D
RAFT VERSION
J
ANUARY
25, 2022
Typeset using L
A
T
E
X
twocolumn
style in AASTeX631
The Unanticipated Phenomenology of the Blazar PKS 2131
−
021: A Unique Super-Massive Black Hole Binary Candidate
S. O’N
EILL
,
1
S. K
IEHLMANN
,
2, 3
A.C.S. R
EADHEAD
,
1, 3
M.F. A
LLER
,
4
R. D. B
LANDFORD
,
5
I. L
IODAKIS
,
6
M.L. L
ISTER
,
7
P. M
R
́
OZ
,
8
C. P. O’D
EA
,
9
T. J. P
EARSON
,
1
V. R
AVI
,
1
M. V
ALLISNERI
,
10
K.A. C
LEARY
,
1
M. J. G
RAHAM
,
11
K.J.B. G
RAINGE
,
12
M.W. H
ODGES
,
1
T. H
OVATTA
,
13, 14
A. L
̈
AHTEENM
̈
AKI
,
14, 15
J.W. L
AMB
,
1
T. J. W. L
AZIO
,
10
W. M
AX
-M
OERBECK
,
16
V. P
AVLIDOU
,
2, 3
T. A. P
RINCE
,
11
R.A. R
EEVES
,
17
M. T
ORNIKOSKI
,
14
P. V
ERGARA DE LA
P
ARRA
,
17
AND
J. A. Z
ENSUS
18
1
Owens Valley Radio Observatory, California Institute of Technology, Pasadena, CA 91125, USA
2
Department of Physics and Institute of Theoretical and Computational Physics, University of Crete, 71003 Heraklion, Greece
3
Institute of Astrophysics, Foundation for Research and Technology-Hellas, GR-71110 Heraklion, Greece
4
2 Department of Astronomy, University of Michigan, 323 West Hall, 1085 S. University Avenue, Ann Arbor, MI 48109, USA
5
Kavli Institute for Particle Astrophysics and Cosmology, Department of Physics,20 Stanford University, Stanford, CA 94305, USA
6
Finnish Center for Astronomy with ESO, University of Turku, Vesilinnantie 5, FI-20014, Finland
7
Department of Physics and Astronomy, Purdue University, 525 Northwestern Avenue, West Lafayette, IN 47907, USA
8
Astronomical Observatory, University of Warsaw, Al. Ujazdowskie 4, 00-478 Warszawa, Poland
9
Department of Physics and Astronomy, University of Manitoba, Winnipeg, MB R3T 2N2, Canada
10
Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, USA
11
Division of Physics, Mathematics, and Astronomy, California Institute of Technology, Pasadena, CA 91125, USA
12
Jodrell Bank Centre for Astrophysics, University of Manchester, Oxford Road, Manchester M13 9PL, UK
13
Finnish Centre for Astronomy with ESO (FINCA), University of Turku, FI-20014 University of Turku, Finland
14
Aalto University Mets
̈
ahovi Radio Observatory, Mets
̈
ahovintie 114, 02540 Kylm
̈
al
̈
a, Finland
15
Aalto University Department of Electronics and Nanoengineering, PO Box 15500, 00076 Aalto, Finland
16
Departamento de Astronom
́
ıa, Universidad de Chile, Camino El Observatorio 1515, Las Condes, Santiago, Chile
17
CePIA, Astronomy Department, Universidad de Concepci
́
on, Casilla 160-C, Concepci
́
on, Chile
18
Max-Planck-Institut f
̈
ur Radioastronomie, Auf dem H
̈
ugel 69, D-53121 Bonn, Germany
ABSTRACT
Most large galaxies host supermassive black holes in their nuclei and are subject to mergers, which can
produce a supermassive black hole binary (SMBHB), and hence periodic signatures due to orbital motion. We
report unique periodic radio flux density variations in the blazar PKS 2131
−
021, which strongly suggest an
SMBHB with an orbital separation of
∼
0
.
001
−
0
.
01
pc. Our 45.1-year radio light curve shows two epochs of
strong sinusoidal variation with the same period and phase to within
<
∼
2%
and
∼
10%
, respectively, straddling
a 20-year period when this variation was absent. Our simulated light curves accurately reproduce the “red
noise” of this object, and Lomb-Scargle, weighted wavelet Z-transform, and least-squares sine wave analyses
demonstrate conclusively, at the
4
.
6
σ
significance level, that the periodicity in this object is not due to random
fluctuations in flux density. The observed period translates to
2
.
082
±
0
.
003
years in the rest frame at the
z
= 1
.
285
redshift of PKS 2131
−
021. The periodic variation in PKS 2131
−
021 is remarkably sinusoidal.
We present a model in which orbital motion, combined with the strong Doppler boosting of the approaching
relativistic jet, produces a sine-wave modulation in the flux density which easily fits the observations. Given
the rapidly-developing field of gravitational wave experiments with pulsar timing arrays, closer counterparts to
PKS 2131
−
021 and searches using the techniques we have developed are strongly motivated. These results
constitute a compelling demonstration that the phenomenology, not the theory, must provide the lead in this
field.
Keywords:
galaxies: active, galaxies: jets, galaxies: BL Lacertae objects: individual (PKS 2131
−
021)
Corresponding author: Anthony Readhead
acr@caltech.edu
arXiv:2111.02436v2 [astro-ph.HE] 24 Jan 2022
2
O’N
EILL ET AL
.
1.
INTRODUCTION
The identification of supermassive black hole binaries
(SMBHBs) will open the field to multimessenger astronomy
through the gravitational radiation they produce. Pulsar tim-
ing arrays provide a powerful technique for searching for
nanohertz signals from gravitational waves from SMBHBs
through the timing of millisecond pulsars (Holgado et al.
2018; Burke-Spolaor et al. 2019). However, in spite of the
fact that galaxy mergers are not uncommon, there are rela-
tively few instances of two galaxies with supermassive black
holes in their nuclei being seen in the actual process of the
merging of the spheres of influence of their SMBHs, or of
the following stages, when an SMBHB forms by ejecting
stars from the merging central clusters, and spirals in more
closely due to gravitational radiation, and finally coalesces
(Begelman et al. 1980). A particularly fine example of the
early stage of possible evolution towards an SMBHB is that
of 3C 75 (Owen et al. 1985), where both SMBHs are produc-
ing radio jets, and their projected separation is 7.2 kpc. On
pc scales the best SMBHB candidate is B3 0402+379 (Ro-
driguez et al. 2006; Bansal et al. 2017), with a projected sep-
aration of 7.3 pc, a deduced period of
3
×
10
4
yr, and a de-
duced SMBHB mass of
≈
1
.
5
×
10
10
M
. The strongest
SMBHB candidate with a separation of
1
pc is OJ 287
(Sillanpaa et al. 1988; Valtonen et al. 2016; Dey et al. 2021),
for which the separation
∼
0
.
1
pc, the deduced primary
mass
≈
1
.
8
×
10
10
M
, and the deduced secondary mass
≈
1
.
5
×
10
8
M
. At separations
1
pc, even with high-
frequency very long baseline interferometry (VLBI), for all
but the closest active galactic nuclei (AGN), we lack the an-
gular resolution required to demonstrate the existence of an
SMBHB through imaging, and we have to look for other sig-
natures.
In principle, light curves offer a way forward (Haiman
et al. 2009; Burke-Spolaor et al. 2019), because SMBHBs
may reasonably be expected to exhibit periodicities. How-
ever, it has been pointed out (Vaughan et al. 2016; Covino
et al. 2019) that, notwithstanding the rich literature on pe-
riodicities and quasi-periodic oscillations (QPOs) in blazars
going back over five decades, there are very few statistically
solid results. In their detailed analysis of ten blazars in which
QPOs have been reported, Covino et al. (2019) show that no
strong cases for
∼
year-long periodicities can be confirmed.
They are all consistent with the power spectra of the varia-
tions in these objects. It requires careful modelling of the red
noise in the power spectrum of a blazar to evaluate the sig-
nificance of any claimed periodicity. Sandrinelli et al. (2017,
2018) had estimated that
∼
10% of bright
γ
-ray blazars are
QPOs, but after their detailed analysis they withdrew this es-
timate (Covino et al. 2019).
In a search for strong periodic signals showing at least 1.5
cycles in the optical light curves of 243,500 quasars, Graham
et al. (2015a) found 111 candidates, of which the strongest is
PG 1302
−
102 (Graham et al. 2015b). In this object approx-
imately sinusoidal variations have been seen over a span of
∼
20 years. However Vaughan et al. (2016) have challenged
the SMBHB interpretation of the PG 1302
−
102 optical light
curve, attributing the periodicity to red noise.
Since blazar light curves at radio wavelengths have a red
noise spectrum with a non-Gaussian probability density func-
tion (PDF) (Liodakis et al. 2017), except where stated other-
wise, we assume a red noise PDF throughout this paper.
Given the history and the persisting problems, it is clear
that great caution is needed in the identification of periodici-
ties and quasi-periodicities in active galactic nuclei. For this
reason we regarded the
P
⊕
= 4
.
69yr
±
0
.
14
yr earth-frame
observed periodicity reported in the blazar PKS 2131
−
021
by Ren et al. (2021) as an interesting result to follow, but by
no means yet shown to be a strong QPO or SMBHB candi-
date. The Ren et al. (2021) paper is based entirely on 11 years
(2008-2019) of our own 15 GHz monitoring observations of
PKS 2131
−
021 with the Owens Valley Radio Observatory
(OVRO) 40 m Telescope.
We recently came across observations of PKS 2131
−
021
(O’Dea et al. 1986) made at the Haystack Observatory be-
tween 1975 and 1983, which show the same periodicity to
<
∼
2%
, and phase to within 10% of the period. As we show
with extensive tests in this paper, the level of significance
of this periodicity, when 45.1 years of radio monitoring data
are combined, is
4
.
6
σ
, and it is certainly not a red noise
phenomenon. This makes PKS 2131
−
021 a strong QPO
or SMBHB candidate. In addition to the Haystack data,
we have also added the 14.5 GHz light curve of the Uni-
versity of Michigan Radio Astronomy Observatory (UM-
RAO), which covers the period 1980-2012, and is in excel-
lent agreement with the Haystack and OVRO light curves in
the regions of overlap, thereby giving us an uninterrupted,
well-sampled, 45.1-year 14.5 GHz - 15.5 GHz light curve of
PKS 2131
−
021.
All of the data presented in this paper are from targeted ob-
servations of PKS 2131-021, i.e. they are not serendipitous
observations in which PKS 2131
−
031 was observed in the
fields of other objects. Given the inhomogeneous process-
ing procedures, the diverse sets of flux calibrators, and the
observed matched flux densities in the overlapping regions,
it is clear that the unusual light curve of PKS 2131
−
021 is
not a result of faulty processing. These demonstrate the basic
light-curve integrity.
At the start of this project we confirmed four predictions
related to the periodicity seen in the OVRO data. This con-
vinced us that the periodicity is telling us something impor-
tant about the physics of this object and is not simply a ran-
dom variation due to red noise. We began with a least squares
sine wave fit to the OVRO data, on the basis of which we
PKS 2131
−
021: A U
NIQUE
SMBHB C
ANDIDATE
3
predicted that the sine wave oscillations had begun before
the OVRO observations, so we extrapolated the sine wave
backwards and began a search for earlier data. Our predic-
tion was that we would find earlier sinusoidal variations in
phase with the OVRO observations. The first confirmed pre-
diction came when we looked at MOJAVE (Lister et al. 2018)
and UMRAO data going back to 1995 and found an in-phase
peak in 2005, immediately preceding the first OVRO peak.
We subsequently obtained UMRAO data going back to 1980
and we found a second in-phase peak in 1982. This was the
second confirmed prediction. The peak in 1982 was much
larger than the peaks from 2005 to 2020, and we predicted
that had we been observing prior to 1982 we would have seen
a larger sinusoidal oscillation prior to 1982. At the time we
thought there were no data on PKS 2131-021 earlier than the
UMRAO data. However, through a literature search, we dis-
covered the Haystack data, which began in 1975. This did
indeed show an earlier in-phase peak, in 1976. This was the
third confirmed prediction. Furthermore, as predicted, like
the peak in 1982, the peak in 1976 was much larger than the
sinusoidal variations from 2005 to 2020. This was the fourth
confirmed prediction. The chain of events occurred exactly
as described here, and convinced us of the significance of the
OVRO periodicity, which, as we show in this paper, has been
confirmed by rigorous statistical analysis.
There are clearly several possible explanations for the peri-
odicities observed in PKS 2131
−
021. Precession of the rela-
tivistic jet due to misalignment of the spin axis of the SMBH
and accretion disk (Caproni et al. 2004) is one possibility.
Alternatively, the periodicity could be the result of preces-
sion due to misalignment of the orbital plane of a SMBHB
with the accretion disk of the more massive SMBH (Caproni
et al. 2017). Another possibility is precession due to warping
of the accretion disk (Britzen et al. 2018; Abraham 2018).
While these may all be viable explanations of the period-
icity we see in PKS 2131
−
021, there is a more straight-
forward explanation – namely that the periodicity is simply
due to the orbital motion of the SMBHB. We show here that
all the observations can be explained by this simple model,
and that no precession is needed to explain the light curve
of PKS 2131
−
021, although, as we show, it might well ex-
plain the large-scale morphology. It should be noted that,
given the characteristic time-scales for variability in blazars,
which range from months to years, and which are likely dom-
inated by fuelling of the central engine, and given the multi-
ple sites of radio emission along the jets, it is not difficult to
invent models in which the sinusoidal signal switches on and
off. Thus the appearance and disappearance of the sinusoidal
variability is easily accommodated in any model, and for this
reason we do not discuss it further in this paper. Since the
SMBHB orbital motion explanation is the simplest, we apply
Occam’s Razor. We deliberately do not consider other possi-
ble explanations in this paper, since it seems to us that nature
is pointing the way. Adopting the simplest explanation is the
best way to proceed at this early stage in our understanding of
the phenomenology of SMBHBs with relativistic jets. Sim-
ple orbital motion was suggested as an explanation of blazar
periodicities by Sobacchi et al. (2017), but their model is a
complex one, which does not produce sinusoidal variations
in all circumstances, and not at all unless the jet itself is as-
sumed to consist of a fast-moving “spine” surrounded by a
slower-moving “sheath”. While this might well be the case
in PKS 2131
−
021, it is not required by our model, which is
the simplest possible SMBHB-relativistic jet model.
We wish, therefore, to make clear at the outset that in this
paper we deliberately focus on a particular model, to the
exclusion of other viable models, because the phenomenol-
ogy of the sinusoidal flux density variations – which has not
been anticipated in previous studies – is an inevitable and
inescapable consequence of the SMBHB orbital motion of
the relativistic jet, and we are strongly of the opinion that
we should pursue this approach until the phenomenology
re-
quires
more complex models.
In this paper we consider all periodicities with significance
levels below
3
σ
to be red noise, unless other, uncorrelated
and independent, observations raise the significance to the
≥
3
σ
level. We analyze the PKS 2131
−
021 light curve and
show that it is unique amongst AGN that have been consid-
ered as possible SMBHB candidates, and is indeed a prime
SMBHB candidate. In § 2 we describe the observations;
in § 3 we analyze the light curve using three different ap-
proaches and taking great care to model the red noise cor-
rectly and hence to derive robust measures of the significance
of our results, and derive an upper limit to the chirp mass of
the putative SMBHB based on the radio observations alone;
in § 4 we present a model of PKS 2131
−
021, in which the
observed periodicity is the orbital period of the putative black
hole binary, which can account for the sinusoidal shape and
amplitude of the periodic variability that we see; in § 5 we
discuss the expected gravitational wave strain and derive an
upper limit on the chirp mass of the SMBHB based on the up-
per limits derived from pulsar NANOGrav observations (Ag-
garwal et al. 2019).
The redshift of PKS 2131
−
021 is
z
= 1
.
285
(Drinkwater
et al. 1997; Rector & Stocke 2001; Sbarufatti et al. 2006),
which we have recently confirmed (see § 2.6). For consis-
tency with our other papers, we assume the following cosmo-
logical parameters:
H
0
= 71
km s
−
1
Mpc
−
1
,
Ω
m
= 0
.
27
,
Ω
Λ
= 0
.
73
(Komatsu et al. 2009). On this model the co-
moving coordinate distance of PKS 2131
−
021 is 3.97 Gpc,
the angular diameter distance is 1.74 Gpc, and the luminos-
ity distance is 9.08 Gpc. None of the conclusions would be
changed were we to adopt the best model of the Planck Col-
laboration (Planck Collaboration et al. 2020).