X-raying winds in distant quasars: the first high-redshift wind duty cycle
E. Bertola, M. Dadina, M. Cappi, C. Vignali, G. Chartas, B. De Marco, G. Lanzuisi, M. Giustini, E. Torresi
AAstronomy & Astrophysics manuscript no. ebert_q2237 c (cid:13)
ESO 2020April 29, 2020
X-raying winds in distant quasars: the first high-redshift wind dutycycle
E. Bertola (cid:63) , M. Dadina , M. Cappi , C. Vignali , G. Chartas , B. De Marco , G. Lanzuisi , M. Giustini , and E.Torresi Dipartimento di Fisica e Astronomia, Università degli Studi di Bologna, via P. Gobetti 93 /
2, 40129 Bologna, Italy INAF–OAS, Osservatorio di Astrofisica e Scienza dello Spazio di Bologna, via Gobetti 93 /
3, 40129 Bologna, Italy Department of Physics and Astronomy of the College of Charleston, Charleston, SC 29424, USA N. Copernicus Astronomical Center of the Polish Academy of Sciences, Bartycka 18, 00-716 Warsaw Centro de Astrobiología (CSIC-INTA), Camino Bajo del Castillo s / n, Villanueva de la Cañada, E-28692 Madrid, SpainReceived ; accepted ABSTRACT
Aims.
Theoretical models of wind-driven feedback from Active Galactic Nuclei (AGN) often identify Ultra-fast outflows (UFOs)as being the main cause for generating galaxy-size outflows, possibly the main actors in establishing the so-called AGN–galaxy co-evolution. UFOs are well characterized in local AGN but much less is known in quasars at the cosmic time when SF and AGN activitypeaked ( z (cid:39) z sources to test the wind-driven AGN feedbackmodels. Methods.
Here we present a study of Q2237 + z = . Chandra and XMM-
Newton data (as of September 2019).
Results.
We find clear evidence for spectral variability, possibly due to absorption column density (or covering fraction) variabilityintrinsic to the source. We detect, for the first time in this quasar, a fast X-ray wind outflowing at v out (cid:39) . c that would be powerfulenough ( ˙ E kin (cid:39) . L bol ) to significantly a ff ect the host galaxy evolution. We report also on the possible presence of an even fastercomponent of the wind ( v out ∼ . c ). Given the large sample and long time interval spanned by the analyzed X-ray data, we are ableto roughly estimate, for the first time in a high- z quasar, the wind duty cycle as ≈ α emission line with variable energy, which we discuss in the light of microlensing e ff ects aswell as considering our findings on the source. Key words.
Galaxies: high-redshift – quasars: individual: Q2237 +
030 – quasars: absorption lines – X-rays: general
1. Introduction
Since the discovery of the existence of scaling relations betweenthe mass of super massive black holes (SMBH) and the globalproperties of their host-galaxy bulge (e.g. the M BH − σ relation;McConnell et al. 2011, and references therein), feedback fromthe active galactic nucleus (AGN) is often invoked as a key in-gredient in regulating the star formation (SF) activity in the hostgalaxy and the growth of the SMBH itself. Despite its impor-tance, we still lack a full comprehension of AGN feedback (Ko-rmendy & Ho 2013).State-of-the-art models (e.g. King & Pounds 2015) pre-dict that the SMBH / galaxy co-evolution might be establishedfrom the generation of fast accretion-disk winds, which couldevolve into massive galaxy-scale outflows, possibly quenchingthe host galaxy star formation by sweeping out all of its inter-stellar medium (ISM). To provide e ffi cient AGN feedback, theinner winds need to be accelerated at su ffi cient speed (nomi-nally at semi-relativistic velocities) and need to carry a mechan-ical power higher than a minimum threshold set by the modelsat an approximate value of 0.5%–5% L bol (e.g. Di Matteo et al.2005; Hopkins & Elvis 2010). In the past decade, these innerwinds have often been observationally identified with the so- (cid:63) [email protected] called Ultra-Fast Outflows (UFOs; Tombesi et al. 2010, 2011,2012, 2013; Go ff ord et al. 2013), which are the most extremewinds known to date, characterized by the highest outflow ve-locities (up to 0.2–0.3 c ). They have been discovered in the X-ray band, through their characteristic blueshifted iron resonantabsorption lines above ≈ ≈ ff ord et al. 2014) has been observed, with varia-tions happening on time scales as short as few ks. Thus, the cur-rent idea is that UFOs might well be common and widespread,but episodic events. However, we are still rather groping in thedark for what concerns their average properties in the distantUniverse. In particular, it is of extreme relevance to fill the gapat those redshifts where the SMBH / bulge scaling relations arethought to be shaped, i.e. where we expect the feedback pro-cesses to be most relevant and possibly visible. This corresponds Article number, page 1 of 18 a r X i v : . [ a s t r o - ph . H E ] A p r & A proofs: manuscript no. ebert_q2237 to the cosmic time at which starburst and AGN activity peaked,the so-called cosmic noon ( z (cid:39) z ≥ . + + + + + z > . z ≥ . + z GLQ: the Einstein Cross (Q2237 + z Q = .
695 (lens at z L = . / HRI (Wambsganss et al. 1999) but
Chandra was thefirst X-ray facility capable of resolving the four images of thequasar (Dai et al. 2003). Being the first GLQ with a nearbylens to have ever been discovered, it was soon recognized asa unique case to study both macro- and microlensing proper-ties. It has thus been the target of many microlensing monitoringcampaigns, first in the optical, then, after the advent of
Chan-dra , also in the X-rays, which allowed investigating the size ofthe quasar’s emitting regions (e.g. Mosquera et al. 2013; Guer-ras et al. 2017). Gravitational lensing theoretical models for thissystem agree on predicting time delays between the four sourceimages (A, B, C, D, named as in Yee 1988, – see Fig. 1) shorterthan a day ( ∆ t AB ≈ ∆ t AC ≈ −
16 hrs, ∆ t AD ≈ − of µ ≈
16 (Schmidt et al.1998; Wertz & Surdej 2014). Dai et al. (2003) succeeded in con-firming the shortest time delay ( ∆ t AB = . + . − . hrs) through the Chandra data. Q2237 has also been studied in the X-ray bandto assess its spectral properties, either over single observations,from
Chandra (Chen et al. 2012) and XMM-
Newton (Fedorovaet al. 2008), or from
Chandra spectra stacked over multiple ob-servations, both keeping the images separate (Dai et al. 2003;Chen et al. 2012) and summing all the images (Chen et al. 2012;Reynolds et al. 2014). In this work, we intend to carry out thefirst systematic and comprehensive temporally and spatially re-solved X-ray spectral analysis of this source, taking advantageof the rather complementary strengths that characterize the twoX-ray facilities. In Sect. 2 we list the analyzed data and presentthe reduction procedures. The
Chandra and the XMM-
Newton spectra are first analyzed separately in Sects. 3 and 4, respec-tively, then all the results are combined and discussed in Sect.5. Additional results obtained from the
Chandra stacked spectraare presented in Appendix A. Throughout the paper, we assumea flat Λ CDM cosmology (Planck Collaboration et al. 2018), with H = . − Mpc − and Λ = .
2. Data reduction
We collected, reduced and analyzed all available X-ray data ofQ2237 as of September 2019: 40 archival observations in total,37 from
Chandra and 3 from XMM-
Newton (hereafter, XMM2002, XMM 2016 and XMM 2018), spanning over 18 years Macro-magnification of the individual images: µ A (cid:39) . µ B (cid:39) . µ C (cid:39) .
8, and µ D (cid:39) . ( ∼ . ∼ . Chandra observationsshow exposure times ranging from 7 . . Newton are much longer (42 . . . Chandra and XMM-
Newton data; the time elapsedbetween each XMM-
Newton pointing and the closest
Chandra observation ranges from one to six months. Since one of the maingoals of the present work is to search for and robustly assess(through appropriate statistical tests and simulations) the pres-ence of feedback and the significance of wind-related features inthe X-ray spectra of the Einstein Cross, the lack of simultaneitybetween the
Chandra and XMM-
Newton data does not influencethe results of this work. Conversely, this lack turns out to be quiteconvenient in assessing the recurrence of these features at di ff er-ent epochs and will allow us to investigate their presence on amore extended time baseline.Fig. 1: (a) EPIC-pn cleaned image of XMM 2002 in the 0.3–10 keV observed-energy band. The red square marks the 5 (cid:48)(cid:48) re-gion of the Chandra image centered on the quasar and shownin inset (b). (b) Raw
Chandra image of Q2237 +
030 (ObsID431) binned with a binsize of 0 . (cid:48)(cid:48) , color-coded based on theobserved-energy bands: 0.4–2 keV in red, 2–4.5 keV in green,4.5–7 keV in blue. The images are named as A, B, C, and D asin Yee (1988). Given the quasar redshift ( z Q = . (cid:48)(cid:48) sepa-ration corresponds to a distance of 8 .
68 kpc (cosmology values: H = . − Mpc − , Λ = . Chan-dra data, given the satellite’s superb angular resolution, we cancarry out a spectral analysis that is spatially resolved over the sin-gle images of the quasar (see Fig. 1). On the other hand, XMM-
Newton grants high counting-statistics spectra, by means of itslarger e ff ective area, that allows us to investigate the spatiallyintegrated source emission through more complex and physicalspectral models. All the data were reduced through the respec-tive standard pipelines, using CIAO 4.9 and SAS 16.1, so to uni- Article number, page 2 of 18. Bertola et al.: X-raying winds in distant quasars: the first high-redshift wind duty cycle formly apply the latest calibrations to all observations. To ex-tract individual-image spectra from the
Chandra data, we selectthe four circular regions ( r A , B , C = . (cid:48)(cid:48) and r D = . (cid:48)(cid:48) , with en-circled energy fraction – EEF – at 1 . ff set w.r.t. the image centroids soto limit the contamination from the neighbors. The backgroundextraction region was selected as a source-free circle of 50 (cid:48)(cid:48) ra-dius in the same chip as the target. Furthermore, to consistentlyanalyze the data, we adopted the same regions for all the Chan-dra observations, after checking that they actually correspondedto the emission peak of the individual components in each point-ing.Regarding the XMM-
Newton data, we used as extractionregions a 25 (cid:48)(cid:48) radius circle for the source (EEF
Epn (cid:39)
85% at1 . (cid:48)(cid:48) radius circle for the background in all theXMM- Newton observations. While the background in
Chandra is extremely low (below 0.2% of the total counts), the first twoXMM-
Newton observations were significantly a ff ected by soft-p + flares. Operationally, we filtered the data against di ff erentcount rate thresholds (CRT), then we extracted the source andbackground EPIC-pn spectra for each CRT and inspected howthe latter relate to the former. The best GTI filtering thresholdof the EPIC-pn data was then selected as that yielding a devia-tion of at least a factor of 2 between the background-subtractedsource spectrum and the background spectrum in the 2–8 keVobserved-energy range ( ∼ = . − , S cts = ±
30 cts in the 2–8 keV observed-energy band). Regarding XMM 2016 EPIC-pn,mainly due to the combination of high flares and the shorter ex-posure of this observation, the GTI filtering threshold that satis-fies our condition (CRT = . − , S cts = ±
14 cts in the2–8 keV observed-energy band) drastically reduces the sourcenet counts in the energy band of interest. Being XMM-
Newton data the integration over the four images of the quasar, we leftthis observation out of our analysis since the counting-statisticsof the yielded source spectrum is so to undo the advantages pro-vided by using XMM-
Newton data. Regarding the EPIC-MOSdata, we applied the same procedure as for the EPIC-pn, us-ing the same extraction regions (EEF
EMOS (cid:39)
80% at 1.5 keV).We obtained good quality data for both cameras in XMM 2002(CRT = . − , S cts = ±
21 cts and S cts = ±
20 cts,for MOS1 and MOS2 respectively, in the 2–8 keV observed-energy band). For completeness, we applied this procedure alsoon XMM 2016 EPIC-MOS 1 and 2 but, as expected given theresults for the EPIC-pn spectrum and the lower e ff ective areathat EPIC-MOS 1 and 2 provide, we only managed to con-firm the exclusion of this observation due to the low counting-statistic spectra obtained. XMM 2018, instead, shows a morestable background, that allowed us to select the GTI thresholddirectly from the detector light curves (CRT = . − andCRT = . − for EPIC-pn and EPIC-MOS, respectively).The properties of the cleaned XMM- Newton data are listed inTable 2.
We produced the
Chandra single-image multi-epoch light curvesby computing the image mean count rate, shown in Fig. 2 versusthe respective observation date. Each image presents flux varia-tions up to a factor of ≈ z l = . ≈ . × M (cid:12) BH mass(Assef et al. 2011), is much longer than all the image time delaysinduced by the lens (Dai et al. 2003; Schmidt et al. 1998). C o un t r a t e ( c t ss ) Image A C o un t r a t e ( c t ss ) Image B C o un t r a t e ( c t ss ) Image C C o un t r a t e ( c t ss ) Image D
Fig. 2:
Chandra individual-image multi-epoch light curves, themean count rate between 0.4–7.0 keV observed-frame of eachobservation vs time. From up to bottom: image A, image B, im-age C, image D. The error bars are derived by the counts Poissonerror.Therefore, the dissimilarities between the light curves in Fig. 2at given epoch are likely due to microlensing (Chen et al. 2011).The e ff ect of a microlensing event is to selectively magnify theemission arising from that particular portion of the backgroundsource that is behind the caustic. This results in a perturbationof the macrolensing-magnified image flux and is most relevantfor the images with lower macro-magnification. An outflowingabsorber moving along our line of sight (los) may produce de-tectable blueshifted absorption lines of highly ionized iron. Wethen expect microlensing events, when present, to result in a di-lution of these absorption lines since they would magnify theunabsorbed emission regions that do not lie along our los. Mi-crolensing events can then be considered as a competing e ff ectto the detection of the UFO signatures we are mainly interestedin. In this regard, they are unlikely to fake UFO absorption linesin our spectra or to shift their energy. Although a thorough anal-ysis of the microlensing events is beyond the scope of our work,their e ff ect on highly blueshifted absorption lines would be aninteresting numerical simulation topic for a future project. Article number, page 3 of 18 & A proofs: manuscript no. ebert_q2237 T a b l e : I n f o r m a ti ono f eac h C hand r a ob s e r v a ti on a ndd e t a il s o f t h e i nd i v i du a l - i m a g e s p ec t r a . T ho s e w it h m o r e t h a n500 c oun t s a r e m a r k e d w it h a n a s t e r i s k ( h i gh - s t a ti s ti c ss a m p l e - H SS ) . T h e ob s e r v a ti on s a r e li s t e d f o r i n c r ea s i ngd a t e . O b s I DD a t e E xpo s u r e A c t s B c t s C c t s D c t s A CR B CR C CR D CR - - . (cid:63) ± ± (cid:63) ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . (cid:63) ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . (cid:63) ± ± ± (cid:63) ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . (cid:63) ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . (cid:63) ± (cid:63) ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . (cid:63) ± ± ± ± . ± . . ± . . ± . . ± . - - . (cid:63) ± ± ± ± . ± . . ± . . ± . . ± . - - . (cid:63) ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . (cid:63) ± ± ± ± . ± . . ± . . ± . . ± . - - . (cid:63) ± ± ± ± . ± . . ± . . ± . . ± . - - . (cid:63) ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . - - . ± ± ± ± . ± . . ± . . ± . . ± . N o t e s . T h ee xpo s u r e ti m e i s g i v e n i nun it s o f k s . T h e ob s e r v a ti on s a r e li s t e d f o r i n c r ea s i ngd a t e . T h e i m a g e n e t ( i . e . b ac kg r ound - s ub t r ac t e d ) c oun t s ( c t s ) a nd c oun t r a t e s ( CR ) a r e r e f e rr e d t o t h e . e V ob s e r v e d - e n e r gy r a ng e . T h ec oun t r a t e s a r e g i v e n i nun it s o f − c t ss − . Article number, page 4 of 18. Bertola et al.: X-raying winds in distant quasars: the first high-redshift wind duty cycle
Table 2: Information of each XMM-
Newton observation of Q2237 + ±
62 16.3 ± ±
26 10.4 ± ±
81 6.0 ± Notes.
The exposure time and the cleaned exposure are given in units of ks. The source net (i.e. background-subtracted) counts and count rates arereferred to the EPIC-pn spectra in the 0.3–10 keV observed-energy band. The net count rate is given in units of 10 − cts s − . Being the BH mass of Q2237 estimated to be of the order of10 M (cid:12) (Assef et al. 2011), we do not expect much short time-scale variability ( <
3. Chandra spectral analysis
We first fitted each spectrum with a single power-law modifiedby Galactic absorption ( N H = . × cm − ; Kalberla et al.2005) model (Model pl = phabs*zpo ), restricting the spectral fit-ting to the 0.4–7 keV observed-energy range (1–19 keV rest-frame energy range). The analysis of the Chandra spectra wasthen narrowed down to those with the highest counting-statisticsto better constrain the presence of absorption or narrow emis-sion / absorption features. For what concerns the lower SNR data,we analyzed their stacked spectra including all the Chandra ob-servations, as reported in Appendix A.
The
Chandra data allowed us to probe the source spectral vari-ability on timescales of weeks to years. Figure 3 shows the best-fit photon index Γ i ( i = A , B , C , D) obtained with Model pl asa function of time. The maximum photon-index variation, interms of di ff erence between the highest and the lowest Γ i values,changes from image to image, with image A showing the small-est ( ∆Γ A ≈ .
94) and image B the largest ( ∆Γ B ≈ .
63) varia-tions. By fitting with a constant (dashed lines in Fig. 3), the spec-tral slope was found to be significantly variable ( > .
9% con-fidence) in each image. Considering the ratios of the Γ i , we cancheck whether the photon-index variations are intrinsic (if com-mon to all images) or induced by microlensing. We find the ratiosconsistent with being overall constant and approximately equalto one. This agrees with the approximation applied by Chen et al.(2012), who analyzed the first 20 Chandra observations of ourlist linking the photon index of the four images at any givenepoch. Moreover, the maximum Γ variations of all the imagesare consistent with each other when propagating their 1 σ errors,thus no image shows a significantly higher maximum photon-index variation w.r.t. the others. Therefore, we can assume the A Image A B Image B C Image C D Image D
Fig. 3: Variation of the photon index (1 σ errors) for each im-age in the Chandra data as a function of time. The dashed linerepresents the best fit obtained using a constant function.variations of the continuum to be overall coherent between thefour images, i.e. not induced by microlensing but inherent to thequasar.To investigate the presence of any intrinsic spectral variabil-ity, we restrict our analysis to the high-statistics sample (HSS) tobetter constrain the best-fit spectral parameters. This sub-sampleis made of the fourteen spectra that show more than 500 sourcenet counts in the 0.4–7 observed-energy range (those marked bya star in Table 1). The count threshold was selected to allow us toapply the χ statistics after binning the source spectra to at least20 cts / bin. Fourteen spectra were extracted from eleven observa-tions, since in three epochs (ObsIDs 431, 11534, 12831) two im-ages exceed our threshold. In Fig. 4 we show three representativeHSS spectra: that with the most counts (ObsID 12831 A), one ofthose with the least (ObsID 431 C) and one with an intermediatenumber of counts (ObsID 14517). Spectra with lower-countingstatistics were used to produce stacked spectra, both keeping theimages separate and combining the four images, the results ofwhich are presented in Appendix A.1 and Appendix A.2, respec- Article number, page 5 of 18 & A proofs: manuscript no. ebert_q2237 tively. We anticipate that the results obtained with the stackedspectra are overall consistent with the ones presented here.Table 3: Summary of the best-fit parameters for Model pl_a ( phabs*zphabs*zpowerlw ) when applied to the high-statisticssample. Those that actually require extra absorption at a signifi-cance level above 99% according to the F-test are in boldface.ObsID Γ N H ∆ χ Confidence431 A 1 . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . > . + . − . . + . − . > . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . Notes.
Col. 1: ObsID and image; Col. 2: photon index; Col. 3: columndensity in excess to the Galactic value (in units of 10 cm − and placedat z = . ∆ χ w.r.t. Model pl ( ∆ do f = To search for additional spectral continuum complexities forthe HSS, we first modified Model pl by adding a zphabs com-ponent, accounting for photo-electric absorption of the primaryemission in a cold medium (Model pl_a = phabs*zphabs*zpo ),which we placed at the quasar’s redshift. We find that only fourspectra of the HSS require extra absorption at more than 99%confidence level (evaluated through the F-test), while for theother ten we could only derive an upper limit to such addi-tional column density. All the best-fit values and the respectiveF-test significance are listed in Table 3, with those at more than99% confidence level shown in boldface. Among the four spec-tra highly requiring some extra absorption, two are referred totwo images from the same epoch (ObsID 11534 image A andD) and show consistent column densities within 1 σ errors. Con-sidering only the cases where extra absorption is required by thedata, we find the column density to be variable at more than 99%confidence throughout the three epochs. To test the assumptionon the location of the absorber, we compared the results obtainedfor the two spectra (ObsID 11534 A and 11538 A) showing thelargest variation in column-density values by plotting their 90%confidence contours of N H as a function of the photon index. Asshown in Fig. 5, their column densities are not consistent, whiletheir photon indices are. Furthermore, the time interval amongthe two observations (see Table 1) is ∼
275 d in the observerframe, which means ∼
102 d (cid:39) . N H discrepancy and the short time elapsed asindications of the extra absorption being dominated by the com-ponent at the redshift of the quasar. ObsIDs 11534 A and D, 11538 A, 18804 A.
We next searched for emission / absorption features, again only inthe HSS spectra to obtain better constrains.A first blind search is carried out by applying the methoddeveloped by Tombesi et al. (2010). By stepping both the en-ergy and the normalization of a Gaussian component, it allowsus to visualize, as a function of both parameters, the statisti-cal improvement produced by the addition of a narrow feature( σ = .
01 keV) in terms of ∆ χ translated in confidence contoursfor the addition of two parameters. We then build the best-fitmodels of all the HSS spectra by adding a Gaussian componentfor each emission / absorption line indicated at more than 90%confidence by the blind search. Even though it is known not tobe reliable when assessing the significance of narrow features(Protassov et al. 2002), we compute the F-test significance foreach line to have a slightly better constrain than that of the blindsearch and only keep those that are still above 90% confidencelevel. Finally, we evaluate the actual significance of the absorp-tion lines at E rf > ∼ pl or Model pl_a based on the require-ment of extra absorption (see Table 3). We then obtain the best-Table 4: Rest-frame energies and equivalent widths of (a) theemission and (b) the absorption lines detected at more than 90%confidence in the high-statistics sample, based on the F-test.Those showing more than 99% confidence are reported in bold-face. (a) Emission lines ObsID E line EW Cont. F-test431 A 5 . + . − . + −
90% 90.4%11534 A 3 . + . − . + −
90% 93%5 . + . − . + −
90% 93%11534 D 2 . + . − . + −
90% 98.5%12831 A 4 . + . − . + −
90% 93%6 . + . − . + − . + . − . + −
90% 96% (b) Absorption lines
ObsID E line EW Cont. F-test MC431 A 11 . + . − . + −
99% 99.5% . + . − . + −
90% 97% 90.3%12831 A 9 . + . − . + −
90% 95% 84.2%10 . + . − . + −
99% 99.1% 94.9%13961 A 9 . + . − . + −
99% 98.3% . + . − . + −
90% 92% 98.4%
Notes.
Col. 1: ObsID and image; Col. 2: line energy (in units of keV);Col. 3: line rest-frame equivalent width (in units of eV); Col. 4: blind-search confidence level; Col. 5: F-test confidence level; Col. 6: Monte-Carlo-simulation confidence level. All the errors are computed at 90%confidence level for one parameter of interest. The energy width of thelines is set to 0.01 keV.Article number, page 6 of 18. Bertola et al.: X-raying winds in distant quasars: the first high-redshift wind duty cycle N o r m a li z e d c o un t s s k e V ObsID 431 C E rf ObsID 14517 A E rf rf rf ObsID 12831 A
Fig. 4: Data (black) and best-fit model (blue) for three of the
Chandra
HSS spectra, representative of three statistics regimes. Fromleft to right: ObsID 431 C (least counts), 14517 A (intermediate counts), and 12831 A (highest counts). The dashed vertical linesindicate the energies of the emission / absorption lines found adopting the blind-search method of Tombesi et al. (2010). Only thoseabove 90% were included in the best-fit models (i.e. those in Tables 4a and 4b). C o l u m n d e n s i t y ( c m - ) Confidence contoursPhoton index11538 A11534 A
Fig. 5: 90% confidence contours of N H vs. Γ for ObsIDs 11534A and 11538 A, which correspond to the spectra that show thelargest di ff erence in column density, among the four that requirea cold absorber at more than 99% confidence level.fit models by adding a narrow zgauss component for each lineindicated at a confidence above 90% from the blind search. Weonly keep those lines having a significance above 90% confi-dence both from the blind search and the F-test, which are sum-marized in Table 4. This procedure indicated the presence (at90% confidence level) of blueshifted iron resonant absorptionlines in five spectra out of fourteen.Finally, we evaluate through Monte Carlo simulations the ac-tual significance of the absorption lines in Table 4b. FollowingProtassov et al. (2002), each of these five spectra was simulated1000 times through the XSPEC fakeit function from the re-spective null model (Model pl or Model pl_a if extra absorptionrequired – see Table 3). This confirmed that all the absorptionlines at E rf > P = . · − , yielding an overall significance of 99.2% (i.e.slightly below 3 σ ) for the detection of these absorption lines at E rf > ff er-ent epochs, we overlapped the 1 . σ confidence contours of thenarrow emission and absorption lines separately (Figs. 6 and 7,respectively); the 90% confidence contours of the features de-tected at more than 99% confidence are reported in blue, thoseof the other lines are in green.The emission lines (Fig. 6) span over the 2.2–6.5 keV rest-frame energy band. The microlensed Fe K α line found by Daiet al. (2003) in the combined spectra of ObsIDs 431 A and 1632A ( E = . + . − . keV, σ = . − . − . keV) is only marginally de-tected (90% confidence) in the spectrum of ObsID 431 A asa narrow line, probably due to the fact that we are analyzingsingle-epoch spectra while Dai et al. (2003) stacked the first twoobservations. The energy of the highly significant emission linein ObsID 12831 A ( E rf = . + . − . keV) is consistent, within1 . σ errors, with the centroid energy of the skewed line foundby Reynolds et al. (2014) ( E = . ± .
03 keV) in the com-bined spectra of all the images, stacking the first 26 observations(ObsIDs 431 – 14514). Following Dai et al. (2003) and Chartaset al. (2016a, 2017), the remaining emission lines can be associ-ated with microlensed Fe K α lines.The absorption features (Fig. 7) show a group of valuesaround 11 keV and one line at about 8 keV. Each of the con-fidence contours of the lines clustered around 11 keV cover awide range of energy. This could be interpreted as being dueto the blend of two or more lines given the lower SNR at thehigher energies, thus larger energy bins, or ascribed to intrin-sic very broad lines as seen in other high- z lensed and unlensed Article number, page 7 of 18 & A proofs: manuscript no. ebert_q2237 L i n e no r m a li z a t i on Line Energy (keV, rest frame)Emission lines
Fig. 6: 90% energy-normalization confidence contours (1.6 σ )for the emission lines reported in Table 4a (ObsIDs 431 A, 11534A, 11534 D, 12831 A, 13961 A). The blue contour correspondsto the line detected at more than 99% confidence level in ObsID12831 A (based on the F-test significance).quasars, for instance APM 08279 + + . . σ = . + . − . keV) line but the low statisticsprevent us to simultaneously constrain the three line parameters(energy, width, normalization). As a result, we consider all thelines in Table 4b consistent with being narrow.The highly significant absorption lines (blue contours in Fig.7) are both found at energies above 9 keV (rest frame) and areconsistent with each other at 1.6 σ . These features are detectedin observations 431 and 13961, which are separated by twelveyears ( ∼ . E rf > + C + D from ObsIDs 431, 12831 and 13961, i.e. those obser-vations in which image A shows highly significant absorptionand / or emission lines (see Tables 4a and 4b). In absence of mi-crolensing events, since the time delays between the images areshort when compared to the intrinsic variability timescale, onewould expect the B + C + D combined spectra to show the samekind of features as the image A spectra at a confidence at leasthigher than 90%, since the number of counts of the stacked spec-tra is similar to that of the respective image A. We find that theB + C + D spectra do not present any of the lines of Tables 4a and4b at more than 90% confidence. However, the upper limits wederive on their equivalent widths (EWs) are consistent with those
Absorption lines L i n e no r m a li z a t i on Line Energy (keV, rest frame)
Fig. 7: 90% energy-normalization confidence contours (1.6 σ )for the absorption lines reported in Table 4b (ObsIDs 431 A,11534 A, 12831 A, 13961 A, 14514 A). Those in blue corre-spond to the lines detected at more than 99% confidence level(based on the Monte-Carlo-simulation significance).of the respective image A lines. Our interpretation is that one, ormore than one, of images B, C, and D is microlensed, thus theabsorption lines are smeared out in the individual image spectraand the Fe K α emission lines are likely shifted to higher / lowerenergies, getting even more diluted when we stack the imagestogether. – We find clear spectral index variability at a significancelarger than 99% as inferred from the analysis of the whole
Chandra dataset (Fig. 3). From the Γ ratios, those varia-tions seem to be intrinsic and not induced by microlensing.Photon-index variability might also be induced by variableabsorption in some observations, as the HSS analysis pointedout (Table 3). Having a variability timescale of ∼ . – Five spectra of the HSS do also show narrow emission linesbelow 3 . E rf = . + . − . keV; this energy is consistent with that de-tected by Reynolds et al. (2014). The line found by Dai et al.(2003) is only marginally detected in the spectrum of ObsID431 A. – Finally, five spectra show narrow absorption features inthe 3–5 keV observed-frame energy band ( ∼ / or persistence of those features in this energy range is in-dicated by the consistency of their confidence contours (Fig.7), although we note that the error associated with the en-ergy of the lines is typically large. The overall significanceof detecting such features in the HSS is proved to be almost3 σ . Article number, page 8 of 18. Bertola et al.: X-raying winds in distant quasars: the first high-redshift wind duty cycle
4. XMM-Newton spectral analysis
Having assessed the source spectral variability and the pres-ence of spectral complexities through the
Chandra observations,we then analyzed the XMM-
Newton spectra, and attempted tomodel those complexities with reflection and complex absorp-tion models. Both XMM 2002 and XMM 2018 spectra weregrouped in order to obtain at least 20 counts per bin. Giventhe large number of counts (Table 2), we set the minimum en-ergy width of each bin at one third of the CCD resolution usingthe task specgroup within SAS, so not to oversample the en-ergy resolution of the instrument. Due to background-dominatedbins at the higher energies, the spectral fitting of XMM 2002and XMM 2018 data was performed in the 0.3–8 keV observed-energy range (0.8–22 keV rest-frame energy range) and in the0.3–7.0 keV observed-energy range (1–19 keV rest-frame en-ergy range), respectively, so to obtain more reliable results. Forboth observations, we analyze and report here only the EPIC-pn spectra which deliver the best SNR but our results were alsocompared, and confirmed, by checking their consistency with theEPIC-MOS data.
The observed-frame best-fit residuals to Model pl (Fig. 8, panelb and Table 5) show clear spectral complexities throughout thewhole energy band. Those catching the eye are a deficit of countsat ∼ . E rf ∼ . ∼ . E rf ∼ . Chandra spectra. Moreover, an excess of counts at ∼ . rf ∼ . ∼ . E rf ∼ . Chandra obser-vations, we investigated the need for an extra cold absorberat the quasar’s redshift, either uniformly or partially cover-ing the source (Model pl_a = phabs*zphabs*zpo ; and Model pl_pca = phabs*zpcf*zpo , respectively; best-fit parameters inTable 5). The data do require extra absorption at the quasarredshift and are slightly better ( ∆ χ ∼ pl_pca for each of thenarrow features hinted by the residuals in Fig. 8 (panel b), thetwo in absorption turn out to be required by the data ( ∆ χ > pl_pca plus the three narrow Gaussian components; theabsorption line whose actual significance was to be measuredwas then deleted before performing the simulations. For eachof the two null models, we simulated 1000 spectra through the fakeit function in Xspec . The two absorption lines turn out tobe detected at E rf = . E rf = . warmabs (Kallman & Bautista 2001), an analytic XS-TAR model for self-consistent ionized absorption (Model pl_wa = phabs*warmabs*zpo ), which takes into account alsothe production of absorption lines. Based on the results for mea-surements in other high- z lensed quasars (Chartas et al., in prep.),we assumed Solar abundances and a gas turbulent velocity of N o r m a li z e d c o un t s s k e V Fe XXVFe XXVI ? (a)42024 R e s i d u a l (b)0.5 1 2 5Observed-frame Energy (keV)42024 R e s i d u a l (c)1 2 5 10 20Rest-frame Energy (keV) Fig. 8: Panel (a): XMM 2002 data (black) and best-fit model(solid blue line) with Model pl_wa. The dashed blue lines indi-cate the UFO signatures. Panel (b): XMM 2002 best-fit residualsfor Model pl . Panel (c): XMM 2002 best-fit residuals for Model pl_wa . The latter model self-consistently accounts for the ab-sorption line at E rf (cid:39) . / bin, with minimum energy width set to one thirdof the CCD energy resolution. The best-fit parameters are sum-marized in Table 5. Due to background-dominated bins above8 . − , letting the column density ( N H ), the ionization pa-rameter (log ξ ) and the redshift of the absorber ( z o ) to vary. The warmabs model assumes that the AGN emission modeled in Xspec is the very same radiation that illuminates and ionizesthe absorber. The initial conditions (abundances and density) ofthe medium are loaded through a population file, which dependson the power-law slope of the illuminating radiation. To find thebest-fit Γ , we created di ff erent population files ( n = × cm − , v turb = − ) until the incident-radiation and the best-fit power-law photon indexes converged (within the errors).Model pl_wa yielded the best description of the XMM 2002data ( χ = .
20) and best-fit parameters for the ionized absorberof: N H = . ± . × cm − , log (cid:16) ξ/ erg s − cm (cid:17) = . + . − . and z o (cid:39) .
445 (see Table 5 and Fig. 8, panel a and c). The ob-served value z o of the absorber redshift is related to the intrin-sic redshift z a of the medium (i.e. in the source rest frame) as Article number, page 9 of 18 & A proofs: manuscript no. ebert_q2237
Table 5: Summary of the best-fit parameters for each model tested for XMM 2002 data.Model Γ N H CF log ξ z abs R E e . l . ∆ χ ∆ ν χ ( ν ) pl . ± .
04 – – – – – – – – 1.56 (85) pl_a . ± .
07 0 . ± .
12 – – 1.695 – – 22.6 1 1.31 (84) pl_pca . ± .
11 2 . + . − . . + . − . – 1.695 – – 28.9 2 1.25 (83) pl_wa . ± .
04 28 . + . − . – 2 . + . − . (cid:39) .
445 – – 34.1 3 1.20 (82) pl_pca_pex_el . ± .
32 2 . + . − . . + . − . – 1.695 0 . + . − . . + . − . pl19_pca_pex_el .
90 1 . + . − . . + . − . – 1.695 < .
18 5 . + . − . Notes.
Col. 1: model name; Col. 2: photon index; Col. 3: column density in excess to the Galactic value (units of 10 cm − ); Col. 4:covering fraction of the extra absorption; Col. 5: logarithm of the extra-absorption ionization parameter (erg s − cm); Col. 6: observed red-shift of the extra absorption; Col. 7: reflection scaling factor; Col. 8: energy of the emission line (units of keV); Col. 9, 10: ∆ χ , ∆ ν w.r.t Model pl ( χ = . , ν = χ . The energy width of the emission line is set to 0.01 keV. All the errors are computedat 90% confidence level for one parameter of interest. Model list : Model pl = phabs*zpo ; Model pl_a = phabs*zphabs*zpo ; Model pl_pca = phabs*zpcf*zpo ;Model pl_wa = phabs*warmabs*zpo ; Model pl_pca_pex_el = phabs*zpcf*(zpo+pexrav+zgauss) ; Model pl19_pca_pex_el = phabs*zpcf*(zpo+pexrav+zgauss) with Γ = .
9. All the models include the Galactic absorption ( N H = . × cm − ). (1 + z o ) = (1 + z a ) (cid:16) + z q (cid:17) . Thus, the outflow velocity v out can bedetermined from the relativistic Doppler e ff ect formula: 1 + z a = (cid:112) (1 − β ) / (1 + β ), where β = v out / c . Given z q = . z o (cid:39) .
444 is v out = . ± . c .This ionized wind model naturally explains both the structure inthe soft band and the absorption feature at E rf (cid:39) . pl_pca , which, beingcold, cannot originate such line. Given the wind ionization stateand the outflow velocity, this line is consistent with being domi-nated by Fe xxv , that has a rest-frame energy at rest of 6.7 keV.However, this same model fails to account for the second ab-sorption line at E rf (cid:39) . . E rf = . ± . pl_wa . Its energy is consistent withthat of the line we find in Chandra
ObsID 431 A and with thatof the microlensed Fe K α found by Dai et al. (2003). Its width( σ < .
01 keV), however, is not consistent with the broad onefound by Dai et al. (2003) ( σ = . + . − . keV). The results ofFedorova et al. (2008), who analyzed the XMM 2002 observa-tion, tentatively detecting an emission line at E rf = . + . − . keVwith an energy-width upper limit of σ < . . warmabs component leads to a detectionof the outflow that is well above the 99.99% confidence level.This is linked to the fact that the UFO does not only explain theabsorption line by itself but also acts on the shape of the soft-band continuum, as the two lower panels of Fig. 8 show.For the sake of completeness, we also tested a re-flection scenario. Instead of self-consistent reflection mod-els which bind the Fe K α line to the 6.4–6.7 keV range,we built a phenomenological model so to let the emissionline be placed at lower energies (Model pl_pca_pex_el = phabs*zpcf*(zpo+pexrav+zgauss) ). The reflected-power-law photon index was set to that of the intrinsic emission, theabundances equal to Solar, the inclination angle to 60 ◦ , the cuto ff energy to 300 keV and the reflection fraction to be negative, soto only model the reflected emission through the pexrav com-ponent. The best-fit reflection fraction and photon index (Table5) are R = . + . − . and Γ = . ± .
32 ( χ = . E rf = . ± . = + − eV. Following Makishima(1986), this agrees with what is expected given the column den-sity. However, this model is almost in tension with Leahy (2001)for what concerns the reflection fraction R. The steep power-law may be caused by the known photon index - column den-sity degeneracy. Therefore, we tried setting the photon index(Model pl19_pca_pex_el ) to the standard value Γ = .
90 forhigh- z quasars (e.g. Vignali et al. 2005; Just et al. 2007), whichis also consistent to the average Γ of the Chandra
HSS and thatof the absorption models for this spectrum (see Table 5). Whendoing so, the reflection fraction becomes consistent with zero(90% confidence upper limit: R < .
18) and the quality of the fitdecreases ( χ r = . As for the other spectra, we began our analysis of the XMM 2018EPIC-pn spectrum by inspecting the best-fit residuals to Model pl (Fig. 9, panel b, Table 6). Due to background-dominatedbins above 7.0 keV, we restricted the fitting to the 0.3–7.0 keVobserved-energy range (1–19 keV rest-frame energy range). Theresiduals (Fig. 9, panel b) show complexities in the soft-energyband that are likely due to an absorber and indicate a prominentemission line just below 3 keV in the observed frame. At higherenergies, however, the distribution is quite flat, although noisy,suggesting the absence of a significant reflection component. Nohints of absorption lines in the hard-energy band are seen either.Using the same logical steps as for XMM 2002, we startedby adding more complex absorption models to fit the low en-ergy continuum. All the best-fit parameters of the tested modelsare summarized in Table 6. Given the shape of the residuals, theabsorber during this observation could either be cold and par-tially covering the emitting source, or ionized (Model pl_pca and Model pl_wa , respectively). For completeness, we also in-vestigated the case of a cold medium blocking all the intrinsic Chemical abundances and the gas turbulent velocity were set as donefor XMM 2002 (see Sect. 4.1). The best-fit Γ was found using the samemethod as for XMM 2002.Article number, page 10 of 18. Bertola et al.: X-raying winds in distant quasars: the first high-redshift wind duty cycle emission (Model pl_a ) which, as expected, turned out not to berequired by the data. In all three cases, we set the absorber’sredshift to the systemic of the quasar, based on the results ob-tained with the Chandra data (Sect. 3.1) and because no absorp-tion lines above 7 keV rest-frame were found (i.e. there are nohints of outflowing material). When limiting the analysis to onefeasible absorption component, the XMM 2018 spectrum is bestreproduced by a power-law emission modified by a partial cov-ering medium (Model pl_pca , see Table 6). The prominent emis- N o r m a li z e d c o un t s s k e V E rf (a)42024 R e s i d u a l (b)0.5 1 2 5Observed-frame Energy (keV)42024 R e s i d u a l (c)1 2 5 10 15Rest-frame Energy (keV) Fig. 9: Panel (a): XMM 2018 data (black) and best-fit model(solid blue line) with Model pl_pca_el. The dashed blue line in-dicate the emission line at E rf = . ± .
11 keV. Panel (b):XMM 2018 best-fit residuals for Model pl . Panel (c): XMM 2018best-fit residuals for Model pl_pca_el . The data are grouped soto obtain at least 20 cts / bin, with minimum energy width set toone third of the CCD energy resolution. The best-fit parame-ters are summarized in Table 6. Due to background-dominatedbins above 7 . σ < .
53 keV. Thus we find forXMM 2018 a rather thick absorber that blocks part of the intrin-sic emission ( N H (cid:39) . × cm − , CF (cid:39) .
53) and a signifi-cant emission line with rest-frame energy and equivalent widthof E rf = . ± .
11 keV and EW = ±
111 eV. The energy isinconsistent with both that of the marginal detection in the 2002data (see Sect. 4.1) and that of the skewed emission line found by Reynolds et al. (2014). To see whether the production of thisline could be ascribed to the absorber, we evaluated the upperlimit of the medium ionization state through the warmabs model.To mimic the partial absorption of the intrinsic emission, we in-cluded two power-law plus emission-line components, the firstseen directly and the other as scattered by the absorber (Model pl_pcwa = phabs*[(zpo+zga)+warmabs*(zpo+zga)] ), bothmodified by Galactic absorption. The slopes and the lineparameters of the two terms were linked to each other,since the primary emission is the same for both compo-nents. We inferred a 90% upper limit to the ionization pa-rameter of log (cid:16) ξ/ erg s − cm (cid:17) = . = . ± .
33, which is consistent with that obtained withModel pl_pca (Table 6). The inferred ionized state is not consis-tent to that of the UFO detected in XMM 2002, and nor are thecolumn densities. Thus, we find unlikely for the two absorbersto be the same gas that changed in covering fraction.Following Makishima (1986), the absorber ionization state(log (cid:16) ξ/ erg s − cm (cid:17) ≤ .
2) translates into a medium that is dom-inated by iron from Fe i to Fe xx , while, from its energy,the line we detect is consistent with being dominated byFe xxv – xxvi . Thus, this feature cannot be produced by the ab-sorber itself since it would require a much more ionized gas(log (cid:16) ξ/ erg s − cm (cid:17) ≥ . . ff er-ential magnification of a relativistic blurred Fe K α produced bythe Compton reflection in the accretion disk, as proposed in othersources by Chartas et al. (2016b, 2017). Based on this argument,we tested whether a reflection scenario would give a better repre-sentation of the XMM 2018 data, using the same phenomenolog-ical model discussed in section 4.1 (Model pl_pca_pex_el ). Thismodel returns a best fit that on a statistical basis is as good as thatof Model pl_pca but its reflection fraction R is consistent withzero at the 90% confidence level (R ≤ .
16, see Table 6). On theone hand, this result confirms that the Compton reflection is nota dominant component in the Einstein Cross emission but, on theother hand, it does not completely rule out the interpretation ofthe 6 . α line. Standardreflection models, as pexrav , assume the reflection continuumas produced by the whole disk. In the case of a microlensingevent magnifying the inner regions of the approaching side ofthe accretion disk, Popovi´c et al. (2006) demonstrate that onlythe emission associated to the blueshifted part of the Fe K α lineis enhanced, while the reflection continuum is not, unless the mi-crolensing event is monitored for its whole duration. Since thesource crossing time in the Einstein Cross is estimated to be ofa few months (Mosquera & Kochanek 2011), we cannot rule outthe possibility of this emission line to be a microlensed Fe K α . – The XMM 2002 spectra are best physically and statisticallyreproduced by a complex absorber with N H (cid:39) × cm − ,log (cid:16) ξ/ erg s − cm (cid:17) (cid:39) .
0, and outflow velocity of v out ∼ . E rf (cid:39) . xxv . – The 2018 data do not show any similar blueshifted ab-sorption features and are best fitted by a partial-coveringmildly-ionized material, with N H (cid:39) . × cm − , Article number, page 11 of 18 & A proofs: manuscript no. ebert_q2237
Table 6: Summary of the best-fit parameters for each tested model for XMM 2018 EPIC-pn spectrum. The fitting was carried outover the 0.3–7.0 keV observed-energy range because the data is background dominated above 7 keV.Model Γ N H CF log ξ z abs R E e . l . ∆ χ ∆ ν χ ( ν ) pl . ± .
03 – – – – – – – – 1.64 (102) pl_a . ± . < .
04 – – 1.695 – – -0.4 1 1.72 (101) pl_pca . ± .
10 11 . + . − . . ± .
08 – 1.695 – – 47.3 2 1.20 (100) pl_pca_el . ± .
10 10 . + . − . . ± .
09 – 1.695 – 6 . ± .
11 62.4 4 1.07 (98) pl_wa . ± . < .
19 – 2 . + . − . pl_pca_pex_el . + . − . . + . − . . ± .
09 – 1.695 < .
16 6 . ± .
10 63.7 5 1.07 (97)
Notes.
Col. 1: model name; Col. 2: photon index; Col. 3: column density in excess to the Galactic value (units of 10 cm − ); Col. 4: cov-ering fraction of the extra absorption; Col. 5: logarithm of the extra-absorption ionization parameter (erg s − cm); Col. 6: observed redshiftof the extra absorption; Col. 7: reflection scaling factor; Col. 8: energy of the narrow emission line (units of keV); Col. 9, 10: ∆ χ , ∆ ν w.r.t Model pl ( χ = . , ν = χ . The energy width of the emission line is set to 0.01 keV. All the errors are computed at90% confidence level for one parameter of interest. Model list : Model pl = phabs*zpo ; Model pl_a = phabs*zphabs*zpo ; Model pl_pca = phabs*zpcf*zpo ; Model pl_wa = phabs*warmabs*zpo with z ≡ z q ; Model pl_pca_el = phabs*zpcf*(zpo+zgauss) ; Model pl_pca_pex_el = phabs*zpcf*(zpo+pexrav+zgauss) with Γ = .
9. All the models include the Galactic absorption (N H = . × cm − ). CF (cid:39) .
53 and with a 90% ionization-state upper limit oflog (cid:16) ξ/ erg s − cm (cid:17) < .
2. We detect a significant narrowemission line at E rf = . ± .
11 keV, with an equivalentwidth of EW = ±
111 eV. This line is in tension withbeing produced by the absorber itself because, given itsenergy, a much more ionized medium would be required. – Constraining the reflection component is challenging forboth the XMM spectra, also due to the limited energy rangeprovided at high energies. We find that for the 2002 data, areflection component is statistically significant only when avery steep power-law ( Γ (cid:39) .
65) is assumed, while it is neg-ligible when a typical AGN slope ( Γ (cid:39) .
9) is adopted. Re-garding the 2018 data, despite the presence of a prominentemission line at E rf = . ± .
11 keV, a reflection compo-nent is found not to be statistically required.
5. Discussion and results
We have presented the results obtained from the analysis of allthe available X-ray data of the Einstein Cross (Q 2237 + z = .
695 that is gravitationally lensed in four imagesby a foreground spiral galaxy. We analyzed 40 archival observa-tions, 37 taken by
Chandra and three by XMM-
Newton , cover-ing a period of 18 years, for a total of ∼ . Chandra data, we probed the source spectral vari-ability, using the photon-index variations through the epochs asproxy. These are qualitatively consistent among the four images(i.e. intrinsic), which supports the assumption made by Chenet al. (2012), who linked the photon index among the imageswhen fitting spectra extracted from the same observation. To as-sess the origin of such variability, we limited the analysis to theHSS, i.e. the fourteen spectra extracted from 11 observations thatshow the highest number of counts (above 500 cts in the 0.4–7keV observed-frame energy range), which allowed us to betterconstrain the model parameters. We find that an additional coldabsorber is highly required (above 99% confidence) in four of theHSS spectra, corresponding to three di ff erent epochs. Moreover,the column density is consistent with being variable at more than99% confidence between the epochs. Thus, the spectral variabil-ity is likely ascribed to a variable absorber placed at the quasar’sredshift, but we cannot exclude that part of it could be producedby the variation of the source intrinsic power-law emission aswell. The XMM- Newton data are fundamental in investigatingthe need for extra absorption and the nature of the medium,given the much higher counting-statistics they provide. We findthat the XMM 2002 data are consistent with a UFO scenariowith N H = . + . − . × cm − , log (cid:16) ξ/ erg s − cm (cid:17) = . ± . v out = . ± . c , that explains the prominent absorptionline at E rf = . ± . ∼ . xxvi Ly α ( E rf = .
97 keV at rest) produced by an even faster componentoutflowing at ∼ . c . This would not be the first UFO show-ing more than one outflow component and at such extremelyhigh velocities (see, for instance, the case of APM 08279 + L − (cid:39) . × erg s − . Given a magnification factor of µ ≈
16 (Schmidt et al. 1998), the intrinsic absorption-correctedluminosity is L int2 − (cid:39) . × erg s − . From the UV lumi-nosity log ( λ L λ ) (cid:39) .
53 reported in Assef et al. (2011)and assuming a conversion factor of (cid:39) L bol (cid:39) . × erg s − .Based on Lusso et al. (2012) and the recent work by Duraset al. (2020), the predicted 2–10 keV intrinsic luminositywould be L int2 − (cid:39) × erg s − , which is in good agree-ment with the one we measure. Assef et al. (2011) estimatethe black hole mass M BH from the H β broadening to belog ( M BH / M (cid:12) ) = . ± .
39, which leads to an Eddingtonluminosity of L Edd (cid:39) . × erg s − ( λ Edd ≈ . v out (cid:39) . c , we can derive the physical propertiesof the wind, by adopting standard ’prescriptions’ (e.g. Tombesiet al. 2012; Crenshaw & Kraemer 2012) and including the uncer-tainties on the best-fit parameters, so to place this detection in abroader context and compare it with the measurements in otherQSOs at z ≥ . M out = π r µ m p N H v out C g , where m p is the protonmass, µ is the mean atomic mass per proton (1.4 for solar abun-dances), r is the distance from the BH and C g is the global cover-ing factor of the wind. We assume C g ≈ .
5, based on the statisti-cal study carried out by Tombesi et al. (2010), and recently con-firmed by Igo et al. (2020), over a sample of local Seyfert galax-
Article number, page 12 of 18. Bertola et al.: X-raying winds in distant quasars: the first high-redshift wind duty cycle ies. Moreover, we conservatively assume the outflow to be de-tected at the minimum distance from the BH, where the observedvelocity v out equals the escape velocity from the BH potentialwell, thus r min = M BH / v . We obtain r min ≈ . × cm,that corresponds to (cid:39)
100 gravitational radii ( r g = G M BH / c );considering the uncertainties on M BH , we find for r min the range r min ≈ (0 . − × cm. Using r = r min as radial location ofthe outflow, we are estimating the lower limit to the follow-ing quantities. The mass outflow rate, given all the assumptionsabove, turns out to be ˙ M out ≈ M (cid:12) yr − . Taking into accountthe 1 σ uncertainties of the parameters, we find quite a widerange for the mass-outflow rate: ˙M out ∼ (0 . − .
3) M (cid:12) yr − .As a result, all the following quantities derived using ˙ M out willhave likewise wide uncertainties. The outflow mechanical out-put ( ˙ E kin = ˙ M out v ) is ˙ E kin = . × erg s − , which corre-sponds to an outflow e ffi ciency of ˙ E kin / L bol ≈ .
1. We obtain anoutflow momentum rate of ˙ p out = ˙ M out v out ≈ . × cm g s − ,that is approximately twice the radiation pressure ˙ p rad = L bol / c .Therefore, this UFO is consistent with generating e ffi cient wind-driven AGN feedback that might indeed act on the evolution ofthe quasar host galaxy, given the ˙ E kin / L bol > . L bol / c , magnetic forces might be playing a non-secondaryrole in accelerating this UFO. The derived wind parameters (col-umn density, ionization state and outflow velocity) are consistentwith those of UFOs in local AGN (Tombesi et al. 2010) but thekinematic properties, albeit the wide uncertainties, seem to behigher than the average values for local objects (Tombesi et al.2012). They are instead consistent with those of high- z AGN,for instance of PID352 (Vignali et al. 2015), a bright, unlensedsource at z ≈ . L bol ( ∼ erg s − ). Fur-thermore, the properties of this UFO agree with the v out − L bol and L bol − ˙ E kin relations in Fiore et al. (2017), computed for acompilation of local and (few) high-redshift X-ray winds.Interestingly, the XMM 2018 spectra do not show any ab-sorption line in the hard band and seem to be best reproducedby a partial-covering mild-ionized absorber, with ionization pa-rameter of log (cid:16) ξ/ erg s − cm (cid:17) ≤ . L int2 − (cid:39) . × erg s − , approximately 49% ofthat found for the 2002 data. From the upper limit to the ion-ization state, we evaluated the lower limit to the absorber max-imum distance from the central BH (being ξ = L X / N H r max ): r max ≥ . ff e et al. 2004; Burtscher et al. 2013). Since accretion-disk winds are thought to have a global covering fraction lessthan unity, we propose a scenario where the wind has changedits direction w.r.t. the los and the disk between the two XMMobservations, and part of the clouds contained within the molec-ular torus or the BLR intercept the los during the second point-ing. Given the short time elapsed, we think it is unlikely thatthe outflow has been totally suppressed between the two obser-vations. Given that the Chandra observations were taken in be-tween XMM 2002 and XMM 2018, we find that our statementis supported by the UFO signatures in the
Chandra data. More-over, the lowest timescale of the absorber variability obtainedfrom
Chandra
HSS ( (cid:39) . d ≈ c ∆ t ≈ .
09 pc, consistent with theinnermost regions of the torus and of the BLR, thus consistentwith the proposed scenario (e.g. Perola et al. 2002; Risaliti et al.2009; Burtscher et al. 2013). However, the lower statistics of the
Chandra spectra did not allow us to investigate the presence ofmore complex absorbers than a neutral medium, i.e. to model theabsorption lines we detect with a wind model producing them.In the
Chandra
HSS we find emission lines that span from ∼ . ∼ . α lines. Moreover, the energyrange they cover is consistent with the energies of the redshiftedFe K α emission lines found by Chartas et al. (2016b) in RXJ1131-1231. Interestingly, the only highly significant emissionline is consistent with a regular Fe K α line ( E rf = . + . − . keV).Despite this fact, also this line seems to be microlensed since wedo not detect it in the stacked spectrum of images B + C + D fromthe same observation. From a preliminary search for microlens-ing events in the
Chandra multi-epoch light-curve ratios, there isno clear link between the observations in which we detect theseemission lines and microlensing e ff ects. In XMM 2018 we finda significant emission line at (cid:39) . α emission from a narrow inner region of the disk withoutmagnifying the distant reflected continuum (Popovi´c et al. 2006;Krawczynski & Chartas 2017; Krawczynski et al. 2019). Thepresent data, however, do not allow us to verify either the pres-ence of another highly ionized absorption component, since thedata SNR is too low to constrain such a complex model, or thecase of a microlensing event magnifying the inner regions of theaccretion disk, since longer and less sparse Chandra monitoringwould be required.In the
Chandra spectra, we detect, for the first time in thissource, several blueshifted iron resonant absorption lines, withoverall significance slightly below 3 σ . Interestingly, all the mostsignificant lines ( >
99% confidence level) of the HSS (Table 4b)are grouped around the value of 11 keV and have energies con-sistent to the least significant (87% confidence level) feature ofXMM 2002. If confirmed, they would imply a second and moreextreme wind component with v out ≈ . . c . Thus, the Ein-stein Cross could have experienced a multi-velocity UFO event,as found for other quasars, either nearby (e.g. PDS 456, Boissay-Malaquin et al. 2019; IRAS 00521–7054, Walton et al. 2019) ormore distant (e.g. APM 08279 + / absence of the narrow features. In total, we detect UFOsignatures at more than 90% confidence in six observations (5from Chandra , 1 from XMM-
Newton ) out of the thirteen ana-lyzed to this purpose (11 from
Chandra , 2 from XMM-
Newton ).Thus, we find the wind duty cycle to be DC w ≈ .
46 at 90%confidence. If we only consider those observations showing ab-sorption lines at more than 95% confidence (3 from
Chandra , 1from XMM-
Newton ), the duty cycle turns out to be DC w ≈ . DC w represent the wind-duty-cycle lowerlimit over the thirteen observations that provide data with high-enough statistics. Although strictly related to the signal-to-noiseof the spectra, our estimation of this parameter represents thebest we are able to achieve with present-day data. Article number, page 13 of 18 & A proofs: manuscript no. ebert_q2237
With this work we are adding a new piece to the puzzle ofhigh-redshift AGN ( z > .
5) which do show complex spectraand sometimes show variable UFOs. So far, only around ten ob-jects at z ≥ . + ∼
70% (Chartas et al. in prep). Expanding this high-redshift sam-ple will be key in the future to better assess the occurrence andthe properties of these events at the peak of the QSO activity. Tothis aim, the launch of new-generation X-ray observatories (i.e.
Xrism and
Athena ) will provide us the means to obtain spectrawith much higher signal-to-noise ratios and spectral resolutionw.r.t those we can reach nowadays, essential ingredient in unveil-ing UFOs. This would allow us to better understand if and howdisk-driven winds do trigger e ffi cient AGN feedback at a timewhere the scaling relations between SMBHs and host galaxieswere put in place. Acknowledgements.
The authors thank the referee for the careful reading of themanuscript and the helpful comments. The authors also thank Gabriele Pontifor useful discussions. This work is based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directlyfunded by ESA Member States and the USA (NASA). This research has madeuse of data obtained from the
Chandra
Data Archive, and software provided bythe
Chandra
X-ray Center (CXC) in the application package CIAO. CV and MDacknowledge financial support from the Italian Space Agency (ASI) under thecontracts ASI-INAF I / / / / TIC-11733.
References
Assef, R. J., Denney, K. D., Kochanek, C. S., et al. 2011, ApJ, 742, 93Boissay-Malaquin, R., Danehkar, A., Marshall, H. L., & Nowak, M. A. 2019,ApJ, 873, 29Burtscher, L., Meisenheimer, K., Tristram, K. R. W., et al. 2013, A&A, 558,A149Cappi, M., Tombesi, F., Bianchi, S., et al. 2009, A&A, 504, 401Chartas, G., Brandt, W. N., & Gallagher, S. C. 2003, ApJ, 595, 85Chartas, G., Cappi, M., Hamann, F., et al. 2016a, ApJ, 824, 53Chartas, G., Eracleous, M., Dai, X., Agol, E., & Gallagher, S. 2007, ApJ, 661,678Chartas, G., Krawczynski, H., Zalesky, L., et al. 2017, ApJ, 837, 26Chartas, G., Rhea, C., Kochanek, C., et al. 2016b, Astronomische Nachrichten,337, 356Chartas, G., Saez, C., Brandt, W. N., Giustini, M., & Garmire, G. P. 2009, ApJ,706, 644Chen, B., Dai, X., Kochanek, C. S., et al. 2011, ApJ, 740, L34Chen, B., Dai, X., Kochanek, C. S., et al. 2012, ApJ, 755, 24Crenshaw, D. M. & Kraemer, S. B. 2012, ApJ, 753, 75Dadina, M., Vignali, C., Cappi, M., et al. 2018, A&A, 610, L13Dai, X., Chartas, G., Agol, E., Bautz, M. W., & Garmire, G. P. 2003, ApJ, 589,100Di Matteo, T., Springel, V., & Hernquist, L. 2005, Nature, 433, 604Duras, F., Bongiorno, A., Ricci, F., et al. 2020, A&A, 636, A73Fedorova, E. V., Zhdanov, V. I., Vignali, C., & Palumbo, G. G. C. 2008, A&A,490, 989Fiore, F., Feruglio, C., Shankar, F., et al. 2017, A&A, 601, A143Giustini, M., Cappi, M., Chartas, G., et al. 2011, A&A, 536, A49Go ff ord, J., Reeves, J. N., Braito, V., et al. 2014, ApJ, 784, 77Go ff ord, J., Reeves, J. N., Tombesi, F., et al. 2013, MNRAS, 430, 60Guerras, E., Dai, X., Steele, S., et al. 2017, ApJ, 836, 206Hasinger, G., Schartel, N., & Komossa, S. 2002, ApJ, 573, L77Hopkins, P. F. & Elvis, M. 2010, MNRAS, 401, 7Huchra, J., Gorenstein, M., Kent, S., et al. 1985, AJ, 90, 691Igo, Z., Parker, M. L., Matzeu, G. A., et al. 2020, MNRAS, 493, 1088Ja ff e, W., Meisenheimer, K., Röttgering, H. J. A., et al. 2004, Nature, 429, 47Just, D. W., Brandt, W. N., Shemmer, O., et al. 2007, ApJ, 665, 1004Kalberla, P. M. W., Burton, W. B., Hartmann, D., et al. 2005, A&A, 440, 775Kallman, T. & Bautista, M. 2001, The Astrophysical Journal Supplement Series,133, 221 King, A. & Pounds, K. 2015, Annual Review of Astronomy and Astrophysics,53, 115Kormendy, J. & Ho, L. C. 2013, ARA&A, 51, 511Krawczynski, H. & Chartas, G. 2017, ApJ, 843, 118Krawczynski, H., Chartas, G., & Kislat, F. 2019, ApJ, 870, 125Lanzuisi, G., Giustini, M., Cappi, M., et al. 2012, A&A, 544, A2Leahy, D. A. 2001, ApJ, 547, 449Lusso, E., Comastri, A., Simmons, B. D., et al. 2012, MNRAS, 425, 623Makishima, K. 1986, Iron Lines from Galactic and Extragalactic X-ray Sources,ed. K. O. Mason, M. G. Watson, & N. E. White, 249McConnell, N. J., Ma, C.-P., Gebhardt, K., et al. 2011, Nature, 480, 215Mizumoto, M., Izumi, T., & Kohno, K. 2019, ApJ, 871, 156Mosquera, A. M. & Kochanek, C. S. 2011, ApJ, 738, 96Mosquera, A. M., Kochanek, C. S., Chen, B., et al. 2013, ApJ, 769, 53Perola, G. C., Matt, G., Cappi, M., et al. 2002, A&A, 389, 802Piconcelli, E., Jimenez-Bailón, E., Guainazzi, M., et al. 2005, A&A, 432, 15Planck Collaboration, Aghanim, N., Akrami, Y., et al. 2018, arXiv e-prints,arXiv:1807.06209Ponti, G., Papadakis, I., Bianchi, S., et al. 2012, A&A, 542, A83Popovi´c, L. ˇC., Jovanovi´c, P., Mediavilla, E., et al. 2006, ApJ, 637, 620Protassov, R., van Dyk, D. A., Connors, A., Kashyap, V. L., & Siemiginowska,A. 2002, ApJ, 571, 545Reynolds, M. T., Walton, D. J., Miller, J. M., & Reis, R. C. 2014, ApJ, 792, L19Richards, G. T., Lacy, M., Storrie-Lombardi, L. J., et al. 2006, ApJS, 166, 470Risaliti, G., Salvati, M., Elvis, M., et al. 2009, MNRAS, 393, L1Schmidt, R., Webster, R. L., & Lewis, G. F. 1998, MNRAS, 295, 488Schneider, D. P., Turner, E. L., Gunn, J. E., et al. 1988, AJ, 95, 1619Smith, R. N., Tombesi, F., Veilleux, S., Lohfink, A. M., & Luminari, A. 2019,ApJ, 887, 69Tombesi, F., Cappi, M., Reeves, J. N., & Braito, V. 2012, MNRAS, 422, L1Tombesi, F., Cappi, M., Reeves, J. N., et al. 2013, MNRAS, 430, 1102Tombesi, F., Cappi, M., Reeves, J. N., et al. 2011, ApJ, 742, 44Tombesi, F., Cappi, M., Reeves, J. N., et al. 2010, A&A, 521, A57Vignali, C., Brandt, W. N., Schneider, D. P., & Kaspi, S. 2005, AJ, 129, 2519Vignali, C., Iwasawa, K., Comastri, A., et al. 2015, A&A, 583, A141Walton, D. J., Nardini, E., Gallo, L. C., et al. 2019, MNRAS, 484, 2544Wambsganss, J., Brunner, H., Schindler, S., & Falco, E. 1999, A&A, 346, L5Wertz, O. & Surdej, J. 2014, MNRAS, 442, 428Yee, H. K. C. 1988, AJ, 95, 1331 Article number, page 14 of 18. Bertola et al.: X-raying winds in distant quasars: the first high-redshift wind duty cycle
Appendix A: Chandra stacked spectra
After analyzing the spectral properties of each
Chandra ob-servation, we inspected the stacked spectra to see the persis-tence of the features we find in the
Chandra data and howthese compare to those in the XMM-
Newton data. First, we pro-duced the individual-image stacked spectra combining all theepochs (Appendix A.1), then we combined all the images fromall the epochs to obtain the final stacked spectrum (AppendixA.2). All the spectra were combined through the CIAO tool combine_spectra , then grouped so to obtain at least 20 cts / binand analyzed applying the χ statistics. All the best-fit values toeach tested model are summarized in Table A.1. Appendix A.1: Stacked spectra of the individual images
The single-image stacked spectra (total source time of 749 ksper each image) of images B, C and D show a similar number ofcounts in the 0.4–7 keV observed-energy band (5500 cts, 3975˙ctsand 4766 cts, respectively), comparable to that of XMM 2002and 2018, while for image A we obtain a much better statistics(17700 cts).The analysis applied to these spectra follows the same logicused for the XMM-
Newton data. In Fig. A.1 we show the best-fit residuals to Model pl of the four stacked spectra. The best-fitvalues of the photon index are consistent among the four images(see Table A.1). All images show hints of absorption in the soft-energy band but these appear to be more prominent in image A.Moreover, they all present signatures of strong emission linesbetween 5 and 7 keV. Interestingly, images A, B and D showhints of a line around 6 . (cid:39) . (cid:39) . pl_a and Model pl_pca ). All of the four images requireextra absorption and are best reproduced by a partial coveringmedium placed at the quasar’s redshift (best-fit values in TableA.1). By superposing the Γ − N H contours, we find that the col-umn density is consistent within 1 σ errors for all the stackedimages. In terms of photon index, Γ B , Γ C and Γ D are consis-tent within 1 σ , while Γ A , being the steepest, is consistent withthe others only within 2.6 σ . Regarding the superposed N H − CFcontours, the best-fit values to both the parameters are consistentwithin 1.6 σ for all the four images. The best-fit residuals of eachspectrum to Model pl_pca still show hints of the lines discussedabove.Images A, B and D show a highly significant Fe K α emis-sion line between (cid:39) . pl_pca_el , Table A.1).Regarding image C, we find a highly significant narrow lineplaced at lower energies ( (cid:39) . α at 6 . ∆ χ = . ∆ ν = E rf = . ± .
20 keV and σ = . + . − . keV. We marginally detect a broad emission linein image A (E = . ± .
08 keV, σ = . ± .
09 keV) and im- R e s i d u a l Image A (a) R e s i d u a l Image B (b) R e s i d u a l Image C (c) R e s i d u a l Image D (d)
Fig. A.1: Rest-frame best-fit residuals of the individual-imagestacked spectra to Model pl . From top to bottom: residuals ofimage A, B, C and D.age B ( E rf = . + . − . keV, σ = . + . − . keV), while for imageC the data are best reproduced by a narrow line. The resolvedwidth found for the emission lines can be explained as follows.From the HSS (see Sect. 3.2), we know that the Fe K α is prob-ably microlensed, thus its energy likely varies from epoch toepoch at fixed image. Detecting a broad Fe K α line in the single-image stacked spectra can be interpreted as indicating that themicrolensing of the line at energies near the intrinsic energy of6 . ff ect, while those events produc-ing more extreme energy shifts are more rare ore less e ff ective,as also shown in Chartas et al. (2016b) for RX J1131-1231.The narrow absorption line is marginally detected (at 90%-99% confidence) at E rf (cid:39) . . ≈ . Article number, page 15 of 18 & A proofs: manuscript no. ebert_q2237 T a b l e A . : S u mm a r yo f t h e b e s t - fi t p a r a m e t e r s o f eac h m od e lt e s t e d f o r t h e C hand r a s t ac k e d s p ec t r a . M od e l I m a g e Γ N H C F E e l σ e l E a l σ a l χ r ( ν ) p l A . ± . . ( ) p l B . ± . . ( ) p l C . ± . . ( ) p l D . ± . . ( ) p l _ p c a A . ± . . + . − . . ± . . ( ) p l _ p c a B . ± . . + . − . . ± . . ( ) p l _ p c a C . ± . . + . − . . ± . . ( ) p l _ p c a D . ± . . + . − . . ± . . ( ) p l _ p c a _ e l A . ± . . + . − . . ± . . ± . < . . ( ) p l _ p c a _ e l B . ± . . + . − . . ± . . + . − . < . . ( ) p l _ p c a _ e l ( ) C . ± . . + . − . . + . − . . ± . < . . ( ) p l _ p c a _ e l ( ) C . ± . . + . − . . + . − . . + . − . < . . ( ) p l _ p c a _ e l D . ± . . + . − . . + . − . . + . − . < . . ( ) p l _ p c a _ be l A . ± . . ± . . ± . . ± . . ± . . ( ) p l _ p c a _ be l B . ± . . + . − . . + . − . . + . − . . + . − . ––1 . ( ) p l _ p c a _ be l C . ± . . + . − . . − . . ± . . . − . ––1 . ( ) p l _ p c a _ be l D . ± . . + . − . . − . . ± . . + . − . ––1 . ( ) p l _ p c a _ e l _ a l A . ± . . ± . . ± . . ± . < . . + . − . < . . ( ) p l _ p c a _ e l _ a l B . ± . . + . − . . ± . . + . − . < . . + . − . < . . ( ) p l _ p c a _ e l _ a l C . ± . . . − . . + . − . . ± . < . . + . − . < . . ( ) p l _ p c a _ e l _ a l D . ± . . + . − . . + . − . . + . − . < . . ± . < . . ( ) p l _p c a_e l A + B + D . ± . . ± . . ± . . ± . > . . ( ) p l _p c a_be l A + B + D . ± . . ± . . ± . . + . − . . + . − . ––1 . ( ) p l A + B + C + D . ± . . ( ) p l _ p c a A + B + C + D . ± . . + . − . . ± . . ( ) p l _p c a_e l A + B + C + D . ± . . ± . . ± . . ± . < . . ( ) p l _p c a_e l _a l A + B + C + D . ± . . ± . . ± . . ± . < . . ± . < . . ( ) p l _p c a_be l A + B + C + D . ± . . ± . . ± . . ± . . ± . . ( ) p l _p c a_be l _a l A + B + C + D . ± . . ± . . ± . . ± . . + . − . . ± . < . . ( ) p l _p c a_be l _ba l A + B + C + D . ± . . ± . . ± . . + . − . . ± . . ± . . + . − . . ( ) p l _p c a_be l _e l _a l A + B + C + D . ± . . ± . . ± . . ± . . ± . . ± . < . . ( ) . ± . < . N o t e s . C o l . : m od e l n a m e ; C o l . : c o m b i n e d i m a g e s ; C o l . : c o l u m nd e n s it y i n e x ce ss t o t h e G a l ac ti c v a l u e ( un it s o f c m − ) ; C o l . : c ov e r i ng fr ac ti ono f t h ee x t r aa b s o r p ti on ; C o l . : e n e r gyo f t h ee m i ss i on li n e ( i nun it s o f k e V ) ; C o l . : w i d t ho f t h ee m i ss i on li n e ( i nun it s o f k e V ) ; C o l . : e n e r gyo f t h ea b s o r p ti on li n e ( i nun it s o f k e V ) ; C o l . : w i d t ho f t h ea b s o r p ti on li n e ( i nun it s o f k e V ) ; C o l . : R e du ce d χ ( d e g r ee s o ffr ee do m ) . A llt h ee rr o r s a r ee v a l u a t e d a t % c on fi d e n ce . M od e lli s t : M od e l p l = phabs*zpo ; M od e l p l _p c a = phabs*zpcf*zpo ; M od e l p l _p c a_e l = phabs*zpcf*(zpo+zgauss) ; M od e l p l _p c a_e l _a l = phabs*zpcf*(zpo+zgauss+zgauss) . W h e n t h e w i d t ho f t h ee m i ss i on / a b s o r p ti on li n e i s a fr ee p a r a m e t e r o f t h e m od e l , iti s r e po r t e d a s a b e l / b a l c o m pon e n ti n it s n a m e . Article number, page 16 of 18. Bertola et al.: X-raying winds in distant quasars: the first high-redshift wind duty cycle sistent with being narrow) and a marginally detected Fe K α . Inaddition, it does not even show hints of the absorption line at (cid:39) . α from all the stacked images (also that of the marginal detectionin image C) are consistent within 1 σ , and so they remain if weconsider the energy of the broad lines.Moreover, being the stacking of 37 epochs, we should bearin mind that these broad emission lines might not be intrinsi-cally broad. They could likely be produced by the combinationof single-epoch microlensed Fe K α , which we actually see in thesingle-image spectra (Sect. 3.2). In this sense, the di ff erences insignificance we find between the four images are to be inter-preted as the result of di ff erent microlensing events occurring inthe respective image through the epochs.In conclusion, when stacking all the single-image spectrafrom all the epochs into one single spectrum, we should takeall the properties found above into account, especially those wefind in image C. Appendix A.2: Stacked spectra of all images
The final stacked spectrum sums up to a total exposure timeof almost 3Ms and 32032 source net counts in the 0.4–7 keVobserved-energy range. Its best-fit residuals to Model pl areshown in Fig. A.2, panel a. As expected from the results inthe previous section, we find evidence of absorption in the soft-energy band, a prominent emission line at ≈ . ≈ . R e s i d u a l (a)0.5 1 2 5Observed-frame Energy (keV)42024 R e s i d u a l (b)1 2 5 10 15Rest-frame Energy (keV) Fig. A.2: Best-fit residuals of the final
Chandra spectrum toModel pl ( panel a ) and Model pl_pca_bel_el_al ( panel b ).improves the quality of the fit (see Table A.1). The best-fit pa-rameters we obtain agree with those of the single images and theresiduals still show an excess at ≈ . ≈ . E rf = . ± .
07 keV( ∆ χ / ∆ ν = /
2) and an absorption line at E rf = . ± .
09 keV( ∆ χ / ∆ ν = / pl_pca_bel_al ). We find a much better best-fit( ∆ χ = . ∆ ν = E rf = . ± .
14 keV and σ = . + . − . keV. When allowing also theabsorption-line width to vary (Model pl_pca_bel_bal ), we findit significantly ( ∆ χ = . ∆ ν =
1) consistent with being broad( σ = . + . − . keV) and placed at E rf = . ± .
10 keV. Thischanges also the centroid and the width of the emission line,which become E rf = . + . − . keV and σ = . ± .
16 keV.However, the best-fit residuals of Model pl_pca_bel_bal indi-cate that the model is overestimating / underestimating the dataat energies lower / higher than the best-fit centroid of the emis-sion line. This could be the indication of a relativistically blurredFe K α line. In fact, the energy and width we find using Model pl_pca_bel_al are consistent with those found by Reynolds et al.(2014), which they interpret as the indication of a relativisticallybroadened line. If we exclude the absorption line from the model,we find the same skewed centroid energy and width as in the pre-liminary analysis done by Reynolds et al. (2014).However, based on the single-image stacked-spectra results,the width of the emission line could be artificially produced bythe stacking, since the line in image C is placed at lower en-ergies w.r.t the other three (see Fig. A.1). Thus, we produceda new stacked spectrum, combining all the epochs of only im-ages A, B and D. Considering the absorber and the emissionfeature (Model pl_pca_bel ), we find that the line is less broad( σ = . + . − . keV) and, more importantly, its centroid is placedat higher energies ( E rf = . + . − . keV). Moreover, constrain-ing the width of the line to be lower than the CCD resolutionor letting it to vary freely make almost no di ff erence on a sta-tistical basis (see Table A.1). Finally, if we compute the F-testsignificance for the addition of the width as a free parameter, wefind that it is not required by the data. Thus, the skewed emis-sion line we find at ≈ .
13 keV is most probably generated bythe blending of two distinct lines.Given the results obtained with the combined spectrum ofimages A + B + D and those from the individual-image stackedspectra, we tried to model the skewed emission line as a nar-row component plus a broad component, the first at ≈ . ≈ . pl_pca_bel_el ). This pro-duced a ∆ χ = . pl_pca_bel ), that according to the F-test translates in a de-tection above 99 .
99% confidence of the narrow line placed at E rf = . ± .
08 keV. The broad Fe K α is now detected at E rf = . ± .
07 keV with σ = . ± .
07 keV, which is incon-sistent to the centroid energy of the relativistically skewed linefound by Reynolds et al. (2014) ( E rf = . ± .
03 keV). We alsodetect the narrow absorption line at E rf = . ± .
10 keV (Model pl_pca_bel_el_al ) at 99.8% confidence (from the F-test). Thismodel also corresponds to the one that returns the best repre-sentation of the data, both on the basis of the distribution of theresiduals (see Fig. A.2, panel b) and in terms of statistical im-provement. This result corroborates our statement of the skewedline being the blending of the two lines we find in the individual-image staked spectra. Thus, when stacking the spectra of a grav-itationally lensed quasar, checking the properties of each imageis fundamental.In conclusion, the spectral features of the stacked spectraconfirm the presence of two distinct outflow components basedon the following arguments. The absorption lines at higher en-ergies ( ≈ . Chandra spectra are ab-sent in the stacked spectra, whereas we detect (marginally or sig-nificantly) the absorption line at ≈ . . Article number, page 17 of 18 & A proofs: manuscript no. ebert_q2237 that of the wind producing the 7 . ff er-ent radii w.r.t. the central engine, i.e. the one associated withthe 11 .8 keV being produced closer to the BH. Moreover, thisscenario would also agree with that proposed in Sect. 5 basedon the significance of the two lines in XMM 2002. This wouldimply the absence of lines in XMM 2018 as the indication thatduring the 2018 observation either both outflows are weak (interms of velocity component along the line of sight) or that theoutermost is weak and the other is so extremely ionized that itbecomes undetectable.