The Origins of a Rich Absorption Line Complex in a Quasar at Redshift 3.45
aa r X i v : . [ a s t r o - ph . C O ] J u l Mon. Not. R. Astron. Soc. , 000–000 (0000) Printed 1 November 2018 (MN L A TEX style file v2.2)
The Origins of a Rich Absorption Line Complex in a Quasar atRedshift 3.45
Leah E. Simon ⋆ and Fred Hamann Department of Astronomy, University of Florida, 211 Bryant Space Science Center, Gainesville, FL 32611, USA
ABSTRACT
We discuss the nature and origin of a rich complex of narrow absorption lines in thequasar J102325.31+514251.0 at redshift 3.447. We measure nine C IV ( l − , and full widths at half minimumranging from 16 to 350 km s − . We also detect other absorption lines in these systems, in-cluding H I , C III , N V , O VI , and Si IV . Lower ionisation lines are not present, indicating agenerally high degree of ionisation in all nine systems. The total hydrogen column densitiesrange from . . to 10 . cm − . The tight grouping of these lines in the quasar spectrumsuggests that most or all of the absorbing regions are physically related. We examine severaldiagnostics to estimate more directly the location and origin of each absorber. Four of thesystems can be attributed to a quasar-driven outflow based on line profiles that are smoothand broad compared to thermal line widths and to the typical absorption lines formed in in-tergalactic gas or galaxy halos. Several systems also have other indicators of a quasar outfloworigin, including partial covering of the quasar emission source (e.g., in systems with speedstoo high for a starburst-driven flow), O VI column densities above 10 cm − and an apparentline-lock in C IV (in two of the narrow profile systems). A search for line variability yieldednull results, although with very poor constraints because the comparison spectra have muchlower resolution. Altogether (but not including the tentative line-lock) there is direct evidencefor 6 of the 9 systems forming in a quasar outflow. Consistent with a near-quasar origin, eightof the systems have metallicity values or lower limits in the range Z > − Z ⊙ . The low-est velocity system, which has an ambiguous location based on the diagnostics mentionedabove, also has the lowest metallicity, Z . Z ⊙ , and might form in a non-outflow envi-ronment farther from the quasar. Overall, however, this complex of narrow absorption linescan be identified with a highly structured, multi-component outflow from the quasar. The highmetallicities are similar to those derived for other quasars at similar redshifts and luminosities,and are consistent with evolution scenarios wherein quasars appear after the main episodes ofstar formation and metal enrichment in the host galaxies. Key words: quasars: general — quasars: individual — quasars: absorption lines — galaxies:evolution
Quasars represent episodes of rapid supermassive black hole(SMBH) growth and probably a unique period in the early evo-lution of galaxies. They may directly follow a major merger(P´erez-Gonz´alez et al. 2008; Hopkins et al. 2008) or a big blowoutof gas and dust. However, the nature of the relationship betweenSMBH growth and galaxy formation is not well understood. Feed-back from quasar outflows may play an important role in the evo-lution of this relationship. We are using narrow absorption lines ⋆ email: [email protected] (LES); [email protected] (FH) (NALs) in quasar spectra to study quasar outflows and environ-ments across a range of scales.NALs have full widths at half minimum (FWHMs) less thanseveral hundred km s − , and they appear in a variety of ultravio-let (UV) resonance transitions, including C IV l l V l l III l IV l l a l b l g l d l e l c (cid:13) L. E. Simon and F. Hamann and its extended environment (Ganguly et al. 2001; Vestergaard2003; Trump et al. 2006; Nestor et al. 2008; Wild et al. 2008;Gibson et al. 2008).The statistical excess of C IV NALs at velocity shifts v > − ,
000 km s − , where negative v indicates motion towards theobserver, indicates that many of these absorbers are directly re-lated to quasar environments (Weymann et al. 1979; Nestor et al.2008; Wild et al. 2008). The excess is largest at v & − − ,where roughly 80% of C IV systems with rest equivalent widthREW(1548 ˚A) > . − , Nestor et al. (2008) estimate that &
43% of C IV NALs with REW(1548 ˚A) > ∼ − ∼ I Lyman series, in order to obtain moreand better constraints on the absorber ionisations, column densi-ties and metal abundance. We are interested in using abundancesto discern rough star formation histories of quasar host galaxies inorder to make inferences about the relationship between the quasar,the growth of the central black hole, and the evolution of the hostgalaxy.Broad emission lines (BELs) have been used most of-ten to study quasar abundances. The most reliable results sug-gest metallicities of at least solar, and up to a few times so-lar, which requires significant previous star formation in thehost (Hamann & Ferland 1999; Hamann et al. 2002; Dietrich et al.2003; Warner et al. 2004; Nagao et al. 2006; Simon & Hamann2010). The metal-rich BEL result is true even for the highestredshifts studied, e.g. Pentericci et al. (2002); Jiang et al. (2007);Juarez et al. (2009), with redshifts out to 6.4. The most reliableresults based on BAL column densities suggest metallicity rangesbetween solar and ten times solar (Arav et al. 2001). Previous stud-ies of NALs in low redshift samples have found super-solar metal-licities and highly ionised gas, and have successfully probed sev-eral other NAL outflow characteristics (Hamann & Ferland 1999;Ganguly et al. 2003; D’Odorico et al. 2004; Hutsem´ekers et al.2004; Ganguly et al. 2006; Gabel et al. 2006). These lower redshiftsamples cover a wide range of luminosities, observed in the ultra- violet (UV) spectral range, where many useful metal and Hydrogenlines occur.NALs offer certain advantages in the study of metallicities andother gas characteristics in the near-quasar environment. Their nar-row widths mean the C IV doublets, separated by 500 km s − , areresolved. We use resolved absorption line doublets to disentanglesaturation effects, and to obtain accurate line optical depth and col-umn density measurements. NALs also form in a range of physicallocations, providing a more complete picture of the regions nearquasars. Because the NAL methods are completely independentfrom the BEL methods, requiring only column densities and ion-isation corrections, the NAL metallicities provide an independenttest of the BEL results.Here we present results for the particular luminous quasarJ102325.31+514251.0 (hereafter J1023+5142) at a redshift of z em = . IV NAL systems at velocities from -1400 to-6200 km s − . The density and diversity of lines in this complexmerits special attention. We will argue below that most (or all) ofthese systems form in a highly structured quasar–driven outflow.To interprete the metallicities and other data provided by theseNALs, we examine several diagnostics that can identify intrinsicNALs that form in quasar–driven outflows (Hamann et al. 1997;Barlow et al. 1997). In particular, 1) variability studies have foundintrinsic absorbers varying on relatively short timescales of monthsto years, providing strong evidence for these absorbers belongingto outflows either crossing the line of sight to the quasar or expe-riencing changing ionisation with the variations in the continuumemission (Hamann et al. 1997; Barlow et al. 1997; Aldcroft et al.1997; Narayanan et al. 2004; Misawa et al. 2007). 2) Detection ofpartial coverage of the background light source along the line ofsight strongly implies gas forming very near the source. This phe-nomenon occurs when the absorbing ’clouds’ are smaller than thebackground source, allowing part of the light from the source toreach the observer unabsorbed. This partial covering is easily de-tected in multiplets like the C IV doublet where the optical depth ra-tio between the two lines is fixed by the oscillator strengths. Whenthe source is partially covered, some light fills in the bottom ofthe absorption line, and makes the apparent optical depth ratio ap-pear different than the real optical depth ratio. 3) Outflow linestend to have profiles that are broad and smooth compared to ther-mal widths (Hamann & Ferland 1999; Srianand & Petitjean 2000;Ganguly et al. 2006; Schaye et al. 2007). In well studied NALs,these three indicators (variability, partial covering and broad pro-files) tend to appear together, which further increases the proba-bility that the occurrence of an individual indicator accurately pre-dicts an outflow origin very near the quasar for a given absorptionline (see also Hamann & Simon (2010) and references therein). Wealso note that super–solar metallicities are consistent with an in-trinsic origin for the gas. There are examples of high–metallicitygas in intervening systems, but not of low–metallicity intrinsic gas(Prochaska et al. 2006; Schaye et al. 2007).We describe the data acquisition and reduction in §
2, the iden-tification and fitting of the absorption lines in § § §
4. We discuss the argumentsfor the locations, probable intrinsic origins and quasar–driven out-flow properties of the gas in § § c (cid:13) , 000–000 arrow Absorption Line Complex at z = 3.45 Figure 1.
Region of the spectrum of J1023+5142 with C IV absorption. Individual C IV doublets are labelled by number. The lower x-axis is observedwavelength in Angstroms, while the upper x-axis is velocity shift of the shorter wavelength doublet line at 1548.20 ˚A from the quasar rest frame in kilometresper second. The flux units are normalised so that the continuum has a value of one. The gap between 6788 and 6797 ˚A is a gap between Echelle orders in thespectrograph. The longer wavelength line of system 8 falls in this gap. Strongly blended lines are considered components of a single system, e.g. systems 7and 9. We observed the quasar J1023+5142 on March 29, 2007 withthe Keck I HIRESr Echelle spectrograph as part of an observingcampaign to measure spectra of high redshift quasars with knownnarrow associated absorption lines. We used an 0”.86 wide slitfor a spectral resolution of R ∼ ,
000 or velocity resolution of ∼ − . Our data span the wavelength range from 3700 to8100 ˚A corresponding to 830 to 1820 ˚A in the quasar rest frame.This spectral range covers a variety of interesting lines, includingrest-frame Ly g
970 ˚A, C IV I Lyman series down to the Lyman limit at 912 ˚A. We use fourexposures totalling 2 hours on the source. The spectral region from ∼ ∼ a forest, by first averaging to-gether several adjacent spectral orders into a single spectrum. Then,we visually inspect the region for small sections of continuum notaffected by absorption or obvious noise spikes, and interpolate be-tween these sections, fitting the entire region with a low order poly- Figure 2.
Region of Ly a forest spectrum with the continuum fit over-plotted. The regions spanning the Ly b and O VI NALs are labelled abovethe spectrum. nomial. Our continuum fit for a region of the Ly a forest containingthe Ly b and O VI NALs is shown in Figure 2. The continuum place-ment has an uncertainty of ∼
10% in the forest and 2-3% at otherwavelengths.
The broad, flat shape of the emission features in the spectrum ofJ1023+5142 make an accurate emission redshift difficult to deter-mine. The redshift provided by the SDSS spectrum is z em = . IV l III ] l IV +O IV ] l c (cid:13) , 000–000 L. E. Simon and F. Hamann centroid respectively by -824, -730 and +36 km s − to correct forknown offsets from the nominal quasar redshift ([O III] l z em = . D z = . ∼ − − . These results and our efforts to measurethe inherently uncertain emission line centroids suggest that theuncertainty of the redshift measured by SDSS is not more than D z em .
02, corresponding to D v − − . We adopt theSDSS value throughout the remainder of this paper.We identify nine distinct C IV absorption line systems within6200 km s − of the quasar redshift. We will refer to these as sys-tems 1–9, as indicated in Figure 1 and Table 1 below. Other C IV systems are present at -16,800 and -33,800 km s − in the spec-trum, but they have narrow widths, complete covering, and blend-ing problems in the Ly a forest, which, along with their high veloc-ities, make them likely candidates for intervening gas and excludethem from further analysis in this work.After identifying the C IV doublets, we search the spectrumfor other common NALs such as Si IV , N V , C III , O VI , and H I Lyman series lines at the same redshift. We also search for lowerionisation species, such as C II and Si II, but find none. All ofthe systems, except possibly system 1, appear to have relativelyhigh ionisations based on the presence and absence of high and lowionisation species respectively. Each set of absorption lines at oneredshift is considered a system, as labelled in Figure 1. Several ofthese systems are blends of two or three components, which are notindividually labelled in the figure.Systems 1 and 2 (Figures 3 and 4) appear to be line-locked inC IV . The velocity offset between the l l < − ) com-pared to the FWHMs of these lines ( ∼
30 km s − ) and the velocityshifts from the quasar systemic, ∼ − ∼ − − .If this overlap between the C IV lines in systems 1 and 2 repre-sents a physical line-lock, where the velocities of the two systemsare actually separated by exactly their doublet separation, and nota chance alignment in the spectrum (see Ganguly et al. (2003) andBraun & Milgrom (1989) for full discussions of the possible physi-cal nature of line-locking), then it provides evidence for these linesforming in a quasar outflow driven by radiation pressure (see § We fit each NAL system with a Gaussian optical depth profile. Thenarrowest absorption lines are at least 1.5 times broader than thespectral resolution and the other lines are significantly broader thanthis. The absorption lines are, therefore, fully resolved, and suchGaussian optical depth profile fits are sufficient to determine accu-rate optical depths and covering fractions. The optical depths andcovering fractions are held constant across the width of each lineprofile. Gaussian fits are actually essential to distinguish individ-ual absorption features in the crowded Ly a forest, and also usefulto disentangle blended absorption in other areas of the spectrum.Furthermore, Gaussian fits smooth over noise spikes and large op-tical depth and covering fraction uncertainties in the wings of thelines. We also use Gaussian fits to simultaneously fit and lock to-gether various parameters including redshift, doppler b parameter,covering fraction and a 2:1 optical depth ratio based on oscillatorstrength ratios for doublets such as the C IV , Si IV , N V , and O VI .To measure accurate optical depths, we must consider the pos- sible effects of partial coverage of the emission source by the ab-sorbing gas. The line of sight covering fraction affects the observedline intensity as follows: I v = ( − C f ) I + C f I e − t v (1)where 0 C f I is the emitted (unabsorbed) intensity and I v and t vare the observed intensity and line optical depth at each velocityshift v. This equation assumes that the background light source hasa uniform brightness given by I and the foreground absorber ishomogeneous with a single value of t v. The viability of this as-sumption is discussed by Hamann & Sabra (2004) and Arav et al.(2005). We assume that all lines in a given multiplet have the sameC f at a given velocity. We do not explicitly attempt to distinguishbetween partial covering of the continuum source and of the BELregion as discussed by Ganguly et al. (1999). However, we estimatefrom the SDSS spectrum that the C IV BEL peaks 20% above thecontinuum, which implies that the BEL can only account for partialcovering of 0.8 or higher.We attempt to fit each system with the smallest possible num-ber of Gaussian components. This minimizes the number of freeparameters and provides a more robust characterization of columndensities, ionisations and abundances in absorbing regions whoseinternal velocities might be more complex than simple Gaussians(e.g., in outflows). We fit each absorption line with a single Gaus-sian unless 1) the system clearly has multiple components distin-guished by inflection points that stand out significantly above thenoise fluctuations in the spectrum (e.g. system 7), or 2) a singleGaussian would miss a significant fraction, & ∼
25% threshold is somewhatarbitrary, but it ensures that we achieve a good fit to the observedline and that significant portions of absorption (i.e., large enoughto change the column density measurements) are not missed. Forthese exceptional cases we use the minimum number of Gaussianspossible to achieve an accurate fit to the data. If a system is fit withtwo or more Gaussians, each Gaussian is labelled as a component.We assume that the covering fraction is the same for all componentsin a given system, such that the optical depths in Equation 1 simplyadd together in regions of component overlap (see Hamann et al. inpreparation for further discussion). This simplifying assumption iswell justified by the excellent fits to all the systems, with the possi-ble exception of system 6, which we discuss in more detail in § § t v and C f across the line profiles for several systems with eithernon-Gaussian profiles (System 8) or C f < C f , central opticaldepth, t , the Doppler b parameter, and redshift for each compo-nent in each system. We fit the C IV absorption lines first, thenbase the fit for other absorption lines on the C IV fit parameters.To ensure we are analysing the same gas in different ions, we fixthe redshift for all absorption lines in a system to the C IV redshift.We further exclude unwanted contributions from blends or complexmulti-phase gas by fixing the b parameter of all ions with ionisa-tions lower than C IV to that of C IV . Higher ionisation lines, suchas N V and O VI , are allowed to be broader. However, we cap the b -values of the H I profiles at 140% of C IV . This cap is importantfor the abundance analysis below ( § c (cid:13) , 000–000 arrow Absorption Line Complex at z = 3.45 derived H I column densities do not include gas with dramaticallydifferent kinematics than C IV . We choose to cap the H I b -valuesat 140% of the C IV b -values instead of the much higher percent-age expected for purely thermal broadening because the widths ofthe C IV lines exceed the thermal widths expected for a gas pho-toionised by either the quasar or the inter-galactic UV spectrum.Therefore, we assume the b -values are dominated by non-thermalbroadening effects. On the other hand, setting the cap at 140% in-stead of something smaller, such as 100%, allows for some contri-bution of thermal broadening to b in the narrower systems (whichwould affect H I more than C IV ). Overall, our fits to the Lymanlines should lead to reasonable but generously large estimates ofthe amount of H I gas that coexists with C IV , and therefore, toconservatively low estimates of the C/H abundance.As stated above, the covering fraction is a free parameter inthe Gaussian optical depth fits of each doublet. In cases where thebest fit profile has C f <
1, we repeat the fit with C f = C f < C f match the data. Incases where the fits are comparable, we assume C f =
1, otherwise,the best fit is chosen. For example, we confirm that the C f < C f = IV and N V . We assume C f = IV doublet coveringfraction. This is necessary because the observed line ratios withinthe Lyman series are too severely affected by blending in the Ly a forest to yield their own independent measures of C f .Table 1 lists fit parameters for all of the absorption lines thatyield useful constraints for the ionisation and abundance analysisdescribed in § VI doublet is listed, as O VI is neverused in the abundance analysis because of either line saturation orstrong blending in these lines in all systems. However, the strengthof O VI is still useful as an indicator of the ionisation of the gas.In systems where N V is not present, we list upper limits for thestronger member of the N V doublet for completeness. Table 1 liststhe central wavelength ( l ) and doppler b parameter values alongwith column densities and rest equivalent widths (REW) derivedfrom the Gaussian optical depth profile fits. Systems 4 and 7 eachhave two blended components. The values of l , b , and log N arelisted separately for these components in Table 1, but the REWs,listed only with the first component data, apply to the entire blend.We measure upper limits on H I column densities in all cases whereall the Lyman series absorption lines are blended with interveningabsorption lines in the Ly a forest. The same is true for singlet ionswith upper limits on the column densities.We estimate uncertainties for the column densities by placingthe continuum at the top and bottom of the noise around the fittedcontinuum, corresponding to the reasonable maximum/minimumvalues ( ∼ s uncertainties) for continuum placement. We mea-sure 3 s uncertainties for the H I column densities of 0.18 dexon average. The covering fraction is C f = C f . However, these uncertaintiesare much smaller than the informal uncertainties, which are domi- Table 1.
Individual absorption lines. abs ID l rest l REW b log N v (km s − ) ˚A ˚A km s − cm − g
973 4307.03 0.162 27.8 15.05-1442 C III 977 4326.88 0.076 10.8 a b a e
938 4143.82 0.011 13.9 g
973 4297.29 0.035 13.9 . a e
938 4142.98 0.059 15.4 g
973 4296.43 0.164 15.4 a e
938 4142.72 * 46.8 g
973 4296.15 * 46.8 a nated by uncertainty in continuum placement. We estimate cover-ing fraction uncertainties due to continuum placement uncertain-ties by first shifting the continuum near each red doublet mem-ber up and down by the 3 s continuum uncertainty, and then fittingthe doublets with this new continuum. The actual uncertainties areprobably smaller than the uncertainties we derive in this way, be-cause a similar shift in the continuum around both doublet members(a more likely occurrence) produces smaller changes in C f . We find C f = . ± .
15 for system 5 and C f = . ± .
20 for system 6. Ifwe fix the covering fraction in systems 5 and 6 at C f = C f < IV and N V doubletmembers based on the intensity of the shorter wavelength member,combined with the 2:1 t -ratio derived from the oscillator strengths c (cid:13) , 000–000 L. E. Simon and F. Hamann
Table 1. Continued5 a a b g
973 4281.07 0.055 31.7 14.41-3254 Ly b a d
950 4178.26 0.008 27.3 a d
950 4177.31 * 60.0 b a b a g
973 4239.50 0.039 58.4 b a a System 5 has covering fractions C f ( H I ) = . C f ( N V ) = . C f ( C IV ) = . b System 6 has covering fractions C f ( H I ) = . C f ( N V ) = . C f ( C IV ) = . of each line. The predicted shorter wavelength member will onlymatch the data if C f =
1. These predictions are shown in Figure 10.The observed data for both doublet members are plotted with solidcurves and the predicted longer wavelength doublet member is plot-ted with a dot-dashed curve. The predicted shape of the longerwavelength member of C IV and especially of N V is much weakerthan the observed shape for system 5. In system 6, the line centresof C IV , and more clearly N V , are stronger in the observed data forthe longer wavelength members than in the predictions. We con- clude that C f < t -ratio analysis that covering fractions are notalways constant across a single line profile. To better account forthis, and to determine if the guassian fits find reasonable averagevalues for C f , particularly for lines with very non-Gaussian shapes,we use a point-by-point method in addition to the Gaussian fittingmethod to determine t v and C f across the line profiles in severalsystems. We fit systems 5 and 6, the two systems with C f < C f and t v at each step. The point-by-point fitsare shown in Figures 11 and 12. The solid curve shows the shorterwavelength doublet member, while the dot-dashed curve shows thelonger wavelength doublet member. The covering fraction at eachpoint is represented by the filled circles. The steps used for sys-tem 6 are three resolution elements wide, which is wide enoughto smooth over the noise but narrow enough to avoid blending thewings and core of the line. However, system 5 is narrow enoughthat using bins three resolution elements wide, or wider, across thewings of the line would blend too much information from the coreand the continuum. Furthermore, the spectrograph resolution couldbe blending the covering fraction in the wings of the line with thecontinuum. Thus, for the narrow system 5, we measure only the 3resolution element bin at the line core. The step size for system 8is four resolution elements. This larger step size further smoothsover noise, and can be used because the line is much broader thanthe other systems, lessening the impact of blending of the core andwings of the line. We derive formal covering fraction uncertain-ties ( s C f ), represented as error bars at each point in Figures 11 and12. The average C IV and N V central covering fraction from thepoint-by-point method matches the C IV and N V covering fractionderived from the Gaussian fitting method to within 10% in systems5 and 6.Based on this result, the results of the t -derived doublet ra-tio analysis, and uncertainties derived from the Gaussian fittingmethod, we are confident of the accuracy of the C f < IV and N V column densities found in system 6 using the point-by-point method match the C IV and N V column densities derivedfrom the Gaussian fitting method to within 0.14 dex and 0.06 dexrespectively. The same comparison for N V in system 8 yields adifference of 0.18 dex between the two methods. We conclude thatthe Gaussian technique is sufficient for comparing different sys-tems and generally provides accurate column density and coveringfraction results for the purposes of this work. The abundance ratios can be derived from the ratio of measured col-umn densities corrected for the degree of ionisation in the gas. Forexample, the relative carbon to hydrogen abundance normalised tosolar is given by: (cid:20) CH (cid:21) = log (cid:18) N ( C IV ) N ( HI ) (cid:19) + log (cid:18) f ( HI ) f ( C IV ) (cid:19) + (cid:20) HC (cid:21) ⊙ (2) c (cid:13) , 000–000 arrow Absorption Line Complex at z = 3.45 where f is the ionisation fraction of a given ion, N is the columndensity and the final term on the right-hand side is the logarithmicsolar abundance ratio of hydrogen to carbon listed in Grevesse et al.(2007). The second term on the right is the ionisation correc-tion (IC). These correction factors can be large when comparinga highly ionised metal like C IV to H I . The exact values dependon the ionisation mechanism. Photoionisation by the quasar spec-trum is by far the most likely scenario based on the arguments in § . The calculationsalso assume that the absorbing gas is optically thin in the Lymancontinuum, which is appropriate for the column densities we mea-sure in the absorption lines of J1023+5142 (Table 1).Ideally, we would constrain the absorber ionisations by com-paring the ratios of observed column densities in different ions ofthe same element, such as N ( C III ) / N ( C IV ) or N ( N III ) / N ( N V ) ,to the theoretical results in Hamann et al. (2010). However, theseconstraints are only marginally usable in our data because N ( C III ) and N ( N III ) are always blended in the Ly a forest and are there-fore only ever constrained as upper limits. Therefore we estimatethe IC from ratios such as N ( N V ) / N ( C IV ) or N ( Si IV ) / N ( C IV ) ,with the additional assumption that the relative metal abundancesare approximately solar. The specific ionisation constraints used foreach system sometimes lead to upper limits, lower limits or specificvalues for the abundance ratios, and are described in more detailfor individual systems in § C / H abundances based on these constraints are all super–solar, except insystem 1. Table 2, which contains several different abundance indi-cators for each NAL system, lists these estimated ’best’ abundancesin column 3, titled [ C / H ] best , for the nine systems.We also calculate robust lower limits on the metal to hydro-gen abundance ratios by applying minimum values of the ionisa-tion correction (IC min , Hamann et al. (1997)) to the measured C IV ,Si IV and N V column densities, when available. Each metal ionhas a unique global IC min that occurs near the peak of its ownionisation fraction. For example, f ( H I ) / f ( C IV ) peaks approxi-mately where f ( C IV ) is largest. We use the values of IC min listedin Hamann et al. (2010). Applying these minimum correction fac-tors to the observed column density ratios (Equation 2) leads tothe firm lower limits listed for [ C / H ] min , [ Si / H ] min and [ N / H ] min abundances in columns 4–6 of Table 2. The minimum ionisationcorrections provide firm lower limits on the abundances that do notdepend on the ionisation uncertainties or the possibility of a multi–phase gas. In particular, any gas components not at an ionisationcorresponding to IC min would have the effect of raising the actualvalue of IC and thus also the actual abundance. The calculations in Hamann et al. (2010) apply to gas that is pho-toionised by a typical quasar spectrum. We perform additional CLOUDY(Ferland et al. 1998) calculations using the inter-galactic background spec-trum in CLOUDY, which is based on Haardt & Madau (2005, private com-munication). We find that the ionisation fractions of interest in the presentwork have only negligible differences between the two calculations, e.g.,compared to uncertainties in the measured quantities or derived ionisationconstraints. Therefore, our analysis of the ionisation and abundances inJ1023+5142 should apply whether the absorbers are located near the quasaror outside the quasar’s radiative sphere of influence.
Table 2.
Metal Abundance and Total H Column Density abs [C/H] best [C/H] min [N/H] min [Si/H] min log(N(H)) cm − -0.47 > -2.25 – – 17.622 3.42133 > +0.32 > -0.76 > -0.47 – > +0.38 > -0.82 – > -0.34 > +0.79 > -1.32 – > -0.66 > +0.02 > -0.10 – 18.216 3.40196 +0.50 > -0.58 > -0.32 – 19.137 3.39889 > +0.40 > -0.30 > +0.01 – > +0.94 > -0.07 > -0.17 – > +0.14 > -0.86 > +0.11 – We derive total H column densities for each NAL system fromthe H I column densities listed in column 7 of Table 1 and the bestionisation correction described above. We uselog N ( H ) = log N ( H I ) − log f ( H I ) , (3)where log N ( H ) is the total H column density, log N ( H I ) is thecolumn density of H I and log f ( H I ) is the H I fraction used toobtain IC. log N ( H ) for each system is listed in column 7 of Table 2.The uncertainties in these results are dominated by uncertain-ties in the IC. In addition to the limited constraints provided by thedata, a few well-studied cases have shown that individual absorberscan span a range of ionisations and have a range of IC values (e.g.Hamann et al. (1997)). We assume a single ionisation state for eachabsorption line system. We discuss the individual systems briefly in § System 1, v = − km s − The C IV in system 1 appears line-locked with the C IV in sys-tem 2, as discussed further in § VI is not present, or is veryweak, implying that the ionisation is low. Further evidence for lowionisation is the weak C IV combined with strong H I measured inLy a and Lyman g , as seen in Figure 3. Our best ionisation con-straint comes from an upper limit on C III , which means the best C / H abundance is an upper limit as well. This gas is a likely can-didate for host galaxy halo gas based on the weakness of the metallines and the low abundances. System 2, v = − km s − The C IV in system 2 appears to be line-locked with system 1,as mentioned above and discussed further in § I could beshifted to a lower velocity by as much as 30 km s − from the metallines in this system, indicating a multi–phase gas, but heavy blend-ing obscures the precise shift of the lines as can be seen in Figure 4.The Ly a absorption line is poorly constrained. The resulting H I optical depth and doppler b parameter are upper limits, resulting inlower limits for the best estimate of C / H abundance. We constrainthe ionisation by the relative strengths of C IV and N V , assumingsolar abundance ratios. System 3, v = − km s − The H I lines in system 3 are blended with those from system4, but appear consistent with the metal lines, shown in Figure 5.Because of the relatively poor constraints on the H I absorptionlines, the H I optical depth and doppler b parameter are upper limits,resulting in lower limits for the best estimate of C / H abundance. c (cid:13) , 000–000 L. E. Simon and F. Hamann
Figure 3.
Line profiles in the normalised spectrum J1023+5142 for system1. The central velocity for the Gaussian profile fit is v = − − .The velocity scale is with respect to the rest frame of the quasar based on z em = .
45, where negative velocities denote motion towards the observerand away from the quasar. The velocity range is 400 km s − for this andfigures 4 through 7 and 9. The solid curve in each panel is the Gaussianoptical depth fit to individual lines. The dashed vertical line is the centralvelocity of the system. All of the lines used to derive or constrain columndensities with Gaussian fits are shown in the figure. We constrain the ionisation by the relative strengths of C IV andSi IV , assuming solar abundance ratios. System 4, v = − , − km s − The C IV and Si IV doublets in system 4 are fit with twoblended Gaussian components to accommodate the asymmetricprofile. We use the central velocity of each component to identifythe system. The H I absorption lines are poorly constrained due toblending with system 3. The resulting H I optical depth and doppler b parameter are upper limits, resulting in lower limits for the bestestimate of C / H abundance. We constrain the ionisation by the rel-ative strengths of C IV and Si IV , assuming solar abundance ratios.This system is broad and asymmetric, which is indicative of a windor outflow feature (see § System 5, v = − km s − The H I in system 5 is well constrained by Ly a . The O VI is strongly blended with that of system 6 as shown in Figure 6.The covering fraction in the doublet is ∼ .
7. The C f = . Figure 4.
Line profiles in the normalised spectrum J1023+5142 for system2. The central velocity for the Gaussian profile fit is v = − − .The symbols and ranges are the same as in Figure 3. seen in Figure 6 in the longer wavelength members of the C IV andN V doublets, which have a much sharper central feature than theirshorter wavelength counterparts. We constrain the ionisation by therelative strengths of C IV and N V , assuming solar abundance ratios.The partial coverage in this system indicates that it is intrinsic to thequasar. The partial coverage in this system and in system 6 are ex-amined qualitatively with the t -ratio predicted doublets, shown inFigure 10, and further with the point-by-point analysis, illustratedin Figure 11. Both analyses confirm similar C f < § System 6, v = − km s − H I is well-constrained in system 6, with three mostly blend-free Lyman lines. The covering fraction in the doublets is C f ∼ . C f < VI lines are blended withthe O VI lines in system 5. The covering fraction in H I appears tobe C f = a reaches zero intensity. We constrain the c (cid:13) , 000–000 arrow Absorption Line Complex at z = 3.45 Figure 5.
Line profiles in the normalised spectrum J1023+5142 for sys-tems 3 and 4. The central velocities of the Gaussian profile fits are v = − − for system 3 and v = − − for system 4. The sym-bols and ranges are the same as in Figure 3. ionisation by the relative strengths of C IV and N V , assuming solarabundance ratios. The broad smooth shape, along with the partialcoverage indicate that this system is part of an outflow.This system appears somewhat asymmetric and the t -ratioanalysis in Figure 10 suggests further that there may be two com-ponents, one with partial covering near the line-centre, and a sec-ond broader component with complete covering in the blue wing.Although one component does not provide the best possible fitto all the lines in system 6, it is not clear that adding a seconddistinct component would provide a better characterisation of theactual conditions in the absorber. We test this by fitting the sys-tem with one and two Gaussian components, where the two com-ponent fit still assumes the same covering fraction in both com-ponents. Both fits produce similar column densities in all ions, D N ( C IV ) = .
15 dex, D N ( N V ) = . D N ( H I ) = . Figure 6.
Line profiles in the normalised spectrum J1023+5142 for sys-tems 5 and 6. The central velocities of the Gaussian profile fits are v = − − for system 5 and v = − − for system 6. The solidcurve is the C f < IV , N V and H I absorption lines.System 5 is the narrower system. The ranges are the same as in Figure 3. fits. Also, by using the single Gaussian fit, we ignore parts of theLy a absorption which do not correspond directly to C IV absorb-ing gas, and therefore retain the ability to directly compare H I andC IV column densities for the abundance analysis. System 7, v = − , km s − The H I column density is constrained as an upper limit in sys-tem 7 because of blending in the Lyman lines, shown in Figure 7, c (cid:13) , 000–000 L. E. Simon and F. Hamann
Figure 7.
Line profiles in the normalised spectrum J1023+5142 for sys-tem 7. This system has two blended components with central velocities ofv = − − . The symbols and ranges are the same as inFigure 3. resulting in lower limits for the best estimate of C / H abundance.We fit this broad system with two Gaussian components to bettermatch the absorber shapes, and identify the system by the centralvelocities of the two components. The ionisation is constrained bythe relative strengths of C IV and N V , assuming solar abundanceratios. Figure 8.
Line profiles in the normalised spectrum J1023+5142 for system8. The Gaussian profile fit has a central velocity of v = − − . Thevelocity range is 1100 km s − . The symbols and ranges are the same as inFigure 3. System 8, v = − km s − The longer wavelength member of the C IV doublet in system8 falls on a gap between orders of the spectrograph between 6785and 6795 ˚A, but the N V doublet is present in the spectrum, as is theshorter wavelength member of the C IV doublet. The N V doubletis used to determine the C f and the Doppler b parameter for bothdoublets. This system is almost broad enough to be a mini-BAL,and is likely an outflow system based on the shape and strength ofthe line profile, shown in Figure 8. The H I appears to be relativelyweak in this system compared to the metal lines, although there issevere blending in the Ly a forest. This blending means the H I absorption is poorly constrained with an upper limit, and thereforethe best estimate for C / H abundance is a lower limit. We constrainthe ionisation by the relative strengths of C IV and N V , assumingsolar abundance ratios.We use the Gaussian fit to compare system 8 to other systems,but the profile of system 8 is distinctly non-Gaussian. Therefore wealso fit the central trough of the line with a point-by-point analysis,shown in Figure 12. The C / H abundance found by the Gaussian c (cid:13) , 000–000 arrow Absorption Line Complex at z = 3.45 Figure 9.
Line profiles in the normalised spectrum J1023+5142 for sys-tem 9. This system is a blend of three components with central velocities,v = − − . Although all three components arefit with Gaussian profiles, only the central component is considered in theabundance analysis. The symbols and ranges are the same as in Figure 3. fit is consistent within 10% of the C / H abundance found using thepoint-by-point method. System 9, v = − , − , − km s − System 9 has three components, but we chose to analyse onlythe central component for abundances, as the two outer components
Figure 10. t -predicted line profiles for systems 5 and 6. System 5 is the nar-rower system. The dot-dashed curve shows the smoothed C IV and N V pre-dicted long wavelength doublet member, based on doublet optical depth ra-tio from short wavelength member, assuming C f =
1. The actual smootheddata for the shorter and longer wavelength doublets are shown as the boldand thin solid curves, respectively. The longer wavelength data are strongerthan the predictions, indicating partial covering, especially for the N V dou-blet in system 5 and in the centre of the N V doublet in system 6. are very poorly constrained, as shown in Figure 9. This system hasthe highest velocity shift out of the group of narrow absorptionlines, and lies just nominally outside of the velocity shift regionfor associated lines (v > − − ), at ∼ − − . TheLyman lines could be shifted up to 20 km s − from the metal lines,indicating a possible multiphase gas, but the line are too weak to de-termine their precise centroids. The weakness of the Lyman lines,along with blending in the Ly a forest mean the H I column densi-ties are upper limits, so the best estimate of the C / H abundance isa lower limit. We constrain the ionisation by the relative strengthsof C IV and N V , assuming solar abundance ratios. J1023+5142 has nine NAL systems with a range of column den-sities from N ( H ) . to 10 . cm − , velocities from –1400to –6200 km s − , C IV doppler b values from 7 to 150 km s − ,C IV REW(1548 ˚A) from 0.02 to 0.81 ˚A and two systems with par-tial covering of either the continuum source or the broad emissionline region (BLR), C f ≈ .
7, which imply absorber diameters of .
03 pc or . § IV REW integrated across the doublet,with completeness limits of 0.3–0.5 ˚A. The NAL systems all ap-pear to be highly ionised; none of the systems exhibit low ionisa-tion species such as Si II, C II or Si III, whereas all contain C IV and some contain higher ionisation species such as O VI and N V . c (cid:13) , 000–000 L. E. Simon and F. Hamann
Figure 11.
Point-by-point covering fractions for C IV and N V in system6 and the centre of system 5 with step size of three resolution elements.System 5 is the narrower system. The solid curve is the smoothed shorterwavelength line, the dashed curve is the smoothed longer wavelength line,with their respective error spectra below. The circles represent 1 − C f ateach step so that a point at zero flux has complete coverage, and a point atthe continuum flux of one has no coverage. The circles are located at thecentre of the average velocity steps. Figure 12.
Point-by-point covering fractions for N V in system 8 with stepsize of four resolution elements. The symbols are the same as in Figure 11. Systems 5, 6, 8 and 9 exhibit high ionisation (O VI ) absorption, andothers may also have absorption at these wavelengths that is notobservable due to blending in the Ly a forest. Systems 2–9 exhibitsupersolar metallicities ranging from Z > > Z ⊙ . System 1has a slightly lower metallicity of Z . Z ⊙ . We examine severaldiagnostics to estimate directly the location of each system. The tight grouping and similar high metallicities (see § IV absorption line systems in J1023+5142 suggest a possible physi-cal connection between the absorbers. The proximity of this NALcomplex to the quasar redshift suggests further that the physical re-lationship includes the quasar itself. The velocity span across thegroup is too large to be explained by a single galaxy or even a largecluster of galaxies. It might be consistent with some larger cosmicstructure connected to the quasar, but then we would expect the ve-locity distribution to include the red side of the quasar systemic.A more likely explanation is that the NAL complex formed in amulti-component outflow from the quasar.There are several indirect arguments for an intrinsic origin forthe gas in this NAL complex. i) 8 of the 9 systems have super–solar metallicities, discussed in detail in § VI absorption may indicate intrinsicgas near the quasar. Fox et al. (2008) carry out a detailed study ofO VI absorption in 2 < z < N ( O VI ) > . − ) indicates an intrinsic origin, supported by evi-dence for partial covering in most of these systems. We measure N ( O VI ) in four systems in J1023+5142. Two of them (6 and 8) areabove the intrinsic thereshold defined by Fox et al., while the othertwo (5 and 9) are very near this threshold at log N ( O VI ) > . VI , that is systems 5, 6, 8 and 9, alsohave strong N V compared to C IV . Strong N V , especially com-pared to C IV , is often (though not always) present in intrin-sic gas (e.g. Weymann et al. (1981); Hartquist & Snijders (1982);Hamann et al. (1997); Kuraszkiewicz & Green (2002); Fox et al.(2008)). iv) The presence of strong O VI and N V , especially withthe absence of low ionisation species such as C II in these absorp-tion systems is consistent with gas exposed to the intense ionisingradiation field near a quasar.If the NALs in J1023+5142 are intrinsic to the quasar en-vironment, the most likely origin is in a quasar–driven out-flow. Other possible intrinsic origins all have lower velocities: i)starburst-driven outflows typically have 100 < v < − (Heckman et al. 2000), and in Seyfert galaxies have maximumoutflow speeds of 600 km s − , and more typical speeds of 100-200 km s − (Rupke et al. 2005), ii) other galactic/halo gas shouldhave velocities near the typical velocity dispersion for such galax-ies ( s ∼
300 km s − ), iii) gas in the narrow line region of thequasar has typical velocities of v − , and maxi-mum velocities of v − (Ruiz et al. 2001, 2005;Veilleux et al. 2005), and iv) intra-cluster galaxy motions areshown by Popesso & Biviano (2006) to generally have velocity dis-persions s v < − or less for clusters with higher num-bers of active galactic nuclei (Richards et al. 1999; Heckman et al.2000; Vestergaard 2003; Nestor et al. 2008).Statistical studies, Nestor et al. (2008) (see also Wild et al.(2008)), have shown that >
43% of NALs at − > v > − − with REW(1548 ˚A) > . ∼
57% for the nar-rower range of − > v > − − , spanned by theNALs in J1023+5142. The percentage reaches ∼
72% for the nar-row range of − > v > − − , which encompassessystems 1 through 4 in J1023+5142. These percentages are proba-bly lower for weaker lines (Nestor et al. (2008) and private com-munication). Misawa et al. (2007) also find that for C IV NALswith REW(1548 ˚A) > .
056 ˚A at velocities v < − the intrinsic (outflow) fraction is >
33% and at higher velocities,5000 < v < − , the intrinsic fraction is > − c (cid:13) , 000–000 arrow Absorption Line Complex at z = 3.45 We search for direct signatures of quasar outflow origin via1) line variability, 2) partial covering and 3) broad profiles (seeHamann et al. (2010); Hamann & Simon (2010) and referencestherein for more discussion).1) We have only very poor constraints on the variability. Wecompare C IV and N V REW results measured from the guassianfits to the SDSS and Keck spectra ( D t rest ≈
11 months) in searchof variability in the absorption lines. System 8 is the only individ-ual C IV and N V system resolved in the SDSS spectrum, whilethe weaker lines are not detected in the SDSS spectrum. System8 is the strongest of the nine systems, and did not vary in REWby more than 15% in C IV and N V between the SDSS and Keckobservations. For the eight weaker systems, we conclude only thatvariability greater than a factor of 2 to 3 did not occur.2) There is partial covering in two (systems 5 and 6, Figures 6and 11) out of the nine systems. Absorption lines with partial cover-ing of the luminosity source are attributed to gas near the quasar be-cause partial covering is not expected to occur in intervening cloudsor galaxies (Hamann et al. 2010). The presence of partial coveringin these lines strongly suggests that the gas is intrinsic and locatedin the near quasar environment.3) The profiles of systems 4, 6, 7, 8, and 9 shown in Fig-ures 5, 6, 7, 8 and 9 have C IV and N V b values between 33and 155 km s − and O VI b values between 42 and 150 km s − .These b values are broad and smooth compared to the ther-mal widths for gas at the highest expected temperature near T = K (Arnaud & Rothenflug 1985; Hamann et al. 1995)for a photoionised gas near a quasar (33 km s − for H and lessthan 10 km s − for C and N). They are also broader than typi-cal non-damped Ly a intervening C IV , N V and O VI absorp-tion lines, which have on average b < −
14 km s − for O VI ,and b < −
12 km s − for C IV and N V (Tzanavaris & Carswell2003; Bergeron & Herbert-Fort 2005; Schaye et al. 2007; Fox et al.2008). These profiles, therefore, exhibit morphologies consistentwith formation in an outflow.These three characteristics, variability, partial covering, andbroad profiles are often found together in a single object, furthersupporting the idea that each individual characteristic likely indi-cates an outflow. One well studied NAL outflow in J2123-0050(Hamann et al. 2010) is a prime example of all three, exhibit-ing variability, partial covering, and broad profiles that still haveFWHMs that are as narrow or narrower than many of the systemsin J1023+5142.There is more tentative evidence for a quasar outflow originin the apparent line–lock between the C IV doublets in systems 1and 2. Line–locking, where the difference in outflow velocities oftwo systems is exactly the velocity separation of the doublet, meansthat the lines are being radiatively accelerated directly towards theobserver. The reality of the line–locking in this case is unclear, dueto the difference in derived metallicities between the two systems.Nevertheless, the incredibly small velocity offset ( § IV in systems 1 and 2 suggests that they are both part ofan outflow and that these weak C IV lines play a significant role inradiatively driving the flow. If this is really the case here, the gasprobably originated near the source of radiative acceleration, i.e.the quasar.Finally, we note a trend in line width with velocity shift awayfrom the quasar. The narrowest lines, with FWHM ∼
20 km s − are closest to the quasar redshift. The lines appear progressivelybroader as the velocity shift increases, with the broadest system de- scribed as a (narrow) mini-BAL with FWHM =
270 km s − , shownin Figure 1. A similar phenomenon has been observed before inother quasars with multiple C IV absorption lines clearly forming inoutflows (Hamann et al. 1997; Steidel 1990; Hamann et al. 2010).Although it provides no direct information on the absorber loca-tions, the appearance of this pattern in J1023+5142 supports theidea that at least some of the systems form in a quasar outflow. Thetight grouping of all nine of the systems also suggests a relationshipbetween them. Ganguly et al. (2003) determine that the probabilityof six similarly grouped NALs in the quasar RX J1230.8+0115 allforming in intervening (uncorrelated) gas at similar velocity shiftsis extremely small. The similarities between those NALs and theNAL complex in J1023+1542 implies a similarly small probabilityfor all nine NALs in this complex forming independently in inter-vening gas. Although there could be up to several interlopers in theNAL complex of J1023+5142 that might form in nearby galaxiesin the line of sight, the density of these galaxies required to formall of the absorbers in the complex is beyond any expectations ofcluster density at this redshift.Overall, we conclude that at least six out of the nine systemsoriginate in a highly structured outflow driven by the quasar, be-cause they exhibit one or more of the following properties: par-tial covering, broad profile shapes, large line strengths, tight group-ing with other systems, and proximity to the quasar redshift. Sys-tems 5 and 6 are the most likely outflow candidates because theyexhibit partial covering as well as several of the other propertieslisted above. Systems 4 and 8 are likely outflows because of theirstrong, broad, asymmetric and smooth profiles and systems 7 and 9are probably outflows because of their broad and smooth shapes.Systems 1 through 3 are more ambiguous in origin, with nar-row widths, complete covering, lower velocity shifts and smallerstrengths. However systems 1 and 2 exhibit line-locking, whichcould be evidence of an outflow. Ultimately, we find strong evi-dence that systems 4, 5, 6, 7, 8 and 9 are part of a quasar outflow,whereas systems 1, 2, and 3 could consist of intervening gas fromthe IGM or other galaxies in the line of sight. As described in § − . Several of these systems (4, 6, 7 and 8) also havesuper–thermal line widths, indicative of large turbulence or strongradial velocity sheer across the outflow structure.Some of the outflow structures, represented by systems 5 and6, must be spatially small to produce partial covering of the back-ground emission source. These lines lie on top of the very weakC IV broad emission line. Therefore, nearly all ( > C f = .
80 can be ascribed to the continuum source and notthe much larger BLR. We estimate the diameter of the accretiondisk continuum source at 1550 ˚A to be d ≈ .
03 pc and the diam-eter of the C IV broad line region to be d ≈ . in Hamann & Simon (2010). To partially cover We estimate the luminosity from the rest–frame flux at 1450 ˚A measuredin the SDSS spectrum, which, combined with the luminosity distance andthe bolometric correction factor L=3.4* n L n (1450), gives l L l (1450 ˚A). Wec (cid:13) , 000–000 L. E. Simon and F. Hamann the emission source, the absorbing clouds should have characteris-tic sizes similar to or less than the BLR diameter, and possibly evenless than the accretion disk diameter.If the absorbers are discrete clouds, their small sizes and sub-stantial velocity dispersions should lead to fairly rapid dissipa-tion in the absence of an external pressure (see Hamann & Simon(2010)). In particular, the characteristic size of d ∼ .
03 pc or d ∼ . b =
45 km s − in the partial covering sys-tem 6 indicate a dissipation time of roughly t dis ∼ d / b ∼
660 yrs or t dis ∼ ,
400 yrs. At the measured velocity of v = − − ,this gas component would travel just ∼ ∼
60 pc before dissi-pating. A thorough discussion of the creation and survival of theseabsorbing structures is beyond the scope of this paper. However,these simple arguments suggest that at least some of the outflowcomponents we measure are very near their point of creation.It is useful to compare the basic properties of this NAL out-flow to BALs. The ionisations in both types of outflows are simi-lar, with very little low ionisation gas (e.g. C
III ). The outflow ve-locities of the NALs in J1023+5142 are lower, − ,than the typical outflow velocities in BALs, which can reachup to > ,
000 km s − (Korista et al. 1993), but they do over-lap. The NALs have total H column densities ( N ( H ) . to10 . cm − , individual values listed in Table 2), more than 1000times lower than typical BAL H column densities ( N ( H ) > − cm − , and probably higher). By definition, these NALs alsohave FWHMs around 1000 times narrower than typical BALs, andmuch smaller REWs as well. Nonetheless, NALs like this might bepart of the same general outflow phenomenon as BALs, viewed atdifferent angles (Elvis 2000; Ganguly et al. 2001).This complex of weak NAL outflows appears to be dramati-cally different from typical BAL outflows, and constitutes a nearlyunexplored part of the quasar outflow phenomenon, with a rangeof physical parameters and kinematics more complex and variedthan previously thought. It is well-known that NALs are a commonfeature of quasar spectra. Previous surveys have found that 40% ofquasars have C IV NALs, in particular 25% have strong C IV NALswithin v > − − (Vestergaard 2003), 60% have quasar-driven outflows in some form, either BALs, NALs, or somethingin-between (Ganguly & Brotherton 2008; Rodr´ıguez Hidalgo et al.2010), and including high-velocity outflows raises the percentageto 70% (Misawa et al. 2007). If the coverage fraction of these out-flows is less than 100%, which is likely, they could be ubiquitousin the near quasar environment, and could potentially play an influ-ential role in the physical processes occurring therein.Finally, we would like to understand what role the NAL out-flow in J1023+5142 might have in feedback to galaxy evolution.The low speeds and small column densities, e.g., compared to BALflows, suggest that its feedback contribution is negligible. How-ever, there are large uncertainties relating to the outflow locationand geometry. For one particular NAL outflow at a derived radialdistance of ∼ ∼ − measure the C IV emission line FHWM in the SDSS spectrum, and usingEquation 7 in Vestergaard & Peterson (2006), derive a black hole mass oflog M BH = . M ⊙ . Based on these values, we calculate an Eddington lu-minosity fraction of 0.8. The black hole mass and Eddington luminosityfraction are then used in the scaling relations in Hamann & Simon (2010)to calculate the size of the continuum and broad line emission regions. tion of the NAL outflow in J1023+5142 is not known well enoughto make these estimates. A more sensitive search for variability inthese NALs could be very helpful for refining both the locationand the total energy yield (see Hamann et al. (2010) and referencestherein). We find greater than or consistent with solar abundances in allof the absorption systems in J1023+5142 except in system 1, inagreement with previous studies of narrow associated absorptionat lower redshifts (Petitjean & Srianand 1999; Hamann et al. 2001;D’Odorico et al. 2004; Gabel et al. 2006). These high metallici-ties are consistent with the results of other studies of intrinsic gasas well, including BEL gas (Hamann et al. 2002) and thereforeconsistent with our interpretation that the gas is intrinsic to thequasar. Intervening absorbers generally have very low metallici-ties, with Z no more than a few hundredths solar, although thereare cases where high metallicity intervening gas has been observed(Prochaska et al. 2006; Schaye et al. 2007). We argue that the highmetallicities found in 8 of the 9 systems in this quasar are consistentwith locations near the quasar, however, we do not rely solely onthis argument to determine the gas location. Instead, we considerthat high metallicities could be a general phenomenon found in allgas in the quasar host environment (Prochaska & Hennawi 2009).The high metallicities of the NAL systems in J1023+5142require that its host galaxy had vigorous star formation in theepoch before the quasar was observable, leading to metal-rich gasin the quasar outflows (Falomo et al. 2008). This evidence, alongwith previous studies of BELs lead us to conclude that the gener-ally accepted paradigm of quasar-host galaxy evolution is correct,where a major merger leads to a vigorous burst of star formation,which then funnels gas to the centre of the galaxy and ignites aquasar that eventually blows out obscuring gas and dust to becomevisibly luminous (P´erez-Gonz´alez et al. 2008; Hopkins et al. 2008;Ramos Almeida et al. 2009). However, larger samples are neededto examine the full range of NAL properties and study their rela-tionships to quasar outflows and host galaxy environments. Mea-surements at high redshifts are particularly valuable because this isthe main epoch of host/massive galaxy formation when the NALgas might have a close relationship to ongoing or recent star forma-tion in the hosts. We use NALs to improve our understanding of the evolutionaryrelationship between the central black hole and its host galaxythrough the study of their location, origin and abundance informa-tion in high redshift quasars. Here, we examine the properties ofnine NAL systems in the quasar J1023+5142 and find N ( H ) from . to 10 . cm − , velocities from –1400 to –6200 km s − ,C IV doppler b values from 7 to 150 km s − , C IV REW from 0.02to 0.81 ˚A and two systems with partial covering of either the con-tinuum or the BLR, at the level of C f ≈ .
7, which imply absorberdiameters of .
03 pc or . IV and some contain higher ionisation speciessuch as O VI and N V .The C IV absorption NALs are tightly grouped, suggesting thatthey have a physical relationship to one-another, and the proximity c (cid:13) , 000–000 arrow Absorption Line Complex at z = 3.45 of the NAL complex to the quasar redshift suggests that the phys-ical relationship includes the quasar itself. The range in velocityacross the complex is larger than can be easily explained by a sin-gle galaxy or even by a large cluster of galaxies. A more likelyexplanation is that the NAL complex formed in a multi-componentquasar-driven outflow.We estimate directly the location of each system inJ1023+5142 through the use of several diagnostics and find strongevidence (partial covering, broad and smooth profiles compared tothermal widths, velocities greater than galaxy dispersion velocities,supersolar metallicities) that systems 4, 5, 6, 7, 8 and 9 are part ofa quasar outflow. Systems 1, 2, and 3 have more ambiguous ori-gins because they exhibit narrow widths, lower velocity shifts, andsystem 1 has a lower metallicity, so these systems could consist ofintervening gas from the IGM or other galaxies in the line of sight.Systems 2–9 in J1023+5142 exhibit supersolar metallicitiesranging from Z > > Z ⊙ . System 1 has a lower metallicity of Z . Z ⊙ . The high metallicities are consistent with scenarios ofgalaxy and black hole formation and evolution.The NALs in outflows appear to be part of a related outflowcomplex, which is very different than other known outflow regionssuch as BAL outflows, and constitutes a relatively unexplored partof the quasar outflow phenomenon. The outflows in J1023+5142could be important for feedback between the black hole and thehost galaxy, depending on the radial distance of the gas from thequasar.The narrow widths of NALs mean that detailed studies of in-dividual objects like this are the only way to make progress in un-derstanding this type of outflow. Variability studies could be usefulto add more examples of NAL outflows to the current sample avail-able for similar detailed analysis.We will add significantly to this sample in future work, in-cluding detailed studies of the full range of NAL properties in 24quasars at high redshift, during the main epoch of host/massivegalaxy formation. This work was supported in part by a NASA Chandra award (TM9-0005X) and a grant from the Space Telescope Science Institute(HST-GO-11705).
REFERENCES
Aldcroft T., Bechtold J., Foltz C., 1997, in Arav N., Shlosman I.,Weymann R. J., eds, Mass Ejection from Active Galactic Nu-clei Vol. 128 of Astronomical Society of the Pacific ConferenceSeries, Variability in Radio-Loud Quasar Associated AbsorptionLines. pp 25–+Anderson S. F., Weymann R. J., Foltz C. B., Chaffee Jr. F. H.,1987, AJ, 94, 278Arav N., de Kool M., Korista K. T., Crenshaw D. M., van BreugelW., Brotherton M., Green R. F., Pettini M., et al. 2001, ApJ, 561,118Arav N., Kaastra J., Kriss G. A., Korista K. T., Gabel J., Proga D.,2005, ApJ, 620, 665Arnaud M., Rothenflug R., 1985, A&AS, 60, 425Barlow T. A., Hamann F., Sargent W. L. W., 1997, in Arav N.,Shlosman I., Weymann R. J., eds, Mass Ejection from Active Galactic Nuclei Vol. 128 of Astronomical Society of the Pa-cific Conference Series, Partial Coverage and Time Variabilityof Narrow-Line Intrinsic QSO Absorption Systems. pp 13–+Bergeron J., Herbert-Fort S., 2005, ArXiv Astrophysics e-printsBraun E., Milgrom M., 1989, ApJ, 342, 100Dietrich M., Hamann F., Shields J. C., Constantin A., Heidt J.,J¨ager K., Vestergaard M., Wagner S. J., 2003, ApJ, 589, 722D’Odorico V., Cristiani S., Romano D., Granato G. L., Danese L.,2004, MNRAS, 351, 976Elvis M., 2000, ApJ, 545, 63Falomo R., Treves A., Kotilainen J. K., Scarpa R., Uslenghi M.,2008, ApJ, 673, 694Ferland G. J., Korista K. T., Verner D. A., Ferguson J. W., King-don J. B., Verner E. M., 1998, PASP, 110, 761Foltz C. B., Weymann R. J., Peterson B. M., Sun L., MalkanM. A., Chaffee Jr. F. H., 1986, ApJ, 307, 504Fox A. J., Bergeron J., Petitjean P., 2008, ArXiv e-prints, 806Gabel J. R., Arav N., Kim T.-S., 2006, ApJ, 646, 742Ganguly R., Bond N. A., Charlton J. C., Eracleous M., BrandtW. N., Churchill C. W., 2001, ApJ, 549, 133Ganguly R., Brotherton M. S., 2008, ApJ, 672, 102Ganguly R., Eracleous M., Charlton J. C., Churchill C. W., 1999,AJ, 117, 2594Ganguly R., Masiero J., Charlton J. C., Sembach K. R., 2003, ApJ,598, 922Ganguly R., Sembach K. R., Tripp T. M., Savage B. D., WakkerB. P., 2006, ApJ, 645, 868Gibson R. R., Brandt W. N., Schneider D. P., Gallagher S. C.,2008, ApJ, 675, 985Grevesse N., Asplund M., Sauval A. J., 2007, Space Science Re-views, 130, 105Hamann F., Barlow T. A., Beaver E. A., Burbidge E. M., CohenR. D., Junkkarinen V., Lyons R., 1995, ApJ, 443, 606Hamann F., Barlow T. A., Junkkarinen V., Burbidge E. M., 1997,ApJ, 478, 80Hamann F., Beaver E. A., Cohen R. D., Junkkarinen V., LyonsR. W., Burbidge E. M., 1997, ApJ, 488, 155Hamann F., Ferland G., 1999, ARA&A, 37, 487Hamann F., Kanekar N., Prochaska J. X., Murphy M. T., Miluti-novic N., Ellison S., Ubacks W., 2010, in prep.Hamann F., Korista K. T., Ferland G. J., Warner C., Baldwin J.,2002, ApJ, 564, 592Hamann F., Sabra B., 2004, in Richards G. T., Hall P. B., eds,AGN Physics with the Sloan Digital Sky Survey Vol. 311 of As-tronomical Society of the Pacific Conference Series, The DiverseNature of Intrinsic Absorbers in AGNs. pp 203–+Hamann F., Simon L., 2010, in prep.Hamann F. W., Barlow T. A., Chaffee F. C., Foltz C. B., WeymannR. J., 2001, ApJ, 550, 142Hartquist T. W., Snijders M. A. J., 1982, Nat, 299, 783Heckman T. M., Lehnert M. D., Strickland D. K., Armus L., 2000,ApJS, 129, 493Hopkins P. F., Hernquist L., Cox T. J., Kereˇs D., 2008, ApJS, 175,356Hutsem´ekers D., Hall P. B., Brinkmann J., 2004, A&A, 415, 77Jiang L., Fan X., Vestergaard M., Kurk J. D., Walter F., KellyB. C., Strauss M. A., 2007, AJ, 134, 1150Juarez Y., Maiolino R., Mujica R., Pedani M., Marinoni S., NagaoT., Marconi A., Oliva E., 2009, A&A, 494, L25Korista K. T., Voit G. M., Morris S. L., Weymann R. J., 1993,ApJS, 88, 357Kuraszkiewicz J. K., Green P. J., 2002, ApJ, 581, L77 c (cid:13) , 000–000 L. E. Simon and F. Hamann
Misawa T., Charlton J. C., Eracleous M., Ganguly R., Tytler D.,Kirkman D., Suzuki N., Lubin D., 2007, ApJS, 171, 1Misawa T., Eracleous M., Charlton J. C., Kashikawa N., 2007,ApJ, 660, 152Moe M., Arav N., Bautista M. A., Korista K. T., 2009, ApJ, 706,525Nagao T., Marconi A., Maiolino R., 2006, A&A, 447, 157Narayanan D., Hamann F., Barlow T., Burbidge E. M., CohenR. D., Junkkarinen V., Lyons R., 2004, ApJ, 601, 715Nestor D., Hamann F., Rodriguez Hidalgo P., 2008, MNRAS, 386,2055Pentericci L., Fan X., Rix H.-W., Strauss M. A., NarayananV. K., Richards G. T., Schneider D. P., Krolik J., Heckman T.,Brinkmann J., Lamb D. Q., Szokoly G. P., 2002, AJ, 123, 2151P´erez-Gonz´alez P. G., Rieke G. H., Villar V., Barro G., BlaylockM., Egami E., Gallego J., Gil de Paz A., Pascual S., ZamoranoJ., Donley J. L., 2008, ApJ, 675, 234Petitjean P., Srianand R., 1999, A&A, 345, 73Popesso P., Biviano A., 2006, A&A, 460, L23Prochaska J. X., Hennawi J. F., 2009, ApJ, 690, 1558Prochaska J. X., O’Meara J. M., Herbert-Fort S., Burles S.,Prochter G. E., Bernstein R. A., 2006, ApJ, 648, L97Ramos Almeida C., Rodr´ıguez Espinosa J. M., Barro G., GallegoJ., P´erez-Gonz´alez P. G., 2009, AJ, 137, 179Richards G. T., York D. G., Yanny B., Kollgaard R. I., Laurent-Muehleisen S. A., vanden Berk D. E., 1999, ApJ, 513, 576Rodr´ıguez Hidalgo P., Hamann F., Nestor D. B., Shields J., 2010,in prep.Ruiz J. R., Crenshaw D. M., Kraemer S. B., Bower G. A., GullT. R., Hutchings J. B., Kaiser M. E., Weistrop D., 2001, AJ, 122,2961Ruiz J. R., Crenshaw D. M., Kraemer S. B., Bower G. A., GullT. R., Hutchings J. B., Kaiser M. E., Weistrop D., 2005, AJ, 129,73Rupke D. S., Veilleux S., Sanders D. B., 2005, ApJS, 160, 115Schaye J., Carswell R. F., Kim T.-S., 2007, MNRAS, 379, 1169Shen Y., Strauss M. A., Oguri M., Hennawi J. F., Fan X., RichardsG. T., Hall P. B., Gunn J. E., Schneider D. P., Szalay A. S.,Thakar A. R., Vanden Berk D. E., Anderson S. F., Bahcall N. A.,Connolly A. J., Knapp G. R., 2007, AJ, 133, 2222Simon L. E., Hamann F., 2010, submitted to MNRASSrianand R., Petitjean P., 2000, A&A, 357, 414Steidel C. C., 1990, ApJS, 72, 1Trump J. R., Hall P. B., Reichard T. A., Richards G. T., SchneiderD. P., Vanden Berk D. E., Knapp G. R., Anderson S. F., Fan X.,Brinkman J., Kleinman S. J., Nitta A., 2006, ApJS, 165, 1Tytler D., Fan X.-M., 1992, ApJS, 79, 1Tzanavaris P., Carswell R. F., 2003, MNRAS, 340, 937Veilleux S., Cecil G., Bland-Hawthorn J., 2005, ARA&A, 43, 769Vestergaard M., 2003, ApJ, 599, 116Vestergaard M., Peterson B. M., 2006, ApJ, 641, 689Warner C., Hamann F., Dietrich M., 2004, ApJ, 608, 136Weymann R. J., Carswell R. F., Smith M. G., 1981, ARA&A, 19,41Weymann R. J., Williams R. E., Peterson B. M., Turnshek D. A.,1979, ApJ, 234, 33Wild V., Kauffmann G., White S., York D., Lehnert M., HeckmanT., Hall P. B., Khare P., Lundgren B., Schneider D. P., vandenBerk D., 2008, MNRAS, 388, 227 c (cid:13)000