On strong correlation between shifted velocity and line width of broad blue-shifted [OIII] components in quasars
aa r X i v : . [ a s t r o - ph . GA ] J a n Draft version January 8, 2021T ypeset using L A TEX twocolumn style in AASTeX63
On strong correlation between shifted velocity and line width of broad blue-shifted [O iii ] components in quasars
XueGuang Zhang School of Physics and technology, Nanjing Normal University, No. 1, Wenyuan Road, Nanjing, 210023, P. R. China
Submitted to ApJABSTRACTIn this manuscript, we report strong linear correlation between shifted velocity and line width of the broadblue-shifted [O iii ] components in SDSS quasars. Broad blue-shifted [O iii ] components are commonly treatedas indicators of outflows related to central engine, however, it is still an open question whether the outflowsare related to central accretion properties or related to local physical properties of NLRs (narrow emission lineregions). Here, the reported strong linear correlation with the Spearman Rank correlation coefficient 0.75 can beexpected under the assumption of AGN (active galactic nuclei) feedback driven outflows, through a large sampleof 535 SDSS quasars with reliable blue-shifted broad [O iii ] components. Moreover, there are much differentdetection rates for broad blue-shifted and broad red-shifted [O iii ] components in quasars, and no positivecorrelation can be found between shifted velocity and line width of the broad red-shifted [O iii ] components,which provide further and strong evidence to reject possibility of local outflows in NLRs leading to the broadblue-shifted [O iii ] components in quasars. Thus, the strong linear correlation can be treated as strong evidencefor the broad blue-shifted [O iii ] components as better indicators of outflows related to central engine in AGN.Furthermore, rather than central BH masses, Eddington ratios and continuum luminosities have key roles onproperties of the broad blue-shifted [O iii ] components in quasars.
Keywords: galaxies:active - galaxies:nuclei - quasars:emission lines INTRODUCTIONActive Galactic Nuclei (AGN) driven outflows, as theprobe of AGN feedback, have been studied in detail formore than two decades (Crenshaw et al. 2003; Veilleux et al.2005; Fabian 2012; Page et al. 2012; King & Pounds 2015;Cheung et al. 2016; Martin-Navarro et al. 2018). AGNfeedback driven outflows not only show clear linkagesbetween AGN and host galaxies, such as AGN feed-back model expected M-sigma relations (Ferrarese & Merritt2000; Gebhardt et al. 2000; Kormendy & Ho 2013) indicat-ing mass outflowing leading to the strong physical connec-tions between central black hole (BH) masses and host galaxyproperties, but also strongly indicate outflows from centralregions have apparent and important effects on structures ofemission/absorption lines from both NLRs (narrow emissionline regions) and BLRs (broad emission line regions).Blue-shifted [O iii ] emission features have been widelytreated as indicators of outflows on scale of kpcs aroundNLRs in AGN, besides broad absorption lines in UV and
Corresponding author: XueGuang Zhang
X-ray bands as better indicators of outflows coming fromcentral accretion disk winds in AGN on scale of light-monthsto light-years (around central BLRs) (Ganguly et al. 2007;Tombesi et al. 2015). Sergeev et al. (1997) have shown thatbroad wings of [O iii ] emission lines could be emitted fromthe outer BLRs and suggested the presence of an outflowcomponent, through study of variations of the [O iii ] line pro-files in NGC5548. Tadhunter et al. (2001) have shown thatkinematics of the broad blue-shifted [O iii ] emission linesare consistent with outflow in an inner NLRs in PKS1549-79. Gupta et al. (2005) have shown that kinematic proper-ties of the blue-shifted absorption-line system (relative to theemission-line system) are similar to the blue-shifted [O iii ]lines, indicating strong connections with outflowing mate-rials in 3C48. Holt et al. (2008) have discussed fast out-flows in compact radio sources, through properties of broadblue-shifted [O iii ] components. Mullaney et al. (2013) haveshown that the [O iii ] profiles of type-1 and type-2 AGNshow the same trends in terms of line width, but type-1 AGNdisplay a much stronger blue wings, which can be well in-terpret as evidence of outflowing ionized gases. Perna et al.(2015) have detected galaxy-wide outflows in high redshift lu-
ZHANG minous obscured quasars by properties of broad [O iii ] lines.Zakamska et al. (2016) have shown that broad [O iii ] emis-sion regions on a few kpc scales can be affected by extremeoutflow from central regions through a sample of high redshiftred quasars. However, not similar as broad absorption linestightly related to outflows from central disk winds, local phys-ical properties in NLRs can also lead to blue-shifted featuresin [O iii ] emission lines, such as local flows in NLRs relatedto stellar winds. Certainly, there are some reported resultsindicating blue-shifted [O iii ] components could be possiblyrelated to central engine. Boroson (2005) have shown weakdependence of shifted velocities of blue-shifted [O iii ] com-ponents on line width of [O iii ] lines through a sample of about800 quasars in SDSS DR1 (Sloan Digital Sky Survey, DataRelease 1) (Abazajian et al. 2003). Schmidt et al. (2018) haveshown much loose correlation between shifted velocities ofblue-shifted [O iii ] components and line widths of [O iii ] linesthrough a sample of 28 narrow line Seyfert I galaxies, muchsimilar weak trends can also be found in Woo et al. (2016);Eun et al. (2017) in samples of type 2 AGN and hidden type-1AGN.More and more evidence have shown that broad blue-shifted [O iii ] components rather than the core [O iii ] com-ponents are more tightly related to central engine in AGN,such as our previous results in Zhang et al. (2017) and pre-vious results in Zakamska et al. (2016), strongly indicatingthat broad [O iii ] emission regions are very nearer to centralregions and have stronger luminosity dependence on centralcontinuum emissions in AGN. In the manuscript,unless other-wise stated, the broad [O iii ] components mean the emissionsfrom the non-BLR regions. Thus, it is interesting to checkwhether are there apparently stronger evidence to support thebroad blue-shifted [O iii ] emission components, rather thanthe blue-shifted asymmetric properties determined from thefull [O iii ] as discussed in the literature, treated as indicators ofoutflows from central regions in AGN. Certainly, after consid-ering probably serious obscuration on broad [O iii ] emissionsin type-2 AGN, we will do our study through a sample ofbroad line AGN. And the manuscript is organized as follows.In section 2, we show our sample selection and emission linefitting procedure. In Section 3, we show our main resultson properties of broad [O iii ] components in a large sam-ple of SDSS quasars in DR12 (Data Release 12, Alam et al.(2015)). In section 4, the main discussions are given. Then,in Section 5, we show our main conclusions. And in thismanuscript, we have adopted the cosmological parameters of 𝐻 = · s − Mpc − , Ω Λ = . Ω m = . SAMPLE SELECTION AND EMISSION LINEFITTING PROCEDUREThe work is based on a large sample of quasars with broadshifted [O iii ] emission components relative to narrow core [O iii ] components. So that, there are three steps to createour main sample. The first step is to create a parent sample ofquasars. The second step is to measure necessary line param-eters, especially parameters of [O iii ] emission lines. And thethird step is to create our final main sample including objectswith reliable broad blue-shifted [O iii ] emission components,based on reliable measured parameters.SDSS SkyServer provided SQL (Structured Query Lan-guage) Search tool (http://skyserver.sdss.org/dr12/en/tools/search/sql.aspx)is firstly applied to conveniently collect SDSS quasars(Richards et al. 2002; Ross et al. 2012) from SDSS DR12to create the parent sample. The applied query is as follows:
SELECT p l a t e , f i b e r i d , mjd
FROM
S p e c O b j a l l
WHERE c l a s s = ’QSO’ and z < 0 . 6 5 andz w a r n i n g = 0 and snmedian > 20 , where "SpecObjall" represents the SDSS provided datasetincluding basic information (redshift, classification, signal-to-noise (SN) of spectra, etc.) of all objects in SDSS DR12.The query leads to collection of 3735 SDSS quasars withreliable redshift less than 0.65 and with high quality spectrawith median SN larger than 20. The criterion of 𝑧 < . iii ] emission lines totally coveredin SDSS spectra. The criterion of median 𝑆𝑁 >
20 onthe full observed spectrum can lead to more neat and cleanspectroscopic emission line features.The second step is to measure necessary emission lineparameters for the quasars in the parent sample. Simi-lar as what we have done in Zhang (2014); Zhang et al.(2016); Rakshit et al. (2017); Zhang et al. (2019), the mostcommonly accepted SSP method (Bruzual & Charlot 2003;Kauffmann et al. 2003; Cid Fernandes et al. 2005) has beenfirstly applied to a small number of the collected SDSSquasars (especially the low redshift quasars with 𝑧 < .
35) ofwhich spectra probably include apparent contributions of stel-lar lights, by considering broadened stellar templates plus apower law component. The 39 simple stellar population tem-plates from Bruzual & Charlot (2003) have been exploitedwith the population ages from 5 Myr to 12 Gyr and withthree solar metallicities (Z = 0.008, 0.05, 0.02), which canbe used to well-describe the characteristics of almost all theSDSS galaxies as detailed discussions in Bruzual & Charlot(2003). Through the Levenberg-Marquardt least-squaresminimization technique (the MPFIT procedure) applied tothe SDSS spectra with the emission lines being maskedout, the stellar velocity dispersions can be determined bythe broadened velocities to the stellar templates, and thenthe stellar component and the AGN power law continuumcomponent can be clearly determined and separated. Here,we have not only masked out all the listed 24 narrow emis-sions lines with rest central wavelengths larger than 3700Å inhttp://classic.sdss.org/dr1/algorithms/speclinefits.html road Blue-shifted [O iii] components in quasars Figure 1.
Left panel shows the SSP method determined stellar lights in the quasar SDSS 0438-51884-0243. Right panel shows the best fittedresults to the emission lines around H 𝛽 in the quasar SDSS 0856-52339-0050. In left panel, from top to bottom, solid lines in black and in redshow the observed SDSS spectrum and the best fitted results by broadened stellar templates plus a power law component, double-dot-dashed redline shows the determined power law AGN continuum emissions, solid blue line shows the determined stellar lights, solid green line shows thepure line spectrum after subtractions of both stellar lights and the AGN continuum emissions. When the SSP method is applied, the emissionlines masked out are marked by the vertical lines in purple and in the two areas filled by purple lines. From left to right, the vertical linesin purple mark the following emission features masked out, including [O ii ] 𝜆 𝜃 , H 𝜂 , [Ne iii ] 𝜆 i 𝜆 iii ] 𝜆 ii ] 𝜆 𝛿 , H 𝛾 , [O iii ] 𝜆 i 𝜆 i ] 𝜆 , ii lines, broad and narrow H 𝛽 and [O iii ]doublet, and the area filled by purple lines around 6550Å shows the region masked out including the emission features of broad and narrow H 𝛼 ,[N ii ] and [S ii ] 𝜆 , 𝜒 value of 1.29 for the best fitted results is marked in the top-left corner in left panel.In the right panel, from top to bottom, solid black line shows the observed line spectrum, solid red line shows the determined best fitted results,double-dot-dashed red line shows the determined power law continuum emissions, solid green line shows the determined broad H 𝛽 , solid purpleline shows the determined optical Fe ii lines, dashed green line shows the determined broad He ii line, solid pink line shows the determinedcore [O iii ] components, and thick blue solid line shows the determined broad blue-shifted [O iii ] components. The top-right corner shows theobserved blue spectrum of SDSS 0856-52339-0050, in order to clearly show there are few contributions of stellar lights in the spectrum. Andthe calculated 𝜒 value of 1.04 for the best fitted results is marked in the top-left corner in the right panel. Figure 2.
Left panel shows the distribution of 𝜒 for the best fitted results to the emission lines in all the collected 3735 quasars (histogram inblue) and in the 535 quasars with reliable broad [O iii ] components (histogram in dark green). Vertical red line shows the position of 𝜒 = iii ] components ofthe 535 quasars. ZHANG with widths of about 450km / s, mainly including the[O ii ] 𝜆 𝛼 , narrow H 𝛽 , narrow H 𝛾 , narrowH 𝛿 , [O iii ] 𝜆 iii ] 𝜆 , i ] 𝜆 , ii ] 𝜆 , ii ] 𝜆 , ii lines and the broad H 𝛼 and H 𝛽 .Based on the model determined stellar velocity dispersionsand uncertainties (the returned best-fit parameters and the re-turned PERROR in the MPFIT procedure), the criterion thatstellar velocity dispersions larger than 70km / s and smallerthan 350km / s at least 5 times larger than their correspondinguncertainties have been applied to determine that the deter-mined stellar components are reliable enough. Here, we donot show further discussions on the SSP method, but theleft panel of Fig. 1 shows an example on the SSP methoddetermined stellar lights in the SDSS quasar PLATE-MJD-FIBERID=0438-51884-0243.After necessary subtractions of contributions of stellarlights, emission lines in line spectrum can be well described.Here, we mainly consider emission lines around H 𝛽 withrest wavelength range from 4400Å to 5600Å, including broadH 𝛽 , narrow H 𝛽 , core and broad [O iii ] components, broadHe ii and broad optical Fe ii lines. The emission featuresare fitted simultaneously by the following model functions.There are two (or more if necessary, after checking fittedresults) broad Gaussian functions 𝐺 B1 + 𝐺 B2 applied to de-scribe broad H 𝛽 . Here, each Gaussian function includes threeparameters of central wavelength, second moment (width ofthe component) and line flux. And we accepted the centralwavelengths of the two broad H 𝛽 components in the rangeof 4800Å to 4900Å. There are three narrow Gaussian func-tions 𝐺 NH + 𝐺 CO31 + 𝐺 CO32 applied to describe the narrowH 𝛽 and core [O iii ] components, with the central wavelengthof narrow H 𝛽 in the range of 4840Å to 4880Å, and withthe flux ratio of core [O iii ] component tied to the theoret-ical value of 3 (Dimitrijevic et al. 2007), and with the cen-tral wavelengths of the three narrow components tied to be4862.81Å:4960.295Å:5008.24Å, and with the core [O iii ]components to have the same line width. There are two an-other Gaussian functions 𝐺 BO31 + 𝐺 BO32 applied to describethe broad [O iii ] components, with the central wavelength ofthe broad [O iii ] 𝜆 iii ] components to have the same redshift,the same line width, and to have the flux ratio tied to thetheoretical value of 3. There is one broad Gaussian function 𝐺 HeII applied to describe weak He ii line, with the centralwavelength in the range of 4600Å to 4730Å. There is onepower law function 𝑃 𝜆 = 𝛼 × ( 𝜆 / ) 𝛽 applied to describeAGN continuum emissions. The broadened and scaled Fe ii templates discussed in Kovacevic et al. (2010) 𝐹𝑒 temps is ap-plied to describe probable optical Fe ii lines. Finally, the detailed model functions are 𝑌 model = 𝐺 B1 + 𝐺 B2 + 𝐺 NH + 𝐺 CO31 + 𝐺 CO32 + 𝐺 BO31 + 𝐺 BO32 + 𝐺 HeII + 𝑃 𝜆 + 𝐹𝑒 temps (1). Based on the widely applied Levenberg-Marquardt least-squares minimization technique, the best fitted results to theemission lines can be well determined, and the line parame-ters and corresponding uncertainties can also be well deter-mined. Here, the uncertainties are the formal 1 𝜎 errors com-puted from the covariance matrix for the final determinedbest-fit model parameters returned by MPFIT procedure(http://cow.physics.wisc.edu/~craigm/idl/idl.html). Then, forthe objects with the calculated 𝜒 = 𝑆𝑆𝑅 / 𝐷𝑜 𝑓 > ∼ 𝑆𝑆𝑅 and
𝐷𝑜 𝑓 as summed squared residuals and degree offreedom, respectively) , the best fitted results have been care-fully re-checked by eyes, in order to determine whether thefitting procedure should be re-applied with more than twobroad components to describe the broad H 𝛽 . Right panelof Fig. 1 shows one example on the best fitted results to theemission lines in the quasar SDSS 0856-52339-0050. Andthe distribution of final determined 𝜒 has been shown in theleft panel of Fig. 2 for all the 3735 quasars.Furthermore, in order to find more robust uncertainties ofthe line parameters, the commonly applied Maximum Like-lihood Method (MEM) through the MCMC (Markov ChainMonte Carlo) technique (Foreman-Mackey et al. 2013) hasbeen accepted to re-fit the emission lines with the startingvalues of model parameters to be the values determined bythe MPFIT procedure. Then, the MCMC technique deter-mined posterior distribution can provide optional uncertaintyof each model parameter. Here, as an two examples, Fig. 3shows the MEM determined best fitted results (which aretotally similar as the MPFIT procedure determined best fit-ted results) to the emission lines around H 𝛽 in SDSS 0856-52339-0050 and in SDSS 1782-53299-0170. For the broad[O iii ] component in SDSS 0856-52339-0050, the centralrest wavelength and the line width can be determined as4997 . ± .
40Å and 781 ± / s based on the MCMCtechnique determined posterior distributions. Meanwhile,the MPFIT procedure determined rest central wavelength andline width of the broad [O iii ] component are 4997 . ± . ± / s. There are similar uncertainties in the linewidth by the MCMC technique and by the MPFIT procedure,but smaller uncertainties in the rest central wavelength by theMCMC technique. However, for the broad [O iii ] compo-nent in SDSS 1782-53299-0170, the determined rest centralwavelength and line width of the broad [O iii ] component are(5007 . ± .
05Å and 5007 . ± . ± / s and535 ± / s) through the MPFIT procedure and through the Among the collected 3735 SDSS quasars, there are 447 quasars with thebest fitted results leading to 𝜒 larger than 2. road Blue-shifted [O iii] components in quasars Figure 3.
Two examples on the MEM determined best fitted results to the emission lines around H 𝛽 in SDSS 0856-52339-0050 (left panel) andin SDSS 1782-53299-0170 (right panel). In each panel, line styles have the same meanings as those shown in the right panel of Fig. 1, but thesolid red line shows the MEM determined best fitted results. MCMC technique, respectively, the uncertainties through theMCMC technique are about 4 times larger than the uncertain-ties determined by the MPFIT procedure. Therefore, differenttechniques can lead to different uncertainties of the model pa-rameters. And in the manuscript, between the uncertaintiesdetermined by the MCMC technique and by the MPFIT pro-cedure for each model parameter, the larger uncertainty hasbeen accepted as its final uncertainty.Based on the measured line parameters and correspondinguncertainties of the core and the broad [O iii ] components,the broader component is accepted as the broad [O iii ] com-ponent, and the narrower component is accepted as the core[O iii ] component, and the shifted velocities of broad [O iii ]components relative to the core [O iii ] components can becalculated by Δ 𝑉 = 𝜆 , core − 𝜆 , broad , with the uncertaintiesof Δ 𝑉 determined by 𝛿 ( Δ 𝑉 ) = 𝛿 ( 𝜆 , core ) + 𝛿 ( 𝜆 , broad ) , where 𝜆 , core and 𝜆 , broad mean the measured central wavelengths ofthe core and the broad [O iii ] components, and 𝛿 ( 𝜆 , core ) and 𝛿 ( 𝜆 , broad ) are the corresponding uncertainties of the centralwavelengths. Then, the following three criteria are appliedto collect quasars with reliable broad blue-shifted [O iii ]components. First, the measured parameters of broad H 𝛽 areat least 5 times larger than their corresponding uncertainties.Second, the measured line parameters of both core and broad[O iii ] components are at least 5 times larger than their corre-sponding uncertainties. Third, the shifted velocities Δ 𝑉 areat least 5 times larger than their corresponding uncertainties: Δ 𝑉 > × 𝛿 ( Δ 𝑉 ) . Based on the three criteria, in our finalmain sample, there are 535 quasars with reliable broad blue-shifted [O iii ] components. Best fitted results to the emis- sion lines in all the 535 quasars can be downloaded fromhttps://pan.baidu.com/s/1gG85QpuXBDfa4NRb5dQHyg.The measured line parameters of the 535 quasars have beenlisted in Table 1. And the distribution of final determined 𝜒 for the best fitted results to the emission lines has been alsoshown in the left panel of Fig. 2 for the 535 quasars.Meanwhile, right panel of Fig. 2 shows distributions of themeasured line widths of the broad and core [O iii ] compo-nents of the collected 535 quasars. In the manuscript, wehave accepted second moments as the line widths of the coreand broad [O iii ] components. And the maximum line widthis about 507km / s of the core [O iii ] components, but themaximum line width is about 1981km / s of the broad [O iii ]components. Fig. 4 shows distributions of redshift and con-tinuum luminosity at 5100Å ( 𝐿 con = 𝜆𝐿 ) of the 535quasars. The continuum luminosity 𝐿 con is calculated basedon the continuum flux at rest wavelength of 5100Å throughthe determined power law function to describe the AGN con-tinuum emissions underneath broad H 𝛽 (such as the compo-nent shown as double-dot-dashed red line in the right panelof Fig. 1), after subtractions of necessary stellar lights. The The access code is xip3. There are three files included in the compressedfile "Blue_O3.tar.gz". One file "par_all_tex.list" includes the necessaryparameters of the 535 quasars. One PDF file "all_spec.pdf" (13M) showsthe results on the determined stellar lights in the spectra of the 535 SDSSquasars with reliable broad blue-shifted [O iii ] components. Among the535 SDSS quasars, there are 26 SDSS quasars of which spectra includeapparent contributions of stellar lights. In the 26 panels with reliable stellarlights, line styles have the same meanings as those shown in the left panelof Figure 1. In the other panels, only the full SDSS spectra are shown. OnePDF file "all_line.pdf" (21M) shows the results on the corresponding bestfitted results to the emission lines around H 𝛽 in the line spectra. And theline spectra of the 26 quasars are determined by subtractions of the stellarlights from the observed SDSS spectra. There are 17 pages in each PDFfile, 32 panels per page. ZHANG mean redshift and continuum luminosity are about 0.323 and3 . × erg / s, respectively.Before proceeding further, there is one point we shouldnote. In some quasars, the broad [O iii ] components haveline width 𝜎 around 1000km / s, such as the line width about 𝜎 ∼ / s of the broad blue-shifted [O iii ] componentsin the SDSS 0856-52339-0050 shown in the right panel ofFig. 1, which are very large values for narrow emission lines,leading to the question whether should the determined broad[O iii ] components be actually as part of broad H 𝛽 fromnormal BLRs? We rejected the possibility by the followingconsideration. We have checked the low-redshift quasars with 𝑧 < .
35 and with reliable broad [O iii ] components. If thebroad [O iii ] components were part of broad H 𝛽 , then similarand corresponding stronger features could be found in thered wings of broad H 𝛼 . However, none low redshift quasarshave shown such strong features in the red wings of broadH 𝛼 . Thus, in the manuscript, we safely accepted that thedetermined broad [O iii ] components are truly from [O iii ]emission regions. MAIN RESULTSBased on the well measured line parameters, properties ofthe broad blue-shifted [O iii ] components can be well checkedin the 535 quasars. Fig. 5 shows the correlation betweenblue-shifted velocity Δ 𝑉 and line width 𝜎 of the broad blue-shifted [O iii ] components. A strong linear correlation canbe confirmed with Spearman Rank correlation coefficient of0.75 with 𝑃 𝑛𝑢𝑙𝑙 < − . Under considering uncertainties inboth coordinates, through the Least Trimmed Squares (LTS)robust technique discussed in Cappellari et al. (2013), thecorrelation can be well described bylog ( 𝜎 km / s ) = ( . ± . ) + ( . ± . ) × log ( Δ 𝑉 km / s ) (2), with rms scatter of about 0.096 in the space of log ( 𝜎 ) versuslog ( Δ 𝑉 ) . In order to ensure the strong linear correlationwithout effects of quality of measured parameters, among the535 quasars, the 160 quasars with parameters (line width andshift velocity) at least 10 times larger than their correspondinguncertainties are applied to show the correlation again inFig 5 as red dots. The strong linear correlation can be re-confirmed with Spearman Rank correlation coefficient of 0.64with 𝑃 𝑛𝑢𝑙𝑙 ∼ − for the 160 high-quality quasars, with thecorresponding rms scatter of about 0.082.It is very interesting that it is so-far the first report on sostrong linear correlation between Δ 𝑉 and 𝜎 in broad blue-shifted [O iii ] components. The strong linear correlation notsimilar as the weak trends in previous references (such as theresult in Eun et al. (2017); Schmidt et al. (2018)) are mainlydue to the following two points. On the one hand, the lineparameters are measured from the well determined Gaussian broad [O iii ] components, not similar as the asymmetric pa-rameters previously determined from the full [O iii ] lines. Onthe other hand, quasars rather than type-2 AGN (or hiddentype-1 AGN) are considered, leading to much wide parame-ter range of the shifted velocities of broad blue-shifted [O iii ]components. Based on the strong linear correlation shown inFig 5, the following points have been considered.First and foremost, whether the strong linear correlationcan be treated as strong evidence for outflows related to cen-tral engine but not from local flows in NLRs? Actually, localflows in NLRs could lead to similar detection rates for broadblue-shifted [O iii ] components and for broad red-shifted[O iii ] components in quasars. However, based on the simi-lar criteria to collect quasars with reliable broad blue-shiftedbroad [O iii ] components, the criteria are applied to collectquasars with reliable broad red-shifted [O iii ] components: 𝑃 H 𝛽 > × 𝛿 ( 𝑃 H 𝛽 ) , 𝑃 core > × 𝛿 ( 𝑃 core ) , 𝑃 broad > × 𝛿 ( 𝑃 broad ) and Δ 𝑉 = 𝜆 , broad − 𝜆 , core > × ( 𝛿 ( 𝜆 , core ) + 𝛿 ( 𝜆 , broad )) where 𝑃 and 𝛿 ( 𝑃 ) represent the measured parameters andthe corresponding uncertainties, the suffixes of "H 𝛽 ", "core"and "broad" mean the parameters and the uncertainties forthe broad H 𝛽 , the core [O iii ] and the broad [O iii ] com-ponents, then there are only 20 quasars with reliable broadred-shifted [O iii ] components. The much different detec-tion rates strongly and naturally indicate that the broad blue-shifted [O iii ] components are due to outflows related to cen-tral engine rather than due to local flows in NLRs in quasars.Furthermore, for the broad red-shifted [O iii ] components, thecorrelation between Δ 𝑉 and 𝜎 is very weak with correlationcoefficient of about -0.28. Besides the much different detec-tion rates, the much different correlations between Δ 𝑉 and 𝜎 for the broad blue-shifted and the broad red-shifted [O iii ]components can also strongly indicate the broad blue-shifted[O iii ] components are due to outflows related to central en-gine rather than due to local flows in NLRs.Besides, whether the large shifted velocities of broad blue-shifted [O iii ] components can be expected by outflows relatedto central engine? As the shown results in Fig. 5, the meanshifted velocity of the broad blue-shifted [O iii ] componentsof the 535 quasars is about 374km / s with minimum valueabout 56km / s and maximum value about 1440km / s. Thestrong dependence of radial velocities of outflowing cloudson distance to central region clearly show that radial velocitieson scale of kpcs are about several hundreds of kilometersper second (see the discussed results in King et al. (2011);King & Pounds (2015); Tombesi et al. (2015)), similar as thevalues shown in Fig. 5. Thus, the large shifted velocities arereasonable under the assumption of AGN-feedback drivenoutflows.Last but not the least, whether AGN-feedback driven out-flows can be applied to explain the strong correlation shownin Fig 5 for the broad blue-shifted [O iii ] components in road Blue-shifted [O iii] components in quasars Figure 4.
Distributions of redshift (left panel) and continuum luminosity at 5100Å (right panel) of the 535 quasars in our final sample.
Figure 5.
On the correlation between Δ 𝑉 and line width 𝜎 of the broad [O iii ] components in the quasars. Blue dots plus error bars show theresults for all the 535 quasars, red dots plus error bars show the results for the 160 high-quality quasars. Solid blue line and dashed red linesrepresent the best fitted results for all the 535 quasars, and the corresponding 5 𝜎 confidence bands, respectively. Green dots plus error barsrepresent the results for reliable broad red-shifted [O iii ] components in the 20 quasars. ZHANG quasars? Under the framework of AGN-feedback driven out-flows, on scale of kpcs, radial velocities do strongly dependon distance 𝑅 to central regions, Δ 𝑉 ∝ 𝑅 − 𝛼 with 𝛼 > iii ]components were due to gravitational potential of central BHbecause of the emission regions nearer to central BH, wecould have 𝜎 ∝ 𝑅 − . Therefore, we can expect the strongcorrelation between Δ 𝑉 and 𝜎 shown in Fig. 5: 𝜎 ∝ ( Δ 𝑉 ) . ,if 𝛼 ∼ DISCUSSIONSIn the section, some further discussions have been shownon the dependence of the strong linear correlation shown inFig. 5 on the other physical parameters, such as Eddingtonratio, BH mass, etc.. The virial BH mass 𝑀 BH can be welldetermined by the line parameters of broad H 𝛽 (Greene & Ho2005; Vestergaard & Peterson 2006; Shen et al. 2011) underthe Virialization assumption to broad line emission clouds(Peterson et al. 2004), 𝑀 BH M ⊙ = . × × ( 𝐿 H 𝛽 erg / s ) . × ( 𝐹𝑊 𝐻 𝑀 H 𝛽 / s ) (3), where 𝐿 H 𝛽 and 𝐹𝑊 𝐻 𝑀 H 𝛽 represent the line luminosityand full width at half maximum of broad H 𝛽 , respectively.The Eddington ratio ¤ 𝑀 Edd can be applied to trace intrinsicaccretion rate and estimated by ¤ 𝑀 Edd = × 𝐿 . × 𝑀 BH / M ⊙ (4). Here, the accepted optical bolometric correction 𝐿 bol ∼ × 𝐿 is mainly from the statistical properties of spec-tral energy distributions of a sample of low redshift quasarsdiscussed in Richards et al. (2006) and from the more recentdiscussed results in Netzer (2020) based on theoretical cal-culations. Moreover, Duras et al. (2020) have shown that theoptical bolometric corrections appear to be fairly constant.Therefore, as commonly applied in the literature (Kaspi et al.2000; Shen et al. 2011), we safely accepted the optical bolo-metric correction 𝐿 bol ∼ × 𝐿 .Based on the calculated 𝑀 BH and ¤ 𝑀 Edd , the Spearman Rankcorrelation coefficients are about 0.15 ( 𝑃 null ∼ × − ) and0.36 ( 𝑃 null ∼ . × − ) for the correlation between 𝑀 BH and Δ 𝑉 and for the correlation between ¤ 𝑀 Edd and Δ 𝑉 , re-spectively, shown in Fig. 6. It is clear that rather than centralBH masses, accretion rates have more important effects onthe strong linear correlation shown in Fig. 5: more power-ful accretion rates lead to more stronger blue-shifted [O iii ]components. Meanwhile, besides the moderate dependence of Eddington ratios on the strong correlation between Δ 𝑉 and 𝜎 , Fig. 7 shows the correlation between Δ 𝑉 and continuumluminosity at 5100Å. The correlation is stronger, with Spear-man Rank correlation coefficient of about 0.48 for all the 535quasars and of about 0.49 for the 160 high-quality quasars.Therefore, rather than BH masses, Eddington ratios and con-tinuum luminosities have more important roles on propertiesof broad blue-shifted [O iii ] components in quasars.Moreover, we have checked effects of extinctions tracedby Balmer decrements (flux ratios of narrow H 𝛼 to narrowH 𝛽 ). Fig. 8 shows the correlation between Balmer decre-ment and Δ 𝑉 of the 151 low-redshift quasars with reliablebroad blue-shifted [O iii ] components and with reliable nar-row H 𝛼 and narrow H 𝛽 emission lines, and of the 10 low-redshift quasars with reliable broad red-shifted [O iii ] compo-nents and with reliable narrow H 𝛼 and narrow H 𝛽 emissionlines. The Spearman Rank correlation coefficients are about0.12 and -0.07 for the quasars with broad blue-shifted andwith broad red-shifted [O iii ] components, respectively. Themean Balmer decrements are about 4.71 and 4.77 for thequasars with broad blue-shifted and with broad red-shifted[O iii ] components, respectively. Therefore, there are feweffects of extinctions on the linear correlation between Δ 𝑉 and 𝜎 . And moreover, in spite of less number of quasarswith broad red-shifted [O iii ] components, there should beno different effects of extinctions on broad blue-shifted andbroad red-shifted [O iii ] components. Here, the emissionlines around H 𝛼 are measured as what we have done to mea-sure emission lines around H 𝛽 . There are two (or more, ifnecessary) Gaussian components applied to describe broadH 𝛼 , one narrow Gaussian component applied to describenarrow H 𝛼 , two Gaussian components applied to describe[N ii ] 𝜆 , ii ] 𝜆 , 𝛼 in the quasar SDSS 1846-54173-0426.Furthermore, as more recent discussed properties of [O iii ]emissions in Mullaney et al. (2013); Harrison et al. (2014);Kakkad et al. (2016); Wylezalek et al. (2020), the parameter 𝑤
80 has been well applied to parameterise the velocity widthof asymmetric [O iii ] emission lines, which refers to the ve-locity width that encloses 80% of the total [O iii ] flux. Here,two parameters of 𝑤
80 and 𝑤
20 of the [O iii ] emission linesare calculated in the SDSS quasars, where 𝑤
20 refers to thevelocity width that encloses 20% of the total [O iii ] flux.Then, rather than the line width of broad [O iii ] componentsapplied , Fig. 10 shows the correlation between 𝑤
80 ( 𝑤 road Blue-shifted [O iii] components in quasars Figure 6.
On the correlation between BH mass and Δ 𝑉 (left panel) and between Eddington ratio and Δ 𝑉 (right panel). In each panel, Dots pluserror bars are for all the 535 quasars, red dots plus error bars are for the 160 high-quality quasars. Figure 7.
On the correlation between continuum luminosity and Δ 𝑉 . Dots plus error bars are for all the 535 quasars, red dots pluserror bars are for the 160 high-quality quasars. Figure 8.
On the dependence of Δ 𝑉 on Balmer decrements, sym-bols in blue and in red are for the quasars with broad blue-shifted[O iii ] components, and for the quasars with broad red-shifted [O iii ]components, respectively. Figure 9.
An example on the best fitted results to the emission linesaround H 𝛼 in the quasar SDSS 1846-54173-0426. In the figure,solid line in black and solid line in red show the line spectrumand the best fitted results, double-dot-dashed red line shows thedetermined power law continuum emissions, solid green line showsthe determined broad H 𝛼 , solid purple lines and in blue show thedetermined [N ii ] and [S ii ] 𝜆 , 𝛼 , respectively. In the top-rightcorner, the full spectrum with few contributions of stellar lights hasbeen shown. and Δ 𝑉 , which can be well described bylog ( 𝑤 / s ) = ( . ± . ) + ( . ± . ) × log ( Δ 𝑉 km / s ) log ( 𝑤 / s ) = ( . ± . ) + ( . ± . ) × log ( Δ 𝑉 km / s ) (5), through the LTS technique. It is apparent that there are alsotwo linear correlations, but the Spearman Rank correlationcoefficient is about 0.79 ( 𝑃 null < − ) for the correlationbetween 𝑤
80 and Δ 𝑉 , larger than the coefficient of 0.63 forthe correlation between 𝑤
20 and Δ 𝑉 . The stronger correlationon the parameter of 𝑤
80 indicates broad blue-shifted [O iii ]component rather than the core [O iii ] components have more0
ZHANG
Figure 10.
Left panel shows the correlation between 𝑤
80 and Δ 𝑉 . Right panel shows the correlation between 𝑤
20 and Δ 𝑉 . In each panel, solidblue line shows the best fitted results, dashed red lines show the corresponding 5 𝜎 confidence bands, respectively. In each panel, dots are forall the 535 quasars, red dots plus error bars are for the 160 high-quality quasars. important contributions to the correlation between 𝜎 and Δ 𝑉 shown in Fig. 5. CONCLUSIONSFinally, we give our main conclusions as follows. Based ona large sample of 535 SDSS quasar with reliable broad blue-shifted [O iii ] components, a strong linear correlation withSpearman Rank correlation coefficient 0.75 can be clearlyconfirmed between shifted velocity and line width of the broadblue-shifted [O iii ] components, which can be explained un-der the assumption of AGN-feedback driven outflows. Mean-while, through similar criteria, there are only 20 SDSS quasarswith reliable broad red-shifted [O iii ] components, and thereis no positive correlation between shifted velocity and linewidth of the broad red-shifted [O iii ] components, providingstrong evidence against the blue-shifted [O iii ] componentsrelated to local flows in NLRs. Therefore, strong broad blue-shifted [O iii ] components can be treated as better indicator of outflows related to central engine in AGN. Moreover, strongerdependence of Eddington ratios and continuum luminositiescan be found on the correlation between Δ 𝑉 and 𝜎 of thebroad blue-shifted [O iii ] components in quasars, indicatingEddington ratios and continuum luminosities have more im-portant roles on properties of blue-shifted [O iii Abazajian, K.; Adelman-McCarthy, J. K.; Agueros, M. A.; et al.,2003, AJ, 126, 2081Alam, S., et al., 2015, ApJS, 219, 12Boroson, T., 2005, AJ, 130, 381Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000Cappellari, M.; Scott, N.; Alatalo, K.; et al., 2013, MNRAS, 432,1709Cid Fernandes, R., Mateus, A., Sodre, L., Stasinska, G., Gomes, J.M., 2005, MNRAS, 358, 363Cheung, E.; Bundy, K.; Cappellari, M.; et al., 2016, Natur, 533, 504Crenshaw, D. M.; Kraemer, S. B.; George, I. M., 2013, ARA&A,41, 117 Dimitrijevic, M. S.; Kovacevic, J.; Popovic, L. C.; Dacic, M.; Ilic,D., 2007, MNRAS, 374, 1181Duras, F.; Bongiorno, A.; Ricci, F.; et al., 2020, A&A, 636, 73Eun, D.; Woo, J.-H.; Bae, H.-J., 2017, ApJ, 842, 5Fabian, A. C., 2012, ARA&A, 50, 455Ferrarese, F. & Merritt, D., 2000, ApJL, 539, L9Foreman-Mackey, D.; Hogg, D. W.; Lang, D.; Goodman, J., 2016,PASP, 125, 306Ganguly, R.; Brotherton, M. S.; Cales, S.; et al., 2007, ApJ, 665,990Gebhardt, K.; Bender, R.; Bower, G.; et al., 2000, ApJL, 539, L13Green, J. E., Ho, L. C., 2005, ApJ, 630, 122Gupta, Neeraj; Srianand, R.; Saikia, D. J.. 2005, MNRAS, 361, 451 road Blue-shifted [O iii] components in quasars Table 1.
Parameters of the 535 SDSS quasars with broad blue-shifted [O iii ] componentspmf z Δ 𝑉 𝜎 log ( 𝐿 con ) log ( 𝑀 BH ) log ( ¤ 𝑀 Edd ) 𝑤 𝑤 / s km / s erg / s M ⊙ km / s km / s0274-51913-0388 0.256 392 ±
51 478 ±
28 44.07 ± ±
15 330 ±
19 44.09 ± ±
50 669 ±
75 44.30 ± ±
17 364 ±
22 44.58 ± ±
41 595 ±
40 43.35 ± ±
74 756 ±
37 44.61 ± ±
165 958 ±
98 44.83 ± ±
36 573 ±
83 44.94 ± ±
42 421 ±
25 43.93 ± ±
13 320 ±
24 43.76 ± iii ] components relative to the core [O iii ] components, Col(4) shows the line width of the broad [O iii ] components, Col(5) showsthe information of continuum luminosity, Col(6) shows the information of virial BH masses, Col(7) shows the information of Eddington ratios,Col(8) and Col(9) show the information of 𝑤
80 and 𝑤𝑤