Discovery of a Metal-Poor Red Giant Star with the Highest Ultra-Lithium Enhancement

Observational evidence has accumulated on the decrease of the $lithium$ (Li) photospheric abundance in low-mass stars as they age. Li depletion is observed to occur already on the pre-main sequence for the very low-mass stars, and along the main sequence with mass and age-dependent efficiency that can not be explained by so-called classical stellar-evolution models (e.g., Charbonneau & Michaud, 1990; Soderblom et al., 1993; Lyubimkov, 2016; Tognelli et al., 2021; Binks et al., 2022). The observed Li patterns instead reveal the occurrence of internal transport processes of chemical elements other than convection in low mass stars. (e.g., Deliyannis et al., 2000; Talon & Charbonnel, 2010). Models including atomic diffusion and turbulence are almost fully bridging the gap between the primordial (BBN) Li abundance and the Li observed in metal-poor warm turnoff stars along the Spite plateau and in globular clusters (e.g., Richard et al., 2005; Korn et al., 2006; Nordlander et al., 2012; Gruyters et al., 2016; Gao et al., 2020; Deal & Martins, 2021). Also, the Li depletion in more metal-rich stars like the Sun and F- and G-type dwarfs in open clusters can be self-consistently reproduced when accounting for different hydrodynamical processes that transport both matter and angular momentum in stellar interiors (e.g., Charbonnel & Talon, 2005; Dumont et al., 2021a, b, and references therein).

The photospheric Li keeps decreasing during the so-called first dredge-up when low mass stars evolve towards the red giant branch (RGB; e.g., Iben, 1967), as evidenced both in metal-poor stars from the halo and globular clusters and in metal-rich stars in the field and in open clusters (e.g., Lèbre et al., 1999; Charbonnel et al., 2000, 2020; Lind et al., 2009; Canto Martins et al., 2011; Magrini et al., 2021a; Aguilera-Gómez et al., 2022; Mucciarelli et al., 2022). Later, Li decreases again sharply as soon as the stars reach the luminosity of the bump on the upper RGB (e.g., Charbonnel et al., 1998; Gratton et al., 2000; Lind et al., 2009; Charbonnel et al., 2020). This last Li depletion episode has been suggested to be attributed to thermohaline mixing, which is a double diffusive process that has been suggested to connect the base of the convective envelope to the hydrogen-burning shell of low-mass red giants and expected to decrease their photospheric Li abundance and carbon isotopic ratio simultaneously (Charbonnel & Zahn, 2007; Charbonnel & Lagarde, 2010). Angelou et al. (2015) showed that their models, however, predicted that the mixing would start at a higher luminosity than predicted by that data, and that it was not possible to simultaneously reproduce the evolution of carbon and lithium abundance on the RGB. Given the universality of the first dredge-up in low mass stellar evolution, and the sensitivity of lithium to mixing processes as evidenced for both metal-poor and metal-rich dwarf and giant stars, one could reasonably expect to observe low lithium abundance in all low-mass giant stars, independently of their initial metal-content. While this is generally the case, a small fraction of low-mass red giants exhibit exceptionally high Li abundances.

The enhancement of lithium in red giant stars is a rare and continuously not understood phenomenon since its discovery by Wallerstein & Sneden (1982a). Lithium being easy to observe in cool stars, many spectroscopic surveys have looked for Li-rich stars (Brown et al., 1989; Jasniewicz et al., 1999a; Charbonnel & Balachandran, 2000; Monaco et al., 2011a; Lyubimkov et al., 2012; Martell & Shetrone, 2013; Liu et al., 2014a; Bharat Kumar et al., 2015; Zhou et al., 2018; Smiljanic et al., 2018; De Silva et al., 2015; Gao et al., 2019a; Charbonnel et al., 2020; Deepak et al., 2020; Martell et al., 2020; Magrini et al., 2021b). Depending on the adopted criteria for a star to be classified as Li-rich (see discussions in Charbonnel et al. 2020 and Chanamé et al. 2022), all surveys show that roughly ${1\%}$ of red giant stars have enhanced photospheric lithium abundances with respect to their Li-depleted counterparts. Of the lithium-rich giants discovered in the last three decades (e.g., Casey et al., 2019; Gao et al., 2019b; Martell et al., 2020; Cai et al., 2023), only about 3% have been observed to be super Li-rich, with an abundance of A(Li)¹¹1A(Li)$=\log_{10}(N_{Li}/N_{H})+12$ $>3.3$  dex, i.e., higher than the proto-solar Li abundance. Accounting for non-local thermodynamic equilibrium (NLTE) for Li abundances, only a handful of stars have been found to have abundance of A(Li) $>4$ (Martell & Shetrone, 2013; Casey et al., 2016; Yan et al., 2018; Balachandran et al., 2000; Strassmeier et al., 2015; Singh et al., 2019; Susmitha et al., 2024).

Two general ideas have been explored to account for the excess amounts of lithium present in this minority of stars, namely, the production of fresh Li inside the stars themselves, or the accretion of Li-rich material from an external source. Internal Li production in red giants requires a fast transport process in the radiative layers between the hydrogen-burning shell where the unstable ⁷Be can be produced through the pp-chains and the base of the convective envelope where Li can survive²²2This is the so-called Cameron & Fowler (1971) process which is expected to occur in the convective envelope of Li-rich stars on the Asymptotic Giant Branch (AGB; Sackmann & Boothroyd 1992; Forestini & Charbonnel 1997). (Sackmann & Boothroyd, 1999; Charbonnel & Balachandran, 2000; Denissenkov & Weiss, 2000; Palacios et al., 2001; Denissenkov & Herwig, 2004; Silva Aguirre et al., 2014; Yan et al., 2018; Casey et al., 2019; Mori et al., 2021). A number of plausible external mechanisms have also been suggested, including planetary engulfment or novae debris contaminating the outer layers of these giants (Alexander, 1967; Siess & Livio, 1999; Andrievsky et al., 1999; Aguilera-Gómez et al., 2016; Casey et al., 2016; Mallick et al., 2022). The occurrence and the efficiency of these different mechanisms are expected to depend on the evolution stage of the Li-rich stars. Recently, Deepak & Lambert (2021); Yan et al. (2021a); Chanamé et al. (2022) showed using asteroseismology that the majority of the Li-rich giant stars are in their core helium burning phase (Red Clump), while a smaller fraction of Li rich stars have been found to be high on the Red Giant branch (Red Bump), including the extremely high Li-rich star TYC 429-2097-1 (Yan et al., 2018). It is thus yet to be established whether Li-enrichment is strictly attributed to one or more evolutionary status of stars, to be evidenced by the influx of more asteroseismic data for the newly discovered Li-rich star.

It is within this context that we present our discovery of a unique ultra-enhanced lithium giant metal-poor Milky Way halo star, 2MASS J05241392$-$0336543 (hereafter referred to as J0524$-$0336), discovered serendipitously as part of the $R$-Process Alliance (RPA) survey. Initial high-resolution spectroscopic data collected for this star suggested a lithium abundance far greater than the Li-rich standard of A(Li) $\sim 1.5$. We then further collected higher S/N high resolution data for the star to confirm this enhancement, and several epochs of observations to monitor its radial velocity. The fundamental properties of J0524$-$0336 are listed in Table 1, with detailed description explained in the sections to follow.

This paper is outlined as follows: In Section 2, we describe our observations, data reduction, and radial velocity determinations for the candidate star. In Section 3, we thoroughly investigate the fundamental atmospheric stellar parameters, using different methods including (1D, LTE) as well as (NLTE) radiative transfer analyses, and use photometric measurements to establish the stellar evolutionary status of J0524$-$0336. In Section 4, we present the detailed chemical abundance determinations in our star, including abundances of the light, $\alpha$, and $r$-process elements, as well as comparison with previously discovered lithium-rich giants. In Section 5, we discuss the evolutionary status of J0524$-$0336 based on tailor-made stellar evolution models. Finally, we present our discussion of the results and conclusions in Sections 6 and 7, respectively.

J0524$-$0336 was initially identified and vetted as an $r$-process enhanced candidate by the RPA collaboration (Hansen et al., 2018; Sakari et al., 2018; Ezzeddine et al., 2020; Holmbeck et al., 2020), from the Large Sky Area Multi-Object Fibre Spectroscopic Telescope (LAMOST) survey (Deng et al., 2012). The star was observed with the Magellan Inamori Kyocera Echelle (MIKE) spectrograph (Bernstein et al., 2003) on the Magellan-Clay Telescope at Las Campanas Observatory on three separate nights: for 1200 s on 08 March, 2018, for 7600 s, on 26 October, 2019 as well as for 120s on 03 March, 2022. All observations were taken with the 0.7” slit with the $2\times 2$ binning setup, yielding nominal resolving powers of $R\sim 35,000$ in the red ($\lambda>5000$ Å) and $R\sim 41,000$ in the blue, respectively.

The spectra were then reduced using the latest versions of the Carnegie Python distribution³³3https://code.obs.carnegiescience.edu/mike (Kelson, 2003). Each order of each spectrum was afterwards normalized and merged into a final spectrum, covering a wavelength range of $\sim 3320$ Å-9165 Å. For our final analysis of J0524$-$0336, we use the highest S/N spectrum from October 2019, with S/N$\sim 25$ (pixel^-1) at 3950 Å, $\sim 70$ at 4550 Å, $\sim 85$ at 5200 Å, and $\sim 240$ at 6750 Å. The final spectrum was radial-velocity, v_rad, shifted by cross-correlation with the Mg i lines near 5100 Å from the spectrum of the benchmark metal-poor star HD 122563, using the spectroscopic analysis tool Spectroscopy Made Hard (SMH; Casey, 2014). A heliocentric velocity (v${}_{\mathrm{rad}}^{\mathrm{helio}}$) correction was then determined with the rvcorrect package in IRAF⁴⁴4NOIRLab IRAF is distributed by the Community Science and Data Center at NSF NOIRLab, which is managed by the Association of Universities for Research in Astronomy (AURA) under a cooperative agreement with the U.S. National Science Foundation. (Tody, 1986), with v${}_{\mathrm{rad}}^{\mathrm{helio}}=103.10\pm 0.90$ km s^-1. This value is in good agreement with the $Gaia$ DR3 heliocentric radial velocity v${}_{\mathrm{rad}}^{Gaia}=102.79\pm 1.35$  km s^-1. We also determine v${}_{\mathrm{rad}}^{\mathrm{helio}}=102.10\pm 1.00$ km s^-1 from the 2022 data and v${}_{\mathrm{rad}}^{\mathrm{helio}}=101.40\pm 0.90$ km s^-1 from the 2018 data. Based on the present data there is no evidence to support that J0524$-$0336 is in a binary system.

Due to the significant Li enhancement of our star ((Li)(3D,NLTE)${=6.15\pm 0.20}$; see Section 4), we conduct a thorough investigation of its stellar parameters, in an attempt to accurately identify its stellar evolutionary status. We, thus, investigate the fundamental atmospheric stellar parameters (namely, the effective temperature $T_{\rm eff}$, surface gravity $\log g$, metallicity [Fe/H], and microturbulent velocity $\xi_{t}$) of J0524$-$0336 using spectroscopic LTE and NLTE radiative transfer models, as well as with non-spectroscopic methods implementing photometry and fundamental equations. Additionally, we investigate its stellar properties relevant to the current study, including its luminosity, radius, projected rotational velocity $v_{\sin i}$, infrared excess, and H_α emission. Below, we describe the methods and report on each of our derived properties and parameters separately.

We derive abundance measurements, as well as upper limit estimates for the light, $\alpha$, and Fe-peak elements for J0524$-$0336 using the same radiative transfer models and spectroscopic tools described in Section 3. We computed the abundance ratios relative to H and Fe, adopting the solar photospheric abundances from Asplund et al. (2009). The abundances were derived using line-by-line EW and COG analysis, except for Li, C, O, Al, and the neutron-capture elements, where spectral profile synthesis were fit to each line. The linelist was adopted from Roederer et al. (2018) (see their Table 2 for atomic data references) for lines which could be detected in J0524$-$0336. Isotopic ratios were included for Li, C (see below for details), as well as for several neutron-capture elements from Sneden et al. (2008)⁸⁸8https://github.com/vmplacco/linemake and Placco et al. (2021). Hyper-fine structure (HFS) was considered for the Fe-peak elements, including Sc, V, Mn, and Co, when necessary. Upper limits were determined by matching the noise levels in the observed spectral lines with the corresponding synthetic spectral lines. Upon examining abundances from multiple lines for each element, outlying abundances (outside of $1\sigma$) were removed and an the average abundance and standard deviation was recorded for each element. For elements with only 2-5 lines measured, we estimated standard deviations by multiplying the range of values covered by our line abundances with the $k$-factor following Keeping (1962). For elements with one line only, we adopt an uncertainty between 0.1 and 0.3 dex, depending on the data and fit quality. The abundance averages and the corresponding standard deviations are reported in Table 3. The line-by-line abundances of each of these elements is recorded in Table 5. We record systematic uncertainties in our chemical abundances resulting from uncertainties in the model atmospheric parameters, by varying the stellar parameters by typical uncertainties in the positive direction (i.e., $\Delta$$T_{\rm eff}$= +150 K, $\Delta$$\log g$= +0.30 dex, $\Delta$[Fe/H] = +0.30 dex, and $\Delta$$\xi_{t}$=0.2 km s^-1), and recording the change in abundance. This is shown in Table 4. All changes are within expected error bars.

To understand the ultra Li enhancement in J0524$-$0336 (A(Li)$>5$), it is crucial to establish its evolutionary status. We compute tailored stellar evolution models using the latest version of the stellar evolution code STAREVOL (see Dumont et al. 2021b for references and details on the equation of state, opacities, and nuclear reactions). The initial chemical composition accounts for the values of [Fe/H], [C/Fe], [O/Fe], and [Na/Fe] derived in this work (see Section 4 and Table 3 for values). We use a gray atmosphere and define the stellar effective temperature and radius at the optical depth $\tau=2/3$. The mass-loss rate follows Reimers (1975) empirical relation (with $\eta_{R}=0.5$) from the zero-age main sequence up to central He exhaustion and switches to Vassiliadis & Wood (1993)’s prescription on the asymptotic giant branch (AGB). We adopt a mixing length parameter $\alpha_{\rm MLT}=1.5$, and we assume the Schwarzschild criteria for convective stability. We include the effects of the thermohaline instability as described in Charbonnel & Zahn (2007) and Lagarde et al. (2012). We also compute models with thermohaline mixing and solid-body rotation, to predict the evolution of the surface stellar rotation rate under conservative assumptions for such a low-mass star.

The position of J0524$-$0336 on the HR diagram and on the Kiel diagram is well fit by the evolutionary track of a model with initial mass of 0.8 M_⊙ (see Figure 9). Given the uncertainties on the stellar parameters, J0524$-$0336 appears to be located either on RGB above the predicted location of the bump, or on the e-AGB before the occurrence of the thermal pulses on the AGB (TP-AGB). It has a much cooler temperature than the position of the horizontal branch (HB), which is the metal-poor counterpart of the red clump (RC) for metal-rich giants. Although no asteroseismic constraint is available for J0524$-$0336, we can safely conclude that it is currently not undergoing central He-burning.

We computed different global asteroseismic quantities ($\nu\mathrm{max}$, frequency at which the oscillation modes reach their strongest amplitudes; $\Delta\Pi$, asymptotic period spacing of g-modes (for $l=1$), total accoustic radius and acoustic radius at the base of the convective envelope) all along the evolution with the same prescriptions as in Lagarde et al. (2012). At the two evolutionary points on the RGB and early-AGB where our model reaches the luminosity of J0524-0336 and has an effective temperature compatible with the derived values within the error bars, their values are very similar, and 2 to 3 orders of magnitude lower than when the model undergoes central He-burning at a very high effective temperature on the horizontal branch (see Lagarde et al. (2012) for a description of the variations of the global asteroseismic quantities along the evolution of low-mass stars from the pre-main sequence to the end of the TP-AGB).

Within the error bars on the luminosity and effective temperature of J0524$-$0336, both on the RGB and the e-AGB, the model predicts a value of 9.6 for the ¹²C/¹³C ratio (to compare to the value of 42 obtained after the first dredge-up), due to the effect of the thermohaline mixing that is predicted to occur at the RGB bump. This is in very good agreement with the value we derived of ¹²C/¹³C=10, confirming that J0524$-$0336 has already undergone thermohaline mixing when it has previously crossed the RGB bump. The typical rotation rate for stars at this evolution stage and metallicity is 1.2 km s^-1 (Cortés et al., 2009), which is much lower than that derived in this study (see Section 3.5).

The evolutionary status of Li-rich giants has long been debated. Charbonnel & Balachandran (2000) found an accumulation of such stars around the RGB bump and on the e-AGB, this later phase being in agreement with our conclusions for J0524$-$0336 . Interestingly, the next most highly enhanced lithium star (A(Li)=4.51) is probably located at the RGB bump (Yan et al., 2018). Others studies, sometimes using additional constraints based on asteroseismology parameters, concluded that a large fraction of enhanced lithium giants are actually red clump (RC) stars presently burning He in their central convective core, the rest being either close to the RGB bump or randomly distributed along the advanced phases (Kumar et al., 2011; Silva Aguirre et al., 2014; Casey et al., 2019; Yan et al., 2020; Kumar & Reddy, 2020; Martell et al., 2021; Yan et al., 2021b). Recent studies have shown that there could be different paths to Li-enrichment. For example, Chanamé et al. (2022) demonstrated that lithium-rich RC giants are likely to be a younger and more massive population than lithium-rich RGB stars. Sayeed et al. (2023) performed an empirical study of Li-rich giants from the GALAH survey and showed that while Li-rich stars are prevalent on the red clump, other culprits such as binary spin-up and mass-transfer could also be likely mechanisms to enrich stars with Li at the RGB phase. As of today, though, there is still no clear consensus, and different paths to Li-enrichment remain to be explored and confirmed observationally.

Several mechanisms have been suggested in the literature to try to explain Li-richness in giant stars. These can be broadly divided into external and internal mechanisms. External mechanisms include the presence or interaction with a stellar or sub-stellar companion (such as planetary engulfment or binary interactions), while internal mechanisms include non-traditional or enhanced mixing processes. In what follows, we discuss some of these different mechanisms, linking them to the observable properties of J0524$-$0336, to try to explain its extreme lithium enhancement relative to other stars in the Milky Way.

We report on the discovery of J0524$-$0336, an ultra Li-rich star, with A(Li)(3D, NLTE)=${6.15}$ and [Li/Fe]=${+7.64}$, from two strong lithium absorption lines. To our knowledge, this is the most Li-rich giant star discovered to date. Additionally, we derive abundances of 16 neutron capture elements in J0524$-$0336 with [Eu/Fe]$=+1.34$, classifying it as a highly enhanced $r$-process star (rII - [Eu/Fe]$>0.7$; Holmbeck et al., 2020). We conduct a detailed stellar parameters analysis of our star using a high-resolution ($R\sim 35,000$), high S/N spectrum. Based on its stellar parameters and Gaia DR3 distance, we find that J0524$-$0336 lies either on the RGB between the bump and the tip, or on the early-AGB. In any case, it is not currently undergoing core He-burning, which is predicted to occur at a much higher temperature for such a low-mass, low-metallicity star. The star should thus have a relatively low asymptotic period spacing of g-modes. Additionally, we determine a fast projected rotational velocity, $v_{\sin i}\sim 11$km s^-1 , as compared to typical values in red giant stars. Furthermore, both an IR-excess, as well as variable emission in the wings of the H_α profile were detected.

We investigated both internal and external processes of possible lithium enhancement that might explain such an ultra-high lithium abundance in a red giant star. Fast rotational velocity could point toward an external source of enrichment, either due to a sub-stellar companion (e.g., planetary) engulfment or binary interaction. No variation was detected in the star’s radial velocity over several observational epochs, from which no present binary companion can be inferred. Nevertheless, H_α emission and IR-excess has been postulated to be due to mass-loss events due to a binary merger with giant stars. We show that a sub-stellar companion engulfment cannot produce the Li abundance observed in J0524$-$0336, and that a scenario in which this would be possible would require a high planetary mass ($M_{p}\geq 10$M_J) and an unrealistically high intrinsic planetary Li content (A_p(Li)$>7$). On the other hand, it could be possible that the ultra high Li in J0524$-$0336 could be due to a previous merger with a binary star triggering and Li-producing extra mixing in the star. Additional studies on the exact amount of Li that could be produced during these interactions are, however, needed to confirm or refute such an event in J0524$-$0336.

We also investigate possible mixing mechanisms that could drive the so-called Cameron-Fowler process in the radiative layers between the H-burning shell and the convective envelope of J0524$-$0336, and account for its extreme Li abundance. Recent parametric models, taking into account asymmetric upwards and downwards mixing flows and updated nuclear reaction rates, could reach A(Li)=5.07 for the case of a more massive and more metal-rich red giant sitting at the RGB bump. Similar computations should be done for the case of a low-metallicity, low-mass star such as J0524$-$0336, to investigate the mass and metallicity dependence of the driving mixing mechanism. We also considered the possibility of Li production in J0524$-$0336 during the lithium flash as proposed by Palacios et al. (2001), and which is expected to occur at the RGB bump and/or on the early-AGB (Charbonnel & Balachandran, 2000). J0524$-$0336’s evolutionary status, fast rotational velocity of $v_{\sin i}\sim 11$ km s^-1, IR excess, H_α emission, and unusually high lithium content are predicted indicators of the lithium flash as described in Palacios et al. (2001). It is therefore possible that J0524$-$0336 was observed during this rare phenomenon, as well as TYC 429-2097-1, which is the next highest A(Li) red giant star, and which is sitting at the bump. However, additional models and upper limits on the production of Li during the Li flash (preferably in 3D), are needed to confirm this scenario.

J0524$-$0336 sets a new benchmark for lithium-rich metal-poor stars, being the first giant discovered with an A(Li)$>6.0$ and [Li/Fe]$>+7$. It provides a great opportunity to investigate the origin and evolution of Li in the Galaxy, and could be the first of a new population of ultra-rich lithium-enhanced stars expected to be discovered in current and future high-resolution spectroscopic surveys such as the $R$-Process Alliance, as well as others. Asteroseismologic follow-up of metal-poor Li-rich targets will be key to pin down the exact evolutionary status of our and similar stars. However, distinguishing whether J0524$-$0336 is climbing the RGB or the early-AGB would be extremely challenging, since the global asteroseismic parameters should not differ significantly between these two evolution phases. Finally, we hope that the discovery of J0524$-$0336 will open a new avenue to understand the instabilities that can develop and transport matter and angular momentum in red giant stars.