Annual Review of Astronomy and Astrophysics. The Star Planet Composition Connection. Johanna Teske.

Index of Other Science Articles:
https://rumble.com/v3t4yzj-index-of-science.-music-by-dan-vasc.html

Annual Review of Astronomy and Astrophysics.
The Star Planet Composition Connection.
Johanna Teske.
Earth and Planets Laboratory, Carnegie Institution for Science, Washington, DC, USA.

Keywords. Exo-planets, stellar abundances, host stars, exo-planet compositions, planet formation.

Abstract.
The mantra “know thy star, know thy planet” has proven to be very important for many aspects of exo-planet science. Here I review how stellar abundances inform our understanding of planet composition and, thus, formation and evolution. In particular, I discuss how:
The strongest star planet connection is still the giant planet metallicity correlation, the strength of which may indicate a break point between the formation of planets versus brown dwarfs.
We do not have very good constraints on the lower metallicity limit for planet formation, although new statistics from TESS are helping, and it appears that, at low Iron to Hydrogen, alpha elements can substitute for iron as seeds for planet formation.
The depletion of refractory versus volatile elements in stellar photo spheres and particularly in the Sun, was initially suggested as a sign of small planet formation but is challenging to interpret, and small differences in binary star compositions can be attributed mostly to processes other than planet formation.
We can and should go beyond comparisons of the carbon-to-oxygen ratio in giant planets and their host stars, incorporating other volatile and refractory species to better constrain planet formation pathways.
There appears to be a positive correlation between small planet bulk density and host star metallicity, but exactly how closely small planet refractory compositions match those of their host stars, and their true diversity, is still uncertain.

One. INTRODUCTION.
To the best of our knowledge, planets generally form in orbit around stars in disks composed of the gas, dust, and ice left over from star formation, which itself holds a record of billions of years of galactic and stellar evolution. Until approximately 30 years ago, we had only one example of a successful iteration of this process around a Sun-like star, although this example provides the most detailed sources for learning about the outcome of planet formation in Solar System planets, moons, comets, asteroids, and so forth. Now, with more than 5,600 confirmed planets around other stars, see the exo-planet archive website, many in multi-planet systems, it is clear that:
(a) Planet formation is common in the Galaxy and can occur around a range of stellar types,
(b) The architectures of exo-planet systems can be very different from that of the Solar System, and
(c) There is a wider range of bulk densities, a rough proxy for composition, in exo-planets than that found in Solar System planets.
Much of the research in the field of exo-planets is related to understanding the factors and processes driving these three points and how “normal” the Solar System is in terms of how it formed and the resulting planet compositions, including whether or not they present conditions suitable for the emergence of life. Host star properties like mass, see Mulders 20 18 and references therein, activity and age, and multiplicity play significant roles in the characteristics of the exo-planet population we observe today. Obtaining accurate and precise host star properties is also critical for extracting accurate and precise planet properties, deriving fundamental parameters like planet mass, Mp, and planet radius, Rp, from observations requires knowledge of the host star, and some specific host star properties directly influence the efficacy of exo-planet detection, meaning that, for example, we are biased in our exo-planet sample toward inactive and single or wide binary stars.
This review focuses on how host star composition, which, of course, is not independent of other properties like stellar mass and age, is related to planet properties. Understanding the origins of different stellar compositions is a fascinating and dynamic field unto itself, but suffice it to say that stars across the Galaxy display a range of different abundances related to their time and location of formation. In a simplistic sense, considering all populations of stars in the Milky Way, the abundances of elements heavier than Hydrogen and Helium, meaning metals, increase with time, so that the oldest stars are the most metal-poor and the youngest are the most metal-rich.

In actuality, the trends between metallicity and age are more complex and related to other variables like alpha to Iron, as discussed in Section 2 point 2, and stellar orbital properties.
The relative proportions of metals are dictated by the astrophysical processes producing them.
From here onward, I use the term metallicity to indicate the iron abundance, Iron to Hydrogen, of a star. This bracket notation indicates the log abundance relative to the Sun, that is, A to B is log 10 N A to N B of the star, minus log 10 of NA to NB of Sol, where N A and N B are the number densities of elements A and B. The increase in Iron over time is due to its production primarily in type One-A supernovae, S N e, wherein white dwarfs that used to be low-mass stars accrete enough mass from a companion star to trigger Carbon fusion and runaway thermonuclear explosion. Other elements like Magnesium, Silicon, and Calcium, called alpha elements because they form via the alpha-capture process, are produced primarily in massive stars and dispersed earlier by type Two S N E, so the alpha to Iron ratio is an important tracer of different stellar populations. The elements Oxygen and Sulphur are also often included with alpha elements, and both generally decrease relative to Iron with time. The source of Carbon is less certain, type Two S N E, novae, massive star winds, low to intermediate-mass stars, but it changes slightly more than Oxygen with age such that Carbon to Oxygen ratio has a small positive slope with age. For a much more in-depth discussion of the process of Galactic chemical evolution, G C E. Dynamics is also likely important, leading to the term chemodynamical evolution.
This review approaches the connection between star and planet compositions from two related perspectives.
First, how are stellar compositions related to the presence or specific architectures of planetary systems, Section 2.
Second, in what ways do host star compositions inform our interpretation or prediction of planet compositions, Section 4.
Section 3 takes a detour into a few special cases of anomalous abundances, how the formation or presence of planets can change a star’s abundances in a measurable way in comparison to a star without known planets, although of course we now know that most stars host planets and it is actually difficult to pinpoint stars that definitely do not host planets. The review concludes with a few future prospects for new or enhanced ways of using stellar abundances to learn about planet compositions, Section 5.
Before diving in, I give a very brief overview of how host star abundances are derived.
Stellar abundances, representative of the average photosphere, are derived from spectra, which need to be of high enough resolution to separate or measure most individual line intensities.
Higher resolving power and signal-to-noise, S N, spectra allow for more in-depth characterization of the star, including measurement of more elements that have weaker lines. Typically, the minimum values for abundance derivation for reaching the precision limits that we have today are around R greater than approximately 20,000 and Signal to Noise greater than approximately 100, although they depend on the specific case, for example, a small number of abundances but across many thousands of stars. There are nuances in both reducing spectral data. For example wavelength calibration, merging orders in the case of echelle spectra, cosmic ray correction, and preparing them for the extraction of information, for example subtracting telluric absorption from Earth’s atmosphere and normalization, namely blaze function removal. Most abundance research is done with moderate to high-resolution optical spectra, but other wavelength regions can be more appropriate depending on the object, for example, near-IR, N I R for M dwarf stars or U V for polluted white dwarf stars.
The basic measurements used to derive abundances from spectra are the strengths of lines from different species. Line strengths can be based on the measurement of equivalent widths E W’s, by fitting a Gaussian or Voigt profile or simply integrating over the line profile.

These are then compared with the curve of growth associated with specific stellar parameters, where for lines that are not saturated there is a linear relationship between the abundance, number of atoms, and the E W. This method is sensitive to how the continuum around the fitted line is chosen and is not ideal for blended or broadened lines that can give overestimated abundances. Alternatively, line strengths can be compared with synthetic profiles, usually from a predetermined synthetic spectrum with given stellar parameters, of varying abundances until the best match is found. An extension of the line-specific synthesis method is comparing a large part of, or the entire, spectrum with a library of synthetic stellar spectra, or even a library of observed spectra of stars with well-calibrated parameters. However, in practice, synthesis is typically used for special cases of blended or very weak lines, crowded features, and molecular features. The main downside of synthesis analysis is that it is very sensitive to the instrument profile and the wavelength solution, as every point in the observed line profile is fitted. The instrumental profile and broadening due to rotation and micro turbulence have to be carefully included in the full synthesis model, whereas E W’s do not require this step. Spectral synthesis results can also suffer from systematic bias because of correlations between The Effective temperature, Iron to Hydrogen metallicity, and log g when solving for all three quantities at once, which is less of an issue in the curve-of-growth technique. Given the growing number of very large data sets of stellar spectra, there is increasing interest in using machine learning approaches to measuring abundances, although these still rely on a robust training set of labels from either a subset of the observed spectra or synthetic spectra.
Translating these line measurements into actual abundances requires an atmospheric model.
The most widely used models are still calculated under the assumption of 1D local thermodynamic equilibrium L T E, but the availability of 3D non L T E models is increasing, and comparisons between the two have shown the necessity of using the latter for at least some lines, for example the 77 71 to 75 Angstrom Oxygen One triplet. For a recent calculation of corrections between 3D non-LTE and 1D LTE, see, for example Amarsi et al 20 19 or classes of spectra, for example, ultra-metal-poor stars. Modeling the atmospheres of cool stars, less than around 4,000 Kelvin is challenging due to the prevalence of molecular species in the spectra and a lack of accurate line parameters, position and transition probability, the latter a product of the oscillator strength and statistical weight, which require laboratory spectroscopy, theoretical calculations, or astrophysical calibration, combining a calibrated model with an observed line of uncertain oscillator strength.
Indeed, even many atomic lines at optical wavelengths have greater than or equal to 10 percent uncertainty in their transition probabilities. Thus, not surprisingly, which lines one measures can influence the resulting abundances. Lines of a given element spanning a range of excitation potentials and ionization states help probe different parts of the stellar atmosphere, and the goal is to obtain an abundance that is well representative of the average. The most precise stellar abundances, and parameters are derived from differential analyses, particularly of similar stars.
Here I refer to effective temperature, surface gravity g, usually given as log g, and micro turbulent velocity zeta, in addition to Iron to Hydrogen ratio. Deriving these parameters can be and often is also done from the spectra by measuring many Iron lines and by:
(a) Minimizing the dependence of Iron to Hydrogen on the excitation potential Chi of the lines, which sets the effective temperature,
(b) requiring Iron to Hydrogen derived from Iron One and Iron two lines to agree, which sets log g, and
(c) Minimizing the dependence of Iron to Hydrogen on reduced E W’s, log E W over lambda, which sets micro turbulence. Other methods for deriving the effective temperature and log g exist that are independent of the spectra and thus can provide an extra check.
This approach reduces systematic uncertainties arising from the choice of continuum as well as from line selection, as a line-by-line comparison effectively cancels out uncertainties from non-L T E, blends, poorly constrained line parameters, and so on.
Most of the analyses discussed in this article focus on F G K dwarf stars, leaving out cooler and lower mass M dwarfs.

Measuring accurate and precise abundances and stellar parameters for M dwarfs presents extra challenges because of their generally fainter magnitudes in the optical range, where many high-resolution spectrographs operate, and optical spectra overlain with molecular features that blend in with atomic features and make it hard to determine the continuum. Progress has been made by benchmarking abundances of M dwarfs in binary systems to those of their more massive companions and or by using N I R spectra which have more flux and fewer strong molecular blends. Some of these studies have even derived abundances beyond Iron to Hydrogen metallicity. And yet, abundance measurements are still lacking for many M dwarf exo-planet host stars. This is an active area of growth in stellar astrophysics and exo-planet science and will likely change rapidly in the coming years, see Section 5.
The take-home message from this section is that there are many choices that go into deriving stellar abundances that can affect their accuracy and precision. The effects that different methodological choices can have on the resulting abundances are discussed in greater detail by Smiljanic et al, 20 14 and Hinkel et al in 20 16. When studying the relationships between exo-planets and their host stars, it is ideal to use stellar abundances derived from the same source, using the same methodology, which helps at least in a relative comparison sense. It is heartening that multiple large all-sky spectroscopic surveys are commencing or on the horizon, as we may soon have uniform abundances across most exo-planet host stars, see Section 5.

Two. STELLAR COMPOSITION AS AN INDICATOR OF PLANET PRESENCE AND ARCHITECTURE.
Despite the nuances described above, stellar compositions are generally easier to measure than those of their orbiting planets. Thus, there is great potential in using different stellar compositions as indicators of the types of planetary systems that can form or have formed around them. This section presents updates to or fairly new trends in stellar composition versus planet properties that inform how we think about planet formation.
2 point 1. Metallicity, Iron Abundances.
Soon after the first detections of exo-planets orbiting Sun-like stars, what remains the most influential host star planet composition connection was identified:
Short-period gas giant planets tend to orbit more metal-rich, high Iron to Hydrogen metallicity abundance stars. As discussed in Section 3 point 2, the community initially debated whether this trend was a cause or effect of planet formation, but eventually it was interpreted as evidence that the dominant formation mechanism for giant planets, at least those fairly close-in around Sun like stars, is core accretion. Population synthesis modeling of core accretion that includes planet migration, disk evolution, and accounting for detection biases provides a good match to the observed systems, whereas planet formation via the disk gravitational instability method is relatively insensitive to the disk opacity, which is likely related to the metallicity of the parent star. This review focuses on stellar composition, but it is important to remember that stellar metallicity and other properties like mass, luminosity, and age are not independent. In particular, the available mass budget for forming planets via core accretion is a function of both the metallicity and mass of the protoplanetary disk, the latter is often assumed to be related to the stellar mass, even a relatively high metallicity environment may not form giant planets if there is not enough total mass. The following subsections discuss what is known about the Iron to Hydrogen metallicity trends in stars hosting different types of exo-planets.

Figure One.
Occurrence rates of (a) hot (P equals 1 to 10 days) and (b) warm (P equals 10 to 100 days) planets as a function of host star Iron to Hydrogen metallicity.
The occurrence of almost all types of planet seems to depend at least weakly on Iron to Hydrogen metallicity, and in some cases differs depending on the orbital period range covered. The colors indicate different planet sizes.
The lines represent the best fit to a distribution f Of X and P is proportional to P to the power of alpha times 10 to the power of Beta times X, where the period dependence has been integrated over.
Jupiters: planets between around 8 and 24 Earth Radii and between around 0.1 and 10 Jupiter Masses. Those with P less than 10 days and or Equivalent temperature greater than around a thousand Kelvin are known as hot, and those with dayside temperatures greater than around 2,200 Kelvin are known as ultra hot.
Sub-Saturns: planets between around 4 and 8 Earth Radii, generally with masses between around 10 and 100 Earth Masses increasing with radius, at P less than around 100 and likely longer.
2 point 1 point 1. Gas giant planets.
The trend in occurrence of giant planets, Jupiters to sub-Saturns, with host star Iron to Hydrogen metallicity has been extensively studied to extract additional information and refine formation models. It is actually quite a strong function, often parameterized as a power law with an index around two, see Figure 1. Furthermore, it appears that not only the presence of close-in giant planets but also their masses and radii are positively correlated with the host star metal content, although how the bulk metal content of giant planets depends on stellar metallicity is less clear, see Section 4 point 1. On the basis of data mostly from F G K dwarfs, the “break point” in planet mass, when the occurrence of giant planets no longer appears to depend on host star metallicity, is somewhere between 4 and 10 Jupiter Masses, even when differences in host star masses and thus disk masses are accounted for, suggesting that planets above this mass range may form differently, perhaps via gravitational instability. This observational transition provides an independent distinction between a planet and a star that does not depend on an object’s composition or internal structure, and also helps constrain the properties, viscosity and initial mass, of disks forming planets via core accretion such that they must not form planets larger than 10 Jupiter Masses.
A trend of decreasing host star metallicity with increasing planet mass is also observed in a sample of 22 longer-period, P greater than around 1 year, directly imaged giant planets, although in the currently small sample the median mass is around 12.5 Jupiter Masses, with the lowest-mass object at 2 Jupiter Masses, and the range of stellar metallicities is wide, spanning minus 0.65 to 0.30 dex.
Brewer et al in 20 18 confirmed that the metallicity trend for cool giant planets with Planet masses greater than 0.5 Jupiter Masses or Planet radii greater than 0.75 Jupiter Radii, at greater than 0.3 Astronomical Units, detected mostly through radial velocities, tracks that of hot giant planets for Iron to Hydrogen metallicity, greater than around minus 0.3 dex, although there is a hint of an increase in the fraction of planet systems hosting a cool Jupiter planet below this value.

The metallicity, period, eccentricity distribution of giant planets has also been suggested to reveal clues about their formation. Dawson and Murray-Clay in 20 13 set out to investigate the processes that deliver gas giant planets to their short orbital periods, around 10 to 100 days, around 0.1 to 1 Astronomical Units from farther out in the disk, beyond the ice line, where they most likely formed. These processes fall into two basic categories: migration smoothly through a disk that still contains substantial gas or migration more violently through gravitational interactions with stars or sibling planets.
Dawson and Murray-Clay in 20 13 pointed out a statistically significant difference in the eccentricity distributions of radial-velocity detected giant planets orbiting Iron to Hydrogen metallicity, greater than or equal to 0 versus Iron to Hydrogen metallicity less than 0 stars, such that those orbiting the more metal-rich stars have a wider range of eccentricities.
Additionally, they found that eccentric proto hot Jupiters experiencing tidal circularization also orbit primarily Iron to Hydrogen metallicity greater than 0 stars and that there are more short period giant planets around metal-rich versus metal-poor stars. They concluded that planet planet scattering is likely the culprit for the orbits of close-in giant planets around metal-rich stars, around which multiple planets are more likely to form and lead to dynamical interactions instability, whereas smooth disk migration does not have the metal-rich star requirement.
However, Yee and Winn in 20 23 recently used a more homogeneous and complete sample of transiting hot Jupiter (P less than 10 days) exo-planets to demonstrate that the period distributions of metal-rich versus metal-poor host stars do not show any difference. Their constraint on the lack of a detectable difference suggests that one planet delivery pathway, from the prescription of Nelson et al in 20 17, disk migration or high-eccentricity migration, dominates at the greater than or equal to 70 percent level, at least for the closest-in giant planets. However, Yee and Winn in 20 23 did not investigate planet eccentricity distributions in their sample, and they suggest that a new joint period-eccentricity analysis is warranted.
If, broadly speaking, giant planet formation via core accretion is increasingly likely with increasing host star metallicity, is there some cutoff metallicity below which giant planets will simply not form?
Johnson and Li in 20 12 investigated how the dust-to-gas ratio affects the time it takes for dust grains to settle in the disk midplane, after which planetesimal formation is expected to be relatively rapid versus the disk lifetime. Since the temperature and density of the disk change the dust-settling timescale, and these parameters change with distance r from the host star, the authors found that the critical dust-to gas ratio, critical Iron to Hydrogen metallicity, for planet formation depends linearly on log (Of r) and can be estimated with critical Iron to Hydrogen metallicity around minus 1.5 plus log R in Astronomical Units. Any planet systems found below this relation would be candidates for an alternative mode of formation besides core accretion, although Johnson and Li in 20 12 note several important assumptions that would not rule out core accretion for a planet in their “forbidden zone.”
How do observations compare to this prediction? Radial velocity studies of metal-poor solar-type stars, accounting for detection limits, suggest a lower limit of giant planet, around 1 to 4 Jupiter masses, P less than 3 years, formation around Iron to Hydrogen metallicity equals minus 0.7 dex, although they also suggest a flattening out of the occurrence rate metallicity relation below Iron to Hydrogen metallicity equals 0.
Recently, the Transiting Exo-planet Survey Satellite (TESS) provided coverage of a wide range of stellar types and around 85 percent of the sky, allowing Boley et al in 20 21 to constrain the occurrence rate of 0.8 to 2.0 Jupiter Radii, 0.5 to 10 day planets around minus 2.0 less than Iron to Hydrogen metallicity, less than minus 0.6 dex halo stars to be 0.04 to 0.36 percent, with a mean upper limit of 0.18 percent.
The authors proposed a similar lower Iron to Hydrogen metallicity limit as the radial velocity surveys, between minus 0.7 and minus 0.6 dex for hot Jupiter formation. Currently, the most metal-poor star known to host a hot Jupiter has Iron to Hydrogen metallicity equals minus 0.6 plus or minus 0.19 dex.
Several giant planets have been reported at longer orbital periods with lower Iron to Hydrogen metallicity values. The lowest, around Iron to Hydrogen metallicity around minus 0.7 dex, are 24 Boo b and HD 11 755 b, although the former is around an evolved giant star.

All these planets are still well above the forbidden zone predicted by Johnson and Li in 20 12. The metallicity limits of small planet formation are discussed further in Section 2 point 1 point 2, and Section 5 presents future directions in the search for planets around metal-poor stars.

2 point 1 point 2. Smaller planets.
Unlike short-period giant planets, the occurrence of smaller, less than 4 Earth Radii, and less massive, less than around 20 Earth mass planets generally has a weaker to nonexistent, depending on size, dependence on host star metallicity Figure 1. Indeed, it is the probable host star to two Planet Mass sin eye equals 5 to 7 Earth Mass planets that currently holds the lowest metallicity record at Iron to Hydrogen metallicity around minus 0.89 dex. This weak metallicity dependence is predicted by core accretion models, although different model prescriptions can change the expected number of small planets as a function of host star metallicity and or the metallicity at which the frequency of small low-mass planets may start to decrease due to the presence of giant planets.
As discussed by Zhu in 20 19, it is important to distinguish between the fraction of stars with a planet and the average number of planets per star; this distinction becomes especially relevant with small planets because they are often found in multi planet systems.
Zhu in 20 19 outlines tracers for each of these quantities and compares them for planets in the Kepler-LAMOST sample, finding that the occurrence of small planet systems increases slightly with host star Iron to Hydrogen metallicity, from minus 0.2 to 0.2 dex. A factor of 4.5 less than giant planets, but that the number of small planets per star does not depend as strongly on Iron to Hydrogen metallicity and may reach a plateau around Iron to Hydrogen metallicity equals 0.1 dex. However, when systems with at least four transiting planets are excluded, the plateau disappears, these high-multiplicity planet systems are concentrated in the range of Iron to Hydrogen metallicity between zero and zero point one dex, so they bias that bin in metallicity, and both the fraction of stars with planets and the average number of planets per star increase with stellar metallicity.
This finding is related to a key question in planet formation: How do outer giant planets affect inner smaller planets? Several studies have found that cold Jupiter planets appear (of order threefold) more often in systems that also host inner small planets. These studies do not all use the same giant outer and small inner planet definitions, but they find qualitatively consistent results). In particular, Zhu and Wu in 20 18 found a rise in the frequency of cold giant planets in systems with inner small planets around Iron to Hydrogen metallicity greater than 0.1 dex stars and a lack of systems with four or more transiting planets around Iron to Hydrogen metallicity greater than 0.2 dex stars. Brewer et al in 20 18 also found that the fraction of host stars of compact multi planet systems (with three or more planets in orbits less than 1 Astronomical Units is roughly flat in the range of Iron to Hydrogen metallicity from minus 0.3 to 0.4 dex but increases sharply below that range, where giant planets are infrequently found. A giant planet can destabilize a packed multi planet system, reducing the number of planets per star but not the number of planetary systems. Zhu in 20 19 suggests that their occurrence rate metallicity results support the idea that giant planets reduce the frequency of high-multiplicity small planets.

Rodriguez Martinez et al in 20 23 also recently found that M dwarfs hosting multi planet systems are significantly more metal-poor than M dwarfs hosting single-planet systems. These low-mass stars are generally expected to lack giant planets regardless of their metallicity, although this is an active area of research, so an increase in multi planet systems at low metallicity may be due to more than just the absence of giant planets.

Trends in the planet size, planet period, stellar metallicity space have also been illuminating for our understanding of small planet formation. On the basis of Kepler results, hot, P less than around 10 days, likely rocky, Planet Radius less than 1.7 Earth Radii planets seem to preferentially orbit more metal-rich stars, whereas at longer periods, these super-Earth planets show a flat or potentially decreasing occurrence of planets per star, with increasing Iron to Hydrogen metallicity, see Figure 1. While it was initially suggested that these so-called hot rocks might be hot Jupiter remnants, this seems unlikely, as the host stars of ultrashort-period planets have significantly different metallicity distributions than those of hot Jupiters, with the former being similar to stars hosting short-period sub Neptune planets. Thus, it is still possible that these hot rock planets are the remaining solid cores of smaller but still gas enveloped planets. Hot sub Neptunes are also found around more metal-rich stars as their occurrence is just not as strongly tied to metallicity as hot Jupiters, see Figure 1, and at longer periods they still show a slightly increasing occurrence of planets per star, with metallicity.
Alternatively, the correlation between the occurrence of short period small planets and their host star metallicity could indicate that close-in planet formation is more efficient around high-metallicity stars through, for instance, more efficient inward migration of solids, a closer-in dust sublimation radius, or a close-in co-rotation radius, where gas is accreted onto the stellar surface, although the latter two scenarios seem unlikely.
Looking beyond hot rocky planets, Owen and Murray-Clay investigated in 20 18 the same California-Kepler survey sample as Petigura et al in 20 18 in the context of atmospheric loss, specifically at short periods, P less than 2.5 days, where atmospheric evaporation is expected to be very effective, and P greater than 25 days, where atmospheric evaporation should be minimal.
They pointed out that.
(a) both short and long-period sub-Neptunes are larger around higher-metallicity stars,
(b) short-period super-Earths are also larger around higher-metallicity stars, and
(c) long-period super-Earths are more common around low-metallicity stars but their radius distribution is metallicity independent.
The authors provide a physically motivated framework tying together these results: The formation of super-Earth and sub-Neptune planets around higher-metallicity stars results in larger higher-mass solid cores, at short periods, where atmospheres should be mostly stripped, all small planets are larger around higher-metallicity stars. Furthermore, increasing host star metallicity may result in increasingly metal-rich gaseous envelopes for the planets massive enough to accrete them, meaning these planets can better “resist” evaporation at shorter orbital periods versus those in lower-metallicity systems. The longer-period planets with enough mass to be able to collect larger gaseous envelopes and thus increase their size are also larger around more metal-rich stars. The longer-period, 1 to 1.8 Earth Radius planets are more frequent around more metal-poor stars, which presumably have less material available in the disk to form larger cores fast enough to accrete Hydrogen, Helium envelopes. Thus, these planets may have originated as rocky and never accreted a significant nebular gas component, pointing to a later stage of small planet formation after the disk dispersed around more metal-poor stars. This hypothesis presents the intriguing possibility that “born rocky” planets form in significantly different chemical environments and thus may have different internal compositions, a topic revisited below in Section 4 point 2.
2 point 2. Beyond Iron Abundances.
While Iron is the most easily and frequently measured element in most solar-type star spectra and a standard metric of total metallicity, as astronomers we have a plethora of metals to choose to measure in host stars, the abundances of which may also be important for planet formation.

As mentioned in Section 1, there are moderately well characterized correlations between Iron and other metals as a result of G C E. Therefore, to isolate the contribution of other metals, we can consider X to Iron ratios, normalizing the other metal X by the star’s Iron abundance. However, stellar chemistry also traces kinematics and age, so it can be difficult to determine which of these factors is most influential in planet formation. Overall, most comparisons of abundances and abundance ratios, for example, Carbon to Oxygen, Magnesium to Silicon, between stars that host and stars not known to host planets have found no meaningful differences that are not explained simply by G C E, regardless of planet formation, but see Robinson et al in two thousand and six for an example of significant enhancement of Silicon and Nickel in planet host stars when controlling for Iron.
A significant and important finding in planet occurrence beyond just Iron host star abundances is that X to Iron ratios for a suite of elements including Magnesium, Aluminum, Silicon, Scandium and Titanium are systematically higher at low Iron to Hydrogen metallicity, usually less than around minus 0.3 dex, in host stars versus stars not known to host planets. While the enhancement of X to Iron ratios at low Iron to Hydrogen metallicity is found in host stars to Jupiters, Neptunes, and super-Earths, the overall frequency of larger and more massive planets decreases with Iron to Hydrogen metallicity, so the trend is dominated by the smaller planets.
However, Maldonado et al reported in 20 18 that cool Jupiter host stars show an enhancement in alpha elements in comparison to stars hosting hot Jupiters.
Notably, this X to Iron ratios enhancement does not appear to be the case for Oxygen, even though it is an alpha element like Silicon, is incorporated into many rock-forming minerals, and is the major constituent of Earth’s crust. This finding may indicate that, particularly in Iron-poor environments, it is the initial grain nucleation, perhaps more related to the Silicon abundance, versus growth by accretion of ices, perhaps more related to the Oxygen abundance, that serves as the limiting step to planet formation and or that the solid compositions of these planets will be significantly different. Planet host stars are also not preferentially enhanced in other elements typically associated with volatile species like Carbon and Nitrogen, although Wilson et al in 20 22 found a statistically significant trend in Sulphur to Iron ratios with both planet radius and period in their analysis of the APOGEE-KOI program. This trend cannot be explained by other cofounding correlations, and its source remains unknown.
The fact that planet host stars do not stand out in many stellar abundance contexts does not mean that there is a lack of variety in their compositions. Indeed, especially for small planets, the range of chemical environments for planet formation spans the entire Galaxy, and perhaps beyond, if some small percentage of stars currently within the Milky Way actually originated in a neighboring dwarf galaxy. The implications of this chemical variety on planet compositions are explored further in Section 4.

Three. ANOMALOUS ABUNDANCES.
Within the study of host star abundances, a few specific populations stand out as particularly useful and interesting for isolating the effect of planet formation and, thus, for better pinpointing where it occurred and whether or not planets have been detected there, yet.
3 point 1. Solar Twins and Siblings.
While the giant planet metallicity correlation, Section 2 point 1, was identified early on in the history of exo-planet studies, more subtle stellar abundance signatures related to planets, particularly small rocky ones, came with increasing planet detections as well as more precise approaches for measuring stellar abundance differences. As discussed in Section 1, the highest-precision stellar abundances, and, thus, sensitivity to the smallest differences, come from comparing similar stars differentially, since systematic errors in model atmospheres and or line parameters effectively cancel out.

Solar twins: stars with an Effective temperature within 100 Kelvin of solar, surface gravity (log g) within 0.1 dex of solar, and metallicity, Iron to Hydrogen metallicity, within 0.1 dex of solar.

The star we know best and can thus make the most accurate comparisons with just so happens to have formed rocky planets. So, it makes sense to investigate whether the Sun’s abundances offer a signature of rocky planet formation, which could then help narrow down the number of stars around which we should spend significant time looking for similar planetary systems.
The first promising evidence of a main-sequence stellar abundance signature related to rocky planet formation was presented by Melendez et al in two thousand and nine, who conducted a differential spectroscopic study of 11 solar twins using R around 65,000, Signal to Noise around 400, and 340 to one thousand nano meter spectra, all collected with the same instrument. The parameters of the spectra like resolution and Signal to Noise are important because they can influence the systematic errors that can arise from subtle mismatches when using spectra from different instruments or configurations.
Melendez et al measured abundances with very small internal errors around 0.01 dex and showed that the Sun is preferentially enhanced in volatile elements, low condensation temperature, T cond, but depleted in refractory elements, high condensation temperature, compared with the solar twins, see Figure 2. The authors suggested that this abundance, condensation temperature trend in the Sun was indeed a beacon for rocky planet formation, the refractories are “missing” from the Sun because they were incorporated into the terrestrial planets, which Chambers confirmed in 20 10 could be matched by adding 4 Earth Masses of a mixture of Earth-like and carbonaceous chondrite like material to the present-day solar convection zone.
As exciting as this suggestion is, there are several alternative explanations and additional factors to consider. Some were pointed out by Melendez et al in two thousand and nine themselves and others were identified later, as our knowledge of exo-planets grew. First, while the comparison stars were solar twins, they were not necessarily formed at the same time or location as the Sun, so they could have experienced different effects of G C E that influenced their X to Iron ratio, Condensation temperature trends.
When this G C E influence, a linear trend in X to Iron ratio with age, is fitted and removed, the Sun may still be anomalous relative to the majority of its twins, Bedell et al in 20 18 found that the Sun is more deficient in refractories than 82 percent of a larger sample of 68 solar twins corrected for G C E effects, but see the different estimates of G C E effects in Cowley and Yuce from 20 22. Note, however, that it is impossible to tell the difference between refractory depletion in the Sun and refractory enhancement in other comparison stars, as their primordial compositions are unknown. It has also been suggested that the Sun’s refractory depletion signature could arise if the gas from which it formed was initially cleansed of dust by radiation pressure from nearby massive stars.
Second, it is important to consider the timing of any event leading to an enhancement or depletion signature in stellar abundances in relation to the size of the stellar convection zone. The size of the young, less than around 10 Mega year Sun’s convection zone was likely 50 to 100 percent by mass versus 2.5 percent today, which means that more like 80 to 160 Earth Mass of rocky material would be needed to produce the Sun-solar twin difference if it arose during the first around 10 Mega years, much more than is bound in Solar System rocky bodies. The mass of solids in the larger planets in the Solar System could reach the lower values in this range, but given the high fraction of volatile-rich ices in these solids, their missing mass from the Sun does not obviously produce the refractory depletion trend.
However, as discussed below in Section 3 point 2, planet engulfment is a feasible alternative, assuming that it occurs at later stages when less material would be required to produce an abundance change in the stellar photosphere.

Figure 2.
(a) Trends in differential stellar abundances versus condensation temperature, 50 percent in solar composition gas, for various twin binary systems, colored symbols, and for the Sun minus solar twins (black points). Individual elements are listed at the top, and the horizontal gray dashed line indicates zero difference. The figure illustrates the challenge of interpreting abundance condensation temperature trends. The precision on these abundances is extremely good, and yet not all systems show an obvious trend like the refractory depletion observed in the Sun, although the change in X to Iron ratio values generally decrease overall with decreasing separation between the binary pair.
(b) Zoomed-in view of the condensation temperature range investigated by Nibauer et al in 20 21, plotting the two distinct trends found in a sample of around 1,800 solar analogs from APOGEE-2. Most of the stars show a small contrast in the ratio of volatile to refractory abundances, green dashed line, shaded uncertainty, while 10 to 30 percent show a greater contrast between volatile and refractory abundances, blue dashed line, shaded uncertainty. The Sun belongs to the more frequently occurring small-contrast population (green), suggesting that it may not be so anomalous when compared with other solar analogs. In all of the binary star cases, the figure shows hotter, colder star abundances for consistency because, in some cases, both stars host planets.
When the tabulated measurement is X to Hydrogen ratio, it is normalized by each star’s Iron to Hydrogen metallicity to get X to Iron ratio. The neutral species abundance is taken when both neutral and singly ionized are listed, and the atomic abundances are taken when both atomic and molecular are listed. Lithium abundances are not plotted. These are usually measured by synthesis fitting to an individual line. The gray black error bars for the Sun-solar twins include observational errors both in the Sun and solar twins and in the solar twins alone.
Abbreviation: P S, primary secondary.

Solar analogs: stars with similar parameters to the Sun (for example, plus or minus 500 Kelvin, plus or minus 0.3 dex in Iron to Hydrogen metallicity, but not as similar as solar twins. Could be yet-to-be-confirmed solar twins requiring better parameter determination.

Third, we now know that small planets occur relatively frequently and giant planets much less so. So, it is hard to reconcile the fact that a large fraction of Sun-like stars would lack a similar signature of rocky planet formation. Also, multiple studies have been unable to confirm the same rocky planet X to Iron ratio, Condensation Temperature signature in Sun-like, albeit not twin, stars known to host close-in rocky planets.
Instead, perhaps, the X to Iron ratio, Condensation Temperature trend in the Sun relative to solar twins is a beacon of giant planet formation. As described by Booth and Owen in 20 20, when giant planets form early and open gaps in the protoplanetary disk, this process can preferentially trap dust while allowing refractory-poor gas to flow through toward the star. These authors modeled the evolution of protoplanetary disks surrounding young stars and matched the observed around 0.04 dex refractory depletion observed in the Sun, although they did not attempt to reproduce the specific trend with T cond. Additional evidence supports such preferential refractory material trapping from observations of Herbig “A e” stars, young “A” stars with very thin, if any, convection zones at their surfaces, allowing recently accreted material to dominate the photospheric composition.

Solar siblings: stars that formed from the same stellar cluster at the same time as the Sun but may not still be close in distance to the Sun today; they have similar composition and age but not necessarily other parameters.

A comparison of Herbig “A e” stars hosting cold flat disks with stars hosting transitional disks with large inner cavities depleted in millimeter-sized dust grains shows that the second group is depleted by a factor of around 10 in Iron, Magnesium and Silicon but not in Carbon or Oxygen. Further research also showed depletion in Titanium and the intermediate condensation temperature element Sulphur. Additional studies of stars hosting distant gas giants could help shed light on this hypothesis of an X to Iron metallicity and Condensation Temperature trend indicating giant planet formation.
While studies searching for refractory depletion in solar twins have focused mostly on relatively small samples with high resolution and Signal to Noise spectra, Nibauer et al in 20 21 applied a likelihood-based approach including a hierarchical mixture model to the less-precise abundances of 1,800 solar analogs in the APOGEE-2 catalog to assess whether there is evidence for two populations of stars, depleted and not depleted in refractories.
The authors focused on the most precisely determined APOGEE-2 abundances for dwarf stars and measured only the X to Iron metallicity and Condensation Temperature trend across a limited range, Silicon, Magnesium, Nickel, Calcium, and Aluminum, covering T c around 1,300 to 1,650 Kelvin. They indeed found evidence for two populations; the not-depleted population had a significant slope across T cond, Figure 2b. Interestingly, they found that around 85 percent of stars are in their depleted population, with an approximately flat slope across a Condensation Temperature ranging from 1,300 to 1,650 Kelvin, Figure 2b, and that while the Sun is still more depleted than around 80 percent of their full sample, it still falls in the middle of a distribution of stars with similar X to Iron metallicity and Condensation Temperature trends.
This finding calls into question just how anomalous the Sun and or some level of refractory-to-volatile depletion is among stars fairly nearby in the Galaxy. Similarly, Behmard et al found in 20 23 that APOGEE-2 abundances of Kepler planet host stars and matching doppelganger stars, with the same fundamental parameters but no knowledge of planets, have indistinguishable condensation temperature trends, and there is nothing special about the planet hosts, although the authors note that they are limited in part by abundance precision.

3 point 2. Binary Stars.

A natural extension of studying stars that are twins of the Sun is studying stars that are twins of one another, binary stars that formed in the same place at the same time and are very similar in mass, metallicity, and temperature. Spectra of these stars can be analyzed via the same strictly differential technique that minimizes systematic errors and thus results in high-precision abundances, which are not subject to different influences of G C E. Thus, differences in otherwise identical binary star abundances could be related to planet formation. Extending this idea even further, the pattern of differences in the abundances might reveal something about the type of planet formation that occurred.
At first, the giant planet metallicity correlation, Section 2 point 1, was thought to potentially be caused by the pollution of the stellar convective envelope by planet or planet-forming material.
This hypothesis motivated astronomers to search for large abundance differences in wide binary systems. Initial line-by-line differential studies of moderate samples of tens of stars found a few outlying cases of large greater than around 0.1 dex abundance differences but indicated that it is uncommon for stars to ingest enough material to cause a significant difference and, thus, the clear enhancement in metallicity among giant planet host stars. As discussed below, planet engulfment may indeed be happening, but its impact on stellar abundances is predicted to be more subtle.

Interest in studying binary stars, particularly twins, was renewed by the solar twin studies discussed above, as well as by increasing numbers of exo-planet detections. It became possible to compare two stars in a binary system in which one was known to host one or more planets, or in which the stars were known to host different types of planets, providing additional context for measured abundance differences. Table 1 presents a selection of detailed binary star abundance differences studies, and Figure 2 plots a subset. Perhaps the most popular example is 16 Cygnus A and B, a pair of stars that themselves are solar twins, wherein B hosts an eccentric super-Jupiter planet at nearly 800 days. While many authors have measured the pair and found the “A” component, is slightly hotter, to be enriched in metals by around 0.04 dex, whether or not there is a trend in the change in X over Iron metallicity and Condensation Temperature has been debated.

The latest results from the highest Signal to Noise spectra indicate a slope in change in X over Iron A minus B metallicity and Condensation Temperature of 1.56 plus or minus 0.24 time ten to the minus five dex per Kelvin. This trend was originally attributed to the effect of the rocky core of 16 Cygnus B b, the B star is preferentially depleted in refractories, but Maia et al in 20 19 also found the A component to be overabundant in Lithium, above what is expected for its age, as well as in Beryllium. This finding suggests instead that the A minus B abundance trend is better explained by engulfment of 2.5 to 3 Earth Masses of Earth-like material by 16 Cygnus A.
Overall, the picture of twin binary star studies is muddy, as summarized in Table 1. Different combinations of systems with without planets, with and without abundance differences, and with and without trends with Condensation Temperature exist in the literature. Most abundance differences between twin binaries, when detected, are at the 0.05 dex level or lower, emphasizing the extreme precision in these studies, Figure 2.
There are a few notable exceptions of more significant differences, the HD 240 430 slash 29 and HIP 34 407 slash 26 systems show differences at the 0.1 to 0.2 dex level, but it is challenging to draw conclusions from the majority of results, especially given that many stars may host planets that remain undetected.
Let us take a step back to answer a more fundamental question: What is the typical difference in binary star abundances? The answer has implications for chemical tagging, using the chemical similarity of stars that are no longer physically close to one another to trace them back to their birth cloud and siblings, as well as metallicity calibrations for cool M dwarf stars, where an M dwarf star is assumed to have approximately the same composition as the higher mass companion.
Aside from planet formation and inhomogeneous molecular clouds, relatively small offsets in The Effective temperature and log g between stars can lead to different surface abundances through different levels of internal turbulence and, hence, differential atomic diffusion. The high-precision astrometry from Gaia has made it possible to greatly expand the sample of wide binaries and co-moving pairs with which we can assess typical binary abundance differences. It appears that within wide, around 300 to 50,000 Astronomical Unit projected separation.
The Effective temperature less than 200 Kelvin binary systems, around 80 percent of pairs are homogeneous at the 0.02 dex level in Iron to Hydrogen metallicity and similarly for a range of other elements, far more similar than randomly paired field stars of similar spectral type. Going further in separation, even very similar co-moving pairs, delta three d velocity less than 2 kilometers per second, separated by greater than about ten to the five Astronomical Units, which are no longer bound, still have a relatively high fraction, around 70 percent, of homogeneity versus random pairings, as least for Iron to Hydrogen metallicity, standard deviation of 0.08 versus 0.23 dex or 0.109 versus 0.215 dex. The larger sample used by Yong et al in 20 23 shows a natural break point in the fraction of chemically homogeneous co-moving pairs at one million Astronomical Units, at closer separations, around 70 percent of their pairs have
Delta Iron to Hydrogen metallicity, over standard deviation of Iron to Hydrogen metallicity less than or equal to 3.0 and delta Iron to Hydrogen metallicity less than 0.04 dex, versus only around 18 percent at larger separations.

Table 1. Selection of detailed abundance studies of binary stars.

The parameters are: System. Average Effective temperature in Kelvins,
Absolute change in Effective temperature in Kelvins, log g average, Absolute delta log g, Average Iron to Hydrogen metallicity dex, Absolute change in Iron to Hydrogen metallicity dex, Separation in Astronomical Units, References, and Comments.

Given that around 20 to 35 percent of binary stars may be chemically distinct at an above-average level, does this mean that all of these systems experienced planet engulfment? Spina et al in 20 21 combined 31 new binary pair measurements with 76 from the literature in a meta-analysis across a wide temperature range, around 5,000 to 6,450 Kelvin and found a higher fraction of chemically distinct pairs, defined as the absolute variation of Iron to Hydrogen metallicity greater than 2 times the max of either the standard deviation of the change in Iron to Hydrogen metallicity, and 0.01, with increasing average Effective temperature of the two stars, which is in line with predictions from planet engulfment events, since hotter stars have thinner convection zones that can be contaminated more easily.
This result was recently bolstered by Yong et al in 20 23, who included a more homogeneous analysis with all abundances measured in the same way, line by line and a larger sample of 125 co-moving pairs and found a similar trend of increasing chemical inhomogeneity with increasing average Effective temperature. However, not all of the pairs in these studies were “twins”, required the pairs to have delta Effective temperature less than 600 Kelvin and delta log g less than 0.6 dex.
Yong et al in 20 23 imposed the absolute variation in BP minus RP less than 0.15 mag and absolute variation in G magnitude less than 1 mag, resulting in delta Effective temperature less than ort equal to 561 Kelvin and delta log g less than 0.4 dex. As mentioned above, different stellar evolutionary states can contribute to abundance offsets from the intrinsic effects of atomic diffusion, unrelated to planet engulfment.
Behmard et al recently addressed the planet engulfment question with MESA, building on past research to model how planet engulfment signatures are influenced by a variety of stellar mixing processes, including convection, thermohaline mixing, diffusion, gravitational settling, and radiative levitation.
Their models tested the impact of the accretion of 1 to 50 Earth Masses of bulk Earth composition material onto 0.5 to 1.2 Solar mass stars at very early-stage, reaching zero-age main sequence (ZAMS), late heavy bombardment, 500 Mega year, disk accretion, 10 Mega year, and late-stage, 300 Mega year to 3 giga-year post-ZAMS times.
The length of time for accretion also varied depending on the time at which it began.
At later times, the convection envelopes of Sun-like stars are smaller and thermohaline mixing decreases, both potentially increasing the detectability of an accretion signature, defined by Behmard et al as greater than 0.05 dex. Subsolar metallicity models also result in more detectable signatures due to the larger contrast of refractory enrichment above the background metallicity and the stars’ smaller convective envelopes, more massive stars also have smaller convective envelopes. Across the range of engulfment scenarios, most do not produce observable abundance difference signatures. The most promising candidates are stars older than 1.5 Giga year with 1.1 to 1.2 Solar masses, particularly those with low metallicity, but even in these cases the engulfment would need to occur well after ZAMS. Behmard et al in 20 23 also applied this planet engulfment model to their own Keck-HIRES measurements of 36 multi-star systems in which at least one star is known to host a planet. They found only a single case (HAT-P-4) that met their criteria for a possible engulfment signature (including matching a model with greater than or equal to 10 Earth Masses of engulfed mass. However, the engulfment must have happened late in the star’s lifetime, within the last 2 Giga-year, and the large projected separation of the components, around 30,000 Astronomical Units suggests that the abundance differences could simply indicate an inhomogeneous, patchy composition of the birth molecular cloud. So, while greater than around 0.02 dex abundance differences between binary stars may occur in something like a quarter of systems, planet engulfment is unlikely to be the culprit in most cases, closer to around 3 percent or around 8 percent.

Also, it is really only in the Delta Effective temperature less than around 200 Kelvin systems that reliable engulfment signatures can be detected, as a result of refractory depletion rates varying with stellar type.
3 point 3. Lithium.
Lithium abundances are a challenging but potentially powerful tracer of the physical processes occurring in stellar interiors. The Lithium that a star is born with is destroyed at relatively low temperatures, around 2.5 million Kelvin in stellar interiors by proton-capture reactions, primarily in the premain-sequence stage. Broadly speaking, for F G K M dwarfs, cooler, less massive stars have lower Lithium than hotter, more massive stars because of their deeper convective zone envelopes, which bring material down to the temperature required to burn Lithium.
By the end of the main sequence, lowermass stars can be depleted by factors of 100 from the original abundance. Standard solar interior models that consider only convective mixing do not predict Lithium depletion in solar-temperature stars, and inferences from helioseismology indicate that the base of the Sun’s present day convection zone envelope is just cool enough to avoid burning Lithium, so the Sun should not show Lithium depletion at the surface. And yet, the Sun itself is depleted in Lithium by more than two orders of magnitude in comparison to the meteoritic values that benchmark Solar System formation, and within the solar neighborhood, Lithium abundances vary widely across stars with similar properties.
Furthermore, while most cooler stars, with deeper convective envelopes, indeed show significant Lithium depletion, even some stars hotter than the Sun, with shallower convective envelopes, still exhibit Lithium abundances lower than traditional models. Lithium abundances also depend on stellar metallicity, age, rotation, and activity level, including in the open cluster M67.
Thus, evidence suggests that additional mixing must be operating inside solar-type stars to destroy Lithium, meaning that Lithium abundances offer an opportunity to understand these mechanisms that are otherwise invisible from the surface.
Many such mechanisms have been proposed, rotational mixing, overshooting mixing, internal gravity waves, and atomic diffusion, but which of them dominates, or which combination and in what proportion, is still an area of active research. Specifically of interest for this review, planet formation has been invoked as a way to influence stellar Lithium abundances. Interactions with the protoplanetary disk can produce rotational breaking in the star, contributing to differential rotation and potentially inducing hydrodynamical instabilities that can enhance Lithium burning.
Planetary migration can deplete Lithium by affecting the angular momentum of stars, creating a shear instability that produces effective mixing. Episodic accretion events might also affect mixing via thermohaline convection and or increase the temperature at the bottom of the convective envelope and thus the effectiveness of Lithium depletion. It is also possible that planet-related processes, accretion of sufficient planetary material, tidal orbital decay, could enhance Lithium if they occur late enough in the star’s lifetime and deliver fresh Lithium after the majority of the original Lithium has been depleted. Lithium enhancement has been observed in both main-sequence and more evolved (giant) stars, but for simplicity, this article focuses on solar-type stars.
If any of these planet-related mechanisms were causing Lithium depletion, or enhancement, we might expect to find that host stars are preferentially Lithium poor or rich versus stars without known planets. However, as described above, isolating the source of any difference in Lithium abundances is difficult, given the multitude of variables at play, and sample biases as well as how the abundances were measured in different studies need to be carefully considered.

Authors have been investigating whether stars with and without known planets show differences in Lithium abundance since not long after the first detection of exo-planets.
Even as samples of high-resolution stellar spectra and exo-planets have grown, the consensus seems to still be that there are no significant differences in Li abundances between host and non host stars, except perhaps within narrow parameter ranges, particularly around the solar effective temperature, where the Lithium abundances are most sensitive to small changes in mixing mechanisms.
There are also one-off cases with large differences in Lithium abundances between binary stars, see Section 3 point 2, that cannot be explained even with more complicated stellar models that include additional transport mechanisms.
Interestingly, Sevilla et al in 20 22 recently showed that, with stellar models including additional mixing processes, convective overshoot mixing, thermohaline mixing, atomic diffusion, planet engulfment can indeed produce detectable Lithium enhancements in solar-type 0.8 to 0.9 Solar mass stars but only for around 1 Giga year after engulfment.
This finding suggests that, in a binary system, only a fairly recent planetary engulfment event can explain a large Change in Lithium abundance, and the interpretation of such an offset would also need to take into account any difference in the stars’ properties, primarily masses.

Four. STELLAR COMPOSITION AS AN INDICATOR OF PLANET FORMATION AND COMPOSITION.

Going beyond looking for correlations between host star abundances and the presence of different types of planets, we can use the stellar comp