Environmental Effects on the UV Upturn in Local Clusters of Galaxies
Sadman Ali, Malcolm Bremer, Steven Phillipps, Roberto De Propris
MMNRAS , 1–8 (2019) Preprint 31 May 2019 Compiled using MNRAS L A TEX style file v3.0
Environmental Effects on the UV Upturn in Local Clustersof Galaxies
Sadman S. Ali, , (cid:63) Malcolm N. Bremer, Steven Phillipps and Roberto De Propris Subaru Telescope, NAOJ, 650 N. Aohoku Place, Hilo, HI 97620, USA H. H. Wills Physics Laboratory, University of Bristol, Tyndall Avenue, Bristol, BS8 1TL, United Kingdom FINCA, University of Turku, Vesilinnantie 5, Turku, 20014, Finland
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
We explore the dependence of UV upturn colours in early type cluster galaxies on theproperties of their parent clusters (such as velocity dispersion and X-ray luminosity)and on the positions and kinematics of galaxies within them. We use a sample of24 nearby clusters with highly complete spectroscopy and optical/infrared data toselect a suitable sample of red sequence galaxies, whose FUV and NUV magnitudeswe measure from archival GALEX data. Our results show that the UV upturn colourhas no dependence on cluster properties and has the same range in all clusters. Thereis also no dependence on the projected position within clusters or on line-of-sightvelocity. Therefore, our conclusion is that the UV upturn phenomenon is an intrinsicfeature of cluster early type galaxies, irrespective of their cluster environment.
Key words: galaxies: formation and evolution – stars: horizontal branch
The Ultraviolet Upturn or Excess is an unexpected rise influx in the spectral energy distributions of early-type galax-ies shortwards of 2500 ˚A. It appears to be a nearly ubiq-uitous property of spheroids and bulge-dominated galaxies(see, e.g., Yi 2008, 2010 for a recent review) although theirgenerally old, metal-rich and quiescent stellar populations(e.g., Thomas et al. 2005, 2010) should contain no sourcescapable of providing significant flux below the 4000 ˚A break.While many candidates have been proposed, the source pop-ulation is generally agreed to consist of hot horizontal branch(HB) stars (Greggio & Renzini 1990; Brown et al. 1998a) andit now appears most likely that this population is helium-rich and formed in situ at high redshift (Ali et al. 2018a,b,c).While such stars are now known to exist in globular clustersin our Galaxy (e.g., see Norris 2004; Piotto et al. 2005, 2007)and likely elsewhere (Kaviraj et al. 2007b; Mieske et al. 2008;Peacock et al. 2017), the origin of such high helium abun-dances is unclear . It has been suggested that the effect maydepend on the environment: stratification of helium in thecentres of clusters might create populations of galaxies withhigh helium abundances (Peng & Nagai 2009). On the other (cid:63) E-mail: [email protected] It must be noted that this hypothesis was originally presentedby Hartwick (1968) and Faulkner (1972), who also commented onthe ‘unpalatability’ of the proposed solution. hand, it is also possible that the helium enrichment dependson the details of early star formation, as in globular clusters.Differences in the originating mechanisms or timescales maybe reflected in different evolutionary histories for galaxies inclusters and the field. Atlee et al. (2009) finds a slow decreasein the strength of the upturn for a sample of bright field ellip-tical galaxies at < z < . compared to the nearly constantcolour for very bright ellipticals in the works of Brown et al.(1998b, 2000, 2003) and brightest cluster galaxies (Loubser& S´anchez-Bl´azquez 2011; Boissier et al. 2018). In our workon much larger samples of cluster early types with luminosi-ties down to L ∗ , we also see no evolution at z < . (Aliet al. 2018a,b,c) but then detect a rapid reddening in theUV colour at z = . ; this may be consistent with the lastdata point in Atlee et al. (2009), despite the large errorsand small number statistics. Le Cras et al. (2016) also findevidence for evolution in the UV upturn at z > . , albeitfrom stacked spectra and using spectroscopic indices sensi-tive to the HB population, for a sample of very luminousBOSS galaxies (as opposed to the UV photometry used byother studies).Ali et al. (2018a) compared the UV upturn colours forgalaxies in the Coma, Fornax and Perseus clusters, andfound no evidence that the UV upturn was affected bythe environment, a conclusion also reached by Smith et al.(2012b) when comparing galaxies in the Coma and Virgoclusters. Similarly, the UV upturn colour of brightest clus-ter galaxies in the samples of Loubser & S´anchez-Bl´azquez © a r X i v : . [ a s t r o - ph . GA ] M a y S. S. Ali et al.
Table 1.
Summary of ObservationsCluster RA (2000) δ (2000) z log L X σ Optical photometry GALEX imageshh:mm:ss dd:mm:ss 0.1–2.4 KeV km s − Abell 930 10 06 54.6 −
37 40 0.0549 35.78 907 SDSS, PS1 AIS 315Abell 954 10 13 44.8 −
06 31 0.0932 37.91 832 SDSS, PS1 MIS DR1Abell 957 10 13 40.3 −
54 52 0.0436 36.60 640 SDSS, PS1 MIS DR1, DR2Abell 1139 10 54 04.3 +01 29 56 0.0398 37.33 503 SDSS, PS1 MIS DR2Abell 1189 11 11 04.0 +01 07 42 0.0962 36.19 814 SDSS, PS1 MIS DR1Abell 1236 11 22 44.9 +00 27 44 0.102 36.42 589 SDSS, PS1 MIS DR1Abell 1238 11 22 58.0 +01 05 32 0.0733 36.49 586 SDSS, PS1 MIS WZN11Abell 1364 11 43 39.6 −
45 39 0.106 35.85 600 SDSS, PS1 GI5–GAMA12, MISGCSNAbell 1620 12 49 46.1 −
35 20 0.0821 34.48 1095 SDSS, PS1 MIS DR1Abell 1663 13 02 50.7 −
30 22 0.0843 37.00 884 SDSS, PS1 MIS DR1Abell 1692 13 12 16.0 −
55 55 0.0842 36.75 686 SDSS, PS1 GI4Abell 1750 13 30 49.9 −
52 22 0.0852 37.50 981 SDSS, PS1 MISGCSNAbell 2660 23 45 18.0 −
58 20 0.0525 35.70 845 PS1 AIS 279Abell 2734 00 11 20.7 −
51 18 0.0618 37.41 780 PS1 GI1Abell 2780 00 29 17.1 −
23 25 0.0988 ... 782 PS1 AIS 280Abell 3094 03 11 25.0 −
53 59 0.0677 36.76 774 PS1 AIS 405Abell 3880 22 27 52.4 −
34 12 0.0548 37.27 855 PS1 MIS2DFSGPAbell 4013 23 31 51.8 −
16 26 0.0500 ... 904 SuperCosmos AIS 394Abell 4053 23 54 46.7 −
40 18 0.0720 ... 994 PS1 AIS 280Abell S0003 00 03 09.5 −
53 18 0.0644 ... 833 PS1 AIS 280Abell S0084 00 09 24.0 −
31 28 0.108 37.41 807 PS1 MIS2DFSGPAbell S0166 01 34 23.4 −
35 39 0.0697 ... 511 SuperCosmos AIS 407Abell S1043 22 36 26.8 −
20 26 0.0340 ... 1345 PS1 MIS2DFSGP (2011) and Boissier et al. (2018) did not appear to dependstrongly on cluster properties. However, it would be interest-ing to extend this to several cluster environments and con-sider the eventual dependence on position within the clusterand on kinematics. For example, if residual star formationcontributes to the UV flux, as argued by Rich et al. (2005);Yi et al. (2005); Salim & Rich (2010) then one would expectto observe a dependence on cluster properties (e.g., X-rayluminosity if ram stripping is important) and/or on posi-tion within the cluster or kinematics (as a proxy for orbitsthat avoid or pass through the cluster core for instance).Although the effect may be stochastic, it should emerge atsome level in the ensemble of several clusters studied here.Furthermore, the Helium sedimentation model of Peng& Nagai (2009) predicted that the strength of the UV up-turn should be stronger in a) larger mass clusters; b) clusterswith cooling flows and; c) in dynamically relaxed clusters.The model also predicts that UV upturn galaxies should bemore prevalent in cluster cores. However, these key environ-mental dependencies of the upturn strength were proven tonot hold true by the studies of Donahue et al. (2010) andLoubser & S´anchez-Bl´azquez (2011), who found no correla-tion between the
FUV − NUV colour (a measure of the upturnstrength) in low redshift cluster galaxies and the aforemen-tioned parameters (see also Yi et al. 2011).In this paper we exploit a highly complete sample ofgalaxies in 24 nearby clusters to derive the dependence ofthe UV upturn color on cluster properties and on the localcluster environment. We describe the sample and photome-try in the next section and present the results in section 3.Discussion of our findings and conclusions are shown in sec-tion 4. We assume the conventional cosmological parametersfrom the latest Planck datasets.
The sample we studied here consists of galaxies in 24 clus-ters from the sample of De Propris (2017). These clustersspan a wide range of properties in terms of velocity disper-sion, X-ray luminosity and Bautz-Morgan type and thereforeallow us to consider how the UV upturn colour is affectedby cluster properties over a large range in environmentaldensity, hot gas content and dynamical indicators (e.g., re-laxed vs. merging clusters). In particular, we can explore theregime between massive systems, such as Coma or Perseus,and poor clusters and rich groups (similar, in some respectto Virgo or Fornax).Each cluster was observed with the CTIO 4m telescopein the K s band, to produce a deep (300s exposure) image ofthe entire cluster across its Abell radius.The K s luminosity isfound, empirically, to be an excellent proxy for stellar mass(Gavazzi et al. 1996; Bell & de Jong 2001; Kettlety et al.2018) and this therefore provides a stellar mass selected sam-ple of galaxies in nearby clusters (mean redshift of 0.075).Galaxies in these clusters down to at least the level of themeasured K ∗ + were identified as spectroscopic members(with typical completeness of about 80% even in the faintestluminosity bin considered) from existing spectroscopic data.This also provides information on the kinematics of thesegalaxies within each cluster.For all galaxies we derived optical colours ( g − r or B J − R F ) from data in the PanStarrs1 survey (Chambers et al.2016; Flewelling et al. 2016; Magnier et al. 2016) or (forobjects below –30 ◦ ) in the SuperCosmos survey (Hamblyet al. 2001a,b).We then matched all confirmed spectroscopic membersto FUV and NUV data in the GALEX database (Morrisseyet al. 2007). We used a . (cid:48)(cid:48) matching radius and only se- MNRAS000
The sample we studied here consists of galaxies in 24 clus-ters from the sample of De Propris (2017). These clustersspan a wide range of properties in terms of velocity disper-sion, X-ray luminosity and Bautz-Morgan type and thereforeallow us to consider how the UV upturn colour is affectedby cluster properties over a large range in environmentaldensity, hot gas content and dynamical indicators (e.g., re-laxed vs. merging clusters). In particular, we can explore theregime between massive systems, such as Coma or Perseus,and poor clusters and rich groups (similar, in some respectto Virgo or Fornax).Each cluster was observed with the CTIO 4m telescopein the K s band, to produce a deep (300s exposure) image ofthe entire cluster across its Abell radius.The K s luminosity isfound, empirically, to be an excellent proxy for stellar mass(Gavazzi et al. 1996; Bell & de Jong 2001; Kettlety et al.2018) and this therefore provides a stellar mass selected sam-ple of galaxies in nearby clusters (mean redshift of 0.075).Galaxies in these clusters down to at least the level of themeasured K ∗ + were identified as spectroscopic members(with typical completeness of about 80% even in the faintestluminosity bin considered) from existing spectroscopic data.This also provides information on the kinematics of thesegalaxies within each cluster.For all galaxies we derived optical colours ( g − r or B J − R F ) from data in the PanStarrs1 survey (Chambers et al.2016; Flewelling et al. 2016; Magnier et al. 2016) or (forobjects below –30 ◦ ) in the SuperCosmos survey (Hamblyet al. 2001a,b).We then matched all confirmed spectroscopic membersto FUV and NUV data in the GALEX database (Morrisseyet al. 2007). We used a . (cid:48)(cid:48) matching radius and only se- MNRAS000 , 1–8 (2019) nvironment and the UV Upturn lected objects with S/N of at least 5 in the GALEX NUVphotometry. These come from a combination of AIS and MISimaging. Exposure times range from a few ks in MIS data toa few hundred seconds in AIS. For blue cloud galaxies in ourclusters, 93% have a NUV detection and 76% have a FUVdetection, whereas 72% of red sequence galaxies have a NUVdetection and 40% have a FUV detection. We can assumethat objects with no NUV detection have red colours, giventhe high detection fraction for blue cloud galaxies.In Table 1 we show the main properties of each clusterused and the relevant information on the sources of photom-etry (see also De Propris et al. 2018). All data were correctedfor Galactic extinction using the latest values from Schlafly& Finkbeiner (2011) and extrapolating to the FUV andNUV bands with a Milky Way extinction law (e.g., Calzettiet al. 1994). Optical and infrared data were corrected forextinction with the same procedures. Colours were also k -corrected using the derived UV spectral energy distribution(SED) from Ali et al. (2018a). This uses a standard modelfrom Conroy et al. (2009) for the optical and a 16000K black-body for the contribution due to the vacuum UV light, acombination that fits the observed spectral energy distribu-tions of Coma cluster galaxies across the whole range from1000 to 10000 ˚A. In our previous papers (De Propris 2017; De Propris et al.2018) we have identified galaxies on the red sequence andblue cloud by fitting a straight line to the colour-magnituderelation (in each cluster, separately) using a minimum ab-solute deviation method that discriminates against outliers(see De Propris 2017; De Propris et al. 2018 for details). The1 σ scatter in g − r for red sequence galaxies was measured tobe 0.05 mag. and we therefore selected galaxies as belongingto the red sequence if they lie within ± . mag. of the best-fitting straight line to the colour-magnitude relation. Bluercluster members are assigned to the blue cloud.Red sequence galaxies (selected in the optical) consistof truly quiescent objects (whose UV light is produced bythe UV upturn) and galaxies with residual star formation(sometimes called the ‘green valley’ – e.g., Rich et al. 2005;Salim & Rich 2010). Several studies have adopted cuts in NUV − optical colours to separate quiescent galaxies fromthose with residual star formation. Kaviraj et al. (2007a)place their selection at NUV − r > . , while Crossett et al.(2014) adopt a somewhat more stringent limit NUV − r > . to their sample of z < . clusters. Here we choose this latterdefinition.Below, we show how this selection is justified from ourdata. We plot NUV − r vs. u − g for all galaxies (this is onlypossible for those with SDSS data, as indicated in Table 1)in Fig. 1 (top panel). Here red dots are red sequence galax-ies and blue dots are galaxies in the blue cloud (as definedabove). These latter objects follow an approximate straightline in this colour-colour plane (i.e. the so-called star-formingmain sequence – e.g., Speagle et al. 2014). We fit this witha straight line and remove this trend from all objects, re-sulting in the middle panel of Fig. 1. By choosing a limit Figure 1.
Top: The colour-colour plot (
NUV − u vs. u − g forgalaxies in 11 of our clusters where SDSS u data are available.Middle: We define a vector y = NUV − r − . ( u − g ) to remove thelinear trend observed in the top panel between NUV − u and u − r and plot this vector vs. u − g to show that star-forming galaxiesgenerally have y < − . . Bottom: We plot the numbers of galaxiesin the initial sample and objects with y < − . (open histograms)and y > − . (filled histograms) as a function of NUV − r colour.Red and orange histograms are for red sequence galaxies (afterand before selection using the y vector) and blue histograms arefor blue cloud galaxies. We also show a line at NUV − r = . .This excludes nearly all star-forming galaxies from the sample. y = ( NUV − u ) − . ( u − g ) > − . we can exclude the vastmajority of objects with star formation from the sample.However, note that we cannot apply this selection to all ourclusters, as those in the South have no u data (this is not pro-vided by PanStarrs1). In the bottom panel of this figure weplot the number counts of galaxies as a function of NUV − r color before and after the above selection. Almost all star-forming galaxies have NUV − r < . and our NUV − r > . criterion appears to isolate a nearly pure sample of quiescentsystems (see also Phillipps et al. 2019, in preparation for adetailed discussion).Arnouts et al. (2013) estimate the star formation ratein galaxies by fitting the full spectral energy distributionsand then derive a vector in the NUV − r vs. r − K plane that MNRAS , 1–8 (2019)
S. S. Ali et al.
Figure 2.
The colour-colour plot (
NUV − r vs. r − K for galaxiesin our clusters (red dots for red sequence galaxies and blue dotsfor blue cloud ones). We plot (thick black line) the NRK vector( NUV − r = . and NUV − r = . ( r − K ) + . defined by Arnoutset al. (2013) to separate star-forming and quiescent galaxies. Thedashed line shows a colour cut NUV − r > . : this always liesabove the NRK vector and therefore selects quiescent galaxieseffectively. separates star-forming and quiescent galaxies (cf. Phillippset al. 2019, where we use MAGPHYS derived star forma-tion rates to the same purpose). They confirm this approachfrom a morphologically selected sample in Moutard et al.(2016a,b). We plot this colour-colour plot in Fig. 2 with thesame colour scheme as in Fig. 1. As we see a colour cut at NUV − r > . lies well above their NRK vector. We thereforeconclude that selecting galaxies with NUV − r > . producesa sample consisting largely of passive galaxies whose UVlight is dominated by upturn sources, whereas the original NUV − r > . cut by Kaviraj et al. (2007a) may still includea small fraction of objects with residual star formation. Fig. 3 shows the derived
FUV − r and NUV − r colours for clus-ter red sequence galaxies plotted against M r . The FUV − r colours mostly range between 5.5 and 7.5 mag. This ob-served ∼ magnitude range in colour is similar to that ofComa (Smith et al. 2012b) and Perseus (Ali et al. 2018a), aswell as that observed in Virgo red sequence galaxies (Boselliet al. 2005). The NUV − r colour varies between 5 and 6.5mag., and once again, the ∼ . magnitude range observed istypical of the aforementioned low redshift cluster galaxies.This can be compared with the well-known small scatter ofoptical-infrared colour-magnitude relations for red sequencesin clusters (e.g., Valentinuzzi et al. 2011). In our sample,the intrinsic scatter is g − r for red sequence galaxies is 0.05mag. (De Propris 2017), i.e., nearly two orders of magni-tudes smaller than in FUV − r or NUV − r . This consistentrange in FUV − r and NUV − r observed between all low red-shift cluster galaxies indicates that the upturn is a universalfeature among all such old, passively evolving systems andalso that the environment must not have any significant in- fluence on this phenomenon, since the clusters studied herehave a wide variety of properties.We further explore this by plotting FUV − r and NUV − r colours for galaxies in clusters vs. the cluster velocity disper-sion (a broad proxy of the cluster mass) and vs. the X-rayluminosity in the ROSAT soft band (Truemper 1993; takenfrom the BAX catalogue - Sadat et al. 2004), which is a mea-sure of the gas density in each cluster, in Fig. 4. We also plotdata for Coma and Perseus for comparison. It is clear thatthere is no obvious trend in these colours with cluster prop-erties, across a wide range of cluster masses. As we expectthat velocity dispersion and X-ray luminosity would affectany residual star formation, the lack of correlation with theseparameters further argues that the UV upturn propertieswere established at early times, prior to the epochs at whichgalaxies first felt the effects of the cluster environment.We next consider whether the UV upturn colour is af-fected by position within the cluster or by kinematics. InFig. 5 we plot FUV − r and NUV − r vs R / R , where R is calculated from Carlberg et al. (1997) using data fromDe Propris (2017). Objects projected closer to the centreare likely to have been in the cluster core longer (Smithet al. 2012a): these are mainly classical ellipticals in theinner 0.3 R / R . The colours appear identical irrespectiveof projected position. Neither do we see any dependence ofthese colours on ∆ V / σ (Fig. 5), where ∆ V is the differencebetween the velocity of each galaxy and the mean velocityof its parent cluster and σ is the cluster velocity dispersion(thus objects with higher ∆ V / σ are likely to be less viri-alised and to move on more radial orbits, and therefore tobe relative newcomers to the cluster environment and to bemore affected by tidal and ram stripping processes). There-fore, these provide further evidence against the existence ofstrong environmental effects, especially those that would af-fect continuing star formation.Finally we plot ∆ V / σ vs. R / R for galaxies colour-coded according to their FUV − r and NUV − r colours, inFig. 6. These caustic plots are related to the orbits of galaxieswithin clusters and we observe no significant dependence onUV upturn colour. This again suggests that the UV upturnis unrelated to the cluster environment. As with Coma and Perseus in our previous paper (Ali et al.2018a), the results obtained here can be best interpretedwith the presence of a He-enhanced sub-population of hotHB stars giving rise to the upturn, superimposed on topof the majority ‘red and dead’ population that makes upETGs. The range in colours can then be explained by vary-ing amounts of He-enhancement. Using the YEPS StellarPopulation Synthesis models of Chung et al. (2017) that in-corporate He-enhancement, Y ≥ . is predicted (assumingsolar metallicity and z f = ) to account for the full rangein FUV − r and NUV − r colours seen in all of these clustersbelow z = . (note that metallicity, unless highly subsolar,which is not the case for these galaxies, does not stronglyaffect the HB morphology and therefore the UV colours forpassive galaxies, so the observed range in colours cannot beaccounted for my metal abundance). Consistent with the ob-servations of Ali et al. (2018c), these models predict that the MNRAS000
FUV − r and NUV − r colours for clus-ter red sequence galaxies plotted against M r . The FUV − r colours mostly range between 5.5 and 7.5 mag. This ob-served ∼ magnitude range in colour is similar to that ofComa (Smith et al. 2012b) and Perseus (Ali et al. 2018a), aswell as that observed in Virgo red sequence galaxies (Boselliet al. 2005). The NUV − r colour varies between 5 and 6.5mag., and once again, the ∼ . magnitude range observed istypical of the aforementioned low redshift cluster galaxies.This can be compared with the well-known small scatter ofoptical-infrared colour-magnitude relations for red sequencesin clusters (e.g., Valentinuzzi et al. 2011). In our sample,the intrinsic scatter is g − r for red sequence galaxies is 0.05mag. (De Propris 2017), i.e., nearly two orders of magni-tudes smaller than in FUV − r or NUV − r . This consistentrange in FUV − r and NUV − r observed between all low red-shift cluster galaxies indicates that the upturn is a universalfeature among all such old, passively evolving systems andalso that the environment must not have any significant in- fluence on this phenomenon, since the clusters studied herehave a wide variety of properties.We further explore this by plotting FUV − r and NUV − r colours for galaxies in clusters vs. the cluster velocity disper-sion (a broad proxy of the cluster mass) and vs. the X-rayluminosity in the ROSAT soft band (Truemper 1993; takenfrom the BAX catalogue - Sadat et al. 2004), which is a mea-sure of the gas density in each cluster, in Fig. 4. We also plotdata for Coma and Perseus for comparison. It is clear thatthere is no obvious trend in these colours with cluster prop-erties, across a wide range of cluster masses. As we expectthat velocity dispersion and X-ray luminosity would affectany residual star formation, the lack of correlation with theseparameters further argues that the UV upturn propertieswere established at early times, prior to the epochs at whichgalaxies first felt the effects of the cluster environment.We next consider whether the UV upturn colour is af-fected by position within the cluster or by kinematics. InFig. 5 we plot FUV − r and NUV − r vs R / R , where R is calculated from Carlberg et al. (1997) using data fromDe Propris (2017). Objects projected closer to the centreare likely to have been in the cluster core longer (Smithet al. 2012a): these are mainly classical ellipticals in theinner 0.3 R / R . The colours appear identical irrespectiveof projected position. Neither do we see any dependence ofthese colours on ∆ V / σ (Fig. 5), where ∆ V is the differencebetween the velocity of each galaxy and the mean velocityof its parent cluster and σ is the cluster velocity dispersion(thus objects with higher ∆ V / σ are likely to be less viri-alised and to move on more radial orbits, and therefore tobe relative newcomers to the cluster environment and to bemore affected by tidal and ram stripping processes). There-fore, these provide further evidence against the existence ofstrong environmental effects, especially those that would af-fect continuing star formation.Finally we plot ∆ V / σ vs. R / R for galaxies colour-coded according to their FUV − r and NUV − r colours, inFig. 6. These caustic plots are related to the orbits of galaxieswithin clusters and we observe no significant dependence onUV upturn colour. This again suggests that the UV upturnis unrelated to the cluster environment. As with Coma and Perseus in our previous paper (Ali et al.2018a), the results obtained here can be best interpretedwith the presence of a He-enhanced sub-population of hotHB stars giving rise to the upturn, superimposed on topof the majority ‘red and dead’ population that makes upETGs. The range in colours can then be explained by vary-ing amounts of He-enhancement. Using the YEPS StellarPopulation Synthesis models of Chung et al. (2017) that in-corporate He-enhancement, Y ≥ . is predicted (assumingsolar metallicity and z f = ) to account for the full rangein FUV − r and NUV − r colours seen in all of these clustersbelow z = . (note that metallicity, unless highly subsolar,which is not the case for these galaxies, does not stronglyaffect the HB morphology and therefore the UV colours forpassive galaxies, so the observed range in colours cannot beaccounted for my metal abundance). Consistent with the ob-servations of Ali et al. (2018c), these models predict that the MNRAS000 , 1–8 (2019) nvironment and the UV Upturn Figure 3.
UV colours (FUV–r on the left and NUV–r on the right) plotted vs M r for all galaxies in the 2dF cluster sample of De Propris(2017). Galaxies with NUV − r > . are convincingly passive systems, while those with . < NUV − r < . may still contain someresidual star formation. Opposite to the ∼ . – mag. scatter in these colours, the scatter in g − r is about 0.05 mag. UV upturn colour is nearly constant to z ≈ . (dependingon Y and the epoch at which the stars were formed) andthen evolves rapidly to the red. This is indeed observed ina z = . cluster (Ali et al. 2018c). It is also interesting thatthe bluest UV upturn colours tend to occur among the mostmassive galaxies, which also tend to be older and more metalrich (Smith et al. 2012b). These galaxies also have resided inthe cluster core for longer times (Smith et al. 2012a). Simi-larly, Ali et al. (2018a) find that the more massive galaxieshave a hotter and more populated HB as well. This maybe due to a greater degree of helium enrichment but alsoto the effect of larger ages in thinning the remaining stellaratmosphere.Under dynamical equilibrium the velocity dispersionand X-ray luminosity of a cluster directly correlate with itsmass. Fig. 4 demonstrates that the upturn strength has nocorrelation with either velocity dispersion or X-ray luminos-ity, and by proxy the mass of the cluster. As such, clusters ofall sizes can have a component of upturn in their early-typepopulation.Within galaxy clusters, star-formation is quenched verystrongly in the centre due to processes such as ram-pressurestripping, harassment and strangulation. The rate of star-formation in cluster galaxies thus tends to increase with in-creasing cluster-centric distance (e.g., Dressler et al. 1997).If star-formation was the sole driving mechanism behind theUV emission in the galaxies in this sample, as argued by Yiet al. (2005), one would expect to see the FUV − r and NUV − r colours become bluer with increasing radius, at least in anensemble. However as seen in Fig. 5 the UV-optical coloursshow no correlation with cluster-centric distance. This re-inforces the idea that the UV emission in these galaxies isindeed from an old, hot HB subpopulation and not fromany residual star-formation. Similarly, there appears to beno correlation between the line-of-sight velocity of galaxiesand the upturn strength, in Fig. 5. Therefore, the influenceof the cluster’s gravitational potential - the main drivingforce behind the velocities - does not affect the strength ofthe upturn. We can see from Fig. 6 that the majority of cluster members are centred around the core of the cluster( R / R < , ∆ V / σ < ), yet these central galaxies showthe full range of FUV − r and NUV − r colours as seen fromthe entire population of galaxies. This suggests that there isno gradient in UV upturn colour with either cluster-centricdistance or line-of-sight velocity.These results are more in line with a He-enhanced HBorigin for the UV upturn, in which case the cluster environ-ment should have little effect on the emergence and preva-lence of the upturn. The observed lack of environmental de-pendence with position within the cluster implies that theextra helium cannot come from stratification as in the modelof Peng & Nagai (2009). In a hierarchical model of structureformation, galaxies form first at < z < and the star for-mation in ETGs rapidly comes to an end by z ∼ (Jørgensenet al. 2017) . Galaxies then accrete in highly over-dense re-gions of the universe to form clusters at z ∼ . (e.g. Wen &Han 2011). Studies have shown that cluster red sequencesare already established between < z < (Newman et al.2014). This indicates that the majority of star-formation inETGs was completed and a mostly passively evolving stel-lar population, that is observed at present, was already es-tablished before these galaxies became part of clusters, orshortly thereafter. While some field ETGs show evidence ofresidual star formation where at least part of their stellarpopulations is formed more recently (e.g., Jeong et al. 2007;Davis et al. 2013), this is unlikely to be the case for clusterETGs: our UV SEDs for ETGs in Coma (Ali et al. 2018a)and Abell 1689 (Ali et al. 2018b), for instance, are incon-sistent with a contribution from star formation to the UVlight.This is particularly important as the oldest stars inETGs, ones that formed at z ≥ , before the galaxies be-came part of clusters (Guarnieri et al. 2018) are the ones thatwould have the necessary time required to evolve from themain sequence on to the red giant branch, and then eventu-ally to the horizontal branch, where they become UV-brightgiven sufficient He-enhancement. Since the cluster environ-ment particularly works to quench the star-formation within MNRAS , 1–8 (2019)
S. S. Ali et al.
Figure 4.
Top: UV colours (FUV–V on the left and NUV–V on the right) plotted vs cluster velocity dispersion for all galaxies in the2dF cluster sample of De Propris (2017). Bottom: UV colours (FUV–r on the left and NUV–r on the right) plotted vs cluster X-rayluminosity in the ROSAT soft band (0.5-2 KeV) for all galaxies in the 2dF cluster sample of De Propris (2017). We also show data fromthe literature (Ali et al. 2018a) for Coma and Perseus. galaxies, the main sequence population in ETGs, which isalready red and passively evolving, is largely unaffected.Hence, the upturn develops intrinsically within these galax-ies irrespective of the cluster environment. This also indi-cates that the large He-enhancement that leads to the even-tual UV upturn in a sub-population of the main sequence inETGs must also occur intrinsically within the galaxies andat very early times.
ACKNOWLEDGEMENTS
The Pan-STARRS1 Surveys (PS1) and the PS1 public sci-ence archive have been made possible through contributionsby the Institute for Astronomy, the University of Hawaii,the Pan-STARRS Project Office, the Max-Planck Societyand its participating institutes, the Max Planck Institutefor Astronomy, Heidelberg and the Max Planck Institute forExtraterrestrial Physics, Garching, The Johns Hopkins Uni-versity, Durham University, the University of Edinburgh, the Queen’s University Belfast, the Harvard-Smithsonian Cen-ter for Astrophysics, the Las Cumbres Observatory GlobalTelescope Network Incorporated, the National Central Uni-versity of Taiwan, the Space Telescope Science Institute,the National Aeronautics and Space Administration underGrant No. NNX08AR22G issued through the Planetary Sci-ence Division of the NASA Science Mission Directorate, theNational Science Foundation Grant No. AST-1238877, theUniversity of Maryland, Eotvos Lorand University (ELTE),the Los Alamos National Laboratory, and the Gordon andBetty Moore Foundation.This work is based in part on observations made withthe Galaxy Evolution Explorer (GALEX). GALEX is aNASA Small Explorer, whose mission was developed incooperation with the Centre National d’Etudes Spatiales(CNES) of France and the Korean Ministry of Science andTechnology. GALEX is operated for NASA by the CaliforniaInstitute of Technology under NASA contract NAS5-98034.This research has made use of the NASA/IPAC Ex-tragalactic Database (NED) which is operated by the Jet
MNRAS000
MNRAS000 , 1–8 (2019) nvironment and the UV Upturn Figure 5.
Top: UV colours (FUV–r on the left and NUV–r on the right) plotted vs R / R (projected) in the 2dF cluster sample of DePropris (2017). Bottom: UV colours (FUV–r on the left and NUV–r on the right) plotted vs ∆ V / σ . We also show data for Coma andPerseus as in Fig. 2. Figure 6.
Caustic plots for all galaxies colour coded according to their FUV–r or NUV–r colour.MNRAS , 1–8 (2019)
S. S. Ali et al.
Propulsion Laboratory, California Institute of Technology,under contract with the National Aeronautics and SpaceAdministration.
REFERENCES
Ali S. S., Bremer M. N., Phillipps S., De Propris R., 2018a, MN-RAS, 476, 1010Ali S. S., Bremer M. N., Phillipps S., De Propris R., 2018b, MN-RAS, 478, 541Ali S. S., Bremer M. N., Phillipps S., De Propris R., 2018c, MN-RAS, 480, 2236Arnouts S., et al., 2013, A&A, 558, A67Atlee D. W., Assef R. J., Kochanek C. S., 2009, ApJ, 694, 1539Bell E. F., de Jong R. S., 2001, ApJ, 550, 212Boissier S., Cucciati O., Boselli A., Mei S., Ferrarese L., 2018,A&A, 611, A42Boselli A., et al., 2005, ApJ, 629, L29Brown T. M., Ferguson H. C., Stanford S. A., Deharveng J.-M.,1998a, ApJ, 504, 113Brown T. M., Ferguson H. C., Deharveng J.-M., JedrzejewskiR. I., 1998b, ApJ, 508, L139Brown T. M., Bowers C. W., Kimble R. A., Ferguson H. C., 2000,ApJ, 529, L89Brown T. M., Ferguson H. C., Smith E., Bowers C. W., KimbleR. A., Renzini A., Rich R. M., 2003, ApJ, 584, L69Calzetti D., Kinney A. L., Storchi-Bergmann T., 1994, ApJ, 429,582Carlberg R. G., et al., 1997, ApJ, 485, L13Chambers K. C., et al., 2016, arXiv e-prints, 1612.05560Chung C., Yoon S.-J., Lee Y.-W., 2017, ApJ, 842, 91Conroy C., Gunn J. E., White M., 2009, ApJ, 699, 486Crossett J. P., Pimbblet K. A., Stott J. P., Jones D. H., 2014,MNRAS, 437, 2521Davis T. A., et al., 2013, MNRAS, 429, 534De Propris R., 2017, MNRAS, 465, 4035De Propris R., Bremer M. N., Phillipps S., 2018, A&A, 618, A180Donahue M., et al., 2010, ApJ, 715, 881Dressler A., et al., 1997, ApJ, 490, 577Faulkner J., 1972, Nature Physical Science, 235, 27Flewelling H. A., et al., 2016, arXiv e-prints, 1612.05243Gavazzi G., Pierini D., Boselli A., 1996, A&A, 312, 397Greggio L., Renzini A., 1990, ApJ, 364, 35Guarnieri P., et al., 2018, arXiv e-prints, 1811.12422Hambly N. C., et al., 2001a, MNRAS, 326, 1279Hambly N. C., Irwin M. J., MacGillivray H. T., 2001b, MNRAS,326, 1295Hartwick F. D. A., 1968, ApJ, 154, 475Jeong H., Bureau M., Yi S. K., Krajnovi´c D., Davies R. L., 2007,MNRAS, 376, 1021Jørgensen I., Chiboucas K., Berkson E., Smith O., Takamiya M.,Villaume A., 2017, AJ, 154, 251Kaviraj S., et al., 2007a, ApJS, 173, 619Kaviraj S., Sohn S. T., O’Connell R. W., Yoon S.-J., Lee Y. W.,Yi S. K., 2007b, MNRAS, 377, 987Kettlety T., et al., 2018, MNRAS, 473, 776Le Cras C., Maraston C., Thomas D., York D. G., 2016, MNRAS,461, 766Loubser S. I., S´anchez-Bl´azquez P., 2011, MNRAS, 410, 2679Magnier E. A., et al., 2016, arXiv e-prints, 1612.05242Mieske S., Hilker M., Bomans D. J., Rey S.-C., Kim S., YoonS.-J., Chung C., 2008, A&A, 489, 1023Morrissey P., et al., 2007, ApJS, 173, 682Moutard T., et al., 2016a, A&A, 590, A102Moutard T., et al., 2016b, A&A, 590, A103Newman A. B., Ellis R. S., Andreon S., Treu T., Raichoor A.,Trinchieri G., 2014, ApJ, 788, 51 Norris J. E., 2004, ApJ, 612, L25Peacock M. B., Zepf S. E., Kundu A., Chael J., 2017, MNRAS,464, 713Peng F., Nagai D., 2009, ApJ, 705, L58Piotto G., et al., 2005, ApJ, 621, 777Piotto G., et al., 2007, ApJ, 661, L53Rich R. M., et al., 2005, ApJ, 619, L107Sadat R., Blanchard A., Kneib J.-P., Mathez G., Madore B., Maz-zarella J. M., 2004, A&A, 424, 1097Salim S., Rich R. M., 2010, ApJ, 714, L290Schlafly E. F., Finkbeiner D. P., 2011, ApJ, 737, 103Smith R. J., Lucey J. R., Price J., Hudson M. J., Phillipps S.,2012a, MNRAS, 419, 3167Smith R. J., Lucey J. R., Carter D., 2012b, MNRAS, 421, 2982Speagle J. S., Steinhardt C. L., Capak P. L., Silverman J. D.,2014, ApJS, 214, 15Thomas D., Maraston C., Bender R., Mendes de Oliveira C., 2005,ApJ, 621, 673Thomas D., Maraston C., Schawinski K., Sarzi M., Silk J., 2010,MNRAS, 404, 1775Truemper J., 1993, Science, 260, 1769Valentinuzzi T., et al., 2011, A&A, 536, A34Wen Z. L., Han J. L., 2011, ApJ, 734, 68Yi S. K., 2008, in Heber U., Jeffery C. S., Napiwotzki R.,eds, Astronomical Society of the Pacific Conference SeriesVol. 392, Hot Subdwarf Stars and Related Objects. p. 3( arXiv:0808.0254 )Yi S. K., 2010, in Bruzual G. R., Charlot S., eds, IAU SymposiumVol. 262, Stellar Populations - Planning for the Next Decade.pp 147–149, doi:10.1017/S1743921310002693Yi S. K., et al., 2005, ApJ, 619, L111Yi S. K., Lee J., Sheen Y.-K., Jeong H., Suh H., Oh K., 2011,ApJS, 195, 22This paper has been typeset from a TEX/L A TEX file prepared bythe author. MNRAS000