north–south, with the black hole at the centre. The positions of the detected maser components M1 and M2 are marked with a green cross ‘×’ and a red plus symbol ‘+’, respectively. The position of the supposed blueshifted maser component M3 is indicated with a blue plus symbol ‘+’ (see Sect. 5.1).

where v_r is the rotation velocity of the disc, which is the velocity difference between M1 and M2, that is, v_r ≈ 470 km s⁻¹; θ is the angular size of half of the disc (0.4 mas; average value of the angular separation between M1 and M2 in the two epochs); and D is the distance of the galaxy (D = 122 Mpc). Therefore, we find

$$\begin{array}{c}\hfill M_{\mathrm{BH}}=1.2\times {10}^7{M}_{\odot },\end{array}$$(2)

with the systematic uncertainty dominated by the galaxy distance error. The value is consistent with the one expected for a supermassive black hole in a LINER or Seyfert galaxy (e.g., Kuo et al. 2010).

Assuming that the above scenario reflects reality, the detection of the main maser component, M1, in both K-band epochs allows us to attempt a preliminary estimate of the velocity drift (usually this estimate is performed through a quasi-regular monthly spaced monitoring program). Because of the lower flux density and poorer spectrum of the second epoch, in order to determine this quantity, we decided to fit the spectra – with no smoothing applied – of the maser component M1 with only one Gaussian feature (see left panel of Fig. 8). The velocities of the Gaussian peaks are (8353.6  ±  0.1) km s⁻¹ and (8354.8  ±  0.5) km s⁻¹ at epochs 2018.16 and 2018.83, respectively. From a linear fit of these values (see right panel of Fig. 8), we derived a velocity drift of dv/dt ≈ 1.8 km s⁻¹ yr⁻¹, with an uncertainty of ±0.6 km s⁻¹ yr⁻¹ estimated from the error propagation. Although this value is roughly estimated, it is consistent with those observed in other maser discs (e.g., Mrk 1419 and NGC 6264; Henkel et al. 2002; Kuo et al. 2013). This value can also be compared with the centripetal acceleration, that is, $a_{\text{c}}=v_{\text{r}}^2/R$, which is derived using the values we obtained for the disc rotation velocity (v_r ≈ 470 km s⁻¹) and for the disc radius (R = 0.24 pc). The computed centripetal acceleration of a_c = 0.95 km s⁻¹ yr⁻¹ is not fully consistent with that obtained for the velocity drift (dv/dt = 1.8 ± 0.6 km s⁻¹ yr⁻¹). However, in order to make our comparison more quantitative, we have to take into account the uncertainties in the parameters involved. While this is not trivial, given the impact of possible disc peculiarities (e.g., a slight deviation from perfectly edge-on orientation, warping, etc.), we can, in any case, associate an uncertainty to our estimate of the centripetal acceleration by assuming an error for R of 0.09 pc (the positional error between M1 and M2) and of 10 km s⁻¹ for v_r (based on the spread in velocity of the redshifted features in Pesce et al. 2015). This yields a value for the uncertainty of ∼0.4 km s⁻¹ yr⁻¹. This computation somewhat reconciles the two values within the errors. However, these latter are likely to be even larger because of the aforementioned necessity to use a single Gaussian to fit the entire M1 systemic feature complex, which impacts on our ability to track the acceleration of each single feature (within a complex that extends over ∼100 km s⁻¹). The preliminary nature of our measurements further reinforces the necessity for an ad hoc monitoring program of the maser emission in order to better constrain the velocity drift of the (systemic) maser lines.

Fig. 8. Spectra and plots of velocity drift. Left panel: spectra of the main maser component M1 in (a) epoch 2018.16 and in (b) epoch 2018.83, with the Gaussian fit (thin red line) used to estimate the velocity drift. Right panel: velocity–epoch diagram. Here, we show the positions in velocity of M1 observed in the two VLBA epochs with the corresponding error bar derived by the Gaussian fit.

In addition, the maser component M1 in VLBA 2018.83 is characterised by a slow ascent and rapid descent (by observing the increasing axis of velocities) and shows an asymmetric shape in its line profile. This is expected for a rotating edge-on disc, simply because of the effects of projection onto the line of sight of the emitting rays, as explained in Schulz et al. (1995).

Similar behaviour has been found in NGC 4258 and in many spectra of maser discs in the catalogue of Braatz et al. (2020). The width of the maser line can be due to turbulent motions in the disc (e.g., Pariev & Bromley 1998) and is speculatively attributable to a magneto-rotational instability, as recently observed in NGC 4258 by Baan et al. (2022). Our hypothesis is that IC 485 is a new candidate galaxy disc maser. However, in order to confirm the disc nature, it is imperative to detect and map the three main maser components at VLBI scales. To achieve this goal, the target should be further investigated through a high-sensitivity VLBI campaign. Furthermore, observations with a high-sensitivity array would also allow us to clarify the nature of the tentative features observed. In particular, the tentative M1B and M1D are presumably associated with a jet or outflow maser, which does not rule out the possibility that the maser is composite in nature (see Appendix A). Adding a new confirmed disc maser at ∼100 Mpc would allow us to use the source in estimations of black hole mass and distance, pushing these kinds of measurements farther out into the Universe, and further strengthening the high potential impact of new upcoming facilities like the SKA and ngVLA, with which we expect to lead similar studies up to high-z (z ≥ 1).

5.2. The origin of continuum emission: AGN and/or star formation

We identified nine radio continuum compact sources, all at K-band, around the water maser emission in the nuclear region of IC 485 (see Fig. 4). The detection of these compact sources at only one of the three observed radio bands (no detections over 5σ at L- and C-band) and at parsec scale can be explained in two ways: (i) Either significant variability is affecting the sources across the different epochs or (ii) all the sources have a highly inverted spectrum. A combination of the two factors may also be taking place, but from a statistical point of view, the possibility that such a large number of sources share the aforementioned characteristics (strong variability and/or inverted spectrum) is unlikely. Also considering the fact that, at K-band all sources detected differ in position between the two epochs, we therefore think that we are dealing with spurious signal or artefacts, especially for the weakest sources. However, this scenario leaves the question regarding the absence of the radio emission that is expected to be present in the core of the AGN open.

Indeed, as reported in Table 5, unresolved and faint radio continuum emission was detected towards IC 485 at kiloparsec (kpc) scales with the VLA at 1.4 GHz within the NVSS (S_peak = 4.4 mJy; Condon et al. 2002) and FIRST¹⁰ (S_peak = 3.01 mJy) surveys. Darling (2017) also detected an unresolved radio-compact source in the nucleus of IC 485 at 20 GHz with the VLA, although with a barely sufficient significance (S_peak = 77 ± 15 μJy, ∼5σ).

Considering the FIRST peak flux density of 3.01 mJy, if we assume (as was the case for IRAS 15480−0534; Castangia et al. 2019) that only 30% of the VLA flux is recovered in VLBI images, we would expect a flux density of about 0.9 mJy in the L band map, which would be well above our 5σ noise level of 90 μJy. Therefore, the non-detection with the EVN suggests that the kpc-scale radio emission observed with the VLA is mostly resolved out at parsec scale, indicating a diffuse morphology. The bulk of the radio emission in IC 485 does not arise from a compact nuclear source, but it is diffused over a region larger than 0.1 arcsec (60 pc), the largest detectable angular scale of EVN observations. This suggests that the AGN in IC 485 is either radio-silent (i.e., radio emission is entirely produced by star formation) or the AGN emits in the radio band but its emission (which might be produced by a jet, a nuclear wind, or a corona) is faint and the large (hundreds of parsecs or kpc-scale) radio emission is dominated by the star forming regions in the host galaxy.

In order to estimate the expected radio emission from the AGN in IC 485, we can take advantage of the fundamental planes of black hole activity that link the radio luminosity of AGNs with their X-ray/optical luminosity and the black hole mass (e.g., Merloni et al. 2003; Baldi et al. 2021). The standard fundamental plane is a correlation between the radio luminosity at 5 GHz, the 2–10 keV X-ray luminosity, and the black hole mass (e.g., Merloni et al. 2003). Kamali et al. (2017) reported a Swift/BAT hard X-ray flux of 1.75 × 10⁻¹¹ erg s⁻¹ cm⁻² (corresponding to a luminosity of ∼3.5 × 10⁴³ erg s⁻¹); however, we note that the large BAT PSF makes it difficult to distinguish whether the emission is arising from IC 485 or, much more probably, from its companion IC 486, which is located ∼5′ away. IC 485 was also observed with Swift/XRT and XMM and a preliminary reduction of the latter data shows that it is only marginally detected with an observed flux density of the order of 10¹⁴ erg s⁻¹ cm⁻² (corresponding to a luminosity of ∼10⁴⁰ erg s⁻¹; Bassani, priv. comm.).

Given the uncertainty in extrapolating a 2–10 keV flux, we prefer to use the fundamental plane of black hole activity in the optical band found for the LeMMINGs sample by Baldi et al. (2021). We derive a rough estimate of the expected 1.5 GHz core luminosity using the expression reported in Baldi et al. (2021, and reference therein):

$$\begin{array}{c}\hfill log{L}_{\mathrm{core}}=(0.83log{L}_{[\mathrm{O}\mathrm{III}]}+0.82log{M}_{\mathrm{BH}})m+q,\end{array}$$(3)

where L_core is the core luminosity, L_[O III] is the [O III] emission line luminosity in units of erg s⁻¹, and M_BH is the black hole mass in units of M_⊙. The slope m and the intercept q differ for different types of galaxies (Seyferts, radio-quiet LINERs, radio-loud AGNs, and non-jetted H II galaxies; Baldi et al. 2021, Table 2). Employing the black hole mass estimated through our maser study, M_BH ∼ 1.2 × 10⁷ M_⊙ (Sect. 5.1), and L_[O III] ∼ 10⁴⁰ erg s⁻¹ (derived form the flux density reported in Zhu et al. 2011), we obtain a core luminosity in the range of 1 × 10³⁶ − 5 × 10³⁷ erg s⁻¹. This luminosity range is consistent with the upper limit obtained from our EVN L-band observations (3.1 × 10³⁶ erg s⁻¹), suggesting that more sensitive VLBI observations or an array with intermediate resolution between those of VLA and VLBI may potentially detect the weak radio emission expected from this AGN.

Using the [OIII] luminosity corrected for extinction as in Bassani et al. (1999), we can estimate the intrinsic X-ray luminosity in the 2–10 keV band from the correlation between L_X and L_[OIII] found by Panessa et al. (2006). Considering the observed L_[OIII] and Balmer decrement, H_α/H_β, reported in Zhu et al. (2011), we obtain L_X ∼ 5 × 10⁴² erg s⁻¹. This luminosity is consistent with the observed flux in the case of a strongly absorbed or a low-luminosity AGN. The low radio luminosity observed in IC 485 can be explained by investigating other properties and the phenomenology of the host galaxy that affects it. Among these, there may be SMBH mass and the accretion rate of material onto the black hole in conjunction with the inefficient accretion system (e.g., Kamali et al. 2019, and references therein). To clarify the spectroscopic classification of IC 485, we cite the studies that explore AGNs and the distinction among their various types, reported in Heckman & Best (2014). These studies indicate how the difference in luminosity may be due to a difference efficiency conversion of the potential energy of the gas accreted by the SMBH. In particular, these latter authors divide AGNs into two main categories: radiative-mode AGNs – historically referred to as Seyfert galaxies –, and jet-mode AGNs, which are associated with LINER galaxies. The low X-ray luminosity that characterises LINER galaxies would be due to a truncated, geometrically thin accretion disc – or the complete absence of one – which would be replaced by a geometrically thick structure. Generally, LINERs and low-luminosity AGNs show an X-ray luminosity in the range of 10³⁹–10⁴¹ erg s⁻¹ (e.g., Awaki 1999; Terashima et al. 2002). For this reason, the high X-ray luminosity observed in IC 485 seems to point towards a Seyfert (type 2) classification. Additionally, the possible presence and the dimensions of the inner radius of the accretion or maser disc of the target (∼0.24 pc, see Sect. 5.1) seem to rule out the idea of a truncated accretion disc and are in better agreement with the above classification (Sy2). Further analysis is therefore needed, not only to confirm but also to study the actual size of the eventual disc.

6. Summary

In this paper, we report a multi-epoch multi-band radio VLBI study of the galaxy IC 485. The observations were conducted in continuum and spectral modes using the VLBA and the EVN arrays. The findings of our work allow us to obtain the distribution (and absolute position) of the maser and continuum emissions in the nucleus of the galaxy at mas scales for the first time.

We detected nine weak radio-compact sources at K-band, but none at L- or C-bands. This indicates that, at these latter two bands, the nuclear radio continuum emission reported at larger scales is diffuse and is resolved-out at our very high resolution. Furthermore, the relative weakness of the radio emission is suggestive of a nuclear region hosting a radio-silent AGN and/or dominated by regions of star formation.

The maser emission is detected in our VLBI maps at 22 GHz and shows two spatially distinguished linearly distributed components: the first is at the systemic velocity (8355 km s⁻¹) and the second is at a redshifted velocity, offset by 472 km s⁻¹ from the systemic velocity. The scenario offered by our analysis supports the idea that the maser is associated with an edge-on accretion disc orientated north–south and with a radius of 0.24 pc. Assuming Keplerian rotation, the estimated enclosed mass is of M_BH = 1.2 × 10⁷ M_⊙, which is consistent with estimates for other galaxies belonging to the same AGN class as IC 485 (LINERs/Seyfert 2s). The tentative detection of some additional maser features – which are displaced from the putative disc – may suggest that the maser is composite in nature (partly associated with the disc and partly of jet or outflow origin) and therefore merits further investigation.

Acknowledgments

We thank the anonymous referee for useful comments on the manuscript. We are also grateful to Loredana Bassani for sharing information on high-energy data of IC 485. E.L. and A.T. would like to thank Liz Humphreys for the constructive discussion during the ‘VLBI-40’ meeting in Bologna.

References

Appendix A: Tentative maser features

In this appendix, we report some tentative features observed in the epochs VLBA 2018.83 and VLBA 2018.16. We define as “tentative” those features detected with S/N < 5σ but in proximity to component M1 (within ∼2 mas ≃ 1.2 pc); we believe these may provide useful information that can be used to build a complete picture of the water maser nature in IC 485. All these maser features (named M1B, M1C, and M1D) were detected with velocities close to the systemic one. The most relevant seem to be the features M1B and M1D, which were identified both in the sqashed map and in the image cube (see Sect. 3.1.1). The spectrum of each tentative maser feature, with a smooth function (8 channel), is shown in Fig. A.1, where the Gaussian fit is also reported. The feature M1C (at (8355  ±  1) km s⁻¹) appears to be strongly linked to component M1 (8354.8  ±  0.5) km s⁻¹ given the similar velocities of both. The features M1B and M1C were detected in the epoch VLBA 2018.83. M1D was detected in the epoch VLBA 2018.16 with a velocity of (8352.8  ±  0.6) km s⁻¹. Table A.1 reports the parameters of these tentative features and Fig. A.2 reports their position with respect to the main maser emission M1 in the zeroth-moment map of this latter. The positions of these tentative detections, that is, displaced from the disc structures, may suggest a jet or an outflow maser with a possible composite nature; such masers can be found in the literature (e.g. NGC1068, Gallimore et al. 1996).

Fig. A.1. Spectrum with smooth function of 8 channel with the corresponding Gaussian fit (thin red line) and the respective label of the tentative maser features observed in the epochs VLBA 2018.16 and VLBA 2018.83.

Fig. A.2. Contour maps of the zeroth moment of M1 observed in epoch 2018.11 (top) and in epoch 2018.83 (bottom) with a cut at 2 σ. The positions of the tentative features (M1B-M1D) are indicated with plus symbols together with M1 (‘+’ symbol) and M2 (‘×’ symbol).

All Tables

All Figures

	Fig. 1. Spectrum of the component M1 detected towards IC 485 with the VLBA in epoch 2018.16. The thin red line is obtained by summing together the three Gaussian features. The three different Gaussian features are also reported in different colours (see Table 3): in violet, green, and blue are features 1, 2, and 3, respectively.
In the text

	Fig. 2. Spectra of the systemic component M1* (top panel) and of the redshifted component (bottom panel) as detected in epoch VLBA 2018.83. Top panel: spectrum covering the velocity range 8100–8600 km s−1. Bottom panel: spectrum covering the velocity range 8700–9000 km s−1.
In the text

	Fig. 3. Contours of the sqash map of the feature maser M1 in the epoch VLBA 2018.16. The cross indicates the position of the peak M1* in VLBA 2018.83. The plus symbol (+) is proportional to the relative error between the two epochs. In the bottom left panel, the clean beam of VLBA 2018.16 is reported: (0.78 × 0.44) mas.
In the text

Fig. 4. Comparison of the absolute positions of the identified radio continuum compact sources (C1–9, see Table 4) and those of the water maser components M1 and M2 (see Table 3). The plus symbols “+” indicate the compact sources while the star symbol is centred at the position of M1, (the dimensions of the map do not allow us to discern M1 from M2, and for this reason the point is denoted M). The sizes of the symbols do not correspond to the absolute position uncertainties.

In the text

Fig. 5. Maps of the zeroth, first, and second moments. Left panel: moment-zero map (colour scale) of the water maser emission in IC 485 superimposed on the sqashed image (contours). Contour levels are: (−3, 3, 6, 9, 12, 15, 24, 26) × 64 μJy beam−1 (1σ rms = 64 μJy beam−1). Centre panel: mean velocity (first moment) map. Right panel: velocity dispersion (second moment) map. These figures were obtained from the epoch VLBA 2018.16. The beam size of (1.16 × 1.02) mas is reported in the bottom left corner of each panel.

In the text

Fig. 6. Spectrum of the water maser detected by Pesce et al. (2015) with the Green Bank Telescope, where we report the velocity coverage (1400 km s−1) of the two IFs used during the VLBA observations in epoch 2018.83. An overlap of about 100 km s−1 is visible. The green and red arrows represent the velocities of the water maser components M1 and M2, respectively, as detected in the present work. The (tentative) blueshifted component reported in Pesce et al. (2015) is highlighted with the blue question mark.

In the text

Fig. 7. Portrayal of the disc geometry based on the water maser components. The disc (in cyan), assumed to be in Keplerian rotation, is edge-on and oriented north–south, with the black hole at the centre. The positions of the detected maser components M1 and M2 are marked with a green cross ‘×’ and a red plus symbol ‘+’, respectively. The position of the supposed blueshifted maser component M3 is indicated with a blue plus symbol ‘+’ (see Sect. 5.1).

In the text

	Fig. 8. Spectra and plots of velocity drift. Left panel: spectra of the main maser component M1 in (a) epoch 2018.16 and in (b) epoch 2018.83, with the Gaussian fit (thin red line) used to estimate the velocity drift. Right panel: velocity–epoch diagram. Here, we show the positions in velocity of M1 observed in the two VLBA epochs with the corresponding error bar derived by the Gaussian fit.
In the text

	Fig. A.1. Spectrum with smooth function of 8 channel with the corresponding Gaussian fit (thin red line) and the respective label of the tentative maser features observed in the epochs VLBA 2018.16 and VLBA 2018.83.
In the text

	Fig. A.2. Contour maps of the zeroth moment of M1 observed in epoch 2018.11 (top) and in epoch 2018.83 (bottom) with a cut at 2 σ. The positions of the tentative features (M1B-M1D) are indicated with plus symbols together with M1 (‘+’ symbol) and M2 (‘×’ symbol).
In the text