Source ot Acquisition NASA Goddard Space Flight Center submitted to the Astrophysical Journal On Iron Enrichment, Star Formation, and Type Ia Supernovae in Galaxy Clusters Michael Loewensteinl Laboratory for High Energy Astrophysics, NASA/GSFC, Code 662, Greenbelt, MD 20771 [email protected] ABSTRACT The nature of star formation and Type Ia supernovae (SNIa) in galaxies in the field and in rich galaxy clusters are contrasted by juxtaposing the build- up of heavy metals in the universe inferred from observed star formation and supernovae rate histories with data on the evolution of Fe abundances in the intracluster medium (ICM). Models for the chemical evolution of Fe in these environments are constructed, subject to observational constraints, for this pur- pose. While models with a mean delay for SNIa of 3 Gyr and standard initial mass function (IMF) are fully consistent with observations in the field, cluster Fe enrichment immediately tracked a rapid, top-heavy phase of star formation although transport of Fe into the ICM may have been more prolonged and - star formation likely continued beyond redshift 1. The means of this prompt enrichment consisted of SNII yielding 2 0.1 Ma per explosion (if the SNIa rate normalization is scaled down from its value in the field according to the relative number of candidate progenitor stars in the 3 - 8 Ma range) and/or SNIa with short delay times originating during the rapid star formation epoch. Star for- mation is > 3 times more efficient in rich clusters than in the field, mitigating the overcooling problem in numerical cluster simulations. Both the fraction of baryons cycled through stars, and the fraction of the total present-day stellar mass in the form of stellar remnants, are substantially greater in clusters than in the field. Subject headings: galaxies: clusters: general - galaxies: evolution - stars: formation - supernovae: general - X-rays: galaxies: clusters 'Allsow ith the University of Maryland Depa.rtment of Astroiiomy f I - 2 - 1. Introduction Rich clusters of galaxies provide a uniquely amenable setting for the study of the complex processes and consequences of galaxy formation and evolution. As the largest (2 3 M,) - virialized objects in the universe, with the deepest potential wells, they retain all of their processed and unprocessed baryonic matter in the form of stars and (predominantly) 3 - 10 keV plasma that are readily studied at infrared/optical/ultraviolet and X-ray wavelengths, respectively. Rich clusters are sufficiently vast to be considered representative volumes of the universe and are often assumed to bc composed of dark matter, stars, and gas in cosmic proportions. However, because clusters originate as extreme peaks in the random field of initial cosmolog- ical density fluctuations, the evolution of their constituent galaxies and by extension the - intergalactic medium that they are coupled to - may proceed in a manner that is distinct from their counterparts in the field. The resulting biases must be understood, and taken into account, when generalizing from cluster-based observables in order to draw conclusions about the universe as whole. Ir, fact, the morphological mix, luminosity function, and star formation history of cluster galaxies all display signatures of the effects of the exceptionally dense environments where they form and develop (Kuntschner et al. 2002; Croton et al. 2005; Schindler et al. 2005; Romeo, Portinari, & Sommer-Larsen 2005). Since galactic outflows as- sociated with the galaxy formation process, in turn, feed back on the cluster environment by injecting energy and metals into intergalactic plasma, evidence of the cluster/field dichotomy is implanted in the intracluster medium (ICM). A powerful approach to quantifying the distinctions between the evolution of baryons in clusters and in the field, is to juxtapose the build-up of heavy metals in the universe inferred from the evolving star formation and supernovae rates with the enrichment history of the ICM in rich galaxy clusters - the dominant reservoir for metals produced by stars in the ensemble of cluster galaxies. Because, under intracluster plasma conditions, abundances of Fe are more easily measured than those of other elements they are derived to redshifts > 1 and provide the strongest current constraints on the evolution of ICM enrichment. Mushotzky & Loewenstein (1997) made an initial investigation along these lines based on their analysis of ASCA spectra of clusters out to z = 0.3, and concluded that ICM enrichment was inconsistent with concurrent estimates of the star formation rate history derived for field galaxies (see, also, Madau, Pozzetti, & Dickinson 1998; Lin 8~ Mohr 2004; Calura & Matteucci 2004). Re-examination of this issue is timely given the extension to higher redshift of ICM Fe abundance measurements made possible with XMM-Newton (Tozzi et al. 2003; Hashimoto et al. 2004; Maughan et al. 2004), and direct estimates of the evolution of supernova rates to comparable redshift in the field (Strolger et al. 2004; Dahlen et sl. 2004; I I -3- Barris & Tonry 2005) and in clusters (Gal-Yam, Maoz, & Sharon 2002) that constrain the level of supernova metal production per unit star formation. Star formation with a standard initial mass function (IMF) produces approximately one core collapse supernova (hereafter, SNII) per hundred solar masses of stars formed, while recent empirical estimates for the specific rate of Type Ia supernovae (SNIa) are an order of magnitude lower. For typical adopted Fe yields of 0.07 Ma for SNII (slightly higher than the empirical estimate of Elmhamdi, Chugai, & Danziger 2003) and 0.7 Mo for SNIa (Iwamoto et al. 1999), the estimated universal fraction of baryons in stars of 0.06 (Fukugita & Peebles N 2004) thus implies an Fe enrichment per baryon of 0.065/(1 - R)r elative to solar (solar N Fe mass fraction = 0.0013) that originates in comparable portions from the two classes of supernov-ae , where R is the stellar mass loss fraction integrated over the age of the universe. Since R 40% for a standard IMF (see below), one expects a mean universal Fe enrichment per baryon to about one-tenth solar. While this may be consistent with the low-redshift (non-cluster) IGM (Prochaska 2004; also, see 54.1 below), it falls short by a factor of 4 N in the ICM (see, also, Page1 2002). A simple way to account for this shortfall is to invoke a top-heavy IMF (David, Forman, & Jones 1991; Arnaud et al. 1992; Elbaz, Arnaud, & Vangioni-Flam 1995; Matteucci & Gibson 1995; Loewenstein & Mushotzky 1996; Gibson &z Matteucci 1997) to enhance the formation efficiency of SNII, and perhaps SNIa. Additional variations (some linked to the IMF) include the following: (1)a higher average star formation rate (more SNII and SNIa), (2) a higher incidence of SNIa per star formed, (3) a higher average Fe yield per (Type Ia and/or Type 11) supernova, (4) a significant enrichment by a pregalactic stellar population (e.g., Loewenstein 2001). The relative dearth of supernovae in clusters out to the highest redshifts for which the rate is presently constrained highlights the cluster abundance paradox, and begs the question of whether one or more of the standard assumptions (e.g., an IMF that is universal and invariant over time) break down in the universe in general. This paper aims to evaluate the plausibility and consequences of possible explanations for ICM Fe enrichment, taking into account constraints based on the characteristics of stellar populations in cluster galaxies as well as the observed mean evolution of the cluster SNIa rate and Fe abundance; and, to examine the resulting implications for our understanding of the star formation history (SFH) in the universe and its dependence on environment. New data on the histories of star formation, supernovae rates, and elemental abundances motivate an approach that goes beyond consideration of the baryon and metal inventories to explicitly model their self-consistent evolution. The framework adopted here for constructing such models, as presented in $2, is kept simple to restrict the number of possible parameters while allowing for consideration of an extensive range in each parameter, and to maintain the I -4- transparency of the effects and implications of various assumptions and scenarios. Bound- ary conditions, source functions, and parameters are initially chosen to be appropriate for the universe as a whole; but; clear paths for adaptations that might be required for rich galaxy clusters are provided. Models for the chemical evolution of Fe in the ICM are pre- sented in $3, and the requirements on the nature of star formation and supernovae in cluster galaxies imposed by observations are indicated and summarized. 54 includes discussion of the distinction between clusters and the field with respect to their intergalactic media and galaxy populations, and more detailed analysis of galactic outflows and Fe enrichment in clusters. Implications for the nature of Type SNIa and its environmental dependence are also presented in this section. Comparisons with results of similar investigations, and pos- sible directions for future model enhancements also appear in $4. Conclusions, with an emphasis on the dichotomy between clusters and the field, are summarized in $5. I adopt a topologically flat cosmology with Hubble constant H, = 70 km s-l, and total and baryonic matter densities, relative to critical, 0, = 0.3 and 0 b = 0.045, respectively (e.g., Fukugita & Peebles 2004). 2. Cosmic Chemical Evolution 2.1. Basic Equations I consider the coupIed evolution of the globally averaged densities and metal abundances of the three following baryonic categories: (1) stars that includes main sequence and evolved stars, substellar objects, and stellar remnants; (2) interstellar gas (ISM) defined as the fraction of total matter in the gas phase where stars may form at any time in the history of the universe (regardless of location); and, (3) intergalactic gas (IGM) defined as the remaining, inert (e.g., non-star-forming) fraction of gas-phase matter2. Stars and ISM are coupled via star formation and mass return, the ISM and IGM via galactic winds (and, potentially, galactic infall - not explicitly considered here). Type Ia and core collapse supernovae directly enrich the ISM. The evolution equations (Tinsley 1980; Matteucci & Gibson 1995; Thomas, Greggio, & Bender 1998) in rest-frame time for the comoving densities of the three cosmic constituents - pstarsp, IsM, PIGM - and for the mass fractions of their “ith” element - fistars3, 2Under these definitions, some (small) fraction of the “IGM” may actually consist of hot galactic halo gas, and any gas originally in circumgalactic or intergalactic space that subsequently accretes onto galaxies and forms stars is considered “ISh4.” 3Note that fistars includes a contribution from metals locked up in remnants that may be significant (Fukugita & Peebles 2004). - 5 - fiISM,f iIGM - are as follows: dPstars PSF PMR, = - dt @ISM -PSF + PMR PGW, = - dt and where the source terms &F, PMR, and ~ G Ware the comoving global mass density rates of star formation, mass return, and galactic wind mass transfer from the ISM to the IGM, PisNI PiSNII respectively; and, and are the density injection rates into the ISM of element “i” from Type Ia and core collapse (mostly Type 11) supernovae, respectively. Since the + universe is a “closed box,” there are five independent variables; pstars ISM PIGM = GbPcrit, where pcrit 3H:/Si7G is the present-day critical density. Also, the total average baryonic 2.2. Functions and Parameters The goal of this paper is to explore the chemical evolutionary implications of the in- creasingly complete and detailed inventory of matter, and inferred star formation and metal production rates, in the universe; and, to evaluate their applicability to the environment of rich galaxy clusters. As such, empirical quantities and constraints are directly applied to the fullest possible extent, and pegged as points of departure in modeling ICM enrichment. 2.2.1. The Stellar Initial Mass Function The initial mass function of forming stars (IMF) 4(m) dN/dm must be specified to determine the normalization of the observed star formation rate and total stellar mass, the I - 6 - stellar mass fraction recycled to the ISM via mass loss from evolved stars, and the number of supernova explosions per unit mass of star formation. Following Kroupa (2001), I adopt a four-part monotonically decreasing piecewise power-law IMF, extending from rnl = 0.01 Ma in the substellar regime to mu = 150 Ma (Weidner & Kroupa 2004; Figer 2005)) with slope (dlogN/dlogrn) a. = 0.3 below the hydrogen-burning mass threshold mH = 0.08 Ma,sl ope al = 1.3 between mH and mB = 0.5 Ma,a nd slopes a2 between mB and mD and a3 above mD. a2 = a3 = 2.3 is adopted as a standard; alternatives are assumed to have a single-slope above mB (a2 = a3), or a break at mo 2 2 m w~h ere the slope changes from a2 = 2.3 to cy3. 2.2.2. Mass Return The mass density rate of material recycled from stars to the ISM is Lomu dm$(m )A m(m )nS(~t - tms( m)). (8) PMR(~) = The turn-off mass mt, is implicitly given by tms(mto=) At where tms(mi)s the main sequence lifetime of a star of mass m (Schaller et al. 1992) and At is time since the onset of star formation. Am(m)i s the mass returned by stars of main sequence mass m, and is assumed mu independent of time. The star formation rate density nSF ~sF/J~,d m$(m)m. Am(m)i s derived using appropriate remnant masses, mremi,n the white dwarf, neutron star, and black hole regimes (Prantzos, Casse, & Vangioni-Flam 1993; van den Hoek & Groenewegen 1997; Fryer & Kalogera 2001; Woosley, Heger, & Weaver 2002) as follows: Am(m)= m - mrem(m); (9) + mrem= 0.45 0.119m, m 5 8; (10) mrem= 1.4, 8 5 m 5 21; (11) + mrem= 1.4 0.717(m - 21), 21 5 m 5 33; (12) mrem = 10, m 2 33, (13) where m is in Ma. The integrated fraction of mass formed into stars that is returned to the ISM up to the present epoch (tnO=w 13.47 Gyr) is Lo 1m ’dm4(m)Am(m) (Low R= s,, dm4 ( m 1 1 mu Til - 7 - where mto(tnow) 0.9 Ma is the present epoch main sequence turn-off mass. Under the N instantaneous recycling approximation (IRA: tms(m) 0 for all m),p ,,(t) = R&F. Since I consider short star-formation timescales and top-heavy IMFs, IRA is found not to be accurate for all models at all times and the exact treatment (equation 8) is adopted. 2.2.3. Star Formation Estimates of PSF, corrected for dust, are compiled and compared in Hopkins (2004)) and fitted to analytic functions in Bell (2005) and Strolger et al. (2004). The star formation rate is renormalized to assure that where pstars(tnowis) the cosmic density in stars observed today, and Pstars(tform) the den- sity of zero-metallicity (Population 111) “seed” stars predating the time, tformw, hen galaxy formation begins and the integration of the set of equations (1)-(6) is initiated. 2.2.4. Core Collapse Supernovae The mass density injection rate of element ‘[i” into the ISM by core collapse supernova of massive stars (“SNII”i)s where ml < m < m2 is the range of SNII progenitor masses and ysN~~z(mis )t he nucleosyn- thetic yield of element “i” from a progenitor of mass m. Instantaneous enrichment is an adequate approximation for these short-lived stars, reducing equation (16) to &II(t) Ks,IIPsF(~)(YsNII’), (17) where ITSNII E ~~im2dm$(m)/~mlmudims t~h(em nu)mmb er of SNII per unit mass of star formation, and (ysNIIi) --= ~m,”2dm$(m)y~~~~i(m)/~mthe, mmeUand SmN~II( ymie)ld of the “ith ” element. 2.2.5. Type la Supernovae In light of the multiplicity of theoretical predictions for the SNIa rate (Barris & Tonry 2005. arid references therein) I adopt the semi-empirical formalism of Strolger et al. (2004) - 8 - and Dahlen et al. (2004), determining the mass density injection rate of element “i” into the ISM by Type Ia supernovae (“SNIa”) from , bkNIa(t) = nSNIa(t)ySNIa a (18) and t nsNIa(t=) u s dt’&(t’)@(t - t’), (19) tforrn where u is the number of SNIa progenitor systems per unit mass of star formation, ~ S N I ~ Z is the SNIa nucleosynthetic yield of element “i” (assumed to be constant), and @(id) is the normalized delay time distribution function parameterized using the Gaussian distribution found by Strolger et al. (2004) and Dahlen et al. (2004) to explain the observed evolution of the SNIa rate, @(td) = (2na2 ) -1 e -[(td-tc)2/2U2] (20) 2.2.6. Star-Formation-Induced Galactic Wind I assume that supernova explosions drive outflow of material from the ISM to the IGM and that the mass loss rate per unit volume is proportional to the total supernova rate and the ISM density, + PGW (i) = KGW(’%NIa(t) KSNIIPSF(t))PISM. (21) Metal-rich gas is not preferentially ejected, and SNIa and SNII energy are assumed to contribute to driving outflows with equal efficiency. Given an observationally determined present-day ISM density, pIsM (tnow)s,e tting the galactic wind strength, KGlv is equivalent to assuming a value for the initial ISM density, PIsM(tform). I calculate this relationship by integrating equation (2) backwards from tnowt o tform. In the absence of infall, ~ ~ ~ ~ ( t f o r ~ ) must be sufficient to account for all the star formation since tformp,s tars(tnow-) pstars(tform), and is bounded above by the total baryon density. That is, allowed values of KGW are those that yield PISM,min < PISM(tforrn) < PISM,maxr (22) where PISM,rnin = PISM( tnow)+(Pstars(tnow)-Pstars(tform)) corresponds to no wind, and PISM,max = ClbPcrit - Pstars(tform) to a wind with maximum integrated mass outflow (“maximum wind”). 2.2.7. Boundary Conditions Equations (l),( 2), (4), (5), and (6) are integrated from tform to tn,,, with metal-free initial conditions: fZstars = fZISM = fiIGM = 0. A small seed stellar density is assumed, - 9 - pstars(tforrn) = L?IIIpcrit; pIsM(tform)i s a free parameter within the limits of equation (22) 2.2.8. Standard and Varying Parameters and Assumptions Standard boundary conditions include the formation epoch of the first Population I1 stars (tform= 4.66 lo8 yr, corresponding to redshift 10 for the adopted cosmology), the (Population 111) stellar density OIIIpcrit = 3 10-6pcrit (Ostriker & Gnedin 1996), and the stel- lar (ps+,ars(tnow) = 0.00267pCrit)a nd ISM ( ~ 1(tsrio~,) = 0.00078pcrit) densities at the present epoch (Fukugita & Peebles 2004). The IMF parameters are the slope at high mass, as, and transition mass, mD, to this slope from a2 = 2.3. The choice of IMF, in turn, determines the mass return fraction R (equation 14) and re-normalization factor for the star formation rate (equation 15), and the number of SNII per unit mass of star formation, KSNII.T he standard model has a3 = 2.3, R = 0.393 (the IRA value is 0.405)) and K~NI=I 0.0104 M,' for ml = 8 Ma and m2 = mu = 150 Ma.T he renormalization factors for the Bell (2005) and Strolger et al. (2004) star formation rate parameterizations are 0.94 and 0.62, respectively. I adopt the Bell (2005) SFH parametrization as standard. Since the Strolger et al. (2004) rate differs most dramatically in shape at low redshift where a relatively small fraction of the integrated star formation occurs, our results are insensitive to this choice (the renormalized functions are compared in Figure 2; see below). Adoption of the standard KS,.JIaIn d &(t) provides consistency with observed SNII rate evolution to redshift 1 (Strolger et al. 2004, Dahlen et al. 2004; see, also, Figure 3 below). There is sufficient theoretical uncertainty in the SNII synthesis of Fe that the IMF-averaged SNII Fe yield, (ysNIIFe) is simply left as a parameter with standard value 0.07 Ma. Standard SNIa parameters include mean delay time t, = 3 Gyr, dispersion = 0.2tC, and normalization v = 1.5 SNIa Ma-'( see, also, Gal-Yam & Maoz 2004, Greggio 2005) that provide the best-fit to the observed SNIa rate evolution for the Bell (2005) star formation rate (these slightly differ from their values in Dahlen et al. 2004 due to the different adopted star formation rate). I focus on Fe in this paper (thus justifying the neglect of non-explosive production in equation 5); a SNIa Fe yield ySNIaFe = 0.7 Ma is adopted as standard. For the above standard parameters and functions, one may vary the galactic wind factor from KGW = 0 to K G W w~h~ere~ St,bP critKcWmax= 3.05 lo3 Ma is such that all of the IGM originates in the ISM. -10- 2.2.9. A Note on Varying the IMF The standard model IMF (with slope 2.3 above 0.5 MD), if assumed universal, provides mutual consistency among observations of the star formation and SNII rates and build-up of stellar mass (Bell 2005; Drory et al. 2005; Gwyn & Hartwick 2005), and with the luminosity density of the universe (Baldry & Glazebrook 2003). If the slope is significantly flatter, (1) the higher stellar mass loss return fraction (Figure 1) implies that a higher star formation rate than is observed would be required to produced the observed amount of stars4 (Figure 2); and, (2) the implied SNII rate would be greater than observed (Figure 3a). Conversely, a steeper high mass IMF slope (i.e., a2 = 2.8; see Kroupa & Weidner 2003) significantly underpredicts the observed SNII rate (Weidner & Kroupa 2005) unless the transition at m D is pushed to high mass (Figure 3b). 2.3. Enrichment in the Standard Model The chemical evolution of Fe in the standard model (see entries “1N2.3” in Tables 1 and 2) provides predictions to be compared with abundance measurements in stars and interstellar gas in field galaxies, and in the intergalactic medium. This serves to indicate possible shortcomings in the model and provide a baseline for evaluating what manner of extension or variation might be required to explain Fe abundances in rich galaxy clusters. Integrated over a Hubble time, the total stellar Fe yield (relative to solar) in the standard model is 1.3 (58/42% from SNIa/SNII) corresponding to an enrichment of 0.13 averaged over all baryons (Table 2). The model-predicted z = 0 SNII and SNIa rates (Table 2) are consistent with observations (Cappellaro, Evans, & Turatto 1999). The distribution of Fe among stellar, ISM, and IGM components as a function of time depends on the strength and time-dependence of galactic outflow. The more numerous SNII dominate the galactic wind term (equation 21), but SNIa contribute the majority of Fe enrichment - and do so with a significant lag with respect to the accumulation of stellar mass and the ejection of SNII-enriched material into the IGM. This results in a relatively recent (z < 1) build-up of Fe in the stars (Figure 4a) and ISM (Figure 4b) and a substantial Fe mass fraction contained in these components at the present epoch. As a result, little more than half of the total Fe production is injected into the IGM for a i[maximum” wind (Figure 5). A reapportionment that boosts the IGM Fe fraction requires extending the 4Neglecting any effects a differelit IMF would have on the observational estimates of the stellar inass 5Pu’ote that this is the yield per >tar fornied and is not renormalized to the present stellar inass.