THE CHEMISTRY AND PROPERTIES OF STARS IN NGC 362

THE CHEMISTRY AND PROPERTIES OF STARS IN NGC 362 A thesis submitted to the University of Manchester for the degree of Master of Science in the Faculty of Engineering and Physical Sciences November 2014 By Lucas Ben Assayag School of Physics and Astronomy

Contents Abstract 8 Declaration 9 Copyright 10 Acknowledgments 12 Conventions and Abbreviations 13 1 Introduction 14 1.1 Motivation................................. 14 1.2 Definitions................................. 14 1.2.1 Abundance............................ 14 1.2.2 Stellar populations........................ 15 1.3 The chemical evolution of the Universe................. 16 1.3.1 Formation of the first stars.................... 17 1.3.2 Low-mass stellar evolution.................... 17 1.3.3 High-mass stellar evolution.................... 19 1.4 Globular Clusters............................. 20 1.4.1 Overview............................. 20 1.4.2 Abundance studies........................ 21 1.5 NGC 362................................. 25 1.5.1 The second parameter problem................. 26 1.5.2 Previous studies.......................... 28 2

1.6 Overview of this thesis.......................... 30 2 Data Processing 31 2.1 Equivalent Width............................. 38 2.2 Equivalent Width measurement..................... 39 2.3 MOOG Software............................. 41 2.4 Results................................... 45 3 Other elements 50 3.1 Results................................... 50 3.2 Sodium.................................. 54 3.3 Silicon................................... 55 3.4 Calcium.................................. 57 3.5 Titanium................................. 58 3.6 α elements................................. 60 3.7 Conclusions................................ 61 4 Error estimation 62 4.1 Statistical abundance uncertainties................... 62 4.2 Abundance uncertainty to model atmosphere parameters....... 67 5 Discussion 69 5.1 Radial dependences of abundances................... 69 5.1.1 Iron abundance.......................... 70 5.1.2 Sodium abundance........................ 74 5.1.3 Silicon abundance......................... 75 5.1.4 Calcium abundance........................ 76 5.1.5 Titanium abundance....................... 77 5.1.6 α-element abundance....................... 78 5.2 Comparison of stellar parameters.................... 81 5.3 Comparison with the literature..................... 86 3

5.3.1 Comparison of the iron abundances............... 87 5.3.2 Comparison of the sodium abundances............. 88 5.3.3 Comparison of the silicon abundances.............. 89 5.3.4 Comparison of the calcium abundances............. 91 5.3.5 Comparison of the titanium abundances............ 92 5.4 Enrichment history of NGC 362..................... 94 6 Conclusion 96 4

List of Tables 1.1 Iron abundance studies of NGC 362................... 25 1.2 Abundance studies of NGC 362..................... 28 2.1 Log of FLAMES observations for NGC 362............... 33 2.2 Line List.................................. 34 2.3 List of Galactic Stars........................... 40 2.4 Derived atmospheric parameters..................... 45 2.5 Metallicity studies of NGC 362..................... 48 3.1 Derived abundances............................ 51 3.2 Comparison between studies of NGC 362................ 54 3.3 Star-by-star comparison between this study and Carretta et al. (2013) 54 4.1 Statistical uncertainties.......................... 63 4.2 Abundance sensitivity to model atmosphere parameters........ 68 5.1 Comparison between spectroscopic and photometric effective temperatures................................... 81 5

List of Figures 1.1 Comparison of spectra between stars having different metallicity... 16 1.2 HR Diagram of a Sun-like star...................... 18 1.3 Example of Na-O anticorrelatoin in different clusters......... 22 1.4 Colour magnitude diagram of the double main sequence in ω Centauri 24 1.5 Colour-magnitude diagrams for NGC 288 and NGC 362........ 26 2.1 Hertzsprung Russell Diagram of NGC 362............... 32 2.2 Image of NGC 362............................ 33 2.3 Equivalent Width............................. 38 2.4 Interface of the first program....................... 39 2.5 Continuum setting............................ 41 2.6 Example of a deblending......................... 42 2.7 Graphical determination of the model atmosphere with MOOG... 43 2.8 [Fe/H] vs T eff............................... 49 3.1 [Na/Fe] vs [Fe/H]............................. 55 3.2 [Si/Fe] vs [Fe/H]............................. 57 3.3 [Ca/Fe] vs [Fe/H]............................. 58 3.4 [Ti/Fe] vs [Fe/H]............................. 59 3.5 [α/fe] vs [Fe/H].............................. 60 5.1 Histogram of the distribution of the distance from the centre of the cluster.................................... 70 5.2 [Fe/H] vs Distance from the centre of NGC 362............ 71 6

5.3 [Fe/H] vs log g.............................. 73 5.4 [Na/Fe] vs Distance from the centre of NGC 362............ 74 5.5 [Si/Fe] vs Distance from the centre of NGC 362............ 75 5.6 [Ca/Fe] vs Distance from the centre of NGC 362............ 76 5.7 [Ti/Fe] vs Distance from the centre of NGC 362............ 77 5.8 [α/fe] vs Distance from the centre of NGC 362............ 78 5.9 [α/h] vs Distance from the centre of NGC 362............. 79 5.10 [Ti/H] vs Distance from the centre of NGC 362............ 80 5.11 T effspec T effphot vs Distance from the centre of NGC 362........ 85 5.12 T effspec and T effphot vs Distance from the centre of NGC 362...... 85 5.13 Comparison of [Fe/H] vs T eff....................... 87 5.14 Comparison of [Na/Fe] vs [Fe/H].................... 88 5.15 Comparison of [Si/Fe] vs [Fe/H]..................... 89 5.16 ɛ(si) vs [Fe/H].............................. 90 5.17 Comparison of [Ca/Fe] vs [Fe/H].................... 91 5.18 ɛ(ca) vs [Fe/H].............................. 92 5.19 Comparison of [Ti/Fe] vs [Fe/H]..................... 93 5.20 Theoretical yields for the chemical enrichment due to Type II SN explosions................................. 95 7

Abstract Abundance studies in Galactic globular clusters increase knowledge of the chemical enrichment of the Universe and the formation of stars or galaxies. This thesis presents chemical abundances for iron, calcium, silicon and titanium in 56 stars in the Galactic globular cluster NGC 362 from spectra obtained with FLAMES, the multi-object spectrograph at the Very Large Telescope. The abundances were derived using equivalent width measurements. This study covered wavelengths from 6300 to 6900 Å. The average iron abundance derived is [Fe/H] = 1.26 ± 0.03 dex, consistent with previous studies. An unexpected scatter was found, explained by the presence of blended stars and the decision to determine each atmospheric parameter (T eff, log(g), v t and [Fe/H]) independently of photometric measurements. It might be better to fix one of them for future studies. The average sodium, silicon and calcium abundances are also in agreement with the literature. We determined [Na/Fe] = 0.11 ± 0.18 dex, [Si/Fe] = 0.25 ± 0.13 dex and [Ca/Fe] = 0.38 ± 0.06 dex. The titanium abundance determined in this thesis is higher by 0.40 dex than previous studies, due to effects caused by the proximity of the 6562 Å Hα line. The measured abundance for this study is [Ti/Fe] = 0.56 ± 0.20 dex. The enhancement of the α elements (Si, Ca and Ti) suggests an enrichment from Type II Supernovae (SNe) and minimal contributions from Type Ia SNe. 8

Declaration No portion of the work referred to in this thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institution of learning. 9

Copyright i. The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the Copyright) and s/he has given The University of Manchester certain rights to use such Copyright, including for administrative purposes. ii. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made. iii. The ownership of certain Copyright, patents, designs, trade marks and other intellectual property (the Intellectual Property) and any reproductions of copyright works in the thesis, for example graphs and tables (Reproductions), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. iv. Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy (see http://www.campus.manchester.ac.uk/medialibrary/policies/intellectual- property.pdf), in any relevant Thesis restriction declarations deposited in the University Library, The University Librarys regulations (see http://www.manchester.ac.uk/library/ 10

aboutus/regulations) and in The University s policy on presentation of Theses. 11

Acknowledgments First I would like to thank my supervisors Dr. Iain McDonald and Prof. Albert Zijlstra. Iain, I cannot believe how much you have done for me throughout the year! You are an incredible supervisor. Albert, thank you for the wise advice you gave me. I would also like to thank the students of the East Wing, especially Skandar, Mike and Chris for their warm welcome and for making this year abroad a very nice experience. This thesis would have never been written without the support I got from my family. Papa, Maman, Ugo and Andy, you have always believed in me and I am very grateful for that. My thanks also go out to all my friends that remained in France, it was so good to see you during the holidays to release the pressure. Thanks for the good laughs you gave me with our discussions. Finally, I wish to thank my girlfriend Salomé for her love and support. You cannot imagine how helpful it was to be able to talk to you everyday. 12

Conventions and Abbreviations The conventional abbreviations for SI units and astronomical quantities are used. The following abbreviations have been used in this thesis: ACS Advanced Camera for Surveys AGB Asymptotic Giant Branch Dec Declination EW Equivalent Width FLAMES Fibre Large Array Multi Element Spectrograph HB Horizontal Branch HR(D) Hertzsprung-Russell (Diagram) HST Hubble Space Telescope ISM Interstellar Medium LTE Local Thermodynamic Equilibrium MEDUSA Multi-Object Spectroscopy mode NGC New General Catalogue of Nebulae and Clusters of Stars RA Right Ascension RGB Red Giant Branch S/N Signal-to-noise (ratio) SN(e) Supernova(e) UT Unit Telescope UVES Ultraviolet and Visual Echelle Spectrograph VLT Very Large Telescope 13

Chapter 1 Introduction 1.1 Motivation Metal-poor stars record the history of the Universe. They are the oldest observable stars and they make up Milky Way globular clusters. If we have access to the primordial composition of the clusters, we may improve our knowledge about their formation and their evolution. By determining the chemical abundance of their most recent generation of stars, we can have a better understanding of the origin of the elements and the chemical enrichment of the Universe. 1.2 Definitions The purpose of this section is to give relevant definitions which will help understanding the other parts. 1.2.1 Abundance The abundance of an element can be found in two forms in the literature: the ɛ notation and the square-bracket notation. In the ɛ notation, the abundance of an element is expressed relative to 10 12 hydrogen atoms. Therefore, given an element A: 14

CHAPTER 1. INTRODUCTION 15 log 10 ɛ(a) = log 10 (N A /N H ) + 12, where N A is the number of atoms of the element A. In the square-bracket notation, the abundance is expressed relative to that of the Sun. For two elements A and B, we then have: [A/B] = log 10 (N A /N B ) star log 10 (N A /N B ). If we select iron as element A and hydrogen as element B, the equation is: [F e/h] = log 10 (N F e /N H ) star log 10 (N F e /N H ). The quantity [Fe/H] is often used as an easily observed substitute for the metallicity 1. 1.2.2 Stellar populations The historical definition of the three different stellar populations is linked to the metallicity. Population I stars are young, metal-rich stars that can be found in the disk of our Galaxy. For instance, the Sun is a Population I star. Population II stars are old, metal-poor stars, mainly found in the halo of galaxies, dwarf spheroidal galaxies and globular clusters. Population III stars refers to hypothetical metal-free stars that would be the first generation of stars. They would have formed from the primordial gas constituted of hydrogen, helium and traces of lithium. These definitions are now outdated, though still in general use, but it has been realized since that a wide range of stellar ages and metallicities are present in the cosmos. Beers & Christlieb (2005) propose a nomenclature for stars of different metallicities. Particularly, they define metal-poor stars as stars having a metallicity less than 1.0. If [Fe/H] < 2.0 the star is called very metal-poor. This continues until [Fe/H] < 5.0 and the hyper-metal-poor stars 2. These stars are studied because they are the oldest stars observed and they record the heavy element abundances produced in the first generation of stars, so if we can measure their chemical abundance and compare it with predictions from stellar evolution we will have a better understanding of the formation and evolution of previous generations of stars. The 1 It must be noted that the evolution of iron may have been different from the evolution of other metals, resulting in iron-poor-stars not necessarily lacking in other metals. 2 These hyper-metal poor stars reveal substantial enhancement of C, N and O (Tumlinson 2007).

CHAPTER 1. INTRODUCTION 16 most metal-poor star which has so far been observed has a metallicity of 5.6 (Frebel et al. 2005). 1.3 The chemical evolution of the Universe This project is about measuring the chemical abundance of stars. To be able to understand the distribution of elements that will be obtained, we need to understand how the elements formed and how the Universe has been enriched. If we look at the mass fraction of the metals (i.e. elements from Li to U) in our Galaxy, we see that it evolved from 2 10 9 (after the Big Bang nucleosynthesis, only Li and Be are found) to 0.02 today (Frebel & Norris 2013). If we look at stars of different metallicities we can have a quick overview of the chemical evolution of stars (e.g. Figure 1.1). Figure 1.1: Comparison of spectra between stars having different metallicity (Aoki et al. 2006): the top spectrum belongs to the Sun ([Fe/H] = 0); the medium spectrum belongs to a very metal-poor star ([Fe/H] = 3.2); the bottom spectrum belongs to the most metal-poor star observed ([Fe/H] = 5.6). Note that the most metal-poor one has only one detectable absorption line in comparison to the multiplicity of absorption lines we can find in the spectrum of the Sun. This shows how different population of stars have been enriched.

CHAPTER 1. INTRODUCTION 17 1.3.1 Formation of the first stars The first elements created were the lightest ones: hydrogen, helium and lithium. They were formed during the Big Bang nucleosynthesis (or primordial nucleosynthesis) that occurred a few seconds after the Big Bang. 7 Be was also formed during that stage, but it fully decayed due to its instability before the first stars formed. The Big Bang theory predicts that the primordial gas clouds contained 75% of H, 25 % of He and 2 10 9 of Li (by mass; Frebel & Norris 2013). As stars form from the gravitational collapse of clouds of gas, the first generation of stars (population III stars) must have the same composition as the primordial gas clouds. The relative instability of nuclei with atomic masses 5 8 prevented the formation of heavier elements through the primordial nucleosynthesis. These elements are synthesised in the core of stars thanks to nuclear fusion. This process, called stellar nucleosynthesis, is related to the stellar evolution. Mass loss from stars returns nuclear-processed material to the interstellar medium (ISM) where it can form metalrich stars. This mass loss is determined by stellar evolution, which is described for a solar mass star in the next Section. 1.3.2 Low-mass stellar evolution From the beginning of its life, up to the main-sequence turn-off point, the Sunlike star burns hydrogen in its core to form helium via the proton-proton chain and releases energy (Iben 1967). When hydrogen is exhausted in the core of the star, the equilibrium between the outward pressure (which resulted from the energy release) and gravity is broken. This leads to a contraction of the core (turn off point) from where the star moves onto the Red Giant Branch (RGB) where hydrogen burns in an envelope (to form helium) around an inert helium core (see Figure 1.2). As the envelope expands and cools it becomes convective. The products of hydrogen fusion are brought to the surface during this stage called first dredge-up. But this phenomenon alone cannot explain the inhomogeneities found in carbon

CHAPTER 1. INTRODUCTION 18 Figure 1.2: HR Diagram of a Sun-like star. The stars presented in this thesis are slightly lighter, but their evolution is qualitatively the same as Sun-like stars. This picture was taken from http://nothingnerdy.wikispaces.com/file/view/ stellar_evolution_on_hr.png. abundances for RGB stars in metal-poor clusters. Indeed, Bell et al. (1979) found that the carbon abundance was a decreasing function of the luminosity in M92. Suntzeff (1981) confirmed this trend for M3 and M13. It implies that some other form of mixing occurs within RGB stars, called deep mixing or extra-mixing. Thermohaline mixing (Angelou et al. 2010) and rotational-induced mixing (Charbonnel & Lagarde 2010) are currently favoured but the process is not fully understood yet. The Sun-like stellar evolution continues as the helium formed in the envelope accumulates at the core, and the core heats because of gravity (constriction). When the temperature reaches approximatively 10 8 K, helium burning can start. Helium then transforms into carbon. This reaction is called the triple-alpha process because it requires three helium atoms to form one atom of carbon. Oxygen can also be formed if four alpha particles fuse. As the star goes down to the horizontal branch (HB), helium keeps forming carbon in the core whereas hydrogen keeps forming helium in the envelope (Iben 1974). When the helium core is exhausted the core contracts and the star enters the Asymptotic Giant Branch (AGB). This is where the abundance of a star changes the most (if it is massive enough). The core is

CHAPTER 1. INTRODUCTION 19 formed by carbon and oxygen, surrounded by a helium burning and a hydrogen burning envelope (Iben & Renzini 1983). If the star is not massive enough (M star < 8M sun ) the core cannot burn and form heavier elements. Another mixing known as third dredge-up occurs during the AGB stage of a low-mass star: as the helium burning shell is highly thermally unstable (due to the fusion rate being strongly dependent on the temperature) it undergoes thermal pulses every 10 4 10 5 years, which allow elements from the intershell region (between the hydrogen and helium burning shells) to mix into the outer envelope. Products of He burning appear near the stellar surface. The outer envelope gets ejected through stellar winds, driven by radial pulsations of the stars mantle on yearly timescales. The C-O core and the envelope separate. The ejected envelope is ionised by the revealed stellar core and becomes a Planetary Nebula. The C-O core is inert and fusion no longer provides energy: the remnant radiates the leftover energy and cools, ending its life as a white dwarf. 1.3.3 High-mass stellar evolution In the case of a massive star, successive steps of exhaustion and contraction occur, stopping with an iron-group core (Fe,Co,Ni), surrounded by lighter elements up to the external envelope made of hydrogen. Elements up to nickel (Z=28) can be formed thanks to the alpha process. However, the iron-group core cannot form heavier elements because the further reactions are endothermic (they require net energy to occur). The elements formed during the lifetime of the high-mass star will be returned to the ISM through supernova explosions. Heavier elements are formed thanks to neutron capture processes, mainly the s-process and the r-process (Burbidge et al. 1957, Cameron 1957). The s-process is a slow neutron capture process (it takes 100 to 10 5 years to capture neutrons). The s-process contains three components: the main s-process which produces elements from Sr up to Pb during the AGB phase of low and intermediate mass stars.

CHAPTER 1. INTRODUCTION 20 the weak s-process which produces elements from Fe up to Sr between the helium burning core and the carbon burning shell in massive stars. the strong s-process which produces Pb and Bi. As the reaction is very slow, the new nucleus will β-decay if it is unstable and form a heavier element (it wins a proton and loses a neutron). The s-process can build elements up to bismuth, but heavier elements are unstable and they only form thanks to the r-process. The r-process is a very quick (rapid) process that occurs when massive stars die as supernovae. It only takes a fraction of second to capture multiple neutrons hence the new nuclei do not β-decay immediately which enables the formation of unstable atoms up to uranium. These heavier elements will then enter the ISM through stellar winds (for s-process elements) or as supernova remnants. All the elements built during the lifetime of a star will enrich the ISM and then allow the next generations of stars to be more metal-rich, as we see with our Sun. 1.4 Globular Clusters 1.4.1 Overview Galactic globular clusters are compact aggregations of many thousands of stars (10 4 10 7 ). They are mainly found in the Galactic halo and contain very old (population II) stars. The stars within a globular cluster have approximately the same age, therefore they make useful places to test theories of stellar evolution. The first observation of a globular cluster was made by Abraham Ihle in 1665, of M22 in Sagittarius (Monaco et al. 2004). Ihle referred to it as a nebula. In 1677, Edmond Halley discovered Omega Centauri (NGC 5139). He also discovered M13 (Halley 1714). The first to use the term globular cluster was Herschel in his second catalogue of 1000 deep sky objects (Herschel 1789). At that time 70 globular clusters were known. Nowadays we know about 157 clusters (Harris 2010). The metallicities of globular clusters

CHAPTER 1. INTRODUCTION 21 range from [Fe/H] = 2.37 to 0.10. One of the first applications of globular clusters was the determination of the Galactic Center (Shapley 1918). He discovered that the Sun was not located in the Galactic Center, in opposition with what was believed until then, but quite far from it. Globular clusters have been well studied since then because they provide information about the earliest times of the Universe. 1.4.2 Abundance studies There are two commonly used ways to derive the chemical abundance of stars from their spectra: the equivalent width measurement and the spectral synthesis. The equivalent width method consists in measuring the area of a spectral absorption line by fitting a Gaussian (or Voigt) profile. Spectral synthesis consists in generating a spectrum via solving the equations of radiation transfer through a model of the star s atmosphere. The user tries to match the generated spectrum with the observed one. Usually astronomers use the equivalent width measurement, but sometimes the shape of the spectrum is too complex (e.g. CN contamination between Al lines, Johnson et al. 2008) and spectral synthesis is required. Curve of growth methods are used to derive the abundance of an element from equivalent width measurements. It was first believed that all the stars in a same globular cluster would have the same abundance. However, studies of abundances within globular clusters have shown many peculiarities (Cohen 1978, Smith 1987, Suntzeff 1993): most of them show that there is a star-to-star variation in some particular elements (light elements) within clusters, the most famous one being the Na-O anticorrelation (see Figure 1.3). One of the first detections of star-to-star variation was made by Cohen in 1978. She derived the abundance of 20 elements in five red giants in M13 and three in M3. The results showed Na and Ca abundance differences between stars of the same cluster. Norris et al. (1981) detected Al inhomogeneity among 69 giant stars in NGC 6752. They also observed CN-weak and CN-strong stars, and proposed to

CHAPTER 1. INTRODUCTION 22 Figure 1.3: An example from Ivans et al. (2001) of Na-O anticorrelation in different clusters (M3, M4, M5, M10, M13, M71). We can clearly see that when [Na/Fe] increases [O/Fe] decreases. split the cluster in two groups of stars. Those stars would have different origins, they could have been different population of stars, but they could also have formed at the same time from a different part of the same gas cloud. Drake et al. (1992) measured abundances for Fe, Ca, Na, Al and O in M4, a cluster known for its CN bimodality (it contains CN-weak and CN-strong stars). They studied four giant stars, two CN-strong and two CN-weak. They found an anticorrelation between Na and O but also between Al and O: the CN-strong stars are rich in Na and Al whereas the CN-weak stars are rich in O. They found no variations between Fe and Ca abundances. Two scenarios can have been put forward to explain this anticorrelation: (Denissenkov et al. (1998) proposed a combination of the two scenarios): the first is a primordial scenario, in which the difference may come from the enrichment of the gas cloud by a previous generation of stars (mass loss from a medium-mass AGB star or massive supernovae). The anticorrelation would exist throughout the star s life. In the second scenario, the difference comes from the deep mixing that occurs during the RGB stage, allowing the star to produce more Na thanks to proton

CHAPTER 1. INTRODUCTION 23 capture reaction. This is called the evolutionary scenario. Gratton et al. (2001) obtained the chemical composition of 14 dwarfs and 12 red giants in NGC 6397 and NGC 6752. They were the first to find an Na-O anticorrelation among main sequence stars. This observation refuted the assumption that Na-O anticorrelation was only due to deep mixing. Indeed, main sequence stars should not go through mixing because that would increase the availability of hydrogen in the core, lengthening the star s main-sequence lifetime beyond what is observed. Thus they try to explain the Na-O anticorrelation by mass loss during the AGB phase of previous stars that would have changed the primordial composition of a second generation of stars. They also make a second hypothesis, related to this one, claiming that the death of a previous star as a planetary nebula would have enriched the interstellar gas from which new stars are formed. In both cases, the presence of multiple generations of stars is identified. Carretta et al. (2004) found the same results for a different cluster: 47 Tuc (NGC 104), giving more support to the premise. More recently, Carretta et al. (2009) studied 202 red giants in 17 globular clusters. This study revealed that the Na-O anticorrelation occurred in all clusters. We may generalize and say that the Na-O anticorrelation is found in all the globular clusters (no exception has been found so far). The Na-O anticorrelation is not the only chemical peculiarity in globular clusters. Shetrone (1996) studied red giants in M13 and halo field stars and found that the abundances of Mg and Al were anticorrelated. Gratton et al. (2001) confirmed this anticorrelation. However, it does not seem to be present in all clusters. The same question about its origin arose, and the same scenarios were presented. The presence of multiple populations of stars was a logical consequence of these abundance variations, but there was still no observation of that phenomenon. It was finally observed (mainly thanks to the progress made in the resolution and the precision of photometric measurements), as Bedin et al. (2004) discovered the double main sequence of ω Centauri (see Figure 1.4).

CHAPTER 1. INTRODUCTION 24 Figure 1.4: Colour Magnitude Diagram of the double main sequence in ω Centauri from the work of Bedin et al. (2004). We can see the two different sequences from the red and blue dots, which correspond to the stars followed in Bedin s study. The ACS (Advanced Camera for Surveys), aboard the Hubble Space Telescope (HST), results proved that the blue main sequence was actually more metal-rich than the red main sequence and that the red main sequence was the one with the most stars. This colour difference is the opposite of what the models of stellar evolution expect. In order to explain this blue main sequence, Bedin et al. proposed an enhanced value of the He abundance. Norris (2004) calculated the enhancement: Y 0.15, where Y is the abundance of He. The origin of this enhancement is not clear at the moment. It could be the mass loss by the first generation of AGB medium-mass stars. An accurate measurement of helium abundance would clarify the problem. But it is not an easy task because cool stars do not have strong He lines. Hot stars cannot be used either because the spectrum is affected by diffusion. Dupree & Avrett (2013) used a transition in He i in the near-infrared (at 1.08 µm) for two red giant stars in ω Centauri. Using a semi-empirical model, they obtained Y = 0.39-0.44 and Y=

CHAPTER 1. INTRODUCTION 25 0.22 for the two stars, deducing an enhancement of Y 0.17 for one star (which also has enhanced aluminium and magnesium). This result can give a quantitative approach concerning the enrichment in clusters but it must be noted that ω Centauri has an exceptionally high helium enhancement. Valcarce et al. (2014) found that Y is typically 0.01 in M4. ω Centauri is not the only cluster where multiple main sequences and other abundance variations have been found. Piotto et al. (2007) discovered a triple main sequence in NGC 2808 based on the ACS on HST. Milone et al. (2012) also found a double main sequence in NGC 6397. 1.5 NGC 362 NGC 362 is a globular cluster located in the constellation Tucana. It was discovered by James Dunlop on August 1, 1826. Different attempts have been made to determine this cluster s age. Dotter et al. (2010) used isochrone fitting from the ACS data and found that it was 11.50 ± 0.50 Gyr. More recently, VandenBerg et al. (2013) derived the age of 55 globular clusters using ACS public photometry. They showed that the age of NGC 362 is 10.75 ± 0.25 Gyr. The metallicity of NGC 362 has been derived by multiple authors. A summary of their results is given in Table 1.1. Table 1.1: Iron abundance studies of NGC 362 Author [Fe/H] Pilachowski et. al (1983) 0.9 Gratton (1987b) 1.2 Caldwell & Dickens (1988) 1.05 ± 0.10 Shetrone & Keane (2000) 1.33 ± 0.01 Kraft & Ivans (2003) 1.31 ± 0.03 Székely et al. (2007) 1.16 ± 0.25 Harris (2010) 1.26 Carretta et al. (2013) 1.171 ± 0.009

CHAPTER 1. INTRODUCTION 26 1.5.1 The second parameter problem NGC 362 is a well studied cluster, mainly because it forms a second parameter pair (maybe the most famous one) with another globular cluster: NGC 288. That means NGC 288 has a blue horizontal branch in the colour-magnitude diagram whereas NGC 362 has a red one (see Figure 1.5), despite their relatively close metallicity: [Fe/H] = 1.39 ± 0.01 dex and [Fe/H] = 1.33 ± 0.01 dex respectively for NGC 288 and NGC 362 (Shetrone & Keane 2000; more recent studies have been performed but Shetrone & Keane derived metallicities for both clusters). Figure 1.5: Colour-magnitude diagrams for NGC 288 (left) and NGC 362 (right), taken from the ESO MAD Science Demonstration Proposal. Note the concentration of stars on the left side of the HB (blue HB) for NGC 288 and on the right side of the HB (red HB) for NGC 362. It has been understood, since Sandage & Wallerstein (1960), that variations in the chemical composition of clusters lead to different horizontal branch stars: they are mainly on the red side in metal-rich globular clusters and on the blue side in metal-poor clusters. However, Faulkner (1966) found an exception: M13 (NGC 6205), considered as a metal-rich cluster, appears to have a blue horizontal

CHAPTER 1. INTRODUCTION 27 branch. This was proof that metallicity could not explain alone the morphology of the horizontal branch and it was the first allusion to the complex second parameter problem. Sandage & Wildey (1967) clearly stated that problem: at least one other parameter besides the metal abundance controls the distribution of stars along the horizontal branch. Many scenarios have been proposed over the past 50 years, but the problem remains only partially solved. The first hypotheses are found in the literature of the late 1960 s. The helium abundance seems to have been the first candidate (Faulkner 1966). Rood & Iben (1968) suggested that the age could be the second parameter. Simoda & Iben (1970) mentioned the abundances of the CNO elements (Z CNO ). Other propositions have been made since then, such as stellar rotation (Freeman & Norris 1981), mass loss (Shetrone & Keane 2000, Catelan 2000) and helium mixing (Catelan 2000). These explanations were accepted and challenged (e.g Pilachowski et al. (1983) showed that Z CNO could not be the second parameter because of the Na-O anticorrelation). Even if the age of the cluster seems to be accepted as a parameter, it does not solve the whole problem. More recently, Gratton et al. (2010) stated that the main reason why this is still an open problem is the existence of more than one second parameter. They recognised that the age is the second parameter and suggested variations of the abundance of helium to be the third parameter. Dotter et al. (2010), following the propositions of Freeman & Norris (1981) and Fusi Pecci & Bellazzini (1997), suggested that two parameters must be taken into account: one global parameter which is different among different clusters such as the cluster s age, and another non-global parameter which varies within a globular cluster such as the internal pollution.

CHAPTER 1. INTRODUCTION 28 1.5.2 Previous studies NGC 362 has been studied for more than 30 years. Previous works have been summarised in Table 1.2. Table 1.2: Abundance studies of NGC 362 Element PSW a (1983) G b (1987b) SK c (2000) WC d (2010) CAG e (2013) [Fe/H] 0.87 ± 0.2 1.18 ± 0.04 1.33 ± 0.01 1.21 ± 0.09 1.17 ± 0.05 [O/Fe] 0.36 ± 0.4 0.04 ± 0.13 0.89 ± 0.18 [Na/Fe] 0.36 ± 0.3 0.08 ± 0.01 0.04 ± 0.15 0.11 ± 0.25 [Mg/Fe] 0.28 0.36 ± 0.05 0.33 ± 0.04 [Al/Fe] 0.31 ± 0.12 0.24 ± 0.19 [Si/Fe] 0.09 ± 0.07 0.36 ± 0.05 0.26 ± 0.04 [Ca/Fe] 0.65 ± 0.3 0.21 ± 0.12 0.18 ± 0.02 0.34 ± 0.02 [Sc/Fe] 0.07 ± 0.2 0.09 ± 0.07 0.07 ± 0.04 [Ti/Fe] 0.30 ± 0.3 0.30 ± 0.09 0.30 ± 0.05 0.16 ± 0.03 [V/Fe] 0.15 ± 0.2 0.19 ± 0.06 0.04 ± 0.01 0.05 ± 0.03 [Cr/Fe] 0.25 ± 0.4 0.03 ± 0.04 [Ni/Fe] 0.15 ± 0.3 0.12 ± 0.19 0.07 ± 0.03 0.09 ± 0.04 [Cu/Fe] 0.20 0.50 ± 0.12 [Y/Fe] 0.30 ± 0.12 0.07 ± 0.11 [Zr/Fe] 0.34 ± 0.12 0.50 ± 0.12 [Ba/Fe] 0.30 ± 0.4 0.17 ± 0.23 0.28 ± 0.13 0.56 ± 0.30 0.18 ± 0.21 [La/Fe] 0.30 ± 0.3 0.36 ± 0.12 0.36 ± 0.12 0.33 ± 0.09 [Eu/Fe] 0.57 ± 0.06 0.78 ± 0.05 0.70 ± 0.07 Stars 3 1 12 13 92 a Pilachowski et al. (1983) b Gratton (1987b) c Shetrone & Keane (2000) d Worley & Cottrell (2010) e Carretta et al. (2013) One of the first abundance studies of NGC 362 was carried out by Pilachowski et al. (1983). They studied three stars in this cluster and the abundance of 14 elements from oxygen to lanthanum. Although their sample was restricted, they detected the Na-O anticorrelation (their Table 7D): even though they found that the average [O/H] and [Na/H] values are the same, we can see that when the oxygen to hydrogen ratio decreases the sodium to hydrogen ratio increases. They detected an overabundance of alpha-elements and also an overabundance of oxygen, confirming

CHAPTER 1. INTRODUCTION 29 that NGC 362 is composed of oxygen-rich stars. This sample also allows comparisons with later studies such as Gratton (1987a, 1987b). His study contains eight clusters including NGC 362. He obtains the abundance of ten elements from oxygen to barium of one star in NGC 362 (I-23). He finds that his results concerning the clusters they had in common do not correspond to Pilachowski s ones: he derives metallicities 0.30 dex lower. Gratton proposes substituting Pilachowski s results with his results because of the use of more accurate detector technology (photographic spectrographs versus CCD cameras). Unfortunately the comparison with this study will be very inaccurate because it is only composed of one star. A wider study has been carried out by Shetrone & Keane (2000) who tried to compare red giants between NGC 362 and NGC 288. They published abundances of 13 elements in 12 giant stars (their Table 4). They found an Na-O anticorrelation (like Pilachowski et al. 1983), but also an Al-O anticorrelation. The comparison with Pilachowski s results reveals differences of 0.4 dex in abundances. They also argue that the difference comes from the method of observation (specifically the difficulty of working with image tube spectra recorded on glass plates ). The comparison with Gratton s results also reveals differences of about 0.1 dex in abundances. As the resolution and signal-to-noise ratio are higher for their sample, their measurements should be more accurate. Worley & Cottrell (2010) determined the abundance of heavy elements of 13 giant stars in NGC 362. They compared barium, lanthanum and europium with previous studies (e.g Pilachowski et al. (1983), Gratton 1987b, Shetrone & Keane 2000). They also derived the abundances of strontium, yttrium, zirconium and neodymium (their Table 15). Their abundance of barium was not very accurate due to sensitivity to microturbulence. The difference with all previous studies is therefore quite significant. The lanthanum abundance was only available in Pilachowski s study. The values correspond although Pilachowski studied more metal-rich stars. The europium abundance was only available in the study by Shetrone & Keane. The europium abundance is 0.2 dex higher in Worley s study than Shetrone s one,

CHAPTER 1. INTRODUCTION 30 stated to be due to the lack of hyperfine correction in Worley s study. Recently, Carretta et al. (2013) measured the abundance of 21 elements for 92 red giant stars. They confirmed the observations of Shetrone & Keane (2000) and found that the abundance of neutron capture elements (heavy elements) is constant throughout the study, proving that these elements come from a previous generation of stars. They considered two generation of stars. They derived abundances using a different method than this study, which in the end was more accurate (it will be discussed in Section 5.3). The results obtained by Carretta s study will serve as an important comparison to this study because they used more precise instruments than the previous studies. Indeed, multi-object spectrographs saves a lot of time, as they can point up to 130 stars at a time (FLAMES). New telescopes, such as the 8.2m telescope used by Carretta and this study, also save substantial time: it would take at least four times as long to observe a star with a 4m telescope. Fifteen years ago multi-object telescopes did not exist, astrophysicists would have to point each star at a time. It would have taken at least four month s worth of telescope time to do the observations that Carretta did, which is impossible to ask for. 1.6 Overview of this thesis This research project aims to determine the fundamental parameters of stars and abundances of their elements from medium-resolution spectroscopy using FLAMES, one of the VLT spectrographs. Chapter 2 describes the software and methods used to obtain the iron abundance of the stars. Chapter 3 details the results for all the elements and compares them with previous studies. Chapter 4 presents how the error calculations were carried out and also reveals the impact of a change in stars parameters on the abundance ratios. In Chapter 5 the results from the previous chapters are discussed. Chapter 6 summarises the conclusions of this work.

Chapter 2 Data Processing This study contained 120 HB, AGB and RGB stars, observed with the Fibre Large Array Multi Element Spectrograph (FLAMES) and the Ultraviolet and Visual Echelle Spectrograph (UVES). FLAMES and UVES are intermediate and high-resolution, multi-objects, fibre-fed spectrographs mounted on the Nasmyth A and Nasmyth B (respectively) foci of the VLT Unit Telescope 2 (UT2). The Hertzsprung Russell Diagram of NGC 362 is shown in Figure 2.1. S/N varied between stars because of brightness differences. It ranged from 16 to 243 and the average value was 112. Stars were selected from the photometrically derived Hertzsprung Russell diagram presented in Boyer et al. (2009). AGB stars were specifically targeted but comprise a relatively small proportion of the sample. Spare MEDUSA / UVES fibres were used on the most luminous giant stars (MEDUSA is the multi-object spectroscopy mode on UVES). The HR13, HR14A and HR15 gratings were used during the observations. High resolution (R 20 000) was necessary to be able to deblend the spectral lines when needed. The total observation time was approximately 5.4 hours. The log of the observations is given in Table 2.1. Observations in the HR13 grating (6100 6400Å) contained most of the lines (35 out of 79 Fe lines and six out of 10 lines from the other elements: Na, Si, Ca and Ti). 31

CHAPTER 2. DATA PROCESSING 32 Figure 2.1: Hertzsprung Russell Diagram of NGC 362. abundances were derived in this study are plotted in blue. The 56 stars for which Observations in the HR14A grating (6380 6600Å) contained 34 Fe lines, one Ca line and two Ti lines. Observations in the HR15 grating (6600 6950Å) and contained 10 Fe lines and one Si line. An image of NGC 362 is given in Figure 2.2. The field of view is 37x25. For comparison, FLAMES has a 25 field of view in diameter. The pixel size in this picture (1.4 ) is similar to the MEDUSA fibre diameter (1.2 ), therefore this picture highlights the blending in the centre of the cluster (which will be discussed in Section 5.2) as it can be the stars are overlapping.

CHAPTER 2. DATA PROCESSING 33 Table 2.1: Log of FLAMES observations for NGC 362 Date UT Exposure (sec) Grating Airmass 2011-10-30 01:45:57 1200 HR15 1.488 2011-10-30 02:06:51 1200 HR15 1.468 2011-10-30 02:27:44 1200 HR15 1.454 2011-10-30 04:13:39 1420 HR14A 1.464 2011-10-30 04:38:12 1420 HR14A 1.486 2011-10-30 05:02:45 1420 HR14A 1.516 2011-12-21 00:49:40 2000 HR13 1.464 2011-12-21 01:30:37 2000 HR13 1.505 Figure 2.2: Image of NGC 362. This picture was taken from http://www.verschatse.cl/clusters/ngc362/details.htm.

CHAPTER 2. DATA PROCESSING 34 The line list was adapted from Johnson & Pilachowski (2010) and reproduced in Table 2.2. The third column gives the excitation potential (E.P.) and the fourth column gives the weighted oscillator strength for the lines of the selected elements. Table 2.2: Line List Ion Wavelength E.P. log(gf) (Å) (ev) (dex) Fe I 6120.25 0.92 5.860 Fe I 6151.62 2.18 3.309 Fe I 6157.73 4.08 1.170 Fe I 6159.37 4.61 1.840 Fe I 6165.36 4.14 1.474 Fe I 6173.34 2.22 2.820 Fe I 6180.20 2.72 2.726 Fe I 6187.40 2.83 3.988 Fe I 6187.99 3.94 1.570 Fe I 6200.31 2.61 2.317 Fe I 6207.23 4.99 1.989 Fe I 6213.43 2.22 2.532 Fe I 6219.28 2.20 2.333 Fe I 6226.74 3.88 2.050 Fe I 6229.23 2.85 2.805 Fe I 6232.64 3.65 1.103 Fe I 6240.65 2.22 3.250 Fe I 6246.32 3.60 0.643 Fe I 6252.56 2.40 1.717 Fe I 6265.13 2.18 2.460 Fe I 6270.23 2.86 2.544 Fe I 6290.54 2.59 4.300

CHAPTER 2. DATA PROCESSING 35 Table 2.2 Continued Ion Wavelength E.P. log(gf) (Å) (ev) (dex) Fe I 6290.97 4.73 0.504 Fe I 6297.79 2.22 2.670 Fe I 6301.50 3.65 0.588 Fe I 6302.49 3.69 1.083 Fe I 6330.85 4.73 1.200 Fe I 6335.33 2.20 2.237 Fe I 6336.82 3.69 0.696 Fe I 6355.03 2.85 2.340 Fe I 6380.74 4.19 1.326 Fe I 6385.72 4.73 1.850 Fe I 6392.54 2.28 3.940 Fe I 6393.60 2.43 1.562 Fe I 6400.00 3.60 0.470 Fe I 6400.32 0.92 4.178 Fe I 6408.02 3.69 1.128 Fe I 6411.65 3.65 0.755 Fe I 6412.20 2.45 5.063 Fe I 6419.64 3.94 2.580 Fe I 6419.95 4.73 0.280 Fe I 6430.85 2.18 1.886 Fe I 6436.41 4.19 2.410 Fe I 6469.19 4.83 0.260 Fe I 6475.62 2.56 2.832 Fe I 6481.87 2.28 2.934 Fe I 6483.94 1.49 5.638

CHAPTER 2. DATA PROCESSING 36 Table 2.2 Continued Ion Wavelength E.P. log(gf) (Å) (ev) (dex) Fe I 6494.50 4.73 1.176 Fe I 6494.98 2.40 1.313 Fe I 6495.74 4.83 1.060 Fe I 6496.47 4.79 0.650 Fe I 6498.94 0.96 4.489 Fe I 6518.37 2.83 2.620 Fe I 6533.93 4.56 1.380 Fe I 6546.24 2.76 1.556 Fe I 6551.68 0.99 5.970 Fe I 6556.79 4.80 1.638 Fe I 6569.21 4.73 0.350 Fe I 6574.23 0.99 4.923 Fe I 6581.21 1.49 4.789 Fe I 6592.91 2.73 1.603 Fe I 6593.87 2.43 2.342 Fe I 6597.56 4.79 1.040 Fe I 6609.11 2.56 2.632 Fe I 6646.93 2.61 3.990 Fe I 6677.98 2.69 1.418 Fe I 6703.57 2.76 3.160 Fe I 6705.10 4.61 1.392 Fe I 6750.15 2.42 2.621 Fe I 6810.26 4.61 0.986 Fe I 6858.15 4.61 0.930 Fe I 6861.94 2.42 3.890

CHAPTER 2. DATA PROCESSING 37 Table 2.2 Continued Ion Wavelength E.P. log(gf) (Å) (ev) (dex) Fe I 6916.68 4.15 1.450 Fe II 6149.26 3.89 2.711 Fe II 6238.39 3.89 2.434 Fe II 6247.56 3.89 2.315 Fe II 6416.92 3.89 2.447 Fe II 6432.68 2.89 3.587 Fe II 6456.38 3.90 2.155 Na I 6154.23 2.10 1.560 Na I 6160.75 2.10 1.210 Si I 6155.13 5.62 0.774 Si I 6721.85 5.86 1.016 Ca I 6161.30 2.52 1.246 Ca I 6169.04 2.52 0.837 Ca I 6169.56 2.53 0.628 Ca I 6455.60 2.52 1.557 Ti I 6554.23 1.44 1.150 Ti I 6556.07 1.46 1.060

CHAPTER 2. DATA PROCESSING 38 2.1 Equivalent Width The equivalent width, W, characterizes the strength of a spectral line. It is defined as a rectangle included between the continuum (being taken as unity) and the zero intensity level, having the same area as the line profile (e.g. Figure 2.3). It is expressed in the same units as the abscissa (wavelength or frequency), in this work in Ångströms. The equivalent width can be calculated via the equation: W = λ2 λ 1 I 0 I λ I 0 dλ (2.1) where I 0 is the continuum intensity and I λ is the line intensity at the wavelength λ. This equation reveals that the continuum intensity must be determined very accurately, especially for weak lines, because it will greatly affect W. The equivalent width will be positive for an absorption line and negative for an emission line. We are only concerned in this work with absorption lines. Figure 2.3: Illustration of the equivalent width by Dyson & Williams (1997). The abscissa is expressed in frequency here. I ν is the intensity of the line at the frequency ν.

CHAPTER 2. DATA PROCESSING 39 2.2 Equivalent Width measurement The abundances have been derived by comparing the equivalent widths of individual spectral lines to stellar atmosphere models. Two programmes have been used. The first program is an updated version of the software used by Johnson et al. (2008). This FORTRAN program goes through the input line list and stellar spectrum and allows the user to measure individual equivalent widths (EWs) by fitting a Gaussian profile or multiple Gaussian profiles if deblending is needed (Figure 2.6) to the given absorption lines (see Figure 2.4). A second spectrum can be used as a guide to uncatalogued spectral lines. We used the high resolution Arcturus spectrum (Hinkle et al. 2000). Figure 2.4: Interface of the first program. The black spectrum is the spectrum of the star being analysed; the blue spectrum is the reference star: Arcturus (Hinkle et al. 2000); the purple curve is the Gaussian fit matching the star spectrum; the green line is the continuum. Elements are shown to help identification in case of blending. Some preliminary setup is required before proceeding with the EW measurement. The first step is to remove the Doppler shift in the spectrum. NGC 362 has a significant heliocentric radial velocity, v r = 223.5 km/s (Harris 2010). Consequently, it introduces a Doppler shift of around 5Å. As the exact shift differs from star to

CHAPTER 2. DATA PROCESSING 40 star, the target spectrum is cross-correlated with the Arcturus spectrum to remove this shift. This process also allows identification of Galactic stars which are not part of NGC 362: these stars have a different radial velocity and need a different shift to match the Arcturus spectrum 1. Seven Galactic stars were identified this way and are reported in Table 2.3. Radial velocities were not measured directly for these stars as the visual shift in the spectrum was obvious. Table 2.3: List of Galactic Stars Star RA (J2000) DEC (J2000) 671 01:03:52.87 70:55:17.0 665 01:03:50.65 70:55:35.0 683 01:04:07.06 70:50:59.0 185 01:03:04.77 70:54:49.0 645 01:03:40.76 70:49:11.0 689 01:04:10.30 70:52:14.0 448 01:03:19.50 70:56:03.0 Once this is done the continuum has to be defined. Continuum placement can substantially affect the EW derived for weak lines, but strong lines are relatively unaffected. A major challenge is the determination of where it should be set. This involves deciding whether small-scale variations in the data are treated as noise or spectral features. The first idea was to set the continuum level locally, evenly separating the white noise. Indeed, the white noise has a random distribution therefore it should be equally distributed around the continuum level. But the dissociation between the noise and the weak spectral features was not evident. After a lengthy period of trial and error, and detailed discussions with C. I. Johnson, it was decided that the majority of weak spectral features were unresolved lines, and that the continuum should be placed on top of these (Figure 2.5). 1 Contamination from the SMC can be ruled out due to the difference in distance between NGC 362 and the SMC. NGC 362 is 8.6 kpc from Earth while the SMC is about 60 kpc, seven times farther. The mean luminosity of the stars in this study is 70, meaning that a star in the SMC would have to be 3500 L in order to mimic a 70 L star. Only a very few, short-lived post-agb can attain this luminosity at the temperature of the stars we have examined. A star in the SMC would also have a lower surface gravity since g MT 4 L.