Draft version February 1, 2017 PreprinttypesetusingLATEXstyleemulateapjv.5/2/11 STATISTICAL DETECTION OF THE HEII TRANSVERSE PROXIMITY EFFECT: EVIDENCE FOR SUSTAINED QUASAR ACTIVITY FOR >25 MILLION YEARS Tobias M. Schmidt1*, Gabor Worseck1, Joseph F. Hennawi1,2, J. Xavier Prochaska3, Neil H. M. Crighton4 Draft version February 1, 2017 ABSTRACT The Heii transverse proximity effect – enhanced Heii Lyα transmission in a background sightline causedbytheionizingradiationofaforegroundquasar–offersauniqueopportunitytoprobethemor- 7 phologyofquasar-driven Heii reionization. Weconduct acomprehensivespectroscopicsurveytofind 1 z 3 quasars in the foreground of 22 background quasar sightlines with HST/COS Heii Lyα trans- 0 m∼ission spectra. With our two-tiered survey strategy, consisting of a deep pencil-beam survey and a 2 shallowwide-fieldsurvey,wediscover131newquasars,whichwecomplementwithknownSDSS/BOSS n quasars in our fields. Using a restricted sample of 66 foreground quasars with inferred Heii photoion- Ja ipzeartfioornmrattheesfigrrsetatsetrattihsatincatlhaeneaxlpyseicsteodf UthVe Hbaecikigtrroaunnsdveartsethpersoexirmeditsyhieftffsec(Γt.HQeSOIOIu>r r5es×ul1ts0−s1h6osw−1q)uwale- 0 itative evidence for a large object-to-object variance: among the four foreground quasars with the 3 highest ΓHeII only one (previously known) quasar is associated with a significant Heii transmission QSO spike. We perform a stacking analysis to average down these fluctuations, and detect an excess in the ] averageHeiitransmissionneartheforegroundquasarsat3σ significance. Thisstatisticalevidencefor A the transverse proximity effect is corroborated by a clear dependence of the signal strength on ΓHeII. G QSO Ourdetectionplacesapurelygeometricallowerlimitonthequasarlifetimeoft >25Myr. Improved Q . modeling would additionally constrain quasar obscuration and the mean free path of Heii-ionizing h photons. p - Subject headings: dark ages, reionization, first stars – intergalactic medium – quasars: general o r t s 1. INTRODUCTION libriumuptothepresentday(Boltonetal.2006;Furlan- a etto & Oh 2008; McQuinn 2009; Furlanetto & Dixon The double ionization of helium at z 3, known as [ Heii reionization, is the final phase trans∼ition of the in- 2010; Furlanetto & Lidz 2011; Haardt & Madau 2012; Meiksin & Tittley 2012; Compostella et al. 2013, 2014). 1 tergalactic medium (IGM) and substantially influences The exact structure of Heii reionization crucially de- v the thermal and ionization state of the baryons in the 9 Universe. According to the currently accepted picture pends on the duration of the bright quasar phase. To 6 (Haardt & Madau 2012; Planck Collaboration et al. match a given luminosity function, quasars either have 7 2016) hydrogen was reionized at redshifts z 8 by the tobeverycommonobjectswhichindividuallyshineonly 8 UV photons emitted from stars. Their spectra∼, however, forashorttimeorrareobjectsifthelifetimeislong(e.g. 0 Martini2004). Thequasarlifetimealsoplacesimportant were not hard enough to supply sufficient numbers of . constraints on the physics and accretion processes of ac- 1 photons with energies > 4Ry required to doubly ionize 0 helium. Therefore, Heii reionization was delayed until tive galactic nuclei (AGN). All massive galaxies are ex- pectedtohostasupermassiveblackhole(SMBH)intheir 7 the quasar era, culminating in the completion of helium center but most of them are inactive and do not accrete 1 reionizationatz 2.7(Madau&Meiksin1994;Reimers : et al. 1997; Mira≈lda-Escud´e et al. 2000; McQuinn 2009; mattermostofthetime. Itishoweverassumedthatallof v themwentthroughanactiveAGNphaseatleastoncein Faucher-Gigu`ereetal.2009;Worsecketal.2011;Haardt i their history and SMBHs at least partially grew to their X & Madau 2012; Compostella et al. 2013, 2014; Worseck current mass during bright quasar phases (Soltan 1982; r et al.2016). Since quasarsare brightbut raresources, it a is expected that Heii reionization is a very patchy pro- Shankar et al. 2009; Kelly et al. 2010). Also, quasars might have a substantial influence on their host galax- cess. The general picture is that quasars create ionized ies, and simulations of galaxy formation usually invoke bubblesaroundthemwhichexpandwithtime,andeven- quasarfeedbacktomatchthegalaxyluminosityfunction tually overlap to form the relatively homogeneous UV at the massive end. Constraining the duration of the backgroundthatkeepstheIGMinphotoionizationequi- quasar phase might shed light on the largely unknown triggersforAGNactivityandtherebytestquasarmodels *e-mail: [email protected] (e.g. Springeletal.2005;Hopkinsetal.2005;Schawinski 1Max-Planck-Institut fu¨r Astronomie, K¨onigstuhl 17, D- et al. 2015). 69117Heidelberg,Germany 2DepartmentofPhysics,UniversityofCalifornia,SantaBar- There have been several attempts for measuring bara,CA93106,USA the quasar lifetime (e.g. Martini 2004, and references 3DepartmentofAstronomyandAstrophysics,UCO/LickOb- therein). An upper limit of 109 yr comes from de- servatory,UniversityofCalifornia,1156HighStreet,SantaCruz, mographic arguments and the evolution of the AGN CA95064,USA 4Centre for Astrophysics and Supercomputing, Swinburne population. Methods based on the quasar luminosity UniversityofTechnology,POBox218,VIC3122,Australia function and clustering statistics are in principle able 2 T. M. Schmidt et al. to determine the average time galaxies host luminous be active for at least the equilibration timescale. This is quasars, described as the duty cycle (t ) of quasars. for Hi of the order of tHI 104 yr (ΓHI 10 12s 1, dc eq ∼ UVB ∼ − − However, studies so far only deliver weak constraints of Becker & Bolton 2013). A very similar effect is ex- tdc 106 109 yr (Adelberger & Steidel 2005; Croom pected when a background sightline intersects the prox- ∼ − et al. 2005; Shen et al. 2009; White et al. 2012; Conroy imity zone of a foreground quasar, called the transverse &White2013;LaPlante&Trac2015)duetotheuncer- proximity effect. It would enable robust estimates of the taintyinhowquasarspopulatedarkmatterhalos. Stud- quasarlifetimepurelybasedongeometricargumentsand ies focusing on the mass assembly of SMBHs are sensi- the light travel time from the foreground quasar to the tivetothetotalblackholegrowthtimewhichmightalso background sightline (e.g. Dobrzycki & Bechtold 1991; include non-luminous or obscured phases. Constraints Adelberger2004). However,uptonowthereisnounam- from such studies (e.g. Yu & Tremaine 2002; Kelly et al. biguous detection of the transverse proximity effect the 2010) are between 30 and 150Myr. Hi Lyα forest, which might be attributed to the cosmic These methods determine the quasar duty cycle or overdensities in which quasars are hosted or the obscu- growthtimeforafullpopulation,reflectingthetotaltime ration of the quasar radiation (Liske & Williger 2001; quasarsareactive. Thismightbedistinctfromthedura- Schirber et al. 2004; Croft 2004; Hennawi et al. 2006; tionofsingleburstsifquasarsgothroughseveralactivity Hennawi & Prochaska 2007; Kirkman & Tytler 2008). periods over a Hubble time. The duration of these can In addition to the hydrogen Lyα forest, there is also be described by the episodic lifetime tQ. Methods which a helium Lyα forest at frequencies 4 higher, caused by aim to constrain the episodic lifetime usually focus on singly ionized helium (Heii). Studyi×ng proximity effects individual quasars and use the presence of a tracer that in helium is for several reasons advantageous over prob- is sensitive to the quasar luminosity at some time in the ing the same effect in hydrogen. At z (cid:46) 5.5, the Uni- past. This measures for how long an individual quasar verse is transparent to 1Ry photons, resulting in a high hasbeenactivepriortotheobservation,basicallyitsage. and quasi-homogeneous UV background (e.g. Miralda- We denote this as tage and stress that it is distinct from Escud´e et al. 2000; Meiksin & White 2004; Bolton & the quasar episodic lifetime tQ but clearly sets a lower Haehnelt 2007; Worseck et al. 2014). A single quasar limit on it (tage tQ). therefore causes a significant increase over the back- ≤ Constraints for sustained activity over at least 106 yr ground only within a relatively small zone of influence. comes e.g. from the presence of enormous Lyα nebulae At z >2.7, before Heii reionization is complete, helium around luminous quasars (Cantalupo et al. 2014; Hen- offersamuchlargercontrastsincethebackgroundislow, nawi et al. 2015). On slightly larger scales geometric themeanfreepathshortandsinglequasarscanproduce constraints of t (cid:46) 8Myr6 might be derived from fluo- a stronger enhancement over the background, resulting Q rescent Lyα emission of galaxies caused by the UV radi- inamuchlargerregionwherethetotalionizationrateis ation of a nearby quasar as claimed by Cantalupo et al. higherthantheUVbackground,farbeyondtheregionin (2012),Trainor&Steidel(2013)orBorisovaetal.(2016). which the host halo causes a substantial enhancement of Gon¸calvesetal.(2008)investigatemetalabsorptionsys- the cosmic density field (Khrykin et al. 2016b). Highly temsinaz =2.5quasartripletandclaimoverionization saturated Heii spectra reaching effective optical depths due to the foreground quasars’ radiation and hence life- τeff 5 on scales ∆z = 0.04 (Worseck et al. 2016) con- time constraints of tQ > 25Myr and 15 < tQ < 33Myr, strai≈ntheHeiifractiontoxHeII ≈0.02atz ≈3(Khrykin butdonotplaceempiricalconstraintsonthegasdensity, etal.2016b,a). Undertheseconditions,arelativelysmall considertheeffectsofcollisionalionization,orexplorethe decrease in the Heii fraction can cause a large increase effects of non-solar relative abundances. Therefore, cur- in the transmission. rent constraints on the quasar episodic lifetime suggest Limited by the Galactic and extragalactic Hi Lyman valuesbetween106 and108 yr(butsee,Schawinskietal. continuum(e.g.Worsecketal.2011),thez >2HeiiLyα 2015). Allstudiesgivingmorespecificconstraintsareei- absorption is observable from space with far UV (FUV) ther based on single or very few objects, or are indirect spectrographs. Over the last two decades the Hubble and therefore rely on assumptions. Space Telescope (HST) has obtained Heii Lyα spectra The direct influence of quasars on the IGM can be foralimitednumberof Heii-transparentquasars(Jakob- clearly detected in quasar spectra as statistically lower sen et al. 1994; Reimers et al. 1997; Heap et al. 2000; IGM Lyα absorption in the vicinity of a quasar. This Shull et al. 2010; Worseck et al. 2011; Syphers & Shull enhanced ionization is called the line-of-sight proximity 2013,2014;Zhengetal.2015;Worsecketal.2016). Many effect(Carswelletal.1982;Bajtliketal.1988;Scottetal. of these exhibit prominent line-of-sight proximity effects 2000; Dall’Aglio et al. 2008; Calverley et al. 2011). The whichitselfsetsaconstraintonthequasarlifetime,simi- presenceofaline-of-sightproximityeffectinquasarspec- lartothehydrogencase(Khrykinetal.2016b). ForHeii, tra sets a lower limit on the episodic lifetime of quasars. this is however approximately three orders of magnitude To cause an observable effect in the IGM they have to longer (tQ (cid:38)10Myr) than for Hi due to the lower HeII photoionizing background (ΓHeII 10 15s 1, Faucher- UVB ∼ − − 6 Giventheluminosityofhyperluminousquasars(mr 16mag Gigu`ereetal.2009;Haardt&Madau2012)andtherefore at z = 3) and the typical sensitivity of narrowband≈surveys longer equilibration timescale (Khrykin et al. 2016b). Ta(mcrhaNiienBvoe(cid:46)rd&2in6S.5ltoemnidgaegiln,t25e0g×1r3a1)t0,i−otnh1s8e(efl∼rugo5sr−hes1)ccoemnnt−1e02mmfoisrscilaoans∆sfλtreo=lmes5c10o.p0˚A(cid:48)e(cid:48)sfisl(iteze.ergd). waTshfeoufinrsdtienvitdheencHeefoiirsaigHhetliiinteratnosvQer0s3e0p2r−o0x0im3itthyaetffeexct- galaxies can only be detected for separations from the illuminat- hibitsatransmissionspikeatz =3.05(Heapetal.2000) ing quasar smaller than 1.2Mpc. This allows one to constrain and a foreground quasar at practically the same redshift quasarlifetimes(cid:46)7.8My≈r. Therequiredintegrationtimeincreases (Jakobsen et al. 2003). This prominent example repre- asthefourthpoweroftheprobedseparation. Statistical Detection of the Heii Transverse Proximity Effect 3 sents the prototype case for the Heii transverse prox- ticular important to constrain the quasar lifetime. At imity effect. From the required transverse light crossing redshift z 3, this angular separation corresponds to ≈ timeinthisassociation,onecaninferageometricallimit a physical distance of only 4.7pMpc or a light crossing of the quasar lifetime of t 10Myr. Although this es- time of 15Myr. Therefore, we also conducted the shal- Q ≈ timate is only based on a single object, it demonstrates lowerandwidersurveyon4mclasstelescopesextending the feasibility of deriving lifetime constraints using the out to ∆θ 90. (cid:48) Heii transverse proximity effect. In the same sightline The sam≈ple from our own surveys was complemented Worseck&Wisotzki(2006)comparedHeiiandHispec- by quasars from the Sloan Digital Sky Survey (SDSS, tra and computed the hardness of the radiation field, York et al. 2000) and the Baryon Oscillation Spectro- basedontherelativeabsorptionstrengthofthetwoions, scopic Survey (BOSS, Eisenstein et al. 2011; Dawson sensitivetoionizationat1Ryand4Ry. Theydetectthe et al. 2013), specifically from the twelfth data release proximity effect for at least one other foreground quasar (DR12, Alam et al. 2015) spectroscopic quasar catalog which sets a lower lifetime limit of 17Myr. Syphers & (Pˆaris et al. 2016). In the context of our study, SDSS Shull (2014) claim a transverse proximity effect for an- and BOSS cover a similar parameter space (magnitude otherquasar34MlyawayfromtheQ0302 003sightline, r (cid:46) 21mag) as our wide survey. However, the SDSS − but this case is degenerate with the proximity effect of quasarselectionissubstantiallyincompleteatz 3(Fan Q0302 003 itself. In the limited sample of Heii spectra 1999; Richards et al. 2002b, 2006; Worseck & P≈rochaska − and foreground quasars Furlanetto & Lidz (2011) see in- 2011). This makes it necessary to conduct our own wide dications against very long (t >108 yr) and very short survey and find the quasars not identified by SDSS. Q (t < 3 106 yr) quasar lifetimes, by simply counting Q Heiitran×smissionspikesundertheassumptionthatthey 2.1. Heii Sightlines are associated with foreground quasars. Our foreground quasar survey targeted fields around The installation of the Cosmic Origins Spectrograph 22Heii-transparentquasarsobservedwithHST/COSor (COS) on HST with unprecedented FUV sensitivity of- HST/STIS(Table1). Worsecketal.(2016)describethe fers for the first time the opportunity for an extended homogeneous data reduction and analysis of 16 of these, surveyof Heiitransparentquasars(Worsecketal.2011; including a much improved HST/COS background sub- Syphers et al. 2012; Zheng et al. 2015; Worseck et al. traction (dark current, quasi-diffuse sky emission, scat- 2016). We use a dataset of 22 Heii sightlines and com- tered light) and suppression of geocoronal contamina- plement it with a dedicated optical spectroscopic survey tion compared to the default CALCOS pipeline reduc- to discover foreground quasars around these sightlines. tions from the HST archive. This improved reduction These allow us for the first time to statistically quan- ensures that weak excess Heii transmission due to the tify the Heii transverse proximity effect and to derive a transverse proximity effects is not affected by zero-level constraint on the quasar episodic lifetime. The survey calibration errors. Five Heii-transparent quasar sight- is described in §2 and a few individual objects are dis- lines observed in HST Cycle 20 were reduced analo- cussed in §3. In §4 we describe our statistical search for gously (Worseck et al., in preparation). Almost all spec- the presence of a Heii transverse proximity effect and tra (20/22) have been taken with the HST/COS G140L constrain the quasar lifetime. In § 5 we investigate the grating (R = λ/∆λ 2000 at 1150˚A, 0.24˚A per object-to-object variation ofthe transverse proximity ef- ∼ (cid:39) Nyquist-binned pixel). Their signal-to-noise ratio per fect within our sample. We discuss the implications of binnedpixelatHeiiLyαofthebackgroundquasarvaries our findings in §6 and summarize in §7. between 2 and 15, mostly depending on whether the Throughout the paper we use a flat ΛCDM cosmology sightlinewasknownbeforetobeHeii-transparent(Shull with H =70kms 1Mpc 1, Ω =0.3 and Ω =0.7 0 − − m Λ et al. 2010; Syphers & Shull 2013, 2014; Zheng et al. which is broadly consistent with the Planck Collabo- 2015), or had been discovered in recent HST/COS sur- ration et al. (2015) results. We use depending on the veys (Worseck et al. 2011; Syphers et al. 2012; Worseck situationproperdistancesorcomovingdistancesandde- et al. 2016). The sightline to HS 1157+3143 was ob- note the corresponding units as pMpc and cMpc, re- served with the HST/STIS G140L grating (R 1000, spectively. We denote the different quasar lifetimes with ∼ 0.6˚Apixel 1; Reimers et al. 2005). For the Q 0302 003 tdc for the duty cycle (total activity over Hubble time), sightline w−e used the higher-quality HST/COS G1−30M t for the quasar episodic lifetime, and t for the age oQf a quasar. Magnitudes are given in the aAgBe system. data (R 18,000, 0.03˚A per Nyquist-binned pixel; Syphers &≈ Shull 201(cid:39)4) instead of the HST/STIS data 2. DESCRIPTIONOFTHESURVEY presented in Worseck et al. (2016). We checked our re- As part of a comprehensive effort to study Heii reion- ductionbycomparingthemeasuredHeiieffectiveoptical ization, we conducted an extensive imaging and spec- depthsintheQ0302 003sightlinetothosepresentedby − troscopic survey to identify foreground quasars around Syphers & Shull (2014), finding very good agreement. 22 Heii-transparent quasar sightlines for which science- AsdetailedinWorsecketal.(2016)wesuppressedgeo- grade HST/COS spectra are available. The survey con- coronal emission lines by considering only data taken sisted of a deep narrow survey covering the immediate during orbital night or at restricted Earth limb angles vicinity of the Heii sightline (∆θ 10) down to a mag- in the affected spectral ranges. Nevertheless, decontam- (cid:48) ≤ nitudeofr 24.0magbasedondeepimagingandmulti- inated regions had to be excluded from our analysis due objectspect≤roscopicfollow-upon8m-classtelescopes,as to the vastly reduced sensitivity to strong Heii absorp- well as a wider survey targeting individual quasars on tionandsometimesextremelyweakgeocoronalresiduals 4m telescopes. Finding quasars that have a separation in the coadded data. Geocoronal Lyα emission was ex- from the Heii sightline of more than 10 is in par- cluded. In addition we apply a liberal signal-to-noise (cid:48) ≈ 4 T. M. Schmidt et al. Table 1 OverviewoftheFUVspectrausedforthiswork. Quasar Instrument R Program PI References PC0058+0215 COSG140L 2000 11742 Worseck Worsecketal.2016 HE2QSJ0233 0149 COSG140L 2000 13013 Worseck Worsecketal. inprep.,thispaper Q0302 003 − COSG130M 18000 12033 Green Syphers&Shull2014,thispaper SDSSJ−0818+4908 COSG140L 2000 11742 Worseck Worsecketal.2016 HS0911+4809 COSG140L 2000 12178 Anderson Syphersetal.2011,2012;Worsecketal.2016 HE2QSJ0916+2408 COSG140L 2000 13013 Worseck Worsecketal. inprep,thispaper SDSSJ0924+4852 COSG140L 2000 11742 Worseck Worsecketal.2011,2016 SDSSJ0936+2927 COSG140L 2000 11742 Worseck Worsecketal.2016 HS1024+1849 COSG140L 2000 12178 Anderson Syphersetal.2012;Worsecketal.2016 SDSSJ1101+1053 COSG140L 2000 11742 Worseck Worsecketal.2011,2016 HS1157+3143 STISG140L 1000 9350 Reimers Reimersetal.2005;Worsecketal.2016 SDSSJ1237+0126 COSG140L 2000 11742 Worseck Worsecketal.2016 SDSSJ1253+6817 COSG140L 2000 12249 Zheng Syphersetal.2011;Zhengetal.2015;Worsecketal.2016 SDSSJ1319+5202 COSG140L 2000 12249 Zheng Zhengetal.2015;Worsecketal.2016 Q1602+576 COSG140L 2000 12178 Anderson Syphersetal.2012;Worsecketal.2016 HE2QSJ1630+0435 COSG140L 2000 13013 Worseck Worsecketal. inprep.,thispaper HS1700+6416 COSG140L 2000 11528 Green Syphers&Shull2013;Worsecketal.2016 SDSSJ1711+6052 COSG140L 2000 12249 Zheng Zhengetal.2015;Worsecketal.2016 HE2QSJ2149 0859 COSG140L 2000 13013 Worseck Worsecketal. inprep.,thispaper HE2QSJ2157−+2330 COSG140L 2000 13013 Worseck Worsecketal. inprep.,thispaper SDSSJ2346 0016 COSG140L 2000 12249 Zheng Syphersetal.2011;Zhengetal.2015;Worsecketal.2016 HE2347 43−42 COSG140L 2000 11528 Green Shulletal.2010;Worsecketal.2016 − (S/N)cut. SincemostofourpathlengthiswithinGunn- observedwithLBT/LBC.ForthefieldofHE2347 4342 − Peterson troughs where the flux is consistent with zero weobtainedmultibandimaging(Ugri)withthe36 36 (cid:48) (cid:48) × we cannot apply a real limit on the S/N but instead Mosaic II camera at the 4m Blanco Telescope at the require the continuum level to correspond to at least Cerro Tololo Inter-American Observatory (Muller et al. five counts per spectral bin. This mostly affects the 1998). The field of SDSSJ1237+0126 was imaged in short wavelength end of the spectra where the efficiency gri with Mosaic 1.1 at the 4m Mayall Telescope at the of the instrument drops dramatically. The background Kitt Peak National Observatory, and in U with Mag- quasar proximity zones were excluded by measuring the ellan/Megacam (McLeod et al. 2015). Exposure times dropping Heii transmission from the proximity zone to were increased to achieve an imaging depth similar to the strongly saturated Heii absorption in the IGM (e.g. our LBC observations. Zheng et al. 2015). By excluding the entire background Data reduction, mosaicing of the individual dithered quasar proximity zone we may have excluded the trans- exposures, stacking, astrometry and photometry was verse proximity zones of foreground quasars at similar doneusingourowncustompipelinebasedonIRAFrou- redshifts (Worseck & Wisotzki 2006; Syphers & Shull tinesandSCAMP,SWarfandSExtractor7 (Bertin2006; 2014), but as these cases are degenerate and therefore Bertin et al. 2002; Bertin & Arnouts 1996). An example require detailed modeling we make sure that our sample forareducedrbandimageisgiveninFigure1. Forfields isnotaffectedbyanybackgroundquasarproximityzone. covered by SDSS the astrometric solution is tied to the SDSS reference frame using SDSS star positions, while thephotometriccalibrationistiedtoSDSSubercalpho- 2.2. Deep Survey on 8m class Telescopes tometry (Padmanabhan et al. 2008) to define the zero For our deep imaging survey we used the Large point and to correct for non-photometric observations. Binocular Cameras at the Large Binocular Telescope For HE2347 4342 that lies outside the SDSS footprint (LBT/LBC,Spezialietal.2008;Giallongoetal.2008)to we had to re−ly on the USNO-B1.0 catalog and photo- obtain optical multiband photometry (USpec, g, r and i) metricstandards. Weusedthereddestbands(r ori)for over an area of 23(cid:48) 25(cid:48) approximately centered on the source detection and used forced photometry to extract targeted Heii sight×line. Imaging for 10 Heii sightlines fluxesattheidenticalpositionsintheotherbands. Mag- was obtained over several runs in 2009, 2011 and 2013 nitudes were corrected for Galactic extinction assuming (Table 2). We observed in binocular mode with USpec the reddening terms from SDSS (Stoughton et al. 2002) and g (r and i) filters on the blue (red) camera. Individ- and E(B V) for the background quasar from Schlegel ual exposure times were short (around 120 seconds) to etal.(199−8),i.e. notaccountingforreddeningvariations limit saturation of bright stars. Depending on the field across the field. Star-galaxy classification was also done and the observing conditions, total exposure times were inthereddestobservedbandssincetheyhadthebestim- 70 to 100 minutes in the USpec filter, 10 to 30 minutes in agequalitywithafullwidthathalfmaximum(FWHM) g,18to35minutesinr and38to54intheifilter. With of 0.8 . The 5σ point source imaging depth of our (cid:48)(cid:48) this strategy we reached a homogeneous depth in USpec, LB(cid:39)T/LBC images is typically 26.5mag in USpec and which is the limiting factor for our target selection due g, and 26.0mag in r and i, re(cid:39)spectively. to the expected colors of z 3 quasars (USpec g 2). Selec(cid:39)tion of quasar candidates was done by applying ∼ − ∼ Duetotheobservationinbinocularmode, theredfilters cutsintheU g vs.g rcolorspaceasshowninFig- Spec naturally reached a sufficient depth. − − Due to declination and scheduling constraints, the fields of HE2347 4342 and SDSSJ1237+0126 were not 7 http://www.astromatic.net/software − z=2.51 Statistical Detection of the Heii Transverse Proximity Effect 5 450 z=2.99 5 0 z=2.60 40 0 z=2.62 z=2.20 z=3.11 0) 350 00 z=2.58 2 (J z=2.23 c e D PC0058+0215 z=2.89 z=2.73 z=2.73 30 0 z=2.51 +2 25 ◦ 0 z=2.89 z=2.54 20s 01m00s 40s 1h00m20s z=2.64 RA(J2000) qFuigasuarreis1.maLrBkeTd/LinBtCher-cbeanntedri(mreadg)e.oTfhtheeafipeplrdoxairmouantedpPoCsit0i0o2n8s+o0f2t1h5e(f2o3u(cid:48)r×qu25a(cid:48)d,rran<ts2o6fmthaegVaItM5OσS).fiTelhdeoHfeviiie-wtraanrsepianrdeinctatbeadc.kgIrnouthnids areaourdeepspectroscopicsurveydiscoveredsixforegroundquasars(yellow). TwoadditionalquasarsoutsidetheVIMOSfootprintwere foundbyourwidesurveywithNTT/EFOSC2(blue)andtwoquasarsarefromSDSS(gray). z=2.04 ure 2. A theoretical color track and contours have been pectedrangeincolorforz 3quasarswhilelimitingthe ∼ computed from SDSS mock photometry of quasars in- stellarcontamination. Lower-prioritycandidateswerese- cluding the spread in color due variations in the spectral lected from an extended selection box (g r)<0.6 that − energydistribution(SED)andIGMabsorption(Worseck overlapped with the stellar locus (Figure 2, dashed re- & Prochaska 2011). The stochastic IGM Lyman con- gion). tinumm absorption leads to a large scatter around the Spectroscopicverificationofthequasarcandidateswas median track, and in particular for z 3 the range of done with the VIsible MultiObject Spectrograph (VI- ≈ expected quasar colors overlaps substantially with the MOS, Le F`evre et al. 2003) at the Very Large Telescope stellar locus, highlighting again the difficulties of quasar (VLT), whose four 7 8 quadrants cover most of the (cid:48) (cid:48) × color selection at these redshifts (Richards et al. 2002b; 23 25 LBC field of view (Figure 1). Custom designed (cid:48) (cid:48) × Worseck & Prochaska 2011). We used a selection box of focal-plane slit masks were used to simultaneously take the form low-resolution spectra (LR Blue grism, R 180, wave- ≈ (U g)>0.3 length range 3700–6700˚A) of 32 candidates per quad- Spec− ∧ rant. Eachofour10imagedfie≈ldswascoveredbyasingle [(g r)<0.25 (g r)<0.5(U g) 0.25] − ∨ − Spec− − VIMOS pointing with a typical exposure time of 2 30 (1) × minutes. The data were reduced with the standard Es- which is visualized in Figure 2. It accounts for the ex- 6 T. M. Schmidt et al. 1.5 with the DEep Imaging Multi-Object Spectrograph (DEIMOS, Faber et al. 2003) at Keck Observatory. We however did not find any additional quasar in this at- 1.0 tempt. ag) 2.3. Wide Survey on 4m Class Telescopes m ( 0.5 3.3 r Ourdeepmulti-objectspectroscopywascomplemented − g 3.0 byindividuallongslitobservationsofbrighterquasarcan- 2.7 didates at larger angular separations (10 (cid:46) ∆θ (cid:46) 90). 0.0 (cid:48) (cid:48) Duetothelargeareaontheskyandsparsetargetdistri- Selectionbox bution, longslit spectroscopy of single targets is prefer- Mediancolortrack able to multi-object spectroscopy. It requires, however, 0.5 80%completeness − amuchhigherselectionefficiencythanpossibleforz 3 0 1 2 3 4 ≈ quasarsfromopticalphotometryalone. Weselectedcan- u g(mag) − didates from the XDQSOz catalog (DiPompeo et al. Figure 2. Candidate selection for our deep survey. Shown are 2015) based on the extreme deconvolution technique all point-like objects with photometric detections in all bands (S/N> 5) as dots or contours (blue). A theoretical color track (Bovy et al. 2011, 2012) and the KDE catalog (Richards for 2.7 < z < 3.5 quasars and corresponding completeness con- et al. 2015). Both catalogs are based on SDSS ugriz toursareshowninblack(Worseck&Prochaska2011). Weselected imaging but also incorporate infrared photometry from high-priorityquasarcandidatesfromaboxshowninredandlower- the Wide-field Infrared Survey Explorer (WISE, Wright prioritycandidatesfromthedashedregion. Confirmedquasarsare overplottedasstarsymbols. Colorsfromgreentoorangeindicate et al. 2010). The WISE 3.6µm and 4.5µm bands are theredshift(z<2.5, 2.5<z<3, 3<z<3.5, z>3.5). Quasars sensitive to the emission of hot dust surrounding AGN havingonlyalimitinU (S/N<5)haveablackticktotheright. (Stern et al. 2012; Assef et al. 2013). This characteris- tic feature allows for very efficient separation of quasars 4 NTT/EFOSC2 Field: PC0058+0215 from galaxies and stars. Both catalogs give a photomet- RA=15.0655◦ ric redshift estimate or even a probability distribution. 3 DE=+2.4008◦ r=20.7mag Weusedthisinformationtomaximizetheprobabilityfor 2 z=2.890 candidatestobeconfirmedwitharedshiftcoveredbythe Heii Lyα absorption spectra of the background quasars 1 andprioritizedtargetsaccordingly. Wealsodesignedour 1− 0 wide survey to maximize the expected Heii transverse ˚A 12scm−−00..46 VLT/VIMOS Field: PRDCAEr0===05+182522+...375040m29213a65◦g◦ pprarhtooextioΓmmHQieteSyItOIriectffheracettd.sehBviefatrs,yewdoefonetshtteihsmeeappteoudstiattithoivnee,pbqhruoiagtoshaitornnseiszwsaotauinolddn erg z=2.730 cause at the background sightline (see § 2.6 for a def- 17−0.2 inition of ΓHeII). We then selected and prioritized ac- 0 QSO in10.0 cordingtocutsinΓHQeSIOI,primarilytargetingobjectswith Flux 5 SDSS Field: PC0058+0215 expected ΓHQeSIOI >0.5×10−15s−1. 4 RA=15.3541◦ For spectroscopic confirmation we used the ESO 3.5m 3 DEr==+212..166m3a7g◦ New Technology Telescope Faint Object Spectrograph z=2.600 and Camera (NTT/EFOSC2, Buzzoni et al. 1984) and 2 the Calar Alto Observatory (CAHA) 3.5m telescope 1 TWIN spectrograph. We used the EFOSC2 grating 0 g782 (R = 180–450, wavelength range 3700–9000˚A). 4000 4500 5000 5500 6000 6500 For TWIN we used only the blue arm with grating T13 WavelengthinA˚ (R = 620–1000, wavelength range 3900–7000˚A). A slit Figure 3. Spectraofthreerepresentativeforegroundquasarsob- width between 1.2 and 1.5(cid:48)(cid:48) was used and the slit ori- servedwithdifferenttelescopesandinstrumentsusedforthisstudy. ented at the parallactic angle. With these setups the All three quasars are located around the Heii sightline toward limiting magnitude for both instuments was r 21mag PC0058+0215 and marked in Figure 1. The observed flux den- ≈ sityisshowninblueandthe1σ errorarrayinred. Overplottedis for the longest used integration times of one hour. The aquasartemplate(VandenBerketal.2001)shiftedtotheredshift CAHA/TWIN spectra were taken in 32 nights between oftheobservedquasarsbutnotadaptedtomatchthedifferentline November 2014 and August 2015, while NTT/EFOSC2 strengthsoftheshownquasars. observations were performed during 5 nights in Decem- oRexVIMOSpipeline8 towhichweaddedcustommask- ber 2014. ing of zeroth-order contamination which is unavoidably Data were reduced using the XIDL Low-Redux pack- present in the raw frames. An example spectrum of a age9. An example spectrum of a quasar confimed with quasar discovered by our VIMOS survey is shown in the NTT/EFOSC2 is shown in Figure 3. Overall, 36% of middle panel of Figure 3. the observed targets were confirmed as quasars and 11% For the field of Q0302 003, additional quasar can- had a redshift within the covered Heii Lyα forest of the didates outside our VIM−OS pointing were observed background quasar. 8 http://www.eso.org/sci/software/cpl/esorex.html 9 http://www.ucolick.org/~xavier/LowRedux/ Statistical Detection of the Heii Transverse Proximity Effect 7 As part of the wide survey we also verified a 912˚A and extrapolate to the Heii Lyman limit at 228˚A quasar with an uncertain redshift in the vicinity assuming the specific luminosity to follow a power law of HS1700+6416 (Syphers & Shull 2013). We L να. The quasar spectral slope α beyond 912˚A is ν used the Keck Low Resolution Imaging Spectrometer not∝very well constrained since the frequencies between (Keck/LRIS, Oke et al. 1995; McCarthy et al. 1998) to the extreme UV and soft X-rays are basically unobserv- confirmitsredshift. Datawereaswellreducedusingthe able. We adopt a value of α = 1.7 as determined by XIDL Low-Redux package. Lusso et al. (2015) (however with−a large uncertainty of 0.6),whichisconsistentwiththeindependentmeasure- 2.4. Selection from SDSS and BOSS ± mentofStevansetal.(2014),aswellastheslopebetween For Heii sightlines within the SDSS footprint we also UVandX-rayregime(Lussoetal.2015),butdiffersfrom use quasars from the SDSS DR12 catalog (Alam et al. the value of α = 0.73 0.26 reported by Tilton et al. 2015; Pˆaris et al. 2016). This has the advantage that we (2016). The unce−rtainty±in α and the long range of ex- can include large numbers of quasars out to very large trapolation from 912˚A beyond 228˚A causes substantial separation from the background sightline. For our ini- uncertainty in the inferred photoionization rate, e.g. up tial input catalog we selected all quasars within 240 (cid:48) toafactorof2.5forα= 1.7 0.6. However,thismostly of the background sightlines and with z > 2.5. For affects the absolute scali−ng of±ΓHeII. A relative compar- HS1700+6416 whose COS spectrum covers lower red- QSO ison of different foreground quasars will not be severely shifts we adapted the latter criterion accordingly. This affected. ensures that we include all objects that may contribute WeconvertquasarluminositytofluxdensityF atthe significantly to the ionizing background at the location ν background sightline according to of the background sightline. At later stages in our anal- ysiswewillimposecutsontheexpectedphotoionization 1 D rate at the background sightline. The vast majority of Fν =Lν 4πD2 e−λmfp . (2) the selected quasars have r <21mag. We note that Q0302 003 and SDSSJ2346 0016 ThisisafunctionofthetransversedistanceDbetween − − lie within SDSS Stripe 82 that was imaged multiple theforegroundquasarandthebackgroundsightlinemea- times, offering photometry approximately two magni- sured at the redshift of the foreground quasar, and the tudes deeper than the standard SDSS imaging (Abaza- meanfreepathtoHeii-ionizingphotonsintheIGMλ . mfp jian et al. 2009), and was also targeted by addi- However, quasars change the ionization state of the sur- tional spectroscopy, using different selection algorithms roundingIGMbycreatinglargeproximityzones. There- (e.g. variability, see Butler & Bloom 2011; Palanque- fore, the mean free path relevant for us is not the one at Delabrouille et al. 2011). We actually find a higher den- random locations in the IGM but an effective mean free sity of foreground quasars near the SDSSJ2346 0016 pathwithintheproximityzone(e.g.McQuinn&Worseck − sightline but not for Q0302 003. 2014; Davies & Furlanetto 2014; Khrykin et al. 2016b). − The Heii transverse proximity effect should in principle 2.5. Systemic Quasar Redshifts be able to constrain the mean free path, but at present Quasarredshiftsdeterminedfromtherest-frameultra- it is not well constrained by observation or simulation. violetemissionlines(redshiftedintotheopticalatz 3) Current studies (e.g Davies & Furlanetto 2014) suggest candifferbyupto1000kms 1 fromthesystemicfra∼me, λ (cid:38) 50cMpc at z 3, longer than the scales probed − mfp because of outflowing/inflowing material in the broad by our study (D (cid:46) 40≈cMpc). We therefore ignore IGM lineregionsofquasars(Gaskell1982;Tytler&Fan1992; absorptionfornowandassumepuregeometricaldilution VandenBerketal.2001;Richardsetal.2002a;Shenetal. of the radiation by setting λ = . mfp ∞ 2007, 2016; Coatman et al. 2017). We estimate systemic Implicitly,wehaveassumedisotropicemissionandinfi- redshiftsbycombiningtheline-centeringprocedureused nite quasar lifetime with constant luminosity. Although in Hennawi et al. (2006) with the recipe in Shen et al. current constraints suggest t (cid:46) 108yr, we assume for Q (2007) for combining measurements from different emis- simplicity no time dependence in our fiducial model and sionlines. Theresultingtypicalredshiftuncertaintiesus- constrain t later. Also, the widely used unified AGN Q ing this technique are in the range σz 270–770kms−1 models (see, e.g. Antonucci 1993; Elvis 2000) assume a (cid:39) depending on which emission lines are used. But given large-scaleanisotropyoftheUVandopticalemissionand thelowS/Nratioof 5ofmanyofourspectra, wecon- relatethedichotomybetweenbroadlinequasars(TypeI) ≈ servatively assume our estimates of the systemic quasar andAGNdisplayingonlynarrowemissionlines(TypeII) redshift to be not better than 1000kms 1. to the presence of obscuring material that blocks the di- − 2.6. Estimate of the Heii Photoionization Rate rect view on the accretion disc and broad line region if observed from certain directions. Studies focusing on To estimate the impact a foreground quasar has on the numbers of TypeIs vs. TypeIIs suggest (with large the ionization state of the IGM we calculate the Heii uncertainties) approximately equal numbers (e.g. Brusa photoionization rate at the location of the background etal.2010;Lussoetal.2013;Marchesietal.2016)which sightline. We use the Lusso et al. (2015) quasar tem- suggestsopeninganglesof 120degreesifoneassumesa plate which is based on HST UV grism spectroscopy, bi-conical emission. In con≈sequence, the TypeI quasars corrected for IGM absorption and covers the restwave- from our survey might be obscured towards parts of the length range down to 600˚A. We redshift and scale this background sightlines. However, we have no way to ei- template to match our r band photometry which always ther infer the orientation or the exact opening angle of falls redwards of Lyα and measures the quasar contin- theforegroundquasarandthereforenootherchoicethan uum flux. From the scaled template we infer the flux at toassumeisotropicemissionforourfiducialmodel. Ade- 8 T. M. Schmidt et al. tailedmodelingofobscurationeffectsisbeyondthescope 10−13 VLT/VIMOS of this paper. NTT/EFOSC Wethereforeremainfornowwiththesimplestisotropic CAHA/TWIN model and convert UV flux density to Heii photoioniza- ) Literature tion rate by 1(s−O10−14 SDSS/BOSS ΓHeII = ∞ Fν σν,HeII dν Fνo σHeII (3) ΓQS QSO (cid:90)νo hPν ≈ hP (3−α) Rate in which h denotes Planck’s constant and ν the fre- n P 0 o quency of the Heii ionization edge. For the second part zati10−15 we have assumed the quasar spectra to follow the power ni lfaowr tdheescHriepitiiocnroFsνs-=secFtνio0n×t(hν/eν0a)pαpirnotxriomdautcieodnaσbove and Io ν,HeII ≈ eσtHaeIl.I,01×99(6ν)/.ν0)−3withσHeII,0 =1.58×10−18 cm2(Verner 10−161 2 4 8 16 32 64 This calculation condenses the observed parameters TransverseSeparationD(pMpc) (angularseparationfrombackgroundsightline,apparent Figure 4. Propertiesofourforegroundquasarsample. Thepanel magnitude, redshift) into one convenient number which shows the transverse separation of the foreground quasars from can be used to estimate the possible influence of a fore- thebackgroundsightlineandtheirestimatedphotoionizationrate ground quasar on the background sightline. Although ΓHeII atthelocationofthebackgroundsightline. Colorsindicate QSO there is substantial uncertainty in the spectral slope and theoriginoftheobjects. Horizontaldottedlinesgivetherangesfor our calculation of the photoionization rate assumes sim- which we expect the objects to have a substantial impact on the plifications like infinite lifetime, no IGM absorption and background sightline (ΓHQeSIOI >2×10−15s−1) or at least include completely ignores obscuration effects, ΓHQeSIOI represents themfortheextendedselections(ΓHQeSIOI >0.5×10−15s−1). Only objects shown as green framed squares fall in usable parts of the an important physical quantity and is (on average) a Heii spectra (see Figure 5) and can be included in our statistical good proxy for the strength of the transverse proximity analysis. AnarrowindicatestheJakobsenetal.(2003)quasar. effect (§ 4.3). We will use ΓHeII extensively throughout QSO rates (ΓHeII >2 10 15s 1). Figure 4 also shows that the paper to select subsamples from our quasar catalog. QSO × − − with our wide survey (blue and orange symbols) we find Althoughweimposenouniversalthreshold, wetypically numerousquasarsthathavenotbeendiscoveredbySDSS consider foreground quasars with ΓHeII >2 10 15s 1, QSO × − − and BOSS. This substantially expands our sample in i.e. similar to or exceeding current estimates of the UV the region of interest, meaning foreground quasars with Mbaacdkagurou2n01d2;(ΓKHUheVIrBIyk≈in 1e0t−a1l5.s2−0116abt).z ∼We3,cauHtiaoanrdttha&t ΓHQeSIOI >0.5×10−15s−1. In fact, the two usable quasars with the highest photoionization rates in our sample aUVdirbeacctkgcroomupnadrimsoondeolfs oisuraffΓeHQceStIOeIdebstyimsuatbesstatnotiaglloubna-l (ΓHQeSIOI =18.2 and 16.9×10−15s−1, see § 3.3 for details) are from our wide survey. Such bright (r 19) quasars certainty due to different quasar SED parameterizations ∼ outside the field of view of our deep survey are particu- (see Lusso et al. 2015 for a discussion). It gives, how- larly important to constrain the quasar lifetime. ever, a rough indication of the value above which we Spectraofall Heii sightlines areshownin Figure5to- expect a foreground quasar to substantially impact the gether with their respective foreground quasars. Gaps IGM ionization structure. We also include objects with in the spectra are due to masking of geocoronal resid- e0ff.5ec×t1o0f−w1e5ask−e1r<quΓaHQsaeSrIOIs.<Se2eק140.3−1fo5rs−a1dettoaielexdplaonrealytshies uaraelsin(§dic2a.1t)e.dTwhitehraegliiognhst gthreaetnpsatsrsipoeuartqtuhaelitbyotctroimteroiaf and § 5 for the impact of individual objects. the plots. Overplotted are the positions of foreground quasars. 2.7. Final Quasar Sample After restricting the combined quasar sample to have 3. SEARCHFORTHETRANSVERSEPROXIMITYEFFECT science-grade Heii Lyα coverage along the background ININDIVIDUALSIGHTLINES wsigithhtl6in6efo(r§eg2r.o1u)nadndquΓaHQsaeSrIOIs,>su0m.5m×ar1iz0e−d15ins−T1abwlee2e.ndHeurpe anAallythsiosuogfhththeeHmeiaiintraanimsveorfsethpirsoxwimoriktyiseffaecstt,awtiestwicialll we have included quasars discovered in previous dedi- briefly discuss a few special objects to build intuition cated surveys in these fields (Jakobsen et al. 2003; Stei- about the proximity effect signal. deletal.2003;Hennawietal.2006;Worseck&Wisotzki 2006; Worseck et al. 2007; Syphers & Shull 2013). Our 3.1. HS1700+6416 survey yields in total 131 new quasars in the projected Near HS1700+6416 four foreground quasars were dis- vicinity of the 22 Heii-transparent quasars (Table 3), 27 covered by Syphers & Shull (2013), one of which had an wofitwhhsicchienhcaev-egrΓaHQdeSeIOIH>ei0i.5Ly×α1a0b−s1o5rsp−t1ioann.dfallinregions udnetceecrtteadinermedissshioifntalisnseig.nmWeintht(ozu=r K2.e6c2k5/)LdRuIeStsopaecstirnugmle Our final foreground quasar sample is illustrated in we confirmed this quasar at a redshift z =2.628 0.003, Figure 4. With our deep survey (yellow symbols) we are matching a broad Heii transmission peak in th±e COS able to detect quasars 2mag fainter than SDSS, and spectrum. OurCAHAsurveydiscoveredthreeadditional ≈ preferentially discover quasars close to the background foregroundquasarsatz >2.3,oneofwhichlinesupwith sightline (D (cid:46) 8pMpc) with still high photoionization a narrow Heii spike at z =2.614. Statistical Detection of the Heii Transverse Proximity Effect 9 Table 2 OverviewoftheHeiisightlinesandthenumberofforegroundquasarswithHeiiLyαcoverageinthebackgroundsightline. Name R.A.(J2000) Decl. (J2000) zBG LBT/LBC NTT/ CAHA/ SDSSa Literatureb hh:mm:ss hh:mm:ss VLT/VIMOS EFOSC2 TWIN PC0058+0215 01:00:58.40 +02:31:32.0 2.890 2 1 0 1 HE2QSJ0233 0149 02:33:06.01 01:49:50.5 3.314 1 0 0 4 Q0302 003 − 03:04:49.85 −00:08:13.4 3.286 1 0 2 2 SDSSJ−0818+4908 08:18:50.02 −+49:08:17.2 2.957 0 1 HS0911+4809 09:15:10.00 +47:56:59.0 3.350 − − 2 2 HE2QSJ0916+2408 09:16:20.85 +24:08:04.6 3.440 − −2 3 SDSSJ0924+4852 09:24:47.36 +48:52:42.8 3.027 1 SDSSJ0936+2927 09:36:43.51 +29:27:13.6 2.930 − − 1 HS1024+1849 10:27:34.13 +18:34:27.6 2.860 0 0 0 0 SDSSJ1101+1053 11:01:55.74 +10:53:02.3 3.029 0 1 0 4 HS1157+3143 12:00:06.25 +31:26:30.8 2.989 2 5 SDSSJ1237+0126 12:37:48.99 +01:26:07.0 3.154 1 0 0 SDSSJ1253+6817 12:53:53.70 +68:17:14.4 3.481 5 7 SDSSJ1319+5202 13:19:14.19 +52:02:00.3 3.930 − − 0 1 SBS1602+576 16:03:55.93 +57:30:54.5 2.862 − − 0 1 HE2QSJ1630+0435 16:30:56.34 +04:35:59.4 3.788 − − 0 HS1700+6416 17:01:00.61 +64:12:09.0 2.751 1 −0 1 SDSSJ1711+6052 17:11:34.40 +60:52:40.5 3.834 − − 1 0 HE2QSJ2149 0859 21:49:27.77 08:59:03.6 3.259 −1 − 2 1 HE2QSJ2157−+2330 21:57:43.63 −+23:30:37.3 3.143 1 3 1 SDSSJ2346 0016 23:46:25.66 00:16:00.4 3.512 0 0 0 6 HE2347 43−42 23:50:34.21 −43:25:59.6 2.887 0 0 0 1 − − − Note. —Foreachtelescope/instrumentorcatalogweonlygivethenumberofdetectedorusedquasarsthatareincluded inourstatisticalanalysis,thereforeonlycountingquasarsforwhichwehaveusablecoverageintheHeiiLyαspectrum(e.g. positionnotcontaminated,notintheproximityzoneorbehindtheHeiiquasar,seeFigure5)andwhichcauseanestimated photoionizationrateattheHeiisightlineofΓHQeSIOI >0.5×10−15s−1 (see§2.6). a SDSSDR12,Alametal.(2015);Pˆarisetal.(2016) b Literature objects from Jakobsen et al. (2003); Steidel et al. (2003); Hennawi et al. (2006); Worseck & Wisotzki (2006); Worsecketal.(2007);Syphers&Shull(2013). However,attheselowredshiftswhereHeiireionization higher photoionization rate (ΓHeII = 12 10 15s 1 vs. QSO × − − is likely to be already completed, it is unclear whether ΓHeII =2.5 10 15s 1)theJakobsenetal.(2003)quasar these features are caused by the additional ionizing ra- seQeSmOs to be×the−dom−inant source of Heii-ionizing pho- diation of foreground quasars or arise due to density tons,butgiventheuncertainimpactofobscuration,life- fluctuations. Given the generally high Heii transmis- timeandquasarSEDeffectsontheresultingphotoioniza- sion at z < 2.7, a random association of a foreground tionratefromeitherquasar(§2.6), wecannotmakeany quasar with such a region is much more likely to occur judgment about their relative contribution. The second than at higher redshifts. Chance alignments of the two quasar could still have a significant impact and it might foreground quasars with the HS1700+6416 transmission be that the prominent Heii transmission spike is the re- spikes thus cannot be excluded. sult of the combined ionizing power of both quasars. In anycase,itdemonstratesthataone-to-oneassignmentof 3.2. Q0302 003 transmission spikes to foreground quasars is an oversim- − The sightline towards Q0302 003 presents the proto- plifiedapproachthatdoesnotcapturethefullcomplexity typical case for the Heii transv−erse proximity effect. of the Heii transverse proximity effect. At z <3.19, i.e. outside the large line-of-sight proxim- ityzoneofthebackgroundquasar( 80cMpc),theHeii 3.3. Foreground Quasars with the Highest Observed ≈ spectrumshowsalmostnotransmissiondowntoredshifts Photoionization Rates 2.9. The exception is a striking transmission feature at Our sample contains three new foreground quasars z =3.05(Heapetal.2000)withawidthof 450kms 1 ≈ − for which we infer a photoionization rate substantially (5.7cMpc) and a peak transmission of 80% 10. Jakobsen higher than the Jakobsen et al. (2003) quasar. The re- et al. (2003) found an r = 20.6mag foreground quasar spective regions in the Heii spectra are shown in Fig- withacorrespondingredshiftz =3.050 0.0036.5 away ± (cid:48) ure 6. Near the sightline toward SDSSJ1253+6817 we from the sightline and argued that this results from the have discovered a z = 3.20 quasar with an estimated Heii transverse proximity effect. ΓHeII =16.9 10 15s 1, 40%largerthanfortheJakob- Our survey uncovered a second quasar at a very sim- QSO × − − sen et al. (2003) quasar. Despite this, there is no trans- ilar redshift of z = 3.047, but further away from the mission spike observed even remotely comparable to the background sightline (8.5) and 0.7mag fainter than the (cid:48) one in the Q0302 003 sightline. The spectrum in this Jakobsen et al. (2003) quasar (Figure 6). Based on its − region is consistent with zero transmission (τ > 4). eff Despite the somewhat larger separation from the back- 10 Only for Q0302 003 we use a high-resolution COS G130M spectrum. Thespikew−ouldbemarginallyresolvedatG140Lreso- groundsightlineandtheslightlyhigherredshift,thisisa lutionandshowalowerpeaktransmissionbutwouldstillbeiden- verysurprisingresult. Asimilarsituationisobservedfor tifiedasanoutstandingspectralfeature. HE2QSJ2149 0859, where we find a foreground quasar − 10 T. M. Schmidt et al. WavelengthinA˚ 1100 1120 1140 1160 1180 Transmission 011...505 4137..20630 12.320 16.0337.06 04.970 zP=C2102.320.085980 0141200151in10s−−SO Q 0.0 Γ 2.65 2.70 2.75 2.80 2.85 2.90 1150 1200 1250 1300 mission 11..05 03.390 03.750 29.01 02.060 09.150 01.870 12.040 HEz2=Q3S.J30142023.3930 141015110s−− Trans 0.5 0 inSO Q 0.0 Γ 2.70 2.75 2.80 2.85 2.90 2.95 3.00 3.05 3.10 3.15 3.20 3.25 3.30 3.35 1150 1200 1250 1300 Transmission 0112....5050 11.170 02.960 219.26050.3 1253.09 zQ=033.02285 0141200151in10s−−SO Q 0.0 Γ 2.75 2.80 2.85 2.90 2.95 3.00 3.05 3.10 3.15 3.20 3.25 3.30 1120 1140 1160 1180 1200 n 1.5 SDzS=S2J.9058718 201s− Transmissio 01..50 02.91260.410 12.150 0141015in10−SO Q 0.0 Γ 2.70 2.75 2.80 2.85 2.90 2.95 1150 1200 1250 1300 Transmission 011...505 Lyβ101.15012.110 14.710 36.01 13.230 03.170 04.780 zH=S039.31510 0141200151in10s−−SO Q 0.0 Γ 2.65 2.70 2.75 2.80 2.85 2.90 2.95 3.00 3.05 3.10 3.15 3.20 3.25 3.30 3.35 3.40 1150 1200 1250 1300 1350 Transmission 011...505 Lyβ 03.48096.04 1212.13.000007.16102.100 41.030 HEz2=Q3S.J4040916 0141200151in10s−−SO Q 0.0 Γ 2.70 2.75 2.80 2.85 2.90 2.95 3.00 3.05 3.10 3.15 3.20 3.25 3.30 3.35 3.40 3.45 3.50 1120 1140 1160 1180 1200 1220 n 1.5 SDzS=S3J.0029724 101s− missio 1.0 02.780 14 1510− Trans 0.5 0 inSO Q 0.0 Γ 2.65 2.70 2.75 2.80 2.85 2.90 2.95 3.00 3.05 Redshift Figure 5. Galleryofthe22Heiisightlinesandtheforegroundquasarsample. TheHST/COSG130MspectrumofQ0302 003hasbeen binned to 0.15˚A per pixel for visualization purposes. Overplotted are positions of foreground quasars which are verticall−y displaced as indicatedb≈ytherightaxistovisualizetheirestimatedHeiiphotoionizationrateatthebackgroundsightline. Forforegroundquasarswith high impact, ΓHQeSIOI is also given as a label in units of 10−15s−1 (bold top number) and the separation from the background sightline in arcminutes(bottom). ForegroundquasarswithΓHQeSIOI <0.1×10−15s−1 areomitted. Colorsdenotequasarsdiscoveredbyuswithdifferent instruments(yellow: VLT/VIMOS;orange: CAHA/TWIN;blue: ESONTT/EFOSC2;purple: Keck/LRIS),quasarsfromSDSSandBOSS (gray), and quasars from previous dedicated surveys (black). A green strip on the bottom edge of the plots indicates the regions of the Heiispectraweuseinthestatisticalanalysis.