An Empirical Measurement of the Initial-Final Mass Relation with Gaia White Dwarfs

TL;DR

This paper uses Gaia DR2 data to empirically determine the initial-final mass relation (IFMR) for white dwarfs, revealing non-linearity and substructure in the CMD, with implications for understanding stellar evolution and cluster observations.

Contribution

It presents the first clear observational evidence of non-linearity in the IFMR using Gaia data and a generative model for the white dwarf CMD.

Findings

The best-fit IFMR flattens between 3.5 and 5.5 solar masses.

A secondary peak in WD mass distribution at 0.8 solar masses.

Bimodal CMD features are only visible in mixed-age populations.

Abstract

We use data from Gaia DR2 to constrain the initial-final mass relation (IFMR) for field stars with initial masses $0.9 \lesssim m_{\rm in}/M_{\odot} \lesssim 8$. Precise parallaxes have revealed unprecedented substructure in the white dwarf (WD) cooling sequence on the color-magnitude diagram (CMD). Some of this substructure stems from the diversity of WD atmospheric compositions, but the CMD remains bimodal even when only spectroscopically-confirmed DA WDs are considered. We develop a generative model to predict the CMD for DA WDs as a function of the initial mass function (IMF), stellar age distribution, and a flexibly parameterized IFMR. We then fit the CMD of 1100 bright DA WDs within 100 pc, for which atmospheric composition and completeness are well-understood. The resulting best-fit IFMR flattens at $3.5 \lesssim m_{\rm in}/M_{\odot}\lesssim 5.5$, producing a secondary peak in the WD mass distribution at $m_{\rm WD} \sim 0.8 M_{\odot}$. Our IFMR is broadly consistent with weaker constraints obtained from binaries and star clusters in previous work but represents the clearest observational evidence obtained to date of theoretically-predicted non-linearity in the IFMR. A visibly bimodal CMD is only predicted for mixed-age stellar populations: in single-age clusters, more massive WDs reach the bottom of the cooling sequence before the first lower-mass WDs appear. This may explain why bimodal cooling sequences have thus far evaded detection in cluster CMDs.