Spectroscopic Measurements of Lithium in Late-type Stars

Abstract

Stars are fundamental to our understanding of the Universe, from the largest galaxies to the smallest planets. Low mass stars retain the chemical imprint of the gas cloud they were born from. Therefore, stellar abundances provide insight into the evolution of stars and chemical elements throughout the life of galaxies, such as our own Milky Way. Galactic archaeology uses stars and their chemical fingerprints like fossils to trace cosmic history. Out of the known elements, lithium (Li) is a uniquely interesting element, produced in Big Bang Nucleosynthesis (BBN) and cosmic ray spallation, and moderated in stellar evolution.

Stellar abundances cannot be directly observed or measured, instead they must be inferred from models; with 3D hydrodynamical models more accurate and computationally expensive than 1D hydrostatic models; and local thermodynamic equilibrium (LTE) an approximation on non-LTE (NLTE) spectral synthesis. I calculate a grid of 3D NLTE Li spectra for FGK-type stars and provide an interpolation package Breidablik, allowing for accurate measurements of Li abundances from spectroscopic observations. I apply these models to high fidelity observed spectra and to a large spectroscopic survey.

I test the limits of 3D NLTE modelling through ESPRESSO observations of three metal-poor dwarf halo stars. I fit 3D NLTE line profiles of Li, Fe, and K, and show that the line shape of 3D NLTE line profiles matches with high fidelity ESPRESSO observations, thereby confirming the accuracy of 3D NLTE spectral synthesis at levels previously only seen for the Sun. I apply my 3D NLTE Li grid to over 600k stars observed as part of the GALAH survey. I developed a method for empirically taking into account the effects of blends on the weak Li line. It is important to derive realistic uncertainties for Li because the upper limits depend on them - not only are the detections close to upper limits in abundance, but also because two thirds of the dataset are upper limits. This is the largest catalogue of 3D NLTE Li abundances to date.

Throughout this thesis, I investigate Li phenomena in late-type stars using my 3D NLTE synthetic spectra. First, the so-called cosmological lithium problems: there is a discrepancy between the predicted BBN 7Li and 6Li abundance compared to the abundance measured in metal-poor halo stars. I find that there is no discrepancy in 6Li, which implies that the solution to the discrepancy in 7Li lies in stellar evolution. I measure the Li-dip, a narrow region of main sequence turn off stars which are depleted in Li. I show that the Li-dip extends up the subgiant branch, implying that the physical mechanism that forms the Li-dip not only depletes Li but also destroys Li. I find a meltdown of the the warm plateau, formed by stars warmer than the Li-dip, due to a large number of false detections. Similarly, I find that the metal-poor giant plateau is not reliably measured due to false detections. In order to study these plateaus, higher signal-to-noise ratio spectra is required. I then investigate Li in giant stars, where I observe more Li in red giant branch stars compared to red clump stars, implying that there is no ubiquitous production of Li in the He-flash. I find more Li rich red clump stars than Li rich red giant branch stars, but no strong correlation with any stellar parameters, leaving the origin of these Li rich stars unknown.

This thesis is a study of Li abundances in stars using stellar spectra from both a modelling and observational perspective. Although physically accurate 3D NLTE models are computationally expensive, I show that 3D NLTE spectral syntheses are accurate when compared to high fidelity observations and that it is now possible to apply these models to large spectroscopic surveys. The detailed study of Li phenomena from a large set of stars with accurate abundance measurements probes our understanding of the physical mechanisms at play during stellar evolution.