Constraints on the optical polarization source in the luminous non-blazar quasar 3C 323.1 (PG 1545+210) from the photometric and polarimetric variability

0.67 MB·

by  Mitsuru Kokubo

MNRAS000,1–15(2017) Preprint17January2017 CompiledusingMNRASLATEXstylefilev3.0
Constraints on the optical polarization source in the
luminous non-blazar quasar 3C 323.1 (PG 1545+210) from
the photometric and polarimetric variability
Mitsuru Kokubo,1,2⋆
1Department of Astronomy, School of Science, the Universityof Tokyo, 7-3-1 Hongo, Bunkyo-ku,Tokyo 113-0033, Japan
2Institute of Astronomy, the Universityof Tokyo, 2-21-1 Osawa, Mitaka, Tokyo181-0015, Japan
7
1
0
2 Accepted2017January10.Received2017January08;inoriginalform2016November29

n
a
J ABSTRACT

3
1 We examine the optical photometric and polarimetric variability of the luminous
type 1 non-blazar quasar 3C 323.1 (PG 1545+210).Two optical spectro-polarimetric
] measurements taken during the periods 1996−98 and 2003 combined with a V-band
A
imaging polarimetric measurementtaken in 2002 revealthat (1) as noted in the liter-
G ature,thepolarizationof3C323.1isconfinedonlytothecontinuumemission,thatis,
the emission from the broad line region is unpolarized; (2) the polarized flux spectra
.
h show evidence of a time-variable broad absorption feature in the wavelength range
p
of the Balmer continuum and other recombination lines; (3) weak variability in the
o- polarizationpositionangle(PA)of∼4degoveratime-scaleof4−6yearsisobserved;
r and(4)theV-bandtotalfluxandthe polarizedfluxshowhighlycorrelatedvariability
t
s over a time-scale of one year. Taking the above-mentioned photometric and polari-
a metric variability properties and the results from previous studies into consideration,
[ we propose a geometrical model for the polarization source in 3C 323.1, in which an

equatorial absorbing region and an axi-asymmetric equatorial electron-scattering re-
1
v gion are assumed to be located between the accretion disc and the broad line region.
8 Thescattering/absorbingregionscanperhapsbeattributedtotheaccretiondiscwind
9 or flared disc surface, but further polarimetric monitoring observations for 3C 323.1
7 and other quasarswith continuum-confinedpolarizationare needed to probe the true
3 physical origins of these regions.
0
. Key words: accretion, accretion discs – galaxies: active – galaxies: nuclei – polar-
1
ization – quasars:individual (3C 323.1)
0
7
1
:
v
i 1 INTRODUCTION type 2 objects is generally perpendicular to the radio jet
X
structure (note that the jet axis is thought to be parallel
It has long been known that the ultraviolet(UV)-optical
r to the accretion disc’s rotation axis), as expected by the
a emission of non-blazar active galactic nuclei (AGNs)
polar scattering model (e.g., Antonucci 1983; Brindle et al.
often shows weak linear polarization. The polarization
1990; Smith et al. 2004). On the other hand, in the case
observed in type 2 AGNs is considered to be the result of
of type 1 AGNs with a prominent radio jet structure,
electron/dust scattering of the AGN nuclear emission from
polarization is generally known to be shown parallel to the
the polar scattering region; photons from the accretion
radiostructure(e.g.,Stockman et al.1979;Antonucci1983;
disc and the broad line region (BLR) behind the obscuring
Stockmanet al. 1984; Schmidt& Smith 2000; Smith et al.
dust torus are scattered into the observer’s line of sight
2002; Kishimoto et al. 2004). Even for radio-quiet objects,
by the polar scattering region and, thus, the blue disc
high-spatial-resolution imaging observations have revealed
continuum and BLR emission appear in the polarized
there to be a significant correlation between the direction
flux spectra (Antonucci& Miller 1985). Observations have
of the extended emission region (which is another good
revealed that the polarization position angle (PA) in
tracer of the accretion disc’s rotation axis) and the po-
larization PA, in that the former tends to be parallel to
the latter in type 1 objects and perpendicular in type 2
⋆ E-mail:[email protected]
(cid:13)c 2017TheAuthors
2 M. Kokubo
objects (Zakamskaet al. 2005; Borguet et al. 2008). It is scattering produces no net polarization in the observed
generally difficult to explain the “parallel” polarization BLR emission because the polarization is cancelled out
observed in type 1 objects by the polar scattering or the when the polarized BLR flux vectors are averaged over
polarization induced within the atmosphere of a plane- a wide range of angles, and thus the polarization due to
parallel scattering-dominated disc (e.g., Antonucci 1988; equatorial electron scattering is confined only to the accre-
Kishimoto et al. 2003). In addition, some authors have tion disc continua (see also Kishimoto et al. 2008a). One
claimed that the intrinsic quasar accretion disc continua interestingconsequenceofKishimoto’sinterpretationisthat
must be completely depolarized because of the strong because the electron (Thomson) scattering cross-section is
Faraday depolarization with magnetic fields in the disc wavelength-independent,thepolarized fluxspectra in these
atmosphere (Agol & Blaes 1996; Silant’ev et al. 2009, and quasars should reflect the spectral shape of the intrinsic
references therein). Instead, the observed optical polar- accretion disc continua hidden under the strong BLR
ization properties in type 1 quasars can be understood emission (Antonucci 2002; Kishimoto et al. 2003, 2004;
by assuming a geometrically and optically thin equatorial Hu& Zhang2012;Marin & Goosmann2013).Interestingly,
electron-scattering region located inside the dust torus; Kishimoto et al.(2004)showedthatthepolarizedfluxspec-
photons produced inside this electron-scattering region traofquasarswithcontinuum-confinedpolarization sharea
can be scattered into our line of sight, resulting in a similar spectral break feature at λrest ∼3600A˚, which they
net linear polarization “parallel” to the disc rotation axis interpreted as the Doppler-broadened hydrogen Balmer
(e.g., Stockman et al. 1979; Antonucci 1988; Smith et al. absorption edge imprinted in the intrinsic accretion disc
2004, 2005; Goosmann & Gaskell 2007; Kishimoto et al. thermalemissionduetothediscatmosphericopacityeffect.
2008a; Batcheldor et al. 2011; Gaskell et al. 2012; Kishimoto et al. (2004) further claimed that the overall
Marin & Goosmann 2013; Hutsem´ekerset al. 2015; spectral shape of these quasars (including the Balmer
Silant’ev et al. 2016). It is currently believed that all edge-like spectral break) can be well described by the
AGNs/quasars have both equatorial and polar scattering non-local thermodynamic equilibrium radiative-transferred
regions, and that differences in the observed polarization thermal accretion disc model of Hubenyet al. (2000) (see
propertiesbetweentype1andtype2objectsareduetothe also Hu & Zhang 2012).
orientation effect (e.g., Smith et al. 2004). However,Kokubo(2016)haspointedoutthatthespec-
In most type 1 Seyfert galaxies, it is observationally tral shape of the accretion disc continua of quasars with
known that not only are the UV-optical continua polarized continuum-confined polarization revealed as the variable
but also the BLR emission lines. polarization of the BLR component spectra (see Section 2.1) contradicts that re-
emissionlinesisoftenobservedtobeatalowerpolarization vealedasthepolarizedcomponentspectra,whichmayimply
degreeandadifferentPAthanthecontinuumemission,and thatequatorialelectronscatteringofthedisccontinuaalone
therotationofthePAcanbeseenasafunctionofthewave- probably cannot explain the optical polarization properties
length (e.g., Smith et al. 2002, 2004) This implies that the in these quasars. Moreover, it should be noted that, before
equatorial electron scattering region is more or less similar the deep spectro-polarimetric observations conducted by
in size to the BLR in these AGNs (e.g., Angel& Stockman Kishimoto et al. (2004), Schmidt& Smith (2000) had car-
1980;Smith et al.2005;Kishimoto et al.2008a;Baldi et al. riedoutspectro-polarimetryforquasarsamplesoverlapping
2016). However, optical spectro-polarimetric observations with those used by Kishimoto et al. (2004) and suggested
for luminous type 1 quasars1 with“parallel”optical polar- that the polarization in quasars with continuum-confined
ization carried out by Antonucci (1988), Schmidt& Smith polarizationmaybeattributedtoaweakopticalsynchrotron
(2000) and Kishimoto et al. (2003, 2004, 2005, 2008b) have componentfromthemisdirectedblazarcorecomponents.In
revealed that there is a population of quasars whose BLR fact,allofthefivequasars(3C323.1,Ton202,B21208+32,
emission lines are essentially unpolarized, i.e., the polariza- 3C 95, and 4C09.72) confirmed as showing continuum-
tion of these quasars is confined only to the continuum. confined polarization by both Schmidt& Smith (2000) and
The observed null polarization of the BLR emission rules Kishimoto et al.(2004)areradio-loudobjects;thus,itisun-
out the possibility that the optical polarization in these surprisingthatthereisafluxcontributionfromthejetsyn-
quasars is predominantly attributed to scattering processes chrotroncomponenttotheobservedopticalspectra(seee.g.,
outside the BLR (e.g., Smith et al. 1993; Kishimoto et al. Impeyet al. 1989; Smith et al. 1993 and Afanasiev et al.
2003).Thecontinuumpolarization PAsofthesequasarsare 2015 for the cases of 3C 273 and 3C 390.3). However, it
mostly wavelength-independent, indicating that there is a should be noted that subsequent work by Kishimoto et al.
single dominant mechanism producing the observed polar- (2008b) has identified two radio-quiet quasars (Q0144-3938
ization. and CTS A09.36) with continuum-confinedpolarization.
Kishimoto et al. (2003, 2004) interpreted the As discussed above, the optical polarization source in
continuum-confined polarization observed in several type 1 quasarswith continuum-confinedpolarization has yet tobe
quasars such that the equatorial electron scattering region fully understood. To probe the true nature of the optical
inthesequasarsissmallerinsizecomparedtotheBLR,but polarization mechanism, and consequently to examine the
stillmuchlargerthantheUV-opticalemittingregionsofthe accretion disc physics in luminous type 1 quasars, detailed
accretion disc.In sucha geometrical configuration,electron studiesof thequasar polarimetric variability must be a key
observable to constrain the geometry of the polarization
source (e.g., Stockmanet al. 1979; Schmidt& Smith 2000;
1 In this paper, we use the term“quasars”to refer to optically Gallagher et al. 2005; Gaskell et al. 2012; Afanasiev et al.
identified bright AGNs, including the radio-loud and the radio- 2015;Rojas Lobos et al.2016).Inthisworkweexaminethe
quietobjects. variability of the total flux, polarization degree, polariza-
MNRAS000,1–15(2017)
Photometric and polarimetric variability of 3C 323.1 3
tionPA,andthepolarized fluxintheluminoustype1non- and its position angle is ∼ 20 deg (Miley & Hartsuijker
blazar quasar 3C 323.1, which is a quasar with continuum 1978; Kellermann et al. 1994; Dennett-Thorpeet al. 2000;
confined polarization identifiedby Schmidt& Smith (2000) Schmidt& Smith 2000; Kishimoto et al. 2004). As is the
and Kishimoto et al. (2004), bycollecting archival data. case with the other quasars with the continuum-confined
The three polarimetry data sets used in this work, polarization, the radio jet axis of 3C 323.1 is known to
including the two optical spectro-polarimetric measure- be parallel to the direction of the optical polarization vec-
ments taken by Schmidt & Smith (2000) (Bok/SPOL in tor within . 10 deg (see Section 4; Stockman et al. 1979;
1996-1998) and by Kishimoto et al. (2004) (Keck/LRIS in Schmidt& Smith 2000; Kishimoto et al. 2004).In theopti-
2003) and a V-band imaging-polarimetry measurement by calwavelengthrange,3C323.1isobservedasapointsource
Sluseet al. (2005) (ESO3.6-m/EFOSC2 in 2002), are de- with an apparent V-band magnitude of V ∼ 16 mag. The
scribed in Section 2. The absolute V-band magnitudes at optical multi-band colour of 3C 323.1 is within the spec-
the epochs of the spectro-polarimetric measurements are troscopic target selection criteria in the Sloan Digital Sky
carefully estimated by using broad-band light curves taken Survey(SDSS)LegacySurvey(Richardset al.2002),mean-
from theliterature. In Section 3,we first examine thespec- ing that this object is“normal”in terms of its optical spec-
tral variability of the polarimetric properties between the trum(seealsoBoroson & Green1992).Theblackholemass
twospectro-polarimetricmeasurements,andidentifytheori- (M )andtheEddingtonratio(definedastheratioof the
BH
gin of the different conclusions between Schmidt& Smith bolometric luminosity L to the Eddington luminosity L )
E
(2000) and Kishimoto et al. (2004) regarding the polariza- of3C323.1wereestimatedtobelog(MBH/M⊙)=9.07and
tion source in the quasars with continuum-confined polar- L/L =0.10, respectively, based on the single-epoch SDSS
E
ization describedabove. WethenexaminetheV-bandpho- spectrum obtained at MJD=53886 (Shenet al. 2011).
tometric and polarimetric variability of 3C 323.1 using all Recently,Kokubo(2016) has presented the multi-band
three polarimetric measurements in order to constrain ob- (u,g,r,i,andz-band)photometriclightcurvesfor3C323.1
servationallythegeometryoftheopticalpolarizationsource obtainedin2015-2016,andhasexaminedthespectralshape
in3C323.1inSection4.Finally,asummaryandconclusions of the UV-optical variable component spectrum in this ob-
are provided in Section 5. ject. Kokubo (2016) showed that the variable component
spectrum in 3C 323.1 is consistent with a power-law spec-
trum with α ∼ +1/3 (in the form of F ∝ ναν). The
ν ν
spectral index of α ∼ +1/3 is commonly detected in the
2 DATA ν
variable component spectra of (mostly radio-quiet) quasars
Throughout thispaper, we assume a cosmology of H0 =73 (e.g., Pereyra et al. 2006; Ruan et al. 2014; Kokuboet al.
km/s/Mpc,Ωm =0.27,andΩΛ =0.73(Spergel et al.2007). 2014; MacLeod et al. 2016; Hunget al. 2016; Buisson et al.
None of the data are corrected for Galactic extinction. In 2017), and is usually attributed to the well-known predic-
this study,“the total flux”(Fλ or Fν) has the same mean- tion of the spectral index αν = +1/3 of the standard thin
ing as the un-normalised Stokes parameter I expressed in thermal accretion disc emission (Shakura& Sunyaev 1973;
units of the flux. q and u represent the normalised Stokes Novikov& Thorne 1973; Franket al. 1992). This means
parametersdefinedasQ/I andU/I,whosepolarizationPA that it is natural to consider that the flux variability of
isdefinedusingtheequatorialcoordinatesystem(aPAof0 3C 323.1 is due to the intrinsic variations of the disc emis-
degcorrespondstoNorth-South,andthePAincreasesfrom sion itself, in line with other non-blazar quasars; neverthe-
NorthtowardsEast).Thepolarizationdegreep≡ q2+u2 less, the precise mechanisms for the quasar flux variability
obtained from the observational data with finitepmeasure- are still under debate (see Kokubo 2015, 2016, and refer-
ment errors2 is known to be a biased estimate of the true encestherein).Theobservedvariabilitytime-scaleofmonths
polarization degree; therefore,thede-biasedestimateof the to years in 3C 323.1 is also consistent with the variability
polarization degree, pˆ≡ p2−σ2 (e.g., Plaszczynski et al. of the quasar accretion disc emission. As discussed in Sec-
2014),isusedthroughoutpthisstudy.Therotatednormalised tion 4,theattributionof thefluxvariability totheintrinsic
Stokesparametersq′ andu′ aredefinedinacoordinatesys- disc emission variability, combined with the observed cor-
temrotatedtothesystemicpolarizationpositionanglePA related variability between the total flux and the polarized
c
of the target; note that all of the polarization, in principle, flux,impliesthatthepolarizedfluxisdirectlyrelatedtothe
falls in the single Stokes parameter q′ when the object has discemissionanddoesnotarisefromtheopticalsynchrotron
a wavelength-independentPA. fluxcontribution.
2.1 Observational properties of 3C 323.1: the
2.2 Polarimetry data
variable component spectrum
In this work, we use the three archival polarimetric
3C 323.1 (PG 1545+210) is a Fanaroff-Riley II (FRII)
measurements for 3C 323.1 obtained at different epochs
radio-loud quasar at z = 0.264 (the luminosity distance
withBok/SPOL,ESO3.6-m/EFOSC2,andKeck/LRIS.The
d = 1293.3 Mpc; Wright 2006). The lobe-dominant ra-
L
dio jet structure of 3C 323.1 has a size of ∼ 300 kpc, ESO3.6-m/EFOSC2 observation is V-band imaging po-
larimetry (Sluseet al. 2005), and the Bok/SPOL and the
Keck/LRIS observations are optical spectro-polarimetry
2 Forthepolarimetricobservationsforweaklypolarizedobjects, (Schmidt& Smith2000;Kishimoto et al.2004).Thecompi-
themeasurementerrorsassociatedwithqanduisessentiallythe lation ofthesepolarimetric datafrom theliteratureenables
same(σ≡σq ≃σu). to examine the two-epoch spectral variability (Section 3)
MNRAS000,1–15(2017)
4 M. Kokubo
Rest−Frame Wavelength [Å] and the three-epoch V-band variability (Section 4) of the
3000 3500 4000 4500 5000 5500 polarimetric propertiesof 3C 323.1.
4
nt] Bok/SPKOecLk i/nL R1I9S9 6in− 21090938 TheV-bandmagnitudesof thetotalfluxat theepochs
ce 3 Difference of the three polarimetric measurements are also estimated
per (Sections 2.2.1 and 2.3). As discussed in Section 4, one of
q’ [ 2 the most important subjects of this work is to examine the
s
ke correlationbetweenthetotalfluxandthepolarizedfluxlight
o 1
St curves, which eventually provides strong constraints on the
d
ate 0 optical polarization source in 3C 323.1.
Rot Following the same procedure of Kishimoto et al.
−1 (2004), we correct the three polarimetric measurements for
4000 5000 6000 7000
the effects of Galactic interstellar polarization (ISP). The
Observed Wavelength [Å]
details of the method of the ISP correction is summarized
Rest−Frame Wavelength [Å]
3000 3500 4000 4500 5000 5500 in Section A. Throughout this paper, the characters with
4 nosubscriptrepresentISP-correctedvaluesunlessotherwise
ent] Bok/SPKOecLk i/nL R1I9S9 6in− 21090938 stated.
c 3 Difference
er
p
u’ [ 2
s 2.2.1 ESO3.6-m/EFOSC2 V-band imaging-polarimetry
e
k
o 1
St Sluseet al.(2005)presentedtheresultsofV-bandimaging-
d
ate 0 polarimetry for 3C 323.1 obtained from the 3.6-m tele-
Rot scope at the European Southern Observatory (ESO),
−1 La Silla, equipped with the ESO Faint Object Spectro-
4000 5000 6000 7000
graph and Camera (v.2) (EFOSC2) on the 1st May, 2002
Observed Wavelength [Å]
(MJD=52396.3; see also Hutsem´ekerset al. 2005). As part
Rest−Frame Wavelength [Å]
ofthis,theytookasetoffour60-secintegrationimageswith
3000 3500 4000 4500 5000 5500
4 thehalf-waveplatepositionanglesof0.0deg,22.5deg,45.0
Bok/SPOL in 1996−1998
Keck/LRIS in 2003 deg, and 67.5 deg.
3 Difference
We downloaded the ESO3.6-m/EFOSC2 imaging-
er cent] 2 FploalatrPimoleItmray fdraatmaeosff3roCm32th3.e1EaSnOd tShceieanscseocAiartcehdiveBIFAaScialintdy
^p [p 1 (Program ID 68.A-0373). After applying bias and flat cor-
rections, the V-band polarization degree pˆ and the polar-
V
0
ization position angle PA of 3C 323.1 were calculated by
V
adopting the aperture photometry and the“ratio”method
−1
4000 5000 6000 7000 (e.g., Lamy & Hutsem´ekers 1999; Bagnulo et al. 2009). By
Observed Wavelength [Å] subtracting the instrumental polarization using the val-
Rest−Frame Wavelength [Å] ues presented in Sluse et al. (2005), we obtained the ISP-
3000 3500 4000 4500 5000 5500 uncorrectedV-bandpolarizationdegreeandpolarizationpo-
40
Bok/SPKOecLk i/nL R1I9S9 6in− 21090938 sition angle as pV,uncorr =1.14±0.13 %, pˆV,uncorr =1.13 %
30 Difference and PAV,uncorr = 17.88 ± 3.39 deg, which are consistent
e] 20 with those reported by Sluseet al. (2005)3. By correcting
gre for the Galactic ISP (Section A), we obtain pˆV = 1.94 %
e 10
A [d and PAV =10.03 deg (Table 1).
P 0 To evaluate the V-band magnitude at the epoch
of the EFOSC2 polarimetry measurement, we applied
−10
differential photometry by using two field stars (SDSS
−20 J154736.44+205141.7 and SDSS J154752.41+204931.8) si-
4000 5000 6000 7000
multaneously imaged on the same EFOSC2 frames along
Observed Wavelength [Å]
with 3C 323.1 as reference stars. By using the SDSS g
Figure1.ComparisonoftherotatedStokesparametersq′andu′, and r-band point spread function (PSF) photometry and
de-biasedpolarizationdegreepˆ,andthepolarizationpositionan- the transformation equation by Jester et al. (2005) from
gle(PA)of 3C 323.1 between the spectro-polarimetricmeasure- the SDSS g and r-band magnitude to V-band magnitude,
ments of Bok/SPOL (Schmidt&Smith 2000) and Keck/LRIS V =g−0.59(g−r)−0.01magwithanuncertaintyof0.01
Vega
(Kishimotoetal. 2004). The difference spectra between the two mag,theV-bandmagnitudesofSDSSJ154736.44+205141.7
measurementsarealsoplotted.Thespectraarebinnedintocom-
and SDSS J154752.41+204931.8 are estimated as V =
mon wavelength bins with10 A˚width to allow easy comparison Vega
between the two measurements. Note that q′ and u′ are defined
byassumingthesystemicPAasPAc=14.2degand9.6degfor 3 Sluseetal.(2005)reportedtheISP-uncorrectedV-bandpolar-
Bok/SPOLandKeck/LRISdata, respectively.
ization degree and the polarization position angle of 3C 323.1
as pV,uncorr = 1.15 ± 0.13%, pˆV,uncorr = 1.14% and
PAV,uncorr=18±3deg.
MNRAS000,1–15(2017)
Photometric and polarimetric variability of 3C 323.1 5
17.0516 mag and V = 16.6971 mag, respectively. For Figure 9 of Kishimoto et al. (2004). As listed in Table 6 of
Vega
each image, aperture photometry was applied to combine Kishimoto et al. (2004), the systemic polarization position
the two orthogonally polarized fluxes of 3C 323.1 and the angle PA of 3C 323.1 at the epoch of the Keck/LRIS ob-
c
tworeferencestars.Byaddingthe0.01magerrorduetothe servationattherestframewavelengthrangeof4000-4731˚A
transformationequationofJester et al.(2005)tothephoto- is PA =9.6 deg.
c
metricerrors,weevaluatedtheweightedaveragemagnitude Figure 1 shows the ISP-corrected spectra of the ro-
of3C323.1anditsuncertaintyasV =15.7779±0.0106 tated Stokes parameters q′ and u′, de-biased polarization
Vega
mag or V = 15.7339±0.0113 mag, where the AB offset degree pˆ, and PA of 3C 323.1 obtained with Bok/SPOL
AB
anditsuncertaintyaretakenfromFrei& Gunn(1994).The (Schmidt& Smith 2000) and Keck/LRIS (Kishimoto et al.
polarimetric and photometric measurements for 3C 323.1 2004). A discussion of the spectral variability of the polar-
obtained with ESO3.6-m/EFOSC2 are summarised in Ta- ization propertiesbetweenthesetwomeasurementsisgiven
ble 1. in Section 3.2.
To compare the Bok/SPOL and Keck/LRIS spectro-
polarimetricmeasurementswiththeESO3.6-m/EFOSC2V-
2.2.2 Bok/SPOL and Keck/LRIS optical
band imaging-polarimetry data, the V-band polarization
spectro-polarimetry
degree pˆ and the polarization position angle PA are
V V
3C 323.1 was spectro-polarimetrically observed by calculated from the Bok/SPOL and Keck/LRIS spectro-
Schmidt& Smith (2000) and Kishimoto et al. (2004) polarimetricdatabyconvolvingthetotalandpolarizedflux
during1996-1998 and 2003, respectively. spectra (Fλ, qλ ×Fλ and uλ×Fλ) with the V-band filter
Schmidt & Smith (2000) carried out spectro- transmissioncurvetakenfrom Bessell(1990).Theerrorson
polarimetric observations of 3C 323.1 at the Steward pˆV andPAV areevaluatedasthesamplestandarddeviation
Observatory 2.3-m Bok Telescope using the SPOL CCD oftheestimates from the1000trials of theMonteCarlo re-
Imaging/Spectropolarimeter (Schmidtet al. 1992) with samplingofthespectra.TheV-bandpolarimetricproperties
a low-resolution grating. The observed wavelength range of 3C 323.1 are summarized in Table 1.
was λ = 4000 − 8000 ˚A. The data presented in Theabsolutefluxcalibrationforthetotalflux(andthe
obs
Schmidt& Smith (2000) comprise 10 observations from polarizedflux)spectraofthetwospectro-polarimetricmea-
several epochs taken in June 1996 (5 observations) and surementsis discussed in Section 2.3.
April 1998 (5 observations), and thus the combined data
represent the average polarization properties of 3C 323.1
at MJD∼50235 and 50933. Since the raw Bok/SPOL data 2.2.3 Note on other historic polarimetric measurements
of 3C 323.1 are unavailable, in this work, we directly use for 3C 323.1
the combined calibrated polarization spectra of 3C 323.1
Otherthanthethreepolarimetric datadescribedabove,we
presented in Schmidt& Smith (2000), which are kindly
can find several historic polarimetric measurements in the
providedbyG.D.SchmidtandP.S.Smithinelectricform.
literature (Stockmanet al. 1984; Wills et al. 2011). How-
The Galactic ISP is corrected as described in Section A.
ever, most of these measurements are obtained with white-
Following the definition of Kishimoto et al. (2004), the
light (no-filter) configuration, and thus, it is difficult to di-
systemic polarization position angle PA of 3C 323.1
c rectly compare them with those obtained with Bok/SPOL,
at the epoch of Bok/SPOL observations is evaluated to
ESO3.6-m/EFOSC2, and Keck/LRIS because of theuncer-
be PA = 14.2 deg as a weighted average of the PA
c tainty of the wavelength coverage of the data; because the
at the rest-frame wavelength range of 4000-4731A˚; thus,
polarimetricpropertiesof3C323.1showwavelengthdepen-
the rotated Stoke parameters q′ and u′ are defined in a
dence(seeSection3),itiscriticalforthestudyofthepolari-
coordinate system rotated to PA =14.2 deg.
c metricvariability tousedatatakenin thesamewavelength
Kishimoto et al.(2004)carriedoutspectro-polarimetry
range. Also, unlike the three polarimetric data used in this
for 3C 323.1 on the 4th May, 2003 (MJD=52763.6), with
work, it is impossible to evaluate the magnitude values of
theLow Resolution ImagingSpectrometer(LRIS)mounted
thetotal fluxof 3C 323.1 at theepochsof thesewhite-light
on the 10-m Keck-I telescope at the W. M. Keck Observa-
polarimetricdatabecauseoflackofreferenceablephotomet-
tory(Okeet al.1995;Goodrich et al.1995;McCarthy et al.
ric measurements. For these reasons, we do not include the
1998),usingtwogrismsof300l/mmand400l/mm.Theob-
historicwhite-lightpolarimetricmeasurementsfor3C323.1
servationsof3C323.1(76minutesintotal)consistedoftwo
inthemaintext;theyareinsteadsummarizedanddiscussed
setsoffourwave-platepositions(0.0deg,22.5deg,45.0deg,
in Section B.
and 67.5 deg) × 7.5 minute observations with 300 l/mm
and a single set of four wave-plate positions × 4 minute
observations with 400 l/mm. As noted in Kishimoto et al.
2.3 Spectro-photometric flux calibration for the
(2004), the observed spectra at the wavelength range at
Bok/SPOL and Keck/LRIS data
λ &6500 ˚A may be affected by second-order light con-
obs
tamination; therefore, the usable wavelength range for the The absolute flux calibration of the spectroscopic data is
analysis is λ = 3500 − 6500 ˚A. In this work we use generally not quite as accurate as that of the broad-band
obs
the calibrated polarization spectra of 3C 323.1 presented photometry data. To confidently evaluate the variability of
in Kishimoto et al. (2004), which are kindly provided by the total and polarized fluxes in 3C 323.1 at the epochs
M.Kishimotoinelectricform(seeKishimoto et al.2004,for of the spectro-polarimetric observations by Bok/SPOL and
details of thedataanalysis). The ISP-correctedKeck/LRIS Keck/LRISdescribedinSection2.2.2,wecollectbroad-band
data used in this work are the same with those shown in photometric data from the literature and publicly available
MNRAS000,1–15(2017)
6 M. Kokubo
Table 1.SummaryoftheISP-correctedV-bandpolarimetricandphotometricvaluesfor3C323.1.
Date MJD Instrument pˆV[%] PAV[deg] V-bandABmag.ofthetotalflux
1996-06and1998-04 50235-50933 Bok/SPOL 2.06±0.01 13.75±0.20 15.8935±0.0358
2002-05-01 52396.3 ESO3.6-m/EFOSC2 1.94±0.13 10.03±3.39 15.7339±0.0113
2003-05-04 52763.6 Keck/LRIS 2.11±0.01 9.61±0.10 15.5838±0.0226
databases(Figure2),andusethemtoestimate theV-band tra should have additional errors due to flux variability
magnitudeof the total flux. during the two Bok/SPOL observations in June 1996 and
April 1998. Nevertheless, the magnitude uncertainty of the
Bok/SPOLdatadoesnotaffectthefinalresult ofthiswork
2.3.1 B- and R-band light curves in 1991-1998 inthelongtermbecausetheobservationalconstraintonthe
polarization source from the well-correlated total flux and
3C 323.1 was photometrically observed for seven years
thepolarizedfluxvariability(discussedinSection4)canbe
(1991-1998) in the Johnson-Cousins B- and R-band with
verified even from only the two measurements of ESO3.6-
theWiseObservatory1-mtelescope(seeGiveon et al.1999,
m/EFOSC2andKeck/LRISwhosemagnitudescanbecon-
for details). The B- and R-band light curves of 3C 323.1
fidentlyevaluated.
presented by Giveon et al. (1999)4 are shown in Figure 2.
In the same figure, the epoch of Schmidt& Smith (2000)’s
Bok/SPOL spectro-polarimetric observations carried out in 2.3.2 SDSS photometry and LINEAR light curve during
June1996 and April 1998 is indicated byvertical bars. 2003-2008
Althoughthereisnoavailablebroad-bandphotometric
3C323.1wasphotometricallyobservedbytheSDSSLegacy
measurement obtained simultaneously with the Bok/SPOL
Survey, using the SDSS imaging camera mounted on the
data, we can infer the broad-band magnitude of 3C 323.1
Sloan Foundation 2.5-m telescope at the Apache Point Ob-
at the periods of the Bok/SPOL observations from the
servatory (Gunnet al. 1998; Yorket al. 2000; Gunn et al.
photometry data of Giveon et al. (1999). We assume that
2006).WeusetheSDSSg-bandPSFmagnitudeof3C323.1,
the weighted average of the four R-band measurements of
g = 15.5606 ± 0.0174 mag, observed on the 12th June,
Giveon et al. (1999) from MJD=50174 to 50690, R =
Vega
15.6793±0.0353 mag or R =15.7963±0.0358 mag (the 2004(MJD=53168.3),retrievedfromSDSSDataRelease12
AB
(DR12)SkyServer(Alam et al.2015).WecheckedtheSDSS
AB offset and its uncertainty are taken from Frei & Gunn
photometry flag, and confirmed that the g-band photome-
1994), represents the magnitude value at the epoch of the
try satisfies the “clean photometry” criteria recommended
observation of Schmidt& Smith (2000). Here, we assume
on theSDSSweb page5.
the uncertainty in R as 0.0353 mag, which is derived
Vega
In addition, 3C 323.1 was observed between 2003 and
not from thestatistical error butfrom thesample standard
2008 as part of the Massachusetts Institute of Technol-
deviation of thefour measurements.
ogy Lincoln Laboratory Lincoln Near-Earth Asteroid Re-
The Bok/SPOL total flux spectrum for 3C 323.1 is
search (LINEAR) survey using the 1-m LINEAR telescope
scaled so that the R magnitude calculated by convolv-
AB
(Stokeset al. 2000; Sesar et al. 2011). The LINEAR obser-
ingtheBok/SPOL totalfluxspectrumwith theR-bandfil-
vations were carried out with an unfiltered set-up, and the
ter transmission curve taken from Bessell (1990) coincides
magnitudevaluesstored in theLINEARSurveyPhotomet-
with R = 15.7963 mag. Then, the polarized flux spec-
AB ric Database 6 are given in the LINEAR recalibrated mag-
tra are defined as the product of the polarization degree
nitude(Sesar et al. 2011).
spectra in Figure 1 and thescaled total fluxspectrum.The
The SDSS photometry data and the LINEAR light
V-bandmagnitudeofthescaledtotalfluxBok/SPOLspec-
curve for 3C 323.1 from MJD=52665 to 54582 are plot-
trum is calculated as V = 15.8935 mag, by using the
AB
tedinFigure2.ThemultipleLINEARmeasurementstaken
V-band filter transmission curve taken from Bessell (1990).
on the same night are binned to a weighted-average mag-
Because the measurement errors on the Bok/SPOL total
nitude for each observation night. In the same figure, the
flux spectrum are negligible compared to the R-band mag-
epoch of the Keck/LRIS spectro-polarimetric observation
nitude uncertainty of 0.0358 mag, we adopt 0.0358 mag as
byKishimoto et al.(2004)on4thMay,2003(Section2.2.2),
the error of the estimated V-band magnitude. The spec-
tralmonochromaticluminosityat5100˚AisλL (5100˚A)= is indicated by a vertical bar. As shown in Figure 2, since
λ
1.1499(±0.0379)×1045 erg/s. The scaled Bok/SPOL total theLINEARphotometrydataweretakenapproximatelysi-
multaneouslywith boththeSDSSdataandtheKeck/LRIS
fluxspectrumandthepolarized fluxspectrumareshownin
spectro-polarimetry data, we are able to use the LINEAR
Figure 3.
photometry data to estimate the broad-band magnitude of
It should be noted that, since there is a time separa-
3C323.1attheepochoftheKeck/LRISobservation,asde-
tionbetweentheobservationsfromthebroad-bandmeasure-
scribed below.
ments in Giveon et al. (1999) and the Bok/SPOL spectro-
Because quasar spectra are significantly different from
polarimetric measurements, the spectro-photometric cali-
bration for the Bok/SPOL total and polarized flux spec-
5 http://www.sdss.org/dr12/algorithms/photo flags recommend/
6 TheLINEARSurveyPhotometricDatabaseisavailableatthe
4 http://wise-obs.tau.ac.il/givon/DATA/ SkyDOTWebsite(http://skydot.lanl.gov/)
MNRAS000,1–15(2017)
Photometric and polarimetric variability of 3C 323.1 7
15.2
15.4
15.6
4
Magnitude 11 561..682 02 May 1 2003 May GGiivveeoonn++11999999,, RB
une pril 20 Sesar+2011, SmDLSINSE,A Rg
16.4 6 J 8 A VAB in 1996-1998 (Bok/SPOL)
16.6 199 199 VAB on 20V0A2B M oany 2 10 0(E3 SMOa3y. 64- m(K/eEcFkO/LSRCIS2))
48000 49000 50000 51000 52000 53000 54000 55000
Modified Julian Date (MJD)
Figure 2. Compilationof the broad-band light curves for 3C 323.1. TheB- and R-band lightcurves at MJD=48355-50690 are taken
from Giveon et al. (1999), and the Lincoln Near-Earth Asteroid Research (LINEAR) survey recalibrated magnitudes (mLINEAR) at
MJD=52665-54582arefromSesaretal.(2011).TheSloanDigitalSkySurveyg-bandphotometryatMJD=53168.32isalsoplotted.The
twoperiodsoftheBok/SPOLspectro-polarimetryandtheepochoftheKeck/LRISspectro-polarimetryareindicatedbydashed-dotted
vertical bars. The estimated VAB magnitude at the epochs of the three polarimetric measurements (Bok/SPOL, ESO3.6-m/EFOSC2,
andKeck/LRIS) isalsoshown(Table1;seeSection2.2.1andSection2.3fordetails).
stellar spectra due to their strong broad emission lines, we Because the measurement errors on the Bok/SPOL total
do not adopt the LINEAR-to-SDSS photometric transfor- flux spectrum are negligible compared to the g-band mag-
mationequationderivedbyusingtheSDSSstellarphotom- nitude uncertainty of 0.0226 mag, we adopt 0.0226 mag as
etryinSesar et al.(2011).Instead,weassumethattheLIN- the error of the estimated V-band magnitude. The spec-
EAR magnitude m is related to the SDSS g-band tralmonochromaticluminosityat5100˚AisλL (5100˚A)=
LINEAR λ
magnitude as g = mLINEAR + mg,0. This assumption is 1.7392(±0.0362)×1045 erg/s. The scaled Keck/LRIS total
justified by the fact that the magnitude difference in the fluxspectrumandthepolarized fluxspectrumareshownin
LINEARmeasurements between theepochsof theobserva- Figure 3.
tionofKishimoto et al.(2004)andtheSDSSobservationis
as small as ∼0.07 mag, and thus the colour variability of
3C 323.1 between these epochs is also expected to be small
3 SPECTRAL VARIABILITY OF THE
(see e.g., Schmidtet al. 2012). The LINEAR magnitude at
POLARIMETRIC PROPERTIES
MJD=53165.26ism =15.4747±0.0123 mag,which
LINEAR
canbedirectlycomparedtotheSDSSphotometryobtained Inthissection,thespectralvariabilitybetweenthetwosets
threedayslater,atMJD=53168.3,withtheassumptionthat of spectro-polarimetry data obtained with Bok/SPOL and
the quasar flux variability on a time-scale of several days Keck/LRISis examined.
is essentially negligible. From these values, the magnitude
shifts are calculated as
3.1 Spectral variability of the polarization degree
g = mLINEAR+0.0859 (±0.0213) [mag]. (1) and the polarization position angle
Keck/LRIS spectro-polarimetry for 3C 323.1 was carried Figure 1 clearly shows that, as noted by Schmidt & Smith
out at MJD=52763.6, and the LINEAR photometry data (2000)andKishimoto et al.(2004),thepolarizationdegrees
are available at MJD=52753.35 and MJD=52771.30 as (q′, u′, and pˆ) at the wavelengths of the BLR emission
m = 15.4014 ± 0.0091 mag and m = are significantly diminished, making the polarization de-
LINEAR LINEAR
15.4110 ± 0.0140 mag, respectively (the weighted-average gree spectra strongly wavelength-dependent; in particular,
is m = 15.4043 ± 0.0076 mag). With the use of the unpolarized “small blue bump”, which is composed of
LINEAR
Equation 1, the g-band magnitude of 3C 323.1 at the theBalmer continuumand theUV FeIIpseudocontinuum
epoch of the Keck/LRIS observation can be estimated as from the BLR, is responsible for the decrease in the polar-
g=15.4902±0.0226 mag. ization degrees at λrest < 4000 ˚A. Although the spectra of
The Keck/LRIS total flux spectrum for 3C 323.1 is polarization degree do not show strong variability between
scaled so that the g magnitude calculated by convolving the two measurements, it should be noted that the polar-
the Keck/LRIS total flux spectrum with the g-band filter izationdegree(i.e.,pˆ)attherest-framewavelengthrangeof
transmission curve taken from Doi et al. (2010) coincides λrest ∼ 3600 ˚A shows variations between the two spectro-
with g = 15.4902 mag. Then, the polarized flux spectra polarimetric measurements(see Section 3.2 for details).
are defined as the product of the polarization degree spec- Each PA spectrum in the bottom panel of Figure 1 is
tra in Figure 1 and the scaled total flux spectrum. The V- nearly wavelength-independent. It is clear that, unlike the
band magnitude of the scaled total flux Keck/LRIS spec- mainly time-constant polarization degrees, the PA spectra
trum is calculated as V = 15.5838 mag, by using the show clear evidenceof variability. The difference in the PA
AB
V-band filter transmission curve taken from Bessell (1990). evaluated at the rest-frame wavelength range of 4000-4731
MNRAS000,1–15(2017)
8 M. Kokubo
Rest−Frame Wavelength [Å] like feature at around λrest ∼ 3600 ˚A. Kishimoto et al.
3000 3500 4000 4500 5000 5500 (2004) interpreted this feature as the Balmer continuum
8
Bok/SPOL in 1996−1998 absorption, but they also noted the possibility that other
7 Keck/LRIS in 2003
Å] Difference (scaled) higher-orderBalmer series absorption lines and weak metal
2/ 6
m absorption lines also contribute to the absorption feature.
s/c 5 However, this spectral feature seems to be weak or absent
−150 erg/ 34 tinailtehdesBhoakp/eSoPfOthLepˆtλwo×pˆFλλ×sFpeλcstpruemct.raInatlesroessthionwglsy,evtihdeendcee-
[1 2 of variability at the wavelength range of the Hβ emission
Fl 1 line(λrest ∼4861˚A);thereisabroadabsorption-likefeature
0 at around the wavelength range of the Hβ emission line in
4000 5000 6000 7000 theKeck/LRISpolarizedfluxspectrum,butitisnotclearly
Observed Wavelength [Å] seen in the Bok/SPOL polarized flux spectrum. The spec-
Rest−Frame Wavelength [Å] tral variability of pˆ ×F at the wavelength ranges of the
λ λ
3000 3500 4000 4500 5000 5500 Balmer continuum and the Hβ emission line can clearly be
10
Bok/SPOL in 1996−1998 seen in the difference spectrum shown in the bottom panel
2m/Å] 8 Keck/LRISa inn= 2+010/33 of Figure 3.
s/c 6 Difference (scaled)
g/
er 4
7
1
−0 2
1
[ Thespectro-polarimetricvariabilityof3C323.1seenin
x Fl 0 thepˆλ×Fλ spectra(Figure3)isverysimilartothatdiscov-
^p −2 eredbyKishimoto et al.(2004)inTon202;Kishimoto et al.
(2004) have identified spectro-polarimetric variability be-
4000 5000 6000 7000
tweenthetwosetsofKeck/LRISdataforTon202obtained
Observed Wavelength [Å]
in 2002 and 2003 (one year apart), where the Balmer-edge
Figure 3. The total flux (top) and the polarized flux spectra absorption-likefeatureseeninthepolarizedfluxspectrumin
(bottom)obtainedwithBok/SPOLandKeck/LRIS.Galacticex- 2002haddisappearedby2003.Kishimoto et al.(2004) sug-
tinction is uncorrected. The arbitrarily-scaled difference spectra gested that the spectro-polarimetric variability of Ton 202
between thetwomeasurementsarealsoplotted. Thespectraare
might be related to the time-variability of the geometry of
binned into 10-A˚-wide bins. For comparison, a power-law spec-
theequatorialelectronscatteringregion,butdidnotdiscuss
trumwithαν =+1/3isalsoplottedinthebottom panel. howthechangesingeometryresultedinthevariabilityofthe
polarimetric properties around the Balmer continuum and
˚A is ∆PA = 4.6 deg, and the PA variability is nearly emission lines. Apparently,if the broad absorption features
c
wavelength-independent(thebottompanelofFigure1).Be- in the polarized flux spectrum are assumed to be intrinsic
causethesystematicuncertaintiesinthegridofpolarization to the disc thermal emission, it is very difficult to explain
standards are at most ∼ 1 deg (e.g., Schmidtet al. 1992), reasonablythetime-variabilityoftheabsorptionfeaturesby
the difference in the observed PA between the Bok/SPOL the changes in the geometry of the scattering region alone.
and Keck/LRIS measurements cannot be attributed to er- Instead, as discussed in detail in Section 4.2.3, we propose
rors of the PA calibration by the use of different polarized that the absorption features are imprinted by an absorbing
standard stars, and therefore in this work the observed PA region with a time-variable structure, which is assumed to
variability is assumed to be the intrinsic variability of the bespatiallyseparatedfromtheUV-opticalemittingregions
polarization properties of 3C 323.1. of theaccretion disc.
Although PA variability is a general property of the
blazar-like synchrotron emission (e.g., Ikejiri et al. 2011),
Section 4.1 shows that synchrotron emission is not respon-
siblefortheopticalpolarization observedin3C323.1. Ifwe
In the next section (Section 4), we focus on the V-
assume that the optical polarization in 3C 323.1 is due to
band polarimetric and photometric variability. As can be
thescatteringoftheaccretiondisccontinuum,theobserved
clearly seen in Figure 3, the polarized flux variability in
PA variability requires some time-variable axi-asymmetric the wavelength range of the V-band [λ ∼ 5000−5900
obs
structure in the scattering region of 3C 323.1, as discussed ˚A at full width at half maximum (FWHM)] is clearly de-
in detail in Sections4.2.2 and 4.2.3.
tected with a high signal-to-noise ratio. It should be noted
that the V-band is sampling the rest-frame wavelengths of
3.2 Spectral variability of the polarized flux λrest ∼ 4000−4700 ˚A, within which the observed flux is
dominated by the continuum emission. Therefore, the V-
In Figure 3, it is clear that the broad Balmer emission band polarimetric variability mostly reflects the variability
lines seen in the total flux spectra do not appear in either ofthepolarizedcontinuumcomponent.Thismeansthatthe
theBok/SPOL ortheKeck/LRISpolarized fluxspectra; in final results of this work discussed in Section 4 are not af-
otherwords,theBLRemissionisunpolarizedin3C323.1.As fectedbytheputativechangestothepolarizationproperties
discussedbyKishimoto et al.(2004),theKeck/LRISpolar- around the wavelength range of the Balmer continuum and
izedfluxspectrum(pˆ ×F )showsaclearbroadabsorption- otherrecombination lines discussed above.
λ λ
MNRAS000,1–15(2017)
Photometric and polarimetric variability of 3C 323.1 9
4 V-BAND POLARIMETRIC AND
PHOTOMETRIC VARIABILITY AND ITS V, Polarization Degree
2.4
INTERPRETATION
Figure 4 shows the photometric and polarimetric measure- 2.2
mentsfor 3C 323.1 evaluated in theV-bandsummarised in ent]
Table 1 as a function of time. As shown in the top panel er c 2
p
of Figure 4, the polarization degree pˆdoes not show strong ^p [
variability during the three observations taken during the 1.8
period 1996-2003. The total flux F of 3C 323.1 shows ∼
ν
0.3 mag variability during the same period, and the small 1.6
variabilityinpˆresultsinthehighlycorrelatedvariabilitybe-
0 500 1000 1500 2000 2500 3000
tweentheV-bandtotalfluxandthepolarized flux(pˆ×F )
ν MJD-50000
(thebottom panel of Figure 4).
25
The middle panel of Figure 4 shows the V-band V, Position Angle
PA, compared with 3C 323.1’s radio jet axis of 20 deg Radio Axis
20
(Kishimoto et al. 2004). As has already been noted in Sec-
tion 3.1, the observed PA differs between the Bok/SPOL
and Keck/LRIS measurements by ∼ 4 deg, although the gree] 15
difference between the PA and the radio axis is kept small de
(i.e., 3C 323.1 remains to be“parallel”polarization). A [ 10
P
Belowwediscussthegeometricalconstraintsontheop-
tical polarization source in 3C 323.1 derived from the po- 5
larimetric andphotometricvariability seenin Figure4,and
thepossibleinterpretationsoftheopticalpolarizationmech- 0
0 500 1000 1500 2000 2500 3000
anism.
MJD-50000
0.06 2.5
V, Polarized Flux
V, Total Flux
4.1 Evidence against the synchrotron origin of the y]
J
optical polarization in 3C 323.1 F [mn 0.05 2 mJy]
AshasalreadybeenmentionedinSection3,ontheonehand, ^p x F [n
tahbesoKrpetciko/nLfReaItSurpeosl,abriuztedonfluthxesoptehcetrruhmansdh,otwhsecBleoakr/SbProOaLd Flux Flux
ptroulmar.izeBdasfleudxsopnectthruemBioska/lSmPoOstLadsmatoao,thScphomwiedrt-l&awSsmpietch- olarized 0.04 1.5 Total
(2000) have suggested that the weak flux contribution of P
the synchrotron emission to the optical wavelength range 0.03 1
can explain the observed optical polarization in 3C 323.1 0 500 1000 1500 2000 2500 3000
MJD-50000
(see Section 1 for details). Even for the Keck/LRIS data,
the synchrotron contribution scenario for the observed po-
Figure 4. Top and middle: light curves of polarization degree
larized flux cannot be ruled out with only the single epoch pˆ(toppanel)andpolarizationpositionanglePA(middlepanel)
data;itispossiblethattheabsorptionfeaturesareimprinted of 3C 323.1 given inTable 1. 3C 323.1’s radiojet axis of 20 deg
in the intrinsically smooth synchrotron emission spectrum (Kishimotoetal. 2004) is indicated as a dotted line in the mid-
somewhere along theline of sight (Kishimoto et al. 2003). dlepanel. Bottom: the lightcurves of the V-bandpolarized flux
However, we point out here that the observed polari- pˆ×Fν (solid line) and total flux Fν (dotted line) of 3C 323.1;
metric and photometric variability strongly suggests that Bok/SPOLin1996-1998, ESO3.6-m/EFOSC2on1stMay,2002,
andKeck/LRISon4thMay,2003.
theopticalsynchrotronemission isnottheorigin ofthepo-
larized flux component in 3C 323.1. As shown in Figure 4,
the three V-band measurements reveal that both the total
andthepolarizedfluxesvaryalmostsimultaneously;inother for the case of Ton 202, we can also confirm the diffi-
words, the polarization degree is nearly time-constant (see culty in attributing the observed polarized fluxin 3C 323.1
also Figure 1). Because thevariable component of thetotal solely to the flux contribution from the optical synchrotron
flux must be dominated by the disc continuum emission as emission in terms of its core-only radio-loudness. In gen-
discussedinSection2.1,theobservedhighlycorrelatedvari- eral, the radio core emission of lobe-dominated quasars be-
abilitybetweenthetotalandpolarized fluxesindicatesthat comesopticallythinatν >30GHz(Antonucciet al.1990),
the polarized flux has a strong relationship with the accre- and, thus, the ratio of the radio core flux to the opti-
tion disc emission. Therefore, theobservation of the polari- cal synchrotron flux (measured at 5000 ˚A) is expected to
metric and photometric variability essentially excludes the be Flux(radio core)/Flux(opt. sync.)∼20,000 by assuming
possibility that the optical polarization source in 3C 323.1 f ∝ν−1 (Kishimoto et al. 2003). On the other hand,from
ν
is theoptical synchrotron emission. 3C 323.1’s radio-loudness for the core region Rcore = 41
In addition to the above mentioned evidence, with ref- (Schmidt& Smith 2000) and theobserved optical polariza-
erencetothediscussion providedinKishimoto et al.(2003) tion degree p ∼ 2%, the flux ratio of the radio core flux
MNRAS000,1–15(2017)
10 M. Kokubo
to the optical polarized flux is Rcore/p ∼ 41/0.02 =2050. be interpreted as the change in geometry of the scattering
Therefore, if we assume that the optical polarized flux in region. Kishimoto et al. (2004) pointed out that the puta-
3C323.1originatesfromthesynchrotronemission,theratio tive equatorial electron scattering region is assumed to be
oftheradiocorefluxtotheopticalsynchrotronfluxmustbe so small in size that its geometrical configuration can be
Flux(radio core)/Flux(opt. sync.)∼2050×p <2050, where changed within a one-year time-scale. If the putative equa-
s
p (<1) is the fractional polarization degree of the puta- torial electron scattering region in 3C 323.1 is not an ax-
s
tive optical synchrotron radiation. The discrepancy in the isymmetric disc-like structure but has an axi-asymmetric
expected (∼20,000) and required (∼2050×p ) ratio of the clumpy density distribution, the PA variability (with lit-
s
radio core flux to the optical synchrotron flux implies that tle variability in the polarization degree) may be a natural
theamountoffluxcontributionfromtheopticalsynchrotron consequence of the orbital or bulk motion of the scattering
in3C323.1 istoosmall(byatleast an orderofmagnitude) region (see Section 4.2.3 for details). According to this in-
to account for theobserved polarized flux. terpretation,thePAvariabilityobservedin3C323.1during
Considering the lines of evidence discussed above, we theBok/SPOLandKeck/LRISobservations(thetwoobser-
conclude that the optical synchrotron emission is not rele- vationsare5-7yearsapart,i.e.aquasarrestframetime-lag
vantfortheobservedopticalpolarizationin3C323.1.InSec- of < 6 years) implies that the maximum extent of the size
tion4.2,weshowthattheobservedpolarimetricandphoto- of thescattering region Rsca is
meleecttrriconvasrciaatbtielirtiyngprfroopmerttiheeseoqfu3aCto3r2ia3l.1scaartetecroinnsgisrteegnitonw,itahs Rsca(PAvar.) < 6× vsca [light years]
(cid:16) c (cid:17)
proposed by Kishimoto et al. (2004) (see Section 1). How-
ever,itisapparentlydifficulttoaccountfortheobservedPA = 0.20× vsca [light years(]3)
(cid:18)10000km/s(cid:19)
variabilityseeninthemiddlepanelofFigure4bythesimple
equatorialscatteringscenario.Weattempttoconstructage- wherevsca representsthetypicalvelocityofthepolarization
ometrical model to explain all of the observed polarimetric source.Ifweassumethatthepolarizationsourceissmallerin
and photometric propertiesof 3C 323.1 in Section 4.2. sizethantheBLR,itisreasonabletoconsiderthatthevalue
of vsca is larger than the velocity width of the broad emis-
sion lines.SincetheHβ broademission lineof3C323.1 has
4.2 Constraints on the geometry of the scattering an FWHM of ∼ 7030 km/s (e.g., Boroson & Green 1992),
region in 3C 323.1 wetakevsca =10,000km/stobeareferencevalueinEqua-
tion 3.
In this section, we first obtain observational constraints on
the radial extent of the scattering region inferred from the
polarimetric and photometric variability via causal reason- 4.2.2 Comparisons of the radial extent between the
ing,andcomparethemwiththesizeoftheBLRandthedust polarization source, BLR, and the dust torus
torus of 3C 323.1. We then propose a geometrical model of
the scattering/absorbing regions, which can potentially ac- The geometrical constraints on the scattering region Rsca
count for all of the observed polarimetric and photometric derivedinSection4.2.1 shouldbecomparedwith theradial
properties of 3C 323.1. extentof theaccretion disc, BLR, and thedust torus.
UndertheassumptionoftheShakura& Sunyaev(1973)
accretion disc model, the disc radius at which the disc
4.2.1 Radial extent of the scattering region inferred from temperature matches the wavelength λrest as kBTλrest =
the polarimetric and photometric variability hc/λrest (h and kB are the Planck constant and the Boltz-
mann constant, respectively) can be evaluated as (e.g.,
A robust constraint on the geometry of the scattering re-
Morgan et al. 2010)
gion can be derived from the highly correlated variability
of the V-band total flux and the polarized flux. As has al- Rdisc,λrest ≃ 0.01 [light years]
rmeaaddye bbyeeEnSpOoi3n.6te-md/oEutFOinSSCe2ctaionnd4K.1e,ctkh/eLRtwIoSoobnse1rsvtaMtioanys, × λrest 4/3 MBH 2/3 L 1/3,(4)
(cid:18) µm (cid:19) (cid:18)109M⊙(cid:19) (cid:18)ηLE(cid:19)
2002, and 4th May, 2003, respectively (separated by one
year, i.e., by 0.8 years in the quasar rest frame), show co- where η≡L/(M˙c2) indicates the radiative efficiency of the
ordinatedvariabilitybetweenthetotalandpolarizedfluxes. disc. The theoretical values of the radiative efficiency, in-
With the assumption that the polarization in 3C 323.1 is cluding general relativistic corrections, range from 6% to
due to scattering of the disc continuum, the time lag be- 42% as a monotonically increasing function of the black
tweenthetotalfluxvariation (disccontinuumflux)andthe holespin(e.g.,Shapiro& Teukolsky1983;Frank et al.1992;
polarized flux variation (scattered flux) corresponds to the Katoet al. 2008). By substituting log(MBH/M⊙) = 9.07
light travel time across the accretion disc and the scatter- and L/L = 0.10 taken from Shen et al. (2011) (see Sec-
E
ingregion(e.g.,Gaskell et al.2012).Therefore,theobserved tion 2.1), the disc radius of λrest =5100 ˚A in 3C 323.1 can
coordinated variability of the total and polarized fluxes di- becalculated to bewithin therange
rectlyconstrainstheradialdistanceofthescatteringregion
as Rdisc,5100A˚=0.003−0.005 [light years]. (5)
RecentobservationsofquasarmicrolensingeventsandAGN
Rsca(pF var.) < 0.8 [light years]. (2)
continuumreverberationmapping(e.g.,Morgan et al.2010;
On the other hand, within the framework of the equa- Edelson et al. 2015) haverevealed thattheactual accretion
torial scattering scenario, the PA variability may possibly discsizesinquasarsarelargerbyafactorof∼4thanthesize
MNRAS000,1–15(2017)