1. ESO Large Program 165-L0263: Distances, Ages and Metal Abundances in Globular Cluster Dwarfs Raffaele Gratton Osservatorio Astronomico di Padova, INAF, ITALY

2. PI: R. Gratton co-authors: P. Bonifacio, A. Bragaglia, E. Carretta, V. Castellani, M. Centurion, A. Chieffi, R. Claudi, G. Clementini, F. D’Antona, S. Desidera, P. Francois, F. Grundhal, S. Lucatello, P. Molaro, L. Pasquini, C. Sneden, M. Spite, F. Spite, O. Straniero, M. Zoccali VLT2 (Kueyen)+UVES 12 nights in June and September 2000 12 nights in August and October 2001 6 nights in August 2002 ESO Large Program 165-L0263:

3. Aims Distances and Absolute Ages of Globular Clusters The O-Na anticorrelation among globular cluster TO-stars Lithium abundances in TO-stars and subgiants of globular clusters

4. Clusters selected for observations The closest globular clusters (but M4 for which differential reddening is important) cluster V(TO) [Fe/H] NGC6397 16.4 -1.82 NGC6752 17.2 -1.42 47 Tuc 17.6 -0.70

5. Stars selected for observations: TO stars and early subgiants (below the RGB clump)

6. Field star sample: 34 metal-poor stars with good parallaxes from the Hipparcos satellite Green points: single stars Red squares: binaries

7. ANALYSIS T eff ’s from spectra: Balmer line profiles Analysis procedure strictly identical for field and cluster stars  Reddening free

8. Our spectra have R~40,000, and S/N~80-100 for stars in NGC6397, S/N~20-60 for stars in NGC 6752 and 47 Tucanae.. The spectral range is 3500-9000 Å. We show the correlation between EWs measured with an authomatic procedure on spectra of two TO stars in NGC6752 (upper panel) and NGC6397 (lower panel) Typical errors are  3 mÅ for stars in NGC 6397, and  5 mÅ for stars in NGC 6752 and 47 Tucanae Accurate EWs can be derived from our spectra

9. T eff ’s from spectra: - Balmer line profiles Analysis procedure strictly identical for field and cluster stars  Reddening free

10. Comparison between T eff ’s from H  and from colours (calibration by Kurucz, model without overshooting) Zero point error  27 K r.m.s.=159 K Reddening zero point error:  E(B-V)=0.008 (yielding an error of 0.04 mag in the distances and 0.5 Gyr on the ages)

11. Our T eff scale agrees very well with that of Alonso et al. based on the IRFM Average difference is T(Us)-T(A)= 8  11 K (r.m.s.= 83 K, 58 stars) Eliminating nine outliers: r.m.s.= 38 K

12. Results Distances and Ages of Globular Clusters Impact of microscopic diffusion on models of low mass starsImpact of microscopic diffusion on models of low mass stars The O-Na anticorrelation among globular cluster TO-starsThe O-Na anticorrelation among globular cluster TO-stars Lithium abundances in TO-stars and subgiants of globular clustersLithium abundances in TO-stars and subgiants of globular clusters Comparison between abundances in GC and field starsComparison between abundances in GC and field stars Rotation of TO-stars in globular clusters

13. Globular Cluster Ages Absolute ages - lower limit to the age of the Universe - put formation of the Milky Way in a cosmological framework Cluster distances: a step to the extragalactic distance scale Relative Ages - GCs as probes to reconstruct early history of Galaxy formation

14. Comparison between confidence range for globular cluster ages and values allowed by Universe geometry

15. Costraints on the epoch of formation of globular clusters

16. True distance modulus to the LMC according various methods

17. Scenarios of MW formation Dissipational collapse (Eggen, Sandage & Lynden Bell (1962) Accretion (Searle & Zinn 1979) Numerical models suggest that both mechanism may be active in a galaxy: the relative weight may be crucial in determining galaxy morphology (spiral vs elliptical)

18. Field stars For field star kinematics and chemistry can be used to show the presence of two metal-poor populations: -a dissipative collapse component -an accretion component (Gratton et al., in preparation) time

19. GC relative ages Rosenberg et al. (1999): - metal-poor GCs all have the same age -some age spread for metal rich GCs End of Thick disk? Early phases of collapse?

20. Absolute Ages for GCs TO luminosity End of WD cooling sequence Nucleocosmochronology Require distances

21. End of WD cooling sequence Hansen et al. (2002): 12.7  0.7 Gyr for M4 However, De Marchi et al. (2002) analysis of this data only indicates an age larger than 10 Gyrs

22. Nucleocosmochronology Cayrel et al., Nature, 409, 691, 2001 Age (14  3 Gyr) of the extremely metal-poor star CS31082-001 from nucleocosmochronology with first identification of UII lines

23. Uncertainties in ages from TO -Observational: Distances: 0.07 mag 1 Gyr -Theoretical: Microscopic diffusion: about 1 Gyr

24. Distances to GCs Currently, the most accurate method is the main sequence fitting method In perspective, dynamical distances obtained combining proper motions and radial velocities (+ a dynamical model for the cluster) may provide distances accurate to a few percent within a few years from now

25. NGC6397 King et al. 1997 Potentiality of GC dynamical distances: few per cent accuracy

26. Errors in dynamical distances Anderson & King (2002) showed that astrometric accuracy of about 1 mas can be achieved with WFPC2 on HST. Over 10 yrs, the accuracy on proper motion is equivalent to errors of about 2 km/s in the transverse motion of stars within GCs Coupled with radial velocities with accuracies of about 1 km/s for some hundred GC stars (with e.g. FLAMES), and a model for internal motions, this may yield distances accurate to a few per cent for most GCs

27. Previous Globular Cluster distances from Main- Sequence fitting to local subdwarfs

28. Systematic effects and total error budget associated with previous MS fitting distances to Globular Clusters Effect  (m-M) Malmquist bias negligible Lutz-Kelker correction  0.02 Binaries (in the field)  0.02 Binaries (in clusters)  0.03 Photometric calibrations (0.01 mag)  0.04  Reddening scale (0.015 mag)  0.07  Metallicity scale (0.1 dex)  0.08  Total uncertainty (1  )  0.12 Reddening free T eff calibration  2 Gyr

29. Our results:

30. Line is not best fit, but the prediction of models by Chieffi & Straniero Colour of the main sequence at M V =6

31. Comparing the Teff-colour relations for field and cluster stars: Source E(B-V) NGC 6397 E(B-V) NGC6752 E(B-V)47Tuc (b-y) 0.178  0.007 0.045  0.007 0.021  0.005 (B-V) 0.186  0.006 0.035  0.007 0.035  0.009 1 average 0.183  0.005 0.040  0.005 0.024  0.004 Harris 0.18 0.04 0.05 Schlegel et al maps 0.187 0.056 0.032 1 Including correction in the photometry by Hesser et al. suggested by Percival et al. 2002 Astro-ph 0203157: (B-V) = 1.091 (B-V) Hesser – 0.048 Reddenings toward NGC6397, NGC6752 and 47 Tuc

32. NGC6397 E(B-V) 0.183  0.005 [Fe/H] -2.03  0.04 NGC6752 E(B-V) 0.040  0.005 [Fe/H] -1.42  0.04 47 Tucanae E(B-V) 0.024  0.004 [Fe/H] -0.66  0.04 Main sequence fitting distance to NGC6397 and NGC6752

33. Parameter NGC6397 NGC6752 47 Tuc [Fe/H] -2.03  0.04 -1.43  0.04 -0.66  0.04 [  /Fe] 0.34  0.02 0.29  0.02 0.30  0.02 [M/H] -1.79  0.04 -1.22  0.04 -0.45  0.04 (m-M) V 12.57 13.38 13.47 (from B-V) (m-M) V 12.62 13.15 13.57 (from b-y) (m-M) V 12.60  0.08 13.26  0.08 13.52  0.08 (average) (m-M) V 12.58  0.08 13.24  0.08 13.50  0.08 (bin. corr) V(TO) 16.56  0.02 17.39  0.03 17.68  0.05 (new measure) V(HB) 13.11  0.10 13.84  0.10 14.13  0.10 (using Rosenberg  V) M V (TO) 3.98  0.08 4.15  0.08 4.18  0.08 M V (HB) 0.53  0.13 0.60  0.13 0.63  0.13 Age (Gyr) 14.2  1.1 14.1  1.1 11.5  1.1 (No diffusion)11.5  1.1 Age (Gyr) 13.8  1.1 13.7  1.1 11.1  1.1 (Diffusion).11.1  1.1 Main parameters for NGC6397, NGC6752 and 47 Tuc

34. Comparison with previous data: Main Sequence Fitting Method [Fe/H] E(B-V) (m-M) V NGC 6397 Reid 1998 -1.82 0.19 12.83  0.15 Us -2.03  0.04 0.183  0.005 12.58  0.08 NGC6752 Reid 1998 -1.42 0.04 13.28  0.15 Carretta 2000 -1.43 0.035  0.005 13.34  0.04 Us -1.43  0.04 0.040  0.005 13.24  0.08 47 Tucanae Reid 1998 -0.70 0.04 13.68  0.15 Carretta 2000 -0.67 0.055  0.007 13.57  0.09 Percival 2002 -0.67 0.055  0.007 13.37  0.11 Us -0.66  0.04 0.024  0.004 13.50  0.08

35. Comparison with other data: White Dwarf cooling sequence Distances from white dwarf cooling sequence are independent on metallicity, but have a dependence on reddening similar to that from Main Sequence Fitting NGC 6752 Renzini et al. 1996 E(B-V)=0.04  0.02 13.17  0.03  0.10 Us 0.040  0.005 13.24  0.02  0.08 47 Tucanae Zoccali et al. 2001 E(B-V)=0.055  0.02 13.27  0.03  0.10 Us 0.024  0.004 13.50  0.02  0.08

36. The age difference between 47 Tuc and the two other clusters is real? A similar age difference is given by the horizontal method The horizontal age parameter is from Rosenberg et al.

37. Calibration of Relative ages from the horizontal method 14 Gyr 12 Gyr 8 Gyr 10 Gyr Relative ages from Rosenberg et al. 1999 End of Thick disk? Early phases of collapse?

38. Effect  (m-M) Malmquist bias negligible Lutz-Kelker correction  0.02 Binaries (in the field)  0.02 Binaries (in clusters)  0.03 Reddening scale (0.008 mag)  0.04  Metallicity scale (0.04 dex)  0.03  Total uncertainty (1  )  0.07 Reddening free T eff calibration Systematic effects and total error budget associated with the MS fitting distances to Globular Clusters  1 Gyr

39. Model Error Budget Error SourceDistributionLimits (Gyr) ConvectionFlat-0.4,0.4 CodeFlat-0.4,0.4 DiffusionGaussian0.4 (@-0.4) Solar M v Flat-0.3,0.3

40. Absolute Age Error Budget Distance modulus  0.07 mag  1.0 Gyr Model uncertainties (Carretta et al. 2000):  0.6 Gyr Chaboyer et al. (2001): ages are likely 4% smaller due to diffusion Best age estimate: 13.7  0.8  0.6 Gyr This corresponds to a redshift of z  4 and very likely >1

41. Epoch of formation of GCs z>3 for the oldest GCs z>1.3 for the youngest GCs

42. Limit on  M  M <0.57 at 95%

43. HB and RR Lyrae magnitudes M v (HB) = (0.22  0.05)([Fe/H]+1.5)+(0.56  0.07) This distance scale is 0.12 mag shorter than that previously obtained from the Main Sequence Fitting Method (Carretta et al. 2000) It is 0.03 mag shorter than the best distance scale proposed by Carretta et al. (2000)

44. Clementini et al. 2003 Dependence of the RR Lyrae magnitude on metallicity for variables in the bar of the LMC (photometric data from the Danish 1.5 m telescope and spectroscopic data from FORS1 at VLT UT1)

45. Distance estimates for the LMC

46. Comparison between various distance estimates for the LMC Old New

47. Microscopic diffusion is a basic physical mechanism, that should be included in stellar models It is needed to adequately reproduce the run of the sound speed within the solar interior as derived from helioseismogical data Microscopic diffusion

48. Kraft, Sneden and coworkers: The O-Na anticorrelation for giants in globular clusters

49. Diffusion causes sedimentation of heavy elements, mainly He Timescale for sedimentation is given by:   K M cz /(M T cz 3/2 ) where K is a constant, M cz is the mass and T cz the temperature at the base of the convective envelope, and M the star mass Effects of microscopic diffusion

50. Due to the low mass of the convective envelope, in low mass (M~0.8 M 0 ), metal-poor ([Fe/H]  -2) stars near the TO, also O and Fe are expected to be depleted significantly The net effects of sedimentation are: - ages are reduced by about 10% - Li abundances may be significantly reduced with respect to the original value Observations of TO and subgiants in NGC6397 (M~0.8 M 0, [Fe/H]=-2.0) allow to costrain sedimentation effects

51. Star S/N [Fe/H] [O/Fe] TO-stars 1543 91 -2.02 0.16 1622 82 -2.02 0.11 1905 92 -2.06 0.11 201432 97 -2.00 0.08 202765 59 -2.02 0.21 <> -2.02  0.01 Subgiants 669 91 -2.01 0.26 793 105 -2.04 <0.26 206810 85 -2.10 0.48 <> -2.05  0.03 Abundances in stars of NG6397

52. Model  [Fe/H] TO-subgiants Castellani et al. 2001 -0.25 for [Fe/H]= -2.0 Salasnich et al. 2000 -0.29 for [Fe/H]= -1.3 -0.78 for [Fe/H]= -2.3 Chieffi & Straniero 1997 -0.38 for [Fe/H]= -2.3 Chaboyer et al. 2001 <-0.28 for [Fe/H]= -2.0 NGC6397 +0.03  0.04 for [Fe/H]= -2.0 Conclusion: Models predict much larger sedimentation due to microscopic diffusion than actually observed. There should be some mechanism that prevents sedimentation Prediction of models with microscopic diffusion (0.8 M o )

53. - Diffusion computations assume full ionization and neglect the effects of radiation pressure: they may then overestimate sedimentation - Chaboyer et al. (2001) propose that mixing occurs at the base of the outer convective envelope. They found that our observations could be explained by an (ad hoc) mixing region of 0.005 M o near the stellar surface, where diffusion is inhibited. When diffusion is included according to this recipe, ages are reduced by 4% with respect to model without diffusion Discussion

54. Richard et al. 2002 Models with diffusion and radiation pressure predicts large overabundances of Fe at the end of the MS. These overabundances disagree with observations

55. These large overabundances may be eliminated if some turbulence at the base of the outer convective envelope is introduced Richard et al. 2002

56. When turbulence is included in the models, only a moderate depletion of Li (<0.2 dex) is predicted for stars on the Spite plateau

57. Li in NGC 6397 Bonifacio et al. 2002 A&A, 390, 91

58. Li doublet in TO-stars of NGC6397 Line strength is the same in all stars

59. Average Li abundance: log n(Li)=2.34 r.m.s=0.056 dex Maximum intrinsic scatter 0.035 dex This is to be fulfilled by stellar models which predict Li depletion.

60. Li abundances in field and (Na-poor) cluster stars. They occupy the same location Dilution factor is about 15 for both field (Gratton et al. 2000) and cluster stars, in agreement with theoretical predictions Spite’s plateau

61. Lithium abundances and primordial nucleosynthesis (figure from Suzuki et al. 2000) If our Li abundance in NGC6397 (log n(Li)=2.34) is primordial Li then the baryonic density is:  b h 2 =0.016  0.004 or  b h 2 =0.005  0.002 WMAP:  b h 2 =0.0224  0.0009 WMAP

62. Variations among MS stars in 47 Tuc (Briley et al. 1994) Variations in the strength of CH and CN bands Noticed since early seventies (Osborn 1971) from DDO photometry and spectroscopy Bimodal distribution along the RGB (Norris & Smith 1980s) NGC6752

63. Presence of elements processed through the complete CNO-cycle. At these temperatures 22 Ne+p  23 Na (Denissenkov & Denissenkova 1990; Langer & Hoffman 1995; Cavallo et al. 1996). At higher temperatures, also 26 Mg+p  27 Al From Langer et al. 1993

64. A first mixing episode occurs at the base of the RGB, due to the inward penetration of the outer convective envelope in regions where some H-burning (through uncomplete CN-cycle) occurred during the latest phases of MS evolution (first dredge-up: Iben 1964). First dredge up causes only minor effect in metal-poor stars At the same phases, dilution (by a factor of ~20) of the surface Li abundance occurs Mixing episodes along the RGB evolution of small mass stars

65. The maximum inward penetration of the outer convective envelope at the base of the RGB creates a discontinuity in molecular weight (  -barrier) that prevents further mixing, until is canceled by the outward expansion of the H-burning shell (RGB bump) (Sweigart & Mengel 1979; Charbonnel 1994). Further deep mixing (due e.g. to meridional circulations activated by core rotation) is possible only after the RGB bump Role of the molecular weight barrier

66. Molecular weight-barrier along the RGB (from Charbonnel et al. 1998)

67. Field stars conform this theoretical paradigma (Gratton et al. 2000) However abundances of O and Na are not affected:  mixing is not deep enough to reach regions where complete CNO cycle occurs

68. There is a systematic difference between field and cluster stars Important: this might be correlated with the 2nd parameter effect - Systematic different core-rotation  core and total mass at He-flash - Mixing of He It may also affect HB magnitudes (and then distance scales) Possible hints for a correlation between the 2nd parameter and the Na-O anticorrelation may be suggested by these graphs by Carretta et al. (1996)

69. What is going on in cluster stars? There are mainly two scenarios: - Deep mixing episodes: may only occur along the RGB, after the bump (temperature is not large enough in TO-stars) - Pollution: should be present independent of the evolutionary phase (the material comes from now extincted TP AGB stars,TP AGB stars undergoing hot bottom burning). Pollution might occur:. on protostars (Cottrell & Da Costa). on already formed stars (D’Antona, Gratton & Chieffi) Not distinguishable from observations of bright giants Observations of stars fainter than the bump

70. Mass lost by TP-AGB stars If we represent the mass function as: f(m)  k m -(1+  ) and  1, then the mass lost by TP-AGB stars is comparable to the mass in stars with mass m<1 M o (those presently seen in GC). This mass is lost at low velocity, and could perhaps be kept within the cluster. These stars might have hot bottom burning, at temperatures where complete CNO cycle occurs

71. The O-Na anticorrelation among globular cluster stars There are mainly two scenarios: - Deep mixing episodes: may only occur along the RGB (temperature is not large enough in TO-stars) - Pollution: should be present independent of the evolutionary phase (the material comes from now extincted TP AGB stars, undergoing hot bottom burning). Pollution might occur:. on protostars (Cottrell & Da Costa). on already formed stars (D’Antona, Gratton & Chieffi) Our observations of TO-stars in NGC6752 (a cluster which exhibits a clear O-Na among giants) allows to make a definitive test on the deep mixing scenarios

72. Na doublet at 8183-94 Å in TO-stars of NGC6752 (these stars have virtually identical atmospheric parameters) There is a clear star-to-star variation in Na abundances

73. OI triplet at 7771-75 Å in TO-stars of NGC6752. These stars have virtually identical atmospheric parameters. There is a clear star-to-star variation in O-abundances, anticorrelated with variations in Na abundances

74. The O-Na anticorrelation among stars in NGC6752. Filled squares: TO stars Empty squares: subgiants. The observed anticorrelation is very similar to that observed in giants

75. The Mg-Al anticorrelation among stars in NGC6752. Upper panel: TO stars Lower panel: subgiants. Na rich stars are Al-rich and Mg-poor. This is most clear among subgiants.

76. C-N anticorrelation in subgiants of NGC6752 CN-band at 3883 ÅG-band Stars are ordered according to increasing Na abundance [N/Fe]=1.0 [N/Fe]=1.1 [N/Fe]=1.3 [N/Fe]=0.0 [N/Fe]=1.2 [N/Fe]=1.3 [N/Fe]=1.2 [N/Fe]=1.45 [N/Fe]=1.5 [C/Fe]=-0.05 [C/Fe]=-0.40 [C/Fe]=-0.15 [C/Fe]=-0.35 [C/Fe]=-0.65 [C/Fe]=-0.60 [C/Fe]=-0.25 [C/Fe]=-0.35

77. C and N abundances in NGC6752 subgiants [(C+N)/Fe]=0 All O transformed into N

78. C and N abundances in subgiants of NGC6397 [N/Fe]=1.4 [N/Fe]=1.3 [N/Fe]=1.5 [C/Fe]=+0.05 [C/Fe]=-0.10 [C/Fe]=0.0 Very high N abundance ! [O/Fe]=+0.21  0.05 but [(C+N+O)/Fe]=+0.58  0.10

79. Conclusions: The O-Na anticorrelation is present among TO-stars and subgiants in NGC6752. For the same stars, also a Mg-Al anticorrelation is observed This clearly rules out deep mixing as explanation for the O- Na anticorrelation The sum of C+N abundances is not constant: a substantial fraction of O is transformed into N in some NGC6752 stars N is overabundant by a large factor in subgiants of NGC6397: while O is almost solar, the sum of C+N+O is overabundant as in halo field stars

80. Comparison between abundances in GC and field stars Abundances for a sample of 140 field stars with good parallaxes

81. Aims Comparison between abundances in field and cluster stars Comparison between abundances in thick disk and halo stars

82. The field star sample 140 subdwarfs and subgiants (this selection allows to reduce concern related to reddening and to derive homogenous abundances and kinematics)  /  <0.2 from Hipparcos M V >2.5

83. Data sources Parallaxes and proper motion from Hipparcos Radial velocities from Simbad Equivalent widths from high quality spectra: UVES, McDonald, SARG, Nissen et al., Fullbright, Prochaska et al. B-V and b-y from Simbad catalogue Information about binarity from Simbad and Carney et al. surveys

84. Abundance analysis Kurucz model atmospheres without overshooting Temperatures from colours (zero point given by the H  profile fitting) gravities from parallaxes, bolometric corrections from Kurucz, and evolutionary masses (assuming an age of 12 Gyr) microturbulent velocities eliminating trends of abundances with expected equivalent widths Non-LTE effects for O and Na

85. Thick disk vs Halo Halo stars: eccentricity > 0.5 OR Maximum height above plane > 2Kpc OR [Fe/H]<-2 Thick disk stars: not halo stars with: eccentricity > 0.2 OR Maximum height above plane > 0.8 Kpc OR [Fe/H]<-0.8

86. RESULTS

91. Papers - Na-O anticorrelation and tests of microscopic diffusion: Gratton et al. A&A, 369, 87 - Li in NGC 6397: Bonifacio et al. 2002 A&A, 390, 91 - Distances and Ages of Globular Clusters: Gratton et al. 2003 submitted to A&A - Rotation in TO-stars: Lucatello et al. 2003, A&A, in press - Abundances in field stars – data: Gratton et al. 2003, A&A, in press - Abundances in field stars – discussion: Gratton et al. 2003, A&A, in press - Abundances in 47 Tuc: Carretta et al., 2003, in preparation - Abundances in M30 and M55: Carretta et al. 2003, in preparation - n-rich elements in NGC6752: James et al. 2003, in preparation - Li in 47 Tuc: Pasquini et al. 2003, in preparation