THE UNIVERSITYOF NEW SOUTH WALES

DOCTORAL THESIS

Observational aspects of globular cluster and halo stars in the GALAH survey

Student: Supervisor: Mohd Hafiz MOHD SAADON Associate Professor Sarah MARTELL

A thesis submitted in fulfillment of the requirements for the degree of

Doctor of Philosophy

in the

School of Physics Faculty of Science The University of New South Wales

12 March 2021 i ii iii iv

Declaration of Authorship

I, Mohd Hafiz MOHD SAADON, declare that this thesis titled, “Observational aspects of globular cluster and halo stars in the GALAH survey” and the work presented in it are my own. I confirm that:

• This work was done wholly or mainly while in candidature for a research degree at this University.

• Where any part of this thesis has previously been submitted for a degree or any other qualification at this University or any other institution, this has been clearly stated.

• Where I have consulted the published work of others, this is always clearly at- tributed.

• Where I have quoted from the work of others, the source is always given. With the exception of such quotations, this thesis is entirely my own work.

• I have acknowledged all main sources of help.

Signed:

Date: v

“In loving memory of my grandmothers, Ramlah and Halijah, who wished to see me this far but could not be here anymore.”

– your grandson.

“To Amani, my sweet little angel, this thesis is your sibling.”

– your father. vi

THE UNIVERSITY OF NEW SOUTH WALES

Abstract

Doctor of Philosophy

Observational aspects of globular cluster and halo stars in the GALAH survey

by Mohd Hafiz MOHD SAADON

This thesis is a study of the observational aspects of globular cluster and halo stars in the Galactic Archaeology with HERMES (GALAH) survey. GALAH is a major astronomical survey project that derives stellar parameters (Teff, log g, [Fe/H]) and up to 30 elemen- tal abundances for a very large sample of stars in the Milky Way from high-resolution spectroscopy. This thesis uses the third GALAH internal data release, which includes over 600 000 stars, to test the validity of GALAH’s reported quantities, especially for cool metal-poor globular clusters and halo stars. We also examine the light-element abun- dance patterns of stars in globular clusters, which have known peculiarities, to search for stars in the Galactic halo with globular cluster-like abundance patterns, and to chart the age-abundance relation in halo stars using elements that have been reported as particu- larly useful age indicators.

In this first application of GALAH data to globular clusters and halo chemodynamics, we explore the behaviour of 30 elemental abundances for 340 stars in four globular clusters, which are NGC 104, NGC 5139, NGC 6397 and NGC 7099, which we found inconsisten- cies in the light-element abundances. We also discover that approximately 1% (4/445) of metal-poor halo giants in the sample are likely globular cluster escapees with enhanced [Al/Fe] and depleted [Mg/Fe]. Finally, we chart the age-abundance relations of metal- poor halo giants and identify stars with likely extragalactic origins using their kinematics and abundances. vii

Acknowledgements Praise be to the Lord of this magnificent universe.

Before anything else, I would like to thank my funder for granting me the SLAB fel- lowship under the University of Malaya (UM) and the Ministry of Higher Education Malaysia so that I would be able to pursue my study.

Most importantly, no words can truly express how thankful I am to my wonderful su- pervisor and mentor, Associate Professor Sarah Martell, for her guidance, support and especially patience throughout my PhD, who have taught me about the Galactic archae- ology and make me enjoy the topic that still amazes me. She is always making time to read, edit, and comment on my works and tirelessly provide feedback whenever possi- ble. I am incredibly thankful that she took me under her wings when I was lost, desperate and had nowhere to go. I am also grateful for her wisdom, selfless time, positivity and passion that kept me moving even through the most challenging times. I pray that all good things will always be bestowed upon her.

I am also grateful to Dr Jeffrey Simpson, Nicholas Borsato, Kirsten Banks, Aditya Gu- dalur Balasubramaniam, and other UNSW Milky Way Group members, for their helpful insights and discussions during our weekly meetings. I would also like to thank Prof. Chris Tinney for helping me by overseeing my progress reviews. A special thanks to my postgraduate coordinators Assoc. Prof. Kim Vy-Tran and Prof. Richard Newbury, and also Prof. Michael Ashley and Prof. Jeremy Bailey. They became my panels for the progress review meetings during my candidature, and gave a lot of suggestions and en- couragement so that I can stay on track, be realistic on my milestones and finish my PhD. I want to thank other postgraduate students and the staff of School of Physics, GRS and UNSW for their assistance in the campus and faculty. For Johannes Böttcher, thank you for providing me this thesis template.

I also thank my former supervisors who welcomed me when I first arrived in Australia, Prof. Rob Wittenmyer and Assoc. Prof. Dennis Stello. I am also grateful to GALAH researchers and collaborators, especially Dr Sven Buder, Dr Sanjib Sharma, Jane Lin, and Prof. Martin Asplund for sharing their knowledge about the GALAH survey. My sincere thanks also go to Dr Chris Lidman, Dr Tayyaba Zafar and Dr Lee Spitler and other as- tronomers, engineers and technicians of AAT and AAO during my observation nights. I will always remember those Christmas and New Year’s Eve.

I am also indebted to Dr Muhammadin for his advice and guidance on being a sane human when I was so down, and his reminders and tips to be humble but wise. Another sincere thank to the alumni of the late Prof. Rahim for their motivational supports and viii prayers. Also not to be ignored, to the members of Facilitator Alumni Club KMJ, thank you for being my platform to have a laugh and gossip about our silly past.

Also, I am thankful to the Heads of Department of Fiqh and Usul UM, Dr Ridzwan, Dr Luqman and Assoc. Prof. Sa’adan; Directors of the Academy of Islamic Studies UM, Prof. Mohd Yakub @ Zulkifli and Prof. Raihanah; and to the program coordinator of Is- lamic Astronomy UM, Dr Sailful Nawawi and Dr Raihana. Thank you for offering me the fellowship and checking on me. Even not to be forgotten, to my comrades of House Sec- retaries of Malaysia Hall Sydney, Azmilnidzam, Yow Hun Yen, Nur Iman, Noor Alyssa, Muhammad Haris, Chng See Yi, Rabiatul Adawiyah, Lo Sin Kuang and Ezwan Shah. Thank you for your cooperation, your understanding, your eyes and ears, your energy and efforts. I also want to thank my housemate, Chng Zhi Yee; the Directors and the At- taches of Education Malaysia Australia; and the residents of Malaysia Hall. They make me feel like home during my time in Australia.

More recently, I also would like to thank Assoc. Prof Andrew Cole and Professor Renee James for being the thesis reviewers and giving a lot of helpful insights and recommen- dations to improve the thesis.

Finally, I want to express my heartfelt gratitude and love to my parents, Mohd Saadon and Siti Hajar; my wife, Nur Liyana Atikah; and my big family for their unconditional love and support (even financially) despite me being so far away. Thank you for being so patient and understanding during this whole journey. And to my dear little daughter, Nur Nasirah Amani, daddy is home. ix

Contents

Declaration of Authorship iv

Acknowledgements vii

List of Figures xii

List of Tables xviii

List of Abbreviations xix

List of Symbols xx

1 Introduction1 1.1 Galactic archaeology...... 1 1.1.1 Chemical tagging...... 2 1.1.2 Halo stars and halo streams...... 4 1.1.3 Globular clusters...... 6 1.1.4 Age-abundance relations and metal-poor stars...... 8 1.1.5 The [α/Fe] ratio...... 9 1.2 Large-scale surveys...... 11 1.2.1 Galactic Archaeology with HERMES...... 12 GALAH in comparison with other surveys...... 12 1.2.2 Gaia ...... 13 1.3 Stellar abundances...... 14 1.3.1 Metallicity...... 15 1.3.2 Spectroscopic analysis...... 16 1.4 Research objectives...... 19 1.5 Thesis structure...... 21

2 Globular clusters in GALAH 22 2.1 The role of globular clusters in Galactic archaeology...... 22 2.2 Stellar abundances in globular clusters...... 27 2.2.1 Light-element abundance variations...... 27 x

Mg-Al anticorrelation...... 29 2.2.2 Metallicity variations...... 29 2.2.3 Neutron-capture abundance variations...... 30 2.3 Globular clusters observed by the GALAH survey...... 32 2.3.1 Determining cluster membership...... 33 Position cut...... 33 Radial velocity cut...... 34 Proper motion cut...... 34 Parallax cut...... 34 Other criteria...... 37 2.4 Globular cluster abundances in the GALAH survey...... 40 2.4.1 Metallicity...... 42 2.4.2 Light elements...... 48 2.4.3 An extension to silicon...... 52 2.4.4 Neutron-capture elements...... 55 2.4.5 Lithium...... 56 2.4.6 The clusters in detail...... 62 NGC 104...... 62 NGC 5139...... 70 NGC 6397...... 77 NGC 7099...... 83 2.5 Summary and conclusion...... 89

3 Globular cluster stars in the halo 91 3.1 The role of halo stars in Galactic archaeology...... 91 3.1.1 Kinematics of halo stars...... 92 3.1.2 The chemical compositions of halo stars...... 93 3.2 Selecting halo stars in the GALAH data set...... 95 3.3 Abundance patterns of the GALAH halo giant stars...... 99 3.3.1 Al-Mg-Si variation...... 102 3.4 Globular cluster escapees in the halo from GALAH...... 105 3.5 Summary and future works...... 112

4 Age-abundance relations in the halo 114 4.1 Introduction...... 114 4.2 Elements of interest for age estimation...... 116 4.2.1 [s/α] ratios...... 117 4.3 Ages and abundances in the GALAH halo data...... 118 4.3.1 The data set...... 118 xi

4.3.2 The age-metallicity relation...... 122 4.3.3 Age and other abundances...... 123 4.4 The origins of GALAH halo stars...... 132 4.4.1 Identifying accreted halo stars based on kinematics...... 132 4.4.2 Identifying accreted halo stars using chemical tagging...... 137 4.5 Conclusion...... 143

5 Conclusions 145 5.1 Thesis conclusion...... 145 5.1.1 Globular clusters in GALAH...... 145 5.1.2 Halo stars with globular cluster-like abundances...... 146 5.1.3 Age-abundance relations in halo giants...... 146 5.2 Future extensions of this work...... 147 5.3 Research acknowledgement...... 148

Bibliography 149 xii

List of Figures

2.1 On-sky positions of the 12 globular clusters with GALAH iDR3 stars pass- ing the positional match are shown as large coloured. Stars from the full GALAH iDR3 catalogue, the K2-HERMES program, and the TESS-HERMES program are shown as grey points.)...... 36 2.2 Histograms of radial velocity for the 4 globular clusters with members ob- served by GALAH...... 38

2.3 Teff versus log g for stars in the 4 clusters observed by GALAH, with the colour variation showing the signal to noise ratio per pixel for each star.. 39 2.4 Metallicity as determined by GALAH and Carretta et al.(2009b) for the same stars. The plot shows a best-fit line in blue and a 1:1 line in red. GALAH appears to slightly overestimate [Fe/H] for metal-poor stars, and the scatter about the best-fit line has an RMS of 0.15 dex...... 42

2.5 Plot of [Fe/H] against surface temperature Teff, colour coded by the signal to noise ratio per pixel, for globular cluster stars in GALAH...... 45

2.6 A zoomed-in look at [Fe/H] versus Teff, with stars divided into three bins in effective temperature. NGC 104 shows the clearest trend between [Fe/H]

and Teff, but there are indications of similar trends in the smaller data sets for NGC 6397 and NGC 7099...... 46

2.7 Plot of [Fe/H] against Teff for NGC 104, using data from Carretta et al., 2009a. Here, there is no correlation between the two quantities, suggesting an accurate spectroscopic analysis...... 47 2.8 Schematic of the Ne-Na and Mg-Al fusion chains showing the nuclei in- volved in the process and the products, including Si. The dashed lines indicate possible leakages in out of the chains and the dashed circles rep- resent unstable isotopes. This figure is a colorized version based on similar figures from Mowlavi and Meynet (2000) and Karakas and Lattanzio (2003). 48 2.9 [Al/Fe] versus [Mg/Fe] for globular cluster stars in our data set as blue cir- cles and data from Mészáros et al.(2020) in the background in grey circles. The horizontal line indicates [Al/Fe] = 0.30 dex...... 49 xiii

2.10 The histogram of [Al/Fe] distribution for our four clusters in 0.05 dex bins. The colours denote the first and second populations: the first pop- ulation with [Al/Fe]<0.3 is shaded red and the second population having [Al/Fe]>0.3 is coloured blue...... 50 2.11 Si-Mg variation for globular cluster observed by GALAH...... 53 2.12 Al-Si variation for globular cluster observed by GALAH with red line in- dicating [Al/Fe] = 0.30 dex...... 54 2.13 The variation of the r-process element Eu on the Mg-Al plane for our glob- ular cluster stars. There is not a clear trend in [Eu/Fe] with the clus- ter populations in NGC 104, but NGC 5139 has a shift in [Eu/Fe] at the [Al/Fe]= 0.3 dividing line between the first and second population.... 57 2.14 The variation of the s-process element Ba on the Mg-Al plane for our glob- ular cluster stars. [Ba/Fe] is higher in the first-population stars in NGC 104 and NGC 5139, but results are more ambiguous for the clusters with only second-population stars...... 58 2.15 The variation of the light- to heavy-s-process ratio [Ba/Y] on the Mg-Al plane for our globular cluster stars. [Ba/Y] is slightly higher in the first- population stars in NGC 104, shows a wide range in both populations in NGC 5139, is positive in the second-population stars in NGC 6397, and is negative in the second-population stars in NGC 7099...... 59 2.16 Li abundance A(Li) as a function of [Fe/H] for globular clusters in the GALAH samples. The stars are divided into effective temperature bins

and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K),

and red (Teff ≤ 4500 K)...... 61 2.17 Abundances for α and 8 other individual light elements in NGC 104. The stars are divided into effective temperature bins and colour coded in blue

(Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Li abundance for all globular clusters are displayed in Figure 2.16...... 63 2.18 Abundances for 9 individual iron-peak elements in NGC 104. The stars are

divided into effective temperature bins and colour coded in blue (Teff ≥

5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K)...... 64 2.19 Abundances for 12 individual n-capture elements in NGC 104. The stars

are divided into effective temperature bins and colour coded in blue (Teff ≥

5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K)...... 65 xiv

2.20 The abundance of the neutron-capture elements Ba and Eu versus the [Ba/Eu] ratio in NGC 104, showing the variation of the relative amounts of s- and r-process enrichment in the cluster members. The lines at [Ba/Fe]= 0 and [Ba/Eu]= 0 mark the separation between n-capture-normal stars ([Ba/Fe]< 0) and s-/r-process-enhanced stars, as described in Table 1.1. The stars are

divided into effective temperature bins and colour coded in blue (Teff ≥

5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K)...... 68 2.21 The abundance ratio of heavy and light s-process elements versus [Al/Fe] for NGC 104, with a red line at [Al/Fe]= 0.3 to separate the first and sec- ond populations...... 69 2.22 Abundances for α and 8 other individual light elements in NGC 5139. The stars are divided into effective temperature bins and colour coded in blue

(Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Li abundance for all globular clusters are displayed in Figure 2.16...... 71 2.23 Abundances for 9 individual iron-peak elements in NGC 5139. The stars

are divided into effective temperature bins and colour coded in blue (Teff ≥

5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K)...... 72 2.24 Abundances for 12 individual n-capture elements in NGC 5139. The stars

are divided into effective temperature bins and colour coded in blue (Teff ≥

5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K)...... 73 2.25 [Na/Fe] plotted against [Al/Fe] (left panel) and [O/Fe] (right panel). The expected correlation on the left is fairly clear; uncertainties in the GALAH determination of oxygen abundances adds scatter to the plot on the right. 74 2.26 Neutron-capture elements Ba (upper panel) and Eu (lower panel) versus the [Ba/Eu] ratio, showing the variation of the s- and r-process enrichment in NGC 5139...... 76 2.27 Abundances for α and 8 other individual light elements in NGC 6397. The stars are divided into effective temperature bins and colour coded in blue

(Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Li abundance for all globular clusters are displayed in Figure 2.16...... 78 2.28 Abundances for 9 individual iron-peak elements in NGC 6397. The stars

are divided into effective temperature bins and colour coded in blue (Teff ≥

5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K)...... 79 2.29 Abundances for 12 individual n-capture elements in NGC 6397. The stars

are divided into effective temperature bins and colour coded in blue (Teff ≥

5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K)...... 80 xv

2.30 Neutron-capture elements Ba and Eu versus the [Ba/Eu] ratio in NGC 6397. The four cluster members with measured abundances show a range in s- and r-process enrichment, and the low [Ba/Eu] ratio indicates the impor- tance of r-process enrichment in the cluster...... 82 2.31 Abundances for α and 8 other individual light elements in NGC 7099. The stars are divided into effective temperature bins and colour coded in blue

(Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Li abundance for all globular clusters are displayed in Figure 2.16...... 84 2.32 Abundances for 9 individual iron-peak elements in NGC 7099. The stars

are divided into effective temperature bins and colour coded in blue (Teff ≥

5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K)...... 85 2.33 Abundances for 12 individual n-capture elements in NGC 7099. The stars

are divided into effective temperature bins and colour coded in blue (Teff ≥

5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K)...... 86 2.34 Neutron-capture elements Ba and Eu versus the [Ba/Eu] ratio in NGC 7099. The two stars we have these abundances for are highly Eu-enriched but not particularly Ba-enriched, indicating that r-process enrichment is relatively more important than s-process enrichment in this cluster...... 88

3.1 Toomre diagram for our data set, with thin and thick disk stars shown as blue points and halo stars plotted in red. The black line marks the locus −1 where vtot is more than 200 km s different from the Local Standard of Rest and the contour lines show the density of points in the plot...... 97 3.2 A Lindblad diagram, showing orbital energy and vertical angular momen- tum for our final data set of metal-poor halo giants (black points), globu- lar clusters from Gaia DR2 (Gaia Collaboration et al., 2018a) (blue crosses), and all other kinematically selected halo stars (red points). In the coordi- nate system we use, prograde orbits are to the right of the diagram and retrograde orbits to the left...... 98 3.3 The abundance ratio of all available elements in the GALAH survey for the selected stars as a function of [Fe/H]. Star J19183781-1011360 is em- phasized with a red star...... 100 3.4 Abundance of Al vs Mg for our halo giants, colour coded by Si abundance. The red line at [Al/Fe] = +0.3 provides a simple separation between the first and second population in globular clusters, and there is clearly an Al- Si correlation. The four stars circled in red are candidate globular cluster escapees, which we discuss in Section 3.4...... 103 3.5 Comparison of halo stars in the [Si/Fe] versus [Mg/Fe] plane against glob- ular cluster members from Carretta et al.(2009b) and Masseron et al.(2019). 104 xvi

3.6 A comparison of the Mg-Al anticorrelation for our four candidate escapees with stars from seven globular clusters with similar [Fe/H]...... 106 3.7 [Al/Fe] versus [Fe/H] for our four candidate escapees, compared with stars from seven globular clusters with similar [Fe/H]...... 107 3.8 A comparison of the Al-Si correlation for our four candidate escapees with stars from seven globular clusters with similar [Fe/H]...... 108 3.9 The four globular cluster escapees are plotted as stars in the Toomre dia- gram and colour-coded by [Fe/H], with the other halo giants in green... 110 3.10 The four globular cluster escapee-candidates in the integrals of motion

plane, LZ versus Etot, comparing their orbital parameters with the other halo giants in green and globular clusters as blue crosses. The vertical red line marks the division between prograde orbits (right) and retrograde or- bits (left)...... 111

4.1 Teff versus log g for our old halo data set, colour-coded by age. The old halo comparison set is shown in the background as pink points...... 120 4.2 A histogram of ages for our halo giants...... 121 4.3 The age-metallicity plot for our old halo data set (black) and the old halo comparison set (pink) with unflagged [Fe/H] values...... 124 4.4 Abundance versus age for our old halo data set (black) and the old halo comparison set (pink). In both data sets, only stars with unflagged (reli- able) abundances are included...... 125 4.5 Unflagged (reliable) abundances of [α/Fe] and 30 individual elements ver- sus metallicity for the old halo giant data set, colour coded by age..... 126 4.6 Lithium abundance versus age for the old halo comparison set (colour- coded by surface gravity) and the old halo data set (green crosses). Main- sequence and turnoff stars with log g > 3.7 tend to have a higher Li abun- dance at all ages than giants...... 129 4.7 A zoomed-in look at abundance versus age for our old halo giant sample (black) and the old halo comparison sample (pink) for [α/Fe], [Mg/Fe], [Si/Fe], [Ca/Fe], [Ti/Fe], and [Na/Fe]...... 130 4.8 Ratios of [Y/Eu], [Ba/Eu], [La/Eu], [Sr/Ba] and [Ba/Y] to examine enrich- ment in the s-process and r-process elements as a function of [Fe/H], with age as the colour variation...... 131 4.9 The chemical clocks [Ba/Mg] and [Y/Mg] versus age for our old halo data set...... 132 xvii

4.10 A schematic of the various structures that have been introduced and iden- tified by Naidu et al.(2020). We have adapted the figure to align with our coordinate system, so that stars on prograde orbits are on the right side of the figure...... 133 4.11 Lindblad diagram of the kinematic space for our old halo data set (with colour variation indicating the age), with the old halo comparison set plot- ted as pink dots in the background...... 134 4.12 The eccentricity, e, against the estimated age for the old halo data set with metallicity as the colour variation (with the old halo comparison set as the background)...... 136 4.13 The eccentricity, e, against the estimated age for the old halo data set with [Mg/Fe] (left) and [Al/Fe] (right) as the colour code...... 136 4.14 [Mg/Fe] versus [Fe/H] for the old halo data set, with age included as the colour variation and the old halo comparison set plotted in the background in light grey. The division line at [Mg/Fe] = −0.2 [Fe/H]+0.05 is a simple criterion suggested by Montalbán et al.(2020) to separate between in situ stars (circled in red) and ex situ stars (circled in blue)...... 138 4.15 Stars with in situ abundances from Figure 4.14 (red dots) and ex situ abun-

dances (colour coded by age) are shown in the same Lz − E plane. The stars with in situ abundances are more likely to have prograde orbits and low total energies than their ex situ counterparts...... 139 4.16 The stars in green are possible members of Gaia-Enceladus based on their high [Mg/Mn] ratio and low [Al/Fe] abundance...... 140 4.17 Stars selected as candidate members of Gaia-Enceladus using the chemical tag from Das, Hawkins, and Jofré (2020) are plotted on the Lindblad dia- gram and colour-coded by their age, with the remaining stars from the old halo data set plotted in the background in maroon...... 141 xviii

List of Tables

1.1 Metal-poor star related definitions from Frebel and Norris (2013) and carbon- enriched metal poor star definitions from Aoki et al.(2007). Carbon-rich stars can appear with or without r- and s-process enhancement...... 17

2.1 Membership criteria for the 12 globular clusters for which GALAH iDR3 has stars within the on-sky cluster radius (half-light radius), based on cat- alogues of globular cluster data from Kharchenko et al.(2013) and Vasiliev (2019) and visual inspection of the data set...... 35 2.2 Mean value of the chemical abundances in NGC 104...... 62 2.3 Mean value of the chemical abundances in NGC 5139...... 70 2.4 Mean value of the chemical abundances in NGC 6397...... 77 2.5 Mean value of the chemical abundances in NGC 7099...... 83

3.1 For each element in the GALAH iDR3 catalog, the number of stars in our halo giant sample with a non-flagged abundance, the mean, standard de- viation in that abundance and the respective typical error...... 101 3.2 Four globular cluster escapees and their stellar parameters...... 109 3.3 Key abundances for our four globular cluster escapees...... 110 xix

List of Abbreviations

The list of abbreviations and acronyms in alphabetical order:

2MASS Two Micron All Sky Survey 4MOST 4-metre Multi-Object Spectrograph Telescope APOGEE Apache Point Observatory Galactic Evolution Experiment AGB Asymptotic Giant Branch AAT Anglo-Australian Telescope CEMP carbon-enhanced metal-poor CoRoT Convection, Internal Rotation and Transiting Planets DESI Dark Energy Spectroscopic Instrument GALAH GALactic Archaeology with HERMES GC Globular Cluster GCE Galactic Chemical Evolution HERMES High Efficiency and Resolution Multi-Element Spectrograph MIST Mesa Isochrones and Stellar Tracks IC Index Catalogue of Nebulae and Clusters of Stars LAMOST Large Sky Area Multi-Object Fibre Spectroscopic Telescope LSR Local Standard Rest MSE Maunakea Spectroscopic Explorer NGC New General Catalogue of Nebulae and Clusters of Stars PFS Subaru Prime Focus Spectrograph RAVE RAdial Velocity Experiment RGB Red Giant Branch SNR signal-to-noise ratio SDSS Sloan Digital Sky Survey TGAS Tycho-Gaia Astrometric Solution TOPCAT Tool for OPerations on Catalogues And Tables WEAVE William Herschell Telescope Enhanced Area Velocity Explorer xx

List of Symbols

α right ascension, ra degree (°) δ declination, dec degree (°) l longitude degree (°) b latitude degree (°) D distance kpc $ parallax mas L orbital angular momentum kms−1 kpc E energy km2s−2 −1 υr radial velocity kms −1 µα proper motion in ra mas y −1 µδ proper motion in dec mas y log g surface gravity cms−2

Teff effective temperature K 1

Chapter 1

Introduction

1.1 Galactic archaeology

The goal of Galactic archaeology is to reveal the history of the Milky Way, which is a typical spiral galaxy, by studying its structure and determining the ages and chemical compositions of its stars. Comparable to archaeologists interpreting the customs of an- cient civilisations based on artifacts, Galactic archaeology involves assembling clues left by events in the Galaxy’s history based on the traces and relics observable today (Recio- Blanco and Thévenin, 2005; Beers and Carollo, 2009).

The Λ cold dark matter (ΛCDM) paradigm, which is often referred to as the standard model of Big Bang cosmology, has become the cosmological framework for Galactic ar- chaeology. It dictates the formation and evolution processes at work in all galaxies, in- cluding the Milky Way. In general, ΛCDM works within the existence of the cosmic microwave background and the large-scale superstructure in the distribution of galaxies (Komatsu et al., 2009; Agertz, Teyssier, and Moore, 2011; Doménech-Moral et al., 2012).

Over the lifetime of our Galaxy, gas has been taken from the larger-scale environment, condensed into stars, returned into the interstellar medium, and recycled into new stellar generations. The overall population of stars at the present day is a mix of many ages and levels of chemical enrichment. Long-lived stars still carry most of the abundances they initially formed with, providing important insights into the formation of low-mass stars across the history of the Milky Way. Because higher-mass stars have shorter lifetimes, we must infer their properties from the chemical enrichment they provided to the Galactic environment and from their compact object remnants. Similarly, the outer regions of the Galaxy have low densities and long dynamical times, allowing kinematic structure to be preserved for many orbits, while the bulge and disk are dynamically active, with resonances and radial mixing constantly redistributing stars in phase space. Chapter 1. Introduction 2

Galaxy formation and evolution are complex, and there are many important observable factors and many relevant physical processes to track. Because we can observe the Milky Way in such great detail, it is the focus of intensive efforts to use its present-day proper- ties to unravel its past. This thesis focuses on old stellar populations in the Milky Way, specifically globular clusters and halo stars, to shed light on the Milky Way in its early development.

There are a number of specific approaches that can be used to make inferences about the Milky Way’s past, including chemical tagging, identification of halo streams and their progenitor systems, searching for patterns in the abundances and kinematics of globular clusters and halo stars, and mapping the elemental abundance patterns of metal-poor stars.

As we are exploring the concept of Galactic archaeology, time is a dimension that plays a crucial role in the background. Hence, despite our focus on chemical abundances in this thesis, we will also study their relations with galactic formation and evolution, and learn how their time scale will be observed from stellar age, nucleosynthetic channels, and abundance ratios.

1.1.1 Chemical tagging

One of the greatest achievements of astrophysics has been the discovery that most of the chemical elements were formed in stars. Nucleosynthesis is a complex process that con- verts existing nuclei into heavier elements. Within a single star, the lightest elements are burnt in the centre, and this process continues until the temperature required for the next reaction chain is higher than the star can produce. Within a population of stars, feedback from supernovae and mass loss during stellar evolution contributes heavy elements to the interstellar medium. These are incorporated into the next generation of stars, with the result that each younger generation of stars builds up heavier elements and abun- dance patterns that depend on their formation environment. Ancient stars in the Milky Way include the most extreme metal-poor stars (e.g., Beers and Christlieb, 2005; El-Badry et al., 2018; Reggiani et al., 2020), which show the Galaxy’s early history with primitive elements. Chemical evolution and stellar abundances will be discussed in depth in Sec- tion 1.3 and subsequent chapters.

Stellar abundances can be used to understand many different aspects of a star or a stel- lar population. Freeman and Bland-Hawthorn, 2002 proposed “chemical tagging” as a way to use stellar abundances to identify groups of stars that had initially formed at the same place and time, which is consistent with our understanding that stars form in clus- ters (Lada and Lada, 2003) from well-mixed gas (e.g., Feng and Krumholz, 2014). Most star formation sites are expected to disperse with time, but if the stars from a single site Chapter 1. Introduction 3 have abundance patterns that match each other closely, and are distinct from other star formation sites, they can be re-connected in abundance space. Ting, Conroy, and Good- man (2015) and Ting, Conroy, and Rix (2016) demonstrated that a large, homogeneous, and precise catalog of abundance data is essential for carrying out this detailed chemical tagging in the Galaxy, as is a robust method for assessing how similar stellar abundance patterns are.

Both the homogeneity and uniqueness of stellar abundances from a single formation site have been the subject of recent research, and so has the development of methods for chemical tagging. We know from work such as De Silva et al.(2007) and de Silva et al. (2009) that open clusters are chemically homogeneous at the 0.05 dex level, but Liu et al. (2016) showed that the abundances in M67 have an intrinsic dispersion below that level.

An interesting study on chemical similarity has been done by Ness et al.(2018), who discuss the phenomenon of “doppelgängers”, which are pairs of stars that have very similar abundance patterns, but were not formed together. The study also predicts that stellar abundances at formation (for disk stars) are a strong function of Galactocentric distance and time, but do not show strong differences between different formation sites at the same distance and time. This is, however, in contrast with studies like Price-Jones et al.(2020), who have some success with chemical tagging.

The process of chemical tagging works within “C-space”, a multidimensional space de- fined by the chemical abundances that we can measure (Freeman and Bland-Hawthorn, 2002). In the simplest verion of C-space, the abundance ratios of the stars being stud- ied are the basis vectors of the space, and we can measure the similarities between those stars through their tendency to cluster in C-space (e.g., Karlsson et al., 2012). Different nucleosynthetic processes follow different paths in C-space, and offer the possibility of transforming to a more effective set of basis vectors through principal component analy- sis (e.g., Ting et al., 2012).

A number of studies have developed and tested potential methods for chemical tagging. Mitschang et al.(2013) demonstrated the use of a Manhattan distance metric as a way to quantify chemical similarity, and Kos et al.(2018) explored the dimensionality reduc- tion technique of t-distributed stochastic neighbour embedding (t-SNE) to visualize and quantify abundance similarities in a two-dimensional space. Dimensionality reduction techniques (including principal component analysis) focus on simplifying the represen- tation of data while maintaining its structure. The t-SNE algorithm identifies the best mapping of a high-dimensional space to a lower-dimensional space while preserving the original grouping information in an information theory sense (e.g., Maaten and Hinton, 2008; Kos et al., 2018; Anders et al., 2018). Meanwhile, Jofré et al.(2017) applied the con- cept of a phylogenetic tree to chart the distribution of stars in C-space and classify their Chapter 1. Introduction 4 similarity, and Casey et al.(2019a) used the Minimum Message Length approach to find the set of “latent factors”, essentially basis vectors in C-space, that describe the distribu- tion of stars in C-space most efficiently. These approaches, and other statistical methods, are critical for reading the history of Galactic chemical evolution from the large current and upcoming sets of stellar abundance data through our understanding of stellar nucle- osynthesis.

Chemical tagging can be used to investigate the formation sites for individual stars, and it can also be used at the population level to separate stars by the larger-scale environment they formed in. This leverages the different chemical evolution that occurs in different environments, for example finding halo field stars that were accreted from dwarf galaxies based on α−elements and other abundances, or finding halo field stars that originally formed in globular clusters based on their light-element abundances. Both of these topics will be covered as we go further in this thesis.

Large-scale surveys like GALAH (De Silva et al., 2015), Gaia-ESO (Gilmore et al., 2012) and APOGEE (Holtzman et al., 2015) have been generating large, homogeneous, high- quality sets of spectroscopic data, which are used to derive high-precision stellar param- eters and abundances. These data sets, and the larger data sets that will come from future surveys like WEAVE and 4MOST, provide a platform for both detailed and population- level chemical tagging, enabling a wide variety of research. For the case of this thesis, however, we do not employ quantitative distance measures in C-space for chemical tag- ging as described earlier in this subsection, but we use correlations between pairs of abundance ratios selected to sample different nucleosynthetic channels.

1.1.2 Halo stars and halo streams

In the widely-accepted hierarchical model of galaxy formation, stellar halos are a hetero- geneous collection of in situ star formation with major and minor accretion events and dynamical evolution. The hierarchical model of galaxy formation dictates that a galaxy is built up from smaller structures, merging together as building blocks to develop larger systems, and eventually, the whole galaxy (Thomas, 1999; Cole et al., 2000; Stringer and Benson, 2007). It is predicted that the merger process should leave behind chemodynam- ical structure, including stellar streams, in the Galactic halo (e.g., Lacey and Cole, 1993; Helmi and White, 1999; Bullock and Johnston, 2005; Font et al., 2008). The stellar stream associated with the Sagittarius dwarf galaxy, which was discovered by chance in 1994, is a massive example of this type of substructure (Ibata et al., 2001, and references therein). Although the accumulated debris from accretion events disperses in coordinate space, stellar streams in the halo can maintain coherent structure in their phase space distribu- tion over long timescales, making the minor merger history of the Milky Way potentially Chapter 1. Introduction 5 observable.

Observational studies in the Milky Way have revealed that the stellar halo is highly struc- tured both spatially (e.g., Ibata, Malhan, and Martin, 2019; Newberg et al., 2020) and kine- matically (e.g., Belokurov et al., 2006; Shipp et al., 2018), and analogous work in M31 by the PAndAS and SPLASH projects have found qualitative similarities along with intrigu- ing differences tied to each galaxy’s individual interaction history (e.g., Martin et al., 2009; Cockcroft et al., 2011; Richardson et al., 2011; Lewis et al., 2013; Kalirai et al., 2009; Gilbert, 2017; Gilbert et al., 2018). Previous work on stellar abundances and kinematics in the halo supports the idea that the outer halo formed primarily through the accretion of smaller systems like dwarf spheroidal galaxies (Searle and Zinn, 1978; Sales et al., 2007; Diemand et al., 2008; Springel et al., 2008; Tissera et al., 2014). There is also a smaller population of “inner halo” stars that may have formed in situ, either from disk heating during merger events (Cooper et al., 2010) or dissipative collapse of gaseous material onto the central region of the Galaxy (e.g., Eggen, Lynden-Bell, and Sandage, 1962; Zolotov et al., 2010; Font et al., 2011). Numerical simulations also predict that most accreted dwarf galaxies should also deposit tidal debris into the Galactic inner halo or even the disk (e.g., Helmi and White, 1999; Abadi et al., 2003), and thus we can expect to trace the ancestry of the stellar halo using local debris remnants.

Many observational studies of halo kinematics have focused on stars in the Solar neigh- bourhood, which are much more observationally accessible than the distant halo. Morri- son et al.(2009) showed that the angular momentum distribution of a nearby sample of metal-poor stars (d ≤ 2.5 kpc, [Fe/H] ≤ −1.0) composed of red giants, red horizontal branch stars, and RR Lyrae variables is not consistent with a smooth halo model, indicat- ing a history of accretion events. Helmi et al.(1999), Kepley et al.(2007) and Smith et al. (2009) all detected discrete overdensities in the distribution of angular momentum and energy for stars in the nearby halo, which they interpret as remnants of past mergers.

Large-scale photometric surveys like the Sloan Digital Sky Survey (e.g., Belokurov et al., 2006), Pan-STARRS (e.g., Bernard et al., 2016; Banik et al., 2019), and the Dark Energy Survey (e.g., Shipp et al., 2018) have dramatically expanded the known set of spatially coherent streams. Spectroscopic followup is returning many new insights into the low- mass systems that have been accreted by the Milky Way and their subsequent evolution, including Wan et al.(2020) on the Phoenix stellar stream, Li et al.(2019) on the ATLAS stream, and Bonaca et al.(2020) on gaps in streams.

The Gaia satellite, with its full sky coverage and extremely precise astrometry, has entirely revolutionised the field of stellar kinematics. The first data release from Gaia generated a number of novel studies of 5-D (right ascension, declination, parallax, proper motions) Chapter 1. Introduction 6 kinematics, including Kushniruk, Schirmer, and Bensby (2017) on the Solar neighbour- hood, Poggio et al.(2017) on the Galactic warp, Oh et al.(2017) on comoving pairs, and Boubert et al.(2018) on hypervelocity stars. The second Gaia data release included radial velocities for approximately 7 million relatively bright stars, bringing the full 6-D phase space in the Solar neighbourhood into sharp focus (e.g., Antoja et al., 2018).

Stellar streams are key archaeological evidence from the halo formation process, and we have recently made major steps forward in our ability to identify and study them. The continued expansion of large photometric, spectroscopic and astrometric surveys will enable us to deepen our understanding of halo stars and their history.

1.1.3 Globular clusters

Both the integrated properties of globular clusters and the properties of their individual stars deliver important information for Galactic archaeology. There is a fairly simple structure in cluster age, metallicity and kinematics in the Galactic globular cluster system, with globular clusters orbiting in the disk and bulge having relatively high metallicities, and globular clusters on halo orbits typically having low metallicities (Zinn, 1985; Zinn, 1993). There is an interesting bifurcation in the age-metallicity relation for halo clusters, where the oldest clusters cover a range of metallicity but only a narrow range in age, and the younger clusters follow an age-metallicity relation that resembles the relation in the Sagittarius dwarf galaxy (Marín-Franch et al., 2009). This parallels the bimodal colour distributions observed in extragalactic globular cluster systems, which are interpreted as a sign of halo assembly through both in situ star formation and accretion (e.g., Brodie and Strader, 2006). A number of authors have used orbital information together with age and metallicity to make inferences about which clusters are likely to have been accreted by the Milky Way, and which may have come from the same progenitor system (e.g., Law and Majewski, 2010; Forbes and Bridges, 2010; Kruijssen et al., 2019; Massari, Koppelman, and Helmi, 2019).

Milky Way globular clusters tend to have quite simple broadband colour-magnitude di- agrams, which indicates that they are much simpler stellar populations than the Solar neighbourhood or the general halo field. Most of them are homogeneous in metallicity at the ≈ 0.03 dex level, which led to a long-standing picture of globular clusters as having experienced a single short burst of star formation and no self-enrichment.

However, a small number of clusters (M54, ω Centauri, M19, M22, M75, M2, NGC 1851, NGC 5286 and NGC 5824) have complex colour-magnitude diagrams and a distinct spread in their metallicity distribution (Da Costa, 2016; Marino, 2017). These clusters with com- plex metallicity distributions tend to be massive, and one possible explanation is that they were originally nuclear star clusters in dwarf galaxies. In this picture, the dwarf Chapter 1. Introduction 7 galaxy experiences multiple episodes of star formation, and retention and recycling of supernova-enriched material builds up a multi-generation cluster at the galaxy’s center. When that galaxy is later accreted by the Milky Way, the nuclear star cluster remains gravitationally bound, but the remaining stars are lost to the halo through tidal tails. This interpretation suits M54, which is currently near the center of the Sagittarius dwarf galaxy, but it is not as successful for ω Centauri, which despite its retrograde orbit has very few extratidal stars plausibly associated with it (e.g., Simpson et al., 2020).

The well-behaved broadband colour-magnitude diagrams of most globular clusters dis- guise a wealth of complexity. Although most Galactic globular clusters display a strict homogeneity in metallicity, they all (with the possible exception of Ruprecht 106, see Vil- lanova et al., 2013; Dotter et al., 2018) show anticorrelations in their light-element abun- dances. This phenomenon is a key characteristic of globular clusters, and is thought to be related to a self-enrichment mechanism resulting from previous generation of stars. There have been extensive observational studies of this phenomenon, including Bekki et al.(2007), Carretta et al.(2009a), D’Orazi et al.(2013a), Mészáros et al.(2015), and Mészáros et al.(2020). This is observable both through high-resolution spectroscopic observations (e.g., Marino et al., 2019) and through narrow- and medium-band photometry (e.g., Piotto et al., 2015; Milone et al., 2017).

We observe anticorrelations in the abundances of elements from He to Si, mainly in the form of C-N, O-Na, and Mg-Al anticorrelations. Bastian and Lardo, 2018 give a thorough review of the phenomenology: the “first population” contains between a third and a half of stars in each globular cluster, and is α-enhanced (Francois, 1991), matching stars of the same metallicity in the halo field (Gratton and Sneden, 1988). The remainder of the stars are the “second population”, and they show depletions in C, O, and Mg coupled with enhancements in N, Na, and Al. He abundance enhancements are usually inferred for the second population from photometry, as they are difficult to determine spectroscopically for most stars (Dupree and Avrett, 2013). Si has been observed to be enhanced only in clusters with particularly high masses and low metallicities (e.g., Yong et al., 2008).

Globular clusters are the only environment where we observe a large population of stars with these abundance patterns, and so we infer that the second population abundance pattern results from some unique aspect of their early history. However, there has not yet been a globular cluster formation model proposed that can reproduce the observed abun- dance behaviour. The unknown origin of multiple populations in globular clusters con- tinues to drive research into questions relevant to Galactic archaeology, including how the properties of the multiple populations depend on cluster mass and metallicity, how a galaxy’s mass affects the ability of its clusters to have multiple populations, and the inter- play between globular cluster formation and galaxy formation at redshift ≈ 3. Detailed Chapter 1. Introduction 8 abundances of stars in globular clusters form a set of interrelated observables that can be used to set constraints and requirements on cluster formation models, illuminating an early stage of the Milky Way’s formation.

1.1.4 Age-abundance relations and metal-poor stars

The Milky Way halo is mainly composed of old and metal-poor stars. It provides us with an important opportunity to directly observe the conditions early in Galactic history, since stars preserve most of their surface abundances throughout their lifetimes. Indeed, the most metal-poor stars can be used to investigate the yields of supernovae from nearly metal-free stars and the initial mass function in the first generation of stars (e.g., Frebel and Norris, 2015; Heger and Woosley, 2010; Nordlander et al., 2019).

Moving up from extremely metal-poor stars ([Fe/H]≤ −3.0) to the typical metallicity range of the halo (−3.0 ≤[Fe/H]≤ −1.0), stellar abundance patterns can still be seen as influenced by only a small number of nucleosynthetic sources. This allows us to use the astrophysics of those sources, including the timescale for chemical enrichment and metallicity-dependent yields, to interpret observed abundance vs abundance planes as a progression over time. A well studied example of this is the [α/Fe] vs [Fe/H] plane, which Venn et al.(2004) presents as a diagnostic of the star formation history in dwarf galaxies Within the Milky Way, McWilliam (1997) also analyzed the abundance of O, Mg, Si, Ca, and Ti and neutron-capture elements in the Galactic disk, halo, and bulge, and demonstrated that chemical evolution is mainly influenced by the stellar environment.

Alpha elements are produced during high-temperature fusion processes in massive stars and in the Type II supernovae that are the end stage of the massive star life-cycle. Mean- while, helium nuclei (α particles) are produced by the photodisintegration of silicon in advanced stellar burning during the CNO cycles and various hot bottom burning cycles (Clayton, 1983) and free α particles are readily captured by the other nuclei in the envi- ronment. The products of these captures are the nuclei between carbon and the iron peak that have even numbers of protons: O, Mg, Si, S, Ca, Ti. Type II supernovae are the main source of α elements in galactic chemical evolution (Timmes, Woosley, and Weaver, 1995; Kobayashi et al., 2006; Gai, 2014). Elements in the iron peak, in contrast, are mainly pro- duced in Type Ia supernovae (Dwek, 1998; Kobayashi and Nomoto, 2009). These result from the detonation of white dwarfs that exceed the Chandrasekhar mass (Tomaschitz, 2018), either through gradual accretion of mass from a binary companion (the so-called ”single-degenerate scenario”) or through the merger of two white dwarfs (the ”double degenerate scenario”) (Pan, Ricker, and Taam, 2013, and references therein).

The progenitors of the two different types of supernova have very different initial masses, and as a result their contributions to galactic chemical enrichment happen at different Chapter 1. Introduction 9 times. The massive stars that will undergo Type II supernovae and produce α elements only live for a short time following their formation, while the low-mass stars that will be- come white dwarfs and become Type Ia supernovae have long main-sequence lifetimes and cannot return iron peak elements quickly following star formation (Arnett, 1971; Ar- nett, 1975; Tinsley, 1979). As a result, successive generations of stars in a galaxy will proceed through the space of α element abundance and metallicity in a way that is pre- dictable based on time and the star formation rate. Other sources of galactic chemical enrichment also have their own delay times, yields, and occurrence rates: asymptotic giant branch stars with different masses and metallicities, neutron star mergers, and cos- mic ray spallation all contribute different elements to the interstellar material at different rates (Eldridge et al., 2018).

1.1.5 The [α/Fe] ratio

The production of α elements occurs mainly in Type II supernovae, which are the evo- lutionary end stage of the most massive stars (M & 10 M ) with very short lifetimes (Matteucci, 1992; Goodwin and Pagel, 2005). Hence, Type II supernovae begin to enrich the interstellar medium shortly after the start of star formation, and stop shortly after star formation ends. Meanwhile, iron-peak elements (e.g., Cr, Mn, Co, Ni) are mainly produced in Type Ia supernovae. Type Ia supernovae have a longer delay time due to the lower mass of their progenitors, meaning that they do not start to contribute to the chemical evolution of their environment until roughly 1 Gyr after the beginning of star formation (Taylor and Kobayashi, 2015; Kobayashi, 2016; Vincenzo, Kobayashi, and Tay- lor, 2018). However, there is some phenomenological variety in Type Ia supernovae (e.g., Hillebrandt et al., 2013) that may correlate with different delay times. Bonaparte et al. (2013) propose that the existence of a minor class (<15-20%) of Type Ia supernovae with delay times on the order of ≈100 Myr could have played a role in early galactic chemical evolution (Hillebrandt and Niemeyer, 2000; Hillebrandt et al., 2013; Scalzo, Ruiter, and Sim, 2014).

The star formation history also determines how the α elements evolve relative to metal- licity. Venn et al.(2004) showed that metallicity alone cannot be used simply to designate a star as “Galactic” or “accreted” because the [α/Fe] ratio is lower in dwarf galaxy stars than it is in halo stars at the same metallicity. Meanwhile, Kirby et al.(2013) found that the mass–metallicity relation, where metallicity was measured in the gas phase, is the same for dwarf elliptical and dwarf irregular galaxies in and around the Local Group as well as for the dwarf spheroidal satellites of the Milky Way. This gives us a strong expec- tation for the α - metallicity - age relation in the halo, since it is formed mainly through accretion of dwarf galaxies. Chapter 1. Introduction 10

Low-mass stars record the chemical state of the environment at the time of their forma- tion and preserve it throughout their lifetime, providing fairly direct observational access to an early stage of Galactic history. Given our best theoretical understanding of low- mass stellar evolution and the sources of chemical enrichment, we can infer stellar ages and observe the progress of the stellar population in the Milky Way through the [α/Fe] vs [Fe/H] plane (e.g., Haywood et al., 2013) or the [s-process/α] vs [Fe/H] plane (e.g., Nissen and Schuster, 2011; Lin et al., 2020). We can also use population-level chemical tagging (Freeman and Bland-Hawthorn, 2002) to identify stars that originally formed in dwarf galaxies and were later accreted by the Milky Way, since the progress of chemical evolution is sensitive to the galaxy-scale environment.

Galactic archaeology uses photometric, spectroscopic, and astrometric data for many Milky Way stars together with state-of-the-art numerical simulations to chart how the universal processes of galaxy formation have interacted with the unique details of the Milky Way to produce the Galaxy we observe. The ongoing boom in observational Galac- tic archaeology work has been enabled by technology development in the form of highly multiplexed spectrographs, dedicated observational facilities, and satellite missions like Gaia, Kepler, and CoRoT. As capabilities have improved, larger sample sizes have en- abled analyses of parameters to go beyond summary statistics to the full distribution functions. There is much to learn from the details of a large sample, and from the out- liers. In the future Galactic archaeology will expand its scope and impact as astronomers improve the connections between Galactic and extragalactic studies, observations and simulations. Chapter 1. Introduction 11

1.2 Large-scale surveys

Our understanding of the Milky Way has long been driven by observational advances. It is the only galaxy where we can study individual stars at all evolutionary phases in great detail, and every boost to our technological abilities in data collection and analy- sis has led to corresponding advances in our theoretical understanding of the physical processes that govern galaxy evolution. Advances in detectors and optics, multiplexed instruments, wider fields of view, and sophisticated methods for data reduction and anal- ysis that take advantage of significant increases in computing power, all enable observa- tional astronomers to “see” a more complete picture of the Galaxy.

Observational studies prior to the large survey era were confined to samples of a few hundreds of stars at most, and would focus on particular components of the Galaxy such as the Solar neighborhood or globular clusters. Eggen, Lynden-Bell, and Sandage (1962), Bond (1970), and Searle and Zinn (1978) are all fundamentally important publications that use samples of tens to few hundreds of stars to explore the formation of the galaxy through the analysis of kinematics and chemical abundances.

Since the 1990s, large-scale survey projects and dedicated observatory facilities have played an increasingly important role in astronomy. The Sloan Digital Sky Survey (SDSS, Gunn et al., 2006) was the first major step in this direction, followed by the Two Mi- cron All Sky Survey (2MASS, Skrutskie et al., 2006) and many others since. The massive observational progress in projects like these enables more sophisticated modelling and complex analysis of the structure of the Milky Way (e.g., Springel et al., 2005; Robin et al., 2014; Rybizki et al., 2020).

There has also been a boom in instrumentation development to enable these large-scale survey operations. Technological advances in highly multiplexed spectroscopy have been particularly important for Galactic archaeology. Major surveys that have bene- fited from these advances include GALAH (De Silva et al., 2015), the Gaia-ESO survey (Gilmore et al., 2012), APOGEE (Majewski et al., 2017), and RAVE (Kordopatis et al., 2013). Each of these projects has aimed to collect spectra for large numbers of stars in the Milky Way to enable new discoveries in Galactic archaeology, and each has its particular scien- tific niche thanks to the details of their latitude, instrumental setup, and target selection strategy. Chapter 1. Introduction 12

1.2.1 Galactic Archaeology with HERMES

GALAH (Galactic Archaeology with HERMES) is a survey project that aims to study the history of the Galaxy using chemical abundance data derived from high quality spec- troscopy of about a million stars. The HERMES spectrograph (High Efficiency and Reso- lution Multi-Element Spectrograph, Sheinis et al., 2015), which is used to collect GALAH spectra, is a fibre multi-object spectrograph on the the 3.9 m Anglo-Australian Telescope. HERMES is fed by the 2dF fibre positioner system (Lewis et al., 2013), which has 392 sci- ence fibres and eight guide bundles on each of two two-degree-diameter field plates that can be rotated into the light path, with a robotic system to arrange fibres on one plate while the other is being observed. HERMES captures light in four simultaneous non- contiguous wavelength bands covering about 1000 Å in total, including the H-α and H-β lines. The main target catalog for GALAH contains all stars with apparent V magnitudes between 12 and 14, at declinations between −80◦ and +10◦. Stars must be located more than 10◦ away from the Galactic plane, in a field with at least 400 stars in the 2dF field of view.

GALAH DR2 is the latest official data release, containing 342 682 stars with abundance measurements for up to 23 elements for each star (Buder et al., 2018). The third GALAH data release will be made public in 2020, and will contain over 600 000 stars with up to 30 measured abundances. The work in this thesis is done with the iDR3 catalog, which is the internal pre-release version available to GALAH team members.

GALAH in comparison with other surveys

The various ongoing Galactic archaeology projects are all very complementary to each other, and can be used side-by-side with each other for data support, comparison, or supplement. As an example, the Gaia DR2 catalog, which is a truly massive amount of highly precise astrometric data (approximately 1.7 billion sources), can be combined with the GALAH data set to analyse the components and evolution of the Milky Way (e.g., Bland-Hawthorn et al., 2019; Khanna et al., 2019; Cotarˇ et al., 2019).

GALAH can also be combined with other spectroscopic surveys. APOGEE is a large-scale spectroscopic survey that has similar science goals, but collects data in the near-infrared region. As a result of its instrument design and its observing strategy, APOGEE has many pencil-beam lines of sight through the Galactic disk, which are an excellent companion data set to GALAH’s dense sampling of the Solar neighbourhood. Stars that have been observed by both surveys can be used to cross-calibrate their results (e.g., Carrera et al., 2019). There is a similar alignment between GALAH and the LAMOST survey (Xiang et al., 2019), which is collecting a larger set of stellar spectra, but at lower resolution and to a fainter limit. In comparison, GALAH has spectral resolution of R ≈ 28, 000 while Chapter 1. Introduction 13

LAMOST has R≈ 1, 800. In addition, LAMOST can reach stars much further away in the Galaxy than GALAH or APOGEE, making it an important component of a combined spectroscopic data set. Cross-calibration between LAMOST and GALAH or APOGEE, with the use of machine learning techniques, makes it possible to combine the data sets fairly directly (e.g., Ho et al., 2017; Wheeler, Ness, and Hogg, 2020). To summarize, the differences in spectral resolution and spectral coverage between the GALAH, LAMOST and APOGEE surveys complement each other as spectral data companions in assembling a comprehensive stellar spectroscopic analysis. Hence, it is typical for the GALAH survey to be mentioned alongside the LAMOST and/or APOGEE project and vice versa.

Over the next decade, a new generation of larger-scale high-resolution surveys will com- mence, using larger telescopes, higher multiplexing, and novel instrument design to col- lect an order of magnitude more data and address questions in Galactic archaeology in finer detail. Projects including WEAVE (Dalton et al., 2014), 4MOST (de Jong et al., 2014), DESI (Levi et al., 2019), MSE (McConnachie et al., 2016) and PFS (Takada et al., 2014) will enable a wider and deeper view of Galactic behaviour and history than is currently possible.

1.2.2 Gaia

Gaia is a satellite mission of the European Space Agency that is collecting images and spectra across the full sky to generate a massive and highly precise astrometric and pho- tometric catalog for the Milky Way and its nearest neighbours, and to determine fun- damental stellar parameters. Gaia data plays an essential role in this thesis, making it possible for us to calculate distances and orbits, and to select halo stars kinematically.

The most recent Gaia data release is DR2, and it includes photometry, positions, proper motions and parallaxes for about 1.3 billion stars (Gaia Collaboration et al., 2018b). eDR3, an early version of the next data release, will be made public in 2020 with expanded catalogs and improved precision. Gaia data enables astronomers to investigate the struc- ture and dynamics of the Milky Way in far greater detail than was previously possible. Studies like Belokurov et al.(2018) and Helmi et al.(2018) were able to investigate ma- jor accretion events in Galactic history using data from Gaia, and combining Gaia data with spectroscopy allows close-up studies of stellar streams (e.g., Koppelman et al., 2019; Meingast, Alves, and Fürnkranz, 2019; Borsato, Martell, and Simpson, 2020) and Local Group dwarf galaxies (e.g., Fritz et al., 2018; Torrealba et al., 2019). Chapter 1. Introduction 14

1.3 Stellar abundances

The physical parameters of a star’s photosphere, including the temperature, surface grav- ity, the velocity of small-scale turbulent motions, and the abundances of various ele- ments, can be derived from analysis of the star’s spectrum. While the first three pa- rameters change significantly over the lifetime of a low-mass star, the abundances at the surface are mostly stable. There are a small number of exceptions to this rule:

• Li, Be, B: These elements are particularly susceptible to proton capture and can be destroyed at the base of the surface convective zone. They can be depleted on the main sequence and in the event of mass transport between the interior and the surface.

• C, N: These elements are produced or destroyed in hydrogen burning at the typical temperatures of low-mass stellar interiors. Their surface abundances are changed if there is mass transport between the interior and the surface, such as in first dredge- up or deep mixing.

• C, s-process elements: These elements are produced in the interiors of AGB stars and transported to the surface during the thermal pulsation phase.

• Mass transfer between the components of a binary star system will affect the surface abundances of the recipient star.

• Diffusion in the atmospheres of main-sequence stars can affect the measured abun- dances of many elements, with some elements showing increases and some show- ing decreases that grow stronger over time (Alecian, 2013; Vauclair, 2013; Michaud, Alecian, and Richer, 2015). For example, Michaud et al.(2004) found that He, Li, Be, B, Mg, P, Ti, Fe, and Ni were those most affected by this mechanism when they analyzed 28 elements for members of M67. First dredge-up re-mixes the stel- lar atmosphere and restores the initial surface abundances, and diffusion does not operate in the lower-gravity environment of red giant atmospheres.

The majority of elements have constant abundance throughout the main sequence and red giant branch phases. Among the exceptions noted above, the processes of first dredge- up, diffusion, and deep mixing can be modeled reasonably well to correct the observed abundances to the initial stellar abundances.

Before continuing into discussion of elemental abundances, it is important to lay out the notation and conventions used in stellar abundance work. Most basically, e(X), the abundance of element X, is presented logarithmically, relative to abundance of hydrogen (H), with log e(H) = 12 by definition. Chapter 1. Introduction 15

log e(X) = log(NX/NH) + 12

For stellar abundances in the literature, results are usually presented relative to their values in the Sun, using the so-called “bracket” notation. In this notation, [X/H] for a given star represents the logarithmic difference between the abundance of element X relative to hydrogen in the star, and that same quantity measured for the Sun.

[X/H] = log (NX/NH)∗ − log (NX/NH) = log e(X)∗ − log e(X) and for two elements X and Y, one then has

[X/Y] = log(NX/NY)∗ − log(NX/NY)

These “bracket” abundances are measured in units of decimal exponentials or dex. If the abundance ratio in the star is the same as in the Sun, then [X/H]=0, and if two stars differ by 1 dex in [X/H], it means the number densities of X atoms relative to H atoms in the two stars differ by a factor of 10.

1.3.1 Metallicity

One fundamental abundance measure is metallicity, which is the abundance sum of all elements heavier than helium. Hydrogen and helium are the main products of Big Bang Nucleosynthesis, and they have always made up the majority of the baryonic Universe. Metallicity, as an integrated measure of the heavier elements in a star, is a reasonable indicator of how far chemical enrichment had progressed prior to its formation. Rather than measuring and summing the abundances of every element in a star, we use the iron abundance [Fe/H] as a proxy for the overall metallicity. Having a single proxy for heavy element abundance is reasonable because many of the elements, including iron, vary together in many stars. [Fe/H] is the proxy because it has many spectral lines, and the line blanketing that occurs at blue wavelengths as those lines become strong were used as early photometric metallicity measures (Wheeler, Sneden, and Truran, 1989). Because of the major role that type Ia SNe plays in chemical evolution and particularly the formation of elements near iron in the periodic table, this pattern also tends to hold for normal stars.

To put the concept of metallicity into context, we use the properties of the major com- ponents of the Milky Way, as compiled by Bland-Hawthorn and Gerhard (2016). The metallicity of the Sun is 0 by definition, and the Sun is fairly typical as a member of the Galactic thin disk. The thin disk is the site of ongoing star formation in the Milky Way. It Chapter 1. Introduction 16 has a scale height of around 300 ± 60 pc and a scale length of 2.5 ± 0.4 kpc , and the Sun is located 8.2 ± 0.1 kpc from the Galactic centre and 25 ± 5 pc above the Galactic mid- plane. The thin disk has undergone significant chemical enrichment, and the maximum metallicity in the thin disk is around +0.5 dex. The thick disk is an older component of the Galaxy, with typical stellar ages of 8 Gyr or more and metallicities in the range −1.5 ≤ [Fe/H] ≤ −0.5. It has a larger scale height than the thin disk (≈ 1 kpc), and its origin is the topic of much ongoing research. While thick disks are a common feature of spiral galaxies (e.g., Dalcanton, Spergel, and Summers, 1997), there are a number of possible processes for forming them, including the capture of a satellite galaxy, dynam- ical heating of a thin disk, and outward migration from the inner disk. The halo is the oldest and largest component of the Galaxy, with a density profile that scales as R−3.5 and a distinct fall-off in stellar density at R ≈ 40 kpc (Deason, Belokurov, and Evans, 2011). The orbits of halo stars have much more variety than those of disk stars in terms of eccentricity, semimajor axis, and orientation. The mean rotation of the halo within a Galactocentric distance of 20 kpc is zero, with a large dispersion, and the mean rotation at larger distances is slightly retrograde. The metallicity distribution function in the halo falls off dramatically at the low metallicity end, with most halo stars in the range −2.5 ≤ [Fe/H] ≤ −1.0. The most metal-poor stars discovered to date in the Milky Way have metallicities as low as [Fe/H]= −6.2 (Nordlander et al., 2019), but they are vanishingly rare.

The GALAH survey has a faint magnitude limit of V = 14, meaning that the main- sequence stars in GALAH all tend to lie within roughly 3 kpc of the Sun, and bright giants are within 8 kpc. There is also no selection in the iDR3 data based on colour or parallax, which provides a fairly simple selection function for Galactic studies and returns a data set that is strongly weighted toward the nearby disk. Halo stars observed by GALAH are all within a few kpc of the Sun, which is something of a limitation as we cannot sample the distant halo directly.

1.3.2 Spectroscopic analysis

The primary goals of analysing stellar spectra are to determine the parameters of the stel- lar photosphere, which are represented by the effective temperature Teff and the surface gravity, log g (Unsöld, 1970). Meanwhile, high-resolution spectroscopic analysis gives the most straightforward method of determining chemical abundances, through mea- suring the equivalent widths of individual spectral lines or by comparing the data with synthetic spectra (Unsöld, 1970; Matteucci, 2012).

For the most part, variations in the abundances of individual elements in a star tend to follow the variations in the iron abundance. However, there are cases where stars do not Chapter 1. Introduction 17

TABLE 1.1: Metal-poor star related definitions from Frebel and Norris (2013) and carbon-enriched metal poor star definitions from Aoki et al. (2007). Carbon-rich stars can appear with or without r- and s-process en- hancement.

Description Definition Population I young (disk) metal-rich stars Population II old (halo) stars formed from low-metallicity gas Population III postulated first stars, formed from zero-metallicity gas Solar [Fe/H] = 0.0 Metal-poor -2.0 ≤ [Fe/H] < -1.0 Very metal-poor -3.0 ≤ [Fe/H] < -2.0 Extremely metal-poor -4.0 ≤ [Fe/H] < -3.0 Ultra metal-poor -5.0 ≤ [Fe/H] < -4.0 Hyper metal-poor [Fe/H] < -5.0 Carbon-rich stars [C/Fe] > +0.7 for log(L/L ) ≤ 2.3 [C/Fe] ≥ (+3.0 - log(L/L )) for log(L/L ) > 2.3 n-capture-rich stars (r-I) 0.3 ≤ [Eu/Fe] ≤ +1.0 and [Ba/Eu] < 0 n-capture-rich stars (r-II) [Eu/Fe] > +1.0 and [Ba/Eu] < 0 n-capture-rich stars (s) [Ba/Fe] > +1.0 and [Ba/Eu] > +0.5 n-capture-rich stars (r/s) 0.0 < [Ba/Eu] < +0.5 n-capture-normal stars [Ba/Fe] < 0 have such a “scaled Solar” abundance pattern, and [Fe/H] does not accurately capture the overall metallicity. Two examples of this are carbon-enhanced extremely metal-poor (CEMP) stars and stars with enhanced α abundances. In both cases, the abundance dif- ferences are significant enough to affect the internal structure of the star. In both cases, the additional atoms provide excess opacity, causing the star to be a little larger and cooler than it would otherwise. In CEMP stars, the extra opacity is from carbon-bearing molecules (CH, CN, CO), and α-enhanced stars provide extra opacity through resonance lines and free electrons.

In the GALAH survey, the abundance analysis was done by the members of a specialist working group within the GALAH team. The spectroscopic analysis is carried out by fitting synthetic spectra to the observed data using the Spectroscopy Made Easy (SME) code (Valenti and Piskunov, 1996; Piskunov and Valenti, 2017), using MARCS model atmospheres (Gustafsson et al., 2008) and non-LTE corrections for O, Na, Mg, Al, Si and Fe from Amarsi et al.(2020).

An initial set of stellar parameters is determined for each star by fitting to a grid of syn- thetic spectra as part of the data reduction pipeline. These are then improved iteratively by re-normalising and fitting carefully selected spectral regions. Once stellar parameters have been finalised, abundances are determined for each element independently by syn- thesising small windows around each of the features in our linelist. The data reduction and analysis is described in the Data Central website for GALAH DR3 (Simpson, 2020). Chapter 1. Introduction 18

Full details of the process also have been published in the DR3 data release paper (Buder et al., 2020), and it is similar to the analysis process followed for the training set data in GALAH DR2 (Buder et al., 2018). Chapter 1. Introduction 19

1.4 Research objectives

This thesis explores the ability of the GALAH Survey to contribute to studies of Galactic globular clusters and their role in halo assembly. In it, we identify GALAH stars be- longing to globular clusters, evaluate how well the light-element abundance complexity known from decades of previous work can be seen in GALAH data, and check for ad- ditional abundance correlations. This last step is made possible by the homogeneous analysis of GALAH data (Buder et al., 2018), which returns stellar parameters (Teff, log g, [Fe/H]) and up to 30 elemental abundances [X/Fe] from all major nucleosynthetic chan- nels. It is difficult to obtain a comprehensive spectroscopic sample of globular cluster stars because of their on-sky density, and it can be difficult to extract detailed abundance information for metal-poor stars from the limited wavelength coverage in GALAH data.

As we already discussed, the Milky Way offers a unique near-field opportunity to investi- gate the hierarchical formation of galaxy-scale structure, and astronomers take many ob- servational and theoretical approaches to the broad topic of Galactic archaeology. These include studying the masses, internal structure, compositions, ages, and orbits of stars in the different morphological components of the Milky Way: the bulge, thick disk, thin disk, and halo.

The hierarchical assembly of a galactic halo is a complex and long-term process, and the origin of the Milky Way’s stellar halo and its population of globular clusters remains an active topic of investigation. There are observable aspects of the progenitor systems accreted during the assembly process that are invariant or approximately so, including orbital actions and stellar chemical compositions, and we can use those to differentiate and study those progenitors. Recent studies using the Galactic globular cluster system to trace back halo assembly include Kruijssen (2015), Posti and Helmi (2019), and Massari, Koppelman, and Helmi (2019).

In this thesis, we aim to answer these questions:

1. What is the sample of globular cluster stars observed in the GALAH survey?

2. What is the abundance profile of those globular cluster stars?

3. Can we find halo stars in GALAH with abundance patterns resembling those in globular cluster stars?

4. What is the relationship between age and abundance for the halo giants?

5. What chemical or kinematic evidence can we see in GALAH for accretion events in the halo? Chapter 1. Introduction 20

Accordingly, the first project of this thesis is to explore how well the GALAH iDR3 cat- alog captures key observable features in globular cluster abundances and to check the reliability of the GALAH data set. From that point, we take the elemental abundances that are well determined in GALAH globular cluster stars and use them for chemical tagging. Our goal is to make new identifications of stars in the Galactic halo that origi- nally formed in globular clusters and later escaped into the halo field. Finally, we explore the behaviour of age-sensitive abundance ratios in the set of GALAH halo giants, and investigate whether we can see signatures of accretion events in the halo. Our goal in considering the age-metallicity and age-abundance relations in the Galactic halo is to make a preliminary investigation into how well the GALAH abundance data can follow the different chemical enrichment pathways at work. Chapter 1. Introduction 21

1.5 Thesis structure

The first chapter of this thesis introduces the essential scientific goals and tools of Galactic archaeology that are used throughout the rest of the chapters. The overall context of this thesis research is that key aspects of Galactic history can be inferred from present-day stellar properties, and the specific focus is on globular cluster stars and their connection to the Galactic halo. We describe the GALAH Survey, which is our primary data source, and the possibilities for studying the abundance behaviour in globular cluster and halo stars using that data set. Galactic archaeology, and chemical tagging in particular, are important pathways for deepening our understanding of globular clusters and halo stars.

As set in the research objectives, this thesis presents three research chapters, each of which begins with a specific introduction that situates the project in the context of pre- vious work. Chapter2 is an in depth look at Milky Way globular clusters as seen in the GALAH data set, particularly focusing on the light element abundance anticorrelations that have been established in the literature.

Chapter3 focuses on kinematically selected halo stars in the GALAH data set and uses chemical tagging to search for halo field stars that may have originated in globular clus- ters, using the results of Chapter2 and following methods from the literature.

Chapter4 applies the GALAH data set to the chemical evolution of the Galactic halo. GALAH iDR3 includes estimates of individual stellar ages, calculated through a proba- bilistic isochrone-based analysis. We evaluate age-abundance relations discussed in the literature using GALAH ages and abundances for kinematically selected halo giants.

The thesis concludes in chapter5 with a summary of the results from the three science chapters and a discussion of future research possibilities.

As a summary, this thesis contain these chapters:

• Chapter1: Introduction to the thesis topic

• Chapter2: Globular clusters observed in the GALAH survey

• Chapter3: Milky Way halo stars with globular cluster-like abundance patterns

• Chapter4: Age-abundance relations in the Galactic halo

• Chapter5: Conclusion 22

Chapter 2

Globular clusters observed in the GALAH Survey

2.1 The role of globular clusters in Galactic archaeology

Globular clusters are a fascinating component of the Galaxy: dense dark matter-free satel- lites that orbit in the halo, disk, and bulge. There are 157 known globular clusters in the Milky Way (Harris, 1996, 2010 edition), and that number is growing thanks to sensitive large-scale photometric surveys (e.g., Camargo and Minniti, 2019). The study is enhanced by surveys such as Dark Energy Survey (DES; Dark Energy Survey Collaboration et al., 2016; Shipp et al., 2018) and the Panoramic Survey Telescope And Rapid Response Sys- tem (Pan-STARRS; Lin, Magnier, and Chen, 2011; Piatti and Carballo-Bello, 2019). They are useful tools for studying Galactic history in several ways. The abundance patterns of individual stars can be used to investigate the chemical enrichment that happened be- fore their formation, and they are also the main tool for building a comprehensive model for the formation of globular clusters and the origin of their abundance anomalies (e.g., Bastian and Lardo, 2018).

Globular clusters can be found throughout the Milky Way, from as close as 0.5 kpc to as far as 125 kpc from the Galactic center. They occupy a wide range of metallicity, from −2.5 ≤ [Fe/H] ≤ 0 (Harris, 1996), and the majority of them have fairly simple stellar populations with ages greater than 10 Gyr. Halo globular clusters tend to be metal-poor and are strongly isolated, with elliptical orbits that take them far from the Galactic centre. Meanwhile, disk and bulge clusters can have metallicities as high as Solar, and a few clusters are as young as 7 Gyr (Marín-Franch et al., 2009). The stars in each cluster are a single-age population to within roughly 0.5 Gyr, which is the limit of our precision for isochrone-based age determination in such old stars. High-precision photometry and spectroscopy (e.g., Milone et al., 2015) have revealed multiple stellar populations in light- element abundances in all globular clusters except one, and metallicity dispersions in Chapter 2. Globular clusters in GALAH 23 only a few.

Globular clusters undergo dynamical evolution through processes such as two-body re- laxation, single and binary stellar evolution, and Galactic tidal stripping. These processes tend to drive mass loss, with one or another dominating depending on the properties of the cluster and its orbit in the Galaxy (e.g., Gnedin and Ostriker, 1997). While some clus- ters appear to have been captured from other galaxies (e.g., Law and Majewski, 2010), some are remnants from the early formation of the Milky Way (e.g., Terzan 5; Gotta et al., 2018). The ages, abundances, and kinematics of globular clusters all hold important information about the process of halo assembly (e.g., Bekki, 2005; Forbes and Bridges, 2010; Myeong et al., 2018; Massari, Koppelman, and Helmi, 2019).

It remains unclear how individual globular clusters formed, and how globular cluster formation fits into the larger process of galaxy assembly. Lotz, Miller, and Ferguson (2004) make the case that globular clusters belonging to dwarf elliptical galaxies formed as subsystems of those halos, and not as independent objects. Studies of more massive elliptical galaxies (Brodie et al., 2012; Usher et al., 2012) find composite populations of “red” and “blue” globular clusters that are spatially and kinematically associated with mergers and in situ formation, respectively. The Galactic halo is predominantly built up by the accretion of smaller galaxies (Bullock and Johnston, 2005; Helmi et al., 2006; Font et al., 2008) which contribute stars and globular clusters, and Peñarrubia, Walker, and Gilmore (2009) predict that some of the globular clusters should survive the accretion process, while some dissolve. Recent models for Milky Way assembly (e.g., Kruijssen, 2015) discuss the overall globular cluster population as the product of the hierarchical merging of smaller systems.

Our model for the assembly of the Milky Way’s globular cluster system is that some clus- ters formed in situ early in the history of the Milky Way and some were captured from smaller galaxies during the hierarchical merging process. Forbes and Bridges (2010) sug- gested that 27–47 globular clusters (about 1/4 of the entire system) were accreted from six to eight dwarf galaxies to form globular clusters in the Milky Way that we observe to- day. This suggestion implies that the migration of stars and star clusters between galaxies through the merger process can be tracked and determined. In addition to that, more re- cently, Kruijssen et al.(2019) carried out a thorough study of Milky Way formation and as- sembly using the age-metallicity distribution of 96 globular clusters. They conclude that about 40 per cent of current Galactic globular clusters formed ex situ, with 6 ± 1 being former nuclear star clusters in nucleated dwarf galaxies. Meanwhile, Myeong et al.(2019) suggested that much of the stellar halo of the Milky Way was built in a major accretion event that also brought in globular clusters such as ω Centauri and FSR 1758. Massari, Koppelman, and Helmi (2019) explore further in Gaia data to associate the Milky Way’s Chapter 2. Globular clusters in GALAH 24 approximately 150 known globular clusters with specific progenitors. They conclude that 40% of the clusters likely formed in situ, while 35% of them can possibly be associated to known merger events, in particular to Gaia-Enceladus (19%), the Sagittarius dwarf galaxy (5%), the progenitor of the Helmi streams (6%), and to the Sequoia galaxy (5%). In these studies, Kruijssen et al.(2019) used ages and metallicities, Myeong et al.(2019) used kinematics, and Massari, Koppelman, and Helmi (2019) used age, metallicity, and kinematics. While there are some claims that are common to multiple studies, there are some clusters assigned entirely different origins by different authors, and consensus has yet to be reached.

Globular clusters also play a role in contributing stars to the general halo population, as was shown by Martell and Grebel (2010). The multiple populations in globular clus- ters exhibit anticorrelated variations in their light-element abundance patterns (e.g., C-N, O-Na, Mg-Al; Bastian and Lardo, 2018) that are not commonly observed in other compo- nents of the Galaxy, and these patterns can be used as a chemical tag to identify stars that have escaped from clusters. Identifying halo field stars that originally formed in glob- ular clusters allows us to investigate the formation process of those clusters and their subsequent stability or dissolution in the Galactic halo. This approach is highly com- plementary to studies of tidal tails and stellar streams originating from globular clusters (e.g., Odenkirchen et al., 2001 on Pal 5, Navin, Martell, and Zucker, 2016 on M3 and M13, Grillmair and Johnson, 2006 on NGC 5466, and Wan et al., 2020 on the Phoenix stream).

The first globular clusters are thought to have formed in dense regions with low metal- licity (−2.5 ≤ [Fe/H] ≤ −1.5) and violent gas interactions (Bland-Hawthorn, Freeman, and Matteucci, 2014). Globular clusters typically have low enough masses and luminosi- ties that they are not detectable at their formation redshift (z≈ 3) in current observations, though they may be observable by the James Webb Space Telescope (Cowley et al., 2018). Predictions for the formation site of globular clusters in the context of cosmological mod- els span a range of possibilities: forming within high-density regions of galactic discs at z ≥ 2 (e.g., Kravtsov and Gnedin, 2005; Wisnioski et al., 2015; Elmegreen, 2018) or gas- rich dwarf galaxies at high redshift (Bekki, 2018), forming the first generation of globular cluster stars in mini-halos during significant merger events (e.g., Trenti, Padoan, and Jimenez, 2015), forming low-metallicity globular clusters in dwarf galaxy mergers (e.g., Lahén et al., 2019), or forming at the centre of primordial dwarf galaxies, producing the nucleated dwarf galaxies still seen today (e.g., Ricotti, Parry, and Gnedin, 2016)).

Milky Way globular clusters were long thought to be a perfect example of a single stellar population: stars formed at the same epoch out of the same well-mixed primordial cloud (e.g., Harris and Canterna, 1979; Mendel, Proctor, and Forbes, 2007). However, over the past few decades, our understanding of globular clusters has changed. More detailed Chapter 2. Globular clusters in GALAH 25 observations continue to uncover new kinds of complexity in them. Photometry in well- selected narrow and medium bands (e.g., Milone et al., 2010) shows multiple sequences from the lower main sequence through the red giant branch. Spectroscopic investigations have found that anticorrelated variations in the light elements (C, N, O, Na, Mg, Al) are extremely common (e.g., Carretta et al., 2009b; Mészáros et al., 2015; Johnson et al., 2017), with arguments for abundance variations in He, F, and Ne that are difficult to measure spectroscopically (D’Orazi et al., 2013a on F, Dupree and Avrett, 2013 on He), occasional correlations with s-process elements (Shingles et al., 2014, and references therein), and in a few cases, multiple groups in metallicity (e.g., Da Costa, Held, and Saviane, 2014). Furthermore, the existence of tidal tails escaping from some globular clusters require a more complex understanding of the internal dynamics (e.g., Sollima, 2020).

The internal properties of some Milky Way globular clusters, in particular stellar abun- dances, have been studied in great detail, with a goal of using present-day abundances. However, the number of globular clusters with in-depth abundance studies is still lim- ited. While there are a large number of small-sample studies, and even in-depth studies in single clusters (e.g., the massive ω Cen study discussed in Johnson et al., 2008; Johnson and Pilachowski, 2010; Johnson et al., 2020), differences in techniques or abundance scales can make it difficult to merge their data together in a sensible way. One major source for homogeneous globular cluster abundance data is Carretta et al.(2009a), which presents the abundances of Fe, Na, and O for 1409 red giant stars in 15 galactic globular clusters, based on an intensive observing campaign with VLT/FLAMES and UVES. Carretta et al. (2009b) extend that work with a homogeneous study of O, Na, Mg, Al, and Si for 202 red giants in 17 globular clusters also from UVES.

Large spectroscopic surveys can produce significant homogeneous data sets on globular cluster abundances. One example of such a contribution is the work of Masseron et al. (2019), who do abundance analysis for globular cluster red giants from the APOGEE sur- vey (Majewski et al., 2017) with the Brussels Automatic Code for Characterizing High ac- cUracy Spectra (BACCHUS) code. They use 885 red giants belonging to globular clusters in the northern sky to analyze the homogeneity of light and neutron-capture elements. This study was followed and expanded by Mészáros et al.(2020), who report abundances of light elements, α elements, the iron peak, and the neutron-capture elements Ce and Nd for 2283 red giant stars in 31 globular clusters from APOGEE-II (Ahumada et al., 2020) survey data. They investigate the intrinsic Fe spread, Al-Mg and N-C anticorrelations as a function of cluster mass, luminosity, age, and metallicity. From there, it can be assumed that the abundances tell the story of formation.

The broad array of abundances available from GALAH observations of globular clus- ter stars has the potential to add depth and detail to this earlier work, and the addition Chapter 2. Globular clusters in GALAH 26 of Gaia astrometric data means that our membership assignments are more certain than ever. This aspect of the GALAH survey data has not yet been explored in a publica- tion, and so in this Chapter we will assess the potential and reliability of the data set for globular cluster abundance studies. We will test the validity of GALAH’s reported quantities, consider the chemical abundances produced by the GALAH survey for the observed clusters, investigate how well the light element anticorrelations can be seen in the data set, and search for additional correlations or unexpected behaviour in the abun- dance data. Chapter 2. Globular clusters in GALAH 27

2.2 Stellar abundances in globular clusters

2.2.1 Light-element abundance variations

Globular clusters have been found to exhibit complexity and inhomogeneities in their chemical abundances, beginning with studies of luminous red giants (e.g., Osborn, 1971). More recent studies have demonstrated that almost all Galactic globular clusters contain multiple stellar populations (Piotto, 2009; Milone et al., 2017). To date, only two globular clusters have been identified that might have single stellar populations: Ruprecht 106 (Villanova et al., 2013) and IC 4499 (Walker et al., 2011). The presence of multiple stel- lar populations has been interpreted as the result of a self-enrichment process, with the abundance patterns of a later generation polluted by elements produced by particular sources in the earlier generation such as supermassive stars (Denissenkov and Hartwick, 2014), massive binaries (de Mink et al., 2009), intermediate mass stars in their AGB phase (Ventura et al., 2001), fast rotating massive main sequence stars losing mass (Decressin et al., 2007), and novae (Maccarone and Zurek, 2012).

The first population of stars in a globular cluster has abundance patterns similar to field stars at the same metallicity (Kacharov, Koch, and McWilliam, 2013; Schiavon et al., 2017). These are generally scaled-Solar abundances and an α enhancement of ≈ +0.3. The sec- ond population is where the “anomalies” are seen; these are enhancements in N, Na and Al coupled with depletions in C, O and Mg. He enhancements in the second population have been inferred from photometry, and Si enhancements are present only in the most massive and metal-poor clusters. Our understanding of this phenomenon has been built through a series of observational studies (e.g., Harris, 1974; Cohen, 1978; Cottrell and Da Costa, 1981; Norris et al., 1981). Norris (1987) found that a bimodal or multimodal CN distribution was common in globular clusters, where the strengths of CN (or NH) and CH molecular bands were anticorrelated, indicating that the N and C abundances are also anticorrelated. Denisenkov and Denisenkova (1990) proposed evolutionary mixing as the primary cause of the C and N inhomogeneities, and Cannon et al.(1998) also inter- preted C-N anticorrelations as a result of the convective dredge-up of processed material from deep within the stars themselves. However, the finding that non-evolved or scarcely evolved main sequence and main sequence turnoff stars also exhibit abundance anoma- lies (Gratton, Sneden, and Carretta, 2004; Briley, Cohen, and Stetson, 2004) challenged that idea, since those stars have not yet experienced internal mixing that would affect their surface C and N abundances. Over time, more light element variations became clear, expanding from the C-N anticorrelation to include O-Na and Mg-Al anticorrela- tions as well. However, the details of the variations differ from cluster to cluster (Kraft, 1994; Gratton, Sneden, and Carretta, 2004; Carretta et al., 2010a; Gratton, Carretta, and Chapter 2. Globular clusters in GALAH 28

Bragaglia, 2012; Mészáros et al., 2020) and the fraction of stars belonging to each popula- tion in any given cluster is not constant.

The details of the light-element abundance behaviour differs between clusters. Carretta et al.(2010b) found that the number of first-population stars on the red giant branch is usually about half the number of their second-population counterparts, but this varies. Campbell et al.(2013) demonstrated that metal-poor NGC 6752 has no Na-rich second- population AGB stars, suggesting that the abundance difference is enough to change the evolutionary path of those stars and prevent them from ascending the AGB. The [Na/Fe] abundance is commonly used to identify second-population stars, and [Al/Fe] can serve the same purpose. Mészáros et al.(2020) gives a thorough examination of the Al ratio in 31 clusters using data from the APOGEE survey. The Mg-Al anticorrelation is not present or clear in all globular clusters because it is heavily reliant on mixing with He-burning material (Bastian and Lardo, 2018). The anticorrelation is also affected by the fact that the Mg-Al cycle requires large temperatures >70 MK to operate, which can only be reached by the core of low metallicity polluters to destroy 24Mg efficiently (Mészáros et al., 2020). The Mg-Al anticorrelation in the yields of AGB stars is sensitive to the metallicity, so if AGB stars contribute to these abundance anticorrelations, we would expect the variations to be stronger at lower metallicity (Ventura et al., 2016).

Hydrogen-burning reactions at high temperature are the main suspect as the source for the light-element anticorrelations (Denisenkov and Denisenkova, 1990; Langer, Hoffman, and Sneden, 1993; Prantzos, Charbonnel, and Iliadis, 2017). However, not every anticor- relation happens in every environment since each reaction network requires a different burning temperature. For example, C-N needs T ≈10 MK, O-Na needs T ≈40 MK, Mg- Al requires T ≈70 MK, Al-Si requires T ≈80 MK, while K-Ca is up to T ≈180 MK. A self-enrichment model for globular clusters requires not only an environment with high enough temperatures for these burning processes to happen, but also that the products of the burning are brought up to the surface of the star by some mixing process and con- sequently lost to enrich the interstellar material by some mass loss mechanism. Some suggested sources for this enriching material are fast rotating massive stars (Decressin et al., 2007), intermediate mass AGB stars (Ventura et al., 2001), massive binaries (de Mink et al., 2009), and supermassive stars (Denissenkov and Hartwick, 2014; Gieles et al., 2018). In addition, to reproduce the observed abundance behaviour, the anticorrelations likely require mixing with material that has pristine chemical composition (e.g., Prantzos and Charbonnel, 2006). The diluting material may be left over from the formation phase (e.g., D’Ercole et al., 2008; Bekki et al., 2007), or it may come from binary stars (Vanbeveren, Mennekens, and De Greve, 2012) or even from less evolved single stars (Gratton et al., 2011). Chapter 2. Globular clusters in GALAH 29

Mg-Al anticorrelation

The Mg-Al anticorrelation is one of the light-element abundance variations found in most globular clusters, where Mg is depleted and Al is enhanced, for example, Mészáros et al. (2015) found that the second population members in globular clusters were Mg-depleted. However, the Mg-Al variation diminishes for the less massive or more metal-rich clusters (Carretta et al., 2009b; Ventura et al., 2016). The presence of this variation indicates the processing of material during H-burning by high-temperature proton capture reactions, namely the Mg-Al cycle. The Mg-Al cycle needs higher temperatures than the C-N-O and Ne-Na cycles, so it allows us to restrict the range of temperatures at which the polluted matter was exposed. Unlike O and Na, for which deep mixing can influence the surface content, the Mg and Al abundances of RGB stars definitively reflect the initial chemical composition of a star. Hence, no change in the surface abundance of these elements occurs during the RGB phase (Denisenkov and Denisenkova, 1990).

One favored model for the second population stars in globular clusters is that they were formed from the winds of AGB stars mixed with pristine gas in the cluster. Hot bottom burning affects the polluters of mass above v 3 M . Hot bottom burning is a nuclear activity that occurs during each AGB interpulse at the base of the convective envelope (Bloecker and Schoenberner, 1991). In determining the degree of nucleosynthetic pro- cessing and the chemical composition of the gas ejected into the interstellar medium, the temperature at the bottom of the convective envelope is the essential quantity. Mg is ef- ficiently destroyed by proton fusion at 100 MK, and the synthesis of Al starts. Part of the synthesised Al can also produce Si at higher temperature via proton capture. Figure 2.8 illustrates how the depletion of Mg and enrichment occur in Mg-Al fusion chains.

We will mainly observe and discuss the presence of this anticorrelation as we go further in Chapter2 and3.

2.2.2 Metallicity variations

A key observation in globular cluster abundances is that the metallicity does not corre- late with the light-element variations. Indeed, for the majority of clusters, the metallic- ity distribution is consistent with having a single value. However, for a small number of clusters, we find extended or multi-modal metallicity distributions. Da Costa (2016) (and references therein) show that ω Cen (NGC 5139), NGC 1851, NGC 5286, NGC 5824, NGC 6273, NGC 6656, NGC 6715, NGC 6864 and NGC 7089 have complex metallicity dis- 5 tributions and intrinsic [Fe/H] ranges. These clusters are also larger than 10 M , and it has been suggested that their peculiar metallicity and other abundance variations are the result of massive globular cluster formation (e.g., Parmentier et al., 1999). However, there 5 are globular clusters with masses > 10 M that do not exhibit internal heavy element Chapter 2. Globular clusters in GALAH 30 variations, such as NGC 2808 (Milone et al., 2015). It has also been hypothesized that these clusters originated as the nuclear star clusters of progenitor substructures that have been accreted into the Milky Way during merger events in the past. This interpretation is most accepted for NGC 6715, which has clear multiple populations photometrically and spectroscopically, and is located at the center of the Sagittarius dwarf galaxy being accreted by the Milky Way (Da Costa and Armandroff, 1995), and for NGC 5139, which has long been suggested to be the core of an accreted dwarf galaxy (e.g., Freeman, 1993; Da Costa and Armandroff, 1995; denBrok et al., 2014).

2.2.3 Neutron-capture abundance variations

Neutron capture processes are responsible for the production of elements heavier than Fe. There are two major nucleosynthetic processes of this type: the slow process (s-process) and the rapid process (r-process). There are also a few heavy isotopes that are synthe- sized through the proton process (p-process; Burbidge et al., 1957), and there is also some observational evidence for an intermediate neutron capture process (Hampel et al., 2016).

In the s-process, the flux of free neutrons is slow enough that nuclei are able to decay to stable forms through beta decay, and the relative abundances of the nuclei produced reflects the beta decay timescales. In the r-process, the neutron flux is 10 orders of mag- nitude higher, which drives all nuclei to the “magic number” of neutrons, where the nuclear shell is full and the cross section to capture another neutron is very small (N=20, 28, 50, 82). Further neutron capture is stalled until the nuclei are able to beta decay all the way back to a stable state, at which point the rapid neutron capture begins again and moves up to the next “magic number”. The “magic number” nuclei are major features in the r-process abundance pattern.

Free neutrons are astrophysically quite rare, as they have a decay timescale of roughly ten minutes. In order for any neutron capture processes to occur, there must be a consistent source of free neutrons. Since the r-process requires such a large flux of free neutrons, it is commonly associated with massive stars, cataclysmic events, and short timescales. Both core-collapse supernovae (type Ib, Ic and II) and neutron-star mergers have been proposed as sites for the r-process, and observations support that suggestion. The prod- ucts of r-process nucleosynthesis are regularly found in all Galactic components includ- ing very metal-poor stars. This prompt pollution indicates that the r-process occurs in stars that evolve very quickly, and that r-process products are distributed widely in the interstellar medium (Sneden, Cowan, and Gallino, 2008). As listed in Table 1.1, the r- process stars can be categorized as the moderately r-process-enhanced (r-I) and highly r-process-enhanced (r-II). The r-II stars display unusual amounts of r-process elements, having [Eu/Fe] more than 10 times the Solar ratio, while nearly 50% of halo stars r-I Chapter 2. Globular clusters in GALAH 31 stars, having [Eu/Fe] > 0.3 (Frebel, 2018). This classification used is for convenience as it helps to understand further the background of the enhancement because the primary nucleosynthetic site of the r-process is still not fully understood (Sakari et al., 2018).

In contrast, the s-process occurs in lower-mass stars, and so the s-process contribution to chemical evolution occurs on a slower timescale than the r-process. The s-process mainly happens during hot shell burning in asymptotic giant branch (AGB) stars, seeded by iron nuclei left by a predecessor supernova (Straniero, Cristallo, and Piersanti, 2014). AGB stars are also a potential site for producing light-element anticorrelations through high- temperature hydrogen burning, so possible correlations between s-process abundances and light-element abundances are a point of interest in the study of globular cluster abun- dances. The typical products of neutron capture nucleosynthesis are Sr, Y, Zr, Ba, La, Ce and Nd for the s-process, and Eu is mainly produced by the r-process.

In the context of globular clusters, the r-process abundances are typically similar to those observed in halo field stars. Variations in s-process abundances are not a typical feature of most of the globular clusters in the Milky Way, but a clear signature of the s-process has been found in some clusters. These include M4 (Yong et al., 2008; D’Orazi et al., 2013b), stars with [Fe/H] > −1.6 in NGC 5139 (Smith et al., 2000; Johnson and Pilachowski, 2010; D’Orazi et al., 2011), M22 (Roederer, Marino, and Sneden, 2011), NGC 1851 (Gratton et al., 2012), and M2 (Lardo et al., 2013). This additional complexity only deepens the mystery of multiple populations in globular clusters.

This thesis will examine the abundances for globular cluster stars that have been ob- served by the GALAH survey. The literature has multiple vocabularies for discussing abundance studies in globular clusters. We choose terminology that makes the fewest assumptions about the origins of the abundance anomalies: stars having field star-like abundances are “first population” or “primordial” stars, and stars with peculiar chemical compositions are referred to as “second population” or “enriched” stars. This designation has become the convention for classifying the multiple populations for globular cluster stars and is used in many related publications (e.g., Bastian and Lardo, 2018; Simpson et al., 2020). Chapter 2. Globular clusters in GALAH 32

2.3 Globular clusters observed by the GALAH survey

In this study we use the GALAH Survey as our data source. As discussed in Chapter1, GALAH is an Australian-led project to carry out a high-resolution spectroscopic survey of a million stars in the Southern sky. The science goals of GALAH, as laid out in De Silva et al.(2015), are to use stellar abundances and kinematics in an effort to unravel the history of star formation, chemical evolution, and accretion in the Milky Way.

From late 2013 through mid-2014 GALAH carried out a Pilot Survey that focused on globular clusters, the structure of the thin and thick disks, stars with asteroseismic data, and planet-hosting stars. Phase 1 survey operations ran from mid-2014 through mid- 2020, targeting all stars with 12 ≤ V ≤ 14, −80◦ ≤ δ ≤ +10◦, |b| ≥ 10◦, and an on-sky density of at least 400 stars per π square degrees, which is the size of the instrumental field of view. Phase 2 survey operations started in mid-2020, with a shift to prioritising stars near the main-sequence turnoff, where ages can be determined to better than 10% precision.

The second public data release from GALAH (Buder et al., 2018) is comprised of radial velocities, stellar parameters, and up to 23 elemental abundances for 342,682 stars. In this work we are using the GALAH iDR3 catalog, which is the private team version of the third data release. It contains radial velocities, stellar parameters, up to 30 elemental abundances, and probabilistically inferred stellar ages and masses for 652,799 stars.

The analysis and figures in this section were done using the Python scripting language with packages such as astropy (Astropy Collaboration et al., 2013; Astropy Collaboration et al., 2018), matplotlib (Barrett et al., 2005; Caswell et al., 2019) and with TOPCAT (Tay- lor, 2005), which provides useful data and table management and catalog matching, with simple but interactive plotting capabilities.

Globular clusters were targeted as one of the four main projects in the Pilot Survey (Martell et al., 2017) as an anchor for the abundance analysis, providing targets from a broad range of metallicity and providing an important sample for testing the abundance precision and detection limits in HERMES data. However, observing globular clusters with 2dF is challenging because the 2dF fibres cannot be placed closer together than 30 arcseconds on the sky. The HERMES spectrograph provides additional constraints on the apparent magnitudes, since it takes an hour of observing time to acquire a signal to noise ratio of 100 per resolution element for a 14th-magnitude target. To build a sufficiently large data set, given the distance moduli of the available clusters, Pilot Survey observa- tions in globular clusters were taken with the apparent magnitude range extended to V = 17, distinctly fainter than normal GALAH survey observations. Pilot Survey spectra for globular cluster stars are coadded across multiple observations, analysed with the Chapter 2. Globular clusters in GALAH 33

GALAH pipeline, and included in the iDR3 catalog. There is a fairly limited set of globu- lar clusters for which a sufficient number of stars as faint as the horizontal branch could be observed given these restrictions. One result of this is that we have not observed the most metal-rich bulge clusters or the outer halo clusters that are most likely to have been accreted from other galaxies. That would affect how many stars we would expect the clusters we are observing to have lost into the field, but not our expectations for the number of escaped stars in Chapter3, since the spatial and kinematic selection for our field stars differs significantly from the selection for our globular clusters.

2.3.1 Determining cluster membership

The GALAH Pilot Survey collected spectra for known or likely cluster members in NGC 104, NGC 288, NGC 362, NGC 1851 and NGC 5139. Targets were selected based on data from Zacharias et al.(2013), Carretta et al.(2011), Zacharias et al.(2010), Bellini et al.(2009), and Skrutskie et al.(2006)), Campbell (priv. comm.), Da Costa (priv. comm.), and Stetson (priv. comm.). As GALAH observing proceeded, additional cluster stars were observed serendipitously as part of regular survey fields. In this study we use both the intention- ally observed stars and the serendipitous stars to maximise our sample size. We apply a single cluster membership selection approach to the full iDR3 data set. This selection uses coordinates and proper motions from Gaia DR2 (Gaia Collaboration et al., 2018b) and radial velocities from the iDR3 catalog. For position and proper motion, we require stars to be similar to known catalogue values for the cluster mean, and for radial velocity we identify and select peaks in the distribution. Cluster mean values and the selection ranges are listed in Table 2.1.

Position cut

Our first requirement is that cluster members must be near a cluster on the sky. We use cluster centre, apparent radius, and proper motion values for each cluster from Kharchenko et al.(2013), which also compiles distances and ages for a comprehensive set of globular clusters in the Milky Way. All stars within the cluster radius r are kept as potential clus- ter members, and there are 12 clusters for which GALAH iDR3 stars pass this position selection: NGC 104, NGC 288, NGC 362, NGC 1851, NGC 4590, NGC 5139, NGC 5986, NGC 6362, NGC 6397, NGC 6541, NGC 6584, and NGC 7099. These clusters are shown in Figure 2.1 as large red dots and labeled with their NGC numbers. The smaller grey points mark the stars in the iDR3 catalogue, the K2-HERMES program (Sharma et al., 2019) and the TESS-HERMES program (Sharma et al., 2018). Cluster centre coordinates and radii are listed in Table 2.1. Chapter 2. Globular clusters in GALAH 34

Radial velocity cut

For our radial velocity selection, we consider the distribution of radial velocities from the GALAH iDR3 catalogue for stars that pass the position selection step. A cut is determined after visually comparing the radial velocity distribution inside the cluster radius with the radial velocity distribution for field stars around the cluster. If a peak in the radial velocity distribution can be attributed to the cluster, we select stars near that peak as potential cluster members. If there is no prominent peak relative to the surrounding field, but the cluster has a known radial velocity, stars close to that radial velocity are selected. Figure 2.2 shows the distribution of radial velocities for stars that pass the position cut in the 12 globular clusters previously mentioned. The velocity selection criteria for each cluster are given in Table 2.1.

Proper motion cut

Gaia DR2 delivers very precise proper motions, and cluster mean proper motions have been calculated by Vasiliev (2019). Our final globular cluster membership selection step is to take all stars in a reasonable range around the cluster mean in the (µα cos(δ), µδ) plane. The reasonable range is different for each cluster, depending on the precision of proper motion measurements in that part of the sky, the motions of foreground stars, distance to the cluster and more. For a nearby cluster, for example, a larger region in the proper motion plane must be used, because velocity perturbations inside the cluster can be resolved. The opposite holds for a distant cluster; perturbations are not resolved and the valid region in the proper motion plane is driven by measurement uncertainties. The mean and range in proper motion used for this step in cluster membership selection are given in Table 2.1.

Parallax cut

While all of our potential cluster member stars have measured parallaxes in Gaia DR2, we do not use parallax as a selection criterion. There are a few stars with negative parallax values, which is not unexpected in that catalogue although it is an unphysical result (Luri et al., 2018). Parallaxes can be unreliable for stars in globular clusters as they tend to be faint, relatively distant, and dense on the sky, all of which can contribute to confusion in the observation-to-source matching for the Gaia catalogue. Furthermore, the previous selections constrain the cluster members well enough that a parallax selection would not remove any candidates. Chapter 2. Globular clusters in GALAH 35 No. of Observ. members 1 − δ µ .( 2013 ) and Vasiliev ( 2019 ) mas y et al ) 1 δ − cos( α µ mas y 1 − r υ Mean Range Mean Range Mean Range and visual inspection of the data set. /° ° km s δ /° α Cen 201.697 -47.480 0.825 232.0 30.0 -3.21 1.50 -6.80 1.40 123 pilot 2.1: Membership criteria for the 12 globular clusters for which GALAH iDR3 has stars within the on-sky cluster ω ABLE T radius (half-light radius), based on catalogues of globular cluster data from Kharchenko Globular cluster Cluster centre Radius NGC ID Other name NGC 104NGC 288NGC 362 47 TucNGC 1851NGC 4590NGC 5139 6.004 -72.081 M68 0.950 13.192 189.867 -26.585 -18.0 15.825 -26.744 78.532 -70.847 0.310 -40.043 20.0 0.195 0.235 -44.2 0.225 5.22 -94.7 219.0 320.0 15.0 1.30 25.0 15.0 17.0 4.21 -2.45 -2.75 6.70 2.13 0.45 1.30 0.45 0.50 -5.66 1.00 1.76 -2.49 177 -0.67 0.40 0.40 0.45 0.80 3 pilot 1 1 2 serendipitous serendipitous pilot pilot NGC 5986NGC 6362NGC 6397NGC 6541NGC 6584NGC 7099 236.512 -37.786 262.978 M30 -67.048 265.176 0.165 -53.674 272.010 325.099 0.230 101.2 -43.715 247.656 -23.180 0.405 -14.2 -52.216 15.0 0.230 0.285 17.6 -164.0 0.155 -4.19 -185.2 4.0 260.6 15.0 15.0 -5.50 0.95 16.0 3.28 35.0 0.35 -4.60 -0.76 0.40 -0.05 1.00 -4.81 0.50 1.00 0.50 -17.54 2.00 -8.84 -7.20 0.40 1.00 -7.22 1 0.55 0.50 4 1.50 19 21 1 pilot 1 serendipitous pilot pilot serendipitous pilot Chapter 2. Globular clusters in GALAH 36 are shown as grey points.) 2.1: On-sky positions of the 12 globular clusters with GALAH iDR3 stars passing the positional match are shown IGURE F as large coloured. Stars from the full GALAH iDR3 catalogue, the K2-HERMES program, and the TESS-HERMES program Chapter 2. Globular clusters in GALAH 37

Other criteria

While the above selections are sufficient to determine cluster membership, we apply a second set of filters to choose the data set we will use for analysis. These focus on data quality and stellar evolutionary phase.

One of the outputs of the GALAH stellar parameter and abundance determination is a set of binary bitmask “flags” that indicate particular issues in the results. The overall data quality flag flag_sp can indicate a number of problems including unreliable Gaia astrometry, problems with continuum normalisation or sky subtraction, or whether the target is a double-lined spectroscopic binary. The flag on each abundance measurement flag_X_fe is used to indicate issues such as weak lines or poor spectrum synthesis fits. In all cases, an “unflagged” result, in which the relevant flag is set to 0, indicates that none of these problems have been encountered in the data. A full explanation of the flagging scheme is given in the GALAH DR3 paper (Buder et al., 2020).

For this study, we require flag_sp= 0, so that we know there are no obvious data prob- lems and the stellar parameters are reliable. Since we will be working with abundances, we require flag_fe_h= 0 for unflagged metallicity, and also flag_o_fe, flag_na_fe, flag_mg_fe, and flag_al_fe= 0 since these are the main elements for studying abun- dance anticorrelations in globular clusters.

Separate from the GALAH data quality flags, we also apply a selection in the signal to noise ratio of the original spectra. Specifically, we require the signal to noise ratio per pixel in the HERMES green channel to be at least 20. This selection is based on a trend we see in the Teff and [Fe/H] determination for NGC 104, which is shown in Figure 2.5 and discussed in the next Section. The Teff versus log g distribution for stars that pass this selection are shown in Figure 2.3, colour coded by the signal to noise ratio per pixel.

Finally, we choose to work only with red giant branch stars. This selection has a fairly minor effect on our data: the targeted Pilot Survey observations of globular cluster stars focused specifically on stars on the red giant branch and the horizontal branch, and any serendipitous observations of globular cluster members in regular survey fields would include only stars with apparent V magnitudes brighter than 14. Given the distance moduli of the 12 clusters in our initial selection, these serendipitously observed stars would have to be fairly luminous. We select only stars with Teff ≤ 5500 K and log g ≤ 4.0 for further analysis. The final column in Table 2.1 gives the number of stars in each cluster in our final data set.

Out of the 12 globular clusters observed by GALAH, there are four that have more than 5 member stars in the iDR3 catalog that meet all of our criteria: NGC 104, NGC 5139, NGC 6397, and NGC 7099 (M 30). Our analysis will focus on these clusters. Chapter 2. Globular clusters in GALAH 38

FIGURE 2.2: Histograms of radial velocity for the 4 globular clusters with members observed by GALAH. Chapter 2. Globular clusters in GALAH 39

FIGURE 2.3: Teff versus log g for stars in the 4 clusters observed by GALAH, with the colour variation showing the signal to noise ratio per pixel for each star. Chapter 2. Globular clusters in GALAH 40

2.4 Globular cluster abundances in the GALAH survey

Groups of stars that form at the same place and time are expected to have almost identical abundances, manifesting the homogeneity of the gas from which they formed (e.g., de Silva et al., 2009), but this expectation is only partially met by globular clusters. A few have extended or multimodal metallicity distributions, and almost all show correlated variations in the abundances of light elements such as C, N, O, Na, Mg, and Al, the origin of which are still not well understood. While more than two populations can be identified based on abundances in some clusters, our goal here is to differentiate the first population from stars with distinctive abundance patterns, and not to draw out details of the enriched stars.

In the upcoming sections we make a first investigation of the abundance behaviour of our four selected globular clusters in the GALAH data set, focusing on metallicity, light elements, and neutron capture elements. In terms of metallicity, we note the mean val- ues and dispersions in comparison with the literature. For the light element abundances, unfortunately, GALAH does not provide us with complete information for the stars in our data set. Nitrogen is not determined from GALAH spectra because there are no N features in the four wavelength bands of the HERMES spectrograph, and carbon and oxygen abundances are only rarely determined for stars on the red giant branch because the features are weak at low gravity. C and O (and a few other elements which will be discussed later in the upcoming subsections) mostly have flag_X_fe> 0 for their abun- dances. This indicates that those abundances are unlikely to be reliable, and prevents us from using them to investigate the C-N and O-Na variations in the GALAH globular clusters. We will therefore base our discussion of multiple abundance populations on Mg and Al abundances, and also evaluate whether there is evidence for Si variations in our data.

Since s- and r-process elements are sometimes also complex in globular clusters (James et al., 2004; Marino, 2013; Bekki and Tsujimoto, 2017), we will also analyse the abundances of s- and r-process in our data set. We will use the criteria that are shown in Table 1.1 in Chapter1, which use [Ba/Fe] for the s-process, [Eu/Fe] for the r-process and [Ba/Y] for the heavy over light s-process element ratio. Finally, we will consider the Li abundances of our cluster stars, which should be significantly depleted in the luminous red giant stars in our data set because Li is destroyed by proton capture at temperatures above 3 MK, especially if there is mixing between the stellar surface and the hot interior, as in red giants (Uttenthaler et al., 2012; Charbonnel et al., 2020).

As mentioned in Chapter1, abundance patterns in globular clusters are expected to fol- low certain anticorrelated patterns that resemble the imprint of feedback from hot hydro- gen burning in an earlier stellar generation (Bekki et al., 2007; Bekki, 2017). This chemical Chapter 2. Globular clusters in GALAH 41 pattern has become the main feature in identifying and distinguishing second population globular cluster members from field stars. Following this investigation of the expected abundance patterns, we will look at each cluster in turn, and describe any interesting be- haviour in any of the other reported abundances. These include the light odd-Z elements (K, Sc), α elements (Si, Ca, Ti), the iron peak (V, Cr, Mn, Co, Ni, Cu), the s-process (Sr, Y, Zr, Mo, Ba, La, Ce, Nd) and the r-process (Rb, Ru, Sm, Eu). Chapter 2. Globular clusters in GALAH 42

FIGURE 2.4: Metallicity as determined by GALAH and Carretta et al. (2009b) for the same stars. The plot shows a best-fit line in blue and a 1:1 line in red. GALAH appears to slightly overestimate [Fe/H] for metal- poor stars, and the scatter about the best-fit line has an RMS of 0.15 dex.

2.4.1 Metallicity

While the majority of Galactic globular clusters have narrow metallicity distributions consistent with a single value, a small number of them exhibit either extended or multi- modal metallicity distributions. Of the four clusters we are considering here, only ω Centauri has such a broad metallicity distribution, and it is known to be quite complex.

We begin with a comparison against the literature, using data from the comprehensive VLT study of Carretta et al., 2009b. There are eleven stars in common between the two data sets: five in NGC 104, one in NGC 288, two in NGC 6397 and three in NGC 7099. Figure 2.4 shows that there is a fairly good correlation between the two sets of metallic- ity values, with an RMS scatter of 0.15 dex. Because GALAH captures relatively small pieces of the full optical wavelength range, abundance determination is more difficult for metal-poor stars than for metal-rich stars, and this can be seen as GALAH tending to overestimate metallicity relative to the literature values at the metal-poor end. Chapter 2. Globular clusters in GALAH 43

Another check on the reliability of our results is shown in Figure 2.5, which shows [Fe/H] versus Teff for the stars in each cluster, colour coded by the signal to noise ratio per pixel.

We expect a range in metallicity for NGC 5139, and a range in Teff for all four clusters. In- deed, we find that in the GALAH sample, NGC 104 has an average metallicity of [Fe/H] = −0.72 ± 0.09, NGC 5139 has an average metallicity of -1.52 ±0.33, NGC 6397 has a mean metallicity of -2.00 ±0.09, and NGC 7099 has a mean metallicity of -2.19 ±0.13. These are all consistent within 1σ with the cluster metallicities listed in the literature, in- cluding Harris (2010), Boyles et al.(2011), Carretta et al.(2009b), and Cordero et al.(2014), and Thygesen et al.(2014). The large range in metallicity for NGC 5139 is well known, and is understood as the result of significant self-enrichment from by numerous episodes of star formation in the cluster (e.g., Ikuta and Arimoto, 2000; Tsujimoto and Shigeyama, 2003; Marcolini et al., 2007; Johnson and Pilachowski, 2010).

However, in NGC 104 we see a correlation between [Fe/H] and Teff, which is not seen elsewhere in the literature. Since Figure 2.5 uses the signal to noise ratio per pixel as the colourmap, we can see that it is also correlated with Teff. This is reasonable as an out- come of our observing procedure, which assigns the same exposure time to stars across a range of apparent magnitudes, and should therefore return the lowest-quality data for the hotter stars lower on the red giant branch. However, we did not expect this to have such a clear and detrimental effect on the derived stellar parameters.

The trend between [Fe/H] and Teff might be caused by a mismatch between photometric and spectroscopic temperatures during the survey pipeline analysis. This could system- atically bias the log g estimate and thus also systematically shift the [Fe/H]. To try to understand this better, in Figure 2.6 we zoom in on the [Fe/H] versus Teff plane and di- vide the stars into three bins in effective temperature. Stars cooler than 4500K are shown as red dots, stars between 4500K and 5000K are in black, and stars hotter than 5000K are drawn with blue dots. This highlights the problem in NGC 104, and also shows that there are less dramatic effects at work in NGC 6397 (where a few hot stars trail off to high metallicity) and NGC 7099 (where a few cool stars extend to low metallicity). Figure 2.7 shows the same figure for NGC 104, made with data from Carretta et al.(2009a), and there is no correlation at all between [Fe/H] and Teff.

We must conclude that there is some systematic error in the GALAH analysis procedure for bright red giants that the survey team should address.1

1This work is based on a preliminary version of the GALAH data analysis. After this thesis was com- pleted, the systematic error in stellar parameters shown in this section was discussed within the GALAH team and found to be an artifact of inaccurate distance estimates. As described in the GALAH DR3 paper (Buder et al., 2020), probabilistic distance estimates from Bailer-Jones et al.(2018) are used as prior infor- mation for GALAH stellar parameter determination. The Bailer-Jones distances are less accurate for more distant stars, and we find quite a large scatter in the distance estimates for the globular cluster stars in GALAH. This is partly because they are distant and partly because they do not follow the smooth spatial Chapter 2. Globular clusters in GALAH 44

The rest of the abundance analysis is carried out relative to the iron abundance, so this temperature-dependent offset in the metallicity will carry through the rest of our analy- sis. One of the effects this will have on our analysis will be a larger scatter in metallicity, which will affect our target selection in Chapter3 and Chapter4. While the accuracy of the metallicity is not extremely important in this Chapter, we continue the Teff-based colour coding in the comprehensive figures showing all elemental abundances versus [Fe/H] for the four globular clusters in our data set (Figures 2.17, 2.18, 2.19, 2.22, 2.23, 2.24, 2.27, 2.28, 2.29, 2.31, 2.32 and 2.33). Another effect, which will be important to keep in mind in all three Chapters, is an increased scatter in the individual elemental abun- dances. This will make our division of the globular cluster data into first and second populations less robust, but it still might be accurate. However, this is only a problem for globular cluster stars in this analysis and not for field stars. While we are not as confident in our separation for first and second populations, it is a solved problem in the DR3 data set (Buder et al., 2020).

distribution used as a prior by Bailer-Jones. Replacing the Bailer-Jones distances for individual globular cluster stars with globular cluster distances from Baumgardt et al.(2019), we find that the systematic error in stellar parameters is resolved. Read section 5 of Buder et al.(2020) for more details. Chapter 2. Globular clusters in GALAH 45

FIGURE 2.5: Plot of [Fe/H] against surface temperature Teff, colour coded by the signal to noise ratio per pixel, for globular cluster stars in GALAH. Chapter 2. Globular clusters in GALAH 46

FIGURE 2.6: A zoomed-in look at [Fe/H] versus Teff, with stars divided into three bins in effective temperature. NGC 104 shows the clearest trend between [Fe/H] and Teff, but there are indications of similar trends in the smaller data sets for NGC 6397 and NGC 7099. Chapter 2. Globular clusters in GALAH 47

FIGURE 2.7: Plot of [Fe/H] against Teff for NGC 104, using data from Car- retta et al., 2009a. Here, there is no correlation between the two quantities, suggesting an accurate spectroscopic analysis. Chapter 2. Globular clusters in GALAH 48

FIGURE 2.8: Schematic of the Ne-Na and Mg-Al fusion chains showing the nuclei involved in the process and the products, including Si. The dashed lines indicate possible leakages in out of the chains and the dashed cir- cles represent unstable isotopes. This figure is a colorized version based on similar figures from Mowlavi and Meynet (2000) and Karakas and Lat- tanzio (2003).

2.4.2 Light elements

Although we talk in general terms about all globular clusters exhibiting anticorrelations in the light element abundances, there is some variety in the observed abundance pat- terns. While the C-N and O-Na anticorrelations are clearly visible for most clusters (e.g., Carretta et al., 2009a; Carretta et al., 2009b; Gratton, Carretta, and Bragaglia, 2012), we are not able to reliably use C, N, or O abundances in GALAH data, as discussed earlier in this Chapter. We will instead use the Mg-Al plane to evaluate how well GALAH data can dis- tinguish the stellar populations in globular clusters. The Mg-Al anticorrelation has been reported in a few clusters such as NGC 5139, NGC 7078, and NGC 6341 by Mészáros et al. (2020) and Carretta et al.(2009a), and Shetrone (1996) reported that some metal-rich clus- ters have a single [Al/Fe] value while the variation in the Al content spans a large range in others. MacLean et al.(2018) find a Mg-Al anticorrelation for both AGB and RGB stars in NGC 6397, which indicates that second-population stars continue to that later stage of stellar evolution, in contrast to the findings of Campbell et al.(2013).

Al and Mg are processed together through the Mg–Al cycle as shown in Figure 2.8, which steadily depletes Mg and produces Al over time. However, stars need high temperatures Chapter 2. Globular clusters in GALAH 49

FIGURE 2.9: [Al/Fe] versus [Mg/Fe] for globular cluster stars in our data set as blue circles and data from Mészáros et al.(2020) in the background in grey circles. The horizontal line indicates [Al/Fe] = 0.30 dex. Chapter 2. Globular clusters in GALAH 50

FIGURE 2.10: The histogram of [Al/Fe] distribution for our four clusters in 0.05 dex bins. The colours denote the first and second populations: the first population with [Al/Fe]<0.3 is shaded red and the second population having [Al/Fe]>0.3 is coloured blue. Chapter 2. Globular clusters in GALAH 51

(> 70 MK) to start the Mg-Al cycle, and these temperatures require particular environ- ments such as the interiors of massive stars or hot shell burning in AGB stars. Meanwhile, the Ne-Na cycle only requires ≈40 MK to operate, while the CNO cycle only needs ≈10 MK. As a result, the C-N, O-Na, and Mg-Al anticorrelations can operate in different astro- physical sites, and their contributions to chemical feedback are not necessarily coupled, in globular clusters or in any environment. Hence, the high [Al/Fe] abundances in some globular cluster members indicate that a previous generation of higher-mass or evolved stars must have contributed to their chemical composition. Since not all globular clusters exhibit Mg-Al anticorrelations, our ability to identify the multiple populations in those clusters will be limited, but we will be to be able to see how well the GALAH abundances replicate literature results, and whether the existing Mg-Al-based population classifica- tions are useful for our data.

Mészáros et al.(2015) used an extreme deconvolution method (Bovy, Hogg, and Roweis, 2011) to divide globular cluster stars into populations based on their light-element abun- dances, and found that Al abundance by itself is effective for separating the cluster mem- bers into the first and second populations. The follow-up study by Mészáros et al.(2020) features a large sample of stars in most of the observed clusters, and uses a simple [Al/Fe] limit to distinguish the populations, based on density maps and histograms in the abun- dance space. Classifications from this more straightforward method agree with the pre- vious method, and so we also adopt their simple separation between the first and second populations at [Al/Fe] = 0.3.

Figure 2.9 shows [Al/Fe] versus [Mg/Fe] for our four globular clusters, with GALAH abundances shown as larger blue circles and data from Mészáros et al.(2020) in the back- ground in grey circles, and the population dividing line at [Al/Fe]= 0.3. In NGC 104, neither we nor Mészáros et al.(2020) see a significant depletion in Mg, but we do both see a range in Al abundance. In our data set there are 62 stars (43%) that we classify as first population, with [Al/Fe]< 0.3, and the remaining 81 (57%) are in the second population. NGC 5139 shows a weak anticorrelation pattern with a large scatter in [Mg/Fe] for stars with [Al/Fe] > 0.8 dex. NGC 5139 occupies a wide range in Mg-Al abundance space in both data sets, and there may be a zeropoint offset between GALAH Mg abundances and those from Mészáros et al.(2020). Meanwhile, in NGC 6397 the stars we have observed are clearly consistent with the second population in the Mészáros et al.(2020) data, and unfortunately NGC 7099 is not included in Mészáros et al.(2020), but Carretta et al., 2009b did not find any Mg-Al anticorrelation in that cluster. Given its similar metallicity to NGC 6397, its stellar populations should be similarly arranged in the Mg-Al abundance plane, and so the stars we observe in NGC 7099 are consistent with belonging to the sec- ond population. Given the difficulty in determining abundances for metal-poor stars in GALAH, it is possible that first-population stars were also observed in NGC 6397 and Chapter 2. Globular clusters in GALAH 52

NGC 7099 by GALAH, but they do not have reliable Mg and Al abundances.

Since our division into first- and second-population stars is simply a function of the alu- minium abundance, we can use histograms of [Al/Fe] to understand the distribution of aluminium abundance and the ratio of the two populations in each cluster. Figure 2.10 shows these histograms, and we see that all four globular clusters exhibit continuous distributions in [Al/Fe]. We agree with Mészáros et al.(2020) that there are multiple populations in NGC 104 based on the range in [Al/Fe] spread, and we find a larger frac- tion of first-population stars than they do. We find a larger range in [Al/Fe] for the two more metal-rich clusters, while NGC 6397 and NGC 7099 have narrower [Al/Fe] distri- butions centered near +1.0 dex. Although it is possible that sensitivity limits prevent us from recognising first-population stars in these two clusters, the presence of Al-rich stars supports the finding by Mészáros et al.(2015) that second-population stars in metal-poor clusters have been heavily affected by high mass, low metallicity AGB polluters.

There are two stars in NGC 5139 that lie in an unexpected region of the [Mg/Fe] versus [Al/Fe] abundance space. These stars, which are highlighted in Figure 2.9 with larger red circles, have [Al/Fe] = 0.54 dex and 0.33 dex, and [Mg/Fe] = -0.66 dex and -0.48 dex respectively. This strong depletion in Mg coupled with enhancement in Al might be caused by very hot proton-capture nucleosynthetic processes occurring above 80 MK temperatures. As explained by Prantzos, Charbonnel, and Iliadis (2017) and observed by Masseron et al.(2019), the depletion of Al occurs when the temperature reaches above 80MK, Si begins to be produced, and the enhancement of Si in the Mg-Al cycle might be observed. This possibility leads to our next investigation.

2.4.3 An extension to silicon

If the Mg-Al fusion cycle is operating at a high enough temperature, we should see the production of Si through “leakage” (Karakas and Lattanzio, 2003) along with the deple- tion of Mg and production of Al. We would therefore expect the abundance of Si from the Mg-Al cycle to correlate with Al and anticorrelate with Mg, and this is observed in NGC 5139 by Mészáros et al.(2020). To investigate whether there is any evidence for Si production in globular clusters in the GALAH data, we produced Figure 2.11 and Fig- ure 2.12. Looking at these two figures, we do not see any clear evidence for these particu- lar correlations. The two stars we noted as having particularly low [Mg/Fe] abundances are not enriched in [Si/Fe], and while there are two stars in NGC 5139 with unusually high [Si/Fe] abundances, they also have fairly high [Mg/Fe] abundances. There is a hint of an Al-Si correlation in NGC 104, and we recover the Al-Si correlation in NGC 5139, both of which are also reported by Mészáros et al.(2020). Chapter 2. Globular clusters in GALAH 53

FIGURE 2.11: Si-Mg variation for globular cluster observed by GALAH. Chapter 2. Globular clusters in GALAH 54

FIGURE 2.12: Al-Si variation for globular cluster observed by GALAH with red line indicating [Al/Fe] = 0.30 dex. Chapter 2. Globular clusters in GALAH 55

For the two metal-poor clusters it is difficult to say whether there is any correlation be- tween Si and Mg or Al. NGC 7099 shows a significant range in [Si/Fe], but it is not ac- companied by Mg depletion, and there are only 6 stars in the cluster with determined Al and Si abundances. We infer that all of the stars we observe in NGC 6397 and NGC 7099 are in the second population, and so we do not have sufficient data to see a difference between the first and second populations. Even if we were sampling from both popula- tions, such a small number of stars makes it difficult to make strong statements about the presence or absence of structure in the data set.

2.4.4 Neutron-capture elements

The s-process and r-process neutron-capture elements are produced in different astro- physical sites under different conditions. For globular clusters, r-process elements are expected to be enhanced uniformly just like halo field stars, but any explanation of glob- ular cluster abundance anomalies that involves chemical feedback from AGB stars has the potential to also include s-process elements in the abundance correlations. Indeed, there are reports on a few enriched s-process stars in globular clusters (e.g., D’Orazi et al., 2010a).

To investigate the abundance behaviour of the neutron capture elements in the globu- lar clusters, we created Figures 2.13, 2.14, and 2.15, which show the Mg-Al abundance plane colour-coded by [Eu/Fe], [Ba/Fe], and [Ba/Y] respectively. In these figures, we only include stars with unflagged abundances of Eu, Ba, or Y, as needed for the figure. Our data set only has [Eu/Fe] abundances for a significant number of stars in NGC 104 and NGC 5139, and since the stars in NGC 6397 and NGC 7099 appear to all belong to the second population, they are not very useful when searching for population-based abundance differences.

For NGC 104 we do not see any meaningful correlation between position in the Mg- Al plane and [Eu/Fe], but we do find that the first-population stars show a higher Ba abundance than the second-population stars. Looking at the [Ba/Y] ratio, which gives the ratio of light to heavy s-process elements, we find a slight difference between the first- and second-population stars in NGC 104, with a mean [Ba/Y] ratio of +0.12 ± 0.26 in the first population and a mean [Ba/Y] = −0.03 ± 0.28 in the second population. However, there are large uncertainties in both populations with considerable overlap. These data are suggestive of a difference in [Ba/Y] between first- and second-population stars, but we cannot make a robust claim.

NGC 5139 seems to divide into three groups in [Al/Fe] when looked at through the lens of neutron-capture abundances: below [Al/Fe]= 0.30 and above [Al/Fe]= 0.9, [Eu/Fe] is lower and [Ba/Fe] is higher than in the range 0.3 <[Al/Fe]< 0.9. This is a consequence Chapter 2. Globular clusters in GALAH 56 of the greater complexity in NGC 5139, where it is not just ordinary globular cluster abundance inhomogeneities at work but also galaxy-like self-enrichment (Johnson and Pilachowski, 2010). The figures also show that first population stars are enhanced in s-process elements, while there are stars with both high and low abundances of r- and s- process elements in the second population. While there is a range in [Ba/Y] in NGC 5139, there is not a clear correlation with position in the Mg-Al plane.

There are only 4 stars in NGC 6397 for which we have unflagged (reliable) [Eu/Fe] abun- dances, and that abundance is significantly enhanced in two of them. As for s-process el- ements, there are 9 stars with unflagged measurements for both [Y/Fe] and [Ba/Fe], and those abundances are mostly sub-Solar. The [Ba/Y] abundance ratio tends to be slightly super-Solar, indicating that light s-process elements are dominant over their heavy coun- terparts.

Meanwhile for NGC 7099, there is only one star with an unflagged [Eu/Fe] abundance, and it is quite enhanced. The five unflagged [Ba/Fe] abundances all fall between ±0.2, and the [Ba/Y] ratio is typically negative, with a mean of ≈ −0.05, indicating that heavy s-process elements are more abundant than light s-process elements.

2.4.5 Lithium

Figure 2.16 shows A(Li) as a function of [Fe/H] for the sample. In contrast to the abun- dance notation for other elements, Li abundance is typically expressed as A(Li) = log(NLi/NH)

+ 12, where NLi and NH are the the number densities of lithium and hydrogen, respec- tively (Bonifacio and Molaro, 1997; Žerjal et al., 2019). This is because the surface Li abundance of the Sun has been depleted from its original level by slow circulation be- tween the surface and the interior (Greenstein and Richardson, 1951; Blöcker et al., 1998; Thévenin et al., 2017). Hence, the Sun is not a meaningful zeropoint for Li abundance the way it is for the other elements.

The lithium abundance is generally quite low in giant stars (de La Reza, 2000), because it is destroyed through proton capture at a temperature of only 2.5 MK. Lithium depletion occurs on the pre-main sequence, the main sequence, at first dredge-up, and as stars evolve along the red giant branch. The primordial lithium abundance in the Universe is A(Li)≈ 2.7 (D’Antona et al., 2012), and typical A(Li) for globular cluster giants is 1.5 or less (Bonifacio and Molaro, 1997; D’Antona et al., 2012).

In our sample, the lithium abundance in NGC 104 correlates with [Fe/H], with a mean value of A(Li) = 0.63 and a standard deviation of 0.49. This is likely to be a consequence of the artificial Teff - [Fe/H] correlation mentioned in Section 2.4.1, and not a real pattern. Chapter 2. Globular clusters in GALAH 57

FIGURE 2.13: The variation of the r-process element Eu on the Mg-Al plane for our globular cluster stars. There is not a clear trend in [Eu/Fe] with the cluster populations in NGC 104, but NGC 5139 has a shift in [Eu/Fe] at the [Al/Fe]= 0.3 dividing line between the first and second population. Chapter 2. Globular clusters in GALAH 58

FIGURE 2.14: The variation of the s-process element Ba on the Mg-Al plane for our globular cluster stars. [Ba/Fe] is higher in the first-population stars in NGC 104 and NGC 5139, but results are more ambiguous for the clusters with only second-population stars. Chapter 2. Globular clusters in GALAH 59

FIGURE 2.15: The variation of the light- to heavy-s-process ratio [Ba/Y] on the Mg-Al plane for our globular cluster stars. [Ba/Y] is slightly higher in the first-population stars in NGC 104, shows a wide range in both popula- tions in NGC 5139, is positive in the second-population stars in NGC 6397, and is negative in the second-population stars in NGC 7099. Chapter 2. Globular clusters in GALAH 60

The mean A(Li) value we find for NGC 104 is low compared to the results in Pasquini and Molaro (1997) and D’Orazi et al.(2010b) and Dobrovolskas et al.(2014).

In NGC 5139, most stars have lithium abundances below A(Li)= 1, and the mean abun- dance is A(Li) = 0.57 ± 0.36. As comparison, the average lithium presented in the litera- ture is A(Li) = 2.19 ± 0.14 dex (Monaco et al., 2010) and stars with A(Li) v 1 dex are found at lower metallicities than [Fe/H] = –1.3 dex and diminish at higher metallicities (Muccia- relli et al., 2018). However, there are two stars with A(Li) > 1.5. These stars could either be lithium-rich giants (e.g., Casey et al., 2019b), which have produced or accreted a large amount of lithium in their atmosphere, or they could be the result of lithium production as part of self-enrichment in the NGC 5139 system before its capture into the Milky Way. Interestingly, we do not see the Li-Na anticorrelation or the Li-O correlation reported by Monaco (2012) and Mucciarelli et al.(2018) in our data set, which might be an artifact of the upper limit flag for [Li/Fe] during GALAH analysis pipeline.

There are many previous studies of lithium abundance in NGC 6397, including Pasquini and Molaro (1996), Bonifacio et al.(2002), González Hernández et al.(2009), and Primas et al.(2010), and Mucciarelli, Salaris, and Bonifacio (2012). The stars near the main-sequence turn-off in NGC 6397 are reported to be rich in Li (e.g., Koch et al., 2012), and the Li deple- tion along the giant branch is very precisely charted by Lind et al.(2009). Our measured values for A(Li) in GALAH have a mean of 0.9 and a standard deviation of 0.3, which is entirely consistent with the literature.

We find a low lithium abundance of A(Li) = 0.82 ± 0.12 in NGC 7099, in keeping with expectations from stellar evolution models. However, with just 5 stars in the sample, it is difficult to compare to larger studies like Gruyters et al.(2016), which follows the evolution of A(Li) from the turnoff to the tip of the red giant branch. Chapter 2. Globular clusters in GALAH 61

FIGURE 2.16: Li abundance A(Li) as a function of [Fe/H] for globular clusters in the GALAH samples. The stars are divided into effective tem- perature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Chapter 2. Globular clusters in GALAH 62

TABLE 2.2: Mean value of the chemical abundances in NGC 104.

Element Mean σ Element Mean σ [α/Fe] 0.28 0.07 [Co/Fe] 0.14 0.23 A(Li) 0.63 0.49 [Ni/Fe] 0.03 0.10 [C/Fe]* 0.48 - [Cu/Fe] -0.04 0.14 [O/Fe] 0.53 0.26 [Zn/Fe] 0.25 0.29 [Na/Fe] 0.25 0.16 [Rb/Fe] 0.36 0.29 [Mg/Fe] 0.35 0.10 [Sr/Fe] 0.84 0.23 [Al/Fe] 0.31 0.13 [Y/Fe] 0.19 0.22 [Si/Fe] 0.29 0.10 [Zr/Fe] 0.23 0.21 [K/Fe] 0.18 0.15 [Mo/Fe] 0.25 0.21 [Ca/Fe] 0.24 0.11 [Ru/Fe] 0.51 0.28 [Sc/Fe] 0.14 0.09 [Ba/Fe] 0.24 0.24 [Ti/Fe] 0.25 0.09 [La/Fe] 0.26 0.16 [V/Fe] 0.17 0.25 [Ce/Fe] -0.11 0.25 [Cr/Fe] -0.09 0.12 [Nd/Fe] 0.42 0.18 [Mn/Fe] -0.26 0.12 [Sm/Fe] 0.15 0.23 [Eu/Fe] 0.34 0.12 (*) sample of stars < 5

2.4.6 The clusters in detail

The globular clusters observed by GALAH can be divided into multiple populations based on their [Mg/Fe] and [Al/Fe] abundances. These multiple populations show some correlation with the abundances of the neutron-capture elements, and the lithium abun- dances in the stars are low, as we would expect for stars on the red giant branch. In this section we will look at [X/Fe] versus [Fe/H] for all of the measured abundances in each cluster that are flagged by the GALAH analysis software as reliable, and discuss individ- ual elements or nucleosynthetic families with interesting or unexpected behaviour.

NGC 104

NGC 104, or 47 Tucanae, is a well-studied globular cluster, and the literature provides many previous studies to compare our results against. We list the mean and standard deviation in abundance for each element in Table 2.2, listing A(Li) instead of [Li/Fe], and not giving a standard deviation value for [C/Fe] because we only have an unflagged (reli- able) carbon abundance for one star in NGC 104. The abundances of stars in NGC 104 for [α/Fe] and 29 individual elements in the GALAH survey are shown in Figure 2.17, 2.18 and 2.19. As mentioned in the previous subsection, the distribution of metallicity in this particular cluster exhibits a correlation with Teff, and this affects the derived abundances both through an inaccurate value of Teff and through an inaccurate value of [Fe/H] when Chapter 2. Globular clusters in GALAH 63

FIGURE 2.17: Abundances for α and 8 other individual light elements in NGC 104. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Li abundance for all globular clusters are displayed in Figure 2.16. Chapter 2. Globular clusters in GALAH 64

FIGURE 2.18: Abundances for 9 individual iron-peak elements in NGC 104. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Chapter 2. Globular clusters in GALAH 65

FIGURE 2.19: Abundances for 12 individual n-capture elements in NGC 104. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Chapter 2. Globular clusters in GALAH 66 forming the ratio [X/Fe]. However, the issue can be settled by resolving the systematic problem.1

As we have noted previously, the oxygen abundances from GALAH for luminous red giants in globular clusters are often flagged as unreliable. In addition, we find that even the unflagged abundances are offset to higher values than we would expect from the literature. The unflagged oxygen abundances are included for completeness in Figures 2.17, 2.22, 2.27, and 2.31, which show all unflagged abundances versus [Fe/H].

The abundance of [α/Fe] shows a narrow distribution, with a mean of 0.28 and a stan- dard deviation of 0.07, and 99% of stars falling in the range 0 <[α/Fe]< 0.5. This result is in agreement within the errors with the literature, where [α/Fe] ≈ 0.4 dex (e.g., Carretta et al., 2004; Fulbright, McWilliam, and Rich, 2007; Koch and McWilliam, 2008; Cordero et al., 2014). It is also in agreement with the abundances of the individual α elements Si, Ca, and Ti.

The light odd-Z elements K and Sc behave rather differently to each other, with [K/Fe] showing a positive correlation with [Fe/H] and a wide scatter and [Sc/Fe] showing scat- ter but no trend. Our mean K abundance is consistent with the [K/Fe]≈ 0.14 reported by Carretta et al.(2013) for NGC 104. Sc is expected to exhibit Solar-scaled abundance in globular clusters and halo stars (e.g., Johnson and Pilachowski, 2010). However, the abundance of [Sc/Fe] in the GALAH sample for NGC 104 appears to be higher, with a mean of 0.14 and a standard deviation of 0.09.

The iron peak elements V, Cr, Mn, Co, Ni, and Cu, show a close trend in abundance with the metallicity. This is what we expect, since their abundances should follow the metallicity closely. We find that [V/Fe] and [Co/Fe] both have a slightly positive mean value and a larger scatter than we would expect, [Cr/Fe], [Ni/Fe], and [Cu/Fe] are all consistent with a value of 0, with the average value of [Ni/Fe] being ≈ 0.14 dex higher than the average of 12 samples described in Thygesen et al.(2014), and a small scatter and slightly negative mean value for [Mn/Fe].

We also see a rather large dispersion in [Y/Fe], [Zr/Fe], [Ba/Fe], and [La/Fe] in NGC 104, indicating that some of the stars in the cluster have been enhanced in s-process elements. The mean abundances of these four elements are also higher than the compiled mean abundances from Marsakov, Koval’, and Gozha (2019), which we interpret as evidence of significant s-process activity in the cluster.

Mo and Ru are elements created by three different processes, the r-, s- and p-process. In our data set, these elements are measured reliably only in cool stars. The mean value for [Mo/Fe], across the 29 stars for which we have an unflagged measurement, is 0.25. This is 0.25 dex higher than the value reported in Thygesen et al.(2014) for a sample of 13 Chapter 2. Globular clusters in GALAH 67 red giants in the same cluster, and also higher than the mean value in Marsakov, Koval’, and Gozha, 2019. Meanwhile for Ru, the two stars hotter than Teff > 4500 K that we have abundances for both seem to be highly enhanced, with [Ru/Fe] values of +1.63 and +1.35. The mean [Ru/Fe] we find for the cluster is +0.51, which is dragged upward a little by the two hot stars, but is still consistent with the results of Thygesen et al.(2014).

Figure 2.20 shows the trends between [Ba/Fe], [Eu/Fe] and the [Ba/Eu] ratio, and it can be seen that NGC 104 is a mixture of stars with normal and enhanced s- and r-process abundances. 32.76% (58/177) of the stars fall within the range 0.0 <[Ba/Eu]< 0.5, which indicates that they are considered r/s enhanced, and 33.33% (59/177) of the stars in the sample are in the abundance range 0.3 <[Eu/Fe]< 1.0 and [Ba/Eu]< 0.0, fulfilling the r-I star criteria. None of the stars have [Ba/Fe]> 1.0 or [Eu/Fe]> 1.0, so we do not find any extremely s-rich or r-II stars in our sample. The large fraction of r-process enhanced stars and low number of extremely s-process enhanced stars in NGC 104 supports the results of James et al.(2004) and Cordero et al.(2015) for the majority of the cluster members.

To take a closer look at the s-process abundances, we plot both [Ba/Fe] and [Y/Fe] against [Al/Fe] in Figure 2.21. Here we see a mild but statistically significant Ba-Al anticorrela- tion, which disagrees with D’Orazi et al.(2010a), who found no variation in the [Ba/Fe] abundance for different stellar populations in globular clusters. There is not a clear an- ticorrelation in the Y-Al plane, but the second-population stars tend to have lower Y abundance than the first-population stars, as we discussed earlier.

The enhancement in [Ba/Eu], which reflects the ratio of s-process to r-process, displays which neutron-capture process is dominant in those stars as they are influenced by metal- licity and neutron bombardment. As mentioned before, 32.76% (58/177) of the stars are within the range 0.0 <[Ba/Eu]< 0.5, which indicates that they are considered r/s en- hanced or double enhanced. The r/s enhanced stars are usually associated with carbon enhanced metal poor (CEMP) stars (e.g., Zhang et al., 2013, and references therein). Since we could not reliably use C in our analysis, we could not properly check their CEMP-r/s status. However, the double enrichments of r-process and s-process elements have been proposed to originate from electron-capture supernova when the cores of massive AGB stars of 8−10 M collapse at the end of the AGB stage (Truran et al., 2002; Barbuy et al., 2005; Abate, Stancliffe, and Liu, 2016; Kobayashi, Karakas, and Lugaro, 2020). The most popular scenario is that the enhancement in r- and s-elements in r/s enhanced stars orig- inate from two independent sources and the current enhanced r/s stars are born in an environment which was polluted in r-elements from nearby supernovas. Meanwhile, the observed s-elements are transferred by stellar winds from AGB companion stars (Abate, Stancliffe, and Liu, 2016, and references therein). Chapter 2. Globular clusters in GALAH 68

FIGURE 2.20: The abundance of the neutron-capture elements Ba and Eu versus the [Ba/Eu] ratio in NGC 104, showing the variation of the relative amounts of s- and r-process enrichment in the cluster members. The lines at [Ba/Fe]= 0 and [Ba/Eu]= 0 mark the separation between n-capture- normal stars ([Ba/Fe]< 0) and s-/r-process-enhanced stars, as described in Table 1.1. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K).

On the whole, we find that the precision of GALAH abundances for NGC 104 is question- able, with a patchy correspondence to the literature. This is likely driven by the unphys- ical correlation between the derived Teff and [Fe/H], which will need to be addressed by the GALAH team. The results in this section will need to be re-evaluated with a future set of analysis results.1 Chapter 2. Globular clusters in GALAH 69

FIGURE 2.21: The abundance ratio of heavy and light s-process elements versus [Al/Fe] for NGC 104, with a red line at [Al/Fe]= 0.3 to separate the first and second populations. Chapter 2. Globular clusters in GALAH 70

TABLE 2.3: Mean value of the chemical abundances in NGC 5139.

Element Mean σ Element Mean σ [α/Fe] 0.28 0.14 [Co/Fe] 0.47 0.44 A(Li) 0.57 0.36 [Ni/Fe] -0.12 0.14 [C/Fe]* 1.48 - [Cu/Fe] -0.421 0.28 [O/Fe] 0.83 0.33 [Zn/Fe] 0.36 0.31 [Na/Fe] 0.11 0.31 [Rb/Fe] 1.07 0.30 [Mg/Fe] 0.13 0.22 [Sr/Fe] 1.37 0.29 [Al/Fe] 0.63 0.31 [Y/Fe] 0.32 0.56 [Si/Fe] 0.30 0.15 [Zr/Fe] 1.26 0.52 [K/Fe] 0.13 0.20 [Mo/Fe] 1.43 0.60 [Ca/Fe] 0.30 0.18 [Ru/Fe] 1.12 0.30 [Sc/Fe] 0.14 0.12 [Ba/Fe] 0.51 0.54 [Ti/Fe] 0.34 0.28 [La/Fe] 0.69 0.32 [V/Fe] 0.18 0.40 [Ce/Fe] 0.35 0.37 [Cr/Fe] -0.03 0.25 [Nd/Fe] 0.86 0.39 [Mn/Fe] -0.36 0.20 [Sm/Fe] 0.57 0.36 [Eu/Fe] 0.57 0.23 (*) sample of stars < 5

NGC 5139

NGC 5139, or ω Centauri, is known for its scatter in abundance and wide metallicity distribution, as reported in a number of studies including Freeman and Rodgers (1975), Norris and Da Costa (1995), Ikuta and Arimoto (2000), Tsujimoto and Shigeyama (2003), Johnson and Pilachowski (2010), and Simpson, Cottrell, and Worley (2012), and Mészáros et al.(2020). Figure 2.22, 2.23 and 2.24 show the distribution [α/Fe] and 29 elemental abundances versus metallicity for NGC 5139, with the same Teff-based colour coding as in the equivalent figure for NGC 104. As expected, the stars in NGC 5139 cover a wide range in abundance for many elements, in line with previous studies that establish its complex history of star formation and self-enrichment. Table 2.3 lists the mean and stan- dard deviation in abundance for each element, listing A(Li) instead of [Li/Fe], and with- out a standard deviation value for [C/Fe] because we only have an unflagged (reliable) carbon abundance for one star in NGC 5139.

Although the current perception of NGC 5139 is that the cluster probably is the remnant of an ancient disrupted dwarf galaxy (e.g., Hilker and Richtler, 2000; Bekki and Freeman, 2003; Da Costa, 2016), abundance ratios the for light elements such as O, Na, Mg, and Al, and even α elements (e.g., Si, Ca, and Ti) display similar patterns to what is seen in normal (i.e., monometallic) globular clusters. The abundances also show a pattern of en- richment from Type II supernovae and proton-capture elements that shape the chemical enrichment in NGC 5139. Chapter 2. Globular clusters in GALAH 71

FIGURE 2.22: Abundances for α and 8 other individual light elements in NGC 5139. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Li abundance for all globular clusters are displayed in Figure 2.16. Chapter 2. Globular clusters in GALAH 72

FIGURE 2.23: Abundances for 9 individual iron-peak elements in NGC 5139. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Chapter 2. Globular clusters in GALAH 73

FIGURE 2.24: Abundances for 12 individual n-capture elements in NGC 5139. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Chapter 2. Globular clusters in GALAH 74

FIGURE 2.25: [Na/Fe] plotted against [Al/Fe] (left panel) and [O/Fe] (right panel). The expected correlation on the left is fairly clear; uncer- tainties in the GALAH determination of oxygen abundances adds scatter to the plot on the right.

We discussed the Mg-Al abundance plane below, and now we will consider the light ele- ments individually. In Figure 2.22 Al shows a “branching” behaviour starting at [Fe/H]= −1.5, with one branch at a constant value of +1.0 and the other negatively correlated with [Fe/H]. This multi-modal distribution in the Al-Fe plane corresponds to the two density peaks in the distribution mentioned by Mészáros et al.(2020), and a similar pattern is also shown by Johnson and Pilachowski (2010), where an underdensity of stars with [Al/Fe]≈ 0.7 can be seen for stars with [Fe/H]> −1.8.

Marino et al.(2010) and Johnson and Pilachowski (2010) both concluded that NGC 5139 has a different chemical evolutionary path for Na than any other star cluster or Galactic component. In this cluster, [Na/Fe] is increasing along with [Fe/H], but it also exhibits the expected correlation with [Al/Fe] and anticorrelation with [O/Fe] (Carretta et al., 2009a; Gratton et al., 2012; Gratton, 2020). Figure 2.25 focuses in on the Na-Al and Na-O planes, which do not seem to be badly affected by any Teff-based analysis uncertainties.

Abundances of the α elements Si, Ca, and Ti all trace each other quite well, showing their common origin and trending upward with increasing metallicity. Comparison of NGC 5139 in the GALAH survey with 6 red giants of the same cluster from Carretta et al.(2013) shows that our mean abundance for [K/Fe] is distinctly lower than in the literature. [Sc/Fe] shows less scatter in NGC 5139 than it did in NGC 104, with a standard deviation of 0.12 dex. Considering the scatter, this is reasonably consistent with the result Chapter 2. Globular clusters in GALAH 75 of Johnson and Pilachowski (2010), who reported Solar-scaled [Sc/Fe] abundances. The iron peak elements exhibit significant variety in NGC 5139, with V and Co showing large scatter and a downward trend in [Co/Fe] with [Fe/H], Cr and Ni showing smaller scatter and scaled-Solar abundances, and Mn and Cu showing smaller scatter and slightly sub- Solar abundances.

One of the unique features of NGC 5139, explored by Simpson, Cottrell, and Worley (2012) and Marino (2013), is the correlation of [Ba/Fe] and [Y/Fe] with [Fe/H]. This can be seen in Figure 2.24, though the other s-process elements [Zr/Fe] and [La/Fe] do not follow the same pattern in our data.

Figure 2.26 shows that only 7 out of 123 stars can be labelled as neutron-capture-normal stars with [Ba/Fe]< 0.0, while 13.8% of the stars (17/123) exhibit s-process enrichment with [Ba/Fe]> +1.0 and [Ba/Eu]> +0.5, and 10.6% of the stars (13/123) can be con- sidered r/s stars with an abundance ratio 0.0 <[Ba/Eu]< +0.5. The [Ba/Fe] trend in NGC 5139 is expected and consistent with previous abundance studies (e.g., Marino et al., 2012). Only a few stars are neutron-capture-normal, which contrasts with NGC 104.

For [Eu/Fe] we find that the r-process abundances rise with decreasing metallicity, and one star can be labelled as r-II star, with [Eu/Fe]> +1.0. This result is in contrast with Johnson and Pilachowski (2010), who find no variation in [Eu/Fe] in NGC 5139.

There are no stars in our NGC 5139 sample that have both [Ba/Fe] and [Eu/Fe] below 0.50 dex for [Fe/H]> −1.0. This indicates that the s-process is a dominant production mechanism for neutron-capture elements in more metal rich stars in the cluster. This contrasts with the neutron capture abundance behaviour found in other globular clusters, but shows a similarity to the chemical evolution seen in dwarf galaxies (e.g., Venn et al., 2004; Johnson and Pilachowski, 2010), supporting the possibility of NGC 5139 having an ex situ origin.

NGC 5139 has less of a problem with a spurious Teff-[Fe/H] correlation than NGC 104.

However, there are some elements that seem to be more susceptible: stars with Teff < 4500 K generally show low abundances in Li, Co, La, Ce, and Nd compared to their hotter counterparts. In general, we find that the GALAH abundances for NGC 5139 are in line with expectations and previous studies. They replicate well-studied patterns from the literature, including the correlation between [Na/Fe], [Ba/Fe], [Y/Fe], and [Fe/H]. A few of the elements have abundance ratios that deviate from the literature, but a future study with improved abundance precision, especially for metal-poor stars, might overcome this problem. Chapter 2. Globular clusters in GALAH 76

FIGURE 2.26: Neutron-capture elements Ba (upper panel) and Eu (lower panel) versus the [Ba/Eu] ratio, showing the variation of the s- and r- process enrichment in NGC 5139. Chapter 2. Globular clusters in GALAH 77

TABLE 2.4: Mean value of the chemical abundances in NGC 6397.

Element Mean σ Element Mean σ [α/Fe] 0.29 0.06 [Co/Fe]* 1.28 0.15 A(Li) 0.90 0.3 [Ni/Fe] 0.00 0.23 [C/Fe]* - - [Cu/Fe] -0.31 0.32 [O/Fe]* 0.84 0.17 [Zn/Fe] 0.18 0.13 [Na/Fe] 0.22 0.17 [Rb/Fe]* - - [Mg/Fe] 0.14 0.09 [Sr/Fe]* - - [Al/Fe] 0.85 0.09 [Y/Fe] -0.22 0.13 [Si/Fe] 0.33 0.08 [Zr/Fe]* 1.65 0.01 [K/Fe] 0.22 0.10 [Mo/Fe]* - - [Ca/Fe] 0.31 0.07 [Ru/Fe]* - - [Sc/Fe] 0.07 0.06 [Ba/Fe] -0.11 0.17 [Ti/Fe] 0.42 0.28 [La/Fe] 0.58 0.37 [V/Fe] 0.25 0.56 [Ce/Fe] 0.21 0.32 [Cr/Fe] -0.09 0.16 [Nd/Fe] 0.64 0.44 [Mn/Fe] -0.38 0.10 [Sm/Fe] 0.42 0.32 [Eu/Fe]* 0.88 0.34 (*) sample of stars < 5

NGC 6397

NGC 6397 is a well-studied metal-poor globular cluster. As for the previous two clusters, Figure 2.27, 2.28, and 2.29 shows the distribution of α and 29 individual elemental abun- dances versus metallicity in NGC 6397. The same colour coding applies as in previous figures like this (e.g., Figure 2.17 and Figure 2.22), and NGC 6397 does not appear to have a significant Teff-[Fe/H] correlation affecting its abundances. Table 2.4 lists the mean and standard deviation in abundance for each element, listing A(Li) instead of [Li/Fe]. Ele- ments for which we have fewer than 5 unflagged abundance measurements are marked with an asterisk, and standard deviations are only given when there are at least three stars with a measured abundance value.

As discussed above, in NGC 6397 we only have unflagged (reliable) abundances for second-population stars, and they have a mean Al abundance of [Al/Fe]= 0.85. Be- cause all of the stars in this data set surpass the 0.30 dex mark for membership in the second population, we do not see the bimodal [Al/Fe] distribution that was reported in Mészáros et al.(2020). Further confirming this, the distribution of [Mg/Fe] is slightly super-Solar and is consistent with the lower end of the abundance distribution in Car- retta et al.(2009b).The abundances of [Na/Fe] are moderately enhanced, and lie within the range of what is reported in Carretta et al., 2009a; Carretta et al., 2009b. However, because of scarcity of the sample—being less than 5 stars with [O/Fe] abundances—we cannot comment on whether we observe the O-Na anticorrelation to be consistent with Chapter 2. Globular clusters in GALAH 78

FIGURE 2.27: Abundances for α and 8 other individual light elements in NGC 6397. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Li abundance for all globular clusters are displayed in Figure 2.16. Chapter 2. Globular clusters in GALAH 79

FIGURE 2.28: Abundances for 9 individual iron-peak elements in NGC 6397. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Chapter 2. Globular clusters in GALAH 80

FIGURE 2.29: Abundances for 12 individual n-capture elements in NGC 6397. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Chapter 2. Globular clusters in GALAH 81 the literature.

There is very little scatter in [α/Fe] or the α element abundances [Si/Fe] and [Ca/Fe], and they all sit near the expected level of +0.3. The light odd-Z element Sc also shows little scatter and no obvious trend with metallicity with a standard deviation of 0.06 dex. The iron-peak elements follow a similar pattern to what we saw in NGC 5139, though with a smaller sample: V and Co enhanced, Cr and Ni roughly Solar, and Mn and Cu mildly depleted.

Evaluating the neutron-capture elements is more difficult with this smaller sample. In NGC 6397 we have only have 4 stars with measured [Eu/Fe] abundance ratios. Two of them are classified as r-I stars, with abundances in the range 0.3 <[Eu/Fe]< 1.0, and the other two are r-II stars with [Eu/Fe]> 1.0. As for s-process elements, both [Y/Fe] and [Ba/Fe] are sub-Solar but the other s-process elements La, Ce and Nd have enhanced abundances with a larger scatter. In Figure 2.15 the positive values of the [Ba/Y] ratio emphasize that the light s-process elements are dominant over their heavy counterparts in at least 5 stars, and in Figure 2.30 it can be seen that r-process enrichment plays a stronger role than s-process enrichment, at least for the four stars with measured abun- dances.

There are important differences between our data for NGC 104 and NGC 5139, and our data for NGC 6397 and NGC 7099. We have collected distinctly larger samples in the for- mer two, which are closer and larger on the sky. Also, the lower metallicity of the latter two makes it more difficult to determine abundances for some of the elements. Never- theless, when we compare against the literature, the GALAH abundances for NGC 6397 complement previous studies including Carretta et al.(2009a) and Carretta et al.(2009b) and confirm the presence of second-population stars in the cluster. However, expanding the data set, particularly by adding abundance results for first-population stars, will al- low us to examine the abundance patterns more robustly and draw stronger conclusions. A new globular cluster observing program has been suggested to the GALAH team by D. Nataf (priv. comm.); however, there are not currently any solid plans to carry out this program in the near future. Chapter 2. Globular clusters in GALAH 82

FIGURE 2.30: Neutron-capture elements Ba and Eu versus the [Ba/Eu] ratio in NGC 6397. The four cluster members with measured abundances show a range in s- and r-process enrichment, and the low [Ba/Eu] ratio indicates the importance of r-process enrichment in the cluster. Chapter 2. Globular clusters in GALAH 83

TABLE 2.5: Mean value of the chemical abundances in NGC 7099.

Element Mean σ Element Mean σ [α/Fe] 0.36 0.21 [Co/Fe]* 1.09 0.11 A(Li) 0.82 0.12 [Ni/Fe] 0.12 0.30 [C/Fe]* - - [Cu/Fe] 0.10 0.36 [O/Fe] 1.32 0.32 [Zn/Fe] 0.28 0.16 [Na/Fe] 0.34 0.23 [Rb/Fe]* - - [Mg/Fe] 0.10 0.14 [Sr/Fe]* - - [Al/Fe] 1.09 0.05 [Y/Fe] 0.03 0.29 [Si/Fe] 0.46 0.23 [Zr/Fe] 1.29 0.25 [K/Fe] 0.17 0.09 [Mo/Fe]* - - [Ca/Fe] 0.32 0.10 [Ru/Fe]* - - [Sc/Fe] 0.17 0.17 [Ba/Fe] -0.03 0.14 [Ti/Fe] 0.56 0.32 [La/Fe] 0.96 0.29 [V/Fe] 0.47 0.55 [Ce/Fe] 0.62 0.36 [Cr/Fe] 0.12 0.23 [Nd/Fe] 0.98 0.59 [Mn/Fe] -0.46 0.20 [Sm/Fe] 0.69 0.38 [Eu/Fe]* 1.06 0.01 (*) sample of stars < 5

NGC 7099

NGC 7099, or M30, has the lowest metallicity of the globular clusters observed by GALAH, which is –2.27 dex, but that does not prevent us from analysing its abundance patterns. Figure 2.31, 2.32, and 2.33 show the distribution of [α/Fe] and 29 elemental abundances versus metallicity for NGC 7099. There is at least some Teff-[Fe/H] correlation at work in these results, since the two coolest stars are assigned the lowest metallicity. Table 2.5 lists the mean and standard deviation in abundance for each element, with A(Li) given instead of [Li/Fe]. Elements for which we have fewer than 5 unflagged abundance mea- surements are marked with an asterisk, and standard deviations are only given when there are at least three stars with a measured abundance value.

We find a large star-to-star scatter for most elements while K, Ca, Ba, Mn show uniformity across metallicity range. In additional, C, Rb, Sr, Mo, Ru are not detected in any of the stars whilst Co and Eu are only measured in 3 and 2 stars, respectively.

As mentioned previously, the [Mg/Fe] and [Al/Fe] abundances indicate that all of the stars we have abundances for in NGC 7099 belong to the second population. The mean abundance ratio of +0.10 in [Mg/Fe] sits at the low end of the distributions from Carretta et al.(2009b) and Cohen, Huang, and Kirby (2011), which both find a mean of about +0.51. At the same time, the mean abundance of [Al/Fe] is considerably high at +1.09, which supports the idea that only second-population stars with high [Al/Fe] and low [Mg/Fe] Chapter 2. Globular clusters in GALAH 84

FIGURE 2.31: Abundances for α and 8 other individual light elements in NGC 7099. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Li abundance for all globular clusters are displayed in Figure 2.16. Chapter 2. Globular clusters in GALAH 85

FIGURE 2.32: Abundances for 9 individual iron-peak elements in NGC 7099. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Chapter 2. Globular clusters in GALAH 86

FIGURE 2.33: Abundances for 12 individual n-capture elements in NGC 7099. The stars are divided into effective temperature bins and colour coded in blue (Teff ≥ 5500 K), black (5500 K > Teff > 4500 K), and red (Teff ≤ 4500 K). Chapter 2. Globular clusters in GALAH 87 are observed in the sample.

Cohen, Huang, and Kirby (2011) present the mean abundance of 28 elements for five red giant members of NGC 7099, which are useful for comparison with the GALAH sample. Of the elements reported in both data sets, the abundances of Si, Ca, Mn, and Ni are consistent with each other within the range of uncertainties. Meanwhile, [Cu/Fe] and [V/Fe] have the most significant difference between the two samples. The GALAH mean abundances are respectively 0.10 ± 0.36 and 0.47 ± 0.55, sitting ≈ 0.8 and 0.7 dex higher than the Cohen, Huang, and Kirby (2011) abundances which are at −0.76± 0.17 and −0.23± 0.09, respectively.

There are only two stars shown in Figure 2.34, as those M30 cluster members for which we have a reliable [Eu/Fe] value. Interestingly, both stars are classified as r-II stars be- cause they are extremely enhanced in the r-process, with [Eu/Fe]> 1.0. [Zr/Fe] and [La/Fe], belonging to the light and heavy s-process peaks respectively, exhibit high abun- dances, with mean value of 1.29 dex and 0.96 dex, respectively. However, other s-process elements such as Y and Ba are not enhanced, and have mean abundances around 0.00 dex. This is the same pattern we saw in the s-process abundances in NGC 6397.

Overall, our results for NGC 7099 are similar to NGC 6397: we cannot measure abun- dances for a number of elements, and it is difficult to assess the reliability of the GALAH abundances for these clusters because of the small sample size. Chapter 2. Globular clusters in GALAH 88

FIGURE 2.34: Neutron-capture elements Ba and Eu versus the [Ba/Eu] ra- tio in NGC 7099. The two stars we have these abundances for are highly Eu-enriched but not particularly Ba-enriched, indicating that r-process en- richment is relatively more important than s-process enrichment in this cluster. Chapter 2. Globular clusters in GALAH 89

2.5 Summary and conclusion

Globular clusters are an important component of the Galactic halo, and they are useful in a number of ways to our understanding of the early history of the Milky Way. Some of the globular clusters discussed in this Chapter were specially targeted early in the GALAH survey, and some were observed serendipitously as part of regular survey operations. In this first exploration of the globular cluster abundance data from GALAH, we have found that they are, for the most part, unexpected and inconsistent. We can see patterns we expect to find in the data, like the light element abundance anticorrelations, and we can reproduce other known abundance behaviour from the literature. In general, the mean [Fe/H] values for the four globular clusters discussed in this Chapter (NGC 104, NGC 5139, NGC 6397, and NGC 7099) agree with literature. A few elements such as Al, Mg, Si, Ba and Eu are very promising. In NGC 5139, we are able to reproduce the unique and well-known correlations between [Na/Fe], [Ba/Fe], and [Fe/H].

However, there are a few limitations we find for GALAH abundances for globular cluster giants, and these do constrain what we can do with the data. The lack of reliable C and O abundances, and the lack of N abundances at all, means that we cannot rely on the more common C-N and O-Na anticorrelations as markers of the first and second populations. We also find that, while some of our elemental abundances match well to the literature and to our expectations, some are more inconsistent, including the s-process elements Zr and Ce, which do not always match well with their more commonly reported counterparts Y and Ba.

We find that placing the dividing line between first- and second-population stars in glob- ular clusters at [Al/Fe]= +0.3, as suggested by Mészáros et al.(2020), works well for the GALAH data set. In NGC 104 and NGC 5139, where we have larger data sets, we find that the s-process abundance and the light- to heavy-s-process ratio both seem to change near that same dividing line. Interestingly, we also find an anticorrelation in the [Ba/Fe] versus [Al/Fe] abundance plane, which contradicts the result of D’Orazi et al.(2010a) regarding the variation of [Ba/Fe] ratios in multiple populations.

We could strengthen our ability to do new globular cluster-based science with GALAH survey data in a number of ways. Increasing the size of the data set would make it pos- sible to more thoroughly sample the complex abundance distributions in each cluster. Increasing the signal to noise ratio in the spectra would allow more confident abundance estimations for weak lines, allowing the determination of a larger number of abundances especially at low metallicity and low surface gravity. The origin of the spurious Teff- [Fe/H] correlation must be discovered and fixed in the spectroscopic analysis process. These improvements would expand the science capabilities of GALAH for all investiga- tions involving distant luminous giants or metal-poor stars, and not just globular cluster Chapter 2. Globular clusters in GALAH 90 abundance studies. Fortunately, this issue has not been seen in the field stars or any other GALAH work as it is an artifact of imprecise distance estimation for distant stars. The field stars in GALAH are all much closer to the Sun, so, they do not suffer from this problem.

Now that we know that GALAH survey data can usefully identify the stellar populations in globular clusters using their abundances, Chapter3 will probe further into the Galactic halo region to identify halo stars with abundance patterns that resemble those of the second population in globular clusters. 91

Chapter 3

Milky Way halo stars with globular cluster-like abundance patterns

3.1 The role of halo stars in Galactic archaeology

The stellar halo of our Milky Way is a large, low-density, spherical distribution of stars.

The density ratio between the halo and the disk ρhalo/ρdisk is about 0.001, and the height of the halo is about 1600 - 2000 pc while the height of the thick disk is only about 300 pc (Norris, 1996). The halo contains the most metal-deficient stars and some of the oldest stars in the Galaxy, and the more information we can gather about the whole ensemble of halo stars, the more carefully we can investigate the earliest stages in Galactic formation and evolution. Hence, the study of the halo enables an archaeological study of the history of the Galaxy, with each substructure serving as a fossil from the Galactic past (Eggen, Lynden-Bell, and Sandage, 1962).

The cosmological framework for Galactic archaeology is the Λ cold dark matter (ΛCDM) paradigm. This grand scheme dictates the formation and evolution processes at work in all galaxies, including the Milky Way (Agertz, Teyssier, and Moore, 2011; Doménech- Moral et al., 2012). Larger galaxies like the Milky Way are theorized to have been built up progressively from the gravitational coalescence of smaller “building blocks” that add up to form larger and more massive galaxies (Springel, Frenk, and White, 2006; Helmi, 2008). However, that orderly and large-scale picture of Galactic formation still holds many unanswered questions at the detailed level, especially in the history of the stellar halo.

The overall structure of the stellar halo is a melting pot of stars with various origins, some heated from orbits in the disk or the bulge (Frankel et al., 2019; Barbuy, Chiappini, and Gerhard, 2018), or escaped from globular clusters (Martell et al., 2011; Lind et al., 2015; Koch, Grebel, and Martell, 2019), or captured from neighbouring dwarf galaxies Chapter 3. Globular cluster stars in the halo 92

(Maxwell et al., 2012; Helmi, 2020). This makes the halo a complex mix of stars with dif- ferent nucleosynthetic influences that currently occupy different families of orbits. This special component of the Galaxy has become a hot topic for researchers in the era of large surveys, and our perspective regarding the halo is evolving quickly as we continue to explore it.

Even though this is an important and active field of study, our understanding of the stellar halo is still incomplete, in part because of the low density of halo stars. Even when working with large survey data sets we must make careful selections in abundance and kinematics to focus our investigations on these fossils, and to build an understanding of the chemodynamics of the stellar halo. This Chapter particularly focuses on the stellar halo in order to identify stars that have abundance patterns that resemble those of stars in globular clusters.

3.1.1 Kinematics of halo stars

As discussed earlier in this chapter, in a ΛCDM universe, the stellar halos of galaxies are complex environments built through a long series of events, including minor mergers, gas accretion and subsequent star formation, and migration of stars from other Galactic components into the halo. Some authors divide the halo into “inner” and “outer” com- ponents with differing kinematics and metallicity (e.g., Ishigaki, Chiba, and Aoki, 2010; An et al., 2015; Belokurov et al., 2018), and numerical simulations of galaxy formation (e.g., Tissera et al., 2014; Somerville and Davé, 2015) can reproduce this dual structure, with distinct halo subpopulations due to early accretion and gas capture. Subsequent in situ star formation tends to produce a halo population that is more gravitationally bound, with lower orbital eccentricity and a smaller median apocentre, while later accre- tion events produce a more spatially extended population with larger dispersion in each velocity component and little net rotation.

Observational efforts to clarify the present-day structure and evolutionary history of the stellar halo in the Milky Way tend to focus on distinctive kinematic tracers, including hor- izontal branch stars or RR Lyrae variables that are reasonably accurate standard candles (Lee, 1992; Bhardwaj, 2018), and on stellar streams. The long dynamical time that gov- erns the remote outer regions of a galaxy allows stars from accreted dwarf galaxies and tidally disrupted star clusters to maintain spatial and kinematic coherence over many or- bits. This makes it possible to investigate the assembly history of a galaxy using stellar streams, as in Fardal et al.(2014), Barba et al.(2019), and Helmi (2020).

Large-scale photometric surveys like the Sloan Digital Sky Survey (e.g., Belokurov et al., 2006), Pan-STARRS (e.g., Bernard et al., 2016; Banik et al., 2019), and the Dark Energy Survey (e.g., Shipp et al., 2018) have been a crucial resource, dramatically expanding the Chapter 3. Globular cluster stars in the halo 93 number of known spatially coherent streams. Kinematic followup is then necessary to separate true stream members from line of sight contamination, and the 5-d astromet- ric solution from Gaia DR2 can be combined with radial velocities from spectroscopy to confirm membership with high accuracy, as in Hasselquist et al.(2019) and Li et al.(2019).

Stellar streams and tidal tails lose their spatial connection over time but, through phase mixing, retain their kinematic coherence. This makes it possible to search for evidence of important accretion or tidal disruption events in the stellar halo by searching the dynam- ical space for overdensities that stand out against the smoothly varying background of the halo field (Helmi and White, 1999; Koppelman et al., 2019; Myeong et al., 2019; Grand et al., 2020; Naidu et al., 2020).

While many of the studies of the Galactic halo’s formation history focus on dwarf galax- ies, there are also some that focus on the contributions of globular clusters to the halo field. There are a number of methods used in this work that combine information on spatial density, photometry, kinematics, and elemental abundances (e.g., Odenkirchen et al., 2001; Martell and Grebel, 2010; Marino et al., 2014; Kuzma et al., 2015; Grillmair, 2017; Malhan, Ibata, and Martin, 2018; Simpson et al., 2020; Wan et al., 2020). In this chapter we expand on previous work that uses chemical tagging to identify halo field stars that have globular cluster-like abundances but are not members of stellar streams or coherent kinematic structures.

3.1.2 The chemical compositions of halo stars

The chemical elements that we observe and know today are the result of nucleosynthesis in complex chains of nuclear reactions inside stars. Every element can play an important role in galactic chemical evolution, as each of them contains clues about the site of its production. As mentioned in Chapter1, the chemical abundances in stars are the result of nucleosynthesis that occurred during the earlier generations of stars. The halo is the Galactic component with the lowest metallicity, with a distribution that peaks around [Fe/H]= −1.5 and has an extended tail to the most metal-poor stars. This means that its chemical enrichment history is less complex, and includes fewer generations of stars, than the disk or the bulge, making the abundances of halo stars distinct (Frebel and Nor- ris, 2013).

Early in the process of star formation in a galaxy, stars with high mass (usually > 8M ) and short lifetimes can greatly influence the chemical enrichment (Tinsley, 1979). Type II supernovae from these massive stars yield enrichment in the α-elements (e.g. Mg, Si, Ca, and Ti), resulting in the enhanced α-element abundances ([Mg/Fe]≈ +0.4) that are typical for metal-poor stars in the Galaxy (McWilliam, 1997). There is some evidence of Chapter 3. Globular cluster stars in the halo 94 scatter in [α/Fe] in the Galactic halo, and a small number of metal-poor stars show partic- ularly low [Mg/Fe] abundances. Detailed studies of abundances in halo stars, including Ivans et al.(2003), Nissen and Schuster (2010), and Helmi et al.(2018), have all found va- riety beyond the typical population. A low-alpha abundance pattern implies that these outliers have a different chemical history from “normal” halo stars, thus suggesting that they may have originated in dwarf galaxies, which harbour many metal-poor stars with low [Mg/Fe] (e.g., Tolstoy, Hill, and Tosi, 2009). The [α/Fe] versus [Fe/H] pattern can be used as a chemical tag for halo stars and link them to possible external progenitors. Chapter4 will go in depth on the chemical evolution for halo stars.

As mentioned in Chapter1, chemical tagging is an essential tool for identifying stars that formed at the same place and time (Freeman and Bland-Hawthorn, 2002). In this Chapter, we will describe the search for stars with globular cluster-like abundance patterns in the halo. This type of chemical tagging is feasible because globular cluster stars have their own characteristic abundance patterns involving the light elements C, N, O, Na, Mg, and Al (e.g., Kraft, Trefzger, and Suntzeff, 1979; Carretta et al., 2009a) that we believe to be uniquely tied to that formation environment, as described in Chapter2. These patterns can be used to identify individual halo field stars likely to have originated in globular clusters, similar to how specific abundance ratios can be used to identify stars likely to have been accreted from dwarf galaxies.

In 2020, the GALAH team will release their third public catalog (GALAH DR3), which includes more than half a million stellar spectra and up to 30 elemental abundances per star. In this Chapter, we use GALAH iDR3, the internal team version of the DR3 catalog, as our data source. We perform a kinematic selection of halo stars using Gaia DR2 proper motions, GALAH radial velocities, and distances from Bailer-Jones et al.(2018), and in- vestigate whether these halo stars display globular cluster-like abundance patterns. We conclude by comparing our results with previous studies. Chapter 3. Globular cluster stars in the halo 95

3.2 Selecting halo stars in the GALAH data set

The majority of stars observed in the GALAH survey belong to the disk, which is a con- sequence of the relatively bright apparent magnitude limit (12 ≤ V ≤ 14) and a lack of selection based on colour or parallax. These choices create a simple selection function that is useful for studying the underlying stellar populations in the Galaxy, but they also create a data set that is weighted toward nearby main-sequence stars. The largest set of halo stars in GALAH will therefore be those in the Solar neighbourhood, rather than the distant halo, and so they must be identified kinematically rather than spatially. With the availability of full space velocities from combining Gaia parallaxes and proper motions with GALAH radial velocities, we can make a more certain identification of halo stars than with proper motion alone.

We begin with the full iDR3 catalog, and select only the stars with appropriate quality flags: flag_sp= 0, which indicates accurate stellar parameters, flag_fe_h= 0 for reliable metallicity, and also flag_o_fe, flag_na_fe, flag_mg_fe, flag_al_fe, since these are the main elements we use in the search for globular cluster abundances in halo stars. The first step in identifying halo field stars with globular cluster-like abundances is to remove potential contamination from globular cluster members in our data set. In Chapter2 we identified likely members of 12 Galactic globular clusters in the iDR3 catalog, and here we take the conservative step of removing all stars within the tidal radius of all Galactic globular clusters from our data set. We use cluster center positions and tidal radii from Vasiliev (2019), who derives them based on data from Gaia DR2. These steps remove 369,764 stars and leaves 283,035 stars.

We calculate stellar kinematic information ourselves using the gala library (Price-Whelan, 2017). We transform the heliocentric phase-space coordinates of right ascension, declina- tion, proper motions, distance and radial velocity to the Local Standard of Rest and then to Galactocentric Cartesian coordinates with astropy (Astropy Collaboration et al., 2018). That coordinate system has orthogonal axes oriented from the Galactic centre to the Sun (X), along the direction of Galactic rotation (Y), and perpendicular to the disk (Z). To make that transformation we use the astropy default values for the Galactocentric dis- tance of the Sun (8.122 kpc), the height of the Sun above the Galactic plane (20.8 pc), and the motion of the Sun relative to the Galactic centre (12.9, 245.6, and 7.78 km s−1 in the U, V, and W directions, respectively). We then integrate the orbits forward in time for 1 Gyr in gala to calculate the total orbital energy Etot, the angular momentum in the vertical direction LZ, and the orbital eccentricity for each star. We used the default “MilkyWay- Potential” as the Galactic gravitational potential, which includes a Hernquist bulge, a Miyamoto-Nagai disk from Bovy (2015), and an NFW halo.

The kinematic selection we apply to identify halo stars in our GALAH data set is done Chapter 3. Globular cluster stars in the halo 96 using a Toomre diagram, shown in Figure 3.1. This diagram compares a star’s velocity in the direction of Galactic rotation (V) against its velocity in the perpendicular direc- √ tion ( U2 + W2). In this plane, stars in the thin disk cluster near V = 245.6 km s−1 and √ U2 + W2 = 0, since they tend to have low orbital eccentricity and small vertical veloc- ity. Stars in the thick disk tend to lag the thin disk in rotational velocity and have larger perpendicular velocity, both because they have higher orbital eccentricity and because of the larger scale height of the thick disk. Halo stars are further from the thin disk in the Toomre diagram than the thick disk is, and some are on retrograde orbits (V < 0 km s−1). An approximate division of the three components can be made (e.g., Nissen, 2004a) in the √ 2 2 2 −1 total velocity vtot = U + V + W , such that stars with vtot less than 50 km s away from the Local Standard of Rest mainly belong to the thin disk, stars in the range 70 km −1 −1 s ≤ vtot ≤ 200 km s are part of the thick disk, and stars with vtot more than 200 km s−1 are almost certainly halo stars. In Figure 3.1 the 200 km s−1 boundary is shown as a solid black line, thin disk and thick disk stars are shown with small blue points, and halo stars are plotted as red circles. While GALAH is mainly a survey of the Galactic disk, it does capture a useful population of halo stars in the Solar neighbourhood. Selecting halo stars by this criterion, we keep 18,914 in our sample and reject 264,121 stars. As noted in Chapter2, the systematic problem in the stellar parameters for globular cluster stars is an artifact of inaccurate distance estimates. The distances are more accurate for these more nearby stars, and there is not a systematic problem in stellar parameters for these field stars. Figure 3.2 shows how the selected metal-poor halo giants (black points) behave dynamically in the orbital energy and vertical angular momentum plane as compared to globular clusters from Gaia DR2 (Gaia Collaboration et al., 2018a) (blue crosses), and all other kinematically selected halo stars (red points).

Because selections in kinematic space are imperfect, we also apply a metallicity filter to our data set to remove interlopers from the disk, only accepting stars with [Fe/H]≤ −1.0. We also require stars to be on the red giant branch, since the globular cluster data set we are comparing against is composed entirely of evolved stars. This selection is Teff ≤ 5500 K and log g ≤ 4.0 cm/s2. Finally, we apply a basic data quality selection, requiring the signal to noise ratio to be at least 20 per pixel in the HERMES green camera. All together, these selections reject 18,469 and keep 445 stars in our final data set.

There are two potential ways to expand the data set for this study. It is possible to ex- pand the selection to include main sequence stars, since they exhibit the same abundance anomalies in globular clusters as red giant branch stars do. We would probably find a smaller number of them than giants, since they would need to be located closer to the Sun to be observed as part of GALAH. However, extending to higher metallicity with this data set would likely not be effective. There are relatively few halo stars with metallicities that high, and (as noted in section 6.4 of Buder et al., 2020) the GALAH abundances for Chapter 3. Globular cluster stars in the halo 97

FIGURE 3.1: Toomre diagram for our data set, with thin and thick disk stars shown as blue points and halo stars plotted in red. The black line marks the −1 locus where vtot is more than 200 km s different from the Local Standard of Rest and the contour lines show the density of points in the plot. red giants at high metallicity are less accurate. Chapter 3. Globular cluster stars in the halo 98

FIGURE 3.2: A Lindblad diagram, showing orbital energy and vertical an- gular momentum for our final data set of metal-poor halo giants (black points), globular clusters from Gaia DR2 (Gaia Collaboration et al., 2018a) (blue crosses), and all other kinematically selected halo stars (red points). In the coordinate system we use, prograde orbits are to the right of the diagram and retrograde orbits to the left. Chapter 3. Globular cluster stars in the halo 99

3.3 Abundance patterns of the GALAH halo giant stars

The halo is the most metal-poor component of the Galaxy, with metallicity typically be- low [Fe/H]= -1.0. This contrasts with the thick disk, where stars are more metal rich (−0.16 >[Fe/H]> −1.25, Sharma et al., 2019), and the thin disk and bulge, where metal- licities can even be super-Solar. While the three components can overlap in metallicity, they behave differently in the [α/Fe] versus [Fe/H] space, which reflects the different conditions of their star formation. Metal-poor halo stars tend to have enhanced abun- dances of α elements (i.e. Mg, Si, Ca, Ti) but at a metallicity of [Fe/H] ≈ −1.0 dex, the [α/Fe] begins to drop from ∼ +0.4 dex to Solar values at [Fe/H] = 0 (Wheeler, Sneden, and Truran, 1989; McWilliam, 1997), since the abundances of stars in the thick disk and the thin disk are more influenced by Type Ia supernovae because they formed at later times (e.g., Sahijpal, 2013).

While none of the stars in our GALAH halo sample is very metal poor (the lowest [Fe/H] in the sample is −1.71 ± 0.10), they are still generally within the range of metallicity for globular clusters, which range from [Fe/H] = 0.00 dex for BH 176 to [Fe/H] = −2.37 for NGC 7078 (Harris, 1996). This is important because we are searching for halo stars that resemble the globular cluster chemical abundance pattern.

Figure 3.3 shows the distribution of our 445 halo giants with chemical abundances from GALAH for [α/Fe] and 30 elements. All of the elements evidently show a continuity to higher metallicity ([Fe/H]> −1) which has been cut off as part of our criteria in selecting metal-poor halo stars. The sample shows the largest spread in [Y/Fe] and [Mo/Fe] with standard deviations of 0.49 dex and 0.48 dex, respectively. A large scatter is also apparent for O, V, Co, Rb, La, Ce, Sm, and Nd, which all have standard deviations larger than 0.25 dex.

There were very few reliable measurements of C abundance in the globular cluster stars discussed in Chapter2. Since our halo giant sample occupies a similar range in metal- licity and evolutionary phase, it is not surprising that only 6 of our 445 halo stars have unflagged C abundances. In contrast, O, Na, Mg, and Sc are detected in all 445 halo stars. However, as mentioned in Chapter2, the O abundance tends to be higher than we would expect. As a result, we could not study the Na-O anticorrelation in globular clusters, and in this Chapter we will continue to use Mg and Al as the indicators of the second-population abundance pattern.

There is one star in our sample with unusual abundances: it has the 2MASS ID J19183781- 1011360, it has a metallicity of [Fe/H]= −1.60, but it has very high abundances in O, Na, Al, Sc, Ti, Cr, Co, Ni, Cu, Sr, Y, Mo and Nd compared to the rest of the halo sample. We are not focusing on this particular star in this Chapter as it does not show any chemical Chapter 3. Globular cluster stars in the halo 100

FIGURE 3.3: The abundance ratio of all available elements in the GALAH survey for the selected stars as a function of [Fe/H]. Star J19183781- 1011360 is emphasized with a red star. Chapter 3. Globular cluster stars in the halo 101

TABLE 3.1: For each element in the GALAH iDR3 catalog, the number of stars in our halo giant sample with a non-flagged abundance, the mean, standard deviation in that abundance and the respective typical error.

Element # stars Mean σ Typical error [α/Fe] 445 0.24 0.04 0.17 A(Li) 43 0.98 0.56 0.40 [C/Fe] 2 1.17 0.26 0.18 [O/Fe] 445 0.64 0.26 0.18 [Na/Fe] 445 -0.09 0.16 0.11 [Mg/Fe] 445 0.24 0.12 0.08 [Al/Fe] 445 0.14 0.23 0.16 [Si/Fe] 443 0.45 0.14 0.10 [K/Fe] 437 0.15 0.13 0.09 [Ca/Fe] 435 0.26 0.12 0.08 [Sc/Fe] 445 0.09 0.11 0.08 [Ti/Fe] 308 0.21 0.14 0.10 [V/Fe] 61 0.01 0.22 0.16 [Cr/Fe] 436 -0.13 0.14 0.10 [Mn/Fe] 440 -0.35 0.11 0.08 [Co/Fe] 191 0.01 0.26 0.18 [Ni/Fe] 383 -0.10 0.12 0.08 [Cu/Fe] 407 -0.40 0.16 0.11 [Zn/Fe] 92 0.18 0.18 0.13 [Rb/Fe] 6 0.57 0.39 0.28 [Sr/Fe] 7 1.50 0.38 0.27 [Y/Fe] 31 0.45 0.49 0.35 [Zr/Fe] 40 0.43 0.22 0.16 [Mo/Fe] 11 0.77 0.54 0.38 [Ru/Fe] 2 0.64 0.27 0.19 [Ba/Fe] 443 0.43 0.35 0.25 [La/Fe] 115 0.47 0.37 0.26 [Ce/Fe] 33 0.09 0.40 0.28 [Nd/Fe] 270 0.53 0.24 0.17 [Sm/Fe] 51 0.20 0.29 0.21 [Eu/Fe] 152 0.46 0.19 0.13 Chapter 3. Globular cluster stars in the halo 102 similarity to globular cluster stars, but it would be interesting in the future to do some further investigation of this metal-poor star with a very odd abundance pattern.

The other GALAH abundances of the halo giants will be discussed further in the next Chapter, as this Chapter only focuses on stars with globular cluster-like abundances.

3.3.1 Al-Mg-Si variation

As mentioned in Chapter2, the Al-Mg anticorrelation is part of the characteristic glob- ular cluster abundance pattern, with the second population of stars enhanced in Al and depleted in Mg. This pattern is not normally seen in halo field stars. Figure 3.4 shows [Al/Fe] versus [Mg/Fe] for our halo giant sample, colour coded by [Si/Fe]. The struc- ture in this figure is quite different to Figure 2.9, which shows globular cluster stars in the same abundance plane. Where NGC 104 and NGC 5139 have wide ranges in Al and very little range in Mg, in the halo field we see a positive correlation between the two elements and a range of at least 0.5 dex in each. The colour coding also shows a clear correlation between both elements and Si. This difference between that chemical evolution in the halo and in globular clusters is exactly what we intend to use for chemical tagging. In Chapter2 we found that the two populations in globular clusters could be well separated at [Al/Fe] = +0.3, and the red line in the figure shows that division. The majority of our halo field stars fall below the line, consistent with the first-population stars in globular clusters. This is also in line with our expectations from the literature. Martell and Grebel (2010) and all subsequent searches for globular cluster escapees in the field have found that the vast majority of halo field stars have normal halo abundance patterns. The stud- ies of Roederer et al.(2014) and Yong et al.(2013) find that it is typical for metal-poor halo stars to have slightly sub-Solar Al abundance and slightly super-Solar Mg abundance.

We saw an Al-Si correlation in NGC 104 and NGC 5139 in Chapter2, and Mg-Si anticor- relations have been found in a small number of massive and metal-poor globular clus- ters before: NGC 6752 (Yong et al., 2005), NGC 2808, NGC 7078 (Carretta et al., 2009a), NGC 6341, and NGC 5139 (Mészáros et al., 2020). This happens when Si is produced by “leakage” from the Mg-Al cycle through the 26Al(p, γ) 27Si(e-,ν) 27Al(p,γ)28Si reactions at high temperature (Karakas and Lattanzio, 2003). Without this leakage, we would expect a simple correlation between Mg and Si since they are both α elements. Figure 3.5 shows the Mg-Si abundance plane, colour-coded by [Fe/H], and there are some key differences between the halo stars and the globular cluster stars. First, there is clearly a Mg-Si corre- lation in the halo stars, and most of the globular cluster stars are not enhanced in Si. There is also a difference in the typical Mg abundance of the first-population globular cluster stars and the halo stars, and this suggests the possibility of doing chemical tagging to recognise first-population globular cluster stars as well as second-population stars, but Chapter 3. Globular cluster stars in the halo 103

FIGURE 3.4: Abundance of Al vs Mg for our halo giants, colour coded by Si abundance. The red line at [Al/Fe] = +0.3 provides a simple separation between the first and second population in globular clusters, and there is clearly an Al-Si correlation. The four stars circled in red are candidate globular cluster escapees, which we discuss in Section 3.4. Chapter 3. Globular cluster stars in the halo 104

FIGURE 3.5: Comparison of halo stars in the [Si/Fe] versus [Mg/Fe] plane against globular cluster members from Carretta et al.(2009b) and Masseron et al.(2019). possible issues with abundance zeropoints between the different data sets would first need to be resolved. Chapter 3. Globular cluster stars in the halo 105

3.4 Globular cluster escapees in the halo from GALAH

Since [Mg/Fe] and [Al/Fe] were the most useful of the light elements for identifying the multiple populations in globular clusters, we will also use them to search for globular cluster escapees in the halo field by finding stars with high Al and low Mg. In Figure 3.4 we have circled the four stars that are positioned the furthest away from the main group of halo stars in the Al-enhanced, Mg-depleted direction. Our selection of these four stars as candidate escapees is similar to the selection of candidate escapees in Martell et al. (2016) and the identification of the "extreme" globular cluster population in Carretta et al.(2009b). It is possible that some of the field stars closer to the main group in the Mg- Al plane are also escapees; confirming that would require a more detailed analysis. In this chapter we proceed with the four stars for which we are most confident. Table 3.2 lists 2MASS IDs, on-sky coordinates, stellar parameters, and Galactocentric distances for these stars, and Table 3.3 lists [Fe/H], [Na/Fe], [Mg/Fe], [Al/Fe], and [Si/Fe].

Figure 3.6 shows the Mg-Al abundance plane for the four candidate escapees along with literature data for globular clusters with similar metallicities, Figure 3.7 shows the same stars in the [Al/Fe] versus [Fe/H] plane, and Figure 3.8 carries the same data points into the [Al/Fe] versus [Si/Fe] plane. The abundances for M3 and M5, which are shown as small crosses, are from Masseron et al.(2019), and the abundances for NGC 288, NGC 1904, NGC 5904, NGC 6121, and NGC 6218 are marked with open circles and taken from Carretta et al.(2009b). We know that the extent of the Mg-Al anticorrelation varies in globular clusters, with less of a range in the more metal-rich or less massive clusters (Carretta et al., 2009b; Ventura et al., 2016). In Figure 3.7 we have plotted the more metal- rich clusters (e.g., NGC 288, NGC 6121) as open circles, and they show a less extended distribution than the more metal-poor clusters such as NGC 1904.

The candidate globular cluster escapees show a particularly nice consistency with the second-population stars from M3, M5, and NGC 1904 in the Al-Mg abundance plane. There seems to be an offset of about 0.2 dex in [Mg/Fe] between the Masseron et al.(2019) and Carretta et al.(2009b) data sets, and the different clusters display different morpholo- gies in this space. M3, M5 and NGC 1904 show definite depletions in Mg along with significant Al enhancement, while the remaining clusters cover a range in [Al/Fe] but do not show any clear signs of Mg depletion. The Al-Fe abundance plane shows that the four candidate escapees all fall within the normal metallicity range for halo globular clus- ters, and that their [Al/Fe] abundances are well within the range of second-population cluster stars. In the Al-Si abundance plane we can also see that [Si/Fe] enhancements are rare in globular clusters, with NGC 6121 showing the clearest sign of an Al-Si correla- tion, and that our four candidate escapees are again perfectly in line with the behaviour Chapter 3. Globular cluster stars in the halo 106

FIGURE 3.6: A comparison of the Mg-Al anticorrelation for our four candi- date escapees with stars from seven globular clusters with similar [Fe/H]. of the second-population stars. Although we cannot include a figure showing the O- Na anticorrelation because of the challenges in deriving [O/Fe] from GALAH spectra for red giant branch stars, we note that all four of our candidate escapees are enhanced in [Na/Fe], which also provides support for the hypothesis that they were originally second-population globular cluster stars.

We can compare to studies in the literature that investigate the idea of globular clusters donating stars to the halo field to further evaluate whether it is reasonable to claim that 1% of our halo data set (4/445 stars) are globular cluster escapees. Martell et al.(2011) and Martell et al.(2016) and Koch, Grebel, and Martell (2019) use a chemical tagging approach similar to ours and find that 2.6% of halo stars have abundances similar to second-population globular cluster stars. The data set considered here is comparable to the samples used in those previous studies: 445 kinematically selected halo giants com- pared to 561 in Martell et al.(2011), 253 in Martell et al.(2016), and 4470 in Koch, Grebel, and Martell (2019). Using models for multi-generation globular cluster formation and subsequent mass loss, they find that the fraction of chemically tagged stars can be trans- lated to a total fraction of halo stars that originally formed in globular clusters between Chapter 3. Globular cluster stars in the halo 107

FIGURE 3.7: [Al/Fe] versus [Fe/H] for our four candidate escapees, com- pared with stars from seven globular clusters with similar [Fe/H]. Chapter 3. Globular cluster stars in the halo 108

FIGURE 3.8: A comparison of the Al-Si correlation for our four candidate escapees with stars from seven globular clusters with similar [Fe/H]. Chapter 3. Globular cluster stars in the halo 109

TABLE 3.2: Four globular cluster escapees and their stellar parameters.

2 Star ID α/° δ/° Teff/K log g/log(cm/s )RGC/kpc 03202568-7318403 50.107222 -73.311202 5324.3 2.603 9.03 16162165-0413068 244.090200 -4.218632 4480.6 2.253 8.82 14132329-2806324 213.346995 -28.109119 5045.6 2.901 8.75 09370287-0724514 144.261972 -7.414322 4738.5 1.889 14.15

10% and 20%, depending on assumptions. They also found that the fraction of halo stars with globular cluster origins declines around ≈ 30 kpc from the Galactic center, which is a distance range that also contains most of the globular cluster system. In contrast, Reina-Campos et al.(2020) use E-MOSAICS simulations of the formation of Milky Way- like galaxies to study the mass fraction of halo stars contributed by globular clusters and predicts a limit of 2-5%. This is a lower fraction than the earlier observational studies found, but our 1% return is potentially consistent with this theoretical result, depending on models for globular cluster formation and mass loss.

We also want to evaluate whether our four escapees are consistent with the Galactocentric distances of known escapees, and with the spatial and kinematic locations of globular clusters. As stated in Table 3.2, the four stars are all located less than 20 kpc from the Galactic centre. Globular clusters can be located anywhere from 0.5 kpc to over 120 kpc from the Galactic centre, and 86% of the clusters are located within 20 kpc (Harris, 2010). Since our four escapees are all taken from GALAH data, they are all relatively close to the Sun, and so they have Galactocentric distances in the range 8-15 kpc, where 21% of globular clusters are located.

We began this study with a set of kinematically selected halo stars, and so it is not a surprise that our four escapees lie in the halo region of the Toomre diagram and the Lindblad diagram, as shown in Figure 3.9 and Figure 3.10. In Figure 3.9 we can see that the most metal-poor escapee is the most different from a disk orbit. In Figure 3.10 we also include globular clusters as blue crosses, using the calculations from Vasiliev (2019). Halo stars and halo globular clusters are distributed quite similarly in this space, with a slight concentration toward lower-energy and more disk-like orbits.

Two of our escapees are located quite close to globular clusters in this diagram: 16162165- 0413068 (blue star, prograde orbit) is near NGC 6352 and 14132329-2806324 (red star) is near NGC 6205. However, the metallicity of 16162165-0413068 is 0.45 dex below the mean metallicity of NGC 6352 (-0.64 dex, Harris, 2010), indicating that the star did not originate from that particular cluster. Interestingly, the metallicity of 14132329-2806324 is just 0.08 dex below the mean for NGC 6205, and so even though the separation between the star and the cluster is 73◦, it is possible that the star was once a member of the cluster. There are a number of further tests that would help to clarify whether our candidate escapees Chapter 3. Globular cluster stars in the halo 110

FIGURE 3.9: The four globular cluster escapees are plotted as stars in the Toomre diagram and colour-coded by [Fe/H], with the other halo giants in green.

TABLE 3.3: Key abundances for our four globular cluster escapees.

Star ID [Fe/H] [Na/Fe] [Mg/Fe] [Al/Fe] [Si/Fe] 03202568-7318403 -1.34±0.14 0.27±0.08 -0.05±0.11 0.63±0.09 0.27±0.08 16162165-0413068 -1.09±0.11 0.49±0.10 -0.17±0.16 1.10±0.11 0.49±0.10 14132329-2806324 -1.45±0.17 0.37±0.10 -0.18±0.16 1.10±0.11 0.37±0.10 09370287-0724514 -1.10±0.09 0.17±0.06 0.17±0.08 0.98±0.07 0.17±0.06 can be pinned to particular globular clusters of origin including an assessment of the likelihood for the cluster to have lost stars based on its structural parameters and orbit (e.g., Gnedin and Ostriker, 1997) and an investigation of whether the star’s current orbit is consistent with a reasonable ejection velocity from the cluster (as in Lind et al., 2015), and this would make for very interesting future work.

Further discussion on the LZ − Etot plane for the halo giants will be done in Chapter4, where we investigate the ages and chemodynamics of halo stars that do not show any resemblance to globular clusters. Chapter 3. Globular cluster stars in the halo 111

FIGURE 3.10: The four globular cluster escapee-candidates in the integrals of motion plane, LZ versus Etot, comparing their orbital parameters with the other halo giants in green and globular clusters as blue crosses. The vertical red line marks the division between prograde orbits (right) and retrograde orbits (left). Chapter 3. Globular cluster stars in the halo 112

3.5 Summary and future works

In this Chapter, we have shown that some of the metal-poor halo stars observed by GALAH do carry the characteristic light-element anticorrelations that are universal in globular clusters. We have restricted our search to stars in the parameter space [Fe/H]< 2 −1.0, Teff ≤ 5500 K and log g ≤ 4.0 cm/s . Given the limitations of GALAH abundances for metal-poor giants, we have found that depletion in Mg and enhancement in Al are the most informative chemical tags for this work.

Using these criteria, we tag 1% (4/445) metal-poor halo stars with [Al/Fe] > 0.5 & [Mg/Fe] < 0.2 which indicated that the stars are possible escapees. This percentage is broadly con- sistent with previous chemical tagging studies (Martell et al., 2016; Koch, Grebel, and Martell, 2019), and the four stars are similar to second-population globular cluster stars in the Mg-Al anticorrelation. The 4 stars also fall into the same metallicity range, RGC range, and Etot − LZ space as globular clusters. These agreements all support previous work in the literature that says that globular clusters provide some of the stars in the in situ halo. Also, with the discovery of escapees in this chapter, we are currently writing this result up for publication (Mohd Saadon et al., in prep) and we will explore further into mass and halo fraction in response to the simulation by Reina-Campos et al., 2020 using observational data from the GALAH survey.

We also found that star 2MASS J19183781-1011360 has peculiarly high abundances for O, Na, Al, Sc, Ti, Cr, Co, Ni, Cu, Sr, Y, Mo and Nd. While we note the oddness of this abundance pattern, we we did not explore it further in this chapter as the star does not show any characteristics of globular cluster chemical abundance. Further investigation of the properties of this star would be an interesting topic for a future project.

There are four key improvements to our methods:

• Confirming the identification of escapees based on [Mg/Fe] and [Al/Fe] by adding more abundance information. We will look more closely at the Na abundances, and incorporate all reasonable O abundances. This step may expand our list of escapees, since more globular clusters exhibit O-Na anticorrelations than Mg-Al anticorrelations.

• Making a more liberal initial selection in the Mg-Al abundance space. We can then down-select to just the most likely escapees based on kinematics and further abun- dance data.

• Updating the kinematic information. A catalogue of kinematic data is being pro- duced as part of GALAH DR3, with updated distances and a somewhat more sophisticated Galactic potential than we used. This will likely move some stars Chapter 3. Globular cluster stars in the halo 113

around in the kinematic space, and so we will re-do our halo star selection and the kinematic comparison against the globular cluster population.

• Compiling the known globular cluster escapees, to enable a comparison of the vari- ous samples in kinematic and abundance space, and to evaluate the selection effects at work.

As part of the work to find stars with globular cluster-like abundance patterns in the Milky Way halo, we also documented and discussed the normal abundance patterns for halo stars. Hence, the next step we take in building our understanding of Galactic ar- chaeology for this thesis is to consider the chemical evolution of the halo. In Chapter4 we will study the age trends for the available elements in the GALAH survey. 114

Chapter 4

Age-abundance relations in the Milky Way halo

4.1 Introduction

The ages of stars are essential for inferring the developmental history of the Milky Way, but they can be difficult to determine precisely for individual stars (e.g., Soderblom et al., 2014; Randich et al., 2018), and ensemble ages are sometimes used as a substitute (Feuillet et al., 2019). The combination of stellar abundances and kinematics with age allows us to track chemical evolution and accretion across Galactic history, providing critical insights into both internal and external factors that have shaped its formation and evolution.

Chemical enrichment proceeds differently in systems with lower mass and metallicity (Grebel, 2012; Frebel and Norris, 2013; Ji, Frebel, and Bromm, 2015), and this can be seen in the abundance patterns of stars in ultrafaint dwarf galaxies, classical dwarf galax- ies, and the halo field. This chapter uses stellar age estimates from the GALAH iDR3 catalogue together with elemental abundances and stellar kinematics to analyse the age- abundance relationships in the Galactic halo. We compare GALAH abundance behaviour to age-abundance trends noted in the observational literature and predicted through so- phisticated chemical evolution modelling. We also consider the likelihood of an accretion origin for our halo stars, and compare the age-abundance behaviour for stars likely to have formed in situ vs stars likely to have been accreted from dwarf galaxies.

This is a longstanding topic of research, and there is a rich literature that applies age- abundance-kinematics data to developing galactic chemical evolution models (e.g., Tins- ley, 1979; Gilmore, Wyse, and Kuijken, 1989; Chiappini, Matteucci, and Gratton, 1997; Schönrich and Binney, 2009; Kobayashi, Karakas, and Lugaro, 2020), tracking the origins of the thin and thick disc components of the Galaxy (e.g., Bensby, Feltzing, and Oey, 2014; Bovy et al., 2012; Haywood, 2013; Necib et al., 2020), and identifying debris from Chapter 4. Age-abundance relations in the halo 115 accretion events (e.g., Helmi et al., 1999; Koposov et al., 2015; Myeong et al., 2019; Krui- jssen et al., 2019; Naidu et al., 2020). Some studies also bring asteroseismology into the mix, since data from space missions like CoRoT and Kepler allow for precise determina- tions of intrinsic stellar properties including age. For example, Sharma et al.(2016) and Sharma et al.(2017) used age-sensitive asteroseismic data from the Kepler mission as part of a comprehensive model of the Galaxy, uncovering a discrepancy in the mass distri- bution of stars between the data and the Galactic models and revising the metallicity of the thick disk upward. Recently, Montalbán et al.(2020) combined asteroseismology with stellar abundances and kinematics to determine ages for stars observed by the Kepler space mission, including ancient stars that were formed inside the Milky Way, but also stars that formed elsewhere and were subsequently accreted into the Galaxy. One of the most intriguing lines of research in this work is exploring to what degree the accreted halo can be separated into discrete progenitors. We will touch briefly on this question in this Chapter, although GALAH is not the ideal data set to answer it.

The chemical evolution and assembly history of the halo is quite distinct from other com- ponents of the Galaxy, and it is our focus in this Chapter. The halo is old, and its main sources of chemical evolution are different to the disk. (e.g., Casali et al., 2020) find that different regions of the Milky Way follow different age-abundance relations. The halo also gives us relatively easy observational access to the chemical evolution process in dwarf galaxies, since accreted dwarf galaxy stars in the halo are much closer and easier to observe than dwarf galaxy stars that are still in dwarf galaxies. Chapter 4. Age-abundance relations in the halo 116

4.2 Elements of interest for age estimation

GALAH provides abundances for a large number of elements along with stellar ages which are estimated as described in Section 4.3. This data enables us to investigate the age-abundance relationship in the halo in terms of chemical evolution processes. The abundances of Fe and the α elements (e.g., Mg, Si, S, Ca, Ti) are widely used to con- strain the chemical evolution of our galaxy (e.g., Haywood et al., 2013) because they are connected quite simply to major nucleosynthetic channels. The [C/N] abundance ratio returns an age estimate for red giant branch stars that have recently experienced first dredge-up (e.g., Martig et al., 2016; Ness et al., 2019). Neutron-capture elements are also useful in the halo for investigating processes of chemical enrichment and stellar origins, and abundance ratios such as [Mg/Mn] can be used as a proxy for stellar age. Neutron capture elements are produced in a number of astrophysical sites, with different yields and timescales, and they can provide insight on processes as different as stellar winds and neutron star mergers. Halo stars, as an old population, provide an important opportunity to study the action of nucleosynthetic processes at early times and low metallicities.

Several authors have developed “chemical clocks”, which are age indicators calibrated to specific elemental abundances or abundance ratios measured in samples of stars with well-determined ages. There are two classes of chemical clocks: those that are based on the ratio between chemical elements generated by different stellar progenitors (and there- fore on different timescales), and clocks based on the ratio between elements modified by stellar evolution, which is heavily influenced by stellar mass and therefore lifetime (e.g., Titarenko, 2018; Casali et al., 2020). The fundamental purpose of the clock is to determine how key abundance ratios vary in stars with calibrated stellar ages, enabling us to derive the ages for any large sample of stars through empirical relations. A study by da Silva et al.(2012) on the relation between chemical abundances and stellar age, which was fol- lowed by Nissen (2015), Spina et al.(2016), Nissen et al.(2017), and Spina et al.(2018) and Delgado Mena et al.(2019), found that [Y/Mg] and [Y/Al] are useful as chemical clocks. Follow-up work by Feltzing et al.(2017) and Delgado Mena et al.(2019) found that these “clocks” are only valid in one particular environment; in this case, for Solar twins near the Sun but not anywhere else. Meanwhile, Skúladóttir et al.(2019) also worked on [Y/Mg], [Ba/Mg] and a few other chemical clocks, but her results are only known to be valid in nearby dwarf galaxies. Using GALAH data, Lin et al.(2020) found a reasonably sharp trend between the [Y/Mg] ratio and age for turnoff stars in the Solar neighbourhood.

Previous chapters have presented the abundances of all the available elements in the GALAH data set. Through the remainder of this Chapter we will focus on elements that are important to understanding the chemical evolution of the halo, and discuss their age- abundance variation. Chapter 4. Age-abundance relations in the halo 117

4.2.1 [s/α] ratios

As discussed in Chapter2, all elements heavier than the iron peak are formed through neutron capture, with the yields that depend on the neutron exposure. The main site for slow neutron capture (the s-process) is late stages of shell burning in AGB stars. The yields of AGB stars are dependent on their mass and metallicity (e.g., Karakas and Lat- tanzio, 2014; Karakas, 2016), and their delay times are intermediate between Type II su- pernovae and Type Ia supernovae. This explains two of the aspects of s-process abun- dances that are important for this Chapter: they are useful as chemical clocks, because they are produced on a different timescale to the α elements, and their age-abundance relation is sensitive to the star-formation history and the environment, because the tim- ing, the amount of enrichment, and the details of the abundance pattern produced are all dependent on the mass and metallicity of the AGB stars responsible. Two particu- lar [s/α] clocks that have been studied are [Y/Mg] and [Ba/Mg], and both ratios are known to increase with metallicity and time. Since Type Ia supernovae do not produce neutron-capture elements, we can use [Fe/H] (which is strongly influenced by Type Ia supernovae) as an independent axis to look at the evolution of s-process abundances and the [s/α] ratio. Chapter 4. Age-abundance relations in the halo 118

4.3 Ages and abundances in the GALAH halo data

4.3.1 The data set

As in previous chapters, we are using abundance data from GALAH iDR3. We begin with the set of halo giants introduced in Chapter3: stars that are not spatially coincident with any globular clusters, that have reliable stellar parameters and metallicities according to the GALAH catalog flags flag_sp and flag_fe_h, reliable measurements for O, Na, Mg, and Al according to the flags flag_o_fe, flag_na_fe, flag_mg_fe, and flag_al_fe, total space velocity more than 200 km s−1 different from the Local Standard of Rest, [Fe/H]≤

−1.0, Teff ≤ 5500 K, log g ≤ 4.0, and signal-to-noise ratio in the green camera of at least 20 per pixel. To this we add age information derived using the Elli code described in Lin et al.(2018).

That code uses a Bayesian approach to isochrone fitting, solving simultaneously for the true parameters of each star in the “posterior” (age (τ), evolutionary phase, bulk metallic- ity ([Fe/H]bulk), K-band extinction (AK) and parallax (ω ¯ sample)) by comparing the real and theoretical values for the observed quantities (Teff, log g, surface metallicity ([Fe/H]surf), apparent K magnitude (mK), and parallax (ω ¯ model)) using a likelihood function. In the Bayesian framework, the posterior can be written as in equation 4.1.

p1(τ, M, [Fe/H]bulk,AK,ω ¯ sample, θ|Teff, logg, [Fe/H]surf, mK,ω ¯ model, D) (4.1) ∝ L(τ, M, [Fe/H]bulk,AK,ω ¯ sample, θ)π

Where π is the prior, D is any additional measurements of the stellar properties which can be used to constrain its parameters, L is the likelihood, and θ is any additional model parameters that might be sampled in the future. For the prior probability in mass, a flat initial mass function has been adopted with the mass range chosen to span the isochrone grid:

 1 1 for 300M > M > 0.1M π(M) = (4.2) 0 else

Further detailed description of the Elli code and Bayesian analysis can be referred to in Lin et al.(2018).

MIST isochrones ((Choi et al., 2016)) are used to translate posterior values to theoretical values for the observed parameters. The emcee code (Foreman-Mackey et al., 2013) is used to carry out an ensemble Markov Chain Monte Carlo analysis, sampling the posterior Chapter 4. Age-abundance relations in the halo 119 space with a population of “walkers” that are moved through that space in steps that are controlled by the likelihood function. The end result of the process is a collection of walker positions across many steps, which can be interpreted as a set of probability distributions for each of the posterior parameters.

Isochrone fitting, however sophisticated the optimization scheme, is always limited in its precision by the density of isochrones at different evolutionary phases and the stellar evolutionary models they are built on. In this case, Elli uses MIST isochrones with −4 ≤ [Fe/H] ≤ +0.5 and [α/Fe] = 0 (Choi et al., 2016). The [α/Fe] enhancement affects the stellar evolutionary tracks of RGB stars, and this acts to artificially inflate the ages derived for old stars (Lin et al., 2018). Ages from Elli are most precise near the main-sequence turnoff and on the subgiant branch, where isochrones of different ages are most different in the observable space. Stars of different ages on the red giant branch are not very widely separated in their observable quantities, with the result that the uncertainties on age estimates for those stars can be quite large. In this study we assign age uncertainties of 50% on the main sequence and red giant branch. Considering the large uncertainties on the ages, and our interest in the chemical evolution of the old halo population, we exclude stars with age below 5 Gyr. Applying this age selection to the halo giants data set from Chapter3 removes 17 stars and leaves 428 stars for this old halo data set. The same age selection applied to the full halo data set removes 580 stars and retains 6621 stars as an old halo comparison set, which we use to contrast against our old halo data set in this Chapter. In summary for this stage, the difference between the data set and the comparison set is that the data set used for the analysis discussion in this Chapter only consists of halo giants with total space velocity more than 200 km s−1 different from the

LSR, [Fe/H]≤ −1.0, Teff ≤ 5500 K, log g ≤ 4.0, age ≥ 5 Gyr, green channel signal-to- noise ratio ≥ 20 per pixel and the consideration of GALAH catalog flags; meanwhile, the −1 comparison sets are the stars that only fulfill the criteria of Etot > 200 km s from the LSR.

Figure 4.1 shows effective temperature versus surface gravity for the old halo data set, with points colour coded by age, and the old halo comparison set included as pink points in the background. The requirements on abundance reliability using flag_X_fe impose a significant selection effect on the data. Although they are kinematically halo stars, and certainly old giants, the more metal-poor and low-gravity red giant branch stars are rejected from the old halo data set. Buder et al.(2018) contains useful discussion of the capabilities of the GALAH spectroscopic analysis process for metal-poor stars. Chapter 4. Age-abundance relations in the halo 120

FIGURE 4.1: Teff versus log g for our old halo data set, colour-coded by age. The old halo comparison set is shown in the background as pink points. Chapter 4. Age-abundance relations in the halo 121

FIGURE 4.2: A histogram of ages for our halo giants. Chapter 4. Age-abundance relations in the halo 122

4.3.2 The age-metallicity relation

Metallicity is a fundamental attribute of individual stars and stellar populations. We know from observations and expect from theory that metallicity is strongly correlated with the stellar mass of galaxies (e.g., Lequeux et al., 1979; Ogando et al., 2008; Tiwari, Mahajan, and Singh, 2020) and over time, as galaxies evolve, so does their metallicity (Terlevich and Forbes, 2002). This correlation, based on external galaxies across a wide mass range, is also valid for smaller-scale structure, including the stellar metallicity in the Milky Way (e.g., Gallazzi et al., 2005).

Hence, many researchers have been studying the age-metallicity relation in the Galactic disk. The increase in metallicity over time is more rapid when the star formation rate is higher (e.g., Strobel, 1991; Finlator and Davé, 2008; Leaman, VandenBerg, and Mendel, 2013; Mackereth et al., 2018; Kobayashi, Karakas, and Lugaro, 2020). The evolution of metallicity with age can also be quantified from absorption-line indices (e.g., Tantalo and Chiosi, 2004). As a result, an observational characterization of the age-metallicity relation provides a valuable reference for Galactic chemical evolution models and cosmological simulations (Feuillet et al., 2019, and references therein). The age-metallicity relation in the Galactic halo is both a simpler and more complex topic than in the disk. Because halo stars are typically older than disk stars, they have been enriched by fewer sources, but because a significant fraction of halo stars were accreted from dwarf galaxies, they have been enriched in a wide variety of environments. Our goal in considering the age- metallicity and age-abundance relations in the Galactic halo is to make a preliminary in- vestigation into how well the GALAH abundance data can follow the different chemical enrichment pathways at work.

Figure 4.2 shows a histogram of the Elli ages for our stars, and they tend to fall near 8 Gyr. This is interesting from the perspective of recent work on the origins of the halo, since Helmi (2020) found that a significant fraction of the local halo could be attributed to the Gaia-Enceladus remnant, and those stars should be no younger than 8 or 9 Gyr. Considering the large uncertainties that must be applied to ages for stars on the red giant branch, our stars are all consistent with having ages up to the age of the Universe. This is reasonable for the halo, which appears to have experienced its last significant accretion event 8 to 9 Gyr ago (Font et al., 2006; Myeong et al., 2019) and has not been the site of active star formation since (Nidever et al., 2019).

Figure 4.3 shows the old halo comparison set (pink circles) and the old halo data set (black circles) in the age versus metallicity plane. The overall halo sample covers a wide range in both age and metallicity, and the lower envelope of the data shows an upward trend with decreasing age, indicating steady chemical evolution in the environments where these stars originally formed. The stars we identify as being old red giants are more Chapter 4. Age-abundance relations in the halo 123 constrained in both quantities, with a concentration in age near 8 Gyr and metallicities almost entirely confined to above -1.5.

The trend of metallicity rising with time meets our theoretical expectations: Kobayashi, Karakas, and Lugaro (2020) find that metallicity decreases with increasing age, especially in stars more than 5 Gyr old. In the halo model shown in Figure 2 of Kobayashi, Karakas, and Lugaro (2020), the age distribution peaks around 8 Gyr, which is consistent with our data. However, our result is younger than the halo age range of 11.5 - 12.5 Gyr reported in Carollo et al. (2016) and Carollo et al. (2018). This may result from the fact that our dataset contains relatively more metal-rich stars, and the uncertainty of about 5 Gyr on our age estimates.

In using red giant stars to study star formation history, we are susceptible to biases in age and metallicity that can cause us to undercount old and metal-poor stars (e.g., Bovy et al., 2014; Manning and Cole, 2017). This effect is likely at work in our study. However, the metallicity distribution in the Kobayashi, Karakas, and Lugaro (2020) halo model peaks around [Fe/H] ≈ −1.6, which is consistent with our old halo comparison sample, but our old halo data sample is restricted to a higher range in metallicity. It is therefore not a representative sample of the halo, and will not provide a useful view of chemical enrichment at low metallicity and early time. As noted above, this is a consequence of the limitations of the GALAH survey analysis process, which currently does not deliver reliable abundance results for metal-poor stars.

4.3.3 Age and other abundances

The age-metallicity relation is a powerful tool for understanding the process of galactic chemical evolution in different environments. GALAH provides us with a large set of abundance information, and in this section we look at how these other elements evolve along with metallicity.

Figure 4.4 shows abundance versus age for [α/Fe] and 30 individual elements from the GALAH iDR3 catalog. As in Figure 4.3, the old halo data set is shown in black, and the old halo comparison set is shown in pink. In each panel, we have applied an additional requirement that the specific abundance flag for that element flag_X_fe must be equal to 0 for a star to be plotted.The number of stars with an unflagged (reliable) abundance varies from element to element.

We can also use the [X/Fe]-[Fe/H] plane, shown in Figure 4.5, to look at the evolution of [α/Fe] and the same 30 elements along with the overall metallicity. [Fe/H] does not map linearly to time, but the precision of our metallicities is better than the precision of our ages, so both figures provide useful information. Chapter 4. Age-abundance relations in the halo 124

FIGURE 4.3: The age-metallicity plot for our old halo data set (black) and the old halo comparison set (pink) with unflagged [Fe/H] values. Chapter 4. Age-abundance relations in the halo 125

FIGURE 4.4: Abundance versus age for our old halo data set (black) and the old halo comparison set (pink). In both data sets, only stars with unflagged (reliable) abundances are included. Chapter 4. Age-abundance relations in the halo 126

FIGURE 4.5: Unflagged (reliable) abundances of [α/Fe] and 30 individual elements versus metallicity for the old halo giant data set, colour coded by age. Chapter 4. Age-abundance relations in the halo 127

In Figure 4.4 and 4.5 we see that A(Li) follows two plateaus in the age-abundance space, with main-sequence and turnoff stars at the higher abundance level and giants at the lower level due to dredge-up and ongoing internal depletion. This is borne out in Fig- ure 4.6, where we colour-code the A(Li) versus age plane by the surface gravity and the sharp drop in A(Li) as stars ascend to the red giant branch is clearly visible. Focusing on the stars in the old halo data set, five of them have A(Li)> 1.5, which is the canonical limit for lithium-rich giant stars, which are rare and thought to have undergone some form of Li enrichment via accretion or a short-lived internal production phase (e.g., Charbonnel and Balachandran, 2000). They are at the metal-rich end of the group, which supports the finding in Casey et al.(2019b) and Martell et al.(2020) that the occurrence rate of lithium- rich giants is higher at high metallicity, but this is really too small of a sample to draw any strong conclusions from.

There are unfortunately very few unflagged (reliable) C abundances in either the old halo data set or the old halo comparison set. Those stars we can determine [C/Fe] for are at the metal-rich end of the distribution, indicating that it is mainly a matter of the strength of the absorption features that limits our ability to determine C abundance. The light odd-Z elements Na, Al, K, and Sc all have Solar abundances in the younger stars, with an upturn for stars older than 10 Gyr and no particular trend with [Fe/H]. This pattern is consistent with the GCE model in Kobayashi, Karakas, and Lugaro (2020) as their abundances in chemical feedback are all dependent on the excess of neutrons from 22Ne and the metallicity of the progenitor.

The overall [α/Fe] and the individual α elements O, Mg, Al, Ca, and Ti all have slightly super-Solar abundances and show a steady rise to older stars, with a small slope with age and no clear correlation with [Fe/H]. This is expected behaviour for in situ halo stars, and the outliers with low abundance relative to the group, especially at low metallicity, are consistent (in this space) with having been accreted from dwarf galaxies.

Figure 4.7 takes a closer look at [α/Fe], Mg, Si, Ca, Ti, and Na (which we expect to evolve similarly to the α elements based on Kobayashi, Karakas, and Lugaro, 2020), since they are known to be meaningful for Galactic chemical evolution. There we can see an in- crease in abundance with age, though the lack of old stars in our data set prevents more thorough investigation. We do see a large scatter in [O/Fe], which we also encountered in the globular cluster data. Interestingly, even though the spread in [O/Fe] is large, the mean value of +0.64 agrees quite well with the NLTE abundance from Kobayashi, Karakas, and Lugaro (2020) and the LTE analysis of Fulbright and Johnson (2003). The mean value of [Na/Fe] is −0.15, and 77% of the stars in our old halo data set (330/428) show sub-Solar [Na/Fe] ratios with also a slight increase in the older stars.The majority of dSph stars also show sub-Solar [Na/Fe] abundance (Shetrone et al., 2003), suggesting Chapter 4. Age-abundance relations in the halo 128 the possibility that some of our stars might have origins in dSph galaxies.

Also in Figure 4.4, the abundances of the iron-peak elements V, Cr, Mn, Co, Ni, and Cu closely follow the Fe abundance because they are all produced together in Type Ia supernovae. Moving to the neutron-capture elements, we see a variety of behaviours. Eu, which is a pure r-process element, is more enhanced in older stars, and declines over time. Sm, which is also an r-process element, is measured in fewer stars but its abundance is consistent with the same pattern as Eu. Sr, an element produced by the s-process, is shown to be enhanced in Figure 4.3. [Sr/Fe] is scattered around 1.3 dex in younger stars (≈ 7 − 8 Gyr), while Mo, Ru, Zr can also be seen in stars in the same age range with abundances around +0.5 and a fairly large scatter. In Figure 4.3, the stars with age > 7 Gyr and enhanced Mo and Ru abundances have lower metallicity than to their older counterparts. Our successful Y measurements in halo giants seem to concentrate at age ≈ 7 Gyr, while for the old halo comparison set we see a higher density of stars with measured Y abundances at 12 Gyr.

Figure 4.8 displays the abundance ratio of s-process over r-process elements ([Y/Eu], [Ba/Eu] and [La/Eu]) and the ratio of light over heavy s-process elements ([Sr/Ba] and [Y/Ba]) versus metallicity, with the points colour-coded by age. In each panel, the num- ber of stars plotted depends on the number of stars with unflagged abundances for both elements. There is more scatter in [s/r] ratios at higher metallicity, but no obvious trend, and there is a tendency for higher [light s/heavy s] ratios in higher-metallicity stars. We see one old star at [Fe/H]= −1.1 that has a markedly low [s/r] ratio and a very high [Sr/Ba] ratio, and a mild tendency for older stars to have lower [La/Eu] ratios. Similar patterns in neutron capture abundances as a function of metallicity can also be seen in Venn et al.(2004) and Cowan and Sneden (2006), where they find that higher-metallicity stars tend to have more enhancement in the light s-process elements compared to their heavier counterpart, Ba.

Finally, we proceed to [s/α] ratios that are used as chemical clocks in the literature. In our sample, we consider [Ba/Mg] and [Y/Mg]. Figure 4.9 shows a slight downward trend for both [Ba/Mg] and [Y/Mg] as we move to older stars, but there is wide scatter, and we would not claim to be able to make a precise age calibration based on these figures. This is probably partially driven by the large age uncertainties, which would horizontally blur any correlation in this space. The young stars with the highest [Ba/Mg] and [Y/Mg] ratios are the most Ba- and Y-enhanced, since the [Mg/Fe] abundances are constrained to a fairly small range.

The slight decline over age in both [s/Mg] abundance ratios agrees qualitatively with Skúladóttir et al.(2019), but our results do not match theirs well in detail, and this is likely the environmental dependence discussed in Feltzing et al.(2017) and Casali et al. Chapter 4. Age-abundance relations in the halo 129

FIGURE 4.6: Lithium abundance versus age for the old halo comparison set (colour-coded by surface gravity) and the old halo data set (green crosses). Main-sequence and turnoff stars with log g > 3.7 tend to have a higher Li abundance at all ages than giants.

(2020). Both ratios are higher for younger stars and more metal-rich stars, but in our data the ratios are higher than in Skúladóttir et al.(2019) even though their data set has a mean [Fe/H]= −0.65 ± 0.35. Chapter 4. Age-abundance relations in the halo 130

FIGURE 4.7: A zoomed-in look at abundance versus age for our old halo giant sample (black) and the old halo comparison sample (pink) for [α/Fe], [Mg/Fe], [Si/Fe], [Ca/Fe], [Ti/Fe], and [Na/Fe]. Chapter 4. Age-abundance relations in the halo 131

FIGURE 4.8: Ratios of [Y/Eu], [Ba/Eu], [La/Eu], [Sr/Ba] and [Ba/Y] to examine enrichment in the s-process and r-process elements as a function of [Fe/H], with age as the colour variation. Chapter 4. Age-abundance relations in the halo 132

FIGURE 4.9: The chemical clocks [Ba/Mg] and [Y/Mg] versus age for our old halo data set.

4.4 The origins of GALAH halo stars

4.4.1 Identifying accreted halo stars based on kinematics

We have mentioned previously that the Galactic halo contains stars that were accreted from outside the Milky Way, and Helmi (2020) indicates that a significant fraction of stars in the local halo were accreted from Gaia-Enceladus1. These stars have metallicities in the range −2.0 <[Fe/H]< −1.0 with sub-Solar [Al/Fe] and super-Solar [Mg/Fe] (Das, Hawkins, and Jofré, 2020), ages of at least 8-9 Gyr, and high orbital eccentricity (e > 0.7).

To examine whether the stars in our old halo data set might belong to Gaia-Enceladus or other known halo substructures, we divide the halo kinematically, using the orbital energy Etot and the eccentricity e, and consider the abundance behaviour of stars that are kinematically consistent with Gaia-Enceladus and the Sagittarius stream. The halo is highly structured, as discussed in Chapter1. Figure 4.10, which we have adapted from Naidu et al.(2020), shows one possible schematic division of the halo into multiple coherent components. If we compare this with Figure 4.11, which shows the positions of our stars in the space of total orbital energy Etot and vertical angular momentum LZ, we find that some of our stars are kinematically consistent with Gaia-Enceladus (which is called “GSE” in their diagram), some are consistent with the Helmi stream (Helmi and White, 1999), and some are kinematically consistent with Sagittarius, although we do not expect to find stars as distant as the Sagittarius stream in the GALAH survey. We do not

1In the recent flurry of studies on kinematic accretion signatures based on Gaia DR2 data, this kinematic structure was dubbed “Gaia-Enceladus”, “The Blob”, and “The Sausage”. For simplicity, we will refer to the stream and its progenitor system as “Gaia-Enceladus”. Chapter 4. Age-abundance relations in the halo 133

FIGURE 4.10: A schematic of the various structures that have been intro- duced and identified by Naidu et al.(2020). We have adapted the figure to align with our coordinate system, so that stars on prograde orbits are on the right side of the figure. Chapter 4. Age-abundance relations in the halo 134

FIGURE 4.11: Lindblad diagram of the kinematic space for our old halo data set (with colour variation indicating the age), with the old halo com- parison set plotted as pink dots in the background. Chapter 4. Age-abundance relations in the halo 135

find stars in our old halo data set, and very few in the old halo comparison set, that fall in the kinematic space they assign to Arjuna, Sequoia and I’itoi. This is likely because of the very different data sets we are using. Naidu et al.(2020) take their data from the H3 survey (Conroy et al., 2019), which has a sparse sampling across a large volume of the halo, while the GALAH halo giants are confined to within ≈ 8 kpc of the Sun. The structure they observe will not necessarily be seen in our local halo data set.

Looking at Figure 4.11, we see that 84 % of the stars in the old halo data set (360/428) are in prograde motion. We can also see that the stars on the highest-energy orbits all tend to be on the young end of the distribution, while the most bound stars (with

Etot < −180 000) are almost all old. There is a mix of older and younger stars above that group, with LZ ≈ 0 and slightly higher Etot. This region of low angular momentum and moderate orbital energy is where the Gaia-Enceladus feature was first identified.

Figure 4.12 shows the distribution of eccentricity e versus age for our old halo data set, which is colour coded by [Fe/H], and for the old halo comparison set, which is plotted in the background with pink dots. Quite a lot of the stars have high eccentricities, and the stars associated with Gaia-Enceladus in the literature have eccentricity greater than 0.7. The stars with those high eccentricities span the full age range in our data set, and we expect members of Gaia-Enceladus to be at least 8 Gyr old (e.g., Vincenzo et al., 2019), so our interpretation is that the stars with Etot and LZ in the correct range, eccentricity above 0.7, and age above 8 Gyr could indeed have originated in Gaia-Enceladus.

To test this possibility, we turn next to the abundances, since we expect that members of Gaia-Enceladus should have [Al/Fe]< 0 and [Mg/Fe]> 0 (Das, Hawkins, and Jofré, 2020; Helmi, 2020). In Figure 4.13 we show the eccentricity versus age for our old halo data set, colour coded by [Mg/Fe] (left) and [Al/Fe] (right). For old stars on high- eccentricity orbits, we do indeed see super-Solar Mg abundances, but a number of those stars also have super-Solar Al abundances. Overall, only 20 of the stars in the old halo data set meet these abundance criteria, and 8 of those have Lz in the right range to be Gaia-Enceladus.

Instead of Gaia-Enceladus, we can consider the Sagittarius stream, since some of the stars in our old halo data set have kinematics similar to it. According to Figure 4.10, Sagittarius can be detected as a grouping at high energy and prograde rotation. We do have a few giants in this kinematic region, and they tend to be relatively young, with ages typically below 10 Gyr. This is consistent with de Boer, Belokurov, and Koposov (2015), who find that the star formation history in the Sagittarius dwarf was at its peak ≈ 7 − 11 Gyr ago and then dropped drastically ≈ 5 − 7 Gyr ago during its accretion encounter with the Milky Way. Hence, selecting stars that fall in this Lz − Etot region, have high eccentricity, e > 0.7, and age between 7 to 11 Gyr, we find 35 stars that are consistent with Chapter 4. Age-abundance relations in the halo 136

FIGURE 4.12: The eccentricity, e, against the estimated age for the old halo data set with metallicity as the colour variation (with the old halo compar- ison set as the background)

FIGURE 4.13: The eccentricity, e, against the estimated age for the old halo data set with [Mg/Fe] (left) and [Al/Fe] (right) as the colour code. Chapter 4. Age-abundance relations in the halo 137 an origin in Sagittarius. Given the large typical distance to the Sagittarius stream and the rather bright apparent magnitude limit for the GALAH survey, we do not expect to have actually observed any Sagittarius stream stars. However, we cannot rule out the possibility entirely, since the leading arm of the Sagittarius stream crosses the Galactic plane near the solar neighborhood (Majewski et al., 2003; Majewski et al., 2004), and there may be RGB members within a kiloparsec of us at the present time. Further investigation of these stars to test their origins is clearly needed.

4.4.2 Identifying accreted halo stars using chemical tagging

We can also make abundance-based divisions between potential in situ and ex situ stars, and then see whether the stars selected by this chemical tagging process have the kine- matics we expect for accreted stars. First we try a simple criterion suggested by Mon- talbán et al.(2020), who suggest that [Mg/Fe]= −0.2 [Fe/H]+0.05 can be used to distin- guish between in situ and ex situ stars. We apply this same criterion to our halo data set and show the results in Figure 4.14. From this figure we would classify 63% of the stars in our old halo data set (270/428) as having ex situ origins, in agreement with the claim from Helmi (2020) that the majority of halo stars originate outside the Galaxy.

We can then compare the kinematic behaviour of the stars identified as ex situ based on the Montalbán et al.(2020) abundance selection versus the stars identified as in situ. In Figure 4.15 we see that the two sets of stars are not drastically different in kinematic space, but the stars with in situ abundances (plotted with red dots) are more likely to be in prograde orbits and more likely to be tightly bound, whereas the stars with ex situ abundances (colour-coded by their ages) are more likely to have high total energies, especially near LZ = 0.

Within the ex situ group of stars, we see a strong tendency for older stars to have lower angular momentum and more radial orbits, while the younger stars have higher angular momentum and higher total energy. The kinematics and ages match well with the ranges given in Helmi (2020) and Naidu et al.(2020) for Gaia-Enceladus (older stars) and the Sagittarius dwarf (younger stars).

Our second attempt at chemical tagging to identify accreted stars uses the abundance space of [Mg/Mn] versus [Al/Fe]. This criterion was used by Das, Hawkins, and Jofré (2020) to identify stars in Gaia-Enceladus. Specifically, we look for stars with high [Mg/Mn] and low [Al/Fe] in our old halo data set. In Figure 4.16 we plot the [Mg/Mn] ratio against [Al/Fe], and select stars with [Mg/Mn]>[Al/Fe]+0.75 and [Al/Fe] ≤ 0.2 as candidate members of Gaia-Enceladus. Chapter 4. Age-abundance relations in the halo 138

FIGURE 4.14: [Mg/Fe] versus [Fe/H] for the old halo data set, with age included as the colour variation and the old halo comparison set plot- ted in the background in light grey. The division line at [Mg/Fe] = −0.2 [Fe/H]+0.05 is a simple criterion suggested by Montalbán et al.(2020) to separate between in situ stars (circled in red) and ex situ stars (circled in blue). Chapter 4. Age-abundance relations in the halo 139

FIGURE 4.15: Stars with in situ abundances from Figure 4.14 (red dots) and ex situ abundances (colour coded by age) are shown in the same Lz − E plane. The stars with in situ abundances are more likely to have prograde orbits and low total energies than their ex situ counterparts. Chapter 4. Age-abundance relations in the halo 140

FIGURE 4.16: The stars in green are possible members of Gaia-Enceladus based on their high [Mg/Mn] ratio and low [Al/Fe] abundance. Chapter 4. Age-abundance relations in the halo 141

FIGURE 4.17: Stars selected as candidate members of Gaia-Enceladus us- ing the chemical tag from Das, Hawkins, and Jofré (2020) are plotted on the Lindblad diagram and colour-coded by their age, with the remaining stars from the old halo data set plotted in the background in maroon. Chapter 4. Age-abundance relations in the halo 142

Figure 4.17 displays the stars selected by the chemical tagging criterion from Das, Hawkins, and Jofré (2020) in the LZ versus Etot plane. In contrast to Figure 4.15, which shows the kinematic comparison for the simpler chemical tagging criterion from Montalbán et al. (2020), in Figure 4.17 we do not find the chemically tagged stars to be preferentially located in the Gaia-Enceladus region of the LZ versus Etot plane. This is perplexing, since the chemical tagging selection and the kinematic selection match quite well in Das, Hawkins, and Jofré (2020). Regardless, we are encouraged that our old halo data set contains hints of the accreted stars we expect to make up a good fraction of the halo. Chapter 4. Age-abundance relations in the halo 143

4.5 Conclusion

In this Chapter we use 428 old red giants in the halo with abundance patterns taken from the GALAH survey and age estimates from Elli (Lin et al., 2018) to make a preliminary exploration of the chemical evolution and possible accretion origins of the Galactic halo. With this data set, we explore the variation of abundance with age and kinematics in the selected sample. Overall, the age-metallicity relation in the halo, as seen by GALAH, is what we would expect from the literature and from galactic chemical evolution models (e.g., Venn et al., 2004; Kobayashi, Karakas, and Lugaro, 2020). However, the large age uncertainties and known issues with the accuracy of abundance determination for low- gravity giants prevents us from saying anything more conclusive.

After discussing the age-metallicity relation for halo stars, we looked more closely at the individual elements. For the most part the elements behave consistently with the literature: Li drops sharply between main-sequence stars and the red giant branch, and the light elements follow patterns we expect from models of galactic chemical evolution such as Kobayashi, Karakas, and Lugaro (2020). The α elements are super-Solar, with a reasonably small scatter and an increase at lower metallicity, and the iron peak elements follow iron closely. We see an indication of a trend between the [r-process/s-process] ratio with age, but with only one old star with the necessary abundances available it is difficult to be certain. [Eu/Fe] is also increasingly enhanced for stars older than 9 Gyr, while [Ba/Fe] is more constant in time, with a significant scatter.

We also investigate whether we can divide our old halo data set into stars with ex situ and in situ origins, both kinematically and using chemical tagging. Beginning with the −1 kinematics, we do find a number of stars with e > 0.7 and −500 < LZ < 500 kpc km s , but only eight of those also have the age ≈ 10 Gyr, [Al/Fe]< 0 and [Mg/Fe]> 0 that Gaia-Enceladus stars have. We also find 35 stars that can be linked kinematically to the −1 Sagittarius stream, with age ≈ 7 − 11 Gyr, LZ ≈ 2000 kpc km s , Etot ≈ 75000, and e > 0.7. Using a simple chemical tagging criterion suggested by Montalbán et al.(2020) (dividing the data set at [Mg/Fe] = −0.2 [Fe/H]+0.05), we find that 63% of our sample have abundances consistent with ex situ origins, and some of those do indeed overlap with Gaia-Enceladus in the kinematic plane. However, we do not see any obvious group of stars with Gaia-Enceladus-like kinematics in the stars that pass the [Mn/Mg] versus [Al/Fe] selection from Das, Hawkins, and Jofré (2020).

This Chapter presents exciting possibilities for investigating stars with extragalactic ori- gins that have not been previously published for the GALAH survey. However, we have been significantly constrained by the small sample size, and in particular the limited metallicity range that remains in our old halo data set when we use a strict selection based on the abundance quality flags. The abundances that are flagged as unreliable Chapter 4. Age-abundance relations in the halo 144 show scatter and systematic errors, and using unflagged abundances could result in spu- rious trends.

The GALAH data set has excellent potential for studies of the Galactic halo. The topic of this Chapter also can be expanded more extensively in the future. Buder et al.(2020) is focusing on the topic of combined kinematic and abundance-based identification of accreted stars in the halo, and we have contributed to that work. Similar to Chapter3, improving the analysis precision for luminous red giants, the abundance reliability at low metallicity, and the age precision will enable important new research in halo assembly with the GALAH data set.

In the next Chapter we will summarise the results of our investigations into the suitability of the GALAH survey for studying Galactic archaeology, in particular with old stars, and we will give our best answers to the research questions that were set in Chapter1. 145

Chapter 5

Summary and conclusions

5.1 Thesis conclusion

Following the thesis objectives laid out in Chapter1, we have studied observational as- pects of globular clusters and halo stars in the GALAH survey. The work in this thesis is based on the GALAH iDR3 catalogue, which contains over 600,000 stars, and there are some updates still on-going that will become available with the upcoming public data release DR3.

To conclude this thesis, we will go through the questions set out in the research objectives and summarize the answers which have been covered in the research chapters.

5.1.1 Globular clusters in GALAH

What is the sample of globular cluster stars observed in the GALAH survey? In Chapter2, we identified globular cluster stars in GALAH data. We found that 12 clusters have stars in the iDR3 data set, and in four clusters there are more than five stars with reliable chemical abundances. These clusters are NGC 104 (also know as 47 Tuc), NGC 5139 (ω Centauri), NGC 6397, and NGC 7099 (M30). Ideally these stars will serve as stepping stones for GALAH to study more clusters in the future.

We found reasonable correspondence between the iDR3 metallicities and the literature, though there seems to be a systematic analysis problem in NGC 104. Limited information on C and O and no N abundances prevented us from examining any C-N-O variation within the sample. The overall abundance set is a bit inconsistent, in that some of the elements agree with the literature and others do not.

What is the abundance profile of those globular cluster stars? Even though light-element anticorrelations are a feature of (nearly) all globular clusters, each cluster is different. We confirm the existence of the correlation in Ba and Y against metallicity in NGC 5139 and the general overabundance of Ba in NGC 5139 irrespective Chapter 5. Conclusions 146 of stellar generation. We also confirm the Na-Li correlation reported by Johnson and Pilachowski (2010).

GALAH abundances for globular clusters in Chapter2 do not show any obvious Mg- Al anticorrelation. However, by using the [Al/Fe] ratio as in Mészáros et al.(2020), we confirm the presence of multiple populations in NGC 104 and NGC 5139, and hypoth- esize that the only stars in the metal-poor clusters NGC 6397 and NGC 7099 that have reliable GALAH abundances are in the second population. We also observe an anticor- relation between [Al/Fe] and [Ba/Fe] in NGC 104, contradicting the results of D’Orazi et al.(2010a).

5.1.2 Halo stars with globular cluster-like abundances

Can we find halo stars in GALAH with abundance patterns resembling those in glob- ular cluster stars? The objective of Chapter3 is to use the light-element anticorrelations investigated in Chapter2 as a reference for finding globular cluster escapees in the halo field. We make a kinematic halo selection, then take a subset that is only metal-poor red giant branch stars and focus on the Mg and Al abundances of those stars, especially those that might fall into the second-population category with [Al/Fe]> +0.3.

We found four stars that show high Al abundance and low Mg abundance out of 445 selected halo giants – approximately 1% of the sample – that are plausibly globular cluster escapees in the halo field. These four stars occupy the same range in metallicity and

Galactocentric distance, and the same region of the Etot − LZ kinematic space as globular clusters. These results all support previous work in the literature finding that globular clusters provide some of the stars in the in situ halo.

5.1.3 Age-abundance relations in halo giants

What is the relationship between age and abundance for the halo giants? Chapter4 uses the same set of kinematically selected halo stars as in Chapter3 with prob- abilistically determined ages from Lin et al.(2018), and explores the abundances of the halo stars more broadly. The metallicity in the data set covers the range −1.7 <[Fe/H]< −1.0, and the age distribution peaks at ≈ 6-8 Gyr. The age-metallicity range is consistent with literature expectations, the α elements tend to be enhanced and trend upward in older stars, and there is an apparent enhancement for some middle aged stars (7-10 Gyr) for s-process elements.

What chemical or kinematic evidence can we see in GALAH for accretion events in the halo? Chapter 5. Conclusions 147

We used kinematic selections for Gaia-Enceladus and Sagittarius from Naidu et al.(2020) and investigated whether the abundances of the selected stars were consistent with the “accreted” abundance patterns described by Helmi (2020). We found that 8 of our 428 old halo stars were consistent with Gaia-Enceladus in terms of kinematics, age, eccentricity, and abundance, and 35 were consistent with the Sagittarius stream when selected on the same criteria. We then inverted the process, using “accreted” abundance selections from the literature and then considering the orbital kinematics of the selected stars. 63% of the stars in our old halo data set have abundances consistent with the Montalbán et al.(2020) ex situ pattern, and some of those overlap kinematically with Gaia-Enceladus. However, none of the stars that meet the Das, Hawkins, and Jofré (2020) ex situ abundance criteria have kinematics that resemble Gaia-Enceladus.

5.2 Future extensions of this work

As noted in this thesis, the majority of the objects in the GALAH catalogue are disk stars. There are some limitations on GALAH’s ability to conquer the halo region as those stars can be quite distant and faint, well below the magnitude limit of the survey. It is a chal- lenge for the GALAH survey to gather large samples of stars in globular clusters, because their high on-sky density restricts the number of stars in the cluster that can be observed at one time.

The availability of so many chemical abundances (30 elements so far) makes it possible to see abundance behaviours for a large number of stars as a common group or as a substructure. Although determining C and O from GALAH data is still a challenge, improving the signal to noise ratio in the spectra may improve the reliability of both abundances. Li in GALAH has become a hot topic for researchers (e.g., Žerjal et al., 2019; Martell et al., 2020; Gao et al., 2020; Deepak, Lambert, and Reddy, 2020, and others). Heavier elements such as s- and r-process products are also observed by GALAH, which makes it possible to investigate more nucleosynthetic contributors to Galactic chemical evolution.

The potential of the GALAH survey is vast: it can make it possible to probe further into Galactic history, widen our understanding of stellar evolution and star formation sites, and use chemical tagging to find stars with similar abundance features, including globular cluster escapees and tidal debris from past merger events. The limitations and challenges that arose in this thesis make it clear that improving the analysis precision, especially for luminous red giants, will be very important as we continue to explore the history of the Galactic halo. Chapter 5. Conclusions 148

This thesis also serves as evidence that a multi-dimensional approach is the best way to establish a comprehensive understanding of the structure and history of the Galaxy, especially the stellar halo. The kinematics, abundances, and ages of stars all provide crit- ical information on the origins of halo stars and globular clusters. Such chemo-chrono- dynamical relations can also allow us to investigate the ex situ and in situ contributions to halo assembly, the interaction history of our Galaxy, and the evolution of its mass and gravitational potential. Updating the kinematic information with future data releases from the Gaia mission, more accurate distances, and a more sophisticated Galactic gravi- tational potential would allow a more precise identification of star clusters, halo streams, and other substructures. Machine learning techniques could allow expanded chemical tagging using GALAH data. A more complete and precise observational view of halo chemodynamics would allow detailed simulations of the formation and the evolution of the Galaxy and its neighbours, and could also answer longstanding questions on the origins of certain elements, including the r-process.

Future survey projects like WEAVE, 4MOST, DESI, and MSE will continue to expand the volume and detail of data available for astronomers to use, acting as a vessel to carry us into uncharted regions that could revolutionize our perspective on Galactic archaeology.

5.3 Research acknowledgement

This thesis is based upon data acquired through the Australian Astronomical Observa- tory. We acknowledge the traditional owners of the land on where the AAT stands, the Gamilaraay people, and pay our respects to elders past and present. We also acknowl- edge the use of data reduction infrastructure developed within the GALAH Survey col- laboration. The GALAH survey web site is www.galah-survey.org. 149

Bibliography

Abadi, Mario G. et al. (Nov. 2003). “Simulations of Galaxy Formation in a Lambda Cold Dark Matter Universe. II. The Fine Structure of Simulated Galactic Disks”. In: The As- trophysical Journal 597.1, pp. 21–34. DOI: 10.1086/378316. arXiv: astro- ph/0212282 [astro-ph]. Abate, Carlo, Richard J. Stancliffe, and Zheng-Wei Liu (Mar. 2016). “How plausible are the proposed formation scenarios of CEMP-r/s stars?” In: Astronomy & Astrophysics 587, A50, A50. DOI: 10.1051/0004-6361/201527864. arXiv: 1601.00976 [astro-ph.SR]. Agertz, Oscar, Romain Teyssier, and Ben Moore (Jan. 2011). “The formation of disc galax- ies in a ΛCDM universe”. In: Monthly Notices of the Royal Astronomical Society 410.2, pp. 1391–1408. DOI: 10.1111/j.1365-2966.2010.17530.x. arXiv: 1004.0005 [astro-ph.CO]. Ahumada, Romina et al. (July 2020). “The 16th Data Release of the Sloan Digital Sky Surveys: First Release from the APOGEE-2 Southern Survey and Full Release of eBOSS Spectra”. In: The Astrophysical Journal Supplement Series 249.1, 3, p. 3. DOI: 10.3847/ 1538-4365/ab929e. arXiv: 1912.02905 [astro-ph.GA]. Alecian, G. (Dec. 2013). “Atomic diffusion in the atmospheres of upper main-sequence stars”. In: EAS Publications Series. Ed. by G. Alecian et al. Vol. 63. EAS Publications Series, pp. 219–226. DOI: 10.1051/eas/1363025. Amarsi, A. M. et al. (Aug. 2020). “The GALAH Survey: Non-LTE departure coefficients for large spectroscopic surveys”. In: arXiv e-prints, arXiv:2008.09582, arXiv:2008.09582. arXiv: 2008.09582 [astro-ph.SR]. Amôres, E. B., A. C. Robin, and C. Reylé (June 2017). “Evolution over time of the Milky Way’s disc shape”. In: Astronomy & Astrophysics 602, A67, A67. DOI: 10.1051/0004- 6361/201628461. arXiv: 1701.00475 [astro-ph.GA]. An, Deokkeun et al. (Nov. 2015). “The Fractions of Inner- and Outer-halo Stars in the Local Volume”. In: The Astrophysical Journal Letters 813.2, L28, p. L28. DOI: 10.1088/2041- 8205/813/2/L28. arXiv: 1510.07630 [astro-ph.GA]. Anders, F. et al. (Nov. 2018). “Dissecting stellar chemical abundance space with t-SNE”. In: Astronomy & Astrophysics 619, A125, A125. DOI: 10.1051/0004- 6361/201833099. arXiv: 1803.09341 [astro-ph.GA]. Bibliography 150

Antoja, T. et al. (Sept. 2018). “A dynamically young and perturbed Milky Way disk”. In: Nature 561.7723, pp. 360–362. DOI: 10.1038/s41586-018-0510-7. arXiv: 1804.10196 [astro-ph.GA]. Aoki, Wako et al. (Jan. 2007). “Carbon-enhanced Metal-poor Stars. I. Chemical Composi- tions of 26 Stars”. In: The Astrophysical Journal 655.1, pp. 492–521. DOI: 10.1086/509817. arXiv: astro-ph/0609702 [astro-ph]. Arnett, W. D. (Feb. 1975). “Supernova remnants and presupernova models.” In: Astro- physical Journal 195, pp. 727–733. DOI: 10.1086/153374. Arnett, W. David (May 1971). “Galactic Evolution and Nucleosynthesis”. In: Astrophysical Journal 166, p. 153. DOI: 10.1086/150947. Astropy Collaboration et al. (Oct. 2013). “Astropy: A community Python package for as- tronomy”. In: Astronomy & Astrophysics 558, A33, A33. DOI: 10 . 1051 / 0004 - 6361 / 201322068. arXiv: 1307.6212 [astro-ph.IM]. Astropy Collaboration et al. (Sept. 2018). “The Astropy Project: Building an Open-science Project and Status of the v2.0 Core Package”. In: Astronomical Journal 156.3, 123, p. 123. DOI: 10.3847/1538-3881/aabc4f. arXiv: 1801.02634 [astro-ph.IM]. Bailer-Jones, C. A. L. et al. (Aug. 2018). “Estimating Distance from Parallaxes. IV. Dis- tances to 1.33 Billion Stars in Gaia Data Release 2”. In: Astronomical Journal 156.2, 58, p. 58. DOI: 10.3847/1538-3881/aacb21. arXiv: 1804.10121 [astro-ph.SR]. Banik, Nilanjan et al. (Nov. 2019). “Evidence of a population of dark subhalos from Gaia and Pan-STARRS observations of the GD-1 stream”. In: arXiv e-prints, arXiv:1911.02662, arXiv:1911.02662. arXiv: 1911.02662 [astro-ph.GA]. Barba, Rodolfo H. et al. (Jan. 2019). “A Sequoia in the Garden: FSR 1758—Dwarf Galaxy or Giant Globular Cluster?” In: The Astrophysical Journal Letters 870.2, L24, p. L24. DOI: 10.3847/2041-8213/aaf811. arXiv: 1812.04999 [astro-ph.GA]. Barbuy, B. et al. (Jan. 2005). “New analysis of the two carbon-rich stars CS 22948-27 and CS 29497-34: Binarity and neutron capture elements”. In: Astronomy & Astrophysics 429, pp. 1031–1042. DOI: 10.1051/0004- 6361:20040415. arXiv: astro- ph/0409213 [astro-ph]. Barbuy, Beatriz, Cristina Chiappini, and Ortwin Gerhard (Sept. 2018). “Chemodynam- ical History of the Galactic Bulge”. In: Annual Review of Astronomy and Astrophysics 56, pp. 223–276. DOI: 10.1146/annurev- astro- 081817- 051826. arXiv: 1805.01142 [astro-ph.GA]. Barrett, P. et al. (Dec. 2005). “matplotlib – A Portable Python Plotting Package”. In: As- tronomical Data Analysis Software and Systems XIV. Ed. by P. Shopbell, M. Britton, and R. Ebert. Vol. 347. Astronomical Society of the Pacific Conference Series, p. 91. Bastian, Nate and Carmela Lardo (Sept. 2018). “Multiple Stellar Populations in Globular Clusters”. In: Annual Reviews of Astronomy & Astrophysics 56, pp. 83–136. DOI: 10.1146/ annurev-astro-081817-051839. arXiv: 1712.01286 [astro-ph.SR]. Bibliography 151

Baumgardt, H. et al. (Feb. 2019). “Mean proper motions, space orbits, and velocity dis- persion profiles of Galactic globular clusters derived from Gaia DR2 data”. In: Monthly Notices of the Royal Astronomical Society 482.4, pp. 5138–5155. DOI: 10 . 1093 / mnras / sty2997. arXiv: 1811.01507 [astro-ph.GA]. Beers, Timothy C. and Daniela Carollo (July 2009). “Galactic archaeology: Overcoming great barriers”. In: Nature Physics 5.7, pp. 463–464. DOI: 10.1038/nphys1327. Beers, Timothy C. and Norbert Christlieb (Sept. 2005). “The Discovery and Analysis of Very Metal-Poor Stars in the Galaxy”. In: Annual Reviews of Astronomy & Astrophysics 43.1, pp. 531–580. DOI: 10.1146/annurev.astro.42.053102.134057. Bekki, K. and K. C. Freeman (Dec. 2003). “Formation of ω Centauri from an ancient nu- cleated dwarf galaxy in the young Galactic disc”. In: Monthly Notices of the Royal Astro- nomical Society 346.2, pp. L11–L15. DOI: 10.1046/j.1365-2966.2003.07275.x. arXiv: astro-ph/0310348 [astro-ph]. Bekki, K. et al. (May 2007). “Origin of abundance inhomogeneity in globular clusters”. In: Monthly Notices of the Royal Astronomical Society 377.1, pp. 335–351. DOI: 10.1111/j. 1365-2966.2007.11606.x. arXiv: astro-ph/0702289 [astro-ph]. Bekki, Kenji (June 2005). “The Galactic Globular Cluster System as a Fossil Record of Reionization”. In: The Astrophysical Journal Letters 626.2, pp. L93–L96. DOI: 10.1086/ 431737. arXiv: astro-ph/0506529 [astro-ph]. – (Aug. 2017). “Globular cluster formation with multiple stellar populations: self-enrichment in fractal massive molecular clouds”. In: Monthly Notices of the Royal Astronomical Society 469.3, pp. 2933–2951. DOI: 10.1093/mnras/stx982. arXiv: 1705.10039 [astro-ph.GA]. – (July 2018). “Formation of globular clusters with multiple stellar populations from massive gas clumps in high-z gas-rich dwarf galaxies”. In: arXiv e-prints, arXiv:1807.02309, arXiv:1807.02309. arXiv: 1807.02309 [astro-ph.GA]. Bekki, Kenji and Takuji Tsujimoto (July 2017). “Formation of Globular Clusters with Inter- nal Abundance Spreads in r-Process Elements: Strong Evidence for Prolonged Star For- mation”. In: The Astrophysical Journal 844.1, 34, p. 34. DOI: 10.3847/1538-4357/aa77ae. arXiv: 1706.01194 [astro-ph.GA]. Bellini, A. et al. (Dec. 2009). “Radial distribution of the multiple stellar populations in ω Centauri”. In: Astronomy & Astrophysics 507.3, pp. 1393–1408. DOI: 10.1051/0004- 6361/200912757. arXiv: 0909.4785 [astro-ph.SR]. Belokurov, V. et al. (May 2006). “The Field of Streams: Sagittarius and Its Siblings”. In: The Astrophysical Journal Letters 642.2, pp. L137–L140. DOI: 10 . 1086 / 504797. arXiv: astro-ph/0605025 [astro-ph]. Belokurov, V. et al. (July 2018). “Co-formation of the disc and the stellar halo”. In: Monthly Notices of the Royal Astronomical Society 478.1, pp. 611–619. DOI: 10.1093/mnras/sty982. arXiv: 1802.03414 [astro-ph.GA]. Bibliography 152

Bensby, T., S. Feltzing, and M. S. Oey (Feb. 2014). “Exploring the Milky Way stellar disk. A detailed elemental abundance study of 714 F and G dwarf stars in the solar neigh- bourhood”. In: Astronomy & Astrophysics 562, A71, A71. DOI: 10.1051/0004- 6361/ 201322631. arXiv: 1309.2631 [astro-ph.GA]. Bernard, Edouard J. et al. (Dec. 2016). “A Synoptic Map of Halo Substructures from the Pan-STARRS1 3π Survey”. In: Monthly Notices of the Royal Astronomical Society 463.2, pp. 1759–1768. DOI: 10.1093/mnras/stw2134. arXiv: 1607.06088 [astro-ph.GA]. Bhardwaj, Anupam (Nov. 2018). “Cepheid and RR Lyrae Variables as Standard Candles and What Else?” In: Stellar Populations and the Distance Scale. Ed. by Joseph Jensen, R. Michael Rich, and Richard de Grijs. Vol. 514. Astronomical Society of the Pacific Conference Series, p. 153. arXiv: 1806.03021 [astro-ph.SR]. Bland-Hawthorn, J., K. Freeman, and F. Matteucci (Feb. 2014). The Origin of the Galaxy and Local Group: Saas-Fee Advanced Course 37 Swiss Society for Astrophysics and Astronomy. Vol. 37. Berlin: Springer Science & Business Media. ISBN: 3642417205. Bland-Hawthorn, Joss and Ortwin Gerhard (Sept. 2016). “The Galaxy in Context: Struc- tural, Kinematic, and Integrated Properties”. In: Annual Reviews of Astronomy & As- trophysics 54, pp. 529–596. DOI: 10 . 1146 / annurev - astro - 081915 - 023441. arXiv: 1602.07702 [astro-ph.GA]. Bland-Hawthorn, Joss et al. (June 2019). “The GALAH survey and Gaia DR2: dissecting the stellar disc’s phase space by age, action, chemistry, and location”. In: Monthly No- tices of the Royal Astronomical Society 486.1, pp. 1167–1191. DOI: 10.1093/mnras/stz217. arXiv: 1809.02658 [astro-ph.GA]. Blöcker, T. et al. (May 1998). “Lithium Depletion in the Sun: A Study of Mixing Based on Hydrodynamical Simulations”. In: Space Science Reviews 85, pp. 105–112. DOI: 10.1023/ A:1005192804708. arXiv: astro-ph/9806310 [astro-ph]. Bloecker, T. and D. Schoenberner (Apr. 1991). “A 7M AGB model sequence not complying with the core mass-luminosity relation.” In: Astronomy & Astrophysics 244, pp. L43–L46. Bonaca, Ana et al. (July 2020). “Timing the Early Assembly of the Milky Way with the H3 Survey”. In: The Astrophysical Journal Letters 897.1, L18, p. L18. DOI: 10.3847/2041- 8213/ab9caa. arXiv: 2004.11384 [astro-ph.GA]. Bonaparte, I. et al. (Nov. 2013). “Galactic and cosmic Type Ia supernova (SNIa) rates: is it possible to impose constraints on SNIa progenitors?” In: Monthly Notices of the Royal Astronomical Society 435.3, pp. 2460–2473. DOI: 10.1093/mnras/stt1457. arXiv: 1308.0137 [astro-ph.CO]. Bond, H. E. (Sept. 1970). “A Search for Metal-Deficient Stars”. In: The Astrophysical Journal Supplement Series 22, p. 117. DOI: 10.1086/190220. Bonifacio, P. and P. Molaro (Mar. 1997). “The primordial lithium abundance”. In: Monthly Notices of the Royal Astronomical Society 285.4, pp. 847–861. DOI: 10.1093/mnras/285.4. 847. arXiv: astro-ph/9611043 [astro-ph]. Bibliography 153

Bonifacio, P. et al. (July 2002). “The lithium content of the globular cluster NGC 6397”. In: Astronomy & Astrophysics 390, pp. 91–101. DOI: 10.1051/0004-6361:20020620. arXiv: astro-ph/0204332 [astro-ph]. Borsato, Nicholas W., Sarah L. Martell, and Jeffrey D. Simpson (Feb. 2020). “Identifying stellar streams in Gaia DR2 with data mining techniques”. In: Monthly Notices of the Royal Astronomical Society 492.1, pp. 1370–1384. DOI: 10.1093/mnras/stz3479. arXiv: 1907.02527 [astro-ph.GA]. Boubert, D. et al. (Sept. 2018). “Revisiting hypervelocity stars after Gaia DR2”. In: Monthly Notices of the Royal Astronomical Society 479.2, pp. 2789–2795. DOI: 10 . 1093 / mnras / sty1601. arXiv: 1804.10179 [astro-ph.GA]. Bovy, Jo (Feb. 2015). “galpy: A python Library for Galactic Dynamics”. In: The Astrophysi- cal Journal Supplement Series 216.2, 29, p. 29. DOI: 10.1088/0067-0049/216/2/29. arXiv: 1412.3451 [astro-ph.GA]. Bovy, Jo, David W. Hogg, and Sam T. Roweis (June 2011). “Extreme deconvolution: Infer- ring complete distribution functions from noisy, heterogeneous and incomplete obser- vations”. In: Annals of Applied Statistics 5.2, pp. 1657–1677. DOI: 10.1214/10-AOAS439. arXiv: 0905.2979 [stat.ME]. Bovy, Jo et al. (July 2012). “The Spatial Structure of Mono-abundance Sub-populations of the Milky Way Disk”. In: The Astrophysical Journal 753.2, 148, p. 148. DOI: 10.1088/ 0004-637X/753/2/148. arXiv: 1111.1724 [astro-ph.GA]. Bovy, Jo et al. (Aug. 2014). “The APOGEE Red-clump Catalog: Precise Distances, Veloci- ties, and High-resolution Elemental Abundances over a Large Area of the Milky Way’s Disk”. In: The Astrophysical Journal 790.2, 127, p. 127. DOI: 10.1088/0004-637X/790/2/ 127. arXiv: 1405.1032 [astro-ph.GA]. Boyles, J. et al. (Nov. 2011). “Young Radio Pulsars in Galactic Globular Clusters”. In: The Astrophysical Journal 742.1, 51, p. 51. DOI: 10.1088/0004-637X/742/1/51. arXiv: 1108. 4402 [astro-ph.SR]. Briley, Michael M., Judith G. Cohen, and Peter B. Stetson (Mar. 2004). “The Chemical Inhomogeneity of Faint M13 Stars: Carbon and Nitrogen Abundances”. In: Astronom- ical Journal 127.3, pp. 1579–1587. DOI: 10 . 1086 / 382100. arXiv: astro - ph / 0312315 [astro-ph]. Brodie, Jean P. and Jay Strader (Sept. 2006). “Extragalactic Globular Clusters and Galaxy Formation”. In: Annual Reviews of Astronomy & Astrophysics 44.1, pp. 193–267. DOI: 10. 1146/annurev.astro.44.051905.092441. arXiv: astro-ph/0602601 [astro-ph]. Brodie, Jean P. et al. (Nov. 2012). “The SLUGGS Survey: NGC 3115, A Critical Test Case for Metallicity Bimodality in Globular Cluster Systems”. In: The Astrophysical Journal Letters 759.2, L33, p. L33. DOI: 10.1088/2041- 8205/759/2/L33. arXiv: 1209.5390 [astro-ph.CO]. Bibliography 154

Buder, Sven et al. (Aug. 2018). “The GALAH Survey: second data release”. In: Monthly No- tices of the Royal Astronomical Society 478, pp. 4513–4552. DOI: 10.1093/mnras/sty1281. arXiv: 1804.06041 [astro-ph.SR]. Buder, Sven et al. (Nov. 2020). “The GALAH+ Survey: Third Data Release”. In: arXiv e- prints, arXiv:2011.02505, arXiv:2011.02505. arXiv: 2011.02505 [astro-ph.GA]. Bullock, J. S. and K. V. Johnston (Dec. 2005). “Tracing Galaxy Formation with Stellar Ha- los. I. Methods”. In: The Astrophysical Journal 635, pp. 931–949. DOI: 10.1086/497422. eprint: astro-ph/0506467. Burbidge, E. Margaret et al. (Jan. 1957). “Synthesis of the Elements in Stars”. In: Reviews of Modern Physics 29.4, pp. 547–650. DOI: 10.1103/RevModPhys.29.547. Camargo, D. and D. Minniti (Jan. 2019). “Three candidate globular clusters discovered in the Galactic bulge”. In: Monthly Notices of the Royal Astronomical Society, p. L17. DOI: 10.1093/mnrasl/slz010. arXiv: 1901.08574 [astro-ph.GA]. Campbell, Simon W. et al. (June 2013). “Sodium content as a predictor of the advanced evolution of globular cluster stars”. In: Nature 498.7453, pp. 198–200. DOI: 10.1038/ nature12191. arXiv: 1305.7090 [astro-ph.SR]. Cannon, R. D. et al. (Aug. 1998). “Carbon and nitrogen abundance variations on the main sequence of 47 Tucanae”. In: Monthly Notices of the Royal Astronomical Society 298.2, pp. 601–624. DOI: 10.1046/j.1365-8711.1998.01671.x. Carollo, D. et al. (Dec. 2016). “The age structure of the Milky Way’s halo”. In: Nature Physics 12.12, pp. 1170–1176. DOI: 10.1038/nphys3874. arXiv: 1607.08628 [astro-ph.GA]. Carollo, Daniela et al. (May 2018). “The Origin of the Milky Way’s Halo Age Distribu- tion”. In: The Astrophysical Journal Letters 859.1, L7, p. L7. DOI: 10.3847/2041-8213/ aac2dc. arXiv: 1803.05154 [astro-ph.GA]. Carrera, R. et al. (Mar. 2019). “Open clusters in APOGEE and GALAH. Combining Gaia and ground-based spectroscopic surveys”. In: Astronomy & Astrophysics 623, A80, A80. DOI: 10.1051/0004-6361/201834546. arXiv: 1901.09302 [astro-ph.GA]. Carretta, E. et al. (Mar. 2004). “Abundance analysis of turn-off and early subgiant stars in the globular cluster 47 Tuc (NGC 104)”. In: Astronomy & Astrophysics 416, pp. 925–940. DOI: 10.1051/0004-6361:20034370. arXiv: astro-ph/0311347 [astro-ph]. Carretta, E. et al. (Oct. 2009a). “Na-O anticorrelation and HB. VII. The chemical composi- tion of first and second-generation stars in 15 globular clusters from GIRAFFE spectra”. In: Astronomy & Astrophysics 505, pp. 117–138. DOI: 10.1051/0004-6361/200912096. arXiv: 0909.2938 [astro-ph.GA]. Carretta, E. et al. (Oct. 2009b). “Na-O anticorrelation and HB. VIII. Proton-capture ele- ments and metallicities in 17 globular clusters from UVES spectra”. In: Astronomy & Astrophysics 505.1, pp. 139–155. DOI: 10.1051/0004-6361/200912097. arXiv: 0909.2941 [astro-ph.GA]. Bibliography 155

Carretta, E. et al. (Oct. 2010a). “Abundances for a Large Sample of Red Giants in NGC 1851: Hints for a Merger of Two Clusters?” In: The Astrophysical Journal Letters 722.1, pp. L1–L6. DOI: 10.1088/2041-8205/722/1/L1. arXiv: 1007.5301 [astro-ph.GA]. Carretta, E. et al. (June 2010b). “Properties of stellar generations in globular clusters and relations with global parameters”. In: Astronomy & Astrophysics 516, A55, A55. DOI: 10.1051/0004-6361/200913451. arXiv: 1003.1723 [astro-ph.GA]. Carretta, E. et al. (Sept. 2011). “Multiple stellar populations in the globular cluster NGC 1851”. In: Astronomy & Astrophysics 533, A69, A69. DOI: 10.1051/0004-6361/201117269. arXiv: 1106.3174 [astro-ph.SR]. Carretta, E. et al. (May 2013). “Potassium in Globular Cluster Stars: Comparing Normal Clusters to the Peculiar Cluster NGC 2419”. In: The Astrophysical Journal 769.1, 40, p. 40. DOI: 10.1088/0004-637X/769/1/40. arXiv: 1303.4740 [astro-ph.GA]. Casali, G. et al. (July 2020). “The Gaia-ESO survey: the non-universality of the age-chemical- clocks-metallicity relations in the Galactic disc”. In: Astronomy & Astrophysics 639, A127, A127. DOI: 10.1051/0004-6361/202038055. arXiv: 2006.05763 [astro-ph.GA]. Casey, Andrew R. et al. (Dec. 2019a). “A Data-driven Model of Nucleosynthesis with Chemical Tagging in a Lower-dimensional Latent Space”. In: The Astrophysical Journal 887.1, 73, p. 73. DOI: 10.3847/1538-4357/ab4fea. arXiv: 1910.09811 [astro-ph.SR]. Casey, Andrew R. et al. (Aug. 2019b). “Tidal Interactions between Binary Stars Can Drive Lithium Production in Low-mass Red Giants”. In: The Astrophysical Journal 880.2, 125, p. 125. DOI: 10.3847/1538-4357/ab27bf. arXiv: 1902.04102 [astro-ph.SR]. Caswell, Thomas A et al. (May 2019). matplotlib/matplotlib v3.1.0. Version v3.1.0. DOI: 10. 5281/zenodo.2893252. Charbonnel, C. and S. C. Balachandran (July 2000). “The Nature of the lithium rich giants. Mixing episodes on the RGB and early-AGB”. In: Astronomy & Astrophysic 359, pp. 563– 572. arXiv: astro-ph/0005280 [astro-ph]. Charbonnel, C. et al. (Jan. 2020). “Lithium in red giant stars: Constraining non-standard mixing with large surveys in the Gaia era”. In: Astronomy & Astrophysics 633, A34, A34. DOI: 10.1051/0004-6361/201936360. arXiv: 1910.12732 [astro-ph.SR]. Chiappini, C., F. Matteucci, and R. Gratton (Mar. 1997). “The Chemical Evolution of the Galaxy: The Two-Infall Model”. In: The Astrophysical Journal 477.2, pp. 765–780. DOI: 10.1086/303726. arXiv: astro-ph/9609199 [astro-ph]. Choi, Jieun et al. (June 2016). “Mesa Isochrones and Stellar Tracks (MIST). I. Solar-scaled Models”. In: The Astrophysical Journal 823.2, 102, p. 102. DOI: 10.3847/0004-637X/823/ 2/102. arXiv: 1604.08592 [astro-ph.SR]. Clayton, Donald D. (1983). Principles of Stellar Evolution and Nucleosynthesis. University of Chicago Press, 519–524. ISBN: 9780226109534. Bibliography 156

Cockcroft, Robert et al. (Apr. 2011). “The M33 Globular Cluster System with PAndAS Data: the Last Outer Halo Cluster?” In: The Astrophysical Journal 730.2, 112, p. 112. DOI: 10.1088/0004-637X/730/2/112. arXiv: 1101.2450 [astro-ph.CO]. Cohen, J. G. (July 1978). “Abundances in globular cluster red giants. I. M3 and M13.” In: The Astrophysical Journal 223, pp. 487–508. DOI: 10.1086/156284. Cohen, Judith G., Wenjin Huang, and Evan N. Kirby (Oct. 2011). “The Peculiar Chemical Inventory of NGC 2419: An Extreme Outer Halo “Globular Cluster””. In: The Astro- physical Journal 740.2, 60, p. 60. DOI: 10.1088/0004-637X/740/2/60. arXiv: 1107.2445 [astro-ph.GA]. Cole, Shaun et al. (Nov. 2000). “Hierarchical galaxy formation”. In: Monthly Notices of the Royal Astronomical Society 319.1, pp. 168–204. DOI: 10.1046/j.1365-8711.2000.03879. x. arXiv: astro-ph/0007281 [astro-ph]. Conroy, Charlie et al. (Sept. 2019). “Mapping the Stellar Halo with the H3 Spectroscopic Survey”. In: The Astrophysical Journal 883.1, 107, p. 107. DOI: 10 . 3847 / 1538 - 4357 / ab38b8. arXiv: 1907.07684 [astro-ph.GA]. Cooper, A. P. et al. (Aug. 2010). “Galactic stellar haloes in the CDM model”. In: Monthly Notices of the Royal Astronomical Society 406.2, pp. 744–766. DOI: 10 . 1111 / j . 1365 - 2966.2010.16740.x. arXiv: 0910.3211 [astro-ph.GA]. Cordero, M. J. et al. (Jan. 2014). “Detailed Abundances for a Large Sample of Giant Stars in the Globular Cluster 47 Tucanae (NGC 104)”. In: The Astrophysical Journal 780.1, 94, p. 94. DOI: 10.1088/0004-637X/780/1/94. arXiv: 1311.1541 [astro-ph.SR]. Cordero, M. J. et al. (July 2015). “A Comparative Study of Two 47 Tuc Giant Stars with Different s-process Enrichment”. In: The Astrophysical Journal Letters 808.1, L10, p. L10. DOI: 10.1088/2041-8205/808/1/L10. arXiv: 1506.07874 [astro-ph.SR]. Cottrell, P.L. and G. S. Da Costa (Apr. 1981). “Correlated cyanogen and sodium anomalies in the globular clusters 47 TUC and NGC 6752.” In: The Astrophysical Journal Letters 245, pp. L79–L82. DOI: 10.1086/183527. Cowan, John J. and Christopher Sneden (Apr. 2006). “Heavy element synthesis in the oldest stars and the early Universe”. In: Nature 440.7088, pp. 1151–1156. DOI: 10.1038/ nature04807. Cowley, William I. et al. (Feb. 2018). “Predictions for deep galaxy surveys with JWST from Lambda CDM”. In: Monthly Notices of the Royal Astronomical Society 474.2, pp. 2352– 2372. DOI: 10.1093/mnras/stx2897. arXiv: 1702.02146 [astro-ph.GA]. Da Costa, G. S. (Aug. 2016). “Are the globular clusters with significant internal [Fe/H] spreads all former dwarf galaxy nuclei?” In: The General Assembly of Galaxy Halos: Struc- ture, Origin and Evolution. Ed. by A. Bragaglia et al. Vol. 317. IAU Symposium, pp. 110– 115. DOI: 10.1017/S174392131500678X. arXiv: 1510.00873 [astro-ph.GA]. Bibliography 157

Da Costa, G. S. and T. E. Armandroff (June 1995). “Abundances and Kinematics of the Globular Cluster Systems of the Galaxy and of the Sagittarius Dwarf”. In: Astronomical Journal 109, p. 2533. DOI: 10.1086/117469. Da Costa, G. S., E. V. Held, and I. Saviane (Mar. 2014). “NGC 5824: a luminous outer halo globular cluster with an intrinsic abundance spread”. In: Monthly Notices of the Royal Astronomical Society 438.4, pp. 3507–3520. DOI: 10.1093/mnras/stt2467. arXiv: 1312.5796 [astro-ph.SR]. da Silva, R. et al. (June 2012). “Accurate and homogeneous abundance patterns in solar- type stars of the solar neighbourhood: a chemo-chronological analysis”. In: Astronomy & Astrophysics 542, A84, A84. DOI: 10.1051/0004-6361/201118751. arXiv: 1204.4433 [astro-ph.SR]. Dalcanton, Julianne J., David N. Spergel, and F. J. Summers (June 1997). “The Formation of Disk Galaxies”. In: The Astrophysical Journal 482.2, pp. 659–676. DOI: 10.1086/304182. arXiv: astro-ph/9611226 [astro-ph]. Dalton, Gavin et al. (July 2014). “Project overview and update on WEAVE: the next gener- ation wide-field spectroscopy facility for the William Herschel Telescope”. In: Ground- based and Airborne Instrumentation for Astronomy V. Vol. 9147. Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, p. 91470L. DOI: 10 . 1117 / 12 . 2055132. arXiv: 1412.0843 [astro-ph.IM]. D’Antona, Francesca et al. (Nov. 2012). “Models for the lithium abundances of multi- ple populations in globular clusters and the possible role of the big bang lithium”. In: Monthly Notices of the Royal Astronomical Society 426.3, pp. 1710–1719. DOI: 10.1111/j. 1365-2966.2012.21663.x. arXiv: 1207.1544 [astro-ph.GA]. Dark Energy Survey Collaboration et al. (Aug. 2016). “The Dark Energy Survey: more than dark energy - an overview”. In: Monthly Notices of the Royal Astronomical Society 460.2, pp. 1270–1299. DOI: 10.1093/mnras/stw641. arXiv: 1601.00329 [astro-ph.CO]. Das, Payel, Keith Hawkins, and Paula Jofré (Apr. 2020). “Ages and kinematics of chem- ically selected, accreted Milky Way halo stars”. In: Monthly Notices of the Royal Astro- nomical Society 493.4, pp. 5195–5207. DOI: 10.1093/mnras/stz3537. arXiv: 1903.09320 [astro-ph.GA]. de Boer, T. J. L., V. Belokurov, and S. Koposov (Aug. 2015). “The star formation history of the Sagittarius stream”. In: Monthly Notices of the Royal Astronomical Society 451.4, pp. 3489–3503. DOI: 10.1093/mnras/stv946. arXiv: 1505.00787 [astro-ph.GA]. de Jong, Roelof S. et al. (July 2014). “4MOST: 4-metre Multi-Object Spectroscopic Tele- scope”. In: Ground-based and Airborne Instrumentation for Astronomy V. Vol. 9147. Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, p. 91470M. DOI: 10.1117/12.2055826. de La Reza, R. (Jan. 2000). “Lithium in Giant Stars (Invited Paper)”. In: The Light Elements and their Evolution. Ed. by L. da Silva, R. de Medeiros, and M. Spite. Vol. 198, p. 310. Bibliography 158 de Mink, S. E. et al. (Nov. 2009). “Massive binaries as the source of abundance anomalies in globular clusters”. In: Astronomy & Astrophysics 507.1, pp. L1–L4. DOI: 10 . 1051 / 0004-6361/200913205. arXiv: 0910.1086 [astro-ph.SR]. De Silva, G. M. et al. (Mar. 2007). “Chemical Homogeneity in Collinder 261 and Implica- tions for Chemical Tagging”. In: Astronomical Journal 133, pp. 1161–1175. DOI: 10.1086/ 511182. eprint: astro-ph/0611832. de Silva, G. M. et al. (June 2009). “O and Na abundance patterns in open clusters of the Galactic disk”. In: Astronomy & Astrophysics 500.2, pp. L25–L28. DOI: 10.1051/0004- 6361/200912279. arXiv: 0905.4354 [astro-ph.GA]. De Silva, G. M. et al. (May 2015). “The GALAH survey: scientific motivation”. In: Monthly Notices of the Royal Astronomical Society 449.3, pp. 2604–2617. DOI: 10 . 1093 / mnras / stv327. arXiv: 1502.04767 [astro-ph.GA]. Deason, A. J., V. Belokurov, and N. W. Evans (Oct. 2011). “The Milky Way stellar halo out to 40 kpc: squashed, broken but smooth”. In: Monthly Notices of the Royal Astronomical Society 416.4, pp. 2903–2915. DOI: 10.1111/j.1365-2966.2011.19237.x. arXiv: 1104. 3220 [astro-ph.GA]. Decressin, T. et al. (Mar. 2007). “Fast rotating massive stars and the origin of the abun- dance patterns in galactic globular clusters”. In: Astronomy & Astrophysics 464.3, pp. 1029– 1044. DOI: 10.1051/0004-6361:20066013. arXiv: astro-ph/0611379 [astro-ph]. Deepak, David L. Lambert, and Bacham E. Reddy (Mar. 2020). “Abundance analyses of Li-enriched and normal giants in the GALAH survey”. In: Monthly Notices of the Royal Astronomical Society 494.1, pp. 1348–1365. DOI: 10.1093/mnras/staa729. arXiv: 2003. 05381 [astro-ph.SR]. Delgado Mena, E. et al. (Apr. 2019). “Abundance to age ratios in the HARPS-GTO sample with Gaia DR2. Chemical clocks for a range of [Fe/H]”. In: Astronomy & Astrophysics 624, A78, A78. DOI: 10.1051/0004-6361/201834783. arXiv: 1902.02127 [astro-ph.SR]. denBrok, Mark et al. (Dec. 2014). “The HST/ACS Coma Cluster Survey - X. Nuclear star clusters in low-mass early-type galaxies: scaling relations”. In: Monthly Notices of the Royal Astronomical Society 445.3, pp. 2385–2403. DOI: 10.1093/mnras/stu1906. arXiv: 1409.4766 [astro-ph.GA]. Denisenkov, P. A. and S. N. Denisenkova (July 1990). “Correlation Between the Abun- dances of Na and the CNO Elements in Red Giants in Omega-Centauri”. In: Soviet Astronomy Letters 16, p. 275. Denissenkov, P. A. and F. D. A. Hartwick (Jan. 2014). “Supermassive stars as a source of abundance anomalies of proton-capture elements in globular clusters”. In: Monthly Notices of the Royal Astronomical Society 437.1, pp. L21–L25. DOI: 10 . 1093 / mnrasl / slt133. arXiv: 1305.5975 [astro-ph.SR]. D’Ercole, Annibale et al. (Dec. 2008). “Formation and dynamical evolution of multiple stellar generations in globular clusters”. In: Monthly Notices of the Royal Astronomical Bibliography 159

Society 391.2, pp. 825–843. DOI: 10.1111/j.1365-2966.2008.13915.x. arXiv: 0809. 1438 [astro-ph]. Diemand, J. et al. (Aug. 2008). “Clumps and streams in the local dark matter distribu- tion”. In: Nature 454.7205, pp. 735–738. DOI: 10.1038/nature07153. arXiv: 0805.1244 [astro-ph]. Dobrovolskas, V. et al. (May 2014). “Abundances of lithium, oxygen, and sodium in the turn-off stars of Galactic globular cluster 47 Tucanae”. In: Astronomy & Astrophysics 565, A121, A121. DOI: 10.1051/0004-6361/201322868. arXiv: 1311.1072 [astro-ph.SR]. Doménech-Moral, M. et al. (Apr. 2012). “Formation of galaxies in Λcold dark matter cos- mologies - I. The fine structure of disc galaxies”. In: Monthly Notices of the Royal Astro- nomical Society 421.3, pp. 2510–2530. DOI: 10.1111/j.1365-2966.2012.20534.x. arXiv: 1201.2641 [astro-ph.CO]. D’Orazi, V. et al. (Oct. 2011). “Chemical enrichment mechanisms in ω Centauri: clues from neutron-capture elements”. In: Astronomy & Astrophysics 534, A29, A29. DOI: 10.1051/ 0004-6361/201117630. arXiv: 1108.5216 [astro-ph.GA]. D’Orazi, V. et al. (Jan. 2013a). “A different view on light-element anticorrelations in globu- lar clusters: fluorine abundances”. In: Memorie della Societa Astronomica Italiana 84, p. 34. D’Orazi, V. et al. (July 2013b). “On the internal pollution mechanisms in the globular cluster NGC 6121 (M4): heavy-element abundances and AGB models”. In: Monthly Notices of the Royal Astronomical Society 433.1, pp. 366–381. DOI: 10.1093/mnras/stt728. arXiv: 1304.7009 [astro-ph.GA]. D’Orazi, Valentina et al. (Aug. 2010a). “Ba Stars and Other Binaries in First and Sec- ond Generation Stars in Globular Clusters”. In: The Astrophysical Journal Letters 719.2, pp. L213–L217. DOI: 10.1088/2041-8205/719/2/L213. arXiv: 1007.2164 [astro-ph.SR]. D’Orazi, Valentina et al. (Apr. 2010b). “Lithium and Proton-capture Elements in Globular Cluster Dwarfs: The Case of 47 TUC”. In: The Astrophysical Journal Letters 713.1, pp. L1– L5. DOI: 10.1088/2041-8205/713/1/L1. arXiv: 1003.0013 [astro-ph.SR]. Dotter, Aaron et al. (Sept. 2018). “Ruprecht 106: A Riddle, Wrapped in a Mystery, inside an Enigma”. In: The Astrophysical Journal Letters 865.1, L10, p. L10. DOI: 10.3847/2041- 8213/aae08f. arXiv: 1808.05582 [astro-ph.SR]. Dupree, A. K. and E. H. Avrett (Aug. 2013). “Direct Evaluation of the Helium Abundances in Omega Centauri”. In: The Astrophysical Journal Letters 773.2, L28, p. L28. DOI: 10. 1088/2041-8205/773/2/L28. arXiv: 1307.5860 [astro-ph.SR]. Dwek, Eli (July 1998). “The Evolution of the Elemental Abundances in the Gas and Dust Phases of the Galaxy”. In: Astrophysical Journal 501, p. 643. DOI: 10.1086/305829. arXiv: astro-ph/9707024 [astro-ph]. Eggen, O. J., D. Lynden-Bell, and A. R. Sandage (Nov. 1962). “Evidence from the motions of old stars that the Galaxy collapsed.” In: The Astrophysical Journal 136, p. 748. DOI: 10.1086/147433. Bibliography 160

El-Badry, Kareem et al. (Oct. 2018). “Where are the most ancient stars in the Milky Way?” In: Monthly Notices of the Royal Astronomical Society 480.1, pp. 652–668. DOI: 10.1093/ mnras/sty1864. arXiv: 1804.00659 [astro-ph.GA]. Eldridge, J. J. et al. (Dec. 2018). “Supernova lightCURVE POPulation Synthesis I: Includ- ing interacting binaries is key to understanding the diversity of type II supernova lightcurves”. In: Publications of the Astronomical Society of Australia 35, p. 49. DOI: 10. 1017/pasa.2018.47. arXiv: 1811.00282 [astro-ph.SR]. Elmegreen, Bruce G. (Dec. 2018). “Two Thresholds for Globular Cluster Formation and the Common Occurrence of Massive Clusters in the Early Universe”. In: The Astrophys- ical Journal 869.2, 119, p. 119. DOI: 10.3847/1538- 4357/aaed45. arXiv: 1810.12940 [astro-ph.GA]. Fardal, Mark A. et al. (Jan. 2014). “Modeling and Fitting Tidal Stellar Streams”. In: Amer- ican Astronomical Society Meeting Abstracts #223. Vol. 223. American Astronomical Soci- ety Meeting Abstracts, p. 309.06. Feltzing, Sofia et al. (Feb. 2017). “On the metallicity dependence of the [Y/Mg]-age re- lation for solar-type stars”. In: Monthly Notices of the Royal Astronomical Society 465.1, pp. L109–L113. DOI: 10.1093/mnrasl/slw209. arXiv: 1610.03852 [astro-ph.GA]. Feng, Yi and Mark R. Krumholz (Sept. 2014). “Early turbulent mixing as the origin of chemical homogeneity in open star clusters”. In: Nature 513.7519, pp. 523–525. DOI: 10.1038/nature13662. arXiv: 1408.6543 [astro-ph.GA]. Feuillet, Diane K. et al. (Oct. 2019). “Spatial variations in the Milky Way disc metallicity- age relation”. In: Monthly Notices of the Royal Astronomical Society 489.2, pp. 1742–1752. DOI: 10.1093/mnras/stz2221. arXiv: 1908.02772 [astro-ph.GA]. Finlator, Kristian and Romeel Davé (Apr. 2008). “The origin of the galaxy mass-metallicity relation and implications for galactic outflows”. In: Monthly Notices of the Royal Astro- nomical Society 385.4, pp. 2181–2204. DOI: 10.1111/j.1365-2966.2008.12991.x. arXiv: 0704.3100 [astro-ph]. Font, A. S. et al. (Oct. 2011). “Cosmological simulations of the formation of the stel- lar haloes around disc galaxies”. In: Monthly Notices of the Royal Astronomical Society 416.4, pp. 2802–2820. DOI: 10.1111/j.1365-2966.2011.19227.x. arXiv: 1102.2526 [astro-ph.CO]. Font, Andreea S. et al. (Feb. 2006). “Chemical Abundance Distributions of Galactic Halos and Their Satellite Systems in a Lambda CDM Universe”. In: The Astrophysical Journal 638.2, pp. 585–595. DOI: 10.1086/498970. arXiv: astro-ph/0507114 [astro-ph]. Font, Andreea S. et al. (Jan. 2008). “The Stellar Content of Galaxy Halos: A Comparison between Lambda CDM Models and Observations of M31”. In: The Astrophysical Journal 673, pp. 215–225. DOI: 10.1086/524102. arXiv: 0709.2076 [astro-ph]. Bibliography 161

Forbes, Duncan A. and Terry Bridges (May 2010). “Accreted versus in situ Milky Way globular clusters”. In: Monthly Notices of the Royal Astronomical Society 404, pp. 1203– 1214. DOI: 10.1111/j.1365-2966.2010.16373.x. arXiv: 1001.4289 [astro-ph.GA]. Foreman-Mackey, Daniel et al. (Mar. 2013). “emcee: The MCMC Hammer”. In: Publications of the Astronomical Society of the Pacific 125.925, p. 306. DOI: 10.1086/670067. arXiv: 1202.3665 [astro-ph.IM]. Francois, P. (July 1991). “Chemical composition of globular cluster stars.” In: Astronomy & Astrophysics 247, p. 56. Frankel, Neige et al. (Oct. 2019). “The Inside-out Growth of the Galactic Disk”. In: The Astrophysical Journal 884.2, 99, p. 99. DOI: 10.3847/1538- 4357/ab4254. arXiv: 1909. 07118 [astro-ph.GA]. Frebel, Anna (Oct. 2018). “From Nuclei to the Cosmos: Tracing Heavy-Element Produc- tion with the Oldest Stars”. In: Annual Review of Nuclear and Particle Science 68.1, pp. 237– 269. DOI: 10.1146/annurev-nucl-101917-021141. arXiv: 1806.08955 [astro-ph.SR]. Frebel, Anna and John E. Norris (2013). “Metal-Poor Stars and the Chemical Enrichment of the Universe”. In: Planets, Stars and Stellar Systems. Volume 5: Galactic Structure and Stellar Populations. Ed. by Terry D. Oswalt and Gerard Gilmore. Vol. 5, p. 55. DOI: 10. 1007/978-94-007-5612-0_3. – (Aug. 2015). “Near-Field Cosmology with Extremely Metal-Poor Stars”. In: Annual Re- views of Astronomy & Astrophysics 53, pp. 631–688. DOI: 10 . 1146 / annurev - astro - 082214-122423. arXiv: 1501.06921 [astro-ph.SR]. Freeman, K. and J. Bland-Hawthorn (Sept. 2002). “The New Galaxy: Signatures of Its Formation”. In: Annual Reviews of Astronomy & Astrophysics 40, pp. 487–537. Freeman, K. C. (Jan. 1993). “Globular Clusters and Nucleated Dwarf Ellipticals”. In: The Globular Cluster-Galaxy Connection. Ed. by Graeme H. Smith and Jean P. Brodie. Vol. 48. Astronomical Society of the Pacific Conference Series, p. 608. Freeman, K. C. and A. W. Rodgers (Oct. 1975). “The Chemical Inhomogeneity of Omega Centauri”. In: The Astrophysical Journal Letters 201, p. L71. DOI: 10.1086/181945. Fritz, T. K. et al. (Nov. 2018). “Gaia DR2 proper motions of dwarf galaxies within 420 kpc. Orbits, Milky Way mass, tidal influences, planar alignments, and group infall”. In: Astronomy & Astrophysics 619, A103, A103. DOI: 10.1051/0004-6361/201833343. arXiv: 1805.00908 [astro-ph.GA]. Fulbright, Jon. P. and Jennifer A. Johnson (Oct. 2003). “Oxygen Abundances in Metal- poor Stars”. In: The Astrophysical Journal 595.2, pp. 1154–1181. DOI: 10.1086/377443. arXiv: astro-ph/0307063 [astro-ph]. Fulbright, Jon. P., Andrew McWilliam, and R. Michael Rich (June 2007). “Abundances of Baade’s Window Giants from Keck HIRES Spectra. II. The Alpha and Light Odd Elements”. In: The Astrophysical Journal 661.2, pp. 1152–1179. DOI: 10 . 1086 / 513710. arXiv: astro-ph/0609087 [astro-ph]. Bibliography 162

Gai, Moshe (Aug. 2014). “Nuclear structure and the fate of core collapse (Type II) super- nova”. In: Nuclear Physics, Section A 928, pp. 313–318. DOI: 10.1016/j.nuclphysa. 2014.05.004. arXiv: 1405.7119 [nucl-ex]. Gaia Collaboration et al. (Aug. 2018a). “Gaia Data Release 2. Kinematics of globular clus- ters and dwarf galaxies around the Milky Way”. In: Astronomy & Astrophysics 616, A12, A12. DOI: 10.1051/0004-6361/201832698. arXiv: 1804.09381 [astro-ph.GA]. Gaia Collaboration et al. (Aug. 2018b). “Gaia Data Release 2. Summary of the contents and survey properties”. In: Astronomy & Astrophysics 616, A1, A1. DOI: 10.1051/0004- 6361/201833051. arXiv: 1804.09365 [astro-ph.GA]. Gallazzi, Anna et al. (Sept. 2005). “The ages and metallicities of galaxies in the local universe”. In: Monthly Notices of the Royal Astronomical Society 362.1, pp. 41–58. DOI: 10.1111/j.1365-2966.2005.09321.x. arXiv: astro-ph/0506539 [astro-ph]. Gao, Xudong et al. (June 2020). “The GALAH survey: a new constraint on cosmological lithium and Galactic lithium evolution from warm dwarf stars”. In: Monthly Notices of the Royal Astronomical Society 497.1, pp. L30–L34. DOI: 10.1093/mnrasl/slaa109. arXiv: 2006.05173 [astro-ph.SR]. Gieles, Mark et al. (Aug. 2018). “Concurrent formation of supermassive stars and globular clusters: implications for early self-enrichment”. In: Monthly Notices of the Royal Astro- nomical Society 478.2, pp. 2461–2479. DOI: 10.1093/mnras/sty1059. arXiv: 1804.04682 [astro-ph.GA]. Gilbert, Karoline (Sept. 2017). “Constraints on the Formation of M31’s Stellar Halo from the SPLASH Survey”. In: Galaxies 5.4, p. 59. DOI: 10.3390/galaxies5040059. Gilbert, Karoline M. et al. (Jan. 2018). “Global Properties of M31’s Stellar Halo from the SPLASH Survey. III. Measuring the Stellar Velocity Dispersion Profile”. In: The Astro- physical Journal 852.2, 128, p. 128. DOI: 10.3847/1538-4357/aa9f26. arXiv: 1711.02700 [astro-ph.GA]. Gilmore, G. et al. (Mar. 2012). “The Gaia-ESO Public Spectroscopic Survey”. In: ESO Mes- senger 147, pp. 25–31. Gilmore, Gerard, Rosemary F. G. Wyse, and Konrad Kuijken (Jan. 1989). “Kinematics, chemistry, and structure of the Galaxy.” In: Annual Reviews of Astronomy & Astrophysics 27, pp. 555–627. DOI: 10.1146/annurev.aa.27.090189.003011. Gnedin, Oleg Y. and Jeremiah P.Ostriker (Jan. 1997). “Destruction of the Galactic Globular Cluster System”. In: The Astrophysical Journal 474.1, pp. 223–255. DOI: 10.1086/303441. arXiv: astro-ph/9603042 [astro-ph]. González Hernández, J. I. et al. (Oct. 2009). “Lithium in the globular cluster NGC 6397. Evidence for dependence on evolutionary status”. In: Astronomy & Astrophysics 505.2, pp. L13–L16. DOI: 10.1051/0004-6361/200912713. arXiv: 0909.0983 [astro-ph.GA]. Goodwin, Simon P. and B. E. J. Pagel (May 2005). “Supernova II enrichment and the star cluster mass function”. In: Monthly Notices of the Royal Astronomical Society 359.2, Bibliography 163

pp. 707–710. DOI: 10.1111/j.1365-2966.2005.08944.x. arXiv: astro-ph/0410068 [astro-ph]. Gotta, Vanessa et al. (Dec. 2018). “Terzan 5, a fossil relic of the Galactic Bulge”. In: The Galactic Bulge at the Crossroads, p. 15. DOI: 10.5281/zenodo.2595334. Grand, Robert J. J. et al. (July 2020). “The dual origin of the Galactic thick disc and halo from the gas-rich Gaia-Enceladus Sausage merger”. In: Monthly Notices of the Royal As- tronomical Society 497.2, pp. 1603–1618. DOI: 10.1093/mnras/staa2057. arXiv: 2001. 06009 [astro-ph.GA]. Gratton, R. G. and C. Sneden (Oct. 1988). “Abundances in extremely metal-poor stars.” In: Astronomy & Astrophysics 204, pp. 193–218. Gratton, R. G. et al. (Oct. 2011). “Multiple populations in ω Centauri: a cluster analysis of spectroscopic data”. In: Astronomy & Astrophysics 534, A72, A72. DOI: 10.1051/0004- 6361/201117093. arXiv: 1105.5544 [astro-ph.GA]. Gratton, R. G. et al. (Nov. 2012). “The Na-O anticorrelation in horizontal branch stars (Corrigendum). II. NGC 1851”. In: Astronomy & Astrophysics 547, C2, p. C2. DOI: 10. 1051/0004-6361/201118491e. Gratton, Raffaele (Jan. 2020). “Spectroscopic studies of stellar populations in globular clusters and field stars: Implications for globular cluster and Milky Way halo forma- tion”. In: IAU Symposium. Ed. by Angela Bragaglia et al. Vol. 351. IAU Symposium, pp. 241–250. DOI: 10.1017/S1743921319007877. arXiv: 1908.06905 [astro-ph.SR]. Gratton, Raffaele, Christopher Sneden, and Eugenio Carretta (Sept. 2004). “Abundance Variations Within Globular Clusters”. In: Annual Reviews of Astronomy & Astrophysics 42.1, pp. 385–440. DOI: 10.1146/annurev.astro.42.053102.133945. Gratton, Raffaele G., Eugenio Carretta, and Angela Bragaglia (Feb. 2012). “Multiple pop- ulations in globular clusters. Lessons learned from the Milky Way globular clusters”. In: Astronomy & Astrophysics Reviews 20, 50, p. 50. DOI: 10.1007/s00159-012-0050-3. arXiv: 1201.6526 [astro-ph.SR]. Grebel, Eva K. (Sept. 2012). “Metal-poor galaxies in the local universe”. In: American In- stitute of Physics Conference Series. Ed. by Masayuki Umemura and Kazuyuki Omukai. Vol. 1480. American Institute of Physics Conference Series, pp. 172–183. DOI: 10.1063/ 1.4754351. arXiv: 1208.2894 [astro-ph.CO]. Greenstein, Jesse L. and Robert S. Richardson (May 1951). “Lithium and the Internal Cir- culation of the Sun.” In: Astrophysical Journal 113, p. 536. DOI: 10.1086/145424. Grillmair, C. J. and R. Johnson (Mar. 2006). “The Detection of a 45 Degree Tidal Stream Associated with the Globular Cluster NGC 5466”. In: The Astrophysical Journal Letters 639.1, pp. L17–L20. DOI: 10.1086/501439. arXiv: astro-ph/0602602 [astro-ph]. Grillmair, Carl J. (Jan. 2017). “Tails from the Orphanage”. In: The Astrophysical Journal 834.2, 98, p. 98. DOI: 10.3847/1538-4357/834/2/98. arXiv: 1612.03181 [astro-ph.GA]. Bibliography 164

Gruyters, Pieter et al. (May 2016). “Atomic diffusion and mixing in old stars. VI. The lithium content of M30”. In: Astronomy & Astrophysics 589, A61, A61. DOI: 10.1051/ 0004-6361/201527948. arXiv: 1603.01565 [astro-ph.SR]. Gunn, J. E. et al. (Apr. 2006). “The 2.5 m Telescope of the Sloan Digital Sky Survey”. In: Astronomical Journal 131, pp. 2332–2359. DOI: 10.1086/500975. eprint: astro-ph/ 0602326. Gustafsson, B. et al. (Aug. 2008). “A grid of MARCS model atmospheres for late-type stars. I. Methods and general properties”. In: Astronomy & Astrophysics 486.3, pp. 951– 970. DOI: 10.1051/0004-6361:200809724. arXiv: 0805.0554 [astro-ph]. Hampel, Melanie et al. (Nov. 2016). “The Intermediate Neutron-capture Process and Carbon- enhanced Metal-poor Stars”. In: The Astrophysical Journal 831.2, 171, p. 171. DOI: 10. 3847/0004-637X/831/2/171. arXiv: 1608.08634 [astro-ph.SR]. Harris, W. E. (Oct. 1996). “A Catalog of Parameters for Globular Clusters in the Milky Way”. In: Astronomical Journal 112, p. 1487. DOI: 10.1086/118116. Harris, W. E. and R. Canterna (July 1979). “On the abundance gradient in the galactic halo.” In: Astrophysical Journal Letters 231, pp. L19–L23. DOI: 10.1086/182997. Harris, William E. (Sept. 1974). “The Unusual Horizontal Branch of NGC 2808”. In: The Astrophysical Journal Letters 192, p. L161. DOI: 10.1086/181616. – (Dec. 2010). “A New Catalog of Globular Clusters in the Milky Way”. In: arXiv e-prints, arXiv:1012.3224, arXiv:1012.3224. arXiv: 1012.3224 [astro-ph.GA]. Hasselquist, Sten et al. (Feb. 2019). “Identifying Sagittarius Stream Stars by Their APOGEE Chemical Abundance Signatures”. In: The Astrophysical Journal 872.1, 58, p. 58. DOI: 10.3847/1538-4357/aafdac. arXiv: 1901.04559 [astro-ph.GA]. Haywood, M. (Jan. 2013). “The structure and formation of the Milky Way disks.” In: Memorie della Societa Astronomica Italiana Supplementi 25, p. 11. Haywood, Misha et al. (Dec. 2013). “The age structure of stellar populations in the solar vicinity. Clues of a two-phase formation history of the Milky Way disk”. In: Astronomy & Astrophysics 560, A109, A109. DOI: 10.1051/0004-6361/201321397. arXiv: 1305.4663 [astro-ph.GA]. Heger, Alexander and S. E. Woosley (Nov. 2010). “Nucleosynthesis and Evolution of Mas- sive Metal-free Stars”. In: The Astrophysical Journal 724.1, pp. 341–373. DOI: 10.1088/ 0004-637X/724/1/341. arXiv: 0803.3161 [astro-ph]. Helmi, Amina (June 2008). “The stellar halo of the Galaxy”. In: Astronomy & Astrophysics Reviews 15.3, pp. 145–188. DOI: 10 . 1007 / s00159 - 008 - 0009 - 6. arXiv: 0804 . 0019 [astro-ph]. – (Feb. 2020). “Streams, substructures and the early history of the Milky Way”. In: arXiv e-prints, arXiv:2002.04340, arXiv:2002.04340. arXiv: 2002.04340 [astro-ph.GA]. Bibliography 165

Helmi, Amina and Simon D. M. White (Aug. 1999). “Building up the stellar halo of the Galaxy”. In: Monthly Notices of the Royal Astronomical Society 307.3, pp. 495–517. DOI: 10.1046/j.1365-8711.1999.02616.x. arXiv: astro-ph/9901102 [astro-ph]. Helmi, Amina et al. (Nov. 1999). “Debris streams in the solar neighbourhood as relics from the formation of the Milky Way”. In: Nature 402.6757, pp. 53–55. DOI: 10.1038/46980. arXiv: astro-ph/9911041 [astro-ph]. Helmi, Amina et al. (Feb. 2006). “Pieces of the puzzle: ancient substructure in the Galactic disc”. In: Monthly Notices of the Royal Astronomical Society 365, pp. 1309–1323. DOI: 10. 1111/j.1365-2966.2005.09818.x. arXiv: astro-ph/0505401 [astro-ph]. Helmi, Amina et al. (Oct. 2018). “The merger that led to the formation of the Milky Way’s inner stellar halo and thick disk”. In: Nature 563.7729, pp. 85–88. DOI: 10.1038/s41586- 018-0625-x. arXiv: 1806.06038 [astro-ph.GA]. Hilker, M. and T. Richtler (Oct. 2000). “ω Centauri - a former nucleus of a dissolved dwarf galaxy? New evidence from Strömgren photometry”. In: Astronomy & Astrophysics 362, pp. 895–909. arXiv: astro-ph/0008500 [astro-ph]. Hillebrandt, W. et al. (Apr. 2013). “Towards an understanding of Type Ia supernovae from a synthesis of theory and observations”. In: Frontiers of Physics 8.2, pp. 116–143. DOI: 10.1007/s11467-013-0303-2. arXiv: 1302.6420 [astro-ph.CO]. Hillebrandt, Wolfgang and Jens C. Niemeyer (Jan. 2000). “Type IA Supernova Explosion Models”. In: Annual Reviews of Astronomy & Astrophysics 38, pp. 191–230. DOI: 10.1146/ annurev.astro.38.1.191. arXiv: astro-ph/0006305 [astro-ph]. Ho, Anna Y. Q. et al. (Feb. 2017). “Label Transfer from APOGEE to LAMOST: Precise Stellar Parameters for 450,000 LAMOST Giants”. In: The Astrophysical Journal 836.1, 5, p. 5. DOI: 10.3847/1538-4357/836/1/5. arXiv: 1602.00303 [astro-ph.SR]. Holtzman, J. A. et al. (Nov. 2015). “Abundances, Stellar Parameters, and Spectra from the SDSS-III/APOGEE Survey”. In: Astronomical Journal 150, 148, p. 148. DOI: 10.1088/ 0004-6256/150/5/148. arXiv: 1501.04110. Ibata, Rodrigo et al. (Feb. 2001). “Galactic Halo Substructure in the Sloan Digital Sky Survey: The Ancient Tidal Stream from the Sagittarius Dwarf Galaxy”. In: Astrophysical Journal Letters 547.2, pp. L133–L136. DOI: 10.1086/318894. arXiv: astro-ph/0004255 [astro-ph]. Ibata, Rodrigo A., Khyati Malhan, and Nicolas F. Martin (Feb. 2019). “The Streams of the Gaping Abyss: A Population of Entangled Stellar Streams Surrounding the Inner Galaxy”. In: The Astrophysical Journal 872.2, 152, p. 152. DOI: 10 . 3847 / 1538 - 4357 / ab0080. arXiv: 1901.07566 [astro-ph.GA]. Ikuta, C. and N. Arimoto (June 2000). “Self-enrichment in omega Centauri”. In: Astronomy & Astrophysics 358, pp. 535–546. arXiv: astro-ph/9912526 [astro-ph]. Bibliography 166

Ishigaki, Miho, Masashi Chiba, and Wako Aoki (Feb. 2010). “Chemical Abundances of Outer Halo Stars in the Milky Way”. In: Publications of the Astronomical Society of Japan 62, p. 143. DOI: 10.1093/pasj/62.1.143. arXiv: 0912.0329 [astro-ph.GA]. Ivans, Inese I. et al. (Aug. 2003). “Chemical Substructure in the Milky Way Halo: A New Population of Old Stars”. In: Astrophysical Journal 592.2, pp. 906–934. DOI: 10.1086/ 375812. James, G. et al. (Dec. 2004). “Heavy elements and chemical enrichment in globular clus- ters”. In: Astronomy & Astrophysics 427, pp. 825–838. DOI: 10.1051/0004-6361:20041512. arXiv: astro-ph/0408330 [astro-ph]. Ji, Alexander P., Anna Frebel, and Volker Bromm (Nov. 2015). “Preserving chemical sig- natures of primordial star formation in the first low-mass stars”. In: Monthly Notices of the Royal Astronomical Society 454.1, pp. 659–674. DOI: 10.1093/mnras/stv2052. arXiv: 1508.06137 [astro-ph.GA]. Jofré, Paula et al. (May 2017). “Cosmic phylogeny: reconstructing the chemical history of the solar neighbourhood with an evolutionary tree”. In: Monthly Notices of the Royal Astronomical Society 467.1, pp. 1140–1153. DOI: 10.1093/mnras/stx075. arXiv: 1611. 02575 [astro-ph.GA]. Johnson, Christian I. and Catherine A. Pilachowski (Oct. 2010). “Chemical Abundances for 855 Giants in the Globular Cluster Omega Centauri (NGC 5139)”. In: The Astro- physical Journal 722.2, pp. 1373–1410. DOI: 10.1088/0004- 637X/722/2/1373. arXiv: 1008.2232 [astro-ph.SR]. Johnson, Christian I. et al. (July 2008). “Fe and Al Abundances for 180 Red Giants in the Globular Cluster Omega Centauri (NGC 5139)”. In: The Astrophysical Journal 681.2, pp. 1505–1523. DOI: 10.1086/588634. arXiv: 0804.2607 [astro-ph]. Johnson, Christian I. et al. (Oct. 2017). “Light and Heavy Element Abundance Variations in the Outer Halo Globular Cluster NGC 6229”. In: Astronomical Journal 154.4, 155, p. 155. DOI: 10.3847/1538-3881/aa86ac. arXiv: 1709.00048 [astro-ph.SR]. Johnson, Christian I. et al. (June 2020). “The Most Metal-poor Stars in Omega Centauri (NGC 5139)”. In: Astronomical Journal 159.6, 254, p. 254. DOI: 10.3847/1538- 3881/ ab8819. arXiv: 2004.09023 [astro-ph.SR]. Juri´c,Mario et al. (Feb. 2008). “The Milky Way Tomography with SDSS. I. Stellar Number Density Distribution”. In: The Astrophysical Journal 673.2, pp. 864–914. DOI: 10.1086/ 523619. arXiv: astro-ph/0510520 [astro-ph]. Kacharov, N., A. Koch, and A. McWilliam (June 2013). “A comprehensive chemical abun- dance study of the outer halo globular cluster M 75”. In: Astronomy & Astrophysics 554, A81, A81. DOI: 10.1051/0004-6361/201321392. arXiv: 1304.4247 [astro-ph.GA]. Kalirai, Jason S. et al. (Nov. 2009). “The SPLASH Survey: A Spectroscopic Analysis of the Metal-Poor, Low-Luminosity M31 dSph Satellite Andromeda X”. In: The Astrophysical Bibliography 167

Journal 705.1, pp. 1043–1055. DOI: 10.1088/0004-637X/705/1/1043. arXiv: 0909.3298 [astro-ph.GA]. Karakas, A. I. (Jan. 2016). “The role of asymptotic giant branch stars in galactic chemical evolution”. In: Memorie della Societa Astronomica Italiana 87, p. 229. Karakas, A. I. and J. C. Lattanzio (Jan. 2003). “Production of Aluminium and the Heavy Magnesium Isotopes in Asymptotic Giant Branch Stars”. In: Publications of the Astro- nomical Society of Australia 20.3, pp. 279–293. DOI: 10.1071/AS03010. Karakas, Amanda I. and John C. Lattanzio (July 2014). “The Dawes Review 2: Nucleosyn- thesis and Stellar Yields of Low- and Intermediate-Mass Single Stars”. In: Publications of the Astronomical Society of Australia 31, e030, e030. DOI: 10.1017/pasa.2014.21. arXiv: 1405.0062 [astro-ph.SR]. Karim, M. T. and Eric E. Mamajek (Feb. 2017). “Revised geometric estimates of the North Galactic Pole and the Sun’s height above the Galactic mid-plane”. In: Monthly Notices of the Royal Astronomical Society 465.1, pp. 472–481. DOI: 10.1093/mnras/stw2772. arXiv: 1610.08125 [astro-ph.SR]. Karlsson, Torgny et al. (Nov. 2012). “The Chemical Signature of a Relic Star Cluster in the Sextans Dwarf Spheroidal Galaxy—Implications for Near-field Cosmology”. In: The Astrophysical Journal 759.2, 111, p. 111. DOI: 10.1088/0004- 637X/759/2/111. arXiv: 1201.5376 [astro-ph.CO]. Kepley, Amanda A. et al. (Oct. 2007). “Halo Star Streams in the Solar Neighborhood”. In: Astronomical Journal 134.4, pp. 1579–1595. DOI: 10.1086/521429. arXiv: 0707.4477 [astro-ph]. Khanna, Shourya et al. (Nov. 2019). “The GALAH survey and Gaia DR2: Linking ridges, arches, and vertical waves in the kinematics of the Milky Way”. In: Monthly Notices of the Royal Astronomical Society 489.4, pp. 4962–4979. DOI: 10.1093/mnras/stz2462. arXiv: 1902.10113 [astro-ph.GA]. Kharchenko, N. V. et al. (Oct. 2013). “Global survey of star clusters in the Milky Way. II. The catalogue of basic parameters”. In: Astronomy & Astrophysics 558, A53, A53. DOI: 10.1051/0004-6361/201322302. arXiv: 1308.5822. Kirby, Evan N. et al. (Dec. 2013). “The Universal Stellar Mass-Stellar Metallicity Relation for Dwarf Galaxies”. In: The Astrophysical Journal 779.2, 102, p. 102. DOI: 10.1088/0004- 637X/779/2/102. arXiv: 1310.0814 [astro-ph.GA]. Kobayashi, Chiaki (Dec. 2016). “Astrophysics: Elemental abundances across cosmic time”. In: Nature 540.7632, pp. 205–206. DOI: 10.1038/540205a. Kobayashi, Chiaki, Amanda I. Karakas, and Maria Lugaro (Aug. 2020). “The Origin of El- ements from Carbon to Uranium”. In: arXiv e-prints, arXiv:2008.04660, arXiv:2008.04660. arXiv: 2008.04660 [astro-ph.GA]. Kobayashi, Chiaki and Ken’ichi Nomoto (Dec. 2009). “The Role of Type Ia Supernovae in Chemical Evolution. I. Lifetime of Type Ia Supernovae and Metallicity Effect”. In: Bibliography 168

Astrophysical Journal 707.2, pp. 1466–1484. DOI: 10 . 1088 / 0004 - 637X / 707 / 2 / 1466. arXiv: 0801.0215 [astro-ph]. Kobayashi, Chiaki et al. (Dec. 2006). “Galactic Chemical Evolution: Carbon through Zinc”. In: The Astrophysical Journal 653.2, pp. 1145–1171. DOI: 10.1086/508914. arXiv: astro- ph/0608688 [astro-ph]. Koch, A. et al. (Jan. 2012). “A super-Li rich turnoff star in NGC 6397 - the puzzle persists.” In: Memorie della Societa Astronomica Italiana Supplementi 22, p. 79. arXiv: 1203 . 6079 [astro-ph.GA]. Koch, Andreas, Eva K. Grebel, and Sarah L. Martell (May 2019). “Purveyors of fine halos: Re-assessing globular cluster contributions to the Milky Way halo buildup with SDSS- IV”. In: Astronomy & Astrophysics 625, A75, A75. DOI: 10.1051/0004-6361/201834825. arXiv: 1904.02146 [astro-ph.GA]. Koch, Andreas and Andrew McWilliam (Apr. 2008). “A New Abundance Scale for the Globular Cluster 47 Tuc”. In: Astronomical Journal 135.4, pp. 1551–1566. DOI: 10.1088/ 0004-6256/135/4/1551. arXiv: 0802.2103 [astro-ph]. Komatsu, E. et al. (Feb. 2009). “Five-Year Wilkinson Microwave Anisotropy Probe Obser- vations: Cosmological Interpretation”. In: The Astrophysical Journal Supplement 180.2, pp. 330–376. DOI: 10.1088/0067-0049/180/2/330. arXiv: 0803.0547 [astro-ph]. Koposov, S. E. et al. (Jan. 2015). “Exposing Sgr tidal debris behind the Galactic disc with M giants selected in WISE and 2MASS”. In: Monthly Notices of the Royal Astronomi- cal Society 446.3, pp. 3110–3117. DOI: 10 . 1093 / mnras / stu2263. arXiv: 1410 . 6482 [astro-ph.GA]. Koppelman, Helmer H. et al. (May 2019). “Characterization and history of the Helmi streams with Gaia DR2”. In: Astronomy & Astrophysics 625, A5, A5. DOI: 10.1051/0004- 6361/201834769. arXiv: 1812.00846 [astro-ph.GA]. Kordopatis, G. et al. (Nov. 2013). “The Radial Velocity Experiment (RAVE): Fourth Data Release”. In: Astronomical Journal 146, 134, p. 134. DOI: 10.1088/0004-6256/146/5/134. arXiv: 1309.4284. Kos, Janez et al. (Feb. 2018). “The GALAH survey: chemical tagging of star clusters and new members in the Pleiades”. In: Monthly Notices of the Royal Astronomical Society 473.4, pp. 4612–4633. DOI: 10.1093/mnras/stx2637. arXiv: 1709.00794 [astro-ph.SR]. Kraft, R. P., C. F. Trefzger, and N. Suntzeff (Jan. 1979). “Is There a Composition Gradient in the Halo?” In: The Large-Scale Characteristics of the Galaxy. Ed. by W. B. Burton. Vol. 84. IAU Symposium, p. 463. Kraft, Robert P. (June 1994). “Abundance Differences among Globular Cluster Giants: Primordial vs. Evolutionary Scenarios”. In: Publications of the Astronomical Society of the Pacific 106, p. 553. DOI: 10.1086/133416. Bibliography 169

Kravtsov, Andrey V. and Oleg Y. Gnedin (Apr. 2005). “Formation of Globular Clusters in Hierarchical Cosmology”. In: The Astrophysical Journal 623.2, pp. 650–665. DOI: 10. 1086/428636. arXiv: astro-ph/0305199 [astro-ph]. Kruijssen, J. M. Diederik (Dec. 2015). “Globular clusters as the relics of regular star for- mation in ‘normal’ high-redshift galaxies”. In: Monthly Notices of the Royal Astronom- ical Society 454.2, pp. 1658–1686. DOI: 10 . 1093 / mnras / stv2026. arXiv: 1509 . 02163 [astro-ph.GA]. Kruijssen, J. M. Diederik et al. (July 2019). “The formation and assembly history of the Milky Way revealed by its globular cluster population”. In: Monthly Notices of the Royal Astronomical Society 486.3, pp. 3180–3202. DOI: 10.1093/mnras/sty1609. arXiv: 1806. 05680 [astro-ph.GA]. Kushniruk, I., T. Schirmer, and T. Bensby (Dec. 2017). “Kinematic structures of the so- lar neighbourhood revealed by Gaia DR1/TGAS and RAVE”. In: Astronomy & As- trophysics 608, A73, A73. DOI: 10 . 1051 / 0004 - 6361 / 201731147. arXiv: 1709 . 06905 [astro-ph.GA]. Kuzma, P. B. et al. (Feb. 2015). “Palomar 5 and its tidal tails: a search for new members in the tidal stream”. In: Monthly Notices of the Royal Astronomical Society 446.4, pp. 3297– 3309. DOI: 10.1093/mnras/stu2343. arXiv: 1411.0776 [astro-ph.GA]. Lacey, Cedric and Shaun Cole (June 1993). “Merger rates in hierarchical models of galaxy formation”. In: Monthly Notices of the Royal Astronomical Society 262.3, pp. 627–649. DOI: 10.1093/mnras/262.3.627. Lada, Charles J. and Elizabeth A. Lada (Jan. 2003). “Embedded Clusters in Molecular Clouds”. In: Annual Reviews of Astronomy & Astrophysics 41, pp. 57–115. DOI: 10.1146/ annurev.astro.41.011802.094844. arXiv: astro-ph/0301540 [astro-ph]. Lahén, Natalia et al. (July 2019). “The Formation of Low-metallicity Globular Clusters in Dwarf Galaxy Mergers”. In: The Astrophysical Journal Letters 879.2, L18, p. L18. DOI: 10.3847/2041-8213/ab2a13. arXiv: 1905.09840 [astro-ph.GA]. Langer, G. E., R. Hoffman, and C. Sneden (Mar. 1993). “Sodium-Oxygen Abundance An- ticorrelations and Deep Mixing Scenarios for Globular Cluster Giants”. In: Publications of the Astronomical Society of the Pacific 105, p. 301. DOI: 10.1086/133147. Lardo, C. et al. (Aug. 2013). “The double red giant branch in M2: C, N, Sr and Ba abun- dances”. In: Monthly Notices of the Royal Astronomical Society 433.3, pp. 1941–1950. DOI: 10.1093/mnras/stt854. arXiv: 1305.3181 [astro-ph.SR]. Law, David R. and Steven R. Majewski (May 2010). “The Sagittarius Dwarf Galaxy: A Model for Evolution in a Triaxial Milky Way Halo”. In: The Astrophysical Journal 714.1, pp. 229–254. DOI: 10.1088/0004-637X/714/1/229. arXiv: 1003.1132 [astro-ph.GA]. Leaman, R., D. A. VandenBerg, and J. T. Mendel (Nov. 2013). “The bifurcated age-metallicity relation of Milky Way globular clusters and its implications for the accretion history of Bibliography 170

the galaxy”. In: Monthly Notices of the Royal Astronomical Society 436, pp. 122–135. DOI: 10.1093/mnras/stt1540. arXiv: 1309.0822. Lee, Young-Wook (Jan. 1992). “RR Lyrae stars as standard candles.” In: Memorie della So- cieta Astronomica Italiana 63, pp. 331–343. Lequeux, J. et al. (Dec. 1979). “Chemical composition and evolution of irregular and blue compact galaxies.” In: Astronomy & Astrophysics 500, pp. 145–156. Levi, Michael et al. (Sept. 2019). “The Dark Energy Spectroscopic Instrument (DESI)”. In: Bulletin of the American Astronomical Society. Vol. 51, p. 57. arXiv: 1907.10688 [astro-ph.IM]. Lewis, Geraint F. et al. (Jan. 2013). “PAndAS in the Mist: The Stellar and Gaseous Mass within the Halos of M31 and M33”. In: The Astrophysical Journal 763.1, 4, p. 4. DOI: 10.1088/0004-637X/763/1/4. arXiv: 1211.4059 [astro-ph.CO]. Li, T. S. et al. (Dec. 2019). “The southern stellar stream spectroscopic survey (S5): Overview, target selection, data reduction, validation, and early science”. In: Monthly Notices of the Royal Astronomical Society 490.3, pp. 3508–3531. DOI: 10.1093/mnras/stz2731. arXiv: 1907.09481 [astro-ph.GA]. Lin, Chien-Cheng, E. Magnier, and W. Chen (May 2011). “Search For And Characteriza- tion Of Galactic Star Clusters With 2MASS And Pan-STARRS 1”. In: American Astro- nomical Society Meeting Abstracts #218. Vol. 218. American Astronomical Society Meet- ing Abstracts, p. 407.12. Lin, Jane et al. (July 2018). “Stellar ages and masses in the solar neighbourhood: Bayesian analysis using spectroscopy and Gaia DR1 parallaxes”. In: Monthly Notices of the Royal Astronomical Society 477.3, pp. 2966–2975. DOI: 10.1093/mnras/sty709. arXiv: 1803. 10875 [astro-ph.SR]. Lin, Jane et al. (Jan. 2020). “The GALAH survey: temporal chemical enrichment of the galactic disc”. In: Monthly Notices of the Royal Astronomical Society 491.2, pp. 2043–2056. DOI: 10.1093/mnras/stz3048. arXiv: 1911.05221 [astro-ph.SR]. Lind, K. et al. (Aug. 2009). “Signatures of intrinsic Li depletion and Li-Na anti-correlation in the metal-poor globular cluster NGC 6397”. In: Astronomy & Astrophysics 503.2, pp. 545– 557. DOI: 10.1051/0004-6361/200912524. arXiv: 0906.2876 [astro-ph.SR]. Lind, K. et al. (Mar. 2015). “The Gaia-ESO Survey: A globular cluster escapee in the Galac- tic halo”. In: Astronomy & Astrophysics 575, L12, p. L12. DOI: 10 . 1051 / 0004 - 6361 / 201425554. arXiv: 1502.03934 [astro-ph.SR]. Liu, F. et al. (Nov. 2016). “The chemical compositions of solar twins in the open cluster M67”. In: Monthly Notices of the Royal Astronomical Society 463.1, pp. 696–704. DOI: 10. 1093/mnras/stw2045. arXiv: 1608.03788 [astro-ph.SR]. Lotz, J. M., B. W. Miller, and H. C. Ferguson (Sept. 2004). “The Colors of Dwarf Ellipti- cal Galaxy Globular Cluster Systems, Nuclei, and Stellar Halos”. In: The Astrophysical Journal 613, pp. 262–278. DOI: 10.1086/422871. eprint: astro-ph/0406002. Bibliography 171

Luri, X. et al. (Aug. 2018). “Gaia Data Release 2. Using Gaia parallaxes”. In: Astronomy & Astrophysics 616, A9, A9. DOI: 10.1051/0004-6361/201832964. arXiv: 1804.09376 [astro-ph.IM]. Maaten, Laurens van der and Geoffrey Hinton (Nov. 2008). “Viualizing data using t- SNE”. In: Journal of Machine Learning Research 9, pp. 2579–2605. Maccarone, Thomas J. and David R. Zurek (June 2012). “Novae from isolated white dwarfs as a source of helium for second-generation stars in globular clusters”. In: Monthly No- tices of the Royal Astronomical Society 423.1, pp. 2–6. DOI: 10.1111/j.1365-2966.2011. 20328.x. arXiv: 1112.0571 [astro-ph.SR]. Mackereth, J. Ted et al. (Aug. 2018). “The age-metallicity structure of the Milky Way disc with APOGEE”. In: Rediscovering Our Galaxy. Ed. by Cristina Chiappini et al. Vol. 334, pp. 265–268. DOI: 10.1017/S1743921317006627. arXiv: 1708.05399 [astro-ph.GA]. MacLean, B. T. et al. (Mar. 2018). “AGB subpopulations in the nearby globular cluster NGC 6397”. In: Monthly Notices of the Royal Astronomical Society 475.1, pp. 257–265. DOI: 10.1093/mnras/stx3217. arXiv: 1712.03340 [astro-ph.SR]. Majewski, S. R. et al. (Sept. 2017). “The Apache Point Observatory Galactic Evolution Experiment (APOGEE)”. In: Astronomical Journal 154, 94, p. 94. DOI: 10.3847/1538- 3881/aa784d. arXiv: 1509.05420 [astro-ph.IM]. Majewski, Steven R. et al. (Dec. 2003). “A Two Micron All Sky Survey View of the Sagittar- ius Dwarf Galaxy. I. Morphology of the Sagittarius Core and Tidal Arms”. In: The Astro- physical Journal 599.2, pp. 1082–1115. DOI: 10.1086/379504. arXiv: astro-ph/0304198 [astro-ph]. Majewski, Steven R. et al. (July 2004). “A Two Micron All Sky Survey View of the Sagit- tarius Dwarf Galaxy. II. Swope Telescope Spectroscopy of M Giant Stars in the Dynam- ically Cold Sagittarius Tidal Stream”. In: The Astronomical Journal 128.1, pp. 245–259. DOI: 10.1086/421372. arXiv: astro-ph/0403701 [astro-ph]. Malhan, Khyati, Rodrigo A. Ibata, and Nicolas F. Martin (Dec. 2018). “Ghostly tributaries to the Milky Way: charting the halo’s stellar streams with the Gaia DR2 catalogue”. In: Monthly Notices of the Royal Astronomical Society 481.3, pp. 3442–3455. DOI: 10.1093/ mnras/sty2474. arXiv: 1804.11339 [astro-ph.GA]. Manning, Ellen M. and Andrew A. Cole (Nov. 2017). “From K giants to G dwarfs: stellar lifetime effects on metallicity distributions derived from red giants”. In: Monthly Notices of the Royal Astronomical Society 471.4, pp. 4194–4205. DOI: 10.1093/mnras/stx1854. arXiv: 1708.01826 [astro-ph.GA]. Marcolini, A. et al. (Nov. 2007). “Modelling the chemical evolution of ω Centauri using three-dimensional hydrodynamical simulations”. In: Monthly Notices of the Royal Astro- nomical Society 382.1, pp. 443–454. DOI: 10.1111/j.1365-2966.2007.12386.x. arXiv: 0708.3445 [astro-ph]. Bibliography 172

Marín-Franch, Antonio et al. (Apr. 2009). “The ACS Survey of Galactic Globular Clusters. VII. Relative Ages”. In: The Astrophysical Journal 694.2, pp. 1498–1516. DOI: 10.1088/ 0004-637X/694/2/1498. arXiv: 0812.4541 [astro-ph]. Marino, A. F. (Jan. 2013). “Anomalous globular clusters: insights from neutron capture elements abundances.” In: Memorie della Societa Astronomica Italiana 84, p. 29. arXiv: 1302.2306 [astro-ph.SR]. Marino, A. F. et al. (Apr. 2010). “Chemical composition of stellar populations in ω Cen- tauri”. In: Light Elements in the Universe. Ed. by Corinne Charbonnel et al. Vol. 268. IAU Symposium, pp. 183–184. DOI: 10.1017/S1743921310004096. arXiv: 1001.1500 [astro-ph.SR]. Marino, A. F. et al. (Feb. 2012). “The C+N+O Abundance of ω Centauri Giant Stars: Impli- cations for the Chemical-enrichment Scenario and the Relative Ages of Different Stellar Populations”. In: Astrophysical Journal 746.1, 14, p. 14. DOI: 10.1088/0004-637X/746/ 1/14. arXiv: 1111.1891 [astro-ph.SR]. Marino, A. F. et al. (Aug. 2014). “The halo+cluster system of the Galactic globular cluster NGC 1851”. In: Monthly Notices of the Royal Astronomical Society 442.4, pp. 3044–3064. DOI: 10.1093/mnras/stu1099. arXiv: 1406.0944 [astro-ph.SR]. Marino, A. F. et al. (Aug. 2019). “The Hubble Space Telescope UV Legacy Survey of Galac- tic Globular Clusters - XIX. A chemical tagging of the multiple stellar populations over the chromosome maps”. In: Monthly Notices of the Royal Astronomical Society 487.3, pp. 3815–3844. DOI: 10.1093/mnras/stz1415. arXiv: 1904.05180 [astro-ph.SR]. Marino, Anna F. (Mar. 2017). “Chemical abundances of multiple stellar populations in massive globular clusters”. In: Formation, Evolution, and Survival of Massive Star Clusters. Ed. by Corinne Charbonnel and Antonella Nota. Vol. 316. IAU Symposium, pp. 267– 274. DOI: 10.1017/S1743921315009485. Marsakov, V. A., V. V. Koval’, and M. L. Gozha (Apr. 2019). “The Chemical Composition of Globular Clusters of Different Nature in Our Galaxy”. In: Astronomy Reports 63.4, pp. 274–288. DOI: 10.1134/S1063772919040048. arXiv: 1904.06256 [astro-ph.GA]. Martell, S. L. and E. K. Grebel (Sept. 2010). “Light-element abundance variations in the Milky Way halo”. In: Astronomy & Astrophysics 519, A14, A14. DOI: 10.1051/0004- 6361/201014135. arXiv: 1005.4070 [astro-ph.GA]. Martell, S. L. et al. (Oct. 2011). “Building the Galactic halo from globular clusters: evidence from chemically unusual red giants”. In: Astronomy & Astrophysics 534, A136, A136. DOI: 10.1051/0004-6361/201117644. arXiv: 1109.3916 [astro-ph.GA]. Martell, S. L. et al. (Mar. 2017). “The GALAH survey: observational overview and Gaia DR1 companion”. In: Monthly Notices of the Royal Astronomical Society 465.3, pp. 3203– 3219. DOI: 10.1093/mnras/stw2835. arXiv: 1609.02822 [astro-ph.IM]. Bibliography 173

Martell, Sarah et al. (June 2020). “The GALAH survey: Lithium-rich giant stars require multiple formation channels”. In: arXiv e-prints, arXiv:2006.02106, arXiv:2006.02106. arXiv: 2006.02106 [astro-ph.SR]. Martell, Sarah L. et al. (July 2016). “Chemical Tagging in the SDSS-III/APOGEE Survey: New Identifications of Halo Stars with Globular Cluster Origins”. In: The Astrophysical Journal 825.2, 146, p. 146. DOI: 10.3847/0004- 637X/825/2/146. arXiv: 1605.05792 [astro-ph.GA]. Martig, Marie et al. (Mar. 2016). “Red giant masses and ages derived from carbon and ni- trogen abundances”. In: Monthly Notices of the Royal Astronomical Society 456.4, pp. 3655– 3670. DOI: 10.1093/mnras/stv2830. arXiv: 1511.08203 [astro-ph.SR]. Martin, Nicolas F. et al. (Nov. 2009). “PAndAS’ CUBS: Discovery of Two New Dwarf Galaxies in the Surroundings of the Andromeda and Triangulum Galaxies”. In: The Astrophysical Journal 705.1, pp. 758–765. DOI: 10.1088/0004-637X/705/1/758. arXiv: 0909.0399 [astro-ph.CO]. Massari, D., H. H. Koppelman, and A. Helmi (Oct. 2019). “Origin of the system of glob- ular clusters in the Milky Way”. In: Astronomy & Astrophysics 630, L4, p. L4. DOI: 10. 1051/0004-6361/201936135. arXiv: 1906.08271 [astro-ph.GA]. Masseron, T. et al. (Feb. 2019). “Homogeneous analysis of globular clusters from the APOGEE survey with the BACCHUS code. I. The northern clusters”. In: Astronomy & Astrophysics 622, A191, A191. DOI: 10.1051/0004-6361/201834550. arXiv: 1812.08817 [astro-ph.SR]. Matteucci, Francesca (Jan. 1992). “The evolution of the alpha-elements in galaxies”. In: Memorie della Societa Astronomica Italiana 63.2, pp. 301–314. – (2012). Chemical Evolution of Galaxies. 1st ed. Berlin: Springer-Verlag Berlin Heidelberg. ISBN: 9783642224911. Maxwell, Aaron J. et al. (Aug. 2012). “Building the Stellar Halo through Feedback in Dwarf Galaxies”. In: The Astrophysical Journal Letters 755.2, L35, p. L35. DOI: 10.1088/ 2041-8205/755/2/L35. arXiv: 1207.2421 [astro-ph.GA]. McConnachie, Alan et al. (May 2016). “The Detailed Science Case for the Maunakea Spec- troscopic Explorer: the Composition and Dynamics of the Faint Universe”. In: arXiv e-prints, arXiv:1606.00043, arXiv:1606.00043. arXiv: 1606.00043 [astro-ph.IM]. McWilliam, Andrew (Jan. 1997). “Abundance Ratios and Galactic Chemical Evolution”. In: Annual Reviews of Astronomy & Astrophysics 35, pp. 503–556. DOI: 10.1146/annurev. astro.35.1.503. Meingast, Stefan, João Alves, and Verena Fürnkranz (Feb. 2019). “Extended stellar sys- tems in the solar neighborhood . II. Discovery of a nearby 120 Degree stellar stream in Gaia DR2”. In: Astronomy & Astrophysics 622, L13, p. L13. DOI: 10 . 1051 / 0004 - 6361/201834950. arXiv: 1901.06387 [astro-ph.GA]. Bibliography 174

Mendel, Jon T., Robert N. Proctor, and Duncan A. Forbes (Aug. 2007). “The age, metallic- ity and α-element abundance of Galactic globular clusters from single stellar popula- tion models”. In: Monthly Notices of the Royal Astronomical Society 379.4, pp. 1618–1636. DOI: 10.1111/j.1365-2966.2007.12041.x. arXiv: 0705.4511 [astro-ph]. Mészáros, Szabolcs et al. (May 2015). “Exploring Anticorrelations and Light Element Vari- ations in Northern Globular Clusters Observed by the APOGEE Survey”. In: Astronom- ical Journal 149.5, 153, p. 153. DOI: 10.1088/0004-6256/149/5/153. arXiv: 1501.05127 [astro-ph.SR]. Mészáros, Szabolcs et al. (Feb. 2020). “Homogeneous analysis of globular clusters from the APOGEE survey with the BACCHUS code - II. The Southern clusters and overview”. In: Monthly Notices of the Royal Astronomical Society 492.2, pp. 1641–1670. DOI: 10.1093/ mnras/stz3496. arXiv: 1912.04839 [astro-ph.SR]. Michaud, G. et al. (May 2004). “Models for Solar Abundance Stars with Gravitational Settling and Radiative Accelerations: Application to M67 and NGC 188”. In: The As- trophysical Journal 606.1, pp. 452–465. DOI: 10.1086/383001. arXiv: astro-ph/0402544 [astro-ph]. Michaud, Georges, Georges Alecian, and Jacques Richer (2015). Atomic Diffusion in Stars. DOI: 10.1007/978-3-319-19854-5. Milone, A. P. et al. (Feb. 2010). “Multiple Stellar Populations in the Galactic Globular Cluster NGC 6752”. In: The Astrophysical Journal 709.2, pp. 1183–1194. DOI: 10.1088/ 0004-637X/709/2/1183. arXiv: 0912.2055 [astro-ph.SR]. Milone, A. P. et al. (July 2015). “The Hubble Space Telescope UV Legacy Survey of Galactic Globular Clusters. III. A Quintuple Stellar Population in NGC 2808”. In: The Astrophys- ical Journal 808.1, 51, p. 51. DOI: 10.1088/0004- 637X/808/1/51. arXiv: 1505.05934 [astro-ph.SR]. Milone, A. P. et al. (Jan. 2017). “The Hubble Space Telescope UV Legacy Survey of Galactic globular clusters - IX. The Atlas of multiple stellar populations”. In: Monthly Notices of the Royal Astronomical Society 464.3, pp. 3636–3656. DOI: 10.1093/mnras/stw2531. arXiv: 1610.00451 [astro-ph.SR]. Mitschang, A. W. et al. (Jan. 2013). “Quantifying chemical tagging: towards robust group finding in the Galaxy”. In: Monthly Notices of the Royal Astronomical Society 428, pp. 2321– 2332. DOI: 10.1093/mnras/sts194. arXiv: 1210.3407. Monaco, L. (Jan. 2012). “The lithium content of the globular clusters omega Centauri and M 4”. In: Memorie della Societa Astronomica Italiana Supplementi 22, p. 72. Monaco, L. et al. (Sept. 2010). “The lithium content of ω Centauri. New clues to the cos- mological Li problem from old stars in external galaxies”. In: Astronomy & Astrophysics 519, L3, p. L3. DOI: 10.1051/0004-6361/201015162. arXiv: 1008.1817 [astro-ph.GA]. Bibliography 175

Montalbán, Josefina et al. (June 2020). “Chronologically dating the early assembly of the Milky Way”. In: arXiv e-prints, arXiv:2006.01783, arXiv:2006.01783. arXiv: 2006.01783 [astro-ph.GA]. Morrison, Heather L. et al. (Mar. 2009). “Fashionably Late? Building Up The Milky Way’S Inner Halo”. In: The Astrophysical Journal 694.1, pp. 130–143. DOI: 10.1088/0004-637X/ 694/1/130. arXiv: 0804.2448 [astro-ph]. Mowlavi, N. and G. Meynet (Sept. 2000). “Aluminum 26 production in asymptotic giant branch stars”. In: Astronomy & Astrophysics 361, pp. 959–976. Mucciarelli, A., M. Salaris, and P. Bonifacio (Jan. 2012). “The primordial Li abundance derived from giant stars”. In: Memorie della Societa Astronomica Italiana Supplementi 22, p. 86. arXiv: 1205.2389 [astro-ph.SR]. Mucciarelli, A. et al. (Oct. 2018). “Lithium abundance in lower red giant branch stars of Omega Centauri”. In: Astronomy & Astrophysics 618, A134, A134. DOI: 10.1051/0004- 6361/201833457. arXiv: 1807.07582 [astro-ph.SR]. Myeong, G. C. et al. (Apr. 2018). “The Milky Way Halo in Action Space”. In: The Astro- physical Journal Letters 856.2, L26, p. L26. DOI: 10 . 3847 / 2041 - 8213 / aab613. arXiv: 1802.03351 [astro-ph.GA]. Myeong, G. C. et al. (Sept. 2019). “Evidence for two early accretion events that built the Milky Way stellar halo”. In: Monthly Notices of the Royal Astronomical Society 488.1, pp. 1235–1247. DOI: 10.1093/mnras/stz1770. arXiv: 1904.03185 [astro-ph.GA]. Naidu, Rohan P. et al. (June 2020). “Evidence from the H3 Survey that the Stellar Halo is Entirely Comprised of Substructure”. In: arXiv e-prints, arXiv:2006.08625, arXiv:2006.08625. arXiv: 2006.08625 [astro-ph.GA]. Navin, Colin A., Sarah L. Martell, and Daniel B. Zucker (Oct. 2016). “New Halo Stars of the Galactic Globular Clusters M3 and M13 in the LAMOST DR1 Catalog”. In: The Astrophysical Journal 829.2, 123, p. 123. DOI: 10.3847/0004- 637X/829/2/123. arXiv: 1606.06430 [astro-ph.GA]. Necib, Lina et al. (July 2020). “Evidence for a vast prograde stellar stream in the solar vicinity”. In: Nature Astronomy. DOI: 10.1038/s41550-020-1131-2. arXiv: 1907.07190 [astro-ph.GA]. Ness, M. et al. (Feb. 2018). “Galactic Doppelgängers: The Chemical Similarity Among Field Stars and Among Stars with a Common Birth Origin”. In: The Astrophysical Journal 853.2, 198, p. 198. DOI: 10.3847/1538-4357/aa9d8e. arXiv: 1701.07829 [astro-ph.SR]. Ness, M. K. et al. (Oct. 2019). “In the Galactic Disk, Stellar [Fe/H] and Age Predict Orbits and Precise [X/Fe]”. In: The Astrophysical Journal 883.2, 177, p. 177. DOI: 10.3847/1538- 4357/ab3e3c. arXiv: 1907.10606 [astro-ph.SR]. Newberg, Heidi Jo et al. (Jan. 2020). “Streams and the Milky Way dark matter halo”. In: IAU Symposium. Ed. by Monica Valluri and J. A. Sellwood. Vol. 353. IAU Symposium, pp. 75–82. DOI: 10.1017/S174392131900855X. Bibliography 176

Nidever, David L. et al. (Apr. 2019). “Exploring the Very Extended Low-surface-brightness Stellar Populations of the Large Magellanic Cloud with SMASH”. In: The Astrophys- ical Journal 874.2, 118, p. 118. DOI: 10.3847/1538- 4357/aafaf7. arXiv: 1805.02671 [astro-ph.GA]. Nissen, P. E. (Oct. 2004a). Origin and Evolution of the Elements: Volume 4, Carnegie Observa- tories Astrophysics Series. Ed. by Andrew McWilliam and Michael Rauch. Vol. 4. Cam- bridge University Press. ISBN: 0521755786. – (July 2015). “High-precision abundances of elements in solar twin stars. Trends with stellar age and elemental condensation temperature”. In: Astronomy & Astrophysics 579, A52, A52. DOI: 10.1051/0004-6361/201526269. arXiv: 1504.07598 [astro-ph.SR]. Nissen, P. E. and W. J. Schuster (Feb. 2010). “Two distinct halo populations in the solar neighborhood. Evidence from stellar abundance ratios and kinematics”. In: Astronomy & Astrophysics 511, L10, p. L10. DOI: 10.1051/0004-6361/200913877. arXiv: 1002.4514 [astro-ph.GA]. – (June 2011). “Two distinct halo populations in the solar neighborhood. II. Evidence from stellar abundances of Mn, Cu, Zn, Y, and Ba”. In: Astronomy & Astrophysics 530, A15, A15. DOI: 10.1051/0004-6361/201116619. arXiv: 1103.4755 [astro-ph.GA]. Nissen, P.E. et al. (Dec. 2017). “High-precision abundances of elements in Kepler LEGACY stars. Verification of trends with stellar age”. In: Astronomy & Astrophysics 608, A112, A112. DOI: 10.1051/0004-6361/201731845. arXiv: 1710.03544 [astro-ph.SR]. Nissen, Poul E. (Jan. 2004b). “Thin and Thick Galactic Disks”. In: Origin and Evolution of the Elements. Ed. by Andrew McWilliam and Michael Rauch, p. 154. arXiv: astro- ph/0310326 [astro-ph]. Nordlander, T. et al. (Sept. 2019). “The lowest detected stellar Fe abundance: the halo star SMSS J160540.18-144323.1”. In: Monthly Notices of the Royal Astronomical Society 488.1, pp. L109–L113. DOI: 10.1093/mnrasl/slz109. arXiv: 1904.07471 [astro-ph.SR]. Norris, J. et al. (Feb. 1981). “The abundance spread in the giants of NGC 6752.” In: The Astrophysical Journal 244, pp. 205–220. DOI: 10.1086/158698. Norris, John (Feb. 1987). “The Ellipticities of Globular Clusters and the Cyanogen Prob- lem”. In: The Astrophysical Journal Letters 313, p. L65. DOI: 10.1086/184832. Norris, John E. (Apr. 1996). “The Relationship Between the Halo and Disk Populations”. In: Formation of the Galactic Halo...Inside and Out. Ed. by Heather L. Morrison and Ata Sarajedini. Vol. 92. Astronomical Society of the Pacific Conference Series, p. 14. Norris, John E. and G. S. Da Costa (July 1995). “The Giant Branch of omega Centauri. IV. Abundance Patterns Based on Echelle Spectra of 40 Red Giants”. In: The Astrophysical Journal 447, p. 680. DOI: 10.1086/175909. Odenkirchen, Michael et al. (Feb. 2001). “Detection of Massive Tidal Tails around the Globular Cluster Palomar 5 with Sloan Digital Sky Survey Commissioning Data”. In: Bibliography 177

The Astrophysical Journal Letters 548.2, pp. L165–L169. DOI: 10 . 1086 / 319095. arXiv: astro-ph/0012311 [astro-ph]. Ogando, Ricardo L. C. et al. (June 2008). “Line Strengths of Early-Type Galaxies”. In: As- tronomical Journal 135.6, pp. 2424–2445. DOI: 10.1088/0004-6256/135/6/2424. arXiv: 0803.3477 [astro-ph]. Oh, Semyeong et al. (June 2017). “Comoving Stars in Gaia DR1: An Abundance of Very Wide Separation Comoving Pairs”. In: Astronomical Journal 153.6, 257, p. 257. DOI: 10. 3847/1538-3881/aa6ffd. arXiv: 1612.02440 [astro-ph.SR]. Osborn, W. (Dec. 1971). “Two new CN-strong globular cluster stars”. In: The Observatory 91, pp. 223–224. Pan, Kuo-Chuan, Paul M. Ricker, and Ronald E. Taam (Aug. 2013). “Evolution of Post- impact Remnant Helium Stars in Type Ia Supernova Remnants within the Single-degenerate Scenario”. In: Astrophysical Journal 773.1, 49, p. 49. DOI: 10.1088/0004-637X/773/1/49. arXiv: 1303.1228 [astro-ph.HE]. Parmentier, G. et al. (Dec. 1999). “The self-enrichment of galactic halo globular clusters. A clue to their formation?” In: Astronomy & Astrophysics 352, pp. 138–148. arXiv: astro- ph/9911258 [astro-ph]. Pasquini, L. and P. Molaro (Mar. 1996). “Lithium abundance in the globular cluster NGC 6397.” In: Astronomy & Astrophysics 307, pp. 761–767. – (June 1997). “Lithium observations in 47 Tucanae.” In: Astronomy & Astrophysics 322, pp. 109–115. arXiv: astro-ph/9612158 [astro-ph]. Peñarrubia, Jorge, Matthew G. Walker, and Gerard Gilmore (Nov. 2009). “Tidal disrup- tion of globular clusters in dwarf galaxies with triaxial dark matter haloes”. In: Monthly Notices of the Royal Astronomical Society 399, pp. 1275–1292. DOI: 10 . 1111 / j . 1365 - 2966.2009.15027.x. arXiv: 0905.0924 [astro-ph.GA]. Piatti, Andrés E. and Julio A. Carballo-Bello (May 2019). “Extra-tidal structures around the Gaia Sausage candidate globular cluster NGC 6779 (M56)”. In: Monthly Notices of the Royal Astronomical Society 485.1, pp. 1029–1035. DOI: 10.1093/mnras/stz500. arXiv: 1902.05824 [astro-ph.GA]. Piotto, G. et al. (Mar. 2015). “The Hubble Space Telescope UV Legacy Survey of Galactic Globular Clusters. I. Overview of the Project and Detection of Multiple Stellar Popu- lations”. In: Astronomical Journal 149.3, 91, p. 91. DOI: 10.1088/0004-6256/149/3/91. arXiv: 1410.4564 [astro-ph.SR]. Piotto, Giampaolo (June 2009). “Observations of multiple populations in star clusters”. In: The Ages of Stars. Ed. by Eric E. Mamajek, David R. Soderblom, and Rosemary F. G. Wyse. Vol. 258. IAU Symposium, pp. 233–244. DOI: 10.1017/S1743921309031883. Piskunov, Nikolai and Jeff A. Valenti (Jan. 2017). “Spectroscopy Made Easy: Evolution”. In: Astronomy & Astrophysics 597, A16, A16. DOI: 10 . 1051 / 0004 - 6361 / 201629124. arXiv: 1606.06073 [astro-ph.IM]. Bibliography 178

Poggio, E. et al. (May 2017). “The kinematic signature of the Galactic warp in Gaia DR1. I. The Hipparcos subsample”. In: Astronomy & Astrophysics 601, A115, A115. DOI: 10. 1051/0004-6361/201629916. arXiv: 1702.04556 [astro-ph.GA]. Posti, Lorenzo and Amina Helmi (Jan. 2019). “Mass and shape of the Milky Way’s dark matter halo with globular clusters from Gaia and Hubble”. In: Astronomy & Astrophysics 621, A56, A56. DOI: 10.1051/0004-6361/201833355. arXiv: 1805.01408 [astro-ph.GA]. Prantzos, N. and C. Charbonnel (Oct. 2006). “On the self-enrichment scenario of galactic globular clusters: constraints on the IMF”. In: Astronomy & Astrophysics 458.1, pp. 135– 149. DOI: 10.1051/0004-6361:20065374. arXiv: astro-ph/0606112 [astro-ph]. Prantzos, N., C. Charbonnel, and C. Iliadis (Dec. 2017). “Revisiting nucleosynthesis in globular clusters. The case of NGC 2808 and the role of He and K”. In: Astronomy & Astrophysics 608, A28, A28. DOI: 10.1051/0004-6361/201731528. arXiv: 1709.05819 [astro-ph.GA]. Price-Jones, Natalie et al. (July 2020). “Strong chemical tagging with APOGEE: 21 candi- date star clusters that have dissolved across the Milky Way disc”. In: Monthly Notices of the Royal Astronomical Society 496.4, pp. 5101–5115. DOI: 10.1093/mnras/staa1905. arXiv: 2004.04263 [astro-ph.GA]. Price-Whelan, Adrian M. (Oct. 2017). “Gala: A Python package for galactic dynamics”. In: The Journal of Open Source Software 2, 388, p. 388. DOI: 10.21105/joss.00388. Primas, Francesca et al. (Jan. 2010). “The lithium history of NGC 6397”. In: Star Clus- ters: Basic Galactic Building Blocks Throughout Time and Space. Ed. by Richard de Grijs and Jacques R. D. Lépine. Vol. 266. IAU Symposium, pp. 143–148. DOI: 10 . 1017 / S1743921309990986. Randich, S. et al. (May 2018). “The Gaia-ESO Survey: open clusters in Gaia-DR1 . A way forward to stellar age calibration”. In: Astronomy & Astrophysics 612, A99, A99. DOI: 10.1051/0004-6361/201731738. arXiv: 1711.07699 [astro-ph.SR]. Recio-Blanco, A. and F. Thévenin (Jan. 2005). “Gaia Stellar Chemical Abundances and Galactic Archaeology”. In: The Three-Dimensional Universe with Gaia. Ed. by C. Turon, K. S. O’Flaherty, and M. A. C. Perryman. Vol. 576. ESA Special Publication, p. 501. Reggiani, Henrique et al. (Oct. 2020). “The Most Metal-poor Stars in the Inner Bulge”. In: The Astronomical Journal 160.4, 173, p. 173. DOI: 10.3847/1538- 3881/aba948. arXiv: 2007.12728 [astro-ph.SR]. Reina-Campos, Marta et al. (Apr. 2020). “The mass fraction of halo stars contributed by the disruption of globular clusters in the E-MOSAICS simulations”. In: Monthly Notices of the Royal Astronomical Society 493.3, pp. 3422–3428. DOI: 10.1093/mnras/staa483. arXiv: 1910.06973 [astro-ph.GA]. Richardson, Jenny C. et al. (May 2011). “PAndAS’ Progeny: Extending the M31 Dwarf Galaxy Cabal”. In: The Astrophysical Journal 732.2, 76, p. 76. DOI: 10.1088/0004-637X/ 732/2/76. arXiv: 1102.2902 [astro-ph.CO]. Bibliography 179

Ricotti, Massimo, Owen H. Parry, and Nickolay Y. Gnedin (Nov. 2016). “A Common Ori- gin for Globular Clusters and Ultra-faint Dwarfs in Simulations of the First Galaxies”. In: The Astrophysical Journal 831.2, 204, p. 204. DOI: 10.3847/0004-637X/831/2/204. arXiv: 1607.04291 [astro-ph.GA]. Robin, A. C. et al. (Sept. 2014). “Constraining the thick disc formation scenario of the Milky Way”. In: Astronomy & Astrophysics 569, A13, A13. DOI: 10.1051/0004- 6361/ 201423415. arXiv: 1406.5384 [astro-ph.GA]. Roederer, I. U., A. F. Marino, and C. Sneden (Nov. 2011). “Characterizing the Heavy El- ements in Globular Cluster M22 and an Empirical s-process Abundance Distribution Derived from the Two Stellar Groups”. In: The Astrophysical Journal 742.1, 37, p. 37. DOI: 10.1088/0004-637X/742/1/37. arXiv: 1108.3868 [astro-ph.SR]. Roederer, Ian U. et al. (June 2014). “A Search for Stars of Very Low Metal Abundance. VI. Detailed Abundances of 313 Metal-poor Stars”. In: Astronomical Journal 147.6, 136, p. 136. DOI: 10.1088/0004-6256/147/6/136. arXiv: 1403.6853 [astro-ph.SR]. Rybizki, Jan et al. (July 2020). “A Gaia Early DR3 Mock Stellar Catalog: Galactic Prior and Selection Function”. In: Publications of the Astronomical Society of the Pacific 132.1013, 074501, p. 074501. DOI: 10.1088/1538-3873/ab8cb0. arXiv: 2004.09991 [astro-ph.IM]. Sahijpal, Sandeep (Jan. 2013). “Influence of Supernova SN Ia Rate and the Early Star For- mation Rate on the Galactic Chemical Evolution”. In: International Journal of Astronomy and Astrophysics 3.3, pp. 344–352. DOI: 10.4236/ijaa.2013.33038. Sakari, Charli M. et al. (Dec. 2018). “The R-Process Alliance: First Release from the North- ern Search for r-process-enhanced Metal-poor Stars in the Galactic Halo”. In: The Astro- physical Journal 868.2, 110, p. 110. DOI: 10.3847/1538-4357/aae9df. arXiv: 1809.09156 [astro-ph.SR]. Sales, Laura V. et al. (Aug. 2007). “Satellites of simulated galaxies: survival, merging and their relation to the dark and stellar haloes”. In: Monthly Notices of the Royal Astronomical Society 379.4, pp. 1464–1474. DOI: 10.1111/j.1365-2966.2007.12024.x. arXiv: 0704. 1770 [astro-ph]. Scalzo, R. A., A. J. Ruiter, and S. A. Sim (Dec. 2014). “The ejected mass distribution of Type Ia supernovae: a significant rate of non-Chandrasekhar-mass progenitors”. In: Monthly Notices of the Royal Astronomical Society 445.3, pp. 2535–2544. DOI: 10.1093/ mnras/stu1808. arXiv: 1408.6601 [astro-ph.HE]. Schiavon, Ricardo P. et al. (Apr. 2017). “APOGEE chemical abundances of globular cluster giants in the inner Galaxy”. In: Monthly Notices of the Royal Astronomical Society 466.1, pp. 1010–1018. DOI: 10.1093/mnras/stw3093. arXiv: 1611.03086 [astro-ph.GA]. Schönrich, Ralph and James Binney (June 2009). “Chemical evolution with radial mix- ing”. In: Monthly Notices of the Royal Astronomical Society 396.1, pp. 203–222. DOI: 10. 1111/j.1365-2966.2009.14750.x. arXiv: 0809.3006 [astro-ph]. Bibliography 180

Searle, L. and R. Zinn (Oct. 1978). “Compositions of halo clusters and the formation of the galactic halo”. In: The Astrophysical Journal 225, pp. 357–379. DOI: 10.1086/156499. Sharma, Sanjib et al. (May 2016). “Stellar Population Synthesis Based Modeling of the Milky Way Using Asteroseismology of 13,000 Kepler Red Giants”. In: The Astrophysi- cal Journal 822.1, 15, p. 15. DOI: 10.3847/0004- 637X/822/1/15. arXiv: 1603.05661 [astro-ph.GA]. Sharma, Sanjib et al. (Feb. 2017). “Stellar Population Synthesis-based Modelling of the Milky Way Using Asteroseismology of Dwarfs and Subgiants from Kepler”. In: The Astrophysical Journal 835.2, 163, p. 163. DOI: 10.3847/1538- 4357/835/2/163. arXiv: 1701.02464 [astro-ph.GA]. Sharma, Sanjib et al. (Jan. 2018). “The TESS-HERMES survey data release 1: high-resolution spectroscopy of the TESS southern continuous viewing zone”. In: Monthly Notices of the Royal Astronomical Society 473.2, pp. 2004–2019. DOI: 10.1093/mnras/stx2582. arXiv: 1707.05753 [astro-ph.SR]. Sharma, Sanjib et al. (Dec. 2019). “The K2-HERMES Survey: age and metallicity of the thick disc”. In: Monthly Notices of the Royal Astronomical Society 490.4, pp. 5335–5352. DOI: 10.1093/mnras/stz2861. arXiv: 1904.12444 [astro-ph.GA]. Sheinis, Andrew et al. (July 2015). “First light results from the High Efficiency and Res- olution Multi-Element Spectrograph at the Anglo-Australian Telescope”. In: Journal of Astronomical Telescopes, Instruments, and Systems 1, 035002, p. 035002. DOI: 10.1117/1. JATIS.1.3.035002. Shetrone, M. D. (Oct. 1996). “Al, Mg and Eu Abundances in Globular Cluster Giants. I. Tests of Deep Mixing”. In: Astronomical Journal 112, p. 1517. DOI: 10.1086/118120. Shetrone, Matthew et al. (Feb. 2003). “VLT/UVES Abundances in Four Nearby Dwarf Spheroidal Galaxies. I. Nucleosynthesis and Abundance Ratios”. In: Astronomical Jour- nal 125.2, pp. 684–706. DOI: 10.1086/345966. arXiv: astro-ph/0211167 [astro-ph]. Shingles, Luke J. et al. (Nov. 2014). “The s-Process Enrichment of the Globular Clusters M4 and M22”. In: The Astrophysical Journal 795.1, 34, p. 34. DOI: 10.1088/0004-637X/ 795/1/34. arXiv: 1409.1227 [astro-ph.SR]. Shipp, N. et al. (Aug. 2018). “Stellar Streams Discovered in the Dark Energy Survey”. In: The Astrophysical Journal 862.2, 114, p. 114. DOI: 10.3847/1538- 4357/aacdab. arXiv: 1801.03097 [astro-ph.GA]. Simpson, Jeffrey D (Nov. 2020). GALAH DR3: Data reduction and analysis. https://docs. datacentral.org.au/galah/dr3/data-reduction-and-analysis/. Simpson, Jeffrey D., P. L. Cottrell, and C. C. Worley (Dec. 2012). “Spectral matching for abundances and clustering analysis of stars on the giant branches of ω Centauri”. In: Monthly Notices of the Royal Astronomical Society 427.2, pp. 1153–1167. DOI: 10.1111/j. 1365-2966.2012.22012.x. arXiv: 1209.0495 [astro-ph.SR]. Bibliography 181

Simpson, Jeffrey D. et al. (Jan. 2020). “The GALAH Survey: Chemically tagging the Fim- bulthul stream to the globular cluster ω Centauri”. In: Monthly Notices of the Royal Astro- nomical Society 491.3, pp. 3374–3384. DOI: 10.1093/mnras/stz3105. arXiv: 1911.01548 [astro-ph.GA]. Skrutskie, M. F. et al. (Feb. 2006). “The Two Micron All Sky Survey (2MASS)”. In: Astro- nomical Journal 131, pp. 1163–1183. DOI: 10.1086/498708. Skúladóttir, Á. et al. (Nov. 2019). “Neutron-capture elements in dwarf galaxies. I. Chem- ical clocks and the short timescale of the r-process”. In: Astronomy & Astrophysics 631, A171, A171. DOI: 10.1051/0004-6361/201936125. arXiv: 1908.10729 [astro-ph.GA]. Smith, M. C. et al. (Nov. 2009). “Kinematics of SDSS subdwarfs: structure and substruc- ture of the Milky Way halo”. In: Monthly Notices of the Royal Astronomical Society 399.3, pp. 1223–1237. DOI: 10.1111/j.1365-2966.2009.15391.x. arXiv: 0904.1012 [astro-ph.GA]. Smith, Verne V. et al. (Mar. 2000). “The Chemical Evolution of the Globular Cluster ω Centauri (NGC 5139)”. In: Astronomical Journal 119.3, pp. 1239–1258. DOI: 10.1086/ 301276. Sneden, C., J. J. Cowan, and R. Gallino (Sept. 2008). “Neutron-capture elements in the early galaxy.” In: Annual Reviews of Astronomy & Astrophysics 46, pp. 241–288. DOI: 10. 1146/annurev.astro.46.060407.145207. Soderblom, D. R. et al. (Jan. 2014). “Ages of Young Stars”. In: Protostars and Planets VI. Ed. by Henrik Beuther et al., p. 219. DOI: 10.2458/azu_uapress_9780816531240-ch010. arXiv: 1311.7024 [astro-ph.SR]. Sollima, A. (May 2020). “The eye of Gaia on globular clusters structure: tidal tails”. In: Monthly Notices of the Royal Astronomical Society 495.2, pp. 2222–2233. DOI: 10.1093/ mnras/staa1209. arXiv: 2004.13754 [astro-ph.GA]. Somerville, Rachel S. and Romeel Davé (Aug. 2015). “Physical Models of Galaxy Forma- tion in a Cosmological Framework”. In: Annual Reviews of Astronomy & Astrophysics 53, pp. 51–113. DOI: 10 . 1146 / annurev - astro - 082812 - 140951. arXiv: 1412 . 2712 [astro-ph.GA]. Spina, L. et al. (Oct. 2016). “Nucleosynthetic history of elements in the Galactic disk. [X/Fe]-age relations from high-precision spectroscopy”. In: Astronomy & Astrophysics 593, A125, A125. DOI: 10.1051/0004-6361/201628557. arXiv: 1606.04842 [astro-ph.SR]. Spina, Lorenzo et al. (Feb. 2018). “The temporal evolution of neutron-capture elements in the Galactic discs”. In: Monthly Notices of the Royal Astronomical Society 474.2, pp. 2580– 2593. DOI: 10.1093/mnras/stx2938. arXiv: 1711.03643 [astro-ph.SR]. Springel, V. et al. (Nov. 2008). “Prospects for detecting supersymmetric dark matter in the Galactic halo”. In: Nature 456.7218, pp. 73–76. DOI: 10.1038/nature07411. arXiv: 0809.0894 [astro-ph]. Bibliography 182

Springel, Volker, Carlos S. Frenk, and Simon D. M. White (Apr. 2006). “The large-scale structure of the Universe”. In: Nature 440.7088, pp. 1137–1144. DOI: 10.1038/nature04805. arXiv: astro-ph/0604561 [astro-ph]. Springel, Volker et al. (June 2005). “Simulations of the formation, evolution and clustering of galaxies and quasars”. In: Nature 435.7042, pp. 629–636. DOI: 10.1038/nature03597. arXiv: astro-ph/0504097 [astro-ph]. Straniero, O., S. Cristallo, and L. Piersanti (Apr. 2014). “Heavy Elements in Globular Clus- ters: The Role of Asymptotic Giant Branch Stars”. In: The Astrophysical Journal 785.1, 77, p. 77. DOI: 10.1088/0004-637X/785/1/77. arXiv: 1403.0819 [astro-ph.GA]. Stringer, M. J. and A. J. Benson (Dec. 2007). “The formation of galaxy discs in a hierar- chical universe”. In: Monthly Notices of the Royal Astronomical Society 382.2, pp. 641–651. DOI: 10.1111/j.1365-2966.2007.12469.x. arXiv: astro-ph/0703380 [astro-ph]. Strobel, A. (July 1991). “An observational age-metallicity relation for the galaxy.” In: As- tronomy & Astrophysics 247, p. 35. Takada, Masahiro et al. (Feb. 2014). “Extragalactic science, cosmology, and Galactic ar- chaeology with the Subaru Prime Focus Spectrograph”. In: Publications of the Astro- nomical Society of Japan 66.1, R1, R1. DOI: 10 . 1093 / pasj / pst019. arXiv: 1206 . 0737 [astro-ph.CO]. Tantalo, Rosaria and Cesare Chiosi (Sept. 2004). “Measuring age, metallicity and abun- dance ratios from absorption-line indices”. In: Monthly Notices of the Royal Astronomical Society 353.3, pp. 917–940. DOI: 10.1111/j.1365-2966.2004.08123.x. arXiv: astro- ph/0305247 [astro-ph]. Taylor, M. B. (Dec. 2005). “TOPCAT & STIL: Starlink Table/VOTable Processing Soft- ware”. In: Astronomical Data Analysis Software and Systems XIV. Ed. by P. Shopbell, M. Britton, and R. Ebert. Vol. 347. Astronomical Society of the Pacific Conference Series, p. 29. Taylor, Philip and Chiaki Kobayashi (Apr. 2015). “The effects of AGN feedback on present- day galaxy properties in cosmological simulations”. In: Monthly Notices of the Royal As- tronomical Society 448.2, pp. 1835–1846. DOI: 10.1093/mnras/stv139. arXiv: 1501.04133 [astro-ph.GA]. Terlevich, A. I. and Duncan A. Forbes (Mar. 2002). “A catalogue and analysis of local galaxy ages and metallicities”. In: Monthly Notices of the Royal Astronomical Society 330.3, pp. 547–562. DOI: 10.1046/j.1365-8711.2002.05073.x. arXiv: astro-ph/0110413 [astro-ph]. Thévenin, F. et al. (Feb. 2017). “Evolution of lithium abundance in the Sun and solar twins”. In: Astronomy & Astrophysics 598, A64, A64. DOI: 10.1051/0004-6361/201629385. arXiv: 1612.01331 [astro-ph.SR]. Bibliography 183

Thomas, Daniel (July 1999). “Abundance ratios in hierarchical galaxy formation”. In: Monthly Notices of the Royal Astronomical Society 306.3, pp. 655–661. DOI: 10.1046/j. 1365-8711.1999.02552.x. arXiv: astro-ph/9901226 [astro-ph]. Thygesen, A. O. et al. (Dec. 2014). “The chemical composition of red giants in 47 Tucanae. I. Fundamental parameters and chemical abundance patterns”. In: Astronomy & As- trophysics 572, A108, A108. DOI: 10.1051/0004- 6361/201424533. arXiv: 1409.4694 [astro-ph.SR]. Timmes, F. X., S. E. Woosley, and Thomas A. Weaver (June 1995). “Galactic Chemical Evolution: Hydrogen through Zinc”. In: The Astrophysical Journal Supplement Series 98, p. 617. DOI: 10.1086/192172. arXiv: astro-ph/9411003 [astro-ph]. Ting, Yuan-Sen, Charlie Conroy, and Alyssa Goodman (July 2015). “Prospects for Chem- ically Tagging Stars in the Galaxy”. In: The Astrophysical Journal 807.1, 104, p. 104. DOI: 10.1088/0004-637X/807/1/104. arXiv: 1504.03327 [astro-ph.GA]. Ting, Yuan-Sen, Charlie Conroy, and Hans-Walter Rix (Jan. 2016). “APOGEE Chemical Tagging Constraint on the Maximum Star Cluster Mass in the Alpha-enhanced Galactic Disk”. In: The Astrophysical Journal 816.1, 10, p. 10. DOI: 10.3847/0004-637X/816/1/10. arXiv: 1507.07563 [astro-ph.GA]. Ting, Yuan-Sen et al. (Apr. 2012). “Principal component analysis on chemical abundances spaces”. In: Monthly Notices of the Royal Astronomical Society 421.2, pp. 1231–1255. DOI: 10.1111/j.1365-2966.2011.20387.x. arXiv: 1112.3207 [astro-ph.SR]. Tinsley, B. M. (May 1979). “Stellar lifetimes and abundance ratios in chemical evolution.” In: The Astrophysical Journal 229, pp. 1046–1056. DOI: 10.1086/157039. Tissera, Patricia B. et al. (Apr. 2014). “Stellar haloes in Milky Way mass galaxies: from the inner to the outer haloes”. In: Monthly Notices of the Royal Astronomical Society 439.3, pp. 3128–3138. DOI: 10.1093/mnras/stu181. arXiv: 1309.3609 [astro-ph.CO]. Titarenko, Anastasia (Sept. 2018). “Stellar chemical clocks in the Gaia era”. PhD thesis. Université Côôte d’Azur. Tiwari, Juhi, Smriti Mahajan, and Kulinder Pal Singh (Nov. 2020). “Age and metallic- ity of galaxies in different environments of the Coma supercluster”. In: Nature Astron- omy 81, 101417, p. 101417. DOI: 10.1016/j.newast.2020.101417. arXiv: 2005.02757 [astro-ph.GA]. Tolstoy, Eline, Vanessa Hill, and Monica Tosi (Sept. 2009). “Star-Formation Histories, Abundances, and Kinematics of Dwarf Galaxies in the Local Group”. In: Annual Re- views of Astronomy & Astrophysics 47.1, pp. 371–425. DOI: 10.1146/annurev- astro- 082708-101650. arXiv: 0904.4505 [astro-ph.CO]. Tomaschitz, Roman (Jan. 2018). “White dwarf stars exceeding the Chandrasekhar mass limit”. In: Physica A Statistical Mechanics and its Applications 489, pp. 128–140. DOI: 10. 1016/j.physa.2017.07.024. Bibliography 184

Torrealba, G. et al. (Sept. 2019). “The hidden giant: discovery of an enormous Galactic dwarf satellite in Gaia DR2”. In: Monthly Notices of the Royal Astronomical Society 488.2, pp. 2743–2766. DOI: 10.1093/mnras/stz1624. arXiv: 1811.04082 [astro-ph.GA]. Trenti, Michele, Paolo Padoan, and Raul Jimenez (Aug. 2015). “The Relative and Absolute Ages of Old Globular Clusters in the LCDM Framework”. In: The Astrophysical Journal Letters 808.2, L35, p. L35. DOI: 10.1088/2041- 8205/808/2/L35. arXiv: 1502.02670 [astro-ph.GA]. Truran, James W. et al. (Dec. 2002). “Probing the Neutron-Capture Nucleosynthesis His- tory of Galactic Matter”. In: The Publications of the Astronomical Society of the Pacific 114.802, pp. 1293–1308. DOI: 10.1086/344585. arXiv: astro-ph/0209308 [astro-ph]. Tsujimoto, Takuji and Toshikazu Shigeyama (June 2003). “Star Formation History of ω Centauri Imprinted in Elemental Abundance Patterns”. In: The Astrophysical Journal 590.2, pp. 803–808. DOI: 10.1086/375023. arXiv: astro-ph/0302300 [astro-ph]. Unsöld, Albrecht O. J. (May 1970). “Stellar Abundaces and the Origin of the Elements (Part 1)”. In: Leaflet of the Astronomical Society of the Pacific 10.491, pp. 321–328. Usher, Christopher et al. (Oct. 2012). “The SLUGGS survey: calcium triplet-based spec- troscopic metallicities for over 900 globular clusters”. In: Monthly Notices of the Royal Astronomical Society 426.2, pp. 1475–1495. DOI: 10.1111/j.1365-2966.2012.21801.x. arXiv: 1207.6402 [astro-ph.CO]. Uttenthaler, S. et al. (Jan. 2012). “Lithium destruction and production observed in red giant stars”. In: Memorie della Societa Astronomica Italiana Supplementi 22, p. 56. arXiv: 1206.2759 [astro-ph.SR]. Valenti, J. A. and N. Piskunov (Sept. 1996). “Spectroscopy made easy: A new tool for fitting observations with synthetic spectra.” In: Astronomy & Astrophysics Supplement Series 118, pp. 595–603. Vanbeveren, D., N. Mennekens, and J. P. De Greve (July 2012). “The effect of intermediate- mass close binaries on the chemical evolution of globular clusters”. In: Astronomy & Astrophysics 543, A4, A4. DOI: 10 . 1051 / 0004 - 6361 / 201118081. arXiv: 1109 . 2713 [astro-ph.GA]. Vasiliev, Eugene (Apr. 2019). “Proper motions and dynamics of the Milky Way globular cluster system from Gaia DR2”. In: Monthly Notices of the Royal Astronomical Society 484.2, pp. 2832–2850. DOI: 10.1093/mnras/stz171. arXiv: 1807.09775 [astro-ph.GA]. Vauclair, S. (Dec. 2013). “Atomic Diffusion, Mixing, and Element Abundances in Main Sequence Stars”. In: Progress in Physics of the Sun and Stars: A New Era in Helio- and Asteroseismology. Ed. by H. Shibahashi and A. E. Lynas-Gray. Vol. 479. Astronomical Society of the Pacific Conference Series, p. 27. Cotar,ˇ Klemen et al. (Aug. 2019). “The GALAH survey: unresolved triple Sun-like stars discovered by the Gaia mission”. In: Monthly Notices of the Royal Astronomical Society 487.2, pp. 2474–2490. DOI: 10.1093/mnras/stz1397. arXiv: 1904.04841 [astro-ph.SR]. Bibliography 185

Venn, Kim A. et al. (Sept. 2004). “Stellar Chemical Signatures and Hierarchical Galaxy Formation”. In: Astronomical Journal 128.3, pp. 1177–1195. DOI: 10.1086/422734. arXiv: astro-ph/0406120 [astro-ph]. Ventura, P. et al. (Nov. 2016). “Evidence of AGB Pollution in Galactic Globular Clusters from the Mg-Al Anticorrelations Observed by the APOGEE Survey”. In: The Astro- physical Journal Letters 831.2, L17, p. L17. DOI: 10.3847/2041-8205/831/2/L17. arXiv: 1610.03243 [astro-ph.SR]. Ventura, Paolo et al. (Mar. 2001). “Predictions for Self-Pollution in Globular Cluster Stars”. In: The Astrophysical Journal Letters 550.1, pp. L65–L69. DOI: 10.1086/319496. arXiv: astro-ph/0103337 [astro-ph]. Villanova, S. et al. (Dec. 2013). “Ruprecht 106: The First Single Population Globular Clus- ter?” In: The Astrophysical Journal 778.2, 186, p. 186. DOI: 10.1088/0004-637X/778/2/ 186. arXiv: 1310.5900 [astro-ph.GA]. Vincenzo, Fiorenzo, Chiaki Kobayashi, and Philip Taylor (Oct. 2018). “On the [α/Fe]- [Fe/H] relations in early-type galaxies”. In: Monthly Notices of the Royal Astronomi- cal Society 480.1, pp. L38–L42. DOI: 10 . 1093 / mnrasl / sly128. arXiv: 1807 . 03416 [astro-ph.GA]. Vincenzo, Fiorenzo et al. (July 2019). “The Fall of a Giant. Chemical evolution of Ence- ladus, alias the Gaia Sausage”. In: Monthly Notices of the Royal Astronomical Society 487.1, pp. L47–L52. DOI: 10.1093/mnrasl/slz070. arXiv: 1903.03465 [astro-ph.GA]. Žerjal, M. et al. (Apr. 2019). “The GALAH Survey: lithium-strong KM dwarfs”. In: Monthly Notices of the Royal Astronomical Society 484.4, pp. 4591–4600. DOI: 10 . 1093 / mnras / stz296. arXiv: 1810.10435 [astro-ph.SR]. Walker, A. R. et al. (July 2011). “Constraints on the formation of the globular cluster IC 4499 from multiwavelength photometry”. In: Monthly Notices of the Royal Astronomical Society 415.1, pp. 643–654. DOI: 10.1111/j.1365-2966.2011.18736.x. arXiv: 1103. 4144 [astro-ph.SR]. Wan, Zhen et al. (July 2020). “The tidal remnant of an unusually metal-poor globular clus- ter”. In: arXiv e-prints, arXiv:2007.14577, arXiv:2007.14577. arXiv: 2007.14577 [astro-ph.GA]. Wheeler, Adam, Melissa Ness, and David W. Hogg (Sept. 2020). “An unsupervised method for identifying X-enriched stars directly from spectra: Li in LAMOST”. In: arXiv e- prints, arXiv:2009.03920, arXiv:2009.03920. arXiv: 2009.03920 [astro-ph.SR]. Wheeler, J. Craig, Christopher Sneden, and Jr. Truran James W. (Jan. 1989). “Abundance ratios as a function of metallicity.” In: Annual Reviews of Astronomy & Astrophysics 27, pp. 279–349. DOI: 10.1146/annurev.aa.27.090189.001431. Wisnioski, E. et al. (Feb. 2015). “The KMOS3D Survey: Design, First Results, and the Evolu- tion of Galaxy Kinematics from 0.7 <= z <= 2.7”. In: The Astrophysical Journal 799.2, 209, p. 209. DOI: 10.1088/0004-637X/799/2/209. arXiv: 1409.6791 [astro-ph.GA]. Bibliography 186

Xiang, Maosheng et al. (Dec. 2019). “Abundance Estimates for 16 Elements in 6 Million Stars from LAMOST DR5 Low-Resolution Spectra”. In: The Astrophysical Journal Sup- plement Series 245.2, 34, p. 34. DOI: 10.3847/1538- 4365/ab5364. arXiv: 1908.09727 [astro-ph.SR]. Yong, D. et al. (Aug. 2005). “Abundances in giant stars of the globular cluster NGC 6752”. In: Astronomy & Astrophysics 438.3, pp. 875–888. DOI: 10.1051/0004-6361:20052916. arXiv: astro-ph/0504283 [astro-ph]. Yong, David et al. (Dec. 2008). “Heavy Element Abundances in Giant Stars of the Globular Clusters M4 and M5”. In: The Astrophysical Journal 689.2, pp. 1031–1043. DOI: 10.1086/ 592600. arXiv: 0808.2505 [astro-ph]. Yong, David et al. (Jan. 2013). “The Most Metal-poor Stars. II. Chemical Abundances of 190 Metal-poor Stars Including 10 New Stars with [Fe/H] <= -3.5”. In: The Astrophys- ical Journal 762.1, 26, p. 26. DOI: 10.1088/0004- 637X/762/1/26. arXiv: 1208.3003 [astro-ph.GA]. Zacharias, N. et al. (June 2010). “The Third US Naval Observatory CCD Astrograph Cat- alog (UCAC3)”. In: Astronomical Journal 139.6, pp. 2184–2199. DOI: 10 . 1088 / 0004 - 6256/139/6/2184. arXiv: 1003.2136 [astro-ph.IM]. Zacharias, N. et al. (Feb. 2013). “The Fourth US Naval Observatory CCD Astrograph Cat- alog (UCAC4)”. In: Astronomical Journal 145.2, 44, p. 44. DOI: 10.1088/0004-6256/145/ 2/44. arXiv: 1212.6182 [astro-ph.IM]. Zhang, Jiang et al. (Dec. 2013). “Puzzling Origin of CEMP-r/s Stars: An Interpretation of Abundance and Enrichment of s- and r-Process Elements from Asymptotic Giant Branch Supernovae”. In: Journal of Astrophysics and Astronomy 34.4, pp. 373–391. DOI: 10.1007/s12036-013-9191-7. Zinn, R. (June 1985). “The Globular Cluster System of the Galaxy. IV. The Halo and Disk Subsystems”. In: The Astrophysical Journal 293, p. 424. DOI: 10.1086/163249. Zinn, Robert (Jan. 1993). “The Galactic Halo Cluster Systems: Evidence for Accretion”. In: The Globular Cluster-Galaxy Connection. Ed. by Graeme H. Smith and Jean P. Brodie. Vol. 48. Astronomical Society of the Pacific Conference Series, p. 38. Zolotov, Adi et al. (Sept. 2010). “The Dual Origin of Stellar Halos. II. Chemical Abun- dances as Tracers of Formation History”. In: The Astrophysical Journal 721.1, pp. 738– 743. DOI: 10.1088/0004-637X/721/1/738. arXiv: 1004.3789 [astro-ph.GA].