SDSS J1138+3517: A quasar showing remarkably variable broad absorption lines

0.49 MB·English

by  Conor Wildy

Mon.Not.R.Astron.Soc.000,1–17(0000) Printed3March2015 (MNLATEXstylefilev2.2)
SDSS J1138+3517: A quasar showing remarkably variable
broad absorption lines
5 C. Wildy1⋆, M. R. Goad1 and J. T. Allen2
1
1Universityof Leicester, Department of Physicsand Astronomy, University Road, Leicester, UK
0
2Sydney Institute for Astronomy, School of Physics, A28, The University of Sydney, NSW 2006, Australia
2

r
a
M Accepted

2
ABSTRACT

]
A We report on the highly variable Si iv and C iv broad absorption lines in SDSS
G J113831.4+351725.2acrossfourobservationalepochs.Usingthe Siiv doubletcompo-
nents,wefindthatthebluecomponentisusuallysaturatedandnon-black,withthera-
.
h tioofopticaldepthsbetweenthetwocomponentsrarelybeing2:1.Thisindicatesthat
p these absorbersdo not fully coverthe line-of-sightandthus a simple apparentoptical
- depth model is insufficient when measuring the true opacity of the absorbers. Tests
o
r with inhomogeneous (power-law) and pure-partial coverage(step-function) models of
t the absorbing Si iv optical depth predict the most un-blended doublet’s component
s
a profilesequally well. However,when testing with Gaussian-fitteddoublet components
[ toallSiivabsorbersandaveragingthetotalabsorptionpredictedineachdoublet,the

upper limit of the power law index is mostly unconstrained. This leads us to favour
2
pure partial coverage as a more accurate measure of the true optical depth than the
v
5 inhomogeneous power law model.
7 The pure-partial coverage model indicates no significant change in covering frac-
9 tion across the epochs, with changes in the incident ionizing flux on the absorbing
4 gas instead being favouredas the variability mechanism. This is supported by (a) the
0 coordinatedbehaviour of the absorption troughs,(b) the behaviour of the continuum
1. at the blue end of the spectrum and (c) the consistency of photoionization simula-
0 tions ofionic columndensity dependencies onionizationparameterwiththe observed
5 variations.EvidencefromthesimulationstogetherwiththeCivabsorptionprofilein-
1 dicatesthattheabsorberliesoutsidethebroadlineregion,thoughtheprecisedistance
: and kinetic luminosity are not well constrained.
v
Xi Key words: galaxies: active-quasars:absorption lines
r
a
1 INTRODUCTION portion of the quasar’s rest-frame spectrum can be blue-
shifted by as much as 0.1c, where c is the speed of light
Blue-shifted quasar absorption lines, which indicate the
in a vacuum, from the corresponding emission line centre.
presence of large amounts of outflowing material from the
Theseabsorptionlinesaredividedintocategoriesaccording
centralengine,havebecomeanincreasinglyimportanttopic
to their velocity width, the widest being broad absorption
over the past two decades. This followed the realisation of
lines(BALs)extendingoveratleast2000kms−1 whereflux
the potential influence these flows can exert on the host
is less than 90 per cent of the continuum (Weymann et al.
galaxybyrestrictingblackholegrowth(Springel et al.2005;
1991).QuasarshostingatleastoneBALareknownasbroad
King 2010) and contributing to the evolution of galac-
absorptionlinequasars(BALQSOs).Absorptioncanalsobe
tic bulges (Magorrian et al. 1998; Silk & Rees 1998). This
seen as narrow absorption lines (NALs), which extend over
feedback process is also supported by evidence that out-
a few hundred km s−1 and mini-BALs, which are of inter-
flows could extend hundreds of parsecs into the surround-
mediatewidthbetweenNALsandBALs(Hamann & Sabra
inginterstellarmedium(Barlow1994),depositingsignificant
2004).
amountsof kineticenergyintheprocess (Arav et al. 2013).
Absorption features observed in the ultraviolet (UV) The population of BALQSOs, of which SDSS
J113831.4+351725.2 (hereafter SDSS J1138+3517) is a
member, account for approximately 10 per cent to 20 per
⋆ E-mail:[email protected] cent of the total quasar population (Reichard et al. 2003;
(cid:13)c 0000RAS
2 C. Wildy, M. R. Goad and J. T. Allen
Knigge et al.2008;Scaringi et al.2009).HoweverSDSStar- thesesubsequentobservationsalsobeingmadeinsuccessive
get selection effects may make this value as high as 41 1-yearrest-frame time intervals.
per cent (Allen et al. 2011). If BAL observability is gov-
erned by line of sight orientation, then large-scale out-
In Filiz Aket al. (2013), the variability of 428 C iv
flows may occur in the vast majority of quasars, even
BALs in 291 quasars was found to produce a ∆EW de-
where no BALs are visible (Schmidt & Hines 1999). Com-
pendenceonrest-frametimeintervalwhichiswellpredicted
mon BALsubtypesincludeHigh-Ionization BALQSOs(Hi-
by a random-walk model (see Fig. 29 of that paper). Us-
BALs) ( 85 per cent) whose spectra only show broad ab-
ing the same BAL identification scheme as that study, the
∼
sorption due to high ionization transitions such as C iv
variation across the two observational epochs examined in
λλ1548,1550 Si iv λλ1394,1403 and N v λλ1239,1243
Wildy et al. (2014) of the most variable C iv BAL region
(Sprayberry& Foltz 1992; Reichard et al. 2003). Low Ion- in SDSS J1138+3517 (spanning 23400 km s−1 to 7500
ization BALQSOs(LoBALs),in additiontohigh-ionization kms−1)is∆EW= 13.8 2.9˚A.−Thisvaluelieswello−utside
BALs,alsoexhibitbroadabsorptionresultingfromlowion- − ±
the random-walk model prediction at the epoch separation
ization species such as Mg ii, Al iii and on rare occasions
time-intervalof360daysinthequasarrest-frame,suggesting
Feii (known as FeLoBALs).
that this object may exhibit a variability mechanism which
Ionizing continuum changes are known to drive
is rare in theBALQSO population.
broad emission line (BEL) variability (Peterson et al. 1998;
VandenBerk et al.2004;Wilhite et al.2006),howeverthere
InFiliz Aket al.(2014),whichusesthesameBALiden-
is as yet no consensus behind the dominant mechanism
tification scheme as above, the variability of C iv BALs
involved in absorption line variablity. Several large sta-
was investigated over timescales ranging from 1 to a few
tistical studies of BAL variability have been published ∼
years, including those in quasars additionally containing
in recent years (Lundgrenet al. 2007; Gibson et al. 2008,
Si iv BALs of overlapping velocity. Again examining the
2010;Capellupo et al.2011,2012,2013;Filiz Ak et al.2013;
most variableCivBALinSDSSJ1138+3517 identifiedus-
Wildy et al. 2014). These studies refer to coordinated vari-
ingtheirdefinition,thevalueof∆EW,alongwithavalueof
ability across widely separated absorption troughs as evi-
∆EW/ EW = 1.3 0.3 would place it among the top few
dence favouring changes in input ionizing flux as the main h i − ±
most variable C iv BALs in their sample of 454 (includes
mechanism behind variability (Filiz Ak et al. 2013), while
C iv BALs both unaccompanied by Si iv absorption and
variability overnarrowvelocity ranges within troughsis at-
overlapping in velocity with a Si iv BAL), highlighting the
tributed to absorbing gas moving across the line of sight
extreme variability of this source. In addition, an investi-
(Gibson et al. 2008; Capellupo et al. 2012).
gation by Gibson et al. (2010) included observations of two
AsinWildy et al.(2014),weuseanovelmethodtore-
quasarsoverasimilarrest-frametimeintervalofbetween0.5
constructtheun-absorbedspectralprofile(Allen et al.2011)
and2years,neitherofwhichshowedvariabilityasdramatic
against which the absorption can be measured. Absorption
as in SDSSJ1138+3517.
isidentifiedbasedon thisemission profileasit containsthe
total input spectral flux (continuum+emission line) enter-
ing the absorbing gas at each wavelength bin. We use this Answering thequestion of what drives thestrong BAL
method to estimate the column density at each epoch for variability observed in thisquasaris theprincipalobjective
the Si iv and C iv ions. This has advantages over purely ofthisinvestigation.Observationsofthesourcearedetailed
continuum-based absorber strength measurements as it al- in Section 2. In Section 3, we quantify Si iv and C iv ionic
lowsabsorptiondepthtobeaccuratelyidentifiedinspectral column densities using different assumptions regarding the
regions where the absorption is overlapping with an emis- line-of-sight geometry, including models where there is in-
sion line profile, allowing absorption features to be studied completecoverageoftheemission region byahomogeneous
to arbitrarily low (less negative) velocities. absorber.ThisisaidedbytheuseofGaussian profilestofit
the Si iv absorption components, which allows calculations
to be performed based on the changes in these components
1.1 The Unusual BAL behaviour of SDSS
even where they are not fully un-blended. In Section 4 we
J1138+3517
compute a grid of photoionization models, allowing exam-
ination of the theoretical variation of ionic column density
Thequasarexaminedin thispaper,SDSSJ1138+3517, has
overarangeofionizationparameteratgivenhydrogennum-
aredshiftofz=2.122(Hewett & Wild2010)andwasinvesti-
ber densities and column densities. These grid models are
gatedaspartofthemulti-BALQSOvariabilitystudyunder-
compared to the best results for the ionic column densities
taken in the rest-frame UV by Wildy et al. (2014). Using a
fromSection3,allowingustotestthehypothesisofchanges
sampleof50quasars,Siivλ1400andCivλ1549BALswere
in ionizing flux being the driver of BAL variability. In Sec-
identified and statistics gathered on their equivalent width
tion5wediscusstheresultsfromSections3and4andtheir
(EW)changesover2epochsseparatedbyquasarrest-frame
implicationsforthevariabilitymechanism.Wealsoattempt
timescales ranging from approximately 1 to 4 years. Of the
to constrain the mass outflow rate and hence kinetic lumi-
50 BALQSOs, SDSS J1138+3517 showed the largest abso-
nosity.
lutechangeinEWinbothSiivandCivabsorptionaswell
asalargefractionalEWchange(∆EW/ EW ),occurring
| h i|
over a rest-frame time interval of 1 year. A third epoch This paper assumes a flat ΛCDM cosmology with
indicated another dramatic variatio∼n in the BALs, while a H0=67 km s−1 Mpc−1, ΩM=0.315 and ΩΛ=0.685, as
fourth epoch showed relatively little change from the third reported from the latest results of the Planck satellite
except for significant variability in one C iv absorber, with (Planck Collaboration et al. 2014).
(cid:13)c 0000RAS,MNRAS000,1–17
A quasar showing remarkably variable BALs 3
2 OBSERVATIONS 3 ANALYSIS
2.1 The Sloan Digital Sky Survey 3.1 Spectral Properties
The epoch 1 observation was obtained from DR6 of the Theobservedspectraateachepochcorrespondtorest-frame
Sloan Digital Sky Survey (SDSS). Beginning in 2000, the UV wavelengths which include the region containing Si iv
SDSShadimagedapproximately10000deg2oftheskyasof andCivemissionandthecorrespondingblueshiftedabsorp-
June2010(Schneideret al.2010),usingthe2.5mTelescope tion lines resulting from those same atomic transitions in
at the Apache Point Observatory, New Mexico, USA (York outflowinggas. Sincethespectralresolution differsbetween
2000).Imaging is carried out usinga CCD camera that op- spectra, in order to accurately identify changes in absorp-
eratesindrift-scanningmode(Gunnet al.1998).Fivebroad tion between each epoch the SDSS and WHT epoch 3 and
bandfilters,u,g,r,iandz,normalised totheABsystem,are epoch 4 spectra were convolved with Gaussians of appro-
used,coveringawavelengthrangeof3900to9100˚Awithan priatefull-widthhalfmaximatoapproximatetheresolution
approximate spectral resolution of λ/∆λ=2000 at 5000 ˚A of the WHT epoch 2 observation (using skyline widths as
(Stoughton et al. 2002). Further information on the SDSS a guideline), followed by resampling of all 4 spectra onto a
data reduction process is given in Luptonet al. (2001) and wavelengthgridof3.7˚A.Thiscloselymatchesthebinwidth
Stoughton et al. (2002). of the epoch 2 observation. Small differences in wavelength
calibration were removed by aligning narrow emission and
absorptionfeaturesintheWHTspectrawiththesamelines
in the SDSS spectrum. The SDSS and both WHT spectra
were corrected for the effects of Milky Way Galactic Ex-
2.2 The William Herschel Telescope tinction usingthemethod of Cardelli et al. (1989) with AV
valuestaken from Schlegel et al. (1998).
The William Herschel Telescope (WHT) ISIS (Intermedi- Using the reconstruction of the unabsorbed SDSS
ate dispersion Spectrograph and Imaging System) double- spectrum from Wildy et al. (2014) originally developed in
armed (red and blue) spectrograph was used for longslit Allen et al. (2011), appropriate reconstructions were also
spectroscopy of the target during semester A 2008, 2011 generated for the three WHT observations. This was
and 2014 corresponding to epochs 2,3 and 4 respectively, achieved as follows: (i) A power-law continuum was fitted
with epochs 3 and 4being observed as part of theINGser- to each spectrum using line-free continuum bands as out-
viceprogrammeandepoch2takenfromthesampleusedby lined in Wildy et al. (2014), (ii) The SDSS continuum was
Cottis et al. (2010). The observations provide spectral cov- subtracted from the SDSS reconstruction, (iii) The WHT
erage from 3000 to 10000 ˚A. For epoch 2 the 5700 dichroic spectrahadtheirrespectivecontinuasubtracted,(iv)Inthe
was used, providing spectra from the two arms with over- continuumsubtractedSDSSreconstruction,theSiivλ1400,
lap in the range 5400 to 5700 ˚A. The R158B and R158R C ivλ1549, O i λ1304 and C iiλ1334 emission linebound-
gratings, giving a nominal spectral resolution of 1.6˚A per aries were identified, (v) The emission lines from part (iv)
pixel in the blue and 1.8˚A per pixel in the red, were also werescaledtomatchtheredside(unabsorbed)oftheemis-
utilised for epoch 2. Epoch 3 and 4 observations employed sion lines in each of the WHT observations, and (vi) The
the 5300 dichroic, giving an overlap range of 5200-5500 ˚A, continuum was added to the scaled emission lines for each
whiletheR300BandR316Rgratingswereusedfortheblue of the WHT observations, creating appropriate reconstruc-
and red observations, providing spectral resolutions of 0.86 tions for epochs 2,3 and 4.
˚Aperpixeland0.93˚Aperpixelrespectively.Inthedichroic For the SDSS reconstruction used in the subsequent
overlap regions, an error weighted mean was used to com- analysis, the procedure was carried out to scale the Si iv
bine the spectra from the two arms for each observation. emission line, as the original reconstruction underestimates
For all WHT observations, a slit width of 1.5” was cho- the Si iv emission (the reconstructions were optimised for
sen to match typical seeing conditions and give reasonable C iv as described in Allen et al. (2011)). The region used
throughput without compromising the spectral resolution. forcalculatingtheappropriateCivlinescalingisrestricted
Multiple exposures were taken (in order to remove cosmic to the interval between line centre and an observed wave-
ray hits) bracketed by arc-lamp exposures for wavelength lengthof4866˚A.ThisistoavoidtheHeii+Oiiiabsorption
calibration.Standardstarspectraweretakenontheobserv- and emission complex immediately longward of C iv. The
ingnightstoallowfluxcalibrationofthetargets.Correction Siivemission linewasshiftedbytwowavelengthbinsblue-
of CCD images using bias and flat field exposures was per- ward in the epoch 2 reconstruction relative to the SDSS
formed within the iraf ccdproc routine. Spectra were ex- reconstruction to match the shifted peak of the emission
tracted,alongwithskybackgroundremoval,usingtheapall line in the epoch 2 spectrum. Based on the synthetic BAL
task within the iraf longslit package. This used an extrac- based error estimation method described in Section 3 of
tion slit width of 10 pixels and optimal weighting, with 2 Wildy et al. (2014), the fractional error on the reconstruc-
pixels per wavelength bin. Wavelength and flux calibration tionforaquasarofthisr-bandS/Nandredshiftisestimated
wereappliedtotheextractedspectrawithinthissamepack- tobe5percentofthefluxvalueateach resolution element
age.Thespectrawereplacedonanabsolutefluxscaleusing for all calculations described in this paper. The observed
spectra of a photometric standard observed at similar air- spectraforall4epochs(solidblacklines),aswellastheirfi-
mass.Noattemptwasmade,suchasthroughtheuseofgrey nalreconstructions(solidredlines),areillustratedinFig.1.
shifts, to adjust the flux scale on the calibrated spectra to- The dramatic variability in both ions’ absorbers across
wards levelswhich more closely matched across theepochs. the first three epochs is clear when the spectra are nor-
Table 1 summarises theSDSS J1138+3517 observations. malised to the reconstructions. The transitions of interest
(cid:13)c 0000RAS,MNRAS000,1–17
4 C. Wildy, M. R. Goad and J. T. Allen
Table 1.Detailsofobservations obtainedofSDSSJ1138+3517
Epoch ObservationDate ∆tqrest Type Exp.time Grating PixelResolution(5000˚A) MeanAirmass
(days) (s) (kms−1)
1 03May2005 0 SDSS 6×2400 Red&Blue 150 1.01
2 27May2008 359 WHT 3×1200 R158B&R158R 108 1.14
3 29May2011 710 WHT 2×1800 R300B&R316R 56 1.80
4 09May2014 1053 WHT 2×1800 R300B&R316R 56 1.06
∆tqrest isquasarrest-frametimeintervalsinceEpoch1
1300 1400 1500 1600
30
SDSS (Epoch 1)
20 1
10 0.8
300 0.6
WHT (Epoch 2)
20 0.4
10 0.2
300 0
WHT (Epoch 3)
20 1
10 0.8
600 0.6
WHT (Epoch 4)
40 0.4
20 0.2
0 0
4000 4200 4400 4600 4800 5000 -5000 0
Figure 1.Spectra for epochs 1 to 4 of SDSS J1138+3517, with Figure2.Siiv(upperpanel)andCiv(lowerpanel)absorption
top panel to bottom panel in order of observation date starting regionsinredcomponentvelocityspace.Epoch1(SDSS)spectra
withtheearliest.Theobservedspectrumisinblackwhilethere- areinblack,epoch2(1stWHT)spectraareinred,epoch3(2nd
constructionoftheunabsorbedspectrumisinred.Verticaldotted WHT)spectraareingreenandepoch4(3rdWHT)spectraarein
lines indicate the laboratory wavelengths of the Si iv and C iv blue. Spectra are normalised to the un-absorbed reconstruction.
emission lines. The vertical dashed line indicates the maximum The Si iv Gaussian components (described in Section 3.3) are
wavelengthusedincalculatingtheCivemissionscaling,emission numericallylabelledinorderofincreasingoutflow velocity
longwardofthispointisnotaccurately reconstructed.
are both doublets, with rest-frame laboratory wavelengths
of 1393.76 and 1402.77 ˚A contributing to Si iv λ1400 and
correspondingly1548.20and1550.77˚AforCivλ1549.Rela-
tivetothelaboratory-framerest-wavelengthoftheredcom-
ponent of each doublet, absorption is significant between 0 hasbeenreportedinpreviousBALstudies(Lundgren et al.
and 13000kms−1forSiivandbetween0and 20000 2007; Capellupo et al. 2011).
km s∼−−1 for C iv, where a negative value indicates∼o−utflow- Due to its relative lack of variation from the previous
ing material (blueshifted). As can be seen in Fig. 2, the observation, especially in the Si iv absorption lines where
C iv troughs are deeper than the Si iv troughs at similar identifiable doublets are present, epoch 4 is left out of the
velocities. The deepest troughs show the least variability analysis in subsequent parts of Section 3 and Section 4.
and are located at the lowest (least negative) velocities, as Epoch 4 is instead discussed furtherin Section 5.3.
(cid:13)c 0000RAS,MNRAS000,1–17
A quasar showing remarkably variable BALs 5
3.2 Lower Limits for Outflow Column Densities
using Direct Integration
1
Minimum values for both the Si iv and C iv column den-
sities can be estimated by assuming unsaturated absorbers 0.8
completely cover theemitting line+continuum region along
0.6
ourlineofsightwithaconstantopticaldepthacrossthetan-
gentialplane(Savage & Sembach1991).TheCivandSiiv 0.4
transitions have a doublet structure with known oscillator
0.2
strengths. The total column density of theion contributing
to a given doublet component may be calculated according 0
1
to
mec 0.8
Nion = πe2fλZ τ(v)dv (1) 0.6
where Nion is the ionic column density, v is the velocity 0.4
relative to the laboratory rest-frame wavelength, me is the
0.2
electron mass, c is the speed of light, e is the elementary
charge,f is theoscillator strength,λisthelaboratory rest- 0
frame wavelength and τ(v) is the optical depth at a given 1
velocity.Ataredshiftofz=2.122,thedoubletseparationat
0.8
zerovelocityisapproximately28.1˚AforSiivand8.0˚Afor
Civ.AstheCivseparationisonlyslightlygreaterthantwo 0.6
wavelengthbins,itispracticallyimpossible,giventheirlarge
0.4
intrinsic widths, to identify individual C iv components.
Though some Si iv components are well-separated, many 0.2
of the doublets are blended to various extents with neigh- 0
bouring absorption features. However, the optical depth of -8000 -6000 -4000 -2000 0
overlapping components will be the sum of the individual
components and the optical depth of the red component is
known to be half that of theblue component for both ions, Figure 3. Epoch 1 (top panel), epoch 2 (middle panel) and
given the ratio of their oscillator strengths. Therefore, by epoch 3 (lower panel) models of Si iv doublet absorption lines
integrating over all of the absorption for a given transition inredcomponentvelocityspace(relateddoubletcomponentsare
using thevalueof f corresponding to thered component, a shown in the same colour). The black dotted line describes the
minimum value of Nion can be found by taking 1/3 of the totalmodelabsorptionprofile,whilethereddottedlinenearthe
bottom ofeachpanel indicates thedifferencebetween themodel
calculated value. The velocity ranges integrated over and
profileandtheobservedprofile.
resulting estimated ionic column densities areshown in Ta-
ble 2.
Since the C iv doublet is unresolved, this method is
sation. In total, five Si iv doublets are identified across the
the only means of estimating a lower limit for the column
3 epochs, spanning the range 100006v60 km s−1. These
density of this ion. For Si iv, methods involving modelling −
are shown in Fig. 3.
thecomponentsneedtobecheckedforconsistencywiththe
Comparing the Gaussian fit to the almost un-blended
lower limit derived here.
red component of the lowest velocity doublet in epoch 2
(shown in red in Fig. 3) to the data points at these veloc-
ities gives a fractional error in the fit of 0.033. To be con-
3.3 Gaussian Components of the Si IV Outflow
servative a larger error of 5 per cent on each point of all
The Si iv doublet velocity separation is 1920 km s−1, Gaussianprofilesisassumedforcalculationsinvolvingthese
∼
allowing some individual components to be identified. As- model profiles. While theEW and occasionally theFWHM
suming narrow components of quasar absorption lines fol- change from one epoch to the next, there is no significant
low an approximately Gaussian profile, e.g. Hamann et al. changein thelinecentrevelocity andhencenoevidencefor
(2011), model profiles can be constructed for the doublet accelerationoftheoutflowingmaterial.Thislackofaccelera-
lines. Spectralfittingisperformed usingthespecfit package tionhasbeennotedinpreviousstudies(Hamann et al.2008;
within iraf,requiringthreefree parameters for each Gaus- Rodr´ıguez Hidalgo et al.2011).Thebest-fitparametersob-
sian component, namely the wavelength at line centre, line tained by specfit for each doublet component are provided
fullwidthhalfmaximum(FWHM)andlineEW.Appropri- in Table 3.
aterestrictionsareappliedtotheseparameters,i.e.theEW Uponexamination ofanindividualcomponent,thefol-
ratio between the blue and red components of the doublet lowingequationcanbeusedtoestimatethecolumndensity
must be between 2:1 and 1:1, the difference in velocities at giving rise to the doublet to which it belongs:
line centre must not differ from the expected doublet sepa-
ration by more than one velocity bin, and the widths must N = √mπee2cfbλτ0 (2)
bethesame.Byusinganinitial’guess’fortheseparameters’
values,spectralfittingisperformedusingchi-squareminimi- whereτ is thepeakoptical depthandb istheDopplerpa-
0
(cid:13)c 0000RAS,MNRAS000,1–17
6 C. Wildy, M. R. Goad and J. T. Allen
Table 2.CivandSiivvelocitylimitsandioniccolumndensitiesforeachepochusingdirectintegrationoftheabsorptionprofile.
Transition VelocityLimit† Epoch1IonicColumnDensity Epoch2IonicColumnDensity Epoch3IonicColumnDensity
(kms−1) (×1014cm−2) (×1014cm−2) (×1014cm−2)
Siivλ1400 -13100 24.7±2.66 7.21±2.37 11.7±2.41
Civλ1549 -19700 190±15.2 95.8±8.64 151±11.3
†Integrationspanszerovelocitytovelocitylimit
Table 3.Listofcomponents withparametersforGaussianmodelprofiles.Doubletsarelistedinorderofincreasinglynegativevelocity
(1islowest,5ishighest).Doubletvelocityismeasuredfromtheredcomponent restwavelength.
Absorber Velocity‡ Epoch1 Epoch2 Epoch3
(kms−1)
FWHM‡ EW† FWHM‡ EW† FWHM‡ EW†
(kms−1) (˚A) (kms−1) (˚A) (kms−1) (˚A)
1(Red) −2300 900 3.18±0.09 900 2.03±0.07 800 2.69±0.04
1(Blue) 3.18±0.16 2.80±0.07 3.21±0.05
2(Red) −3300 800 1.65±0.12 600 0.32±0.07 900 1.01±0.06
2(Blue) 1.76±0.08 0.60±0.08 1.62±0.07
3(Red) −4300 1100 2.62±0.24 500 0.19±0.08 900 0.60±0.11
3(Blue) 2.92±0.18 0.19±0.07 0.72±0.07
4(Red) −4800 1000 2.21±0.22 500 0.16±0.08 700 0.48±0.07
4(Blue) 2.88±0.18 0.21±0.08 0.97±0.06
5(Red) −5900 700 0.77±0.15 700 0.15±0.08 700 0.09±0.06
5(Blue) 1.02±0.13 0.15±0.11 0.16±0.06
†EWinquasarrest-frame
‡Error∼200kms−1
rameter (related to line width by b=√2 FWHM/2.355). will underestimate the true optical depth of the absorbing
×
This assumes a Voigt profile for theabsorbers, which tends gasandconsequentlywillunderestimatethecolumndensity.
towardstheGaussian caseinthelimit oflowopticaldepth. The observed flux Iapp is related to thecovering fraction C
Red components are less likely to be affected by satura- by Iapp = CIout +(1 C)I0, where I0 is the unabsorbed
−
tion since they are weaker than their corresponding blue emission flux and Iout is the true output flux from the ab-
components. As a result, column density measurements sorber.ThusthecoveringfractionCasafunctionofvelocity
are taken from red doublet lines. As in Section 3.2, es- is given by
timates of the absorbing column density are lower limits
as the absorber may not cover 100 per cent of the emis- C(v)= Ir2(v)−2Ir(v)+1 (3)
sion region. Summing the values obtained for the five dou- Ib(v)−2Ir(v)+1
blets listed in Table 3 gives estimates of the total col- whereIr(v)andIb(v)aretheapparentresidualfluxesofthe
umn densities as NSi IV,1=33.5 1.49, NSiIV,2=7.79 0.50 red and blue componentsrespectively at a given velocity v.
and NSiIV,3=14.7±0.47 in units±of 1014 cm−2 for the±first, Since the true optical depth τ = ln(I0/Iout), the following
secondandthirdepochsrespectively.Asexpected,theseare is also true
largerthanthelimitsinTable2sinceblue:reddoubletcom-
ponent ratios are less than 2:1. τ = ln Ir−Ib (4)
− (cid:16)1 Ir (cid:17)
−
whereτ is theopticaldepthoftheredcomponent at aspe-
3.4 A Pure Partial Coverage Model cificvelocity.SinceC61,thevaluesofIrandIbcanbeshown
to be constrained to the range Ir >Ib >Ir2. Assuming this
Manyabsorptionlinesystemshavebeenfoundtobebestde- condition is met, it is then straightforward to calculate the
scribed byasituation in which material of acertain optical column density using Equation 1.
depth covers a fraction of the emission source, leaving the Evidence for pure partial covering can be obtained by
restofthesourceuncovered(Arav et al.1999;deKool et al. comparing the residual flux velocity dependence of a (al-
2001) otherwise known as a pure partial coverage (PPC) most)saturatedabsorptionlinetotheprofileof1 C.Simi-
−
model. Under these conditions the apparent optical depth larprofileswouldsuggesttheshapeoftheabsorptionprofile
τ = ln(1/Ires), where Ires is the residual intensity (nor- isdeterminedbythevelocity-dependentcoveringfractionof
malised to the unabsorbed continuum+line emission flux), anopaqueabsorberratherthanintrinsicdifferencesinopti-
(cid:13)c 0000RAS,MNRAS000,1–17
A quasar showing remarkably variable BALs 7
evidence that partial coverage is a significant effect in this
absorber and possibly others in this quasar.
Given that this is the only doublet where meaningful
results can be obtained on an individual velocity bin ba-
111
sis, and even in this case it can only be done close to the
peak of the absorption components, it is necessary to use
theGaussianmodelstoachieveestimatesofcolumndensity
across agreater proportion oftheSiivabsorption for the3
000...888 epochs.IndividualvelocitybinsintheGaussianmodelshave
nophysicalrealisation,sounlikeforthepreviousexampleit
is impossible to construct 1 C profiles using Equation 3.
−
However if for each doublet component, model fluxes are
averaged over all velocities spanned by the absorption, in-
000...666
dividual components can be treated as individual velocity
bins.Thismethodcansubsequentlybeusedtoestimateav-
erage values of covering fraction and true optical depth for
each doublet. Using this true optical depth in Equation 1
000...444 gives the doublet’s column density along the line of sight
whenNion ismultipliedbythecoveringfraction.Theseval-
uesarelistedinTable4.Duetotheweaknessofabsorbers3
to5inepochs2and3thiscanonlybeperformed acrossall
absorbers in epoch 1, since at small absorber depths opti-
000...222
caldepthandcoveringfractionmeasurementsareunreliable
duetotheirsensitivedependenceondoubletcomponentra-
tio.
000
---333555000000 ---333000000000 ---222555000000 ---222000000000 ---111555000000 ---111000000000 3.5 An Inhomogeneous Absorber Model
AnalternativetothePPCmodelisasituationwherebythe
optical depth varies with spatial location across the line of
Figure 4. The absorber 1 profiles of the red and blue compo-
sight. This was originally developed using a power-law de-
nents interpolated onto the red component’s velocity grid. The
pendent optical depth model in deKool et al. (2002) and
solidred and blue lines represent the Gaussian models, the dot-
was subsequently investigated by Arav et al. (2005), who
ted lines represent the observed fluxes, the green short-dashed
line indicates the 1−C values at these points and the magenta foundpower-lawandGaussian shapedopticaldepthdepen-
long-dashedlinerepresentsthee−τ valuesoftheredcomponent. denciesona1-dimensionalspatiallocationtobeinadequate
Appropriately coloured triangles and squares represent the ob- in describing O vi absorbers in Mrk 279. Here, we attempt
servedpointsinvelocityspace.Whilethe1−C profileisagood asimilar investigation usingapower-law model of theform
match forthe blueabsorber,the e−τ profiledoes not matchthe τ(x,λ)=τmax(λ)xa,whereτ(x,λ)istheopticaldepthatpo-
redprofile. sitionxonthesurfaceoftheabsorberandatwavelengthλ,
τmax(λ) is themaximum optical depthof theabsorber and
a is the power-law index.
Following Arav et al. (2005), we make two simplifying
caldepth.Anadditionaltestistocomparetheprofileofe−τ assumptionstoaidcalculations withoutanyloss ofgeneral-
to the red component of an absorption doublet. If the pro- ity. First, we assume that the surface brightness is uniform
filesdonotmatch,theabsorptionlineprofilecannotbedue across theface oftheemission source suchthatS(x,y,λ)=1,
to an absorber which is homogeneous and completely cov- where x and y are spatial positions in the plane of the line
ers the emission region. In order to carry out these tests, a ofsightandSisarbitrarilynormalised.Wealsosimplifythe
doubletwithunblendedcomponentsisdesirable.Theclosest optical depth spatial dependence by using a 1-dimensional
example available in the dataset is absorber 1 in the epoch model, so that τ(x,y,λ)=τ(x,λ). The x values are fixed to
2spectrum.FromthevaluesoftheEWsoftheredandblue span an interval 06x61 for simplicity. The observed resid-
components(seeTable3),theblue:redratioislessthan2:1, ual flux at a given velocity bin will then be found using
indicating that the blue component is probably saturated. Ires= 1e−τ(x)dx. To assess a broad range of power-law in-
0
After placing the blue component onto the red component dices,Rvaluesfrom0to10withstepsizeof0.1aretested.In
velocity grid by interpolation, six velocity bins are found order to integrate over the x range, it is effectively divided
which follow the general shape of the corresponding Gaus- into1000 bins, giving dx=0.001.
sian models for both the red and blue components. Five of Similar to the procedure in Section 3.4, we first exam-
these pointssatisfy Ir>Ib>I2r and are illustrated in Fig. 4. inetheplausibilityofapower-lawmodeldescribingthereal
From the profile traced by 1 C (Fig. 4, dashed green behaviour of the absorption features by attempting to pre-
−
line), it is clear that the saturated blue component could dicttheprofileofthebluecomponentofabsorber1inepoch
have its profile determined by the uncovered fraction. It is 2 after interpolation of the blue absorption line profile into
also obvious that the residual intensity of the red compo- thevelocity space of thered component.Given avalueof a
nent is a poor match to e−τ. Together these provide strong to be tested, the process is carried out as follows: (i) Find
(cid:13)c 0000RAS,MNRAS000,1–17
8 C. Wildy, M. R. Goad and J. T. Allen
Table 4. Table of covering fractions and column densities calculated forabsorbers 1to 5. Absorbers 3 to5 arenot listedforepochs 2
and3duetotheirlowstrength.
Absorber Epoch1 Epoch2 Epoch3
C N C N C N
ion ion ion
×1014cm−2 ×1014cm−2 ×1014cm−2
1 0.225±0.013 23.3+15.6 0.234±0.042 7.17+3.56 0.236±0.020 12.3+4.40
−14.0 −2.87 −3.881
2 0.187±0.017 11.0+8.68 0.318+0.682 0.773+13.7 0.275±0.165 2.898+4.010
−7.40 −0.318 −0.773 −0.165
3 0.339±0.017 14.1+4.20
−3.88
4 0.366±0.032 8.45+2.34
−2.09
5 0.122±0.043 2.94+3.61
−2.27
Table 5. Power indices calculated by matching the predicted
bluecomponentofaninhomogeneouspower-lawtotheGaussian
modelequivalent(’unc’indicates’unconstrained’).Severalofthe
valueshavereachedtheupperlimitof10,necessarilyleavingthe
111
upperboundunconstrained inthesecases.
Absorber Power-lawindex
Epoch1 Epoch2 Epoch3
1 10.0+unc 9.9+unc 10.0+unc 000...888
−0.9 −3.3 −1.3
2 10.0+unc 7.1+unc 6.5+unc
−1.2 −6.1 −3.6
3 10.0+unc 5.4+unc 1.0+unc
−2.2 −2.9 −0.4
4 6.4+2.6 3.9+unc 4.6+unc
−1.7 −2.2 −3.2
5 0.7+unc 8.3+unc 10.0+unc 000...666
−0.4 −5.1 −6.9
the value of τmax which gives the value of the residual red 000...444
component flux Ir= 1e−τmaxxadx at each velocity bin, (ii)
0
Predict each Ib valuRe using Ib = 1e−2τmaxxadx, and (iii)
0
ComparepredictedIb valuestotheRobservedvaluesandcal-
culate the corresponding χ2 value over the entire velocity
000...222
range. This procedure is repeated for all values of a, the
valueadoptedbeingtheonewhichminimisestheχ2calcula-
tioninstep(iii).Thisprovidesabestvalueofa=3.3+0.4,the
−0.1
correspondingpredictedblueprofileisshowninFig.5.Find-
ingNion requ1iresknowledge oftheaverage optical depthτ¯, ---000333555000000 ---333000000000 ---222555000000 ---222000000000 ---111555000000 ---111000000000
where τ¯= τ(x)dx.As in Fig. 2 and Fig. 3 of Arav et al.
0
(2005), theRdependence of the doublet component residual
intensitiesonτ¯aswellascomparisonswithcompletehomo-
Figure5.Inhomogeneousabsorbermodel(reddottedline)ofthe
geneous coverage and PPC models are shown in Fig. 6.
blue doublet component for a power-law index a=3.3. The solid
From Fig. 5 it is clear that, like the PPC model, an
blue line is the blue component Gaussian model and the dotted
inhomogeneous (power-law) absorber model is a good pre-
blue line is the observed spectrum in velocity range of the blue
dictor of the behaviour of the blue component in this par-
component, both of which have been interpolated onto the red
ticular doublet. As in Section 3.4, this is extended to the component velocity grid. Power-law model error bars are based
Gaussian models with doublet components treated as sin- on the errors of a. Blue triangles and red squares indicate the
gle points with flux averaged over all velocity bins showing points on the observed spectrum and inhomogeneous absorber
absorption. Interestingly, this leaves the upper bound on a modelrespectively.
unconstrainedforall absorption doubletsexceptabsorber4
at epoch 1 (see Table 5).
Given that the best-fit power-law index of the Gaus-
sian model for absorber 1 at epoch 2 is not consistent with
the measured value of a from the six velocity bins in the
observed spectrum, this analysis is considered inconclusive.
The constraints on the strongest doublet (absorber 1) sug-
gest that this feature has a comparatively large value of a. tion,x,approximatesastepfunction,andthuscannoteasily
Atlargevaluesofathedependenceofopticaldepthonposi- bedistinguished from a PPC model.
(cid:13)c 0000RAS,MNRAS000,1–17
A quasar showing remarkably variable BALs 9
spectral index (αox) between 2500 ˚A and 2 keV, and x-ray
1 spectral index (αx) from 2 keV to 100 keV. The αuv value
used is 0, which is within the range found by other studies,
for example Natali et al. (1998) found an average value of
αuv= 0.33 0.59. We adopt a value of αox = 1.98 calcu-
100 lated−using±themethodofWilkes et al.(1994)in−Equation5
0.8 Lo
αox =−1.53−0.11×log(cid:16)1030.5(cid:17) (5)
where Lo is the specific intensity at 2500 ˚A in the quasar
rest-frameinunitsofergs−1Hz−1.Thisistowardsthemore
0.6 negativeendoftherangefoundinGrupeet al.(2010),how-
10 ever tests with values of αox between a more typical value
of 1.6andtheadoptedvalueshownosignificantdifference
−
in Si iv and C iv column densities, so any discrepancies
between our adopted value and the true value should not
0.4 dramatically alter our results. According to Zdziarski et al.
(1996)thevalueofαxistypicallybetween 0.8and 1.0,so
− −
a value of 1.0 is adopted here for simplicity. It is possible
−
1 toestimateT fromtheratioofthebolometrictoEddington
luminosityandthemassoftheblackholeusingthemethod
0.2 of Bonning et al. (2007):
Tmax =105.56M8−14LLEbodld (6)
where Tmax is the maximum accretion disk temperature,
0 0.1 M8 is the black hole mass in units of 108M⊙, Lbol is the
0.1 1 10 100
bolometricluminosityandLEddistheEddingtonluminosity.
The value of LEdd can be found from the black hole mass
usingLEdd =1046.1M8.UsingestimatesofM8andLbolfrom
Figure6.Behaviourofdoubletcomponentswithrespecttoaver- Shenet al. (2011) gives a value of Tmax 240000 K, with
ageopticaldepthforthepower-lawmodeladopted(a=3.3).The the peak flux of the big blue bump occu∼rring at 2.1 1016
residual intensity of the blue component Ib (blue line) is found Hz( 140 ˚A or 87 eV). ×
bydoublingtheopticaldepthofthesimulatedredcomponentIr ∼A grid of models containing column density data for
(red line). The I2 value (magenta line) indicates what the blue
r Si iv and C iv is generated using the input continuum,
component would have been for a homogeneous absorber com-
pletelycoveringtheemissionsource.Thegreenlineis(1-C)fora with grid-points at hydrogen column densities log (NH /
PPCmodelgiventhevaluesofIr andIb ataspecificτ¯.Thera- cm−2)=21,22and23,hydrogennumberdensitieslog(nH /
tioof τPτ¯PC indicatedbytheblackline,whereτPPCistheoptical cm−3)=5,7and9andlogU valuesspanning−5.0to3.0in
depthofthePPCmodelpredictingthe(1−C)curve,showsthe intervals of 0.2 dex. These hydrogen number densities span
divergenceofthepower-lawmodelfromthePPCmodel. typicalnarrowlineregiontobroadlineregion(BLR)densi-
ties,themaximumdensitybeingclosetothecriticaldensity
ofCiii].Thedensityisunlikelytobehigherthanthisasthis
4 PHOTOIONIZATION SIMULATIONS wouldsuggesttheabsorberoriginatesfromwithintheBLR,
conflicting with the fact that individual absorption compo-
4.1 Cloudy Setup nentsarerelativelynarrowincomparisontothebroademis-
sionlines.Thisisalsoapparentfromtheneedforthedeepest
Modelling the column densities and ionization fractions of
troughs to be absorbing both the BEL and the continuum.
ionsinquasaroutflowsbyspecifyingtheshapeoftheioniz-
The ionization parameter range encompasses all reasonable
ing continuum and various gas (column) densities can pro-
values given the existence of the absorbers, while the hy-
vide insight into the physical location of the outflowing
drogen column density range was chosen to span a range
plasmaandthechangingnatureoftherelationshipbetween
above the minimum ionic column density for C iv, N ,
the ionizing photon flux and ionization state, especially if C IV
and below a density at which Thomson scattering becomes
ionization changes are the principal driver of absorption
linevariability.Photoionization modelsinthisinvestigation significant. This rules out column densities of log (NH /
cm−2)620 as these generate N values below the mini-
are performed using cloudy (Ferland et al. 1998), version C IV
mum valuecalculated from direct integration (Fig. 7).
c13.02.
The cloudy code combines these input parameters
withamodeloftheinputspectralenergydistribution(SED)
4.2 Application of Cloudy Models to Estimated
tocalculatethetotalionizingphotonflux.Itmodelsthein-
Column Densities
put SED using input parameters for temperature (T) and
power-law indices determining continuum behaviour of the Thecloudy modelscanbeapplied toresultsfrom thepre-
form Fν να, including optical to near-UV spectral index vioussectionsinthispapertoprovideparametersgoverning
∝
(αuv)foropticalwavelengthsabove2500˚A,UVtosoftx-ray the properties of the quasar continuum source which affect
(cid:13)c 0000RAS,MNRAS000,1–17
10 C. Wildy, M. R. Goad and J. T. Allen
Table6.TotalcolumndensitiesandtheirlimitsforSiivateach
epoch.
19 Epoch NSiIV NSiIV lowerlimit NSiIV upperlimit
(×1014 cm−2) (×1014 cm−2) (×1014 cm−2)
1 59.8 33.5 94.2
2 9.18 7.79 27.5
3 18.3 14.7 27.9
18
tion 4.1, limits for the column densities of Si iv and C iv
mustbefound.Sincetherearenoresolvable doubletsavail-
able for C iv, only the lower limits found from direct inte-
gration (see Section 3.2) can be used for this ion. ForSi iv,
17
thestrength ofthedoubletsat epoch 1meanstotalcolumn
densities can be estimated using the PPC model for all 5
absorbers as listed in Table 4. For epochs 2 and 3, PPC
can be used for absorbers 1 and 2 which are the predomi-
nant contributors to the total line of sight column density
at these epochs, plus a contribution from absorbers 3 to 5
16
usingthepeakopticaldepth(POD)methodoutlinedinSec-
tion 3.3 modified to take account of partial coverage. Here
we assume that the covering fractions of absorbers 3 to 5
at epochs 2 and 3 are the same as those at epoch 1. The
5 6 7 8 9
true optical depth can be found once the covering fraction
is known. Multiplying the original POD-derived N by
Si IV
thecoveringfractionandtheratiooftheintegratedrealop-
Figure 7. Predicted C iv column densities as a function of hy-
ticaldepthovertheintegratedapparentopticaldepthspan-
drogennumberdensityforgivenhydrogencolumndensities.Solid
ningthevelocityrangeoftheredcomponentofthedoublet,
squaresindicatethecloudyinputhydrogendensities,whichare
givesanestimateofthetruecolumndensity.TheN and
linked by solid lines of constant hydrogen column density. The Si IV
horizontal dashed line gives the minimum C iv column density theirupperandlowerlimitsareshowninTable6.Thetotal
predicted by the SDSS observation, which shows the strongest columndensitylowerlimitsareassumedtobethetotalsde-
absorption ofall epochs. It isclear that log(NH /cm−2)620 is rived from the POD method (listed in Section 3.3). Upper
ruledoutbythislimit. limits on the value of N at epoch 2 are comparatively
Si IV
large due to the large uncertainty in the covering fraction
ofabsorber2, howeverthebest valuelies towardsthelower
theoutflowinggas.Initiallyitisnecessarytofindthecorrect end of this interval.
normalisation of the continuum SED described in Section Using the cloudy output at each combination of
4.1, which can be used to find the total ionizing luminos- log NH and log nH, plots can be made of predicted ionic
ityLion andionizing photonemission rateQ(H)and hence column densities across the entire range of ionization pa-
thedistancefromthecontinuumsourcetotheinneredgeof rameter U.The span of log U coversthepeak of both Nion
theoutflowinggas(facingthesource) andthemassoutflow for all combinations, resulting in cases where there are two
rate. The correct ionising SED is found by scaling the out- possible regions in which the ionic column density values
put to the rest-frame corrected flux of the observed SDSS can be located. Given that only lower limits of the C iv
spectrum. Output parameters from the simulation include column density can be found, the column densities for this
the total ionizing photon flux φ(H) which can be used to ionareestimatedbylocatingthepositionsinlogU spaceof
findthetruevaluesoftheionisingluminosity,Lion,therate allowed NSi IV and using the equivalent values for C iv. If
of emission of Lyman continuum photons, Q(H) and the itisassumed thationizingcontinuumchangesarethemain
ionising source distance to theabsorbing gas, R. factor driving variability,it is possible toidentify therange
The calculated value of R will vary depending on of log U which satisfies the ionic column density limits at
the grid parameters (U,NH and nH); however since the each epoch. Although there are nine possible combinations
same SED is used in each case, values of ionizing lu- of log NH and log nH, only 2 are shown (Figures 8 and 9).
minosity and ionizing photon production rate are found ItisapparentfromtheseFiguresthatthelowerregionofal-
which are applied universally; they are Lion=4.17 1046 lowedlogU values(wherethegradientoflogNion vs.logU
erg s−1and Q(H)=9.80 1056 s−1. The elemental ×abun- ispositive)isinvalidastheallowedrangeofN forepoch
Si IV
×
dances used by cloudy are a solar composition de- 3 does not permit any N values above the lower limit.
C IV
rivedfrom Grevesse & Sauval(1998),Holweger (2001),and This was found to be true for all 9 density/column density
AllendePrieto et al. (2001, 2002). combinations, giving one continuous span of allowed log U
In order to apply the grid of models calculated in Sec- valuesin each case.
(cid:13)c 0000RAS,MNRAS000,1–17