Evolution of the Blue Luminosity-to-Baryon Mass Ratio of Clusters of Galaxies Kazuhiro Shimasaku Department of Astronomy, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113 Research Center for the Early Universe, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113 E-mail: [email protected] 0 (Received 1999 September1; accepted 1999 September1) 0 Abstract 0 2 Wederivetheratiooftotalblueluminositytototalbaryonmass,LB/Mb,formassive(Mgas attheAbell n radiusis≥1×1013h−2.5M⊙•)clustersofgalaxiesuptoz ≃1fromtheliterature. Twenty-twoclustersinour a sample are at z >0.1. Assuming that the relative mix of hot gas and galaxies in clusters does not change J during cluster evolution, we use LB/Mb to probe the star formation history of the galaxy population as a 1 3 whole in clusters. We find that LB/Mb of clusters increases with redshift from LB/Mb = 0.024(LB/M)⊙ at z =0 to ≃0.06(LB/M)⊙ at z =1, indicating a factor of 2−3 brightening (we assume H0 =70 km s−1 1 Mpc−1). Thisamountofbrighteningis almostidenticalto the brighteningoftheM/LB ratioofearly-type v galaxies in clusters at 0.02 ≤ z ≤ 0.83 reported by van Dokkum et al. (1998). We compare the observed 1 brightening of LB/Mb with luminosity evolution models for the galaxy population as a whole, changing 2 the e-folding time of star formation τ by 0.1 ≤ τ ≤ 5 Gyr and the formation redshift z by 2 ≤ z < ∞. 5 F F We find that τ = 0.1 Gyr ‘single burst’ models with z ≥ 3 and τ = 5 Gyr ‘disk’ models with arbitrary 1 F 0 zF are consistent with the observed brightening, while models with τ = 1−2 Gyr tend to predict too 0 steep brightening. We also derive the ratio of blue luminosity density to baryon density for field galaxies, 0 adopting Ωbh2 =0.02, and find that blue luminosity per unit baryon is similar in clusters and in fields up / h to z ≃1 within the observational uncertainties. p Key words: Galaxies: clusters of — Galaxies: photometry — Galaxies: X-rays - o r t s 1. Introduction In this paper, we study the global (or average) star a formationhistory of the galaxypopulation as a whole in : v clusters of galaxies. To do so, we derive the ratio of to- Clusters of galaxies are suitable objects for study- Xi ing the evolution of galaxies in the dense environments. tal blue luminosity (LB) to total baryon mass (Mb) for r Recent observations based on large telescopes including clusters of galaxies up to z ≃ 1. The quantity LB/Mb, a including its evolution, should reflect when (and what Hubble Space Telescope (HST) have been revealing the fraction of) baryons (= primordial gas) in clusters are morphology-dependent evolution of galaxies in clusters converted into stars. Similar studies have been done for up to z ∼ 1. The evolution of elliptical and S0 galaxies has been found to be reproduced well by the so-called fieldgalaxies. Thegloballuminositydensityl[L⊙•Mpc−3] invariouswavelengthshasbeenmeasuredbymanywork- single burst model (e.g., Schade, Barrientos,Lo´pez-Cruz ersonthebasisofobservationsoffieldgalaxies(e.g.,Lilly 1997;Ellisetal. 1997;Kodamaetal. 1998;vanDokkum et al. 1996; Madau, Pozzetti, Dickinson 1998). If the et al. 1998). Schade et al. (1996) found that spiral galaxiesinclustersbrightenby∼1magwithredshiftup densityparameterofbaryons(Ωb)isgiven,onecancom- putefroml themeanluminosityperunitbaryonmassin to z ≃ 0.5, and that this brightening is similar to that fields (l/ρ ). We will derive B-band l/ρ in fields up to of field spiral galaxies in the same redshift range (see, b b however, Vogt et al. 1997 and Lilly et al. 1998 for the z ≃1 and compare it with LB/Mb of clusters. evolution of field spiral galaxies). Morphological studies The structure of this paper is as follows. In section 2, based on HST imaging suggest that a transition from we present the data of nearby and distant clusters used spiral galaxies to S0 galaxies may have occurred in clus- to derive LB/Mb. We compare LB/Mb with predictions ters since z ∼ 0.5 (Dressler et al. 1997). An attempt of of simple luminosity evolution models in section 3. A measuring the star formation rate of individual galaxies comparison with l/ρb in fields is also given in section 3. hasalsostarted(e.g.,Baloghetal. 1998;Poggiantietal. We summarize our conclusions in section 4. 1999). We adopt h = 0.7, Ω = 0.2, and λ = 0 throughout 0 0 No. 0] Luminosity Evolution of Galaxies in Clusters 1 thispaperunlessotherwisestated,wherehistheHubble Figure 2 shows L /M as a function of the frac- V gas constantinunitsof100kms−1 Mpc−1,Ω isthedensity tion of luminosity emitted from elliptical and S0 galax- 0 parameter, and λ is the cosmological constant. Under ies to the total luminosity for 18 clusters with type 0 this assumption, the present age of the universe is 11.8 mix data. It is found that L /M is constant for V gas Gyr. The value h = 0.7 is taken from recent determi- Mgas ≥ 1 × 1013h−2.5M⊙• clusters irrespective of type nations of H (e.g., Freedman 1999). To adopt different mix. This implies that L /M is not sensitive to the 0 V gas values for Ω and λ in the observationally reasonable changeinthepopulationsofgalaxiesformassiveclusters. 0 0 ranges of 0.2 ≤ Ω0 ≤ 1 and 0 ≤ λ0 ≤ 0.8 does not Figures 1 and 2 demonstrate that the dependence of significantly change our results. L /M on L (or on M ) is much weak for massive V gas V gas clusters. This is also supported by Renzini (1997) who found that rich clusters have a fairly constant M to 2. Data gas B-band luminosity ratio. The reason why poor clusters have a relatively higher L /M value is not clear, but Wedivideclustersofgalaxiesintonearbyclusters(z ≤ V gas apossibleexplanationisthatasignificantfractionofhot 0.1) and distant clusters (z > 0.1). We assume that gasinpoorclusters hasescapedfromthe clusters during evolutionary effects are negligible for z ≤ 0.1 clusters clusterevolutionowingto theirshallowgravitationalpo- and regard their properties as those of the present-day tentials, resulting in a higher L /M value (e.g., Ren- clusters. V gas zini 1997). We adopt the B band to measure the luminosity of galaxies,andthink thatit is the best compromise. Clus- In any case, in what follows we assume that LB/Mb ter luminosities have been measured mainly in optical is constant for Mgas ≥1×1013h−2.5M⊙• clusters and use bandpassessuchasB,V,orR. Amongtheopticalband- themtoderivetheaverageLB/Mbofnearbyclusters. We willseeinthenextsubsectionthatallthedistantclusters passes,B is mostsensitiveto the luminosityevolutionof galaxies. Though ultraviolet wavelengths such as U are adoptedinthispaperhaveMgas ≥1×1013h−2.5M⊙•. This much better for measuring star formation, data in such promisesafaircomparisonofLB/Mbbetweennearbyand distant clusters. wavelengthsare very few. As for field galaxies,there are alotofmeasurementsofluminositydensityoffieldgalax- For Arnaud et al.’s (1992) clusters which have mor- ies in the B band, which enables us to compare LB/Mb phologicaltypemix,wecomputetotalB-bandluminosity with lB/ρb. (LB)fromLV usingB−V =0.96(E),0.85(S0),and0.68 (S) (see Fukugita, Shimasaku, Ichikawa 1995). For clus- terswithouttype mixdata,weadoptB−V =0.85±0.2 2.1. Nearby Clusters as the average color of galaxies. We compute baryon We use the sample of nearby clusters given in Arnaud mass Mb from gas mass Mgas using: etal. (1992)toderiveLB/Mbofthepresent-dayclusters. Arnaud et al. (1992)compiled a sample of 27 clusters of Mb =Mgas+(M/LB)⋆LB, (1) galaxies, where total V-band luminosity (L ), morpho- V logicaltype mix ofgalaxies(E,S0, andS), and gasmass where (M/LB)⋆ is the mean mass-to-luminosity ratio of within a radius of 1.5h−1 Mpc (the Abell radius) are the stellar population in galaxies. We neglect atomic given. Morphological type mix is available for 18 clus- and molecular gas in galaxies. We adopt (M/LB)⋆ = ters. (6 ± 3)h(M/LB)⊙, which roughly covers the mass-to- Arnaud et al. (1992) found in their clusters a strong luminosityratioofellipticalgalaxies(vanderMarel1991; dependence of L /M on L : L /M ∝ L−0.9. If Pizzella et al. 1997) and of spiral disks (Bahcall 1984; V gas V V gas V such a strong dependence holds in the whole mass range Broeils,Couteau1997). The errorsinLB/Mb contain(i) of clusters, it would very much complicate a compari- errorsinLV whicharegiveninArnaudetal. (1992),(ii) son of LB/Mb among clusters having different masses. errorsin the mean B−V (only for clusterswithout type Thus, we first examine for what clusters such a strong mix data), and (iii) errors in (M/LB)⋆ in equation (1). dependence exists. Figure 1 plots L against M for Since most of the baryons in clusters are in form of hot V gas allthe clusters in Arnaudet al. (1992). The dependence gas, the error in Mb due to (iii) is only about 5 %. found in Arnaud et al. (1992) is shown as the dashed Figure 3 presents LB/Mb as a function of Mgas. The line. The solid line, on the other hand, is a regression filled and open circles indicate Arnaud et al.’s (1992) line between LV and Mgas for Mgas ≥ 1×1013h−2.5M⊙• clusters with and without type mix data, respectively. clusters, L ∝ M0.8. This is close to a linear regres- As seen in figure 1, a clear trend is seen in figure 3 that V gas sion, i.e., a constant LV/Mgas, indicated as the dotted clusterswithMgas∼<1×1013h−2.5M⊙• havesystematically line. Thus,the strongdependence ofLV/Mgas onLV (or higher LB/Mb. To derive LB/Mb of nearby clusters, we equivalently on Mgas) found by Arnaud et al. (1992) is not only remove clusters with Mgas < 1×1013h−2.5M⊙• probably due to the inclusion of less massive clusters. but also remove clusters without type mix data because 2 K. Shimasaku [Vol. 00, the uncertainties in L of these clusters are on the aver- M and L are given in Squires et al. (1997). L is B gas V B agelargerthanthoseforclustershavingtypemix(Toin- computed assuming rest-frame B−V =0.8±0.2. clude the clusters without type mix data hardly changes (iv) CL0500-24(z =0.32) the result, though). M is taken from Schindler and Wambsganss (1997). Twelve out of the 21 clusters with M ≥ 1 × gas gas L is givenin Infante et al. (1994), and L is computed 1013h−2.5M⊙• havetypemixdata,andtheirmeanLB/Mb usVingrest-frameB−V =0.85(ThemeanaBpparentcolor for h=0.7 is of this cluster, V −I = 1.8, reported by Infante et al. 1994 corresponds to B−V =0.85 in the rest frame). LB/Mb =(0.024±0.004)(LB/M)⊙, (2) (v) CL0939+47(z =0.41) which we regard as the representative value for the Schindler et al. (1998) found two substructures in this present-day clusters. The contribution from elliptical cluster,implyingthatthis clusterhasnotbeenvirialized and S0 galaxies to total B luminosity is on the aver- yet. M andL adoptedherearethesumofthevalues gas B age (69±13)% for the 12 clusters, implying that these forthetwosubstructuresgiveninSchindleretal. (1998). clusters are dominated by early-type galaxies (See figure Dressler et al. (1997)reported the fraction in number of 2). ClusterswhichhaveMgas <1×1013h−2.5M⊙• tendto elliptical and S0 galaxies to be 55%. be less dominated by early-type galaxies. For example, theVirgocluster,whichhasMgas =0.44×1013h−2.5M⊙•, (vi) RXJ1347.5-1145(z =0.45) has LB(E+S0)/LB(tot)=45%. Mgas is taken from Sahu et al. (1998) and LB is com- putedfromM andM /L giveninFischerandTyson Forthe12clusters,weestimatetheratioofstellarmass tot tot B to baryon mass to be M⋆/Mb = 0.10±0.05 (h = 0.7), (1997), where Mtot is the total mass of a cluster. using (M/LB)⋆ = (6±3)h(M/LB)⊙. This means that (vii) CL0016+16(z =0.55) only ∼10 % of baryons have been used to form stars to M is taken from Neumann and B¨ohringer (1997) and gas date in rich clusters. L is computed from the r-band total luminosity given B in Carlberg et al. (1996) using rest-frame B−r = 0.97, 2.2. Distant Clusters which corresponds to the observed g −r color of 1.455 (Carlberg et al. 1996). Dressler et al. (1997) reported Searchingfortotalluminosityandgasmassdataofdis- the fractionin number of elliptical and S0 galaxies to be tantclustersintheliterature,wetake22clusters,among 73%. whichthirteenarefromthe CNOC clustersample(Carl- bergetal. 1996;Lewisetal. 1999). Wedonotapplyany (viii) AXJ2019+1127(z =1.01) selection criterion to compile our sample. Data of these M is taken fromHattoriet al. (1997),and L is com- gas B clusters are given in table 1. All the clusters have rest- putedfromL (Benitezetal. 1998)assumingrest-frame V frame luminosity (either B, V, or r band) and gas mass B−V =0.7±0.3,whichroughlycoverstheexpectedcol- measurements. For a cluster whose rest-frame luminos- ors of elliptical and spiral galaxies at z =1. ity is in the V orr band, we use anobservedorassumed CNOC Clusters colorto convertthe luminosity to the rest-frameB-band Carlberg et al. (1996) give rest-frame r-band luminosity luminosity. at the virial radius for 16 clusters at 0.17 < z < 0.55. When computing Mb from Mgas by Mb =Mgas+M⋆, Out of them, 14 clusters have gas mass measurements weuseM⋆/Mb =0.10(h=0.7)whichisthevalueforthe (figure 4 of Lewis et al. 1999). Since the maximum ra- nearbyclusters. DistantclustersmayhavelowerM⋆/Mb dius atwhichgas massis plotted, r =600h−1 kpc for all values than the nearby clusters, but the uncertainties in the clusters,is smallerthanthe virialradii,wederivefor LB/Mb due to this effect are at most ≃ 10%, which is each cluster the r-band luminosity at 600h−1 kpc from negligible for our discussion. Below are the references to the valueatthe virialradiusassumingthatluminosityis the nine clusters and the CNOC clusters. proportional to radius. The ratio of 600h−1 kpc to the (i) Abell 1413 (z =0.14) and Abell 1689 (z =0.18) virialradiiof14clustersis onthe average0.48,implying These two clusters are taken from Cirimele, Nesci, that a large factor of conversion is necessary. For each Tr`evese’s(1997)sample. Thissampleconsistsof12clus- cluster, we then transform the rest-frame r-band lumi- ters with z <0.2 for which M and L within a radius nosityintotherest-frameB-bandluminosityonthebasis gas V of 0.75h−1 Mpc are given. We calculate L assuming of the observed g−r color given in figure 5 of Carlberg B rest-frame B−V =0.85±0.2. et al. (1996). CL0016+16 is among the 14 clusters. As seen in (vii), we have adopted for gas mass of this clus- (ii) Abell 2218 (z =0.18) ter the measurement given in Neumann and B¨ohringer We adopt M and L from Squires et al. (1996). gas B (1997)because their value is at r =1.67h−1 Mpc, which (iii) Abell 2163 (z =0.20) is verycloseto the virialradiuswherer-band luminosity No. 0] Luminosity Evolution of Galaxies in Clusters 3 is measured. The numberofthe CNOCclusters adopted CL0939+47 and CL0016+16) are the CNOC clusters. here is thus 13. Linesindicatemodelpredictions,whichwillbe discussed cluTstheersraadnidi umseodstfoofrtmheemasuarriengsmLaBl/leMrbthdainffetrheamAobneglltrhae- iwnitthhezn,exfrtomsubLseBc/tMionb.=It i0s.0fo2u4n(LdBt/hMat)L⊙B/aMt bzin=cre0asteos dius (See table 1). Unfortunately, it is not clear whether ≃ 0.06(LB/M)⊙ at z = 1, though the error in each dis- tant cluster is fairly large. CL0939+47 deviates largely LB/Mb measured at these small radii represent global from this trend. We suspect, however,that the observed values, i.e., values at the Abell radius, though there is acosntsutdanytthbaettwteheenLaB/rMadbiuosfotfhe≃C0o.m4ha−c1luMsteprcisanndeatrhlye LvaBlu/Me,bboecfatuhsisectlhuestoebrsdeorveesdnoLtBr/eMprbesisentthethvearlueael,fogrlotbwaol substructureswhoseradiiareonlyr =0.14h−1Mpc. The Abell radius (Taguchi et al. 1999). In this paper, we assume that the values of LB/Mb derived here represent TChNiOs Cmacylusrteeflrescsteeumncetortahianvteieas ldauregetroscaatltaerrgeinfaLcBto/Mr obf. the globalvalues of individual clusters (For CL0939+47, the conversion of optical luminosity from the value at see, however, the next section). the virial radius to that at r =600h−1 kpc. Figure 4 plots L againstM for the 12nearby clus- B gas Therearetwooppositeexplanationsfortheincreasein ters and the 22 distant clusters. Both L and M are values at r = 1.5h−1 Mpc, which are deBrived frogmasraw LB/Mb with redshift. One is the brightening of LB due tothe luminosityevolutionofthegalaxypopulationasa values on the assumption that L (r) and M (r) are B gas whole. Notethatgalaxymergings,eveniftheyoccur,do proportional to radius r. The thick solid line indicates not change the total mass of the galaxy population and the best fit of a linear law, L ∝ M , to the nearby B gas thatstarformationwhichcouldbetriggeredbymergings clusters. The thin solid line and the dotted line corre- can be treated in the framework of the ‘pure’ luminosity spond to a similar fit to distant clusters at 0.1<z ≤0.4 evolution of the galaxy population. The other explana- and 0.4 < z ≤ 0.7, respectively. It is found that the tion is that the mass of baryons (≃ hot gas) per galaxy average B luminosity at a given gas mass increases with decreaseswithredshift. However,thisexplanationseems redshift. This should reflect some evolution of L /M . B gas to be less plausible, because no significant evolution has Note that the range of gas mass is similar between the been observationally found for the global properties of nearby and distant clusters: there is no distant cluster in our sample whose gas mass at r =1.5h−1 Mpc is less clusters at z∼<1 (e.g., Schindler 1999) (This result is, than 1×1013h−2.5M⊙•. Note also that no clear depen- however, mainly for X-ray properties, and the evolution of the galaxy distribution in clusters is not well known). dence of L /M on M is seen either in the nearby B gas gas Inwhatfollows,wetaketheformerexplanationasour cluster sample or in the distant cluster sample. We mention here the effects of changing Ω and λ on hypothesis, i.e., we assume that the increase in LB/Mb 0 0 found here is due to pure brightening of the galaxy pop- estimates of LB/Mb for distant clusters. Let dL(z) and ulation as a whole. Then the increase found here corre- dA(z) be the luminosity distance and the angular diam- sponds to brightening by ≃ 1 mag of the galaxy popu- eter distance to a cluster at z, respectively. The ratio L /M is proportional to d−0.5(z), because of L ∝ lation. In the next subsection, we compare the observed B gas L B d2L(z), Mgas ∝d2A.5(z), anddL(z)=(1+z)2dA(z). Thus, brightening with predictions of simple luminosity evolu- L /M depends on Ω and λ through d−0.5(z). Since tion models of galaxies. B gas 0 0 L Mb is dominated by Mgas, the dependence of LB/Mb on Ω0 and λ0 is very close to that of LB/Mgas. Figure 5 3.1. Comparison with Luminosity Evolution Models shows the ratio of L /M (Ω ,λ ) to L /M (Ω = B gas 0 0 B gas 0 We characterize the evolution of L , the B-band lu- 0.2,λ = 0) as a function of redshift for two sets of B 0 minosity summed over all galaxies in a cluster, by two (Ω0,λ0). Since dL(z) is a decreasing function of Ω0 and parameters: the star formation timescale τ and the for- an increasingfunction of λ up to at leastz =1, we find 0 mation redshift z . In other words, we assume that all from this figure that the change in L /M due to the F B gas galaxiesare formed at the same redshift z and that the change in Ω and λ in the ranges of 0.2 ≤ Ω ≤ 1 and F 0 0 0 e-folding time of star formation summed over all galax- 0≤λ ≤0.8 is less than ±10% for z <1 clusters, which 0 ies is τ (Gyr). We compute B-bandluminosityusing the is negligible for our discussion below. populationsynthesiscodedevelopedbyKodamaandAri- moto (1997). The values of τ and z examined here are F 3. Results and Discussion τ =0.1,1,2,3,and 5 Gyr and z =2,3, and ∞. Models F with τ = 0.1 Gyr correspond to elliptical galaxies and Figure 6 shows LB/Mb of clusters as a function of models with τ = 5 Gyr are for spiral disks like that of redshift. The filled circles present the distant clusters ourGalaxy. We donotexamineτ >5 Gyr,since to take and the filled square indicates the average LB/Mb of τ >5 Gyr leads to too blue colorsatz =0 which arein- the nearby clusters. Clusters without errors (but for consistent with the observed colors of galaxies in nearby 4 K. Shimasaku [Vol. 00, clusters (The meanB−R ofthe Virgo andComa clus- VanDokkumetal. (1998)presentobservationsofM/L C tergalaxiesis1.2and1.8,respectively[Andreon1996for ratio inthe B band of early-typegalaxiesin five clusters Coma and Young and Currie 1998 for Virgo], while the at0.02≤z ≤0.83. TheyfindthattheM/Lratioevolves models with τ = 0.1 and τ = 5 Gyr give B−R = 1.6 as ∆logM/L ∝ −0.40z (Ω = 0.3,λ = 0), which is C B 0 0 andB−R =1.1,respectively,atanageof12Gyr). We consistentwithsingle-burstmodelswithz >1.7−2.8. If C F also set the lower limit of z to be 2 following the tra- theevolutionofM/LfoundbyvanDokkumetal. (1998) F ditional pure luminosity evolution models which assume isunderstoodaspureluminosityevolutionofL ,thefor- B that elliptical/S0 galaxies and spiral galaxies are formed mula ∆logM/L ∝ −0.40z implies that L brightens B B at high redshifts (z∼>2) and which broadly succeed in by a factor of 2.5 from z = 0 to 1, which is in excellent reproducing observed properties of these galaxies (e.g., agreement with the brightening of LB/Mb found in this Kodama et al. 1998; Shimasaku and Fukugita 1998). study. Measurements of M/L of individual galaxies are Figure 6 compares the observed LB/Mb with predic- a direct measurement of the effects of luminosity evolu- tions. Predicted values of LB/Mb are normalized to tion occurredin galaxies,while measurements of LB/Mb matchtheobservedvalueatz =0. Inotherwords,mod- of clusters are less direct. Note, however, that informa- els are used to predict relative brightening (or fading) of tioncontainedinLB/Mb isdifferentfromthatinM/Lof LB as a function of z. Panels (a), (b), and (c) are for individual galaxies. The quantity LB/Mb describes the z = ∞,3, and 2, respectively. From panel (a), we find evolutionofluminosity summedoverall galaxiesin clus- F that all the models reproduce the observation. If, how- ters. LB/Mb also gives us a hint to the star formation ever, z = 3 is adopted (panel [b]), models with τ = 1 efficiency in clusters (see below). In any case, the agree- F and 2 Gyr give too steep brightening comparedwith the ment of brightening between M/LB and LB/Mb found observation. This trend is strengthened for the z = 2 here can be regarded as indirect support of our conclu- F case (panel [c]): the τ =0.1 Gyr model also becomes in- sionthatsingle-burstlikemodelsseemtobeplausiblefor consistent with the observation, though the discrepancy describing the evolution of LB/Mb. is at less than 2σ levels. The allowed range for τ is de- The absolute value of LB/Mb tells us a hint to the pendent on z , and we cannot rule out any value of τ star formation efficiency in clusters, i.e., the fraction of F on the basis of the current data if we permit z = ∞, baryonsin clusters used to formstars. If the star forma- F though‘single burst’models with τ =0.1 Gyr and‘disk’ tion efficiency differs among clusters, the absolute value models with τ = 5 (and τ = 3 Gyr models) match the of LB/Mb would also vary from cluster to cluster. The observationforawiderrangeofz towardlowerredshifts fact that there exist models, such as those with τ = 0.1 F than the other (τ =1,2 Gyr) models. Gyr,whichreproducetheobservedLB/Mb ofmanyclus- Itisinterestingthat‘singleburst’(τ =0.1Gyr)models ters at different redshifts within the observational errors and ‘disk’ (τ = 5 Gyr) models are consistent with the suggest that the star formation efficiency is universal observation. In this paragraph, we concentrate on these among clusters up to z ∼1. models,andexaminewhicharemoreconsistentwiththe In order to see when baryons are converted into stars, observed luminosity evolution of individual galaxies in we plot in figure 7 the predicted evolution of M⋆/Mb, clusters. Various observationssuggest that elliptical and where the evolution of M⋆ is calculated from mass-to- S0 galaxies in clusters brighten by ∼ 1 mag from z = 0 luminosity ratios of galaxies predicted by the z = 3 F to z =1, which is consistent with the single burst model models. The values of M⋆/Mb at z = 0 have been nor- (e.g., Schade, Barrientos, Lo´pez-Cruz 1997; Ellis et al. malized so that they are consistent with the observed 1997; Kodama et al. 1998; van Dokkum et al. 1998). LB/Mb of the nearby clusters 1. The predicted values of Spiral galaxies have also been found to brighten by ∼ 1 M⋆/Mb atz =0are0.05−0.13,dependingonτ, andare mag (e.g., Schade et al. 1996), though observations are consistentwiththeobservedvalue(thefilledsquarewith not as many as those of elliptical and S0 galaxies. The an error bar). This simply implies that the predicted amounts of brightening of E/S0 and spiral galaxies are (M/LB)⋆ values at z = 0 fall within (6±3)h(M/LB)⊙, similar to each other, and they are in agreement with which is the adopted value of (M/LB)⋆ for galaxies in the predictions of τ = 0.1 Gyr models (with zF ≥ 3) nearby clusters. As expected, the evolution of M⋆/Mb and τ = 5 Gyr models. However, our distant clusters largely differs among the models. A constant M⋆/Mb is are rich clusters and thus it is likely that elliptical and predictedby theτ =0.1Gyrmodelinthe redshiftrange S0 galaxies dominate in these clusters. Hence, the τ = of this figure, while for the τ =5 Gyr model, half of the 0.1 Gyr models seem to be more plausible for describing stars present today are formed at z <1. the evolution of LB/Mb. This is also supported by the Finally,wementionhowtoputstrongerconstraintson fact that the mean colorof galaxiesin the Coma cluster, modelsusingtheevolutionofLB/Mb. Unfortunately,the which is a very rich nearby cluster and is likely to be a steepness of brightening up to z = 1 is not a monotonic counterpart of the distant clusters studied here, agrees with the color predicted by the τ =0.1 Gyr models. 1 M⋆/Mb(z=0)=(M/LB)p⋆red(z=0)×(LB/Mb)obs(z=0) No. 0] Luminosity Evolution of Galaxies in Clusters 5 functionofτ: thebrighteningisthesteepestforτ =1−2 this paper, we regard the values at z ≃ 0.1 as the local Gyr models, and models with τ =0.1 Gyr and τ =3−5 (z =0) value. Gyr give similar brightening. In order to place further constraintsusingtheevolutionofLB/Mb,oneneedsdata 3.2.1. Local values for LB/Mb and lB/ρb at z > 1: for example, for z = 3 and 2, τ = 0.1 Gyr F The filled circles in figure 8 indicate the LB/Mb of the models predict much steeper brightening at z > 1 than distant clusters of galaxies. The filled square at z = 0 τ =5 Gyr models. corresponds to the average of the nearby clusters. We find that the local LB/Mb agrees with the local lB/ρb: 3.2. Comparison with Evolution of Field Galaxies LB/Mb is(0.024±0.004)(LB/M)⊙ andtheaverageofthe The quantity LB/Mb is the B-band luminosity per three lB/ρb values at z ≃0.1 is (0.026±0.002)(LB/M)⊙ unit baryon in clusters. The corresponding quantity for 3. field galaxies is the ratio of the blue luminosity density This agreement implies that the blue luminosity per lB [LB⊙ Mpc−3] to the mean baryon density ρb [M⊙• unitbaryonmassisveryclosebetweenclustersandfields. Mpc−3]. Inthissubsection,wederivelB/ρboffieldgalax- Then the next questionmay be whether the stellar mass ies up to z ∼ 1 from the literature and compare it with per unit baryon mass (M⋆/Mb), i.e., the star formation LB/Mb of clusters. efficiency, is the same between clusters and fields (Note Data of l are taken (or computed) from recent mea- thatmorphologicaltypemixlargelydiffersbetweenclus- B surements of luminosity function based on redshift sur- ters and fields and that (M/LB)⋆ of galaxies varies with veys: Lilly et al. (1996; CFRS; data points are at morphology). In order to examine this, we do a sim- z = 0.35,0.625,0.875), Ellis et al. (1996; Autofib; z = ple (but more detailed than that given in §2.1) esti- 0.085,0.25,0.55),Colless (1998; 2dF; z =0.11), Loveday mation of M⋆/Mb of clusters and fields below. We as- et al. (1992; APM; z ≃ 0.05), Zucca et al. (1997; ESP; sume the (M/LB)⋆ of elliptical and S0 galaxies to be z ≃ 0.1), and Marzke et al. (1998; SSRS2; z ≃ 0.025). 8h(M/LB)⊙ and that of spiral and irregular galaxies to Lilly et al. (1996) give lB itself while the other papers be4h(M/LB)⊙ 4. Usingthesevaluesandtakingaccount give only luminosity functions in the B band. For those ofthemeanmorphologicaltypemixofthe12clusters,we exceptforLillyetal. (1996),weintegratetheluminosity obtainM⋆/Mb =0.114±0.020. A similarcalculationfor function given in each paper from M = −25 to −10 to the field galaxies gives M⋆/Mb = 0.097±0.009 (We use B obtain lB. thetypemixgiveninColless1998: lB(E+S0)/lB(tot)= Inordertocomputeρb,weadoptΩb =0.02h−2follow- 34.5% and lB(S+Irr)/lB(tot) = 65.5%). Hence, the ingTytlerandhis coworkers’results(e.g.,Burles,Tytler M⋆/Mb of clusters is in agreement with that in fields within errors. Renzini (1997)has alsoobtained a similar 1998). Their estimates ofΩb arebasedonmeasurements of deuterium abundance (D/H) of QSO absorption sys- result from a rough calculation for M⋆/Mb of the Coma cluster and field galaxies. temsathighredshifts. NotethatthemeasurementsofΩb have not completely converged among authors, ranging fromΩbh2 ≃0.01to0.02,thoughTytleretal.’smeasure- 3.2.2. Evolution of LB/Mb and lB/ρb ments seem to be the most reliable (e.g., Turner 1999). Fromfigure 8,it is found that the evolution ofLB/Mb ThelB/ρb offieldgalaxiescalculatedaboveareplotted as a function of redshift is the same as for lB/ρb within in figure 8 as open circles with error bars. A gradual the observational errors. Both ‘brighten’ by a factor of increaseinlB/ρb with redshiftis seen. This is due to the 2−3 from z =0 to z =1. brightening of lB 2. The agreement between the evolution of LB/Mb and AfactwhichneedsattentionisthatthevaluesoflB/ρb lB/ρb may be interpreted as the mean star formation atz∼<0.05aresmallerthanthoseatz ≃0.1byaslargeas history of cluster galaxies being similar to that of field a factorof ≃2. It is unlikely that the luminosity density galaxies. However,wethink thatthis agreementis prob- evolves so rapidly for such a short time from z = 0.1 to ably superficial. The observedglobalstar formationrate the present. A possible explanation for this problem is of field galaxies increases by a factor of ∼ 3−10 from that the number density of galaxiesin the local(z∼<0.1) 3 Anoteshouldbeadded. Theagreement foundheredoesnot universe happens to be lower than the globalvalue (e.g., holdifverydifferentvaluesforΩb andhareadopted, because Marzke et al. 1998), though further investigations are (i)lB/ρb changeslinearlywithΩ−b1 and(ii)thedependenceof neededinorderto proveordisprovethis explanation. In lB/ρb andLB/Mb onhisdifferent(lB/ρb∝hifΩbh2 isfixed, while LB/Mb ∝ h0.5), though we think that the values of Ωb 2 Forfieldgalaxies,theglobalstarformationrate(ρ˙⋆ [M⊙• yr−1 andhadopted inthispaperarethemostprobableatpresent. Mpc−3]) has also been measured from UV and emission-line 4 The value for elliptical and S0 galaxies is based on van der luminosities of galaxies (e.g., Madau et al. 1998; for recent Marel’s(1991)study,whichfoundthattheM/LB ofelliptical observations, see Cowie, Songaila, Barger 1999). These mea- galaxieswith1×1010h−2LB⊙ is8.4h(M/LB)⊙. Thevaluefor surements suggest that ρ˙⋆ increases by a factor of ∼ 3−10 theGalacticdisk,M/LB≃3(M/LB)⊙,isadoptedasthevalue fromz=0to1,butthescatter amongthedataisstilllarge. forspiralandirregulargalaxies(4h≃3forh=0.7). 6 K. Shimasaku [Vol. 00, the present epoch to z = 1, and then has a peak at We have made a simple estimate of star formation ef- z = 1−2 (e.g., Madau et al. 1998; Cowie et al. 1999). ficiency (M⋆/Mb) to find no difference between clusters Thoughthisstarformationhistory,whichreproducesthe andfields. To place further constraintsonthe meanstar observed evolution of lB/ρb as well, should be a solution formation history of cluster galaxies needs new data at for the mean star formation history of cluster galaxies, higherredshiftsordirectmeasurementsofstarformation quite different models can also be solutions, as has been rate. seen in §3.1. Simple models having just two parame- ters(τ andz )wereexaminedin§3.1,andmanymodels Acknowledgements F have been found to reproduce the observed evolution of We thank the anonymous referee for useful comments LB/Mb, and the color of galaxies in nearby rich clusters which improved the paper. This research was supported suggests that τ =0.1 Gyr models are favored (see §3.1). in part by the Grants-in-Aid by the Ministry of Educa- In any case, we cannot give a clear conclusion about tion, Science, Sports and Culture of Japan (07CE2002) the mean star formation history of cluster galaxies on to RESCEU (Research Center for the Early Universe). the basis of the current data. As was mentioned in §3.1, data at z > 1 are useful to place further constraints on References the star formation history. More desirable may be data ofultravioletluminosityper unit baryonmass,LUV/Mb, Andreon, S. 1996, A&A 314, 763 from which one can measure star formation rate (and Arnaud, M., Rothenflug, R., Boulade, O., Vigroux, L., efficiency) directly. Vangioni-Flam, E. 1992, A&A 254, 49 Bahcall, J. N. 1984, ApJ 287, 926 4. Conclusions Balogh, M., Schade, D., Morris, S. L., Yee, H. K. C., Carlberg, R. G., Ellingson, E. 1998,ApJ 504, L75 WehavederivedLB/Mbformassive(MgasattheAbell Benitez, N., Broadhurst, T., Rosati, P., Courbin, F., radiusis≥1×1013h−2.5M⊙•)clustersofgalaxiesuptoz ≃ Squires, G., Lidman, C., Magain, P. 1998, astro- 1 from opticalandX-ray data in the literature. Twenty- ph/9812218 twoclustersinoursampleareatz >0.1. Assumingthat Broeils, A. H., Couteau, S. 1996, in Dark and Visible the relative mix of hot gas and galaxies in clusters does Matter in Galaxies, ASP ConferenceSeries,Vol. 117; not change (i.e., no segregation in hot gas or galaxies) 1997; ed. M. Persic and P. Salucci, p74 duringclusterevolution,weuseLB/Mb toprobethestar Burles, S., Tytler, D. 1998, ApJ 507, 732 formation history of the galaxy population as a whole Carlberg, R. G., Yee, H. K. C., Ellingson, E., Abraham, in clusters. We have found that the LB/Mb of clusters R., Gravel, P., Morris, S., Pritchet, C. J. 1996, ApJ increases with redshift from LB/Mb = 0.024(LB/M)⊙ 462, 32 (z = 0) to ≃ 0.06(LB/M)⊙ (z = 1), indicating a factor Cirimele, G., Nesci, R., Tr`evese,D. 1997, ApJ 475, 11 of ∼ 2−3 brightening. This amount of brightening is Colless, M. 1998, astro-ph/9804079 almost identical to the brightening of the M/L ratio of Cowie, L. L., Songaila, A., Barger, A. J. 1999, astro- B early-typegalaxiesinclustersat0.02≤z ≤0.83reported ph/9904345 by van Dokkum et al. (1998). Dressler, A., Oemler, A., Jr., Couch, W. J., Smail, I., We have compared this result with luminosity evo- Ellis, R. S., Barger, A., Butcher, H., Poggianti, B. lution models for the galaxy population as a whole by M., et al. 1997, ApJ 490, 577 changing the e-folding time of star formation τ by 0.1≤ Ellis, R. S., Colless, M., Broadhurst, T., Hely, J., Glaze- τ ≤5 Gyrandthe formationredshiftz by 2≤z <∞. brook, K. 1996, MNRAS 280, 235 F F We have found that ‘single burst’ models (τ = 0.1 Gyr Ellis,R.S.,Smail,I.,Dressler,A.,Couch,W.J.,Oemler, models) with z ≥3 and ‘disk’ models (τ =5 Gyr) with A., Jr., Butcher, H., Sharples, R. M. 1997, ApJ 483, F arbitraryz areconsistentwiththeobservedbrightening 582 F ofblue luminosity to z =1,while models with1≤τ ≤2 Fischer, P., Tyson, A. J. 1997, AJ 114, 14 Gyrtendtopredicttoosteepbrighteningthoughwecan- Freedman, W. L. 1999, astro-ph/9905222 not rule out these models. Fukugita, M., Shimasaku, K., Ichikawa, T. 1995, PASP We have also derived the ratio of blue luminosity den- 107, 945 sitytobaryondensity,lB/ρb,forfieldgalaxiesuptoz ≃1 Hattori, M., Ikebe, Y., Asaoka, I., Takeshima, T., from various existing data, adopting Ωbh2 = 0.02, and B¨ohringer, H., Mihara, T., Neumann, D. M., have found that the observedevolutionof LB/Mb agrees Schindler, S., Tsuru, T., Tamura, T. 1997, Nature withthatoflB/ρb,includingtheabsolutevalues,fromthe 388, 146 present epoch to z ≃ 1 within the observational uncer- Infante,L.,Fouque,P.,Hertling,G.,Way,M.J.,Giraud, tainties, indicating that blue luminosity per unit baryon E., Quintana, H. 1994, A&A 289, 381 mass is similar between clusters and fields up to z ≃ 1. Kodama, T., Arimoto, N. 1997, A&A 320, 41 No. 0] Luminosity Evolution of Galaxies in Clusters 7 Kodama, T., Arimoto, N., Barger, A. J., Arag´on- Zucca, E., Zamorani, G., Vettolani, G., Cappi, A., Salamanca, A. 1998, A&A 334, 99 Merighi,R.,Mignoli,M.,Stirpe,G.M.,MacGillivray, Lewis, A. D., Ellingson, E., Morris, S. L., Carlberg, R. H., et al. 1997, A&A 326, 477 G. 1999, ApJ 517, 587 Lilly, S. J., Le F`evre, O., Hammer, F., Crampton, D. 1996, ApJ 460, L1 Lilly, S. J., Schade, D., Ellis, R., Le F`evre, O., Brinch- mann, J., Tresse, L., Abraham, R., Hammer, F., et al. 1998,ApJ 500, 75 Loveday,J., Peterson, B. A., Efstathiou, G., Maddox, S. J. 1992, ApJ 390, 338 Madau, P., Pozzetti, L., Dickinson, M. 1998, ApJ 498, 106 Marzke,R.O.,daCosta,L.N.,Pellegrini,P.S.,Willmer, C. N. A., Geller, M. J. 1998, ApJ 503, 617 Neumann, D. M., B¨ohringer, H. 1997, MNRAS 289, 123 Renzini A. 1997, ApJ 488, 35 Pizzella, A., Amico, P., Bertola, F., Buson, L. M., Danziger, I. J., Dejonghe, H., Sadler, E. M., Saglia, R.P.,deZeeuw,P.T.,&Zeilinger,W.W.1997,A&A 323, 349 Poggianti, B. M., Smail, I., Dressler, A., Couch, W. J., Barger, A., Butcher, H., Ellis, R. S., Oemler, A., Jr., 1999, ApJ 518, 576 Sahu, K. C., Shaw, R. A., Kaiser, M. E., Baum, S. A., Ferguson, H. C., Hayes, J. J. E., Gull, T. R., Hill, R. J., Hutchings, J. B., Kimble, R. A., Plait, P., & Woodgate, B. E. 1998, ApJ 492, L125 Schade, D., Carlberg, R. G., Yee, H. K. C., Lo´pez-Cruz, O., Ellingson, E. 1996, ApJ 465, L103 Schade, D., Barrientos,L. F., Lo´pez-Cruz, O. 1997,ApJ 477, L17 Schindler, S. 1999, astro-ph/9908130 Schindler, S., Belloni, P., Ikebe, Y., Hattori, M., Wamb- sganss, J., Tanaka, Y. 1998, A&A 338, 843 Schindler, S., Wambsganss, J. 1997,A&A 322, 66 Shimasaku, K., Fukugita, M. 1998,ApJ 501, 578 Squires, G., Kaiser,N., Babul, A., Fahlman, G., Woods, D., Neumann, D. M., B¨ohringer, H. 1996, ApJ 461, 572 Squires, G., Neumann, D. M., Kaiser, N., Arnaud, M., Babul, A., B¨ohringer, H., Fahlman, G., Woods, D. 1997, ApJ 482, 648 Taguchi, H., Shimasaku, K., Doi, M., Okamura,S. 1999, in preparation Turner, M. S. 1999,astro-ph/9904051 van der Marel, R. P. 1991,MNRAS 253, 710 van Dokkum, P. G., Franx, M., Kelson, D. D., Illing- worth, G. D. 1998, ApJ 504, L17 Vogt, N. P., Phillips, A. C., Faber, S. M., Gallego, J., Gronwall, C., Guzma´n, R., Illingworth, D., Koo, D. C., & Lowenthal, J. D. 1997, ApJ 479, L121 Young, C. K., Currie, M. J. 1998, A&AS 127, 367 8 K. Shimasaku [Vol. 00, Table 1. Distant Clusters. name z L /Ma) Lb) Mc) radiusd) errore) B b B gas A1413 0.14 0.028 7.6 2.42 0.77 21 A1689 0.18 0.042 14.0 2.80 0.78 21 A2218 0.18 0.043 8.4 1.81 0.42 33 A2163 0.20 0.037 3.0 0.74 0.26 48 CL0500-24 0.32 0.037 3.9 0.95 0.53 30 CL0939+47 0.41 0.089 1.8 0.18 0.14 — RXJ1347.5-1145 0.45 0.053 65.7 11.3 1.09 37 CL0016+16 0.55 0.044 50.6 10.4 1.67 — AXJ2019+1127 1.01 0.075 5.6 0.68 0.30 30 A2390 0.23 0.030 11.3 3.3 0.60 — MS0302+16 0.42 0.055 5.1 0.84 0.60 — MS0440+02 0.20 0.027 2.5 0.84 0.60 — MS0451+02 0.20 0.021 6.3 2.7 0.60 — MS0451-03 0.54 0.036 14.7 3.7 0.60 — MS0839+29 0.19 0.054 4.5 0.74 0.60 — MS0906+11 0.17 0.066 9.2 1.3 0.60 — MS1006+12 0.26 0.028 6.4 2.0 0.60 — MS1008-12 0.31 0.051 11.2 2.0 0.60 — MS1224+20 0.33 0.066 7.1 0.98 0.60 — MS1358+62 0.33 0.062 11.4 1.7 0.60 — MS1455+22 0.26 0.021 4.9 2.1 0.60 — MS1512+36 0.37 0.028 4.2 1.4 0.60 — a) In units of h0.5(LB/M)⊙. b) In units of h−2×1011L⊙•. c) In units of h−2.5×1013M⊙•. d) Radius in units of h−1 Mpc adopted to measure L B and M . gas e) Relative error (%). No. 0] Luminosity Evolution of Galaxies in Clusters 9 Figure Captions catethe22distantclustersandthefilledsquareisforthe average LB/Mb of the nearby clusters. The open circles Fig.1. LV plottedagainstMgas forArnaudetal.’s(1992) with error bars correspond to lB/ρb of field galaxies. nearbyclusters. Thefilledandopencirclesindicateclus- ters with and without type mix data, respectively. The dashedandsolidlinesindicatethebestfitofapowerlaw to all and massive (Mgas ≥ 1×1013h−2.5M⊙•) clusters, respectively. The dotted line corresponds to the best fit of L ∝M to the massive clusters. V gas Fig.2. L /M plottedagainstthefractionofluminosity V gas emitted fromelliptical and S0 galaxies to the totallumi- nosity L (E/S0)/L for Arnaud et al.’s (1992) clusters V V havingtypemixdata. Thefilledandopencirclesindicate Mgas ≥ 1×1013h−2.5M⊙• and Mgas < 1×1013h−2.5M⊙• clusters, respectively. Fig.3. LB/Mb as a function of Mgas for Arnaud et al.’s (1992)clusters. The filled and opencircles indicate clus- ters with and without type mix data, respectively. Fig.4. L versus M for the 12 nearby and the 22 dis- B gas tant clusters adopted in this paper. Filled circles indi- cate the nearby clusters. Open circles and crosses are for 0.1 < z ≤ 0.4 and 0.4 < z ≤ 0.7 clusters, respec- tively. The star corresponds to the most distant cluster AXJ2019+1127 at z = 1.01. The thick solid line indi- cates the best fit of L ∝ M to the nearby clusters. B gas The thin solid line and the dotted line correspond to a similar fit to the distant clusters at 0.1 < z ≤ 0.4 and 0.4<z ≤0.7, respectively. Fig.5. DependenceofthemeasurementofL /M onΩ B gas 0 andλ . TheratioofL /M (Ω ,λ )toL /M (0.2,0) 0 B gas 0 0 B gas is plotted as a function of redshift for (Ω ,λ ) = (1,0) 0 0 and (0.2,0.8) cases. Fig.6. Observed LB/Mb of clusters as a function of red- shift. The filled circles present the 22 distant clusters and the filled square indicates the averageLB/Mb of the nearby clusters. Model predictions are overlaid. Pre- dicted values of LB/Mb are normalized to match the ob- served value at z = 0. Panels (a), (b), and (c) are for z = ∞,3, and 2, respectively. Thick and thin solid F lines indicate models with τ = 0.1 and 1 Gyr, respec- tively. Dotted, dashed,andlong-dashedlinescorrespond to models with τ =2,3, and 5 Gyr, respectively. Fig.7. Theratioofstellarmasstobaryonmass,plottedas a function of redshift. The five lines indicate predictions of models with different τ. All models are for z = 3. F The meanings of lines are the same as in figure 6. The filled square corresponds to the observed value. Fig.8. lB/ρb of field galaxies as a function of redshift, comparedwithLB/Mb ofclusters. Thefilledcirclesindi-