1. Thermohaline Mixing in Low and Intermediate Metallicity Globular Clusters George Angelou The Globular Cluster-Galaxy Connection II. ● Prof John Lattanzio (Monash) ● Prof Graeme Smith (UCSC) ● Dr Richard Stancliffe (was Monash now RSAA) ● Dr Ross Church (was Monash now Lund)

2. George Angelou Monash University Kraftfest 2011 2 Outline ● The need for extra mixing. ● Thermohaline mixing. ● Multiple populations in globular clusters. ● Testing themohaline mixing in M3. ● Intermediate metallicity clusters. ● Metal poor clusters. ● Completing the picture.

3. George Angelou Monash University Kraftfest 2011 3 Extra Mixing Solar Type Metal Poor Charbonnel 1994, 1998 M=1.25 Z=0.02

4. George Angelou Monash University Kraftfest 2011 4 Lithium Lind, K., Primas, F., Charbonnel, C., Grundahl, F., & Asplund, M. 2009, A&A, 503, 545 NGC 6397

5. George Angelou Monash University Kraftfest 2011 5 Extra Mixing Smith, Martell PASP Volume 115, Issue 812, pp. 1211-1219

6. George Angelou Monash University Kraftfest 2011 6 Extra Mixing ● Stars are mixing along the RGB. Contrary to canonical theory (mixing by convection only). ● Has a well defined starting point ● “Physically it is the point where the hydrogen burning shell encounters the deepest point to which the convective envelope ever reached, and where the hydrogen content shows a sudden increase. ( Weiss & Charbonnel Mem. S.A.It. Vol. 75, 347, 2004.) ● Mixing efficiency decreases with mass but increases with metallicity.

7. George Angelou Monash University Kraftfest 2011 7 Extra Mixing ● There is a plethora of literature dedicated to finding out what is causing this extra mixing. ● Phenomenological models (Denissenkov & VandenBerg 2003; Wasserburg et al. 1995). ● Many physically based mechanisms have also been explored as a potential mechanisms: ● rotational mixing (Sweigart & Mengel 1979; Palacios et al. 2006) ● magnetic fields (Palmerini et al. 2009; Nordhaus et al. 2008; Busso et al. 2007; Hubbard & Dearborn 1980) ● internal gravity waves (Denissenkov & Tout 2000).

8. George Angelou Monash University Kraftfest 2011 8 Rotation Palacios, A., Charbonnel, C., Talon, S., & Siess, L. 2006, A&A, 453, 261

9. George Angelou Monash University Kraftfest 2011 9 Djehuty Hydrodynamics code developed at LLNL. First Fully 3D simulations of a star. Derived from a hydrogen bomb code. It allows us to deviate away from spherical symmetry. Unaccountable motions inside radiative zones. Realised a molecular weight inversion developed. Was on a scale of  ~10 -4. This was large enough for the inversion to grow unstable and drive extra mixing.

10. George Angelou Monash University Kraftfest 2011 10 A Natural Cause As hydrogen burning shell approaches deepest penetration of the convection zone the first burning to occur is: Not very big! The magnitude of the inversion scales with the mass fraction of 3 He. This reaction decreases  ! It thus creates 1) mean molecular weight inversion 2) excess energy locally Its unstable. But 1D models won't mix (unless you tell them to, and how to do it…)‏ Convection is driven by similarly small super-adiabatic gradients…

11. George Angelou Monash University Kraftfest 2011 11 What is Thermohaline Mixing? ● Doubly diffusive process ● Analogous to the salt finger mixing seen in the oceans. (Stern 1960)

12. George Angelou Monash University Kraftfest 2011 12 1D Parameterisation Initial value First Dredge-Up H shell advances on discontinuity  mixing begins 12 C/ 13 C reduced from ~30 to ~15 As observed! 1 M sun X=0.70 Z=0.02

13. George Angelou Monash University Kraftfest 2011 13 1D Parameterisation Eggleton, P. P., Dearborn, D. S. P., & Lattanzio, J. C. 2008, ApJ, 677, 581

14. George Angelou Monash University Kraftfest 2011 14 Thermohaline Mixing ● Recently has featured prominently in the literature. ● Extra Mixing on the RGB (Eggleton et al. 2006, Charbonnel & Zahn 2007, Eggleton et al. 2008). ● Stancliffe et al. 2009 have been able to explain the dichotomy between metal poor and carbon enriched metal poor stars. ● Cantiello and Langer 2010, Stancliffe 2010 have demonstrated may effect AGB evolution.

15. George Angelou Monash University Kraftfest 2011 15 Thermohaline Mixing ● Mixing depth and mixing speed dictated by a single free parameter in the diffusion coefficient. ● Displays required mass and metallicity dependence. ● Mixing is predicted to begin after the LF bump. ● We thus believe it pertinent to investigate the detailed physics of the mechanism and the best way to do so is by concentrating on a homogeneous population such as seen in Globular Clusters.

16. George Angelou Monash University Kraftfest 2011 16 But as we all know.... ● Globular clusters are not SSP ● Stars are not coeval – multiple populations. ● Stars are not chemically homogeneous. ● Star-to-star heterogeneity WRT C, N, O, Na, Mg, Al. (see Gratton 2004). ● Due to pollution from earlier stellar populations (Smith & Kraft 1996, Cannon et al. 1998, Pancino et al 2010.) ● Due to in situ exta mixing (Smith 2002). Which we have already seen a lot of evidence for.

17. George Angelou Monash University Kraftfest 2011 17 Enrichment Scenario F. Grundahl et al. 2002 A&A 385, L14–L17 ٱ Above the bump + Below the bump

18. George Angelou Monash University Kraftfest 2011 18 Enrichment Scenario ● A first generation imprints the abundance patterns from the various stellar burning sites. ● AGB stars (Fenner et al. 2004) ● SAGB stars (Ventura & D'Antona 2011) ● Massive Binaries (de Mink et al. 2009) ● Massive Rotating Stars (Decressin et al 2007) ● No discernible correlation with magnitude along the RGB. Should see on the MS too and we do. ● Also see CN cycled material and 4 He returned to the medium. ● New stars form from a mixture of pristine and enriched gas. ● A second generation should form with initial C down and N and 4 He up

19. George Angelou Monash University Kraftfest 2011 19 CN bands ● CN bands generally used as a tracer of nitrogen in a stellar atmosphere. ● Therefore expect a 'normal' or first generation which are the CN-weak stars ● Second generation C-depleted N- enriched which are the CN-strong stars (Norris et al. 1981) ● This dichotomy has been detected below the LF bump so can't be due to extra mixing changing the bands (eg Briley et al. 1991, Suntzeff & Smith 1991, Cohen 1999, Briley & Cohen 2001) ● Can use the CN bands to trace the populations.

20. George Angelou Monash University Kraftfest 2011 20 Hybrid Theory of GC evolution ● In order to match observations pollution from a primordial generation must be present in the cluster providing an enriched environment from which a second generation can form. ● Extra mixing must be operating in each generation (Norris et al 1981, Suntzeff & Smith 1991, Dennisenkov et al. 1998, Briley et al. 1999). ● Smith 2002 suggested this is exactly what was occuring in M3 so we decided to see if thermohaline mixing as the extra mixing mechanism could explain the abundance variation.

21. George Angelou Monash University Kraftfest 2011 21 Of all the globular clusters, in all the galaxies in all the super clusters why did we pick Graeme's? ● Considered a typical GC ● [Fe/H] ~ -1.4 (Sneden et al. 2004, Cohen and Melendez 2005) ● Falls in the mode of the metallicity distribution. ● Plenty of evidence for [C/Fe] depletion as a function of magnitude. Reduction isotopic ratio. (Too many studies to list!) ● CN bands studied (Suntzeff 1981, Smolinski et al. 2011) M3

22. George Angelou Monash University Kraftfest 2011 22 [C/Fe] vs 12 C/ 13 C

23. George Angelou Monash University Kraftfest 2011 23 M3

24. George Angelou Monash University Kraftfest 2011 24 Mv Ct=12 Ct=120 Ct=600 Ct=1000 ● Observations suggest Ct~1000 ● Charbonnel and Zahn (2007) based their results on laboratory experiments and require ~1000. ● Kippenhahn et al. (1980) derived a linear theory of thermohaline mixing with spherical blobs ~12. ● Dennisenkov & Merryfield (2011), Traxler et al. 2010. Their hydro simulations say ~12. M3 CN Weak

25. George Angelou Monash University Kraftfest 2011 25 Mv Ct=12 Ct=120 Ct=600 Ct=1000 ● Thermohaline mixing can account for the evolution of [C/Fe] in both the CN-weak and CN-strong stars. ● Only affects the nitrogen abundance in the CN-weak stars. ● CN strong stars have so much nitrogen to begin with any extra mixing does not affect the surface composition. M3 CN Strong 1

26. George Angelou Monash University Kraftfest 2011 26 Mv Ct=12 Ct=120 Ct=600 Ct=1000 ● Furthermore spread in [N/Fe] dominated by variation that was present in the cluster before stars commenced RGB evolution. ● Lack of observations of CN strong stars at low L. Artefact of the original studies used. Other studies have shown that the stars are there. (Norris and Smith 1984, Smolinski 2011) ● See Angelou et al. (2011) M3 CN Strong 2

27. George Angelou Monash University Kraftfest 2011 27 M3 and M13: [Fe/H]= -1.5 Ct=1000, CN Weak Ct=1000, CN Strong

28. George Angelou Monash University Kraftfest 2011 28 M13 ● SGB observations show spread is primordial. ● Thermohaline mixing can model the behaviour of [C/Fe] in both M3 and M13 with the same parameter. ● Significantly more stars N- enhanced ● Third model where half O is cycled into N ● We don't have CN band strength for our sample. But Matching the MS spread we are able to match the RGB. Mv

29. George Angelou Monash University Kraftfest 2011 29 Low Z ● Works for two clusters at intermediate metallicity therefore it must work for all right?!...... ● Obviously youthful blissful optimism! ● There is data in the literature for NGC 5466, M92 and M15 all metallicity ~[Fe/H] =-2.2

30. George Angelou Monash University Kraftfest 2011 30 NGC 5466 Data is from Shetrone et al (2010). Solid line is end FDU according to models Dashed line is photometrically derived LF bump (Fekadu et al 2007) Dotted line is LF bump derived by Martell. et al (2008) from metallicity-bump relation

31. George Angelou Monash University Kraftfest 2011 31 NGC 5466 ● Only require single scaled model to fit the data. ● Shetrone et al (2010) also analysed the CN bands.

32. George Angelou Monash University Kraftfest 2011 32 CN bands at low metallicity ● CN bands at 3880A and 4220A are often weak at this metallicity and luminosity (Cohen et al 2005). ● Obviously less metals at low Z. ● Another example of CN bands at low metallicity....

33. George Angelou Monash University Kraftfest 2011 33 M15 Trefzger et al. (1983, crosses) and Cohen et al. (2005, circles) Smolinski et al. 2011 Evolution

34. George Angelou Monash University Kraftfest 2011 34 CN Bands at low metallicity ● CN bands are not representative of Nitrogen abundance at low metallicity?? ● Not useful for identifying multiple populations. ● Sneden (1997, 2000) have shown 1 dex variation in Na that does not correlate with RGB evolution. So we can infer there are multiple populations ● Will just worry about matching the spread at low Z. No consideration of multiple populations ● Lets see what is happening in M15 and M92

35. George Angelou Monash University Kraftfest 2011 35 M15 and M92: [Fe/H]=-2.2 ● Black solid vertical line is FDU from models ● Dotted line LF bump from Martel et al 2008 ● Dashed line various photometrically derived values of LF bump ● Appears to be greater spread in [C/Fe] in M92 and M15 ● Applied a C depleted model to match spread ● Thermohaline seems to explain NGC 5466 ● Does not explain M92 or M15 Angelou et al. (In prep)

36. George Angelou Monash University Kraftfest 2011 36 ● All is not as it seems ● Pre FDU mixing?....Maybe ● Both look like mixing before FDU. So even if you don't believe thermohaline mixing you have a problem ● Systematic error by using multiple studies? ● C in M15? ● M15 and M92 both show similar star-to-star r-process dispersion (Roederer 2011) M15 and M92 Mv M92 M15 X Trefzger et al. (1983) ◊ Cohen et al. (2005) Carbon et al (1982) Langer (1986, 1988) Bellman et al (2001)

37. George Angelou Monash University Kraftfest 2011 37 Completing the Picture ● Best plan of attack is to fill in the gaps with the clusters we already have observations of ● Low Lum stars in M3 ● Confirm the low luminosity stars divide into CN strong and CN weak the way we expect (ie with the [C/Fe] abundances assumed by the models) ● CN measurements for sample in M13 ● Applying the initial abundances used for the M3 models are justified ● Homogeneous [C/Fe] over a large luminosity range in M15 and M92. ● Does M92 have the same primordial [C/Fe] spread as M15? ● Need to know if there is a problem with stellar theory. Mixing before the bump? ● Is using multiple studies at low luminosity clouding our perception of the mixing history?

38. George Angelou Monash University Kraftfest 2011 38 Completing the Picture ● C in M15 ● Is this telling us about deep mixing? ● [Na/Fe] ● Can you use this in a similar way to CN bands to infer the populations? ● Na-O anticorrelation ● [O/Fe] ● get C+N+O, Tells us about the progenitors and burning sites. ● O-Na anticorrelation. ● HB morphology ● Kraft et al 93, 97, Sneden et al 2004, have shown some strange behaviour in M13 WRT O

39. George Angelou Monash University Kraftfest 2011 39 Completing the Picture ● Lithium gets a slide of its own. Lind et al (2009) and Mucciarelli et al (2010) ● Yes we realise how difficult it is to get reliable measurements. ● Very sensitive ● Identify where extra mixing begins in M15 and M92. ● Must be considered distinct from LF bump in M15 and M92

40. George Angelou Monash University Kraftfest 2011 40 Conclusions ● Thermohaline mixing operating in multiple populations can explain the variation of [C/Fe] and [N/Fe] in M13 and M3 ● Difficult to use CN bands to distinguish populations at low Z ● Can tentatively explain spread of [C/Fe] in NGC 5466 ● Can not explain evolution of [C/Fe] in M92 and M15 which currently appear inconsistent with stellar evolution ● As always us modellers are being whiny and want more observations

41. George Angelou Monash University Kraftfest 2011 41 Prior to first dredge-up After first dredge-up 3 He and the Big Bang in Crisis

42. George Angelou Monash University Kraftfest 2011 42 M3 and M13 Smolinski et al 2011