HI Observations of the Supermassive Binary Black Hole System in 0402+379
ABSTRACT We have recently discovered a supermassive binary black hole system with a projected separation between the two black holes of 7.3 parsecs in the radio galaxy 0402+379. This is the most compact supermassive binary black hole pair yet imaged by more than two orders of magnitude. We present Global VLBI observations at 1.3464 GHz of this radio galaxy, taken to improve the quality of the HI data. Two absorption lines are found toward the southern jet of the source, one redshifted by 370 +/- 10 km/s and the other blueshifted by 700 +/- 10 km/s with respect to the systemic velocity of the source, which, along with the results obtained for the opacity distribution over the source, suggests the presence of two mass clumps rotating around the central region of the source. We propose a model consisting of a geometrically thick disk, of which we only see a couple of clumps, that reproduces the velocities measured from the HI absorption profiles. These clumps rotate in circular Keplerian orbits around an axis that crosses one of the supermassive black holes of the binary system in 0402+379. We find an upper limit for the inclination angle of the twin jets of the source to the line of sight of 66 degrees, which, according to the proposed model, implies a lower limit on the central mass of ~7 x 10^8 Msun and a lower limit for the scale height of the thick disk of ~12 pc . Comment: 20 pages, 7 figures. Accepted on the Astrophysical Journal
arXiv:0902.4444v1 [astro-ph.CO] 25 Feb 2009
HI Observations of the Supermassive Binary Black Hole System
C. Rodriguez1, G. B. Taylor1,2, R. T. Zavala3,Y. M. Pihlstr¨ om,1,2, A. B. Peck4,5
We have recently discovered a supermassive binary black hole system with
a projected separation between the two black holes of 7.3 parsecs in the radio
galaxy 0402+379 (Rodriguez et al. 2006). This is the most compact supermassive
binary black hole pair yet imaged by more than two orders of magnitude. We
present Global VLBI observations at 1.3464 GHz of this radio galaxy, taken to
improve the quality of the HI data. Two absorption lines are found toward the
southern jet of the source, one redshifted by 370 ± 10 km s−1and the other
blueshifted by 700 ± 10 km s−1with respect to the systemic velocity of the
source, which, along with the results obtained for the opacity distribution over
the source, suggests the presence of two mass clumps rotating around the central
region of the source. We propose a model consisting of a geometrically thick disk,
of which we only see a couple of clumps, that reproduces the velocities measured
from the HI absorption profiles. These clumps rotate in circular Keplerian orbits
around an axis that crosses one of the supermassive black holes of the binary
system in 0402+379. We find an upper limit for the inclination angle of the twin
jets of the source to the line of sight of θ = 66◦, which, according to the proposed
model, implies a lower limit on the central mass of ∼ 7 × 108M⊙and a lower
limit for the scale height of the thick disk of ∼ 12 pc .
Subject headings: galaxies: active – galaxies: individual (0402+379) – radio
continuum: galaxies – radio lines: galaxies
1Department of Physics and Astronomy, University of New Mexico, Albuquerque, NM 87131
2Greg Taylor and Ylva Pihlstr¨ om are also Adjunct Astronomers at the National Radio Astronomy Ob-
3United States Naval Observatory, Flagstaff Station 10391 W. Naval Observatory Rd. Flagstaff, AZ
4Joint ALMA Office, Avda El Golf 40, piso 18, Santiago, Chile
5National Radio Astronomy Observatory, 520 Edgemont Rd, Charlottesville, VA 22903
– 2 –
Given that most galaxies harbor supermassive black holes at their centers (Richstone
et al. 1998; Gebhardt et al. 2000), and that galaxy mergers are common, binary black
holes should likewise be common. An understanding of the evolution and formation of these
systems is important for an understanding of the evolution and formation of galaxies in
general (Silk & Rees 1998, Merritt 2006). Theoretical descriptions of supermassive binary
black hole systems and their accretion disks have been investigated by Hayasaki et al. 2007,
2008 and by MacFadyen & Milosavljevi´ c 2008.
Our ability to resolve the supermassive black holes in any given binary system depends
on the separation between them, on their distance from Earth, and on the resolving power of
the telescope used. It is believed that the longest timescale in the evolution of a supermassive
binary black hole system leading up to coalescence is the stage in which the system is closely
bound (∼ 0.1 - 10 pc), meaning that in most of these systems the black hole pair can only
be resolved by VLBI observations (in the case where both black holes are radio loud) which
provides resolutions of milliarcseconds and finer. This could explain why very few such
systems have been found (see review by Komossa (2003a) detailing observational evidence
for supermassive black hole binaries).
Some source properties like X-shaped radio galaxies and double-double radio galax-
ies, helical radio-jets, double-horned emission line profiles, and semi-periodic variations in
lightcurves have been taken as indirect evidence for compact binary black holes though other
explanations are possible. Some wider systems have, however, been found more directly. For
example, the ultra luminous galaxy NGC 6240, discovered by the Chandra X-ray observa-
tory, was found to have a pair of active supermassive black holes at its center (Komossa et al.
2003b), separated by a distance of 1.4 kpc. Another system that has been known for some
time is the double AGN (7 kpc separation) constituting the radio source 3C 75, which was
discovered by the VLA to have two pairs of radio jets (Owen et al. 1985).
The radio galaxy 0402+379 was recently found to contain two central, compact, flat
spectrum, variable components (designated C1 and C2), with a projected separation of 7.3
pc, a feature which had not been observed in any other compact source, making this system
the most compact supermassive binary black hole pair yet imaged by more than two orders of
magnitude, with an estimated system mass of a few 108M⊙(Maness et al. 2004; Rodriguez
et al. 2006).
Maness et al. (2004) performed spectral line observations of 0402+379 at 1.348 GHz.
HI absorption at a redshift of 560 km s−1from the systemic velocity of the source (16,489
± 300 km s−1; Xu et al. 1994) was observed and attributed to a high velocity gas system,
– 3 –
possibly due to a merger. Following this discovery, a spectrum of the HI region was taken in
2004 with the Westerbork telescope (Morganti et al. 2009), which showed two components
with velocities separated by 1000 km s−1. In this article we present Global VLBI observa-
tions taken to improve the quality of the HI data on 0402+379, with the purpose of better
understanding the origin of the deviation from the systemic velocity seen in the HI gas in
At the redshift of 0402+379 of 0.055, H0=75 km s−1Mpc−1and q0= 0.5, a scale of 1
mas = 1.06 pc is obtained.
2.1.Global VLBI Observations
Global VLBI observations1were made on 2007 March 17 at 1.3464 GHz. A single
intermediate frequency with a bandwidth of 16 MHz was observed with 256 channels in
both right and left circular polarizations, resulting in a frequency resolution of 62.5 kHz,
corresponding to a velocity resolution of 15 km s−1. Four level quantization was employed.
The net integration time on 0402+379 was 497 minutes.
Standard flagging, amplitude calibration, fringe-fitting, and bandpass calibration (3C
111 was used for both gain and bandpass calibration) were followed in the Astronomical
Image Processing System (AIPS; van Moorsel et al. 1996). AIPS reduction scripts described
in Ulvestad et al. (2001) were used for a large part of the reduction. Spectral line Doppler
corrections were also applied in AIPS. All manual editing, frequency averaging procedures,
imaging, deconvolution, and self-calibration were done using Difmap (Shepherd et al. 1995).
A clean cube was also produced by first performing a continuum subtraction using the task
UVLSF in AIPS, then using Difmap in order to obtain a cleaned map for each spectral
frequency, and finally combining all the maps into a cube with the task MCUBE in AIPS.
1Global VLBI = Very Long Baseline Array + Effelsberg + Jodrell Bank + Westerbork + Green Bank
Telescope + Onsala.
– 4 –
3.1. Radio Continuum
Figure 1 shows a naturally weighted 1.3 GHz image of 0402+379 from the 2007 Global
VLBI observations. The source consists of two diametrically opposed jets and a central
region containing two active nuclei not resolved at this frequency (see Rodriguez et al. 2006
for a detailed multi-frequency study of this radio galaxy). The northern jet is pointing in
the northeast direction whereas the southern jet is pointing in the southwest direction. The
orientation of the source at this frequency is consistent with that seen by the VLA at both
1.5 and 5 GHz (Maness et al. 2004) and by the VLBA at 0.3 GHz (Rodriguez et al. 2006).
The image shown in panel (a) was tapered and restored with a circular 25 mas synthesized
beam, in order to show better the extended structure in this radio galaxy. The source spans
∼ 1000 mas (∼ 1000 pc). The image shown in panel (b) was tapered and restored with a
8.15 × 3.72 mas synthesized beam. We see structure on scales of ∼ 500 mas (∼ 500 pc).
Figure 2 shows HI absorption profiles taken from four regions of the source. The contin-
uum has been subtracted from the spectra using the task UVLSF in AIPS, which removes a
continuum model from the u,v data of all channels. The contours are taken from the 2007
Global VLBI observations at 1.3 GHz and the color scale from the 2005 VLBA observations
at 5 GHz. Two absorption lines are evident, which appear to be at two different locations
toward the southern jet. From this point on we will designate CW the western component,
where we see the stronger line; and CE the eastern component, where we see the weaker line.
Measurements of these lines are given in Table 1, where we also show the peak opacity and
the column density for both components, which we calculated according to,
NHI(cm−2) = 1.8224 × 1018TS(K)
∼ 1.8224 × 1018TS(K)∆v(km s−1)
where TSis the spin temperature, ∆v is the velocity resolution, and the summation is over
the channels where we detect absorption for CW and CE respectively.
We determined the central velocity of the HI absorption lines to be 16,927 ± 7 km s−1
and 15,856 ± 9 km s−1for CW and CE respectively. From the most recent reported redshift
for 0402+379, as measured from optical emission lines (Rodriguez et al. 2006), we find that
– 5 –
the observed line for CW is redshifted by 370 ± 10 km s−1and for CE is blueshifted by 700
± 10 km s−1from the systemic velocity of the source (16,558 ± 3 km s−1). No absorption
was found at either C1 or C2 (see Figure 2). We calculate a limit on the HI opacity at the
location of C1 and C2, and obtain < 0.17 and < 0.03 repectively. Figure 3 shows a map of
both the central velocity and width of the HI absorption profiles over the source.
Figure 4 shows a velocity slice of the continuum-subtracted cube (right panel), accom-
panied by the HI opacity distribution over the source (left panel). This result shows that
either the two locations where we find absorption are localized regions, rather than being
part of a more extended and perhaps homogeneous structure, or our sensitivity prevented
us from detecting HI absorption over a broader region. In order to explore this question we
calculated the lower limit on the peak opacity across the source assuming an intensity for the
line of 3σ, where σ is the rms noise in a single channel. The result is shown in Figure 5. We
see that outside the region where we detect absorption (compare with Figure 4) we would
need a peak opacity of at least ∼ 0.02, comparable to the peak opacity found for components
CE and CW, in order to have a 3σ detection. The fact that we did not detect this means
that the peak opacity in this region is even smaller than the lower limit found, suggesting
that what we are observing are two different, localized clumps, rather than a more extended
and homogeneous structure. The fact that we do not see absorption against C2 supports
We can estimate the HI mass of the clumps from the calculated column densities (Table
1) and assuming a spherical shape, according to the following relation,
where mHis the hydrogen atom mass and r is the radius of the clump estimated from the
opacity distribution map shown in Figure 4. For CW we used r ∼ 12 pc and for CE r ∼ 8
pc. We find a mass ranging from ∼ (7 − 400) × 103M⊙for component CW, and ∼ (1 −
60) × 103M⊙for component CE, assuming spin temperatures ranging from 100 K to 6000
K. These masses are bigger than typical values of HI cloud masses of 60 M⊙found in the
Milky Way (Stil et al. 2006).
The results found from our 2007 Global VLBI observations at 1.3 GHz show two HI
absorption lines, one blueshifted and the other redshifted with respect to the systemic ve-
locity. The redshifted line (component CW) shows a FWHM of 300 ± 20 km s−1, and the