M5: SDSS Observational(!) Project: AGB stars in Galactic Globular Clusters Are They Chemically Different to Their Fellow RGB, HB and MS Stars? AKA: Why has Simon been staring at spectra of real stars lately? Simon Campbell, Monash Centre for Astrophysics (MoCA)
Collaborators RSAA, Mt Stromlo, Australia: David Yong Elizabeth Wylie de Boer Monash University, Australia: John Lattanzio Richard Stancliffe George Angelou University of Aarhus, Denmark: Frank Grundahl University of Texas, USA: Chris Sneden
Part 1: Background on GC Abundances
Background: Spectroscopic/Chemical Anatomy of GCs: Fe Group Most globular clusters (GCs) have a very uniform distribution of Fe group elements - all the stars have the same [Fe/H]. Fe I Fe II This indicates that the cluster was well mixed when the stars formed. M4: Lisa Elliott 2003 Sc II VI Teff Kraft, et al., 1992: M3, M13
Background: Early Observations: N In contrast to the Fe group, it has been known since the early 1970s that there is a spread in Carbon and Nitrogen in many GCs. Cyanogen (CN) was observed to have a bimodal distribution implying one population has a normal N content whist the second has high N content. Norris et al. 1981 References: eg. Bell & Dickens 1974, Da Costa & Cottrell 1980, Norris et al. 1981. 5
Background 2: The C-N 'Anticorrelation' The first negative correlation (anticorrelation) was found 25 years ago -- C is low when N is high. The anticorrelation is explicable in terms of the C-N cycle, where C is burnt to N14: Smith et al. 2005 (RGBs) This is also observed in halo field stars (eg. Gratton et al, 2000)
Background 3: The C-L Anticorrelation However it has also been observed that the C abundance decreases with L on the RGB (and N increases). This is known as the C-L anticorrelation: It appears to (only?) occur in low-metallicity GCs (eg. M3, M13, M10) Evolutionary Effect => Deep/Extra Mixing must exist! (at least on RGB) [C/Fe] [N/Fe] This is also observed in halo field stars. (eg. Gratton et al, 2000) M3 RGBs, Smith 2002 Mag Mag
Background 2: GC Weirdness; O-Na Anticorrelation An O-Na anticorrellation exists in many GCs. This anticorrelation is readily explained by hot hydrogen burning, where the ON and NeNa chains are operating the ON reduces O, whilst the NeNa increases Na (T~45 million K) Where this nucleosynthesis occurs is still a matter of debate. This is not observed in field stars GC weirdness. O-Na anticorrellation is not observed in halo field stars!! [ N a / F e ] Fiel d [ N a / F e ] Clust er [O/Fe ] Gratton et al, 2000
Part 2: Our Observational Project: AGBs in GCs
More GC Weirdness? CN in AGBs Norris et al. 1981 noted that their sample of AGB stars in NGC 6752 were all CN-weak (triangles = AGBs). Could this be chance, due to the small sample size or is something strange happening on the AGB?? NGC 6752, Norris et al 1981 CN Index AGBs: all CN-weak -S N C Note trend with Temp. ng o tr CN k a e -W Mag References: eg. Bell & Dickens 1974, Da Costa & Cottrell 1980, Norris et al. 1981.
MS/SGB Observations Interestingly the CN band strengths are distributed in the same way in the less evolved populations in GCs the AGB could be a glaring exception. 47 Tuc Cannon et al, 1998 Briley et al, 2004
An Important piece of evidence? If this is true, then it means that something is happening to the stars between the RGB and the AGB! This is an interesting proposition, possibly being an important clue in the abundance anomaly or second parameter problems. So, I went & checked 'all' references relating to AGB stars in GCs on ADS, to see if this had been confirmed yet.
Literature search for CN in GC AGB Stars Intrigued that the AGB may be showing very strange behaviour in GCs, we conducted a literature search to see if the same had been found in other GCs. It appears some GC AGBs had been looked at, but none in any detail. AGB stars were generally a side issue in the studies, due to their low numbers and the difficulty in identifying them. (Ivans et al, 2004) (Smith & Norris,1993) (Mallia 1978 +?) M5 & 47 Tuc the 'Contrary' GCs (Sunzeff, 1981) (Suntzeff, 1981) (Smith & Norris 1993) (Briley et al., 1993) (Lee, 2000) So, the data generally points to CN-weak AGBs, but there is also evidence for CNstrong AGBs... Note however that the sample sizes are quite small...
CN in AGBs Conclusion: The data so far hint towards a possible difference between the RGB and RGB, but we can't say for sure. => It's worth doing if you can get a big enough sample of AGB stars. The problem with getting decent AGB samples is that one needs excellent photometry to split the RGB and AGB in colour. So I asked ADS if there's any photometry out there that can give us decent numbers...
Photometry Search: AGBs in Globular Clusters It turns out that high quality photometry is now making the AGB accessible in GCs, we can now get good numbers of AGBs. M5 (Sandquist & Bolte 2004) 100 AGBs! I B-I M5 (SDSS)
References for the Photometric Studies Used NGC 362: Bellazzini et al. 2001 NGC 6752: Grundahl (private comm.) NGC 288: Grundahl (private comm.) M4: Mochejska et al. 2002 Omega Cen: Sollima et al. 2005 NGC 1851: Walker 1992 M2: Lee 1999 M10: Pollard 2005 M5: Sandquist & Bolte 2004 47 Tuc: Kaluzny et al. 1998 (OGLE survey)
Which Telescope/Instrument? Need low/mid resolution only (R ~ 3000, CN bands are huge), but want to look at many stars. => Our good old friend the AAT, with its multi-fibre-fed spectrographs AAOMega (2 degree field, 400 stars at once) Doing it this way gives a strong benefit to the study the data from all the stars in all the GCs will be homogeneous. 2dF Field Plate
Summary of the Proposal Get low-resolution spectra for statistically significant sample of AGB stars in 3 GCs. Observe some HB stars also, as this may let us know when/if the stars decide to go up the AGB. RGBs will be the control stars as they are very well studied already - and have similar temps (etc) to AGBs. Try for Al might have high enough resolution (?) We will be able to get CH (which is a proxy for C) but we can't get NH (for N) because the range of the spectrograph doesn't go that blue. Proposal history: * 2005A 2df Service proposal: Rating = 4/5 Never executed.. * 2007: WiFes Science Verification proposal: Waiting * 2008B AAOmega full proposal: Rejected * 2008B AAOmega Service proposal: Rating = 3/5 Never executed * 2009B AAOmega full proposal: Successful!! :) Observed Sep/2009 * 2009 Sep: AAT Service Proposal (OmCen): Rating = 3.8/5 : Waiting (exp. 2011)
Run Preparation Example Guide stars Sky (starless) positions Science targets Need very accurate positions for everything cross-reference with 2MASS catalogue. A lot of work!
2dF Field Plate Setup
AAT Run: 5-9/Sep 2009 Rich, George & Liz
Exposure Details for the Observers :)
A Logistical Game!
Part 3: Reducing and Analysing the Data
2dfdr Data Reduction Pipeline Tram map
Bias 2dfdr Data Reduction Pipeline Arc exposure for wavelength calibration
Example of a Reduced Spectrum: AGB Star in NGC 6752 (full blue arm spectrum)
Finding GC non-members - IRAF+fxcor First apply doppler shift due to GC radial velocity to get spectra in rest frame. Ten use Fxcor to get radial velocities (Fourier cross-correlation) Only a few stars found to be non-members, depended on quality of stellar datasets.
Quantifying Cyanogen Abundance: The S(3839) CN Index Cyanogen (CN) is a molecule whose abundance is thought to track that of Nitrogen. It absorbs over a few regions in the spectrum. Here we consider the Blue CN bands. Blue CN Index Norris et al. (1981) Basically you see how much flux is missing in a wavelength region due to CN absorption by comparing to piece of 'continuum' nearby. Only need fairly low (~2 Ang) resolution.
Example Spectra: CN Band Region Norris et al. (1981) Total = 247 AGB + 207 RGB +79 HB + 55 BGB = 688 spectra! (9 GCs)
Part 4: Results
Spectra Collected: Number of AGBs 5 nights on AAT Multi-object spectrograph 2dF/AAOmega Data collected for 247 AGB stars across 9 clusters (plus many RGB & HB stars).
Tons of Data... 247 AGBs + 307 RGBs + 79 Hbs + 55 BGBs = 688 Spectra!!
Results: NGC 6752 The cluster that Norris et al 1981 investigated. RGB nicely bimodal, as expected. And on the AGB... Strong to Weak Ratios *ALL AGBs CN-weak!* RGB = 80:20 AGB = 0:100
Error Bars Internal Due to errors in wavelength calibration, resolution of spectra, velocity dispersion, etc: ~ +/- 30 km/s ~ +/- 0.3 Ang. = +/- 0.005 in CN index Error bars = +/-0.02 However... Between Observations Some stars observed twice. The CN values vary significantly between subsequent observations...! Typical difference: +/- 0.02 in CN index This is main source of error! **Beware of inter-observation & inter-author error bars**
Results: NGC 6752 CMD Where did all the CN-Strong stars go??!! NGC 6752 photometry: Grundahl et al., 1999.
Results: GC Pair Comparison NGC 288 and 362 have similar metallicities ([Fe/H] ~ -1.2) but different HB morphologies compare CN behaviour. Ext-Blue HB Red HB
Results: NGC 288 (Blue HB) The normal CN bimodality is seen on the RGB. And on the AGB... Strong to Weak Ratios RGB = 50:50 AGB = 0:100 A totally CN-weak AGB! just like NGC 6752, which also has a very blue HB NGC 288 photometry: Grundahl et al., 1999.
Results: NGC 362 (Red HB) Strong to Weak Ratios RGB = 60:40 AGB = 40:60 to 60:40 Either a CN-weak dominated AGB, or no change from RGB (hard to define the bimodal split) Different to 288 & 6752. NGC 362 photometry: Bellazzini et al., 2001.
M10: [Fe/H] = -1.1 (Blue HB) Almost all CN-weak AGB. Blue HB like 6752 & 288
NGC 1851, [Fe/H] = -1.2 (Red+Blue HB) Strong to Weak Ratios RGB = 60:40 AGB = 60:40 to 50:50 => Either no change or a paucity of CN-strong AGB stars compared to RGB.
M4, [Fe/H] = -1.2 (Red+Blue HB) Blue + Red HB, like NGC 1851. But looks to be totally CN-weak AGB...!? More like a blue HB GC.. Not many AGB observations though.
M5 The 'Contrary' GC, [Fe/H] = -1.3 M5 was thought to have a CN-strong dominated AGB, however this was based on 8 stars (Smith & Norris 1993). Our data shows it certainly has CN-strong stars on the AGB, but it actually seems to be dominated by CN-weak stars.. Red + Blue HB CN Ind ex More complex than NGC 6752... M5 photometry: Sandquist & Bolte 2004. B_Mag
Metal Poor: M2 Monomodal RGB? M2 is the most metal-poor GC in our sample ([Fe/H] = -1.65) RGB seems almost totally CN-weak rather than bimodal. Maybe this is due to the low metallicity? It looks as if the AGB is more CN-weak, so same process has happened here? Extremely blue HB.
Metal-Rich GC: 47 Tuc (Red HB) Strong to Weak Ratios RGB = 30:70 (CN-weak) HB = 40:60 (intermediate) AGB = 70:30 (CN-strong!!) 47 Tuc photometry: Kaluzny et al., 1998.
Getting Carbon: CH (G-band) Similar process as measuring CN bands, but with 2 pseudo continua. May be able to get an estimate of [C/Fe] from these measurements. AGBs really stand out here!
Summary/Discussion Our preliminary results clearly show there is something strongly effecting the numbers of CN-strong and CN-weak stars between the RGB, HB and AGB. It appears to be related to the HB morphology of the GCs. GCs with red HBs show little or no change in the ratio of CN-strong to CN-weak stars going from the RGB to AGB. However in GCs with very blue HBs it is startling to find that there are zero CN-strong stars on the AGB (eg. 6752, 288) the CN-strong stars seem to disappear when moving from the RGB to AGB. So what is happening?? Maybe the CN-strong stars don't ascend the AGB at all? (an idea suggested by Norris et al. 1981). The fact that this feature is (mainly) seen in GCs with blue HBs suggests this may be the case, since the blue HB stars should have low masses. Primordial abundance variations (eg. He, N) may affect mass loss or other evolution.
Things Left To Do Get more chemical information from current spectra (CH, Al and maybe Li?) Do spectral synthesis to check magnitude of N overabundance, and (maybe) get [C,N,Al/Fe]. Analyse our HB data clues as to where things change? Use these sets of AGB stars for higher resolution observations, to check for additional abundance variations/correlations (Na, O, Mg, etc.) VLT proposal submitted but didn't get time this round :( Models to explain this strange change between the RGB and AGB!
Thanks! Thanks heaps to the observers! Richard Stancliffe Elizabeth Wylie de Boer George Angelou David Yong Thanks to the producers & maintainers of these tools which made life much easier: 2MASS an excellent resource for accurate astrometry. IRAF ViZieR & Aladdin. 2dFdr, the 2dF data reduction software. SIMBAD
The End :) 50
Bellazzini et al. 2001, 122, 2569
288: [Fe/H] = -1.2, BHB only 362: [Fe/H] = -1.2, RHB only M5: [Fe/H] = -1.3, RHB+BHB M2: [Fe/H] = -1.6, EBHB 6752: [Fe/H] = -1.6, EBHB 1851: [Fe/H] = -1.2, no Blue HB tail.
Source of the Observational Project Idea: With all the reading on abundances (and SINS discussions) at the time, we came across an interesting observation in an old paper by Norris et al (1981):
GCs Not so Simple After all! NGC 1851 Han et al. 2008
CN Bimodality in GC Giants Early observations of GC giants showed that the molecule Cyanogen (CN) has a bimodal distribution. This suggests that each population CN-weak and CN-strong have different nitrogen abundances. This is not observed in halo field stars! NGC 6752, Norris et al 1981 eg. Langer et al, 1992) distribution on RGB No. Bimodal CN(see Sta rs C N In de x -S N C Note trend with Temp. g n t ro CN k ea -W Mag References: eg. Bell & Dickens 1974, Da Costa & Cottrell 1980, Norris et al. 1981.
GC Abundance Summary Figuring out why things are different in GCs as compared to the field is a long-standing, difficult problem. MS observations have now been made many of the abundance anomalies are found there too, suggesting a primordial origin (but some anomalies are certainly evolutionary, eg. RGB Extra Mixing). So we have abundance anomalies at each stage of evolution MS, SGB, RGB, HB. However, it seems that the AGBs haven't really been looked at in detail because it is difficult to identify AGB stars & also there are not many of them due to their short lifespans...
An Interesting Proposition It is an interesting proposition that the AGBs of GCs could be chemically distinct from the RGBs (and MS, etc). This is not predicted by standard stellar evolution there is no known evolutionary phase between the RGB and AGB that changes the surface composition (note that these are generally EAGB stars -- so no TDU yet). Furthermore, taking into account Deep Mixing on the RGB, which tends to increase N, the (apparent) general trend seen in the AGBs is the opposite to that which would be expected... However all this speculation is based on small samples of AGB stars, so we can't say for sure an in-depth study is needed to settle the issue hence our observing project :) This is easier said than done, since there are so few AGB stars in GCs, plus it is difficult to separate them in colour from the RGBs...
Excellent Photometry Needed Excellent photometry is needed in order to split the two giant branches. v versus v-y CMD gave good AGB-RGB splitting fro NGC 6752. (NGC 6752 data from Frank Grundahl) BGB=RGB+AGB NGC 6752 AGB v RGB (yellow = raw data) v-y