UNIFORM CHEMICAL ABUNDANCES OF OPEN CLUSTERS USING THE CANNON

by

AMY E. RAY

Bachelor of Science, 2013 Mississippi State University Mississippi State, MS

Master of Science, 2017 Mississippi State University Mississippi State, MS

Submitted to the Graduate Faculty of the College of Science and Engineering Texas Christian University in partial fulfillment of the requirements for the degree of

Master of Science

December 2018

ACKNOWLEDGEMENTS

IwouldliketothankDr.Frinchaboyforbeinganexcellentadvisor. Thank you fellow graduate students, my committee, my parents, and my puppy Andromeda “Annie” Ray. I would like to acknowledge grant support for this research from the National Science Foundation (AST-1311835 & AST-1715662). Funding for SDSS-III has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, and the U.S. Department of Energy Oce of Science. The SDSS-III web site is http://www.sdss3.org/. SDSS-III is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS-III Collaboration including the University of Arizona, the Brazil- ian Participation Group, Brookhaven National Laboratory, Carnegie Mellon University, University of Florida, the French Participation Group, the German Participation Group, Harvard University, the Instituto de Astrofisica de Canarias, the Michigan State/Notre Dame/JINA Participation Group, Johns Hopkins University, Lawrence Berkeley National Laboratory, Max Planck Institute for Astrophysics, Max Planck Institute for Extraterres- trial Physics, New Mexico State University, New York University, Ohio State University, Pennsylvania State University, University of Portsmouth, Princeton University, the Span- ish Participation Group, University of Tokyo, University of Utah, Vanderbilt University, University of Virginia, University of Washington, and Yale University. Funding for the Sloan Digital Sky Survey IV has been provided by the Alfred P. Sloan Foundation, the U.S. Department of Energy Oce of Science, and the Participating Institutions. SDSS-IV acknowledges support and resources from the Center for High- Performance Computing at the University of Utah. The SDSS web site is www.sdss.org. SDSS-IV is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS Collaboration including the Brazilian Participation Group, the Carnegie Institution for Science, Carnegie Mellon University, the Chilean Participation Group, the French Participation Group, Harvard-Smithsonian Center for Astrophysics, Instituto de Astrofisica de Canarias, The Johns Hopkins University, Kavli Institute for the Physics and Mathematics of the Universe (IPMU)/University of Tokyo, Lawrence Berke- ley National Laboratory, Leibniz Institut fur Astrophysik Potsdam (AIP), Max-Planck- Institut fur Astronomie (MPIA Heidelberg), Max- Planck-Institut fur Astrophysik (MPA Garching), Max-Planck- Institut fur Extraterrestrische Physik (MPE), National Astro- nomical Observatories of China, New Mexico State University, New York University, University of Notre Dame, Observatario Nacional/MCTI, The Ohio State University, Pennsylvania State University, Shanghai Astronomical Observatory, United Kingdom Participation Group, Universidad Nacional Autonoma de Mexico, University of Arizona, University of Colorado Boulder, University of Oxford, University of Portsmouth, Univer- sity of Utah, University of Virginia, University of Washington, University of Wisconsin, Vanderbilt University, and Yale University.

ii Contents

ListofCommonAcronymsandTermsinAstronomy viii

1 Introduction 1 1.1 ChemicalAbundances(Metallicity) ...... 2 1.2 Open Clusters ...... 4 1.3 Goals ...... 5

2 Data Collection and Interpretation 9 2.1 VictorBlancoTelescope ...... 9 2.1.1 TheHydraMulti-FiberSpectrograph ...... 11 2.1.2 Data Preparation ...... 11 2.1.3 Cluster Membership Determination from (Frinchaboy & Majewski 2008) ...... 13 2.2 SpectralAnalysis ...... 15 2.2.1 Line Broadening Mechanisms ...... 16 2.2.2 Curve of Growth ...... 19 2.2.3 Boltzmann Equation ...... 20 2.2.4 Saha Equation ...... 21 2.3 Important Stellar Parameters ...... 22 2.3.1 E↵ective Temperature ...... 23 2.3.2 Surface Gravity ...... 23 2.3.3 Metallicity ...... 24 2.3.4 Stellar Parameters and The Cannon ...... 25

3 Results 27 3.1 The Cannon ...... 27 3.1.1 Training Set ...... 28 3.1.2 IndividualStatResultsandDataQuality...... 31 3.1.3 Verification of Cluster Membership and Bulk Cluster Metallities . 33 3.2 TotalClusterSample ...... 33

4 Discussion 35 4.1 Summary ...... 35 4.1.1 Comparison to Other Surveys ...... 36 4.1.2 Discrepancies ...... 42

iii 4.1.3 NewValues ...... 44 4.2 Future Work ...... 44

A The Cluster Sample 46

Vita

Abstract

iv List of Figures

1.1 An example of a star cluster over time. The young cluster in the leftmost panel has most of its stars on the main sequence. With time, more stars move o↵the main sequence, and from the turno↵point an age can be estimated...... 5 1.2 Average iron abundances for a sample of open clusters. The plus symbols connected by red lines indicate measurements for the same cluster. This highlights the wide scatter in abundances across surveys that are discussed in Yong et al. (2012)...... 6

2.1 The Blanco 4 m telescope at CTIO in Chile...... 10 2.2 An example of a Cassegrain telescope. The Blanco 4 m has a similar setup. 10 2.3 The Hydra fiber positioner system that used to be on the Blanco 4 m. . . 11 2.4 A histogram from Frinchaboy & Majewski (2008) that shows the radial velocity distribution of all stars with proper motion data that are in the fieldofthecalibrationclusterNGC2682...... 13 2.5 These graphs show the steps in the membership analysis process for NGC 2682 that was done by Frinchaboy & Majewski (2008). (a) shows the result of the radial velocity distribution convolution with a Gaussian kernel to smooth the histogram shown in Figure 2.4, (b) shows the same smoothing process that was done for stars that fall outside of the cluster radius of NGC 2682, (c) shows the probability distribution for the cluster where non- members appear ouside of the main peak, and (d) shows a 1D Gaussian fit to the cluster distribution in (c)...... 14 2.6 CMD showing all of the stars within the cluster radius for NGC 2682. The larger black circles are members from both RV and proper motion deter- minations, crosses are stars that were found to be non-members, triangles are stars with high RV, but no proper motion values and the small grey circles are other non-member stars (Frinchaboy & Majewski 2008). . . . . 15 2.7 This is an example of curve of growth for the Sun. The dashed lines are separating the three di↵erent regions of the three functional forms (CarrollOstlie)...... 20 2.8 An example of how surface gravity impacts the strength of lines. The surface gravity is lowest at the top spectrum and highest at the bottom spectrum. As surface gravity increases, the pressure broadening wings increase...... 24

v 3.1 One-to-one plots showing a comparison between the values obtained using The Cannon and the APOGEE DR14 labels for the training set...... 32

4.1 A one-to-one comparison of [Fe/H] values for open clusters in common between Santos et al. (2009) and this study. The grey shaded area indicates the 0.17 scatter above and below the blue one-to-one line...... 37 4.2 A one-to-one± comparison of average [Fe/H] values obtained by Reddy et al. (2013; 2015) and this study for the open clusters in common...... 38 4.3 A comparison of [Fe/H] values for open clusters in common from Netopil et al. (2016) and this study...... 40 4.4 A comparison of average [Fe/H] values for in common open clusters from this survey and from the literature compilation in Table 4.4. Blue stars are [Fe/H] averages for open clusters IC 4756 and orange diamonds are [Fe/H]averagesforopenclusterNGC5822...... 41

vi List of Tables

1.1 The open clusters with no prior chemical abundance measurements. . . 7 1.2 The open clusters with chemical abundance measurements that will be used to verify the values obtained in this study for the 12 open clusters listed in Table 1.1...... 8

3.1 Values for Teff ,logg, and [Fe/H] from APOGEE compared to The Cannon output for the training set...... 28

4.1 Average in common open cluster iron abundance from Santos et al. (2009) compared to ( this study)...... 36 4.2 Average in common open cluster iron abundance comparison between Reddy et al. (2013; 2015) and this study...... 38 4.3 Average iron abundances for open clusters in common between Netopil et al. (2016) and this study...... 39 4.4 Average open cluster iron abundance for open clusters in common between this study and other literature studies...... 41 4.5 Average open cluster iron abundance for clusters with no prior measure- ments...... 44

A.1 The output from The Cannon for every star in the open cluster sample. The errors for Teff ,logg, and [Fe/H] are 141 K, 0.29 dex, and 0.17 dex, respectively. Cluster members used in± the average± Fe/H cluster± de- termination are indicated by a “Y” in the last column and stars that were either non-members or dwarf stars are indicated by “N”...... 46

vii List of Common Acronyms and Terms in Astronomy

Metals – All elements heavier than hydrogen and helium • Metallicity – The mass fraction of metals in a star with respect to • hydrogen

– [Fe/H] – Often used as a proxy for metallicity in stellar spectroscopic studies

– [Fe/H] = log(Fe/H)star log(Fe/H) dex – Decimal exponent (deprecated) • – Unit used for abundance value (e.g. [Fe/H] = 0.06 dex)

LTE – Local Thermal Equilibrium • – The total energy of a system is constant and amount is independent of particle position. Particles also follow a Maxwell Boltzmann distribution.

10 A˚ – Angstrom; 10 m • –Wavelength,usuallyinA˚ • ly – 1 lightyear = 9.46 x 1012 km • kpc – 1000 parsecs; 1 parsec (pc) = 3.262 ly • Main Sequence star – Star fusing hydrogen (H) to helium (He) in its core • APOGEE – Apache Point Observatory Galactic Evolution Experiment • SDSS – Sloan Digital Sky Survey • J, K – The near-infrared magnitudes taken from APOGEE • s J K – Di↵erence between magnitudes, also known as the color index • s IRAF – Image Reduction and Analysis Facility •

viii Chapter 1

Introduction

Galactic evolution is an important aspect of understanding how the universe has changed and is changing. One way to get an idea of how galaxies change during their lifetime is by studying the Milky Way. Studying the Milky Way allows us to see intricate details, such as individual stars and gas clouds, that are dicult to observe in the majority of other galaxies. Useful components to study are groups of stars known as open clusters. These clusters generally lie in the disk of the Milky Way and are thought to have formed at the same time out of similar material. By analyzing the properties of these clustersincluding their color, brightness, and chemical composition, we can accurately determine when its stars were formed. By surveying properties of open clusters throughout the Milky Way, we can piece together an evolutionary story for our galaxy that includes a timeline.

The concentration of heavy elements that a star contains, i.e., its chemical abundance, can be used to explore and constrain which stellar processes occurred in the Milky Way.

Astronomers refer to elements heavier than hydrogen and helium as metals. When the earliest stars formed in our galaxy, they contained few metals because the gas they formed

1 from was composed of hydrogen, helium, and lithium. With each new generation of stars, heavier elements build up by various processes, some occurring inside of stars themselves and some as the stars die in supernovae. Large, uniform surveys are advantageous for investigating the chemical history of the Milky Way. This survey aims to provide a uniform set of chemical abundances for a set of open clusters that can be added to similar studies.

In this thesis, we provide a description of work done in this area as well as the contributions of this study. In the rest of this chapter, key concepts are more thoroughly defined, the goals of this study are outlined, and the data used is presented. The process of obtaining observations, reducing the data, and the subsequent analysis methods are discussed in Section 2. Section 3 presents and discusses the results of this study compared to similar studies. Finally, Section 4 explains the results and details the motivations for future projects concerning this research.

1.1 Chemical Abundances (Metallicity)

Examining chemical abundance changes across the Milky Way is particularly useful for understanding its history. As stars cycle from one generation to the next, heavier elements begin to build up. There are several di↵erent ways these elements can build up. The first involves fusion processes that occur within stars such as the proton-proton chain (PP- chain), the carbon-nitrogen-oxygen (CNO) cycle, the alpha process, the slow neutron- capture process (s-process), and the rapid neutron-capture process (r-process).

2 The PP-chain converts hydrogen to helium in the cores of stars and consists of three main parts. The first one is called the PPI branch. This branch occurs the majority of the time and begins with two protons colliding to produce deuterium. Another proton collides with deuterium to produce helium-3. Then two helium-3 fuse to create helium-

4. The other two parts of the PP-chain, or the PPII and PPIII branches also produce helium-4. The PPII branch involves the creation of beryllium-7 and lithium to make helium-4. In the PPIII branch, the beryllium-7 created from the PPII branch reacts with hydrogen to produce boron. This boron then decays to beryllium-8 which then decays to two helium-4.

The CNO cycle is the second reaction where helium-4 is created from hydrogen using carbon, nitrogen, and oxygen. There are two di↵erent branches in this process. The CNOI branch ends with nitrogen-15 and hydrogen producing a carbon-12 and two helium-4. The

CNOII branch begins with the last step in the CNOI branch producing oxygen-16 and a photon instead.

Another fusion reaction is the triple-alpha process. The first step in this reaction is where two helium-4 produce beryllium-8. This nucleus is very unstable and will quickly decay back into two helium-4. Although, if the beryllium-8 nucleus interacts with another helium-4 before it decays, it will produce carbon-12.

Two more processes that produce heavier elements are called the s-process and r- process. The s-process is where a nucleus captures a neutron resulting in an isotope with a higher atomic mass. If the isotope is unstable, it will beta-decay. The reason this process is slow is that it occurs on much larger timescales as opposed to the beta- decay half-life. This process produces mostly stable nuclei. The r-process is similar to

3 the s-process except it occurs on timescales much shorter than the beta-decay half-life.

This process requires a large number of neutrons to occur meaning this process is more relevant during a supernova. A higher number of unstable isotopes are created from the r-process.

1.2 Open Clusters

While measuring the chemical abundance of every star that is observable in the Milky

Way disk is an important aspect when it comes to understanding its chemical distribution and formation history, getting accurate ages for individual stars is quite dicult. We know the age of the Sun from radioactive dating of objects in the solar system, but it is challenging to form a sample of stars to compare to the Sun. Luckily, there is a way to determine relatively accurate ages for some stars. Groups of stars known as open clusters consist of up to several hundred stars that formed at approximately the same time. This makes them ideal for studying the evolution of the galaxy because all of them share a similar age. Ages of open clusters can be determined from what is known as the main-sequence turno↵.

The “main sequence” phase of a star’s evolution is when the star’s primary energy source is through the fusion of hydrogen into helium in its core and when this energy production is able to generate enough outward forcethrough a combination of thermal pressure or radiation pressureto balance against the gravitational inward force of the star’s own weight. This sequence is illustrated on a Hertzsprung-Russell (H-R) diagram in the leftmost panel of Figure 1.1, which appears as a continuous diagonal line in this

4 luminosity versus temperature plot. The main sequence spans from low temperatures and luminosities to higher temperatures and luminosities. Also shown in Image 1.1, is the evolution of a cluster where all of the stars begin on the main sequence, and as time continues, the stars move above the main sequence starting with the higher mass stars and going to the lower mass ones.

Figure 1.1: An example of a star cluster over time. The young cluster in the leftmost panel has most of its stars on the main sequence. With time, more stars move o↵the main sequence, and from the turno↵point an age can be estimated.

1.3 Goals

Large uniform samples of open clusters are an excellent way to investigate galactic trends.

An example is searching for signatures of star-formation history in the Milky Way by investigating chemical abundance gradients in open clusters across the disk of the galaxy.

One way to improve the current knowledge in this area is to increase the number of clus- ters with known chemical abundances. There are roughly 2000 known open clusters, but

5 only a small portion of them have been analyzed. Even the ones with known values have substantial uncertainties from study to study. A few reasons for such large uncertainties are due to varying data quality, the type of data, and di↵erent data analysis methods between studies. This is illustrated in Figure 1.2 from Yong et al. (2012) where iron abundances for a set of clusters are plotted. Another issue that arises is which catalog each survey used for distances to open clusters, as there are several that have determined substantially di↵erent distance results. This di↵erence translates into widely varying re- sults when attempting to determine a chemical abundance gradient across the disk of the

Milky Way. Yong et al. (2012) and Donor et al. (2018) highlight this problem in their abundance gradient research.

Figure 1.2: Average iron abundances for a sample of open clusters. The plus symbols connected by red lines indicate measurements for the same cluster. This highlights the wide scatter in abundances across surveys that are discussed in Yong et al. (2012).

Our goal was to put together a large uniform sample that matched on the Apache

Point Observatory Galactic Evolution Experiment Data Release 14 (APOGEE DR14)

6 Table 1.1: The open clusters with no prior chemical abundance measurements.

Cluster l b d log(age) diameter Name (deg) (deg) (pc) (yr) (arcmin) Collinder 205 269.2091 1.8434 1853 7.200 5 Collinder 258 299.9710 +1.9654 1184 8.032 5 NGC2437 231.8575 +4.0644 1375 8.390 20 NGC 2546 254.8551 1.9859 919 7.874 70 NGC2579 254.6741 +0.2126 1033 7.610 7 NGC 2669 267.4854 3.6250 1046 7.927 20 NGC 5281 309.0102 2.4915 1108 7.146 7 NGC 6124 332.9179 3.1668 512 8.147 39 NGC6167 338.4047 +1.2106 1108 7.887 7 NGC 6250 341.9974 1.5166 865 7.415 10 NGC 6885 66.1352 6.3113 597 9.160 10 Ruprecht 119 333.2758 1.8794 956 6.853 8 system in order to correct the problems with current surveys that were listed above.

This sample could also be combined with open clusters from Donor et al. (2018) to form a more extensive dataset for galactic abundance studies. The sample used here consists of 31 open clusters. There were 12 open clusters, listed in Table 1.1, that did not have chemical abundances measured before. Stellar parameters for all cluster stars were measured on a standard system developed by Ness et al. (2015) known as The Cannon, using medium resolution data with low signal-to-noise spectra. Spectral resolution is defined as R = / where is the wavelength and is the smallest wavelength interval that can be resolved. Low resolution spectra have R<7, 000, medium resolution spectra have 7, 000 R 20, 000, and high resolution spectra have R>20, 000. We   verified that those clusters determinations were reliable by comparing to 19 open clusters,

Table 1.2, that had been studied before. Positions, distances, and ages for all of these clusters were obtained from the Dias et al. (2002) catalog of open clusters.

7 Table 1.2: The open clusters with chemical abundance measurements that will be used to verify the values obtained in this study for the 12 open clusters listed in Table 1.1.

Cluster l b d log(age) diameter Name (deg) (deg) (pc) (yr) (arcmin) IC 4651 340.0881 7.9068 888 9.057 10 IC 4756 36.3807 +5.2422 484 8.699 39 NGC 1662 187.6949 21.1142 437 8.625 20 NGC 2215 215.9932 10.1024 1293 8.369 7 NGC 2354 238.3683 6.7918 4085 8.126 18 NGC2423 230.4835 +3.5368 766 8.867 12 NGC2447 240.0386 +0.1345 1037 8.588 10 NGC 2516 273.8157 15.8558 409 8.052 30 NGC2539 233.7053 +11.1115 1363 8.570 9 NGC2548 227.8724 +15.3928 769 8.557 30 NGC 2682 122.9232 27.0400 908 9.409 25 NGC 3680 124.9390 1.2226 938 9.077 5 NGC5617 317.5264 +2.0851 1533 7.915 10 NGC5822 324.3610 +1.7201 917 8.821 35 NGC6067 127.7404 +2.0870 1417 8.076 14 NGC 6134 335.2223 1.4272 913 8.968 6 NGC 6281 345.2791 3.0564 479 8.497 8 NGC 6405 356.9316 1.5491 487 7.974 20 NGC 6705 15.3951 9.5927 1877 8.302 13

8 Chapter 2

Data Collection and Interpretation

2.1 Victor Blanco Telescope

The data used in this project was collected from the Blanco 4 m telescope, shown in

Figure 2.1, at the Cerro Tololo Inter-American Observatory (CTIO). This observatory is located in Chile just 80 km East of La Serena. Atmospheric interference of light is greatly decreased due to the altitude which is 2200 m or close to 7200 ft. This telescope was designed as a Southern hemisphere equivalent to the 4m telescope being planned for

Kitt Peak National Observatory (KPNO) and was made possible by donations from the

Ford Foundation and the National Science Foundation (NSF). Excavation for the 4 m began in 1967 and in 1976 astronomers began taking the first observations. It was the largest optical telescope in the Southern hemisphere from 1976 until 1998.

The Blanco telescope design is known as a Cassegrain reflector. An example of this type of telescope is shown in Figure 2.2. Light is first collected by the primary parabolic mirror. In this case, it is the 4 m mirror for which the telescope gets part of its name

9 and has a total light collecting area of 10 m2. The light is then reflected to a smaller secondary mirror that has a hyperbolic shape. From here, the light is reflected back through a small hole in the primary mirror that lies in the focal plane. Behind the primary mirror is where the light is then focused. This is also where a plethora of detection instruments are placed. For this project, the Hydra Multi-Fiber Spectrograph was used to collect data.

Figure 2.1: The Blanco 4 m telescope at CTIO in Chile.

Figure 2.2: An example of a Cassegrain telescope. The Blanco 4 m has a similar setup.

10 2.1.1 The Hydra Multi-Fiber Spectrograph

The Blanco 4 m telescope Hydra multi-fiber spectrograph was utilized to collect data for this project. Light directed to the focal point was collected by a fiber positioner mounted behind the primary mirror. This fiber positioner was made of 138 individual movable fibers that could be placed on separate targets, in this case, stars. The Hydra

fiber system is shown in 2.3. The fibers were fed into a spectrograph located in a separate room below the telescope where they simultaneously dispersed onto a 2048 4096 pixel ⇥ CCD using a di↵raction grating that had 1200 lines per mm (Frinchaboy & Majewski

2008). A spectral range of 7740–8740Awascovered.˚

Figure 2.3: The Hydra fiber positioner system that used to be on the Blanco 4 m.

2.1.2 Data Preparation

Once the observations were completed, there were several data reduction steps that needed to be taken in order to obtain important information. IRAF was used to complete all of the following steps. The first was removing the bias level from the CCD. This was

11 done by taking several bias frames, or images that contain no photo or thermally excited electrons. The frames were averaged together to produce a main bias frame which was then subtracted from science images.

Next, pixel-to-pixel variation in the CCD must be removed. This was done using flat

field images. These images are short exposures of a uniform white light source. They were also bias subtracted and then averaged-combined to produce a master flat field image. Finally, the master flat field was also subtracted from the science images.

Then, the wavelength scale was set for all of the science images. This step was accom- plised using the IRAF task dispcor which allowed the beginning and ending wavelengths to be set as well as the wavelength per pixel.

Another correction that had to be performed was the Doppler correction. A velocity with respect to the Sun, also known as a heliocentric velocity, was determined for each target using the task rvcorrect. With this heliocentric velocity, the spectra were shifted back to the appropriate wavelengths.

One of the last steps was to trim distortions at the edges of the science images. This was also done to double check that they all had the same starting and ending wavelength across the same number of pixels. The science images were required to be uniform in shape so they would be compatible with The Cannon, which is explained in more detail in Section 2.3.1.

12 2.1.3 Cluster Membership Determination from (Frinchaboy &

Majewski 2008)

Radial velocity was more sensitive to determining cluster membership because of the small

RV errors (Frinchaboy & Majewski 2008). This velocity describes how fast an object is moving towards or away from an observer along the line of sight. It can be determined by using a Doppler shift in spectra. Stars that are moving away from an observer are

“red shifted” and stars that are moving towards an observer are “blue shifted.”

The RV values were determined using software called the Image Reduction and Anal- ysis Facility (IRAF). The first step was to put all of the stars associated with a cluster into an RV histogram like the one for NGC 2682 shown in Figure 2.4. This distribution was then compared to a similar RV distribution of stars outside the radius of the cluster.

Aprobabilitydistributionwasdeterminedandthenfitwitha1DGaussiantodetermine the membership probability. The steps for this process are shown in Figure 2.5.

Figure 2.4: A histogram from Frinchaboy & Majewski (2008) that shows the radial velocity distribution of all stars with proper motion data that are in the field of the calibration cluster NGC 2682.

13 Figure 2.5: These graphs show the steps in the membership analysis process for NGC 2682 that was done by Frinchaboy & Majewski (2008). (a) shows the result of the radial velocity distribution convolution with a Gaussian kernel to smooth the histogram shown in Figure 2.4, (b) shows the same smoothing process that was done for stars that fall outside of the cluster radius of NGC 2682, (c) shows the probability distribution for the cluster where non-members appear ouside of the main peak, and (d) shows a 1D Gaussian fit to the cluster distribution in (c).

Proper motions were also used to constrain membership probabilities. This motion describes the angular motion of an object across the sky. They are expressed as arcmil-

4 liseconds per year. One arcsecond equals 2.78 10 degrees. Stars that were members of ⇥ aclustersharedcommonpropermotions.Thepropermotionmembershipprobabilities were determined similarly as the RV probabilities except the distributions were in 2D

(Frinchaboy & Majewski 2008).

Figure 2.6 shows the final results of the membership anaysis for observed stars in comparison to all of the stars in the field of NGC 2682. Color-magnitude diagrams

(CMD) are variants of H-R diagrams and are also useful when looking at stars that might be probable members. In Figure 2.6 the KS and J represent the apparent magnitudes of stars in di↵erent filters. The K filter wavelength range is between 1.9 2.3µm,andthe S

14 Jfilterwavelengthrangeisfrom1.1 1.3µm. Apparent magnitude is defined as how bright an object appears from Earth. The y-axis is the KS magnitude, and the x-axis is the di↵erence between J and KS. This di↵erence corresponds to temperature. Figure 2.6 shows the locations of stars that were determined to be likely members for NGC 2682.

Figure 2.6: CMD showing all of the stars within the cluster radius for NGC 2682. The larger black circles are members from both RV and proper motion determinations, crosses are stars that were found to be non-members, triangles are stars with high RV, but no proper motion values and the small grey circles are other non-member stars (Frinchaboy &Majewski2008).

2.2 Spectral Analysis

In order to determine the abundance of heavier elements in stars, we must used measure- ments of spectral lines. Line profiles can be impacted by many di↵erent factors that can occur at the same time. The three main broadening mechanisms are natural, pressure of collisional broadening and Doppler broadening.

15 2.2.1 Line Broadening Mechanisms

2.2.1.1 Natural Line Broadening

Natural line broadening is a result of the Heisenberg uncertainty principle. The orbital energies will not have a single value but will have a spread of energies. So, an electron has the possibility of transitioning from anywhere within the energy spread of the orbitals which means the emitted or absorbed photon has a spread in energy.

For natural line broadening in the case of absorption, the line profile can be determined by considering a classical model for plane electromagnetic waves interacting with dipoles.

The electromagnetic wave, E,mustsatisfythewaveequationandisgivenby

E = E exp i![t (✏/✏ )1/2x/c]. (2.1) 0 0

Next, the displacement x can be determined by using the harmonic oscillator equation with a driving term,

d2x dx e + + !2x = E exp i!t. (2.2) d2t dt 0 m 0

Here, is the damping constant and x will have the form x = x0 exp i!t. After plugging x in and solving, several other manipulations can be performed to find E.

Then, since intensity is proportional to EE⇤,theabsorptioncoecient↵ can be found and is given as

e2 /4⇡ ↵ = . (2.3) mc ⌫2 +(/4⇡)2 ✓ ◆

16 Gamma is the lifetime of the level and can also be written in terms of the Einstein probability coecient

=4⇡ Aul. (2.4) X where u and l are upper and lower levels, respectively. The uncertainty in the upper electron level is defined as

W t h/2⇡ or W 2h A . (2.5) u u ul Xl

So, energy level u has a width of W and the lower level has a similar width. The probability of an electron being in one of the energy bands is given by ↵u or ↵l.The total absorption coecient is a convolution between ↵u and ↵l which produces a larger profile.

2.2.1.2 Collisional/Pressure Broadening

Pressure/collisional broadening occurs when atoms collide and perturb di↵erent orbitals.

The change in orbital separations while an atom is emitting a photon changes the energy of the photon. Spectral line widths from pressure broadening are dependent on the number density of the atoms.

Pressure broadening profiles are very similar to natural line broadening profiles in that it can also be described by a Lorentzian profile. The change in energy as a function

n of separation R is E =constantR where atoms of the same kind will have n =3and atoms of di↵erent kinds will have n =6.Thischangeinenergycanbeconvertedtoa

17 change in frequency, E E = h⌫ or ⌫ = C /Rn,whereC is a constant unique u l n n to each transition. When an atom is emitting a photon and collides with another atom, the phase of the photon is changed. So, if we have several pieces of the original photon that have undergone a phase change, each piece has a di↵erent duration W .Thismeans that each piece will also have a di↵erent ⌫ since ⌫ =1/W .Takingtheweighted average of all of these pieces gives an absorption coecient of

/4⇡ ↵ =constant n . (2.6) (⌫ ⌫ )2 +( /4⇡)2 0 n

Here, the collisional damping constant is dependent on the number density, which can broaden the profile.

2.2.1.3 Doppler Broadening

Doppler broadening is due to the motion of atoms or groups of atoms moving toward or away from an observer. This motion causes the emitted light to be either blue-shifted or red-shifted, respectively. As a result, the line is broadened.

The Doppler broadening profile arises from the motions of the atoms which is de- scribed by a Maxwellian distribution. This Doppler shift is given by

⌫ v = = r . (2.7) ⌫ c

AMaxwelliandistributioncanbeusedtodescribethevelocities,

dN 1 vr = exp dvr. (2.8) N v0p⇡ v0

18 The frequency shift is given by

v ⌫ 2kT 1/2 ⌫ = 0 ⌫ = . (2.9) D c c m ✓ ◆

Finally, the absorption coecient is given by

⇡1/2e2 1 ⌫ 2 ↵d⌫ = f exp . (2.10) mc ⌫ ⌫ D  ✓ D ◆

The distribution of wavelengths or frequencies gives the shape of the absorption co- ecient.

2.2.2 Curve of Growth

The curve of growth is a very useful tool for finding the column density (Na), or the number of atoms of an absorbing element lying above a unit area of the photosphere of a star. The shape of an absorption line is dependent on the number of absorbing elements that are present in a star’s atmosphere. Their strength is measured using the equivalent width. This width is defined as the width of a box that extends to the continuum and has the same area as the spectral line. The equivalent width, W ,changeswiththe column density and has three di↵erent regimes which are shown in 2.7. When there are a small number of absorbing atoms, the curve of growth is initially linear in shape, or

W N . This region is known as the optically thin region and has weak absorption lines. / a Increasing the number of absorbing atoms causes the lines to become optically thicker.

Here the absorption line becomes saturated and bottoms out, but the wings remain

19 optically thin. The equivalent width increases and the curve of growth transitions to a logarithmic regime. This relation is given by W pln N . As the number of atoms / a increases, the pressure broadening becomes more relevant than Doppler broadening and the wings of the line profile become deeper. The curve of growth now has a shape described by W pN .Fromthecurveofgrowthandtheequivalentwidth,wecan / a find the number of absorbing atom. This will allow us to use the Boltzmann and Saha equations to convert this value into the total number of atoms of a particular element above a star’s photosphere.

Figure 2.7: This is an example of curve of growth for the Sun. The dashed lines are separating the three di↵erent regions of the three functional forms (CarrollOstlie).

2.2.3 Boltzmann Equation

The ratio of the probability that a system is in state sb to the probability that the system is in the state sa is given by the expression

20 P (s ) e Eb/kT e (Eb Ea) b = = , (2.11) Ea/kT P (sa) e kT

E/kT where e is the Boltzmann factor and the temperature T is common between the two systems. The energy levels of the system can be degenerate, meaning more than one quantum state can have the same energy. In order to have an accurate count of the number of states, we can define a statistical weight, ga,tobethenumberofstateswith energy Ea.ThesamecanbedefinedforthestateswithenergyEb.Theratioofthe probabilities with degeneracy taken into account becomes

P (E ) g e Eb/kT g e (Eb Ea) b = b = b . (2.12) Ea/kT P (Ea) gae ga kT

This ratio of probabilities can also be expressed as the ratio of the number of atoms with a given energy in di↵erent states of excitation, which is given by

N g e Eb/kT g e (Eb Ea) b = b = b . (2.13) Ea/kT Na gae ga kT

This is the Boltzmann equation. The atmosphere of a star is not actually in thermody- namic equilibrium, however, a local thermodynamic equilibrium (LTE) can be defined in order for the Boltzmann equation to be valid.

2.2.4 Saha Equation

Next, the number of atoms in di↵erent stages of ionization need to be considered. This involves the partition function which is the weighted sum of the number of ways an atom

21 can arrange its electrons with the same energy. The configurations that are less probable, the ones with higher energy, are weighted less from the Boltzmann factor when taking the sum. The partition function is given by

1 (Ej E1)/kT Z = gje . (2.14) j=1 X Using the partition functions for the atom in initial state i and final stage of ionization i +1, the ratio of each of the stages can be written as

3/2 Ni+1 2Zi+1 2⇡mekT /kT = e i . (2.15) N n Z h2 i e i ✓ ◆

The number density of free electrons, ne,canbeexpressedasthepressureofthefree electrons since the two are related by the ideal gas law Pe = nekT.TheSahaequation is also only valid for LTE.

2.3 Important Stellar Parameters

There are three main stellar parameters that a↵ect spectral line profiles e↵ective tem- perature (Teff ), surface gravity (log g), and chemical abundance (metallicity). They are unequivocally the most important qualities used to describe the overall behaviour of the spectral flux of stars. These parameters are described in detail below.

22 2.3.1 E↵ective Temperature

The e↵ective temperature is the temperature a star would if it were to emit the same total amount of radiation as a black body. E↵ective temperature can be derived by starting with the Stefan-Boltzmann equation

L = AT 4. (2.16)

Here, L is luminosity, A is area, and T is temperature. As the temperature of a blackbody increases, it emits more energy per second at all wavelengths. For a star, this equation becomes

2 4 L =4⇡R Teff , (2.17)

and R is the radius of the star and Teff is the e↵ective temperature. Stars are not perfect blackbodies. The blackbody curve comes from the interior layer of the star that has an optical depth of 2/3. The deviation from the blackbody is caused the outer layer of the atmosphere that is much less dense than the interior.

2.3.2 Surface Gravity

The second important stellar parameter The Cannon requires is surface gravity. Units for surface gravity are commonly expressed as log[g/(cm/s2)] and values for log g can be obtained using the relation

23 GM g = , (2.18) R2 which requires the mass, M,andradius,R, of the star. Here, G is the universal grav- itational constant. Using the Stefan-Boltzmann equation and the magnitude relation discussed in section 1.4, it is possible to express the surface gravity in terms of e↵ective temperature and absolute magnitude. The e↵ect surface gravity has on spectral lines is shown in 2.8 and a↵ects lines by pressure broadening them.

Figure 2.8: An example of how surface gravity impacts the strength of lines. The surface gravity is lowest at the top spectrum and highest at the bottom spectrum. As surface gravity increases, the pressure broadening wings increase.

2.3.3 Metallicity

The third extremely important stellar parameter obtained from The Cannon is the chem- ical abundance or metallicity. It is a measure of the proportion of elements heavier than

24 hydrogen or helium a star possesses. This proportion is commonly expressed as the ratio of iron to hydrogen, or [Fe/H]. The expression is given below as

NFe NFe [Fe/H] log10 log10 . (2.19) ⌘ NH ? NH ✓ ◆ ✓ ◆

The indicates the value for the Sun and the ? indicates the value for the star in question. This ratio can be determined by examining iron absorption lines and then comparing to the known solar iron abundance. The unit used to describe metallicity is referred to as the dex. Stars with a dex of 0 have metallicities comparable to the Sun. For a dex of

+1, stars have a metallicity 10 times greater than that of the Sun, and for a dex of 1, it is 1/10th of the Sun’s value. Iron is not the only metallicity that is useful. Another common metallicity expression is the abundance of alpha elements. These abundances are commonly expressed as [↵/Fe].

2.3.4 Stellar Parameters and The Cannon

Many surveys use the curve of growth, the Boltzmann and Saha equations, and stellar models to determine chemical abundances. The models, however, usually do not account for many distinctive factors in a star’s atmosphere and research groups can get opposing values for the same star (Ness et al. 2015). The Cannon, developed by Ness et al. (2015), o↵ers a unique way to find stellar parameters without having to use any models. Instead, it takes a subset of stars with known parameters or “labels” and creates a model. This model can be applied to the rest of the set of stars to infer labels for them (Ness et al.

2015). For this survey, the set of parameters used were Teff ,logg, and [Fe/H].

25 The open clusters in this survey were selected from a previous study by Frinchaboy &

Majewski (2008) that collected medium resolution (R 15,000) spectra and determined ⇠ probable members. Two clusters with APOGEE observations as well as observations from the Sagittarius dwarf galaxy were used to get consistent measurements for the total sample. The Sagittarius stars and their chemical abundances are discussed in detail in a study by Hasselquist et al. (2017).

The “labels” used for this study were taken from the Sloan Digital Sky Survey (SDSS)

IV/ APOGEE DR14 (see Majewski et al. 2017, Blanton et al. 2017, Eisenstein et al. 2011,

Abolfathi et al. 2018, Garc´ıa P´erez et al. 2016, Nidever et al. 2015 for more information), which observed the same stars as our CTIO/Hydra-based study. This APOGEE data are high resolution (R 22,500), high signal-to-noise (S/N > 100) spectroscopic survey that ⇠ includes over 200,000 observations in a wavelength range of 1.51–1.70 µm. APOGEE

DR14 was released in 2017 August and included a re-reduction and reanalysis of the original APOGEE data, as well as the first 2 years of APOGEE-2 data (Holtzman et al.

2018). The APOGEE Stellar Parameters and Chemical Abundances Pipeline (ASP-

CAP) automatically derived stellar parameters and chemical abundances for all stars in

APOGEE (Holtzman et al. 2018).

26 Chapter 3

Results

3.1 The Cannon

The Cannon is a data-driven approach developed by Ness et al. (2015) for determining stellar parameters and chemical abundances for spectroscopic surveys. It does not use any models of stellar spectra but instead uses machine learning, which means that The

Cannon learns to identify patterns and build a model based on a set of known values.

These values, also referred to as labels, were e↵ective temperature (Teff ), surface gravity

(log g), and chemical abundance ([Fe/H]) are part of what is called the training set. The

Cannon looks for pixel-to-pixel correlations between this set of labels in the training set and builds a 3D model space (Ness et al. 2015). Other spectra are then analyzed using

The Cannon where the pixel space of the spectra are compared to the model space.

It was named after the famous astronomer Annie Jump Cannon because she similarly classified stellar spectra without knowing the physics involved in stellar atmospheres.

27 3.1.1 Training Set

Before The Cannon could be used on the open cluster sample, it had to be trained

with a set of reference spectra with known values. The stars in this reference set were

chosen based on which ones had stellar parameters determined in APOGEE DR14. This

included stars from two open clusters, NGC 2682 and NGC 6705, as well as observations

of stars in the Sagittarius dwarf galaxy. When the reference set was used to train The

Cannon, a check had to be performed to determine if the output values made sense.

This check was done by creating three one-to-one plots for each of the parameters and is

discussed further in the next section. Values obtained from The Cannon for all 106 stars

of the reference set are listed in Table 3.1 along with the values from APOGEE DR14.

Once The Cannon was trained, the next step was to determine values for the stars in

the set of 31 open clusters.

Table 3.1: Values for Teff ,logg, and [Fe/H] from APOGEE compared to The Cannon output for the training set.

APOGEE The Cannon

2MASS Teff log g [Fe/H] Teff log g [Fe/H] ID (K) (dex) (dex) (K) (dex) (dex)

2M08512898+1150330 4743 2.44 +0.10 4787 2.48 +0.10 2M08514355+1144264 4722 2.59 +0.10 4772 2.43 +0.11 2M08514235+1151230 4779 2.79 +0.07 4761 2.77 +0.10 2M08511704+1150464 4758 2.71 +0.05 4739 2.40 +0.10 2M08521097+1131491 4650 2.56 +0.11 4739 2.41 +0.10 2M08512280+1148016 4798 2.45 +0.12 4780 2.82 +0.11 2M08522636+1141277 4963 3.26 +0.02 4773 2.81 +0.08 2M08504964+1135089 4783 2.81 +0.07 4732 2.71 +0.09 2M08513938+1151456 4943 3.18 +0.09 4771 2.80 +0.08 2M08512377+1149493 5063 3.03 +0.01 4741 2.40 +0.08 2M08514234+1150076 4826 2.99 +0.09 4765 2.78 +0.10

Continued on next page

28 Table 3.1 – continued from previous page

APOGEE The Cannon

2MASS Teff log g [Fe/H] Teff log g [Fe/H] ID (K) (dex) (dex) (K) (dex) (dex)

2M08511269+1152423 4806 2.49 +0.08 4765 2.79 +0.11 2M18510092-0614564 4853 2.51 +0.15 4766 2.79 +0.11 2M18542283-3051089 4197 1.29 0.50 4630 2.16 0.23 2M18543218-3034025 3965 1.11 0.08 4776 2.46 0.03 2M18555179-3030590 4829 2.11 1.08 4772 2.46 1.08 2M18544217-3054051 4006 1.01 0.30 4637 2.17 0.18 2M18562522-3040100 3945 0.83 0.43 4689 2.26 0.27 2M18545671-3031258 4083 1.39 0.07 3788 0.40 +0.06 2M18553079-3028198 4073 1.20 +0.02 4659 2.21 +0.04 2M18562750-3038165 4140 1.40 0.25 4772 2.45 0.28 2M18555553-3018239 3864 0.67 0.13 4665 2.28 0.11 2M18564050-3031153 4024 1.11 0.37 4750 2.41 0.35 2M18565708-3036367 3906 0.71 0.65 4771 2.45 0.54 2M18551334-3052052 4000 0.88 0.31 4759 2.43 0.21 2M18552120-3056411 3921 0.78 0.29 4672 2.32 0.28 2M18542596-3041140 3958 0.76 0.22 4760 2.44 0.19 2M18560149-3027248 3872 0.57 0.44 4789 2.48 0.28 2M18545206-3039058 3771 0.64 0.20 4802 2.46 0.24 2M18540214-3036218 4077 1.01 0.59 4790 2.48 0.54 2M18562652-3029160 4130 1.04 0.38 4804 2.50 0.41 2M18543812-3036116 3966 1.00 0.11 4832 2.55 0.13 2M18562768-3022343 3890 0.55 0.42 4819 2.53 0.30 2M18554423-3034322 3942 1.03 0.14 4821 2.53 0.11 2M18553699-3047403 4104 1.25 0.09 4816 2.54 0.20 2M18545758-3019388 3791 0.38 0.26 4819 2.52 0.16 2M18561757-3043094 3910 1.20 0.30 4829 2.54 0.12 2M18565918-3033127 4337 1.54 0.53 4834 2.54 0.50 2M18542499-3031568 4183 1.25 0.85 4833 2.55 +0.00 2M18544834-3026384 3935 0.77 0.09 4835 2.56 0.08 2M18541366-3033271 4079 1.28 0.18 4836 2.54 0.18 2M18561890-3051098 4052 1.18 0.21 4825 2.53 0.27 2M18545718-3024386 3948 0.95 0.20 4834 2.54 0.10 2M18551047-3053278 3730 0.41 0.54 4820 2.53 0.43 2M18561105-3031111 3984 0.96 0.18 4778 2.44 0.17 2M18551842-3025181 4761 2.02 0.89 4784 2.46 0.86 2M18550098-3048006 3896 0.74 0.04 4819 2.55 0.06 2M18555143-3052577 3729 0.44 0.36 4782 2.46 0.43 2M18544264-3022353 4117 0.98 0.80 4779 2.46 0.86 2M18555503-3035481 4104 1.14 0.20 4804 2.49 0.47 2M18561204-3028225 3882 0.63 0.63 4815 2.52 0.58 Continued on next page

29 Table 3.1 – continued from previous page

APOGEE The Cannon

2MASS Teff log g [Fe/H] Teff log g [Fe/H] ID (K) (dex) (dex) (K) (dex) (dex)

2M18541666-3036019 3991 1.03 0.23 4806 2.50 0.31 2M18544046-3026498 3980 1.03 0.16 4809 2.51 0.19 2M18550374-3055014 3742 0.36 0.25 4774 2.44 0.21 2M18562183-3044144 4343 1.69 0.14 4794 2.48 0.35 2M18564778-3040306 3879 0.81 0.10 4823 2.55 0.14 2M18553412-3020551 3820 0.48 0.48 4788 2.47 0.53 2M18563475-3034037 3974 0.81 0.40 4795 2.50 0.46 2M18560782-3048289 3723 0.40 0.43 4809 2.52 0.44 2M18553108-3051582 3913 0.64 0.54 4798 2.49 0.54 2M18544287-3050556 4105 1.31 0.50 4798 2.49 0.15 2M18562154-3047240 4175 1.37 0.30 4816 2.53 0.39 2M18564729-3027016 3755 0.37 0.27 4799 2.49 0.33 2M18554509-3052455 4085 1.24 0.34 4784 2.47 0.37 2M18544441-3035355 3763 0.64 0.34 4774 2.44 0.38 2M18565257-3042207 3721 0.46 0.30 4808 2.51 0.30 2M18543231-3031336 3894 0.80 0.07 4791 2.49 0.05 2M18383170-2923417 3820 0.48 0.22 4776 2.45 0.29 2M18363253-2923581 3936 0.76 0.37 4826 2.52 0.51 2M18365956-2925366 3849 0.61 0.15 4813 2.50 0.34 2M18375484-2933161 3859 0.23 0.43 4790 2.47 0.45 2M18360489-2938083 3756 0.38 0.30 4785 2.47 0.34 2M18380772-2928068 3780 0.18 0.51 4786 2.46 0.56 2M18460108-2956265 3753 0.42 0.25 4790 2.46 0.32 2M18472852-3018578 3991 0.94 0.01 4812 2.49 0.26 2M18455694-3007349 3955 0.59 0.59 4779 2.46 0.58 2M18474337-2957093 3755 0.37 0.42 4805 2.48 0.37 2M18453128-3020484 3770 0.60 0.37 4764 2.42 0.35 2M18453460-3014528 3801 0.50 0.27 4823 2.51 0.35 2M18474736-3015254 3785 0.42 0.28 4816 2.50 0.32 2M18451363-2959321 3921 0.59 0.42 4794 2.49 0.53 2M18460205-2958267 3788 0.41 0.28 4792 2.46 0.31 2M18472455-3022578 3918 0.63 0.38 4764 2.42 0.48 2M18454995-3014142 4012 0.97 0.16 4736 2.31 0.33 2M18461852-3007565 3853 0.71 0.24 4794 2.49 0.35 2M18464088-3024518 3733 0.40 0.33 4790 2.47 0.34 2M18460165-3010065 3942 0.97 0.17 4649 2.16 0.21 2M18454526-3001311 3830 0.53 0.18 4805 2.51 0.26 2M18455099-3004524 4152 1.04 0.99 4835 2.55 +0.24 2M18462085-3000598 3983 0.90 0.38 4824 2.53 0.41 2M18565793-2829319 3846 0.66 0.22 4799 2.49 0.37 Continued on next page

30 Table 3.1 – continued from previous page

APOGEE The Cannon

2MASS Teff log g [Fe/H] Teff log g [Fe/H] ID (K) (dex) (dex) (K) (dex) (dex)

2M18573014-2837577 3961 0.93 0.42 4791 2.49 0.44 2M18522151-2841088 3983 0.95 0.64 4805 2.51 0.66 2M18515121-2849451 3824 0.60 0.37 4780 2.45 0.42 2M18520026-2833035 3774 0.47 0.47 4796 2.44 0.46 2M18512180-2821559 3819 0.52 0.28 4802 2.52 0.36 2M18504212-2857267 3997 1.04 0.26 4845 2.95 0.41 2M18504871-2826308 3886 0.60 0.35 4808 2.55 0.37 2M18513010-2830552 3875 0.81 0.09 4808 2.53 0.24 2M18514572-2831138 3930 0.90 0.23 4805 2.52 0.42 2M18510625-2842547 3789 0.67 0.29 4825 2.51 0.41 2M18503032-2855221 3844 0.14 0.78 4817 2.50 0.68 2M18552395-3131338 3752 0.44 0.62 4795 2.48 0.52 2M18531136-3109573 3785 0.66 0.28 4797 2.50 0.39 2M18555355-2954274 3751 0.18 0.47 4837 2.52 0.51

3.1.2 Individual Stat Results and Data Quality

One-to-one plots allowed us to check the quality of the results from The Cannon.Dis- played in Figure 3.1, these plots showed that the scatter for Teff ,logg, and [Fe/H] was

141.19, 0.29, and 0.17, respectively. The overall scatter for each “label” is an average on how far the values for each star are away from the on-to-one line. So, if all of the di↵erences between values from APOGEE and The Cannon are small, the overall scatter is small as well. The scatter for each “label” was then used as errors for the rest of the data set.

There are several outliers in the [Fe/H] one-to-one plot that might have contributed to the higher scatter. We also did not have many stars with more metal-rich values,

31 greater than the Sun'sFe/Hvalueof0.0,whichcouldhavegivenahigherscatter.The cause of the discrepancies will be explored in the future.

(a) Teff one-to-one plot. (b) Log g one-to-one plot.

(c) [Fe/H] one-to-one plot.

Figure 3.1: One-to-one plots showing a comparison between the values obtained using The Cannon and the APOGEE DR14 labels for the training set.

32 3.1.3 Verification of Cluster Membership and Bulk Cluster Met-

allities

Before we determined average Fe/H for each cluster, two additional membership critera were examined. First, the dwarf stars had to be identified and removed. They were removed because most surveys only use giant stars since they are more luminous and therefore provide better spectra. Dwarf stars also tend to have less metals than giants since they do not have the same fusion processes occurring in their cores. Stars with

KS magnitudes greater than 10.0 and J-KS values between 0.5 and 1.0 were identified as giant stars.

Once the giant stars were identified, Fe/H values were compared between each star in a given cluster. If values were outside a one sigma range based on the 0.17 dex scatter, they were removed. This resulted in three stars being excluded from average

Fe/H determinations. The remaining stars’ Fe/H values were averaged to determine an an overall value for each cluster. The errors for each average were done using standard error propagation.

3.2 Total Cluster Sample

The next step after obtaining values from The Cannon for the open cluster sample was to check against the ones with literature values. These clusters were displayed in Table

1.2. Three well known spectroscopic surveys with high-resolution data were compared to first. They were the Santos et al. (2009), Reddy et al. (2013), and Netopil et

33 al. (2016) surveys. After these surveys were examined individually, we looked at the remaining studies that covered a wide resolution range and also included photometric studies. Finally, the cluster NGC 2682 was compared to a survey by Donor et al. 2018 that determined abundances using APOGEE data.

All Fe/H values for individual stars are in Appendix A Table 4.1. Several stars were determined to be non-members based on a median metallicity cut. For Collinder 258, one of the stars was removed based on its metallicity and the probabilities determined in the study Frinchaboy & Majewski (2008). The cluster NGC 3680 had members removed based on comparisons to the literature.

34 Chapter 4

Discussion

4.1 Summary

The goal of this survey was to get a uniform sample of metallicities for a set of 31 open clusters. Included in this set were 12 clusters that did not have metal abundances determined before. We also re-determined metallicites for clusters with known values in order to determine if the values The Cannon provided were accurate. Of the clusters with known data, only 4 had inconsistencies. Two were explained by the di↵erence in data used between surveys. The others will be explored further to determine why there were di↵erences.

35 Table 4.1: Average in common open cluster iron abundance from Santos et al. (2009) compared to ( this study).

This Study Santos Cluster Number [Fe/H] Number [Fe/H] Name of Stars (dex) of Stars (dex) IC 4651 8 +0.03 0.06 3 +0.09 0.01 ± ± NGC 2423 7 +0.09 0.06 3 +0.09 0.06 ± ± NGC 2447 8 +0.10 0.06 3 0.03 0.03 ± ± NGC 2539 2 +0.11 0.12 3 +0.08 0.03 ± ± NGC 2682 10 +0.10 0.05 3 +0.02 0.01 ± ±

4.1.1 Comparison to Other Surveys

4.1.1.1 Santos et al. (2009) Survey

The Santos et al. (2009) survey provided iron abundances for 13 open clusters using high-resolution spectra. There were 5 clusters that overlapped with this survey, and they are listed in Table 4.1. Average iron abundances from this survey and Santos et al.

(2009) are also shown. To better illustrate how the two surveys compare, we constructed aplotofvaluesfromThe Cannon vs. values from Santos et al. (2009) which is shown in Figure 4.3. The light blue line indicates one-to-one values, and the grey dashed lines show the 0.20 scatter for [Fe/H[. All of the clusters lie within the scatter which means the measurements between this survey and the Santos et al. (2009) are consistent.

36 Figure 4.1: A one-to-one comparison of [Fe/H] values for open clusters in common be- tween Santos et al. (2009) and this study. The grey shaded area indicates the 0.17 scatter above and below the blue one-to-one line. ±

4.1.1.2 Reddy et al. (2013) Surveys

The next comparison was to the Reddy et al. (2013) and Reddy et al. (2015) studies.

Both examined a total of 12 clusters using high-resolution spectra. Here, there was also an overlap of 5 clusters between this survey and the combined Reddy surveys. The values for each are shown in Table 4.2. A one-to-one plot, shown in Figure 4.2, was also constructed to see how the data agreed. The values obtained by Reddy et al. (2013; 2015) were more metal-poor than the values determined using The Cannon.Thereis,however, a well documented metal-poor trend for these studies. Donor et al. (2018) observed this o↵set when compared to their study as well as multiple other studies, meaning the values for these clusters are consistent.

37 Table 4.2: Average in common open cluster iron abundance comparison between Reddy et al. (2013; 2015) and this study.

This Study Reddy Cluster Number [Fe/H] Number [Fe/H] Name of Stars (dex) of Stars (dex) NGC 1662 2 +0.07 0.12 2 0.10 0.06 ± ± NGC 2354 2 +0.11 0.12 2 0.19 0.04 ± ± NGC 2447 8 +0.10 0.06 3 0.13 0.05 ± ± NGC 2539 2 +0.11 0.12 2 0.60 0.04 ± ± NGC 2682 10 +0.10 0.05 3 0.08 0.04 ± ±

Figure 4.2: A one-to-one comparison of average [Fe/H] values obtained by Reddy et al. (2013; 2015) and this study for the open clusters in common.

4.1.1.3 Netopil et al. (2016) Survey

The last spectroscopic survey compared to was by Netopil et al. (2016). They examined

172 clusters with a variety of data including low, medium, and high-resolution spectra as well as photometric data. The 12 clusters that overlapped had iron abundances deter-

38 mined from high-resolution spectra. All of the iron abundances are listed in Table 4.3.

The one-to-one plot in Figure 4.3 shows that the majority of the iron abundances are consistent and most clusters lie within the scatter range from The Cannon.Thereare two outliers which are discussed in Section 3.2.2.

Table 4.3: Average iron abundances for open clusters in common between Netopil et al. (2016) and this study.

This Study Netopil Cluster Number [Fe/H] Number [Fe/H] Name of Stars (dex) of Stars (dex) IC 4651 8 +0.03 0.06 18 +0.12 0.04 ± ± IC 4756 5 +0.07 0.08 15 +0.02 0.04 ± ± NGC 1662 2 +0.07 0.12 2 +0.00 0.08 ± ± NGC 2354 2 +0.11 0.12 2 0.18 0.02 ± ± NGC 2423 7 +0.09 0.06 3 +0.08 0.05 ± ± NGC 2447 8 +0.10 0.06 3 +0.07 0.03 ± ± NGC 2516 2 +0.09 0.12 2 +0.05 0.11 ± ± NGC 2682 10 +0.10 0.05 28 +0.03 0.05 ± ± NGC 3680 6 0.34 0.07 10 0.01 0.06 ± ± NGC 6134 7 +0.08 0.06 8 +0.11 0.07 ± ± NGC 6281 2 +0.10 0.12 2 +0.06 0.06 ± ± NGC 6705 1 +0.06 0.17 21 +0.12 0.09 ± ±

4.1.1.4 Other Surveys

The remaining clusters with iron abundances from smaller studies were also examined for inconsistencies. Table 4.4 lists all of the values for [Fe/H] determined by The Cannon,the

[Fe/H] values from the literature, and the type of data that was used. Most of the studies listed also used high-resolution spectra. There were several that used photometric data and one that used low-resolution spectra and one that used medium resolution spectra.

Another one-to-one plot, Figure 4.4, was made to look for any outliers. Most of the

39 Figure 4.3: A comparison of [Fe/H] values for open clusters in common from Netopil et al. (2016) and this study. clusters fall within the range of scatter again; however, 3 clusters were not. Reasons for why these values do not appear to agree are given in the next section.

40 Table 4.4: Average open cluster iron abundance for open clusters in common between this study and other literature studies.

This Study Other Studies Cluster Stars [Fe/H] Stars [Fe/H] Typea Citation Name (dex) (dex) IC 4651 8 +0.03 0.06 4 +0.11 0.01 HS Carretta et al. (2004) IC 4756 5 +0.07 ± 0.08 9 0.02 ± 0.01 HS Bagdonas et al. (2018) ± 12 0.01 ± 0.10 HS Ting et al. (2012) 2 +0.01± HS Pace et al. (2010) NGC 2215 1 +0.09 0.17 51 0.40 0.10 P Fitzgerald et al. (2015) NGC 2548 4 +0.08 ± 0.09 91 0.24 ± 0.27 P Balaguer-N´u˜nez et al. (2005) NGC 2682 10 +0.10 ± 0.05 0.08 ± 0.04 LS Chen & Gao (2011) NGC 5617 2 +0.08 ± 0.12 2 0.18 ± 0.02 HS De Silva et al. (2015) NGC 5822 9 +0.09 ± 0.06 5 +0.05± HS Pace et al. (2010) ± 61 0.02 0.02 P Carraro et al. (2011) NGC 6067 6 +0.08 0.07 5 +0.19 ± 0.05 HS Alonso-Santiago et al. (2017) NGC 6134 7 +0.08 ± 0.06 6 +0.15 ± 0.03 HS Carretta et al. (2004) NGC 6405 2 +0.08 ± 0.12 44 +0.07 ± 0.03 MS Kılı¸co˘glu et al. (2016) NGC 6705 1 +0.07 ± 0.17 21 +0.10 ± 0.06 P Cantat-Gaudin et al. (2014) ± ± a HS stands for high resolution spectroscopy, MS stands for medium resolution spectroscopy, LS stands for low resolution spectroscopy, and P stands for photometry.

Figure 4.4: A comparison of average [Fe/H] values for in common open clusters from this survey and from the literature compilation in Table 4.4. Blue stars are [Fe/H] averages for open clusters IC 4756 and orange diamonds are [Fe/H] averages for open cluster NGC 5822.

41 4.1.1.5 NGC 2682

The cluster NGC 2682 is one of the most well known open clusters. That is why it was used as a calibration cluster in this survey. It also has an average [Fe/H] value determined from APOGEE DR14 data from Donor et al. (2018) making it a significant check on how well The Cannon produced values. The average [Fe/H] from Donor et al.

(2018) was 0.07 0.03, and this survey found a value of 0.10 0.05 which was well within ± ± the scatter.

4.1.2 Discrepancies

There were 4 clusters that lie outside of the acceptable scatter in the one-to-one plots.

This section discusses possible reasons why there were di↵erences in iron abundance values.

4.1.2.1 NGC 2215

The value of 0.08 0.10 obtained from The Cannon is significantly di↵erent from the ± metal-poor value of -0.40 0.10 given by Fitzgerald et al. (2015). The most probable ± reason for the disparity is the di↵erence in the type of survey. Photometric data were used in Fitzgerald et al. (2015) which is similar to using low-resolution spectroscopic observations. This method also uses models to fit color-color diagrams, comparisons of stellar magnitudes in di↵erent wavelength regions, which could be biased towards a certain metallicity.

42 4.1.2.2 NGC 2354

This cluster was one of two outliers from the Netopil et al. (2016) survey. One cause could be a problem associated with the stars used in this cluster that went through The

Cannon since there was a broader spread in [Fe/H] values that ranged from 0.07 0.20 ± to 0.11 0.20. ±

4.1.2.3 NGC 2548

The literature value for NGC 2548 was also determined using photometric data. There could be similar reasons for why the values do not agree such as the low-resolution similarities and the models used by Balaguer-N´u˜nez et al. (2005).

4.1.2.4 NGC 3680

NGC 3680 was the second cluster that was an outlier when comparing to the Netopil et al. (2016) survey. The metallicities determined for all of the stars that were members in this cluster had an even number of metal-poor and metal-rich values. The giant stars all had metal-poor values whereas the dwarf stars had metal rich values. One reason for this split could be the membership analysis. If the metal-poor giants are determined to not be members, then the remaining dwarf stars can not be compared to Netopil et al.

(2016) and this cluster will be removed from the sample.

4.1.2.5 NGC 5617

The survey by De Silva et al. (2015) used high-resolution spectra to find a value for

[Fe/H]. The di↵erences in values for this cluster could be due to how De Silva et al. (2015)

43 Table 4.5: Average open cluster iron abundance for clusters with no prior measurements.

Cluster No. of Stars [Fe/H] Name This Study (dex) Collinder 205 1 +0.08 0.17 Collinder 258 1 +0.08 ± 0.17 NGC 2437 3 +0.11 ± 0.10 NGC 2546 1 +0.08 ± 0.17 NGC 2579 2 +0.10 ± 0.12 NGC 2669 1 +0.09 ± 0.17 NGC 5281 1 0.56 ± 0.17 NGC 6124 3 +0.03 ± 0.10 NGC 6167 2 +0.05 ± 0.12 NGC 6250 1 +0.07 ± 0.17 NGC 6885 2 0.38 ± 0.12 Ruprecht 119 1 +0.06 ± 0.17 ± carried out chemical abundance measurements. They used a code known as MOOG that incorporates model atmospheres based on more than just the three parameters used by

The Cannon.

4.1.3 New Values

Since the open clusters with known values agree with most of the literature, the values that were obtained for the clusters that did not have previous measurements have a higher probability of being valid. Their [Fe/H] values are shown in Table 4.5

4.2 Future Work

There are several ways this project can go from here. Since the metallicites for these clusters were all determined based on APOGEE DR14 data, they can be combined with

44 the clusters from Donor et al. (2018) to create a large uniform sample. With this larger sample, we could look at di↵erent trends in the Milky Way in better detail.

There are an additional 22 open clusters that have observations, but not known memberships for those observations. Membership determination could be done using radial velocity and proper motion probability analysis described in Section 1.3.2. The recent Gaia data release has great proper motion measurements which would greatly improve the confidence with which we suspect targets are members or not.

45 Appendix A

The Cluster Sample

Table A.1: The output from The Cannon for every star in the open cluster sample. The errors for Teff ,logg, and [Fe/H] are 141 K, 0.29 dex, and 0.17 dex, respectively. Cluster members used in the average Fe/H± cluster± determination± are indicated by a “Y” in the last column and stars that were either non-members or dwarf stars are indicated by “N”.

ID Cluster Teff log g [Fe/H] Member

09001428-4901040 Collinder 205 4787 2.48 +0.08 Y

12291805-6031347 Collinder 258 3838 0.42 0.53 N 12265230-6047561 Collinder 258 4773 2.43 +0.08 Y

17245416-4953075 IC 4651 4761 2.77 +0.02 Y

17261801-4952535 IC 4651 4739 2.40 +0.07 N

17245259-4956313 IC 4651 4739 2.41 +0.06 N

17252437-4958525 IC 4651 4780 2.82 +0.03 Y

17243568-4956273 IC 4651 4773 2.81 +0.03 Y

17250895-4953571 IC 4651 4732 2.71 +0.03 Y

46 17252354-4955470 IC 4651 4771 2.80 +0.03 Y

17234787-4947221 IC 4651 4741 2.40 +0.06 N

17245776-5001327 IC 4651 4765 2.78 +0.03 Y

17260141-4954283 IC 4651 4765 2.79 +0.03 Y

17251321-4959293 IC 4651 4766 2.79 +0.03 Y

18385292+0520165 IC 4756 4630 2.16 +0.07 Y

18380662+0519502 IC 4756 4776 2.46 +0.06 N

18382656+0509501 IC 4756 4772 2.46 +0.06 N

18384377+0514200 IC 4756 4637 2.17 +0.07 Y

18382075+0526024 IC 4756 4689 2.26 +0.07 Y

18373582+0515379 IC 4756 3788 0.40 0.43 Y 18380515+0524337 IC 4756 4659 2.21 +0.08 Y

18380876+0520554 IC 4756 4772 2.45 +0.07 Y

04482950+1055482 NGC1662 4665 2.28 +0.06 Y

04474415+1106572 NGC1662 4750 2.41 +0.06 N

04480658+1048553 NGC1662 4771 2.45 +0.07 N

04482517+1047057 NGC1662 4759 2.43 +0.07 N

04483208+1057589 NGC1662 4672 2.32 +0.07 Y

04482904+1100117 NGC1662 4760 2.44 +0.07 N

06210343-0719085 NGC2215 4789 2.48 +0.08 N

06204686-0717029 NGC2215 4802 2.46 +0.09 Y

07142988-2552101 NGC2354 4804 2.50 +0.08 N

47 07135193-2544242 NGC2354 4832 2.55 +0.11 Y

07133035-2543342 NGC2354 4819 2.53 +0.08 N

07140636-2536449 NGC2354 4821 2.53 +0.08 N

07132188-2527005 NGC2354 4816 2.54 +0.07 N

07132374-2533462 NGC2354 4819 2.52 +0.10 Y

07372858-1355585 NGC2423 4829 2.54 +0.08 Y

07380534-1404073 NGC2423 4834 2.54 +0.08 Y

07371259-1344553 NGC2423 4833 2.55 +0.08 Y

07374842-1335074 NGC2423 4835 2.56 +0.09 Y

07364647-1350306 NGC2423 4836 2.54 +0.09 Y

07370026-1345381 NGC2423 4825 2.53 +0.09 Y

07361592-1403461 NGC2423 4834 2.54 +0.09 Y

07373533-1357296 NGC2423 4820 2.53 +0.07 N

07421614-1449443 NGC2437 4778 2.44 +0.08 N

07414072-1454283 NGC2437 4784 2.46 +0.08 N

07413760-1443129 NGC2437 4819 2.55 +0.12 Y

07405803-1451026 NGC2437 4782 2.46 +0.08 N

07413200-1451484 NGC2437 4779 2.46 +0.08 N

07411127-1452280 NGC2437 4804 2.49 +0.08 N

07421014-1438511 NGC2437 4815 2.52 +0.07 N

07414194-1447158 NGC2437 4806 2.50 +0.08 N

07420012-1451449 NGC2437 4809 2.51 +0.07 N

48 07415842-1449247 NGC2437 4774 2.44 +0.09 N

07422015-1438296 NGC2437 4794 2.48 +0.09 N

07422286-1442555 NGC2437 4823 2.55 +0.11 Y

07415256-1452141 NGC2437 4788 2.47 +0.08 N

07410541-1445518 NGC2437 4795 2.50 +0.08 N

07414693-1440371 NGC2437 4809 2.52 +0.07 N

07414286-1457022 NGC2437 4798 2.49 +0.09 N

07410834-1451274 NGC2437 4798 2.49 +0.08 N

07413811-1452433 NGC2437 4816 2.53 +0.07 N

07420592-1444335 NGC2437 4799 2.49 +0.08 N

07410253-1447474 NGC2437 4784 2.47 +0.08 N

07412289-1449109 NGC2437 4774 2.44 +0.09 N

07424124-1459514 NGC2437 4808 2.51 +0.11 Y

07413129-1438220 NGC2437 4791 2.49 +0.06 N

07443507-2351416 NGC2447 4776 2.45 +0.09 N

07442573-2349529 NGC2447 4826 2.52 +0.10 Y

07445380-2353151 NGC2447 4813 2.50 +0.10 Y

07444196-2350167 NGC2447 4790 2.47 +0.09 N

07450246-2351137 NGC2447 4785 2.47 +0.09 N

07444142-2351037 NGC2447 4786 2.46 +0.09 N

07443281-2357504 NGC2447 4790 2.46 +0.10 Y

07441844-2353236 NGC2447 4812 2.49 +0.11 Y

49 07441545-2350327 NGC2447 4779 2.46 +0.08 N

07450947-2349325 NGC2447 4805 2.48 +0.10 Y

07442803-2342458 NGC2447 4764 2.42 +0.10 N

07444418-2347222 NGC2447 4823 2.51 +0.09 Y

07435090-2341436 NGC2447 4816 2.50 +0.10 Y

07443373-2348488 NGC2447 4794 2.49 +0.08 N

07442457-2350440 NGC2447 4792 2.46 +0.09 Y

07574523-6055353 NGC2516 4764 2.42 +0.09 N

07561863-6031352 NGC2516 4736 2.31 +0.09 Y

07595779-6044484 NGC2516 4794 2.49 +0.08 N

07573787-6054321 NGC2516 4790 2.47 +0.08 N

07580963-6049007 NGC2516 4649 2.16 +0.09 Y

08104884-1250289 NGC2539 4805 2.51 +0.08 N

08104286-1240117 NGC2539 4835 2.55 +0.11 Y

08102302-1250433 NGC2539 4824 2.53 +0.10 Y

08101609-1254061 NGC2539 4799 2.49 +0.08 N

08100631-1255197 NGC2539 4791 2.49 +0.08 N

08102909-1243329 NGC2539 4805 2.51 +0.08 N

08103495-1251577 NGC2539 4780 2.45 +0.07 N

08111558-3739567 NGC2546 4796 2.44 +0.08 Y

08121540-3731450 NGC2546 4802 2.52 +0.06 N

08141471-3738428 NGC2546 4845 2.95 +0.02 N

50 08112772-3740402 NGC2546 4808 2.55 +0.05 N

08111698-3736417 NGC2546 4808 2.53 +0.05 N

08135440-0558476 NGC2548 4805 2.52 +0.06 N

08142812-0542161 NGC2548 4825 2.51 +0.08 Y

08134483-0548008 NGC2548 4817 2.50 +0.08 Y

08133964-0547148 NGC2548 4795 2.48 +0.07 N

08134771-5372590 NGC2548 4797 2.50 +0.07 N

08141703-0554007 NGC2548 4837 2.52 +0.08 Y

08134291-0535471 NGC2548 4806 2.53 +0.06 N

08123723-0540509 NGC2548 4823 2.51 +0.08 Y

08204217-3613125 NGC2579 4668 2.20 +0.10 Y

08210039-3617163 NGC2579 4729 2.31 +0.09 Y

08463129-5252414 NGC2669 4718 2.29 +0.09 Y

08511710+1148160 NGC2682 4852 2.60 +0.11 Y

08512898+1150330 NGC2682 4848 2.59 +0.10 Y

08511747+1145226 NGC2682 4860 2.62 +0.11 Y

08514235+1151230 NGC2682 4848 2.59 +0.10 Y

08511704+1150464 NGC2682 4849 2.58 +0.10 Y

08512280+1148016 NGC2682 4848 2.59 +0.11 Y

08504964+1135089 NGC2682 4837 2.58 +0.09 Y

08512377+1149493 NGC2682 4842 2.58 +0.08 Y

08512990+1147168 NGC2682 4826 2.56 +0.09 Y

51 08511269+1152423 NGC2682 4851 2.59 +0.11 Y

11252623-4311239 NGC3680 3793 0.35 0.43 Y 11252924-4315477 NGC3680 3798 0.35 0.46 Y 11253582-4311211 NGC3680 4725 2.35 +0.08 N

11253865-4313587 NGC3680 4558 2.04 +0.10 Y

11253807-4316064 NGC3680 3786 0.34 0.44 Y 11250152-4310231 NGC3680 4730 2.36 +0.09 N

11261027-4321592 NGC3680 4722 2.35 +0.09 N

11254853-4309524 NGC3680 3793 0.33 0.45 Y 11254991-4312157 NGC3680 3783 0.34 0.38 Y 11254329-4315563 NGC3680 4674 2.25 +0.09 N

13462566-6255556 NGC5281 3854 0.43 0.56 Y 13460048-6255149 NGC5281 4797 2.49 +0.07 N

14294260-6039313 NGC5617 4768 2.44 +0.08 N

14295016-6041121 NGC5617 4646 2.23 +0.07 Y

14292873-6042188 NGC5617 4619 2.19 +0.08 Y

15035344-5414055 NGC5822 4710 2.32 +0.09 N

15034041-5428128 NGC5822 4703 2.32 +0.09 N

15030047-5428176 NGC5822 4670 2.27 +0.09 Y

15045198-5419086 NGC5822 4732 2.37 +0.09 N

15053113-5426201 NGC5822 4684 2.28 +0.08 Y

15042824-5417416 NGC5822 4747 2.39 +0.08 N

52 15042202-5424032 NGC5822 4729 2.36 +0.09 N

15042125-5423056 NGC5822 4674 2.27 +0.09 Y

15052349-5421348 NGC5822 4678 2.28 +0.08 Y

15040342-5405419 NGC5822 4703 2.33 +0.08 Y

15041417-5425473 NGC5822 4694 2.31 +0.08 Y

15035000-5414304 NGC5822 4659 2.25 +0.09 Y

15031585-5416391 NGC5822 4632 2.20 +0.09 Y

15041709-5423386 NGC5822 4696 2.30 +0.09 N

15031209-5414065 NGC5822 4697 2.30 +0.09 N

15034941-5420105 NGC5822 4625 2.17 +0.09 Y

16131847-5414040 NGC6067 4715 2.33 +0.09 Y

16130280-5421369 NGC6067 4671 2.25 +0.10 Y

16130403-5412187 NGC6067 4658 2.27 +0.08 Y

16133568-5407364 NGC6067 4631 2.21 +0.08 Y

16123981-5417410 NGC6067 4737 2.41 +0.07 Y

16132470-5420055 NGC6067 4759 2.77 +0.03 Y

16253239-4043477 NGC6124 4785 2.49 +0.07 N

16255169-4044280 NGC6124 4772 2.44 +0.07 N

16243712-4029170 NGC6124 4789 2.49 +0.06 N

16271625-4037254 NGC6124 4790 2.48 +0.06 N

16242387-4043492 NGC6124 4788 2.49 +0.06 N

16255980-4033307 NGC6124 4748 2.74 +0.03 Y

53 16243609-4037027 NGC6124 4788 2.51 +0.06 N

16260042-4051026 NGC6124 4742 2.72 +0.03 Y

16270442-4033171 NGC6124 4794 2.51 +0.06 N

16242637-4045386 NGC6124 4745 2.73 +0.03 Y

16244606-4039576 NGC6124 4783 2.49 +0.06 N

16234611-4046305 NGC6124 4781 2.47 +0.06 N

16262353-4039538 NGC6124 4790 2.50 +0.06 N

16260855-4036035 NGC6124 4783 2.48 +0.06 N

16250138-4028260 NGC6124 4779 2.46 +0.07 N

16275628-4910297 NGC6134 4659 2.24 +0.08 Y

16275584-4908108 NGC6134 4704 2.35 +0.07 Y

16280018-4909063 NGC6134 4694 2.31 +0.08 Y

16280553-4912399 NGC6134 4672 2.28 +0.09 Y

16275101-4906559 NGC6134 4736 2.40 +0.07 Y

16273219-4906459 NGC6134 4676 2.28 +0.08 Y

16281579-4907195 NGC6134 4629 2.19 +0.08 Y

16343259-4931418 NGC6167 4652 2.25 +0.07 Y

16350391-4940290 NGC6167 4784 2.49 +0.06 N

16351464-4938348 NGC6167 4800 2.52 +0.05 N

16354597-4931130 NGC6167 4731 2.68 +0.02 Y

16570817-4556281 NGC6250 4799 2.45 +0.07 Y

17042880-3754355 NGC6281 4739 2.37 +0.08 N

54 17043597-3759311 NGC6281 4574 2.09 +0.10 Y

17044867-3806261 NGC6281 4778 2.44 +0.08 N

17044753-3753147 NGC6281 4619 2.17 +0.09 Y

17043637-3754271 NGC6281 4763 2.42 +0.09 N

17043881-3758442 NGC6281 4738 2.38 +0.09 N

17044830-3757244 NGC6281 4745 2.38 +0.08 N

17394124-3217576 NGC6405 4766 2.43 +0.08 N

17401652-3217511 NGC6405 4804 2.50 +0.07 N

17405971-3209342 NGC6405 4801 2.49 +0.07 N

17405855-3212520 NGC6405 4577 2.10 +0.08 Y

17401109-3215237 NGC6405 4784 2.46 +0.08 Y

17400226-3209367 NGC6405 4776 2.46 +0.08 N

18513150-0622062 NGC6705 4803 2.50 +0.06 N

18505407-0616503 NGC6705 4801 2.52 +0.06 Y

18505981-0617243 NGC6705 4797 2.49 +0.07 N

18510975-0610486 NGC6705 4808 2.52 +0.07 N

20113141+2632535 NGC6885 3792 0.34 0.42 Y 20114389+2635071 NGC6885 3759 0.35 0.34 Y 16284223-5131496 Ruprecht 119 4781.56 2.47 +0.06 Y

55 Bibliography

Abolfathi, B., Aguado, D. S., Aguilar, G., Allende Prieto, C., Almeida, A., Ananna, T. T., Anders, F., Anderson, S. F., Andrews, B. H., Anguiano, B., & et al. 2018, Astrophysical Journal Supplement, 235, 42

Alonso-Santiago, J., Negueruela, I., Marco, A., Tabernero, H. M., Gonz´alez-Fern´andez, C., & Castro, N. 2017, Monthly Notices of the Royal Astronomy Society, 469, 1330

Bagdonas, V., Drazdauskas, A., Tautvaiˇsien˙e, G., Smiljanic, R., & Chorniy, Y. 2018, Astronomy and Astrophysics, 615, A165

Balaguer-N´u˜nez, L., Jordi, C., & Galad´ı-Enr´ıquez, D. 2005, Astronomy and Astrophysics, 437, 457

Blanton, M. R., Bershady, M. A., Abolfathi, B., Albareti, F. D., Allende Prieto, C., Almeida, A., Alonso-Garc´ıa, J., Anders, F., Anderson, S. F., Andrews, B., & et al. 2017, The Astronomical Journal, 154, 28

Cantat-Gaudin, T., Vallenari, A., Zaggia, S., Bragaglia, A., Sordo, R., Drew, J. E., Eisloe↵el, J., Farnhill, H. J., Gonzalez-Solares, E., Greimel, R., Irwin, M. J., Kupcu- Yoldas, A., Jordi, C., Blomme, R., Sampedro, L., Costado, M. T., Alfaro, E., Smiljanic, R., Magrini, L., Donati, P., Friel, E. D., Jacobson, H., Abbas, U., Hatzidimitriou, D., Spagna, A., Vecchiato, A., Balaguer-Nunez, L., Lardo, C., Tosi, M., Pancino, E., Klutsch, A., Tautvaisiene, G., Drazdauskas, A., Puzeras, E., Jim´enez-Esteban, F., Maiorca, E., Geisler, D., San Roman, I., Villanova, S., Gilmore, G., Randich, S., Bensby, T., Flaccomio, E., Lanzafame, A., Recio-Blanco, A., Damiani, F., Hourihane, A., Jofr´e, P., de Laverny, P., Masseron, T., Morbidelli, L., Prisinzano, L., Sacco, G. G., Sbordone, L., & Worley, C. C. 2014, Astronomy and Astrophysics, 569, A17

Carraro, G., Anthony-Twarog, B. J., Costa, E., Jones, B. J., & Twarog, B. A. 2011, The Astronomical Journal, 142, 127

Carretta, E., Bragaglia, A., Gratton, R. G., & Tosi, M. 2004, Astronomy and Astro- physics, 422, 951

Chen, L. & Gao, X. H. 2011, in Astronomical Society of the Pacific Conference Series, Vol. 451, 9th Pacific Rim Conference on Stellar Astrophysics, ed. S. Qain, K. Leung, L. Zhu, & S. Kwok, 303

56 De Silva, G. M., Carraro, G., D’Orazi, V., Efremova, V., Macpherson, H., Martell, S., & Rizzo, L. 2015, Monthly Notices of the Royal Astronomy Society, 453, 106

Dias, W. S., Alessi, B. S., Moitinho, A., & L´epine, J. R. D. 2002, Astronomy and Astrophysics, 389, 871

Donor, J., Frinchaboy, P. M., Cunha, K., Thompson, B., O’Connell, J., Zasowski, G., Jackson, K. M., Meyer McGrath, B., Almeida, A., Bizyaev, D., Carrera, R., Garc´ıa- Hern´andez, D. A., Nitschelm, C., Pan, K., & Zamora, O. 2018, The Astronomical Journal, 156, 142

Eisenstein, D. J., Weinberg, D. H., Agol, E., Aihara, H., Allende Prieto, C., Anderson, S. F., Arns, J. A., Aubourg, E.,´ Bailey, S., Balbinot, E., & et al. 2011, The Astronom- ical Journal, 142, 72

Fitzgerald, M. T., Inwood, L., McKinnon, D. H., Dias, W. S., Sacchi, M., Scott, B., Zolinski, M., Danaia, L., & Edwards, R. 2015, The Astronomical Journal, 149, 190

Frinchaboy, P. M. & Majewski, S. R. 2008, The Astronomical Journal, 136, 118

Garc´ıa P´erez, A. E., Allende Prieto, C., Holtzman, J. A., Shetrone, M., M´esz´aros, S., Bizyaev, D., Carrera, R., Cunha, K., Garc´ıa-Hern´andez, D. A., Johnson, J. A., Majew- ski, S. R., Nidever, D. L., Schiavon, R. P., Shane, N., Smith, V. V., Sobeck, J., Troup, N., Zamora, O., Weinberg, D. H., Bovy, J., Eisenstein, D. J., Feuillet, D., Frinchaboy, P. M., Hayden, M. R., Hearty, F. R., Nguyen, D. C., O’Connell, R. W., Pinsonneault, M. H., Wilson, J. C., & Zasowski, G. 2016, The Astronomical Journal, 151, 144

Hasselquist, S., Shetrone, M., Smith, V., Holtzman, J., McWilliam, A., Fern´andez- Trincado, J. G., Beers, T. C., Majewski, S. R., Nidever, D. L., Tang, B., Tissera, P. B., Fern´andez Alvar, E., Allende Prieto, C., Almeida, A., Anguiano, B., Battaglia, G., Carigi, L., Delgado Inglada, G., Frinchaboy, P., Garc´ıa-Hern´andez, D. A., Geisler, D., Minniti, D., Placco, V. M., Schultheis, M., Sobeck, J., & Villanova, S. 2017, The Astrophysical Journal, 845, 162

Holtzman, J. A., Hasselquist, S., Shetrone, M., Cunha, K., Allende Prieto, C., Anguiano, B., Bizyaev, D., Bovy, J., Casey, A., Edvardsson, B., Johnson, J. A., J¨onsson, H., Meszaros, S., Smith, V. V., Sobeck, J., Zamora, O., Chojnowski, S. D., Fernandez- Trincado, J., Garcia-Hernandez, D. A., Majewski, S. R., Pinsonneault, M., Souto, D., Stringfellow, G. S., Tayar, J., Troup, N., & Zasowski, G. 2018, The Astronomical Journal, 156, 125

Kılı¸co˘glu, T., Monier, R., Richer, J., Fossati, L., & Albayrak, B. 2016, The Astronomical Journal, 151, 49

Majewski, S. R., Schiavon, R. P., Frinchaboy, P. M., Allende Prieto, C., Barkhouser, R., Bizyaev, D., Blank, B., Brunner, S., Burton, A., Carrera, R., Chojnowski, S. D., Cunha, K., Epstein, C., Fitzgerald, G., Garc´ıa P´erez, A. E., Hearty, F. R., Henderson, C., Holtzman, J. A., Johnson, J. A., Lam, C. R., Lawler, J. E., Maseman, P., M´esz´aros,

57 S., Nelson, M., Nguyen, D. C., Nidever, D. L., Pinsonneault, M., Shetrone, M., Smee, S., Smith, V. V., Stolberg, T., Skrutskie, M. F., Walker, E., Wilson, J. C., Zasowski, G., Anders, F., Basu, S., Beland, S., Blanton, M. R., Bovy, J., Brownstein, J. R., Carlberg, J., Chaplin, W., Chiappini, C., Eisenstein, D. J., Elsworth, Y., Feuillet, D., Fleming, S. W., Galbraith-Frew, J., Garc´ıa, R. A., Garc´ıa-Hern´andez, D. A., Gillespie, B. A., Girardi, L., Gunn, J. E., Hasselquist, S., Hayden, M. R., Hekker, S., Ivans, I., Kinemuchi, K., Klaene, M., Mahadevan, S., Mathur, S., Mosser, B., Muna, D., Munn, J. A., Nichol, R. C., O’Connell, R. W., Parejko, J. K., Robin, A. C., Rocha-Pinto, H., Schultheis, M., Serenelli, A. M., Shane, N., Silva Aguirre, V., Sobeck, J. S., Thompson, B., Troup, N. W., Weinberg, D. H., & Zamora, O. 2017, The Astronomical Journal, 154, 94

Ness, M., Hogg, D. W., Rix, H.-W., Ho, A. Y. Q., & Zasowski, G. 2015, The Astrophysical Journal, 808, 16

Netopil, M., Paunzen, E., Heiter, U., & Soubiran, C. 2016, Astronomy and Astrophysics, 585, A150

Nidever, D. L., Holtzman, J. A., Allende Prieto, C., Beland, S., Bender, C., Bizyaev, D., Burton, A., Desphande, R., Fleming, S. W., Garc´ıa P´erez, A. E., Hearty, F. R., Majewski, S. R., M´esz´aros, S., Muna, D., Nguyen, D., Schiavon, R. P., Shetrone, M., Skrutskie, M. F., Sobeck, J. S., & Wilson, J. C. 2015, The Astronomical Journal, 150, 173

Pace, G., Danziger, J., Carraro, G., Melendez, J., Fran¸cois, P., Matteucci, F., & Santos, N. C. 2010, Astronomy and Astrophysics, 515, A28

Reddy, A. B. S., Giridhar, S., & Lambert, D. L. 2013, Monthly Notices of the Royal Astronomy Society, 431, 3338

—. 2015, Monthly Notices of the Royal Astronomy Society, 450, 4301

Santos, N. C., Lovis, C., Pace, G., Melendez, J., & Naef, D. 2009, Astronomy and Astrophysics, 493, 309

Ting, Y.-S., De Silva, G. M., Freeman, K. C., & Parker, S. J. 2012, Monthly Notices of the Royal Astronomy Society, 427, 882

Yong, D., Carney, B. W., & Friel, E. D. 2012, The Astronomical Journal, 144, 95

58 VITA Personal Background Amy Elaine Ray Lexington, TN Daughter of Brad and Nancy Ray

Education Diploma,LexingtonHighSchool,Lexington,TN,2008 Bachelor of Science, Physics, Mississippi State University, Mississippi State, MS, 2013 Master of Science, Physics, Mississippi State University, Mississippi State, MS, 2017

Experience Summer research assistant, Mississippi State University, Mississippi State, MS, 2013 Research assistantship, Mississippi State University, Mississippi State, MS, 2013-2015 Teaching assistantship, Texas Christian University, Fort Worth, TX, 2016-2018 Professional Memberships American Astronomical Society American Association for the Advancement of Science ABSTRACT

UNIFORM CHEMICAL ABUNDANCES OF OPEN CLUSTERS USING THE CANNON

by Amy E. Ray, Ph.D., 2018 Department of Physics and Astronomy Texas Christian University

Research Advisor: Peter M. Frinchaboy III, Associate Professor of Physics

Open clusters are key tracers of Milky Way, as both chemical and age tracers. The use of open cluster to provide significant constraints on galaxy evolution, however, has been limited due to discrepancies in measuring abundances from di↵erent studies. This work seeks to add additional southern open clusters into the SDSS/APOGEE metallicity system. We analyze medium resolution spectra of giant stars in 31 open clusters have been obtained and analyzed using the Cannon to determine [Fe/H] metallicities accurate to 0.15 dex. This uniform analysis is compared for roughly half of the clusters with previous results, and we present metallicities for the first time for 12 open clusters.