NASA Goddard Space Flight Center

X-ray Spectra of Clusters and Groups

Richard Mushotzky

NASA/Goddard Space Flight Center, Greenbelt MD 20771 USA

Abstract:

Detailed x-ray imaging spectroscopy has come of age with the launch of the Rosat and ASCA satellites. The high quality spectra allow determination of the abundance of the alpha-burning elements and abundance gradients for low redshift clusters. For high redshift clusters the average metallicity can be well determined out to redshifts > 0.5. The early results show that rich clusters show very similar abundances and have an abundance distribution which is very well matched to that produced by type II supernova. This has important consequences for the formation and evolution of galaxies and clusters. The situation for poor clusters and groups is more complex with a more solar-like element abundance distribution but with variation from object to object. There is little evidence for variation in cluster Fe abundance with redshift out to z~0.5. There is also little variation in the log L(x) vs. log kT relation, Mass profiles from ASCA and Rosat data are robust and in good agreement. Out to radii of ~1 Mpc most isolated, round smooth clusters are isothermal (ignoring the central cooling flow). At larger radii the ASCA data often show that the temperature is dropping at large radii.

Rich clusters of galaxies and groups represent the largest virialized systems in the universe and as such are quite sensitive to the formation and evolution of structure (Bohringer 1995). The x-ray emitting hot gas offers one of the best probes of the form of the gravitational potential, the amount and distribution of the dark matter, the evolution of these systems, the origin and evolution. of the heavy element abundances. As such, x-ray studies of clusters have received much attention, both observationally and theoretically in the last few years. In this meeting there are several other review talks on clusters, in particular H. Bohringer on Rosat imaging results on clusters P. Henry on the structure of merging clusters, T. Ohashi on ASCA spectra of poor clusters and groups and H. Ebeling on Rosat all sky survey results. In this paper I will emphasize what we can learn from imaging x-ray spectra of relaxed systems. However due to the enormous amount of material this review will of necessity have to be sketchy and brief. There were more than 150 papers on clusters in the literature published in the last 2 years.

B. Spectral Observables

A good summary of the observational and theoretical situation before the launch of Rosat and ASCA is in Sarazin 1989. To zeroth order the x-ray emission from clusters is due to hot, metal enriched gas "trapped" in the potential well of the cluster. As such the observables are 1) the average gas temperature and the projected temperature vs. position, 2) the strengths, widths, and energies of the emission lines from the abundant elements, 3) the amount of photoelectric absorption and its variation with position. Given these data one can derive physically interesting properties such as 1) the total mass and mass distribution of the dark matter and baryons, 2) the absolute abundances, abundance ratios and spatial distribution of most of the abundant elements, which enables strong conjectures about the origin of the gas and the evolution of galaxies to be made 3) the physics of cooling flows and the fate of the cooled gas 4) the evolution of the cluster mass function with redshift 5) the evolution of abundance with redshift and 6) estimates of the Hubble constant from the Sunyaev-Zeldovich effect

In this talk I will stress the recent determinations of the abundances of the alpha-burning elements (O, Ne, Mg, Si, S, Ar, Ca) for clusters and groups, the evolution of temperature and abundance and the mass distributions of rich clusters.

2. Enrichment and Energetics of the Intracluster Medium and Elliptical Galaxies

A. Introduction

It has been almost 20 years (Serlemitsos et al 1976, Mitchell et al 1975) since the discovery of Fe in intracluster gas in clusters of galaxies. The existence of large amounts of iron in the gas is a strong constraint (Renzini et al. 1993, Arnaud et al. 1992) on the history of star formation in clusters and the evolution of galaxies and clusters. The total mass of iron, combined with its correlation with the total light from elliptical galaxies in the clusters (Arnaud et al 1992) have given rise to models in which the bulk of the ICM Fe comes from SN driven winds in elliptical galaxies. However the observed mass of Fe (MFe) is too large by factors of 3-10 for the observed present SNIa rate under the assumption that the stars formed with either a Salpeter initial mass function (IMF) which has a slope, X=1.35, or Miller-Scalo IMF (which has an effect value of X=1.7 over the high mass range important for producing the a-elements).

The possible solutions to this problem have included

1) a much higher SNIa rate in the past. In this model most of the Fe is produced by SNIa which do not produce much alpha-burning products.

2) A massive burst of star formation in the distant past in which massive stars produce most of the Fe. Since type II SN-produce a-burning elements as well as Fe one expects a non-solar metallicity ratio. The initial burst of star formation can have either a flat IMF (X< 1.3) or be "bi-modal" such that the less massive stars that we see today do not come from the same IMF as the massive stars that produced the metals (Elbaz et al 1995). It should be noted that for a fixed amount of Fe, type IIs produce a lot more mechanical energy than type Is

Neither of these scenarios can be constrained solely by measurements of the cluster Fe abundance at the present epoch.

The recent ASCA spectra of rich clusters (Mushotzky et al 1995 and Loewenstein and Mushotzky 1995) provides the first detailed measurements of the alpha-burning products (O, Ne, Mg, Si, S, Ar and Ca) as well as Fe in regions outside of the cooling flow. This data, combined with the ASCA measurements of the evolution of the iron abundance with redshift allows a detailed test of the proposed models.. The results for the alpha-burning elements are based on the analysis of the ASCA data for 4 rich clusters A496, A2199, AWM7, A1060. Similar results for MKW3S, A1795 and the Perseus cluster. have been obtained by Hatsukade et al.(1995) and Arnaud et al. 1995.

B. Basic Model

M. Loewenstein (Loewenstein and Mushotzky 1995) has used the newest nucleosynthesis results (Weaver and Woosley 1995, Hashimoto et al. 1995) to calculate the yields of the alpha-burning elements from type II SN with a wide variety of IMF slopes X. For each element i one can calculate a Mass/Light ratio Mi/LB

where Y= rate at which stars leave main sequence and Yi(M)= yield of i^th element from SN II with progenitor mass M. Mup is the upper mass cutoff for massive stars and MTO is the main sequence turn off at the present time. We can then compare the predicted yields as we vary X - the only free parameter in the model. As shown in detail in Loewenstein and Mushotzky (Table 1, figure 1) the ASCA data are most compatible with X< 1.1 for all the clusters indicating a IMF strongly biased to high mass stars compared to the Salpeter or Miller-Scalo IMFs.

These results are robust since the number of Si and O atoms translate directly to a number of type II SN. If we require the same IMF to produce the present day stellar light and the cluster metals then galaxies must have gone through a giant phase of star formation.

Figure 1: The observed abundance distribution for rich clusters (black dots) compared to theoretical abundances for a pure type II enrichment, normalized to Oxygen (Woosley and Weaver 1995) (open box), pure type I enrichment normalized to Fe = solar (Nomoto et al 1984) ( triangles) and pure type I normalized to the cluster Fe abundance (crosses).

C. Implications

These results have very strong implications for the formation and structure of galaxies and clusters

1) The abundances of O, Si, Fe are consistent with Type II origin, however some contribution due to type Is is not ruled out.

2) the mass ejected from galaxies M_eject ~ the present mass in stars in the galaxies

3) The implied luminosity of the stars responsible for the production of the alpha-elements is enormous since L_bolometric (SNII progenitors)~10⁴⁸(t/10⁸yrs)erg/sec/gal and thus these galaxies are roughly 10⁴ times more luminous than at present (depending on the evolutionary timescale of the starburst.) These galaxies are not seen in redshift or color surveys---thus either they exist at very large redshifts or are dust enshrouded.

4) The kinetic energy due to the SNII is ~ 1/4-1/2 of the binding energy of the cluster and larger than the binding energy of the elliptical galaxies ---This phase of metal formation must have had a tremendous influence on the formation and structure of clusters and galaxies. To date numerical simulations of cluster and galaxy formation and evolution have not included this effect and thus cannot be fully representative of the true physical origin of clusters. Furthermore, it has been noted in several theoretical simulations that such an effect is needed to prevent an over concentration of baryons in galaxies (Summers 1993, Navvarro 1995).

E. Abundance variation with redshift

There are now available on the ASCA public data base > 13 ASCA spectra of clusters with z>0.14 (many more will be available within the next 2 years). The relative errors in the bulk properties (kT, abundance) are small and are similar to those derived from proportional counters for the z~0.03 systems. Only for the very largest/brightest systems (e.g. Abell 2163 Markevitch et al 1995) is spatially resolved spectroscopy with ASCA possible. The median value of the Fe abundance for these high z systems is <Fe>=0.32+/-0.09 compared to the low redshift median of <Fe>=0.34+/-0.02. Thus there has been little or no evolution in the average Fe abundance from z~0.3 (1/4-1/3 of the age of universe) till now. (Figure 2). Similar results have been obtained by Tsuru et al (1995). So far none of the observed systems which are cool enough that their Si and O abundances can be determined have had published values.

Figure 2: The Fe abundance versus redshift for clusters observed in the ASCA PV and early AO-1 phases. The ellipse represents the range of values seen for low redshift clusters ignoring the central regions where a few show abundance gradients. Note the lack of any systematic change with redshift.

Figure 3: Comparison of bolometric X-ray luminosity and temperature for z<0.1 (low z clusters) and z>0.15 (high z clusters). The data are obtained primarily from ASCA PV observations and from Tanaka and Mushotzky (1995). q0=0 H0=50 are used.

Figure 4a: The abundances of Fe, Si and S versus distance from the center in the WP23 group. Notice the lack of an abundance gradient and the uniform ratio of Fe:Si:S. Data are from Fukazawa et al 1995.

Since most hierarchical clustering theories (Kauffmann et al. 1994) indicate that clusters "grow" primarily by the absorption of smaller systems (e.g. groups) and their mass increases by a factor of 2 from z~0.2 till the present, the lack of evolution in <Fe> indicates that almost all systems must have about the same Fe abundance. However groups seem to show a wide range in abundance at the present epoch (Fukazawa et al 1995). The lack of evolution in Fe abundance strongly supports ideas that the bulk of the Fe originates at "early" times (type II SN) rather than growing with age as a type I SN origin would indicate. Since the data go back to an effective look-back time of ~1/3 the age of the universe the constraints on the role of type I's in the enrichment of the cluster gas is rather severe.

F. Evolution of Cluster potential

Another area of interest is the relationship between the x-ray luminosity and temperature (e.g. Henry and Arnaud 1991, Kaiser 1991). It is expected, from hierarchical clustering theory, that for a given x-ray temperature, clusters should be less luminous at higher redshifts. However (figure 3) shows that there is no measurable change in the log L(x) vs. log T diagram with redshift. The objects in this figure are not an unbiased sample of the high redshift systems, since before the release of the Rosat all sky survey data (Ebeling 1995), the observations of high z systems was rather biased to the most luminous and/or famous objects. Inclusion of more objects from ASCA AO-1 observations (Tsuru et al 1995) shows a weak trend for the higher z systems to be hotter at a fixed bolometric luminosity.

G. Groups and Low mass clusters.

T. Ohashi has reviewed some of the relevant ASCA data at this meeting. Rather few groups have had well determined abundances. However there are 7 well studied systems (WP23 and HCG 51 (Fukazawa et al 1995), HCG 62 (Ohashi et al 1995), NGC 2300 (Davis et al 1995), NGC 4261 (Davis et al 1995), Fornax (Ikebe et l 1995), Virgo (Matsumoto et al 1995). Their observed Fe abundance ranges from a low of~0.11 (NGC 2300) to a high of ~0.6 (central regions of Virgo) as opposed to the very small range seen in rich clusters (0.25-0.4 solar). What is rather striking is that the observed Si:S:Fe ratios tend to be solar rather than the cluster value. However, the situation is rather complex since some systems show a strong abundance gradient (e.g., Centaurus (Fukazawa et al 1994), M87 (Matsumoto et al 1995)) while others show no gradient at all (WP 23 Fukazawa et al 1995).

Figure 4b: the Si:Fe ratio plotted versus radius for WP23. Notice that the value is a factor of 2 less than that seen in rich clusters or in M87/Virgo.

In addition in the Centaurus cluster the Fe:S:Si ratio varies from solar near the center to a "rich cluster" value in the outer regions, while in M87 this ratio remains constant as a function of radius.

Figure 5. Abundance versus distance from center for the Centaurus cluster. Notice the strong abundance gradient in all the elements and the change in the ratio of Si:S:Fe. In the center the ratio is close to solar while in the outer regions the ratio is consistent with that seen in rich clusters.

It seems clear, that unlike the situation for rich clusters, groups and poor clusters form a varied family.

Figure 6: Distribution of the Fe:Si and S:Si ratios versus distance from the center in M87. Notice the relative constancy of the ratio compared to the situation in the Centarus cluster. Data are from Matsumoto et al 1995.

The fraction of the mass that is in gas also shows a wide variation (Mulchaey et al 1995, Pildis et al 1995) and the baryonic fraction also varies. As yet there is insufficient data to examine patterns such as the relationship of Fe mass to galaxy mass, gas fraction versus abundance etc. We hope that such data will soon be available.

H. Elliptical Galaxies

The ASCA and Rosat data for elliptical galaxies (Awaki et al 1994, Mushotzky et al 1994, Loewenstein et al 1994) show a wide variation in Fe abundance, similar to the situation in groups. In fact, some of the "isolated" ellipticals like NGC 4636, have many properties, such as total mass, gas mass and temperature which are virtually identical to that of groups. While there is some variation in the Si: Fe ratio, most objects tend to have solar ratios. There is also evidence, in the very few objects for which ASCA can obtain spatially resolved spectra, of abundance gradients (e.g. NGC 4636 Mushotzky et al 1994). The ASCA data strongly prefer models in which the abundance is sub-solar, while the Rosat data are not capable of truly determining the abundance in the presence of spectral complexity. While the results are not yet final (Loewenstein 1995) it seems as if there is a true factor of 2 discrepancy between the x-ray determined and optically determined Fe abundance. Given the robustness of the x-ray determined Si and S abundances, which agree with the less well determined (see next section) Fe L abundances this discrepancy will have to be taken seriously.

H. Robustness of Abundance determinations

Recent ASCA observations of cooling flow clusters (Fabian et al 1994) have clearly shown that the predicted strengths of certain of the Fe L lines (in particular Fe XXIV) at certain temperatures are in error in the Raymond-Smith and Mewe-Kaastra plasma codes (Leidahl et al 1995). Since the Fe abundance for groups and elliptical galaxies is determined primarily from measurements of the Fe L lines this has cast into doubt the accuracy of the ASCA abundance determinations for these systems. However it is possible to measure the Fe abundance in certain systems (those with 2<kT< 4keV) from both the Fe L and Fe K lines. As shown in figure 7 the abundance of Fe determined for these systems from the L and K lines are in excellent agreement. In addition with ASCA it is possible to measure the ratio of the Si and S He-like and H-like lines, as well as determine the temperature of the continuum directly. This allows an independent determination of the temperature and abundance of these elements.

The derived values (e.g. Fukazawa et al 1995) agree quite well with that from an analysis with the Raymond-Smith plasma code, but unfortunately disagree with that derived from the "old" Mewe-Kaastra plasma code values.

I. Summary

1) Rich Clusters seem to form a coherent pattern: they apparently have weak or absent abundance gradients; a fairly uniform abundance of the various elements from cluster to cluster; have an abundance distribution whose origin is consistent with almost pure SN type II enrichment; have a narrow distribution in iron abundance, <Fe>~1/3 solar; and show little or no evolution in <Fe> with redshift. However poor clusters and groups show a wider variation: some have abundance gradients, with a Si/S/Fe ratio which varies by a factor of > 4 (indicating different mechanisms for enrichment); the range of Fe abundance seem to be more variable with NGC2300 at A~0.15 solar and the center of Centaurus at A>0.6 solar. However their evolution in abundance with redshift is not known at present. The presence of an abundance gradient may be related to the relative mass contained in the central galaxy since the gradients all seem to occur in the central 100 kpc. Because of the larger redshifts of rich clusters the abundance gradients that appear in the inner regions of some poor clusters will be smoothed out by the point response function of ASCA.

2) Elliptical galaxies show a range of abundances with the observed range varying over a factor > 3 from 0.2-0.8 solar. The Si/Fe ratio is ~1-2 solar, less than in rich clusters. They often show strong abundance gradients and more luminous galaxies often have higher abundances. The abundances derived from Fe L for elliptical galaxies seem to be robust and for those systems where one can measure abundances with both the K and the L lines they are in good agreement.

For rich clusters the abundance pattern is totally consistent with an origin of the metals at high redshifts. For groups and elliptical galaxies the range of variation indicates a "2nd" parameter which is not yet clear. The gas rich groups seem to have higher abundances.

3. Mass Determinations

For many years (Fabricant and Gorenstein 1983) it has been known that the spatially resolved spectra of clusters of galaxies offers one of the best ways to determine the mass of these systems. If the gas is in hydrostatic equilibrium and gas pressure is the predominant pressure component then the total mass can be determined from the observed surface brightness and temperature data. The surface brightness can be determined precisely with Rosat images and the temperature profiles can be measured to better than 10% accuracy with ASCA spectra. To give an idea of the achieved accuracy the error in the ASCA temperature for a rich cluster , like Abell 496, corresponds to an uncertainty of less than 50km/sec in the implied velocity dispersion of the galaxies.

Figure 7: Comparison of the Fe abundance derived from Fe K and Fe L lines in 4 rich clusters (Mushotzky et al 1995) and in various annuli centered on M87 (data from Hwang et al. 1995).

If one selects round smooth isolated clusters, the uncertainties due to mergers and non-equilibrium effects (Schindler 1995, Evrard, Metzler and Navarro 1995) are rather small and the x-ray mass estimates within the "virial radius" are robust and accurate.

Figure 8: Comparison of the derived mass of the cluster AWM 7 from Rosat and ASCA imaging x-ray spectra. The solid lines are the range of allowed masses from the Rosat data (Neumann and Bohringer 1995) while the dots and error bars are from the ASCA data (Loewenstein 1995).

I reviewed this topic last year (Mushotzky 1995) and the situation is more or less similar for the low redshift systems. However it is now clear that for systems of kT< 4 keV the ASCA data are very precise (cf. Tamura et al 1995 for Abell 1060) and that, when appropriate data exist the ASCA and Rosat temperatures are in excellent agreement (figure 8 for AWM 7). The results for ~6 clusters show that out to radii of ~ 1 H50^-1 Mpc the mass of gas is almost always ~20% of the total mass of the system and that the mass derived from the x-ray data are in fairly good agreement with that derived from the virial theorem for those clusters with measurements for more than 50 galaxy velocities and no strong spatial substructure.

Recently M. Markevitch and collaborators have analyzed the ASCA data for several moderate redshift (z~0.06-0.17) clusters. They have found that at radii >1 Mpc the temperature is decreasing with radius in all of these systems. In some of these objects (e.g. Abell 2163 and A665) the temperature profile is steeper than adiabatic. In others (A2256, A2319) the temperature drop is more gradual. Similar results were obtained for A2218 (Arnaud 1995) and A1689. These results combined with the older Coma data (Hughes et al 1988) indicate that all clusters for which spatially resolved spectra exist at large radii show negative radial temperature gradients. Mass estimates derived from the A2163 data (Markevitch et al 1995) severely constrain the dark matter potential. While the sample is somewhat biased to luminous hot clusters the trend is interesting. If this effect is due to cluster formation rather than heating mechanisms it favors low W universes (Evrard, Metzler and Navarro 1995).

4. Conclusions

In this paper we have not had the opportunity to present the ASCA results on cooling flows, the measurement of the form of the x-ray surface brightness at high energies or the Sunyaev-Zeldovich effect results. I hope that the mere smattering of x-ray spectral results presented here will give the reader at taste for the results soon to be published. While the Rosat data have reached a relatively mature stage the ASCA data are still quite young. However the future is extremely rich. With the launch of AXAF, XMM and Astro-E we will have increased angular resolution, increased collecting area and increased energy resolution. The results should prove exciting.

Acknowledgments: I would like to thank M. Loewenstein, K. Arnaud, and U. Hwang for communication of results prior to publication. I also acknowledge a debt to the ASCA team which has made much of this work possible.

References

Arnaud, K. et al. 1994 presented at the HEAD Napa Valley meeting
Arnaud, M, Rothenflug, R., Boulade,O. Vigroux, L. and Vangioni-Flan, E, 1992 Astron+Astrophys 254, 49
Awaki, H. et al. 1994 PASJ 46, L65
Bohringer, H. 1995 these proceedings
David, L., Forman,W., Jones, C., and Daines, S. 1994 Ap.J. 428,544
Davis, D, Mushotzky, R., Mulchaey, J., Worrall, D.,Birkinshaw, M. and Burstein, D. 1995 Ap.J. 444,582
Davis, D. Mulchaey, J., Mushotzky, R., and Burstein, D. 1995 Ap.J. in press
Ebeling, H. 1995 these proceedings
Elbaz, D., Arnaud, M., and Vangioni-Flam,E. 1995 Astron and Astroph in press
Evrard, A., Metzler, C. and Navarro, J. 1995 Ap. J submitted
Fabian, A. S., Arnaud, K. Bautz, M. and Tawara,Y. 1994 ApJ. L 436,L63
Fabricant and Gorenstein, P. 1983Ap J.. 267, 535 Fukazawa, Y. et al. 1994 PASJ 46, L55
Fukzawa et al 1995 PASJ submitted
Hashimoto, M., Nomoto, K., Tsujimoto, T., and Thielmann. 1994. Univ. of Tokyo preprint.
Hatsukade, I., Kawarabata, K., and Takenaka, K. 1995 paper presented at the 11th International Colloquium on UV and X-Ray Spectroscopy of Astrophysical and Laboratory Plasmas at Nagoya
Henry, J.P., and Arnaud, K. 1991 Ap J. 383, 95
Henry, J.P. 1995 these proceedings
Hwang, U.et al. 1995 in preparation
Hughes, J. Yamashita ,K. Okumura,Y. Tsunemi, H. and Matsuoka, M. 1988 Ap.J. 327, 615
Ikebe,Y. 1995 in prep
Kaiser, N. 1991 Ap. J. 383,104
Kauffmann, G.,White, S.D.M., and Guiderdona,B. 1993 MNRAS 264,201
Liedahl D., Osterheld, A., and Goldstein, W. 1995 Ap. J. 438, L115
Loewenstein, M. Mushotzky, R.F. Tamura, T. Ikebe, Y., Makishima, K. Matsushita, K. Awaki, H. and Serlemitsos, P.J. 1994 Ap. J. 436,L75
Loewenstein, M., and Mushotzky, R. F. 1995 Ap J in press
Loewenstein, M. 1995 talk presented at the Napa Valley HEAD meeting
Loewenstein, M., and Mushotzky, R. F. 1995 Ap J in press
Markevitch, M. these proceedings
Markevitch, M. et al Ap. J. in press
Matsumoto, H. et al. 1995 PASJ in press
Mitchell, R., Ives, J., and Culhane, L. 1975 MNRAS 175,29
Mulchaey, J. Mushotzky, R. Davis, D. and Burstein , D. 1995 Ap.J. in press
Mushotzky, R., et al. ApJ. 436,L79
Mushotzky, R., et al. 1995 ApJ in press
Mushotzky, R. 1995 in proceedings of the College Park Meeting on Dark Matter ed. S. Holt, page YY
Navarro, J. 1995 preprint
Neumann, D. and Bohringer,H. 1995 A+A 301,865
Nomoto, K., Thielman, F., and Yokoi, K. 1984 Ap. J. 286, 644
Ohashi, T. 1995 these proceedings
Pildis, R., Bregman, J. and Evrard,A . 1995 ApJ 443,514
Renzini, A., Ciotti, L., D'Ercole, A, Pellegrini, S. 1993 Ap.J. 419, 52
Sarazin, C. 1989 Rev Mod Phys. 58,1
Schindler, S. 1995 A=A submitted
Serlemitsos, P., Smith, B., Boldt, E, Holt, S.S., and Swank. J. 1976, Ap. J., 211, L63
Summers, F. 1993 PhD Thesis Univ of Calf Berkeley
Tamura, T. et al. 1995 PASJ submitted
Tanaka, Y., and Mushotzky R. 1995 in prep
Tsuru, T., Koyama, K, Hughes, J. Arimoto, N., Kii,T., and Hattori, M. et al. preprint
Woosley, S.E., and Weaver, T. 1995 ApJ. Suppl in press