Abstract

The Atacama Large Millimeter/submillimeter Array (ALMA) serendipitously detected H2O |$J_{K_{\rm a}, K_{\rm c}} = 10_{2,9}$|–93, 6 emission at 321 GHz in NGC 1052. This is the first submillimeter maser detection in a radio galaxy and the most luminous 321 GHz H2O maser known to-date with the isotropic luminosity of |$1090\, L_{\odot }$|⁠. The line profile consists of a broad velocity component with FWHM = 208 ± 12 km s−1 straddling the systemic velocity and a narrow component with FWHM = 44 ± 3 km s−1 blueshifted by 160 km s−1. The profile is significantly different from the known 22 GHz 61, 6–52, 3 maser which shows a broad profile redshifted by 193 km s−1. The submillimeter maser is spatially unresolved with a synthesized beam of |${0{^{\prime \prime}_{.}}68} \times {0{^{\prime \prime}_{.}}56}$| and coincides with the continuum core position within 12 pc. These results indicate amplification of the continuum emission through high-temperature (>1000 K) and dense [n(H2O) > 104 cm−3] molecular gas in front of the core.

1 Introduction

Extragalactic H2O masers (megamasers) are known to originate in sub-parsec (sub-pc)-scale disks rotating around supermassive black holes in active galactic nuclei (AGNs) (Miyoshi et al. 1995; Moran et al. 1995; Impellizzeri 2022). Thanks to their compact and bright nature, megamasers allow μas-resolution VLBI observations (Baan et al. 2022) making them powerful probes to weigh black-hole mass, to trace accretion matter, and to measure geometrical distance and constrain cosmological parameters (Humphreys et al. 2013). To date, 150 AGNs are known to harbor the |$J_{K_{\rm a}, K_{\rm c}} = 6_{1,6}$|–52, 3 maser at 22 GHz (Braatz et al. 2018).

The radio galaxy NGC 1052 is a unique megamaser found in an elliptical galaxy with double-sided sub-relativistic jets (Braatz et al. 1994; Claussen et al. 1998), while the majority of megamasers are radio-quiet AGNs in spiral galaxies. NGC 1052 shows the 22 GHz maser spectrum spanning ∼400 km s−1 (Claussen et al. 1998; Braatz et al. 2003; Kameno et al. 2005) which is smooth and broad as for radio-loud minorities such as TXS 2226-184 (Koekemoer et al. 1995), Mrk 348 (Peck et al. 2003) and Centaurus A (Ott et al. 2013), unlike archetypal megamasers which display narrow spikes <20 km s−1 spread over ∼±1000 km s−1 tracing (sub-)Keplerian rotation (Nakai et al. 1993; Humphreys et al. 2008). H2O masers in Mrk 348 and Centaurus A are interpreted to be excited in shocked molecular gas by interaction with jets (Peck et al. 2003; Ott et al. 2013). The maser spots in NGC 1052 appear to have high brightness temperature exceeding 109 K and lie along jets (Claussen et al. 1998). Sawada-Satoh et al. (2008) revealed the spatial distribution of maser spots consisting of two clusters on the line of sight towards the continuum components in the eastern approaching side and western receding side of jets, and found positional coincidence of maser spots and the plasma torus which obscures the core by free–free absorption (FFA) casting a gap between two maser clusters (Kameno et al. 2001, 2003; Vermeulen et al. 2003; Kadler et al. 2004). As the FFA opacity decreases at higher frequency, the gap disappears above 43 GHz, and the core takes over the steep-spectrum jets at higher frequency than 86 GHz (Baczko et al. 2016, 2019; Sawada-Satoh et al. 2016, 2019). In this context, submillimeter (sub-mm) observations are crucial to probe inside the obscuring torus.

The H2O 102, 9–93, 6 maser at 321.225677 GHz is a new probe for AGNs. Excitation of the 321 GHz maser, with the lower energy level of 1846 K in the vibrational ground state, requires a high temperature and high density for an inverted excitation condition. Gray et al. (2016) calculated the excitation condition of possible H2O maser transitions for evolved star environments and clarified that the 321 GHz maser requires a physical temperature of Tk > 1000 K and high density of n(H2O) > 104 cm−3. The 321 GHz maser condition implies a closer location to a central engine than that of the 22 GHz maser, if found in an AGN.

Hagiwara et al. (2013) surveyed five type-2 Seyfert galaxies selected by strong 22 GHz maser emission and discovered the first 321 GHz megamaser in the Circinus galaxy, followed by the second detection in NGC 4945 (Pesce et al. 2016; Hagiwara et al. 2016). Braatz (2019) refers to the third detection in NGC 5643 by D. W. Pesce et al. (in preparation). So far, no radio-loud AGN has been identified as a 321 GHz maser source.

In this letter we present the first discovery of the 321 GHz maser in NGC 1052, coupled with monitoring of the 22 GHz maser to compare the spectral profiles. Since the monitoring was taken 16 years earlier than the 321 GHz observation, we do not attempt to perform a detailed comparison, however. We employ the systemic velocity of Vsys = 1492 km s−1 and the luminosity distance of DL = 17.6 Mpc.

2 Observations

Reduction scripts and reduced data are available in the GitHub repository.1

2.1 GBT

We observed the 22 GHz maser at six epochs: 2005-11-18, 2005-11-28, 2005-12-07, 2005-12-20, 2005-12-27, and 2006-01-13, using the Green Bank Telescope (GBT) with the dual-beam dual-polarization receiving system. We followed the standard spectral calibration procedure using GBTIDL to obtain spectra with a velocity resolution of 0.33 km s−1. The rms noise of the spectrum at each epoch was 2.3, 2.7, 2.1, 1.4, 4.6, and 2.5 mJy, respectively. The continuum flux density was 1.451, 1.131, 1.365, 1.272, 1.505, and 1.168 Jy, respectively, involving systematic errors of ∼10% in flux scaling and off-source subtraction.

2.2 ALMA

The Atacama Large Millimeter/submillimeter Array (ALMA) observations with 41 12 m antennas were carried out on 2022-05-12, as the PI science project 2021.1.00341.S targeting a high-redshift galaxy. The operations software selected NGC 1052 as a phase calibrator within 15° from the target and J0423−0120 as a bandpass calibrator. Total integration time on NGC 1052 was 120 s consisting of 4 30-s scans. Four spectral windows (SPWs) with a 1875 MHz bandwidth were centered at 306.4, 308.2, 318.4, and 320.2 GHz. The spectral resolution was 7.8125 MHz, corresponding to 7.3 km s−1. The image rms was 3.3 mJy beam−1 for each spectral channel with the synthesized beam of |${0{^{\prime \prime}_{.}}68} \times {0{^{\prime \prime}_{.}}56}$| with |$\mathit {PA}=79^{\circ }$|⁠.

3 Results

Continuum-subtracted spectra are shown in figure 1 together with absorption lines of H2O 55, 0–64, 3, |$v$|2 = 1, HCN 1–0, and 4–3 (Kameno et al. 2020, 2023). The 22 GHz maser showed broad spectra in the velocity range of 1450–1850 km s−1 with time-variable narrower components. The centroid of the 22 GHz emission remained at 1685 ± 4 km s−1, which was redshifted with respect to the systemic velocity by 193 km s−1. The 321 GHz H2O emission profile was biased blueward, which was fitted by two Gaussian components: (a) a broad component with the center velocity of 1467 ± 5 km s−1 and FWHM of 208 ± 12 km s−1, and (b) a narrow component with the center velocity of 1333 ± 1 km s−1 with FWHM of 44 ± 3 km s−1. The peak flux density was 0.07 Jy at 1334 km s−1.

Continuum-subtracted spectra of NGC 1052. Observation date and line species are labeled on the left-hand side. The bottom spectrum (blue) is the 321 GHz H2O emission observed with ALMA. Two adjacent SPWs overlap in the velocity range of 1752–1814 km s−1 to show non-uniform channel spacing. The profile is fitted (blue solid line) by two Gaussian components (dashed lines): (a) a broad component with the center velocity of 1467 ± 5 km s−1 and width (FWHM) of 208 ± 12 km s−1 in green; and (b) a narrow component with the center velocity of 1333 ± 1 km s−1 and width (FWHM) of 44 ± 3 km s−1 in red. The second to fourth spectra from the bottom are the absorption features of 89 GHz HCN (1–0) (Kameno et al. 2023), 233 GHz H2O 55, 0–64, 3, $v$2 = 1, and 355 GHz HCN (4–3) (Kameno et al. 2020) transitions, offset by 0.11, 0.16, and 0.2 Jy, respectively. The 233 GHz feature is vertically magnified by 15 times. The top six spectra are the 22 GHz masers observed with GBT, offset by 0.525–0.05i Jy for ith epoch. The vertical dotted line indicates the systemic velocity of 1492 km s−1.
Fig. 1.

Continuum-subtracted spectra of NGC 1052. Observation date and line species are labeled on the left-hand side. The bottom spectrum (blue) is the 321 GHz H2O emission observed with ALMA. Two adjacent SPWs overlap in the velocity range of 1752–1814 km s−1 to show non-uniform channel spacing. The profile is fitted (blue solid line) by two Gaussian components (dashed lines): (a) a broad component with the center velocity of 1467 ± 5 km s−1 and width (FWHM) of 208 ± 12 km s−1 in green; and (b) a narrow component with the center velocity of 1333 ± 1 km s−1 and width (FWHM) of 44 ± 3 km s−1 in red. The second to fourth spectra from the bottom are the absorption features of 89 GHz HCN (1–0) (Kameno et al. 2023), 233 GHz H2O 55, 0–64, 3, |$v$|2 = 1, and 355 GHz HCN (4–3) (Kameno et al. 2020) transitions, offset by 0.11, 0.16, and 0.2 Jy, respectively. The 233 GHz feature is vertically magnified by 15 times. The top six spectra are the 22 GHz masers observed with GBT, offset by 0.525–0.05i Jy for ith epoch. The vertical dotted line indicates the systemic velocity of 1492 km s−1.

We produced channel maps of the 321 GHz H2O line with a velocity slice of 20 km s−1 in 1280 ≤ VLSR ≤ 1580 km s−1 and measured the position by 2D Gaussian fitting in the channel maps. Figure 2 shows the position of each channel registered on the continuum image. These coincided with the continuum position with an upper limit of |${0{^{\prime \prime}_{.}}14}$| corresponding to 12 pc, except one outlier offset by |${0{^{\prime \prime}_{.}}29} \pm {0{^{\prime \prime}_{.}}03}$| at 1560 km s−1 with a signal-to-noise ratio as low as 4.3. No significant velocity gradient was found.

The contours show continuum image at 313.3 GHz combining all SPWs. The contour levels stand for 0.1, 0.5, and 0.9 × the peak intensity of 0.39 Jy beam−1. Note that the continuum source is unresolved (see figure 3) and the contours represent the beam size. Filled circles with crosses indicate continuum-subtracted 321 GHz maser spots with a velocity slice of 20 km s−1. The circle size is proportional to the flux density of the emission line. Crosses indicate 3 × standard errors of the position determined by 2D Gaussian fit. The origin of the map is $(\alpha , \delta ) = ({02^{\rm h}41^{\rm m}04{^{\rm s}_{.}}798}, -08^{\circ }{15^{\prime }20{^{\prime \prime }_{.}}751})$ (J2000.0).
Fig. 2.

The contours show continuum image at 313.3 GHz combining all SPWs. The contour levels stand for 0.1, 0.5, and 0.9 × the peak intensity of 0.39 Jy beam−1. Note that the continuum source is unresolved (see figure 3) and the contours represent the beam size. Filled circles with crosses indicate continuum-subtracted 321 GHz maser spots with a velocity slice of 20 km s−1. The circle size is proportional to the flux density of the emission line. Crosses indicate 3 × standard errors of the position determined by 2D Gaussian fit. The origin of the map is |$(\alpha , \delta ) = ({02^{\rm h}41^{\rm m}04{^{\rm s}_{.}}798}, -08^{\circ }{15^{\prime }20{^{\prime \prime }_{.}}751})$| (J2000.0).

The continuum component was unresolved with flux densities of 0.401 Jy and 0.380 Jy at 307.3 GHz and 319.3 GHz, respectively. The upper limits (⁠|$99\%$| confidence) of the FWHM size were |${0{^{\prime \prime}_{.}}024}$| and |${0{^{\prime \prime}_{.}}12}$| for the continuum and the continuum-subtracted velocity-integrated emission line, respectively, as shown in figure 3.

Visibility amplitudes of 319.3 GHz continuum and line emission as a function of spatial frequency (projected baseline length / wavelength). The 321 GHz H2O line visibility is obtained by integrating the continuum-subtracted spectra in the velocity range of 1300–1550 km s−1. The filled areas indicate the $99\%$-confidence sizes of ${\le }{0{^{\prime \prime}_{.}}024}$ and ${\le }{0{^{\prime \prime}_{.}}12}$, for the continuum and the ${\rm H}_{2}{\rm O}$ line, respectively.
Fig. 3.

Visibility amplitudes of 319.3 GHz continuum and line emission as a function of spatial frequency (projected baseline length / wavelength). The 321 GHz H2O line visibility is obtained by integrating the continuum-subtracted spectra in the velocity range of 1300–1550 km s−1. The filled areas indicate the |$99\%$|-confidence sizes of |${\le }{0{^{\prime \prime}_{.}}024}$| and |${\le }{0{^{\prime \prime}_{.}}12}$|⁠, for the continuum and the |${\rm H}_{2}{\rm O}$| line, respectively.

4 Discussion

4.1 Justification for maser emission

The 321 GHz H2O emission size of FWHM |$\le {0{^{\prime \prime}_{.}}12}$| and peak flux density of 0.07 Jy yields a brightness temperature of TB ≥ 43 K. The ALMA array configuration was too compact (<500.2 m) to rule out the possibility of thermal emission. Nevertheless, ancillary evidence indicates that the emission is a non-thermal maser.

All thermal molecular transitions, including H2O 55, 0–64, 3, and |$v$|2 = 1, appeared in absorption straddling the systemic velocity at the core position of NGC 1052 (Kameno et al. 2020). CO J = 2–1 and 3–2 appeared as the sole thermal emission in the circum-nuclear disk (CND) extended in 230 × 60 pc off the continuum core, while they appeared in absorption toward the core (Kameno et al. 2020). The 321 GHz emission locates within 12 pc from the radio continuum, much more compact than the CND. The integrated flux density of the 321 GHz H2O emission was 10.5 Jy km s−1 exceeding the 1.55 Jy km s−1 and 7.96 Jy km s−1 of CO emissions. The CND cannot harbor the 321 GHz H2O emission.

Since the brightness temperature of the continuum emission exceeds 108 K (Kadler et al. 2004), molecular gas covering the continuum source must be seen in absorption except non-thermal emission. Thus, the 321 GHz H2O emission at the same position of the core is probably non-thermal maser. Longer-baseline (>3500 m) observations would provide more concrete constraints on the brightness temperature >2000 K.

4.2 Spectral profiles and location of masers

Observed maser intensity is a product of the background brightness and amplification gain through population-inverted H2O molecules. A line width becomes progressively narrower than the thermal width as it gains more amplification (Dopita & Sutherland 2003). The wide velocity width in NGC 1052 indicates small amplification gain and requires bright background continuum emission. Thus, the position of maser spots is related to the continuum structure. That characteristic is consistent with the smooth systemic-velocity components and is different from narrow spikes in archetypal megamasers.

Sawada-Satoh et al. (2008) revealed that the position of 22 GHz masers in NGC 1052 coincides with the intensity peaks of the double-sided jets where the FFA opacity is not too high. The continuum core is hidden in the optically thick gap at 22 GHz to generate 22 GHz maser avoidance. The GBT monitoring of 22 GHz maser emission always shows redshifted profiles with respect to the systemic velocity, implying an inward stream of the excited H2O molecule inside the near side of the torus.

The continuum structure at higher frequency is more dominated by a compact core with a smaller contribution from steep-spectrum jets, and the gap due to FFA disappears in millimeter (mm) wavelengths (Baczko et al. 2016, 2019; Sawada-Satoh et al. 2016, 2019). Thus, the 321 GHz maser is considered to pinpoint the compact core and is amplified by population-inverted H2O molecules with a temperature >1000 K. Kameno et al. (2020) found sub-mm molecular absorption features (see HCN 4–3 in figure 1) that implied the presence of a massive molecular torus covering the continuum source. The velocity ranges of the sub-mm absorption lines straddle the systemic velocity and resemble that of the broad component of the 321 GHz maser. This supports that molecular gas in the torus causes absorption and stimulated emission along the line of sight to the compact core.

Kameno et al. (2023) found mm (86–130 GHz) molecular absorption features and identified different line profiles from sub-mm ones (see HCN 1–0 and 4–3 in figure 1). The difference is ascribed to the continuum structure dominated by the jet at mm and the core at sub-mm, as mentioned in the former paragraph. The temperature of SO absorption lines were 26 ± 4 K and 344 ± 43 K for mm and sub-mm, respectively. The difference in temperatures also indicate that mm and sub-mm absorption lines represent different absorbers.

The torus is expected to consist of multiple layers with temperatures of ≥104 K, ≥1000 K, ≥400 K, and 344 ± 43 K that accounts for the FFA, 321 GHz maser, 22 GHz maser, and sub-mm SO absorption region, respectively. Kameno et al. (2023) interpret that SO molecules evaporate from dust grains through jet–torus interaction and the high temperature is maintained by shock heating. The shock heating also explains the excitation of the 321 GHz maser.

Since the mm SO absorber is too cold to excite the maser, downstream of the jets is unlikely to harbor the 321 GHz maser.

4.3 Comparison with the Circinus galaxy and NGC 4945

In this subsection we compare the properties of the H2O maser in NGC 1052 with those in NGC 4945 and the Circinus galaxy, referring to Hagiwara et al. (2021).

The isotropic luminosity is estimated to be |$176\, L_{\odot }$| and |$1090\, L_{\odot }$| for 22 GHz and 321 GHz masers, respectively, using the formula |$L = 1.04 \times 10^{-3} \nu _{\rm rest}, D^2_{L} \int S(v) \ dv$|⁠. These values are 5.5 times and 17 times larger than those of the Circinus galaxy. The ratio of the 321 GHz maser peak flux to continuum flux in NGC 1052 is 0.19, which is significantly smaller than that of 3.2 and 37 in the Circinus galaxy on 2012-06-03 and 2017-05-06, respectively. The ratio in NGC 4945 was 0.29 on 2012-06-03 and there was a non-detection on 2017-05-06. Since the ratio is an indicator of maser amplification gain, the high isotropic luminosity in NGC 1052 is ascribed to the brighter continuum. The broader maser feature is consistent with less narrowing with the lower amplification gain. It is difficult, however, to establish the degree of blending of multiple narrower features making up the spectrum in these observations.

While the Circinus galaxy and NGC 4945 harbor high-velocity components ∼1000 km s−1 with respect to the systemic velocity, such components are not identified in NGC 1052. Moran et al. (1995) interpreted the high-velocity components of 22 GHz masers as self-amplification of long velocity-coherent gain path through the rotating disk without a background continuum source. The presence of the systemic-velocity component and absence of high-velocity components in NGC 1052 also support the lower gain amplification of a brighter continuum.

5 Summary

ALMA observations of NGC 1052 revealed the presence of 321 GHz H2O emission with the isotropic luminosity of |$1090\, L_{\odot }$| which is in all probability the first sub-mm maser in a radio galaxy. The sub-mm maser profile straddles the systemic velocity of the galaxy, similar to sub-mm molecular absorption lines, unlike the redshifted profile of the 22 GHz maser. Although the sub-mm maser distribution is unresolved, the inferred location is between the two clusters of 22 GHz masers pinpointing the compact continuum core.

Follow-up ALMA observations of the 321 GHz emission, in order to establish variability and polarization in that line, as well as searching for emission in other H2O maser lines such as at 183 and 325 GHz, could also provide valuable information on the nucleus of this target. Sub-mm VLBI observations are desired to resolve the structure of the maser and unveil the dynamics and the excitation mechanism in the sub-pc region of AGNs.

Acknowledgements

This letter makes use of the following GBT and ALMA data: GBT 05C-034 and ADS/JAO.ALMA#2021.1.00341.S. The Green Bank Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), MOST and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ. This work is supported by Japan Society for the Promotion of Science (JSPS) KAKENHI 17740115, 18K03712, and 21H01137.

Footnotes

References

Baan
 
W. A.
,
An
 
T.
,
Henkel
 
C.
,
Imai
 
H.
,
Kostenko
 
V.
,
Sobolev
 
A.
 
2022
,
Nature Astron.
,
6
,
976

Baczko
 
A.-K.
 et al.  
2016
,
A&A
,
593
,
A47

Baczko
 
A.-K.
,
Schulz
 
R.
,
Kadler
 
M.
,
Ros
 
E.
,
Perucho
 
M.
,
Fromm
 
C. M.
,
Wilms
 
J.
 
2019
,
A&A
,
623
,
A27

Braatz
 
J.
 
2019
,
ALMA2019: Science Results and Cross-Facility Synergies
,
id. 24 (doi:10.5281/zenodo.3585214)

Braatz
 
J.
 et al.  
2018
, in
IAU Symp. 336, Astrophysical Masers: Unlocking the Mysteries of the Universe
, ed.
Tarchi
 
A.
 et al.
(
Cambridge
:
Cambridge University Press
),
86

Braatz
 
J. A.
,
Wilson
 
A. S.
,
Henkel
 
C.
 
1994
,
ApJ
,
437
,
L99

Braatz
 
J. A.
,
Wilson
 
A. S.
,
Henkel
 
C.
,
Gough
 
R.
,
Sinclair
 
M.
 
2003
,
ApJS
,
146
,
249

Claussen
 
M. J.
,
Diamond
 
P. J.
,
Braatz
 
J. A.
,
Wilson
 
A. S.
,
Henkel
 
C.
 
1998
,
ApJ
,
500
,
L129

Dopita
 
M. A.
,
Sutherland
 
R. S.
 
2003
,
Astrophysics of the Diffuse Universe
(
Berlin
:
Springer
)

Gray
 
M. D.
,
Baudry
 
A.
,
Richards
 
A. M. S.
,
Humphreys
 
E. M. L.
,
Sobolev
 
A. M.
,
Yates
 
J. A.
 
2016
,
MNRAS
,
456
,
374

Humphreys
 
E. M. L.
,
Reid
 
M. J.
,
Greenhill
 
L. J.
,
Moran
 
J. M.
,
Argon
 
A. L.
 
2008
,
ApJ
,
672
,
800

Humphreys
 
E. M. L.
,
Reid
 
M. J.
,
Moran
 
J. M.
,
Greenhill
 
L. J.
,
Argon
 
A. L.
 
2013
,
ApJ
,
775
,
13

Hagiwara
 
Y.
,
Horiuchi
 
S.
,
Doi
 
A.
,
Miyoshi
 
M.
,
Edwards
 
P. G.
 
2016
,
ApJ
,
827
,
69

Hagiwara
 
Y.
,
Horiuchi
 
S.
,
Imanishi
 
M.
,
Edwards
 
P. G.
 
2021
,
ApJ
,
923
,
251

Hagiwara
 
Y.
,
Miyoshi
 
M.
,
Doi
 
A.
,
Horiuchi
 
S.
 
2013
,
ApJ
,
768
,
L38

Impellizzeri
 
C. M. V.
 
2022
,
Nature Astron.
,
6
,
885

Kadler
 
M.
,
Ros
 
E.
,
Lobanov
 
A. P.
,
Falcke
 
H.
,
Zensus
 
J. A.
 
2004
,
A&A
,
426
,
481

Kameno
 
S.
 et al.  
2020
,
ApJ
,
895
,
73

Kameno
 
S.
 et al.  
2023
,
ApJ
,
944
,
156

Kameno
 
S.
,
Inoue
 
M.
,
Wajima
 
K.
,
Sawada-Satoh
 
S.
,
Shen
 
Z.-Q.
 
2003
,
PASA
,
20
,
134

Kameno
 
S.
,
Nakai
 
N.
,
Sawada-Satoh
 
S.
,
Sato
 
N.
,
Haba
 
A.
 
2005
,
ApJ
,
620
,
145

Kameno
 
S.
,
Sawada-Satoh
 
S.
,
Inoue
 
M.
,
Shen
 
Z.-Q.
,
Wajima
 
K.
 
2001
,
PASJ
,
53
,
169

Koekemoer
 
A. M.
,
Henkel
 
C.
,
Greenhill
 
L. J.
,
Dey
 
A.
,
van Breugel
 
W.
,
Codella
 
C.
,
Antonucci
 
R.
 
1995
,
Nature
,
378
,
697

Miyoshi
 
M.
,
Moran
 
J.
,
Herrnstein
 
J.
,
Greenhill
 
L.
,
Nakai
 
N.
,
Diamond
 
P.
,
Inoue
 
M.
 
1995
,
Nature
,
373
,
127

Moran
 
J.
,
Greenhill
 
L.
,
Herrnstein
 
J.
,
Diamond
 
P.
,
Miyoshi
 
M.
,
Nakai
 
N.
,
Inque
 
M.
 
1995
,
PNAS
,
92
,
11427

Nakai
 
N.
,
Inoue
 
M.
,
Miyoshi
 
M.
 
1993
,
Nature
,
361
,
45

Ott
 
J.
 et al.  
2013
,
ApJ
,
771
,
L41

Peck
 
A. B.
,
Henkel
 
C.
,
Ulvestad
 
J. S.
,
Brunthaler
 
A.
,
Falcke
 
H.
,
Elitzur
 
M.
,
Menten
 
K. M.
,
Gallimore
 
J. F.
 
2003
,
ApJ
,
590
,
149

Pesce
 
D. W.
,
Braatz
 
J. A.
,
Impellizzeri
 
C. M. V.
 
2016
,
ApJ
,
827
,
68

Sawada-Satoh
 
S.
 et al.  
2016
,
ApJ
,
830
,
L3

Sawada-Satoh
 
S.
 et al.  
2019
,
ApJ
,
872
,
L21

Sawada-Satoh
 
S.
,
Kameno
 
S.
,
Nakamura
 
K.
,
Namikawa
 
D.
,
Shibata
 
K. M.
,
Inoue
 
M.
 
2008
,
ApJ
,
680
,
191

Vermeulen
 
R. C.
,
Ros
 
E.
,
Kellermann
 
K. I.
,
Cohen
 
M. H.
,
Zensus
 
J. A.
,
van Langevelde
 
H. J.
 
2003
,
A&A
,
401
,
113

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.