SDSS J094533.99+100950.1 - the remarkable weak emission line quasar

0.56 MB·English

by  Krzysztof Hryniewicz

Mon.Not.R.Astron.Soc.000,000–000 (0000) Printed26January2010 (MNLATEXstylefilev2.2)
SDSS J094533.99+100950.1 - the remarkable weak
emission line quasar
0
1 K. Hryniewicz1,2, B. Czerny1, M. Niko lajuk2 and J. Kuraszkiewicz3
0 1Nicolaus Copernicus Astronomical Center, Bartycka 18, 00-716 Warsaw, Poland
2 2Faculty of Physics, Universityof Bia lystok, Lipowa 41, 15-424 Bia lystok, Poland
n 3Harvard-Smithsonian Centerfor Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA
a
J

6 26January2010
2

] ABSTRACT
A
G Weak emission line quasars are a rare and puzzling group of objects. In this
. paper we present one more object of this class found in the Sloan Digital Sky Survey
h
(SDSS). The quasar SDSS J094533.99+100950.1, lying at z = 1.66, has practically
p
no CIV emission line, a red continuum very similar to the second steepest of the
-
o quasar composite spectra of Richards et al., is not strongly affected by absorption
r and the MgII line, although relatively weak, is strong enough to measure the black
t
s hole mass. The Eddington ratio in this object is about 0.45, and the line properties
a are not consistent with the trends expected at high accretion rates. We propose that
[
the most probableexplanationof the line properties in this object, andperhaps in all

2 weak emission line quasars, is that the quasar activity has just started. A disk wind
v is freshly launched so the low ionizationlines which form close to the disk surface are
7 already observed but the wind has not yet reached the regions where high ionization
1 lines or narrow line components are formed. The relatively high occurrence of such a
5 phenomenonmayadditionallyindicatethatthequasaractivephaseconsistsofseveral
3
sub-phases, each starting with a fresh build-up of the Broad Line Region.
.
4
Key words: galaxies:active - galaxies:quasars:emission lines - galax-
0
9 ies:quasars:absorptionlines - galaxies:quasars:individual:SDSSJ094533.99+100950.1
0
:
v
i
X
1 INTRODUCTION since there are no signs of the deep and broad absorption
r troughs in the spectra (Anderson et al. 2001, Collinge et
a
Strong,broademission linesarethecharacteristicsignature al. 2005) and the X-ray absorbing column is low compared
of Active Galactic Nuclei (AGN). However, the Sloan Digi- withBALQSOs(NH <5×1022cm−2,Shemmeretal.2009).
talSkySurvey(SDSS)finds,amongthousandsofnewAGN, 2) Gravitational lensing does not remove the emission lines
rare objects with veryweak oralmost absent emission lines (Shemmeret al. 2006), and strong CIII] and CIV lines are
(Fan et al. 1999, Shemmer et al. 2009, Diamond-Stanic et found in gravitationally amplified quasars (e.g. Dobrzycki
al.2009,Plotkinetal.2010).Mostoftheseobjects,forming et al. 1999, Waythet al. 2005). 3) Relativistic beaming is a
a new class of weak line quasars (hereafter referred to as good explanation for the absence of strong lines in BL Lac
WLQs), have been found primary at high redshifts (z>2). objectsbutWLQs,in contrast toBL Lacs,haveblueropti-
∼
A few sources lyingrelatively close byhavebeen studied in cal/UVcontinua,areradioquietandshownovariability or
greaterdetail:PHL1811(z=0.19;Leighlyetal.2007),PG strong polarization (Shemmer et al. 2009, Diamond-Stanic
1407+265 (z = 0.94; McDowell et al. 1995), and HE 0141- et al. 2009). 4) The difference in the ionizing continuum
3932 (z=1.80; Reimers et al. 2005). The study of WLQs also faces difficulties since at least the observed part of the
optical/UV continuum properties, X-ray and radio emis- continuuminthosesourceslookstypicalforaquasar.How-
sion shows no obvious difference from typical emission line ever, those objects may still emit more vigorously in the
quasarsandnoapparentsimilarity totheweak-lineBLLac unobserved UV, thus having a higher accretion rate than
objects (Shemmeret al. 2009). theaveragequasars, whichmayaffect thelines, ase.g. seen
through the Baldwin effect. No black hole mass determina-
No generally accepted explanation for the weak-
tion or Eddington ratio was possible for the sample of ob-
ness/absence of emission lines has been found so far. Possi-
jectsanalyzed byShemmeret al. (2009) soastoconfirm or
bilities sofar considered include: 1) dustobscuration or ex-
rejectthishypothesis.InthispaperwepresentanewWLQ
treme broad absorption line (BAL) effect, which is unlikely
2 Hryniewicz et al.
object found in the SDSS survey which opens a possibility
Table 1.Continuum andironfittingwindows
to address this issue.
Model Fittingwindows
Name restframewavelength rangein˚A
2 SELECTION PROCEDURES AND
Cont1 1455-1470, 1690-1700, 2160-2180, 2225-2250,
OBSERVED PROPERTIES OF THE
3010-3040, 3240-3270
QUASAR SDSS J094533.99+100950.1
Fe12 2020-2120, 2250-2340, 2440-2650, 2900-3000
SDSS J094533.99+100950.1 was serendipitously discovered
by us while searching the full SDSS DR5 (Adelman- Fe23 2020-2120, 2250-2300, 2500-2650, 2850-3000
McCarthy et al. 2007) spectral database with the aim of
Fe34 1490-1502, 1705-1730, 1760-1800, 2250-2320
finding interesting quasars for which atypical line and/or
2470-2625, 2675-2715, 2735-2755, 2855-3010
continuumpropertieswould prohibitthembeingfoundand
classifiedasquasarsbytheSDSSautomaticpipeline.Inpar- ContFe45 1455-1470, 1685-1700, 2020-2120, 2190-2210
ticular, we looked for quasars with extreme blue continua 2250-2300, 2450-2650, 2850-3070
(thesewouldhavebeenmissedduetocolourcutsusedbythe ContFe56 1440-1470, 1700-1820, 1950-2400, 2450-2700
SDSS to avoid regions of colour space dominated by white 2850-3100
dwarfs,A-starsandMstar+whitedwarfpairs-seeRichards
ContFe67 1455-1470, 1690-1700, 2020-2120, 2160-2180
et al. 2002) and quasars with normal (unreddenedby dust)
2225-2650, 2900-3000, 3010-3040, 3240-3270
continuabutwithnon-standardemissionlineproperties.We
3350-3400
analyzed all, 1048960, spectra available in the SDSS DR5
spectral database, classified by the SDSS automatic proce-
dures into: galaxies, quasars, stars, sky, and unclassifiable Column(1)-nameofmodel;column(2)-wavelengthrangeof
objects (see Adelman-McCarthy et al. 2007). spectralwindowsinwhichthemodelwasfit.
1 -continuum fittedasapowerlawinForsteretal.(2001)
The spectrum of SDSS J094533.99+100950.1 appeared
continuum windows;
as particularly interesting, showing a normal quasar con-
2 -irontemplatefittedinForsteretal.(2001)ironspectral
tinuum but with atypical, weak emission lines, with practi-
windows,excludingwavelengthrange2340-2440˚A (atmosphere
callynonexistentCIVandCIII],andweak,althoughclearly artifact+featureat2350˚A);
visible, MgII. The object was identified as a quasar by 3 -irontemplatefittedinForsteretal.(2001)ironwindows,
the SDSS pipeline due to the presence of the MgII line, withregionsaroundCII]λ2327andOIIλ2441excluded, and
and is hence also included in the Schneider et al. (2007) 2850-2900˚A window(atredsideofMgII)addedforamore
SDSS quasar catalogue. The spatial coordinates are α = consistentfit;
146.39163, δ = 10.163927 and the determined redshift is 4 -irontemplatefittedinspectralwindowsdefinedby
z=1.66160±0.00025. Thequasar’sspectrumwasobserved Vestergaard&Wilkes(2001), withregionsofCII]emissionand
narrowabsorptionatCIV(1427-1505˚A) andMgII
on April25, 2003 and classified as “excellent” bythe SDSS
(2715-2735˚A) excluded;
database.
5 -powerlawcontinuum andironemissionfittedsimultaneously
inthe“Cont”and“Fe1”windows(somewindowsarenarrower
orslightlyshiftedtobettermatchthemodelwiththespectrum);
2.1 Emission line measurements - modelling of
6 -powerlawcontinuum andironemissionfittedsimultaneously
the spectrum
inoptimumwindowschosenbyusandcreatedbycomparisonof
In order to study the weak emission line properties of our theRichardsetal.(2003) compositesandVestergaard&Wilkes
(2001)irontemplate;
WLQwemodeledthespectrumbyfittingapowerlawtothe
7 -powerlawcontinuum andironemissionfittedsimultaneously
underlyingcontinuum,accountingforblendedironemission
inthe“Cont”modelwindowscomplemented witha3350-3400˚A
(theiron template of Vestergaard & Wilkes 2001 was used)
windowandironwindowsfromForsteretal.(2001) .
and fitting single Gaussians to theemission and absorption
lines. Atmospheric emission lines that still remained after
the automatic pipeline reduction were first removed, and
thespectrumwascorrectedforGalacticreddeningusingthe iron windowswhich excludedspectral regions ofCII]λ2327
Cardelli et al. (1989) extinction curve, adopting a colour and OIIλ2441 emission and/or CIV and MgII NAL ab-
excess of E(B−V) = 0.06 determined from the Galactic sorption (models “Fe2” and “Fe3” in Table1). We present
neutral hydrogen column from Dickey & Lockman (1990) theresults ofourcontinuumand iron modelling in Table 2.
and Stark et al. (1992), and assuming a fixed conversion of In model no. 1 (Cont-Fe1-best in Table 2) we fol-
N(HI)/E(B −V) = 5.0×1021cm−2 mag−1 (Burstein & lowed the procedure described in Vestergaard & Wilkes
Heiles 1978). (2001), where first the underlying power law continuum
The underlying power law continuum was fitted to re- was fitted and subtracted from the spectrum, followed by
gions of spectrum uncontaminated by emission lines and the modelling of iron emission. The Vestergaard & Wilkes
away from blended iron emission. The continuum windows (2001)irontemplatespectrumwasbroadenedbyconvolving
used are shown in Table1 (model “Cont”) and taken from with Gaussian functions of widths between 900 kms−1 and
Forsteretal.(2001).Blendedironemissionwasmodeledus- 9000 kms−1 and separated by steps of 250 kms−1, while
ing the UV iron template of Vestergaard & Wilkes (2001) conserving the total flux in each template. We chose the
tospectral windows definedin Forsteret al. (2001) (Table1 broadening value for which χ2 was lowest. Finally, we sub-
model “Fe1”). Wealso experimented with slightly modified tracted the best iron fit from the spectrum and again fit-
WLQ SDSS J094533.99+100950.1 3
CIV
SDSS J094533.99+100950.1
composite no 3 of Richards et al.
binned SDSS J094533
continuum of SDSS J094533
iron emission of SDSS J094533
4
]
1
A(cid:4) 3
1(cid:5)
s(cid:4) SiIII]CIII]
g
r
[e HeII
430
1
(cid:3)
L(cid:2)
2
FeIII MgII
FeII
1
1500 2000 2500 3000
[A(cid:1)]
(cid:0)
Figure1.TherestframespectrumofSDSSJ094533.99+100950.1correctedforGalacticreddening(yellowline),thebinnedspectrumof
thequasar(redline),itsfitted, underlyingcontinuum (blueline)andironemission(greenline).Forcomparison,inblack,theRichards
etal.(2003) compositespectrum(no.3)isshown,greyshiftedtomatchtheSDSSJ094533.99+100950.1 spectrum.
ted the continuum. This procedure was repeated two more (fλ ∝ ναλ) equal to −1.232 ± 0.017, which is equiva-
times. Final parameters are presented in Table 2. lent to αν = −0.77 (fν ∝ ναν). Therefore, the SDSS
In models no. 2-4 (Cont-Fe1,2,3) a similar procedure J094533.99+100950.1 continuum is, in this respect, almost
was used for the continuum fitting, however broadening of identicaltotheno.3quasarcomposite from Richardsetal.
the iron template and the scaling factor of the amplitude (2003).However,wealsocomparedquantitativelytheunder-
was estimated automatically. We experimented with stan- lying continua of SDSS J094533.99+100950.1 and Richards
dard and slightly modified iron windows (models Fe1, Fe2, et al. (2003) composites no. 1-4 by fitting the latter with a
Fe3 from Table1). power law (to the same spectral windows as for our WLQ
In models no. 5-7 (i.e. ContFe4,5,6) power law contin- - the “Cont” model from Table 1) and allowing for inter-
uumandirontemplatewerefittedsimultaneously(asasum) nal reddening. We checked both Gaskell et al. (2004) and
to spectral windows including standard and slightly modi- SMC-bar (Sofiaet al. 2006) extinction curves. Wemodified
fied continuumand iron windows from Forsteret al. (2001) the composites and allowed for either reddening or dered-
(models “ContFe4”, ”ContFe5” and “ContFe6” inTable 1). dening. The lowest χ2 was obtained for composite 4 dered-
dened by E(B −V) = 0.75 for Gaskell et al. (2004) ex-
Thebestcontinuumandironemissionfit,isrepresented
bymodelno.1(Cont-Fe1-best).Ithasthelowestreducedχ2 tinction, which is relatively high due to its weak depen-
dence on wavelength, which thus requires large reddening
value and visually gives the best fit to iron emission in the
toproduceanysignificantchangeinspectralshape.Forthe
vicinityoftheMgIIline,whichthenleadstothebestMgII
emission line fit (as shown by the lowest χ2 and weakest case of SMC-bar extinction curve a lower value is needed,
E(B−V)=0.02. Forcomposite 3of Richardset al. (2003)
residuals). In Fig. 1 we plot this continuum and iron emis-
no reddening/dereddening was required an acceptable fit.
sion fit over the restframe, Galaxy dereddened spectrum of
For the other two composites reddening was required but
SDSSJ094533.99+100950.1.ThefitisalsoincludedinFig.2
thefitwasalwaysworsethanforpure,unreddenedcompos-
and the emission line and central engine properties derived
ite3.Sincetheexactwavelengthdependenceoftheintrinsic
furtherin thetext are madeusing this model.
reddeninginquasarsisstillunderdebate,wedecidedtouse
The best fit continuum has a power law index, αλ
4 Hryniewicz et al.
composite3forcomparison.InFig.1weshowtheRichards 1.6
et al. (2003) composite no. 3. plotted over the rest frame
SDSS J094533.99+100950.1 spectrum, corrected for Galac- 1.4 CIV
tic extinction.
Despite the almost identical continua, the two spectra 1.2
arewidelydifferentwithrespecttotheemissionlineproper-
ties.TheMgIIemissionlineisclearlypresentbutthenarrow
1.0
component in the line profile is absent. The contribution of
theFeIIandFeIIIisalmost typical,whiletheSiIII]+CIII]
0.8
and CIV emission is extremely weak compared with a typ-
ical quasar. There are also two deep but narrow absorption
0.6
components (clearly resolved into doublets in the unbinned
dataplot),blueshiftedby∼8300kms−1withrespecttothe
0.4
CIVandMgIIemissionlinepositionexpectedfromWLQ’s
adopted redshift.
0.2
The emission and absorption lines were fitted using
1.6
Gaussiansandthefitswereperformedforallcontinuumand
ironmodelsdefinedinTable2.Thedifferentmodelshadno
SiIII]+CIII]
1.4
significanteffectontheabsorptionlinefits(resultinginless
than11% differenceinequivalentwidths)andso inTable3
we give thevalues for one model only Cont-Fe1-best. x1.2
u
The MgII emission line was more influenced by the fl
choice of continuum and iron models, and the equivalent d 1.0
e
width varied from fit to fit by a few percent, with a 15% z
differencebetweentheCont-Fe1-bestandtheextremevalue ali0.8
(see Table 4). The measured kinematic width depended on m
thespectraldecomposition,andvariedfrom5390kms−1up or0.6
to 6290 km s−1, dependingon themodel used.However, in N
all studied cases the line was relatively strong and broad. 0.4
Therefore, we consider that the properties of the MgII line
are determined reliably which is important from the point 0.2
of view of black hole mass determination. The CIV line is 1.6
consistently faint in all models, although some fits indicate
the presence of a very weak line while others give values MgII
1.4
consistent with zero intensity.
Forcomparison,inTable5wegivethetypicalemission
1.2
line parameters measured for QSOs. Single component fits
andtwo(narrowandbroad)componentfitsaregivenwhen-
1.0
everpossible.WeseethattheMgIIlineintensityisafactor
of ∼ 2 lower than the mean MgII line intensity of sources
0.8
foundintheLargeBrightQuasarSurvey(LBQS,Forsteret
al. 2001) or observed with the Faint Object Spectrograph
0.6
on theHubbleSpaceTelescopes (FOS,Kuraszkiewicz etal.
2002) as well as the value found for the single component
0.4
MgIIfitstotheRichardsetal.(2003)steepcomposite.How-
ever,CIVline(aswellasCIII])isover20timesfainter.The
weaknessofCIV-typicallyoneofthestrongestquasarlines 0.2
- qualifies the object as a WLQ. However, the presence of
-10000 -5000 0 5000 10000
MgII allows for an estimate of the global physical parame- 1
v [km s(cid:6) ]
ters of our source.
Wealsotestedthepossibilitythattheredshiftdetermi-
nation based on theMgII lineisinaccurate and weallowed Figure 2.TheregionsoftheCIV,SiIII]+CIII] andMgIIlines
thelinecentroidandtheposition oftheiron linecomplexes inthespectrumofSDSSJ094533.99+100950.1(blue).Greenline
toshiftarbitrarily.Thisdidnotimprovethefitsignificantly showsthebroadened(FWHM(FeII)=1650kms−1)ironemission
template,redandcyanshowfittedemissionandabsorptionlines
although a possibility of such a shift by a few hundred km
s−1cannot berejected. respectively, while the black line is sum of all components. Zero
velocitywasatthetheoreticalwavelength ofeachlineatagiven
Since WLQs have been occasionally suggested to be a
redshift.Theplottedfluxwasnormalizedbythecontinuum.
BAL (McDowell et al. 1995), we also calculated the BI in-
dex(Weymannet al.1991) in thevicinityof theMgII line.
Weobtainedavalueconsistentwithzero.AlsotheAIindex
(Hall et al. 2002) for this quasar is zero if we take the un-
binned spectrum where NAL doublets are clearly resolved
WLQ SDSS J094533.99+100950.1 5
Table 2.Propertiesoffittedcontinuum andironemissiontemplatefordifferentfittingwindows.
No continuum FeIItemplate χ2/ndf. fitting
αλ F3000˚A FWHM scalefactor REW windows
[ergs−1 cm−2 ˚A−1] [kms−1] [˚A]
1. Cont-Fe1-best
Cont: −1.232±0.017 (5.24±0.69)·10−16 — — — 1.51 Cont
Fe: — — 1650±130 0.01176±0.00027 82 1.30 Fe1
2. Cont-Fe1
Cont: −1.233±0.017 (5.24±0.69)·10−16 — — — 1.56 Cont
Fe: — — 2150±380 0.01139±0.00027 79 1.30 Fe1
3. Cont-Fe2
Cont: −1.233±0.017 (5.24±0.69)·10−16 — — — 1.58 Cont
Fe: — — 2680±420 0.01208±0.00027 84 1.30 Fe2
4. Cont-Fe3
Cont: −1.240±0.017 (5.24±0.69)·10−16 — — — 1.58 Cont
Fe: — — 3440±550 0.01070±0.00027 75 1.57 Fe3
5. ContFe4
Cont+Fe: −1.194±0.011 (5.40±0.43)·10−16 1140±190 0.00849±0.00043 58 1.55 ContFe4
6. ContFe5
Cont+Fe: −1.243±0.010 (5.22±0.40)·10−16 2500±300 0.01195±0.00053 85 1.71 ContFe5
7. ContFe6
Cont+Fe: −1.215±0.012 (5.35±0.43)·10−16 1179±190 0.01083±0.00043 74 1.69 ContFe6
Column(1)-modelnumber;column(2)-modelname:fitsno.1–4havecontinuum modeledasapowerlawinspectralwindows
definedbymodel“Cont”inTable1,independentlyfromtheironemissionfittedinironspectralwindowsdefinedbymodelsFe1,2,3in
Table1;fitsno.5–7havecontinuum andironemissionfitsimultaneouslytowindowsdefinedbymodelsContFe4,5,6inTable1.
Column(3)liststhepowerlawindex,αλ (wherefλ∝λαλ).Column(4)showstheamplitudeofthecontinuum measuredat3000˚A.
Columns(5)-(7) listrestframefullwidthathalfmaximum(FWHM)oftheFeIItemplate,scalingfactor,andFeIIrestframe
equivalentwidth(REW),respectively. Thereducedχ2 ofthefittingprocedureisshowninColumn(8).Column(9)liststhefitting
windowsusedbyusanddefined inTable1.
Table 3.RestframeabsorptionlinespropertiesofSDSSJ094533.99+100950.1
Line λem λ0 REW FWHM Shift zabs
[˚A] [˚A] [˚A] [kms−1] [kms−1]
fit: Cont-Fe1-best
MgII 2796.357 2721.30±0.51 −3.79+−00..3341 343±82 −8268±57 1.59016
2803.536 2728.26±0.58 290±110 −8272±64 1.59013
CIV 1548.189 1506.39±0.22 −3.35+−00..4427 455±63 −8318±44 1.58975
1550.775 1509.24±0.24 269±69 −8251±49 1.59031
Column(1)-nameofabsorptionline,column(2)-thewavelength ofthelineseeninlaboratory.Theabsorptionlineswerefittedwith
Gaussianstoadereddened, restframe,FeIIsubtracted quasarspectrum.Column(3)-wavelengthoftheminimumintensityoftheline
inthefit.Column(4)-RestframeEquivalentWidth(REW)ofthedoublet(computed asasumoftwoGaussians).Column(5)Full
WidthatHalfMaximum(FWHM)oftheabsorptionline.Column(6)liststhelineblueshiftinkms−1.Wepresentmeasurements for
theCont-Fe1-best modelfromTable2only,asothermodelsgavesimilarresults.
- Fig. 2. There is also no sign of a CIV BAL, where the archivesandsoftwareoftheVirtualObservatory(VO)1and
observed spectrum allows detection of such a component the Galaxy Evolution Explorer (GALEX) 2 database. This
blueshifted upto∼20000km s−1.Weconcludethat SDSS includedopticalu,g,r,iandzphotometryfromSDSS,and
J094533.99+100950.1 is not a BAL or a mini-BAL object. near-IR J, H and K photometry from the Two Micron All
s
Sky Survey (2MASS) and near- (NUV) and far-ultraviolet
2.2 Spectral energy distribution
We compiled the broad band spectral energy distribu- 1 http://www.euro-vo.org/
tion (SED) of SDSS J094533.99+100950.1 using the data 2 http://galex.stsci.edu/
6 Hryniewicz et al.
Table 4.RestframeemissionlinespropertiesofSDSSJ094533.99+100950.1.
Line λ0 REW FWHM zem
[˚A] [˚A] [kms−1]
fit: Cont-Fe1-best
MgIIλ2800 2800.29±0.79 15.3+−11..11 5490±220 1.66187
CIII]λ1909+SiIII]λ1892 1906.7±2.3 1.68+−00..6512 4800±4800 1.65845
CIVλ1549 1544.9±2.3 1.49+−00..8624 4000±1100 1.65451
fit:Cont-Fe1
MgIIλ2800 2800.60±0.81 15.5+−11..11 5690±230 1.66217
CIII]λ1909+SiIII]λ1892 1905.6±2.4 1.69+−00..6533 5000±5000 1.65680
CIVλ1549 1545.6±2.2 1.41+−00..7692 3800±1100 1.65572
fit:Cont-Fe2
MgIIλ2800 2800.85±0.83 15.2+−11..11 5750±240 1.66240
CIII]λ1909+SiIII]λ1892 1904.6±2.5 1.69+−00..6544 5100±5100 1.65539
CIVλ1549 1545.9±2.3 1.36+−00..7680 3800±1100 1.65632
fit:Cont-Fe3
MgIIλ2800 2801.64±0.92 15.9+−11..22 6290±270 1.66316
CIII]λ1909+SiIII]λ1892 1903.9±2.7 1.64+−00..6565 5200±5200 1.65455
CIVλ1549 1545.7±2.5 1.18+−00..7578 3600±1200 1.65597
fit:ContFe4
MgIIλ2800 2799.28±0.91 14.0+−11..21 5670±250 1.66091
CIII]λ1909+SiIII]λ1892 1905.6±2.5 1.37+−00..5498 4400±4400 1.65680
CIVλ1549 1543.7±2.6 1.59+−00..9711 4300±1300 1.65252
fit:ContFe5
MgIIλ2800 2804.77±0.76 16.0+−11..11 5600±220 1.66614
CIII]λ1909+SiIII]λ1892 1902.5±2.5 1.91+−00..6588 5400±5400 1.65252
CIVλ1549 1547.4±2.4 1.11+−00..7535 3400±1100 1.65888
fit:ContFe6
MgIIλ2800 2798.84±0.87 14.1+−11..21 5390±240 1.66050
CIII]λ1909+SiIII]λ1892 1906.0±2.4 1.33+−00..5487 4300±4300 1.65738
CIVλ1549 1543.7±2.6 1.45+−00..8698 4200±1300 1.65244
Column(1)-nameofemissionline,column(2)-wavelengthofthemaximumintensityofthelineinthefit(emissionlineswerefitted
withaGaussiantoadereddened, restframe,FeIIsubtractedquasarspectrum),column(3)-RestframeEquivalentWidth(REW)of
theline,column(4)-FullWidthatHalfMaximum(FWHM)oftheemissionline,Column(5)redshift.
ThefittingwindowforMgIIis2750–2840˚A, forCIV1530–1580˚A, andforCIII] +SiIII]1880–1940˚A.
(FUV) photometry from GALEX (Edge, private communi- The SED of SDSS J094533.99+100950.1 is shown in
cation). Fig. 3. It does not differ significantly from typical quasar
SEDs(e.g.Elvisetal.1994,Richardsetal.2006),consistent
Nootherspectraorphotometricdatapointswerefound withtheconclusionreachedbyDiamond-Stanicetal.(2009)
in the VO within the matching radius of 5”. All available for other WLQs. The deviation from theElvis et al. (1994)
images of thequasarand itsvicinityweredownloaded. The medianSEDinthefarUV(atNUVandFUVGALEXdata
quasarisvisibleintheSDSSand2MASSimages,andalsoon points) is due to the simple interpolation between the UV
the Palomar Observatory Sky Survey (POSS) plates in the and soft X-rays in the Elvis et al. (1994) SED, which does
opticalandnear-IR.NoX-raydata(XMM-Newton,RXTE, not constrain the true QSO SED at λ < 1000 ˚A. However,
INTEGRAL) were found. We searched for X-ray spectra in lower flux in the far-UV remains a possibility in our WLQ,
the ROSAT catalogues, but found no sources with a flux which may lead to low CIV emission due to lower ionizing
significantlyhigherthantheX-raybackground(Sol tan,pri- flux.
vatecommunication).Theradiopropertiesofourquasarand
its neighbourhood were checked in the Parkes-MIT-NRAO
(4850 MHz survey), and in the Faint Images of the Radio
3 GLOBAL PARAMETERS OF SDSS
SkyatTwenty-centimeters(VLAFIRST,1.4GHz).Inboth
J094533.99+100950.1
cases the radio background was found and thespatial reso-
lution was too low to observe the quasar. The same results Our WLQ has a monochromatic luminosity at 3000˚A of
were obtained for IRAS and EUVE catalogues - the signal λLλ(3000˚A) =2.95×1046 erg s−1, (assuming H0 =71 km
did not differsignificantly from thebackground. s−1 Mpc−1,ΩΛ =0.73andΩm =0.27-Spergeletal.2007),
WLQ SDSS J094533.99+100950.1 7
Table 5.Emissionlinepropertiesinother quasarsamples.
Line REW(composite) REW(FOS) FWHM(FOS) REW(LBQS) FWHM(LBQS)
[˚A] [˚A] kms−1 [˚A] kms−1
MgII
single 30.14 136±921 3840±1850 39±22 5160±120
broad 33±8 8230±1850 37±22 8660±300
narrow 29±9 3260±760 28±13 3510±110
CIII] 21.45 21±5 4890±780 28±15 7820±170
CIV
single 26.86 21±161 4430±3070 38±20 7720±150
broad 63±7 11090±1070 45±23 10960±360
narrow 30±4 2900±290 17.8±9.3 2860±110
EmissionlineparametersofSDSSJ094533.9+100950 arecomparedwith:incolumn(2)theRichardsetal.(2003) compositeno.3
(Fig.1)builtonthebasisof770quasarspectra,incolumns(3),(4)the158AGNobservedwithFaintObjectSpectrographonthe
HubbleSpaceTelescopes (FOS;heterogeneous sampleofmixedtypeAGN,Kuraszkiewiczetal.2002)andincolumns(5),(6) the993
quasarsfromtheLargeBrightQuasarSurvey(LBQS;Forsteretal.2001). InColumn(1)“single”meansthatthelinewasfitted bya
singleGaussian,while“broad”and“narrow”refertothebroadandnarrowcomponents oftheline.
1 -ThesingleGaussianfitstoMgIIandCIVemissionlinesinFOSwerebasedonlyonafew,lowqualityspectraandthusmaynotbe
representative.
MBH =3.4×106(cid:18)10λ4L43e0r0g0s˚A−1(cid:19)0.58±0.10(cid:20)F10W00HkMmMs−gI1I(cid:21)2M⊙(1)
from 2MASS from SDSS from GALEX
L 0.57±0.12 FWHM 2
MBH =2.9×106(cid:18)1042MerggIsI−1(cid:19) (cid:20)1000kmMs−gI1I(cid:21) M⊙(2)
whereλL andL arethemonochromaticcontinuum
˚ MgII
3000A
luminosity at 3000 ˚A and the MgII line intensity, respec-
tively. Using the MgII width from the best fit (Cont-Fe1-
bestinTable4)weobtainablackholemassof2.7×109M⊙
fromequation(1).Thisisatypicalvalueforadistantquasar
(e.g.Kellyetal.2008, Vestergaardetal.2008).Usingequa-
tion (2) we obtain a slightly lower value of the black hole
mass, of 1.5×109M⊙. Similar values of MBH are obtained
when using formulae from McLure & Dunlop (2004).
Figure3. ThebroadbandSEDofSDSSJ094533.99+100950.1.
The first value of the black hole mass implies the Ed-
The photometric IR data point (triangles) come from 2MASS
(J,H,Ks colours). Optical photometry (squares) is from SDSS dingtonratioof0.45.Thisvalueisveryconservativefromthe
(u,g,r,i,zfilters).TheNUVandFUVpoints(hexagons)arefrom point of view of possible super-Eddington accretion, which
GALEX. The spatial coordinates of the source differ between ispossiblypresentinsomeAGN(Collin&Kawaguchi2004,
SDSS and 2MASS by ∼ 0.39 arcsec and between SDSS and Jin et al. 2009). It is consistent with several other determi-
GALEX by ∼ 0.48 arcsec. We assume the data from SDSS, nationsofthisratioinquasars.Quasarsintheredshiftrange
2MASSandGALEXcomefromthesameobject.Galacticdered- between 1 and 2 in the study of Sulentic et al. (2006) have
dening was done using the values E(B-V)=0.060, RV = 3.2 and Eddington ratios between ∼ 0.2 and ∼ 1, the value 0.25 is
formulaeofCardellietal.(1989).Neitherintrinsicdereddeningof
the mean favoured by statistical studies of Shankar et al.
thequasarnorsubtractionofFeIIweredone.Forcomparisonthe
(2008).Labitaetal.(2009)foundavalueof0.45asastatis-
lineshowstheSEDfromElvisetal.(1994)markingwithdashed
ticalupperlimitofaccretionrateintheirsampleofquasars.
linethefrequencyrangeofinterpolationbetweenthefarUVand
softX-rayband. Thesecondvaluebasedonblackholemassfromequation(2)
gives a somewhat higher ratio of 0.79, which is less reliable
because of the weakness of the MgII emission line. Wealso
computed the mean and standard deviation of the Edding-
which is high but not extremely high for a quasar. Using a tonratioforallmodelsinTable2andobtainedthefollowing
bolometric correction BC = 5.15 determined at 3000˚A means: 0.429±0.037 using equation (1) and 0.752±0.077
3000
(Shenet al. 2008 and Labita et al. 2009, based on Richards for equation (2). In the newest studies of the relation be-
etal.2006)weobtainabolometricluminosityof1.52×1047 tween MgII line parameters and black hole mass - Wang
erg s−1. et al. (2009) do not use MgII equivalent width but rather
WeusethefollowingequationsfromKongetal.(2006): theMgII FWHM and continuum luminosity at 3000˚A. Us-
to determinethe central black hole mass: ingtherelationfromWangetal.(2009)themeanblackhole
8 Hryniewicz et al.
massofSDSSJ094533.99+100950.1 is(3.09±0.23)×109M⊙ However,when thediskfinally approaches ablack hole, i.e.
and the mean Eddington ratio is 0.396±0.026. whentheinnerradiusmovesfrom ∼10−20R toISCO
Schw
Additionally, since the MgII emission line intensity is (Inner Stable Circular Orbit), the disk’s luminosity rapidly
lowerthanintypicalquasars,thesecondmethodisprobably increases on a hundred year timescale and X-ray emission
less reliable than the first, based on the continuum, and so starts (Czerny 2006).
theEddington ratio of 0.45 is more likely. Theimmediatestageafteristheirradiationoftheouter
disk which happens in the light travel time across the re-
gion(years).Atthisstagetheaccretion diskcontinuumap-
pearsalreadytobetypicalforanAGN.However,neitherthe
4 DISCUSSION
broad-line region (BLR) nor the narrow-line region (NLR)
The weak line quasar SDSS J094533.99+100950.1 is an ex- region, exist.
ceptionalobjectwithalmostnonexistentCIVemission,nor- If our explanation is correct, our WLQ should have no
mal FeII emission and visible, although somewhat fainter narrow[OIII]λ5007emissionbutitsLILHβlineshouldhave
andatypical,MgIIemissionline.Thepresenceofironemis- already formed (as has the MgII line). For our z =1.6 ob-
sion with standard intensity argues against any relativistic ject these lines lie outside the observed wavelength range
enhancement of the continuum in this source. The quasar of the optical telescopes used by SDSS, but could, in prin-
continuumisnormal,similartothesteepcompositeno.3of ciple, be observed using an IR facility. In another WLQ,
Richardset al. (2003). PG 1407+265, (McDowell et al. 1995) the Hβ line is weak,
The lack of strong emission lines (except for MgII) is contrary to ourexpectations. However, significant influence
likelytobeintrinsic,consistentwiththeconclusionreached of the atmosphere above 9000˚A leading to low spectrum
byDiamond-Stanicetal.(2009)forotherWLQs.Aremark- qualityatthosewavelengthsmaypartiallyhidetheline(see
able property of the SDSS J094533.99+100950.1 spectrum Fig. 1 of McDowell et al. 1995). The puzzling properties of
isthepresenceofLowIonizationLines(LILs)suchasMgII, theX-rayemission ofPG1407+265 maybeconsistentwith
whicharethoughttoformclosetotheaccretiondisksurface the young age of the source as the emission significantly
(seee.g.Collin-Souffrinetal.1988)andtheabsenceofHigh varies in time, and is only occasionally dominated by a jet
Ionization Lines (HILs) such as CIV. The narrow compo- (Gallo 2006) indicating that a stable jet has not yet been
nent of MgII, which is thought to form at larger distances launched.
from the nucleus,is also absent from thespectrum. A relatively young age was also suggested for the class
McDowell et al. (1995) listed 10 possible explana- of Narrow LineSeyfert 1(NLS1) galaxies, in thecontextof
tions for the weakness of emission lines in their WLQ, cosmological evolution(Mathur2000)buttheirBLRiswell
PG 1407+265, but rejected most of them on the basis developed. WLQs thus possibly represent an even earlier
of overall quasar properties. Among those remaining are: stage but not necessarily in the sense of a global evolution
high extinction or BAL effects, which are inconsistent with sincetheargument of their youngage is based on BLR for-
SDSSJ094533.99+100950.1 properties,andabnormalioniz- mationstage.Iftheobservedactivityisthefirstactivestage
ingcontinuum,whichremainsapossibility.Inparticular,the of the nucleus then indeed the WLQ stage should be seen
option of high (super-Eddington) accretion rate may seem first,laterreplacedbytheNLS1stage(withefficientgrowth
attractive since the determination of the Eddington ratio ofthecentralblackhole),followedbythebroademissionline
for our object allows (but does not require) for such a pos- stage(quasarorSeyfert1).However,itisalsopossiblethat
sibility (see Sect. 3). However, the expected trends found the observed active stage represents the source’s reactiva-
byPCAanalysisperformedontheoptical/UVemissionline tion. The amount of intrinsic (nuclear) dust along the line
propertiesofPGQSO(Shangetal.2003; seeFig.10) show ofsightdoesnotseemtobelargeandthepastactivitymay
that,withincreasingEddingtonratio,theUVFeIIemission bethenaturalexplanationforsuchaclean environmentfor
(measured around MgII) decreases3, SiIII]+CIII] emission anotherwiseyoungobject.Thelargemassoftheblackhole,
increases, and the widths of MgII and SiIII]+CIII] lines particularlyinPG1407+265 (logMBH =9.8basedonCIV
decrease. The opposite is seen in our object. Additionally FWHM and the formula of Vestergaard & Peterson 2006)
the SiIV line should become stronger, but this is not seen might rather indicate a relatively late evolutionary stage,
in the WLQs of Shemmer et al. (2009). Since the emission favouring the reactivation scenario. A few stages of quasar
line properties of WLQs do not follow the trends for high lifearealsopredictedbyHopkins&Hernquist(2009).They
L/LEdd objects predicted by PCA analysis, something dif- postulatethat thelife of quasarsconsists of amergerand a
ferentthanhighL/LEddiscausingtheatypicalemissionline few episodes of high accretion rate. Theactivity is likely to
behaviour. cease through a “naked” AGN stage (Williams et al. 1999,
Theonlyexplanation,whichinouropinioniscurrently Hawkins2004) sincewhen theaccretion ratedrops, theop-
consistentwiththeintriguingemissionlinepropertiesofour tical disk recedes outward and the component of the BLR
WLQ is that the quasar activity in this source has just be- linked to the disk wind disappears (e.g. Czerny et al. 2004,
gun. Elitzur & Ho 2009).
TheformationofanaccretiondiskinanAGNisalong Anintermittentcharacterforthequasaractivityisalso
time phenomenon, occurring in millions of years (the vis- more consistent with the statistics of WLQ. The sources
coustimescaleintheouterdisk,Siemiginowskaetal.1996). are rare but not as rare as we might expect (based on the
following discussion). The irradiation causes the formation
of a wind in the upper disk atmosphere, and this material
3 not increases, as is the case for optical FeII measured around slowly rises up in the vertical direction. This wind is cus-
Hβ tomarily now considered as the source of the material from
WLQ SDSS J094533.99+100950.1 9
BLR clouds (Murray et al. 1995; Proga et al. 2000; Che- weak/absent. The BAL phenomenon is not visible in the
louche & Netzer 2003; Everett 2005). The initial rise is not spectrum so the lack of CIV line is an intrinsic property of
highly supersonic. Assuming velocities ∼ 100 km s−1 (e.g. thesource.ThepresenceofMgIIandtypicalFeIIemission
Proga 2003, Everett 2005) we can estimate how much time excludes a non-thermal, blazar-like origin of the spectrum.
the material takes to depart significantly from the disk’s ThesignificantwidthofMgIIline(FWHMof5500kms−1)
surface layers. Assumingadisk radiusof 1pc, arise byten and the FeII properties argue in turn against an extremely
percent of theradial distance takes 1000 years, and at that highaccretionrateinthisobject.Objectslikethesearediffi-
stage the low ionization lines forming close to the disk, like culttofindsinceatypicalsearchforWLQs,basedonfinding
MgIIand Hβ appear. The rise to theheight comparable to spectra with extremely weak or non-existent emission lines
the radial distance takes a factor 10 times longer, and only is likely to miss sources where one of the lines is relatively
after that time the highly ionized lines such as CIV ap- strong. However, such objects are important to find/study
pearinthespectrum.Thenarrowemissionlinecomponents astheyarelikelytoshedlightonthethemechanismdriving
take longer to form and are likely to be absent at the ini- all (or at least some) WLQs.
tialstages.Asimilarargumentwasusedfortheexplanation In the present paper we propose the following hypoth-
of the relative faintness of the [OIII] line in GPS quasars esis for the nature of WLQ. AGN are generally classified
which are also thought to be relatively young (Vink et al. accordingtotheirviewingangleandEddingtonratio.How-
2006).Thus,aWLQphaselastsforabout1000years.There ever,thoseobjectsevolveandsomeevolutionarystagesmay
are almost 100 quasars classified as WLQ (Shemmer et al. be too short to produce a consistent equilibrium between
2009,Diamond-Stanicetal.2009,Plotkinetal.2010)inthe various emitting components in the nucleus like the BLR,
whole sample of ∼ 100 000 SDSS quasars. However only a NLR and the accretion disk. It is an interesting possibil-
fractionofthose,includingSDSSJ094533.99+100950.1(this ity that the WLQ class represents such a stage, at the
paper),PG1407+265 (McDowelletal.1995) andHE0141- onset of quasar activity. The relatively high occurrence of
3932 (Reimerset al. 2005), may represent amore advanced these sources may additionally suggest that a quasar’s ac-
stage with partially developed low ionization lines. There- tive phase has an intermittent character. The whole active
fore,theprobabilityofobservingthisevolutionarystagecan phaseconsistsofseveralsubphases,eachstartingwithaslow
be estimated as 10−3−10−4. This would indicate that the development of the BLR region which manifests itself as a
typicaldurationofaquasar’sactivephaseisonly∼106−107 WLQ phenomenon. Further observational tests of this hy-
years.Ontheotherhand,thelifetimeoftheluminousphase pothesis are clearly needed.
of quasars is ∼ 107 < t <∼ 108 yrs (Haiman & Hui,
QSO
2001), but in some sources extendsover a period ∼109 yrs
(Martini&Weinberg,2001).Thetwonumberscanberecon-
ACKNOWLEDGMENTS
ciledifthetotalactivephaseofaquasarconsiststypicallyof
100 separate subphases,each of thosestarting with aWLQ Wewouldliketothankananonymousrefereeforusefulcom-
stage of BLR buildup. ments that improved our paper. We are grateful to Mar-
TheincreasingprobabilityofoccurrenceofWLQamong ianne Vestergaard for providing us with the iron emission
high redshift objects (Diamond-Stanic et al. 2009) agrees template. We would also like to thank Andrzej Sol tan for
withourhypothesisif,intheearlierUniverse,minormergers data search in ROSAT catalogues, Alastair Edge for help-
are more frequent and the typical single active subphase ful remarks about GALEX and 2MASS missions and Be-
lasts shorter. linda Wilkes for advices during improving this paper. This
Thus the SDSS J094533.99+100950.1 quasar with its work makes use of EURO-VO software, tools or services.
remarkable properties may have a Rosetta stone impact on TheEURO-VOhasbeenfundedbytheEuropeanCommis-
our understanding of quasar activity cycle and the nature sionthroughcontractnumbersRI031675(DCA)and011892
of WLQ. (VO-TECH)underthe6thFrameworkProgrammeandcon-
Finally, it is important to note that the definition of tractnumber212104(AIDA)underthe7thFrameworkPro-
a WLQ has not been yet worked out. Some objects with gramme.ThisworkwassupportedinpartbyNN203380136
extremely faint/nonexistent lines, may indeed be BL Lacs. and bythePolish Astroparticle Network 621/E-78/BWSN-
Others, like PHL 1811 (Leighly et al. 2007), considered as 0068/2008.
a WLQ prototype, may be explained by higher Eddington
ratio since the lines are not only fainter but also narrower
(MgII line of 2550±110 km s−1 vs. 5490 ±220 km s−1
REFERENCES
in J094533.99+100950.1). Our hypothesis of young/newly
active sources would not apply to those classes of objects. Adelman-McCarthyJ.K.etal.2007, ApJS,172,634
AndersonS.F.etal.2001,AJ,122,503
BursteinD.,HeilesC.,1978,ApJ,225,40
CardelliJ.A.,ClaytonG.C.,MathisJ.S.,1989, ApJ,345,245
5 CONCLUSIONS
CheloucheD.,Netzer H.,2003, MNRAS,344,223
CollinS.,Kawaguchi T.,2004,A&A,426,797
QuasarSDSSJ094533.99+100950.1 isanexceptionalexam-
Collin-Souffrin S., Dyson J.E., McDowell J.C., Perry J.J., 1988,
ple of the Weak Line Quasar class. The broad band spec-
MNRAS,232,539
trumistypicalforquasars,verysimilartothesecondsteep-
CollingeM.J.etal.2005,AJ,129,2542
est composite spectrum of Richards et al. (2003). It has a Czerny B., 2006, in Gaskell C. M., McHardy I. M.,Peterson B.
relatively well developed broad MgII line and the typical M., S. G. Sergeev S. G. eds, ASP Conf. Ser. Vol. 360, AGN
FeIIcontributionbutitshighionizationlines,likeCIV,are VariabilityfromX-RaystoRadioWaves,SanFrancisco,p.265
10 Hryniewicz et al.
CzernyB.,Ro´z˙an´skaA.,KuraszkiewiczJ.,2004, A&A,428,39 WangJ.-G,DongX.-B.,WangT.-G.,HoL.,YuanW.,WangH.,
Diamond-StanicA.M.etal.,2009,ApJ,699,782 ZhangK.,ZhangS.,ZhouH.,2009,ApJ,707,1334
DickeyJ.M.,LockmanF.J.,1990,ARA&A,28,215 WaythR.B.,O’DowdM.,Webster,R.L.,2005,MNRAS,359,561
DobrzyckiA.,EngelsD.,HagenH.-J.,1999,A&A,349,L29 WeymannR.J.,MorrisS.L.,FoltsC.B.,HewettP.C.,1991,ApJ,
ElitzurM.,HoL.C.,2009,ApJL,701,L91 373,23
ElvisM.etal.,1994, ApJS,95,1 WilliamsR.J.R.,BakerA.C.,PerryJ.J.1999,MNRAS,310,913
EverettJ.E.,2005, ApJ,631,689
FanX.etal.,1999,ApJ,526,L57
This paper has been processed by the authors using the
ForsterK.,GreenP.J.,AldcroftT.L.,VestergaardM.,FoltzC.B.,
HewettP.C.2001,ApJS,134,35 Blackwell Scientific Publications LATEX style file.
GalloL.C.,2006,MNRAS,365,960
Gaskell C. M., Goosmann R. W., Antonucci R. R. J., Whysong
D.H.,2004,ApJ,616,147
HaimanZ.,HuiL.,2001, ApJ,547,27
HallP.B.etal.,2002,ApJS,141,267
HawkinsM.R.S.2004,A&A,424,519
HopkinsPh.F.,HernquistL.,2009, ApJ,698,1550
Jin C.C., Done C., Ward M., Gierlin´ski M., Mullaney J., 2009,
MNRAS,398,16
Kelly B.C., Bechtold J., Trump J.R., Vestergaard M., Siemigi-
nowskaA.,2008,ApJS,176,355
KongM.-Z.,WuX.-B.,WangR.,HanJ.-L.,2006,Chin.J.Astron.
Astrophys,6,396
KuraszkiewiczJ.K.,GreenP.J.,ForsterK.,AldcroftT.L.,Evans
I.N.,KoratkarA.2002,ApJS,143,257
LabitaM.,DecarliR.,TrevesA.,FalomoR.,2009,MNRAS,396,
1537
Leighly K.M., Halpern J.P., Jenskins E.B., Casebeer D., 2007,
ApJS,173,1
MartiniP.,WeinbergD.H.,2001,ApJ,547,12
MathurS.,2000.MNRAS,314,L17.
McDowellJ.C.,CanizaresC.,ElvisM.,LawrenceA.,MarkoffS.,
MathurS.,WilkesB.J.,1995,ApJ,450,585
McLureR.J.,DunlopJ.S.,2004,MNRAS,352,1390
MurrayN.,ChiangJ.,GrossmanS.A.,VoitG.M.,1995,ApJ,451,
498
PlotkinR.M.,etal.,2010,AJ,139,390
ProgaD.2003,ApJ,585,406
ProgaD.,StoneJ.M.,KallmanT.R.,2000,ApJ,543,686
ReimersD.,JanknechtE.,FechnerC.,AgafonovaI.I.,Levshakov
S.A.,LopezS.2005,A&A,435,17
RichardsG.T.etal.,2002, AJ,123,2945
RichardsG.T.etal.,2003, AJ,126,1131
RichardsG.T.etal.,2006, ApJS,166,470
Schneider D.P.etal.,2007,AJ,134,102
ShangZ.,WillsB.J.,RobinsonE.L.,WillsD.,LaorA.,XieB.,
YuanJ.,2003, ApJ,586,52
ShankarF.,CrocceM.,Miralda-EscudeJ.,FosalbaP.,Weinberg
D.H.,2008,arXiv:0810.4919
ShemmerO.etal.2006,ApJ,644,86
ShemmerO.etal.2009,ApJ,696,580
ShenY.,GreeneJ.E.,StraussM.A.,RichardsG.T.,Schneider
D.P.,2008,ApJ,680,169
SiemiginowskaA.,CzernyB.,KostyuninV.1996,ApJ,458,491
Sofia U. J., Gordon K. D., Clayton G. C., Misselt K., Wolff M.
J.,CoxN.L.J.,EhrenfreundP.,2006,ApJ,636,753
SpergelD.N.etal.,2007,ApJS,170,377
Stark A., Gammie Ch.F., Wilson R.W., Bally J., Linke R.A.,
HeilesC.,HurwitzM.1992,ApJS,79,77
Sulentic J.W., Repetto P., Stirpe G.M., Marziani P., Dultzin-
HacyanD.,CalvaniM.2006,A&A,456,929
VestergaardM.,PetersonB.M.,2006, ApJ,641,689
VestergaardM.,WilkesB.J.,2001,ApJS,134,1
Vestergaard M., Fan X., Tremonti C.A., Osmer P.S., Richards
G.T.,2008,ApJ,674,L1
VinkJ.,SnellenI.,MackK.-H.,SchilizziR.,2006,MNRAS,367,
928