Lithium (Li) is an ancient element that is almost as old as the universe itself. The abundances of Li observed in many celestial bodies often disagree with the predictions of classic theories. The Li-rich giant stars, accounting for only 1% of the total number of the low-mass evolved stars, is one example of such conflicts.

It was traditionally supposed that a large fraction of Li-rich low-mass evolved stars were at the red giant branch (RGB) stage, whereas recent studies report that the red clump (RC) stars, in which helium are burning at the cores, are more frequent than RGB stars. By monitoring the ‘heartbeats’ using asteroseismic approach to a large sample of such stars, we provide clear evidence that the majority of Li-rich stars are at the RC stage, and their Li abundances are higher than those of the RGB stars on average.

The combination of spectroscopy and asteroseismology is the key factor for obtaining those results (figure 1). Seven telescopes worldwide, including the Large Sky Area Multi-Object Fiber Spectroscopy Telescope (LAMOST), Subaru, and Kepler, have been used for data collecting. The observations lasted for years from 2015. Here I would like to share the story about the observations in our study.

In the year of 2014, Silva Aguirre et al. found a Li-rich low-mass evolved star with core-helium-burning using the asteroseismic approach. At that time, we had no idea if RC stars accounted for a considerable fraction of Li-rich low-mass evolved stars, or this one was just a particular case among such stars. But one thing was certain, the asteroseismology will definitely play an important role in this filed, as it provides the reliable classification to the evolutionary stage by the ability of detecting the core of a star (figure 2).

It would be interesting to investigate the fraction of RC stars among a large sample of the Li-rich low-mass evolved stars, we thought. Then our research plan at that time was to study Li-rich low-mass evolved stars using only the high-resolution spectra, from which reliable stellar parameters and elemental abundances can be obtained. Based on this plan, we firstly selected about two dozen of Li-rich candidates in the Kepler filed. All of these stars had strong Li lines at 670.8 nm seen from LAMOST spectra, and were bright enough for the high-resolution spectroscopic observations. Meanwhile, they were suitable for the asteroseismic analysis, as most of them had sufficient data from the Kepler. The observation started at 2015, and most of the targets were observed by Subaru, a 8.2-meter flagship telescope located at Mauna Kea Observatory on Hawaii and run by the National Astronomical Observatory of Japan.

Everything went fine, but when we were analyzing the data, the results shocked us. The red clump stars accounted for more than 90% of our whole sample, and only two stars were at RGB stage. This was almost exactly the opposite result from our previous understanding, because at that time, only two Li-rich stars were found to be red clumps.

Then we began to struggle with the preliminary results. The ratio was apparently biased by the selection of targets (stars with strong Li lines were selected) and the limited size of the sample. Besides, we were hardly able to obtain more crucial information from the sample for further constraining the properties of the Li-rich low-mass evolved stars.

We decided to expand our sample. This means that we have to turn to the low-resolution spectra, because the majority of the bright Li-rich stars in the Kepler filed have already been observed with the high-resolution spectroscopy in the first step. Thanks to the powerful spectra-collecting capability of LAMOST, we found that ~40,000 evolved stars were observed through the LAMOST-Kepler project (figure 3), which aimed to systematically survey for the common sources in the LAMOST and Kepler filed. We finally succeeded to identify more than 600 stars with Li abundances over 1.5 dex. We compared the Li abundances and stellar parameters derived from the LAMOST spectra to those from the high-resolution spectra, and found a good consistency.

With this expanded sample, we were able to carry out the statistical analysis to a set of signatures based on the classification from the asteroseismology. For example, we revealed the steep decline of the distribution and the possible upper limit of Li abundance for RGB stars, we also found that the distributions of mass and nitrogen abundance are notably different between RC and RGB stars. The using of asteroseismic data to this uniquely large sample enables us to investigate more detailed properties by classifying individual stars into RC or RGB.

Although the mechanism that enhances Li in RC stars is still unknown, the result that the evolutionary status of such stars has been successfully identified is a big step toward solving the question. It will also be a clue to help answer the long-standing problems in the stellar evolution theories.

In the end, I would like to express my thanks to the staffs of the telescopes used in our research. Besides LAMOST, Subaru, and Kepler, the other four are 3.5-meter telescope at Apache Point Observatory, the Automated Planet Finder telescope at Lick Observatory, the 2.4-meter and 1.8-meter telescopes at Lijiang Observatory. I would also like to thank the editors and anonymous referees who helped to improve this paper.

Published article can be found here: https://www.nature.com/articles/s41550-020-01217-8