1. The chemical enrichment of clusters of galaxies Jelle S. Kaastra Collaborators: Norbert Werner, Jelle de Plaa, Aurora Simionescu, Yan Grange

2. 2 Outline 1.Importance clusters 2.Observational challenges 3.Enrichment by supernovae 4.Enrichment by winds 5.Conclusions

3. 3 1. Importance clusters for abundance studies Largest bound structures Deep potential wells, retains most of the gas Hot gas: no significant “hiding” of metals in dust Spatial extent allows mapping

4. 4 2. Observational challenges 1.Fe bias 2.Non-thermal components 3.Complex temperature structure

5. 5 The Fe bias 1T models sometimes too simple: e.g. in cool cores Using 1T gives biased abundances (“Fe- bias, Buote 2000) Example: core M87 (Molendi & Gastaldello 2001) Multi-T1T

6. 6 Non-thermal components Example: Sérsic 159- 03 Strong soft excess, modeled by non- thermal component Implications on abundances (De Plaa et al. 2006)

7. 7 Complex temperature structure I (de Plaa et al. 2006) Sérsic 159-3, central 4 arcmin Better fits 1T  wdem  gdem Implication for Fe: 0.36  0.35  0.24 Implication for O: 0.36  0.30  0.19

8. 8 Inverse iron bias: how does it work? Simulation: 2 comp, T=2 & T=4 keV, equal emission measure Best fit 1-T gives T=2.68 keV Fitted Fe abundance 11 % too high Due to different emissivity for Fe-L, Fe-K

9. 9 Temperature maps (Hydra A, Simionescu et al. 2008)

10. 10 Complex temperature structure II (Simionescu et al. 2008) Example: Hydra A Central 3 arcmin: Full spectrum: Gaussian in log T (σ=0.2) 1T fits individual regions: also Gaussian Confirmed by DEM analysis (blue & purple)

11. 11 Implications for Fe abundance (Simionescu et al. 2008) Central 3 arcmin Hydra A, 1T models: (errors on Fe 0.01 to 0.02) Band (keV)kT (keV)Fe Full (0.35-10)3.40.50 Low (Fe-L) 0.35-22.80.37 High (Fe-K) 2-73.90.41 Gdem3.4, σ=0.20.45

12. 12 Fitting bias: continua ok? (Werner et al. 2006) Some features subtle Example: 2A0335+096, 130 ks XMM-Newton To determine Cr abundance (0.5±0.2 solar) needs carefull analysis local continuum

13. 13 Fitting bias: calibration uncertainties (de Plaa et al. 2007) Some lines (Si) weaker than calibration uncertainty instruments Important to estimate systematic uncertainties (no “blind” χ 2 fitting) Diffference pn data & best-fit MOS model Sérsic 159-03

14. 14 Quantifying systematics (de Plaa et al. 2007) Correction: fudge MOS area to match pn and vice versa (spline) Remaining difference systematic:  Mg: too uncertain  Si: 11 %  Ni: 19 % Example: Sérsic 159-03

15. 15 3. Enrichment by supernovae

16. 16 Type Ia, Type II and Solar abundances O O

17. 17 Supernova yields: core collapse

18. 18 Supernova yields: Ia

19. 19 Decomposing abundances into SN types (De Plaa et al. 2006) Deep exposure XMM- Newton Sérsic 159-3 Data include RGS ~50 % SN Ia by number Ca problem

20. 20 Another case: 2A 0335+096 (Werner et al. 2006) Use here WDD model Central 3 arcmin: Sn Ia: 25 % Increases to 37 % in 3-9 arcmin annulus Ni: W7 model predicts more Also here Ca problem

21. 21 Analysis of a large sample (De Plaa et al. 2007) 22 clusters, 685 ks net exposure Taken from HIFLUGCS sample (Reiprich & Böhringer 2002) All spectra extracted from within 0.2 R 500 Use wdem model

22. 22 Solution to the Ca problem (De Plaa et al. 2007) Also sample shows Ca excess Problem solved by adopting SN Ia yields based on Tycho SNR (Badenes et al. 2006) Best fit Ia/(Ia+cc) number ratio: 0.44±0.05 WDD Tycho

23. 23 Sample: mean abundance ratios (De Plaa et al. 2007) Ratio Weighted mean (Lodders scale) σ int Si/Fe0.66 ± 0.130.17 ± 0.05 S/Fe0.60 ± 0.060.18 ± 0.06 Ar/Fe0.40 ± 0.030.11 ± 0.05 Ca/Fe1.03 ± 0.040.12 ± 0.08 Ni/Fe1.41 ± 0.310.2 ± 0.2

24. 24 A 2052 (Grange et al. 2008) 90 ks exposure (2001, 2007) Analysis in progress Abundances O to Fe all consistent with sample De Plaa et al. (2007) Only Ca more overabundant: Ca/Fe = 1.51±0.10 (compared to 1.03, σ=0.12) Needs further confirmation

25. 25 Comparison between clusters (Simionescu et al. 2008) 6 clusters with deep exposures, taken from literature Most have 30-40 % contribution Ia Hard to discriminate between Ia models, but see extremes Hydra A / M87

26. 26 Radial profiles: example 2A 0335+096 (Werner et al. 2006) Si FeAr S

27. 27 Comparison between clusters: radial profiles (Simionescu et al. 2008) All elements have decreasing abundances Also valid for O (contrary to earlier suggestions of flat O profile, Tamura et al. 2004)

28. 28 Abundance ratios constant? (Simionescu et al. 2008) Si/Fe flat within 0.1 R 200, maybe break at 0.05R 200 O/Fe increases, but only slightly: per dex in radius, O/Fe increases by 0.25±0.09 (Fe decreases by 0.72) O/Fe Si/Fe

29. 29 Consequences of “flat” oxygen profiles (Simionescu et al. 2008) Flattish O profiles:  not only Ia contribute to core enrichment Ram pressure stripping works already at Mpc scale (compare to 130 kpc core Hydra A) Continued cc SN activity over past 10 10 year? Early central enrichment cc SN?

30. 30 4. Enrichment by winds

31. 31 XMM-Newton RGS results RGS optimal for point sources But still the best for moderately extended sources: Δλ (Å) = 0.138 Δθ (arcmin)

32. 32 RGS results: M 87 (Werner et al. 2006) Exposure time: 169 ks Lines from O, N, & C C/Fe: 0.74±0.13 N/Fe: 1.62±0.21 O/Fe: 0.59±0.04 Ne/Fe: 1.25±0.12 Mg/Fe: 0.60±0.06 Fe: 1.06±0.03  AGB winds for CN! Continuum-subtracted RGS spectrum

33. 33 Nitrogen with RGS: other cases M87: N/Fe = 1.62±0.21 2A 0335+096 (Werner et al. 2006): 1.3±0.4 Sérsic 159-3 (De Plaa et al. 2006): 0.0±0.5 Centaurus (Sanders et al. 2008): 1.5-3 Need for more deep exposures with RGS

34. 34 Other case: Centaurus (Sanders et al. 2008) N/Fe=1.5-3

35. 35 Future: SXC (and of course XEUS etc)

36. 36 5. Conclusions XMM-Newton observations of clusters of galaxies can disentangle contributions different SN types and winds Need take care of systematics, in particular temperature distribution for reliable results Best done using deep exposures