Draft version July 22, 2024

Typeset using LATEX twocolumn style in AASTeX63

The First Evidence of a Host Star Metallicity Cut-off In The Formation of Super-Earth Planets
Kiersten M. Boley,1, ∗ Jessie L. Christiansen,2 Jon Zink,3 Kevin Hardegree-Ullman,4 Eve J. Lee,5
Philip F. Hopkins,6 Ji Wang (王吉),1 Rachel B. Fernandes,7, 8, † Galen J. Bergsten,9 and Sakhee Bhure10
1 Department of Astronomy, The Ohio State University, Columbus, OH 43210, USA
2 Caltech/IPAC-NASA Exoplanet Science Institute, Pasadena, CA 91125, USA
3 Department of Astronomy, California Institute of Technology, Pasadena, CA 91125, USA‡
4 Department of Astronomy, The University of Arizona, Tucson, AZ 85721, USA
arXiv:2407.13821v1 [astro-ph.EP] 18 Jul 2024

5 Department of Physics and Trottier Space Institute, McGill University, Montreal, QC, H3A 2T8, Canada
6 TAPIR, MS 350-17, Caltech, 1200 E. California Blvd, Pasadena, CA 91125
7 Department of Astronomy & Astrophysics, 525 Davey Laboratory, The Pennsylvania State University, University Park, PA 16802, USA
8 Center for Exoplanets and Habitable Worlds, 525 Davey Laboratory, The Pennsylvania State University, University Park, PA 16802,

USA
9 Lunar and Planetary Laboratory, The University of Arizona, Tucson, AZ 85721, USA
10 Centre for Astrophysics, University of Southern Queensland, Toowoomba, QLD 4350, Australia

ABSTRACT
Planet formation is expected to be severely limited in disks of low metallicity, owing to both the
small solid mass reservoir and the low opacity accelerating the disk gas dissipation. While previous
studies have found a weak correlation between the occurrence rates of small planets (≲4R⊕ ) and stellar
metallicity, so far no studies have probed below the metallicity limit beyond which planet formation
is predicted to be suppressed. Here, we constructed a large catalog of ∼110,000 metal-poor stars
observed by the TESS mission with spectroscopically-derived metallicities, and systematically probed
planet formation within the metal-poor regime ([Fe/H] ≤ −0.5) for the first time. Extrapolating
known higher-metallicity trends for small, short-period planets predicts the discovery of ∼68 super-
Earths around these stars (∼ 85,000 stars) after accounting for survey completeness; however, we
detect none. As a result, we have placed the most stringent upper limit on super-Earth occurrence
rates around metal-poor stars (-0.75 < [Fe/H] ≤ -0.5) to date, ≤ 1.67%, a statistically significant (p-
value = 0.000685) deviation from the prediction of metallicity trends derived with Kepler and K2. We
find a clear host star metallicity cliff for super-Earths that could indicate the threshold below which
planets are unable to grow beyond an Earth-mass at short orbital periods. This finding provides a
crucial input to planet formation theories, and has implications for the small planet inventory of the
Galaxy and the galactic epoch at which the formation of small planets started.

Keywords: planet formation, TESS, planet, occurrence,transiting exoplanet, exoplanet evolution

1. INTRODUCTION Lee et al. 2014; Lee & Chiang 2015). This leads to the
Under the core accretion paradigm, planet formation expectation of a strong correlation between the occur-
begins from the coagulation of solid material. Once rence rate of gas giants and the amount of solid material
this rocky core becomes massive enough (i.e., when its in the disk, for which the metallicity of the host star is
Bondi radius exceeds the core radius), gas accretion be- an excellent proxy (Johnson & Li 2012; Hasegawa & Hi-
gins (e.g., Pollack et al. 1996), and the mass of the core rashita 2014; Lee 2019). This giant planet-metallicity
ultimately determines the amount of accreted gas (e.g., correlation has been well constrained from radial veloc-
ity surveys (Gonzalez et al. 2001; Fischer & Valenti 2005;
Udry & Santos 2007; Johnson et al. 2010; Wang et al.
[email protected] 2015). On the other hand, the super-Earth-metallicity
∗ NSF Graduate Research Fellow correlation is expected to be weaker (Lee 2019), and in-
† President’s Postdoctoral Fellow deed only emerges in the much larger sample available
‡ NHFP Sagan Fellow
2

via the Kepler transit survey (Kutra et al. 2021; Lu et al.

2020; Zhu 2019; Petigura et al. 2018; Zink et al. 2023).
While the observed correlation is weak, there are many
theoretical reasons to suggest that there may be a crit-
ical threshold metallicity below which the formation of
super-Earths becomes difficult, if not impossible. A
lower limit to metallicity could arise from the physics
of dust grain-grain collisions and dust growth coupled
with metallicity-dependent disk evolution (Johnson &
Li 2012), instabilities believed to trigger planetesimal
formation (Youdin & Goodman 2005; Johansen et al.
2007; Bai & Stone 2010; Squire & Hopkins 2018; Li &
Youdin 2021), or the necessity for planetary protocores
to accrete solids from their environment via planetesimal
(Kokubo & Ida 1998; Goldreich et al. 2004) or pebble ac-
cretion (Ormel & Klahr 2010; Lambrechts & Johansen
2014; Lin et al. 2018). While the qualitative expecta-
tion of some threshold is generic to most models, actual
predictions of where that threshold should lie vary sig-
nificantly with values from 0.003 times solar metallicity Figure 1. The stellar sample from TESS portrayed in the
(see Johnson & Li 2012, their equation 10, evaluated Tef f and log(g) plane. We show the main sample (-1≤[Fe/H]
at 0.1 AU) to solar or supersolar metallicities (see Li ≤ -0.5) in purple, and the sub-sample (-0.5<[Fe/H] ≤ -0.25)
& Youdin 2021, their equation 14) although the latter in blue.
value is sensitive to the Stokes number of the coagulat-
ing dust grains (e.g., the critical minimum metallicity metallicity below which such a drop in occurrence rate
can be lower than the solar value if the Stokes number arises.
is large ∼0.1). The outline of our paper is as follows: Section 2 pro-
Because of the well-studied stellar age–metallicity cor- vides a description of our stellar sample. In Section 3,
relation in the Galaxy, the value of this critical metal- we discuss our planet detection pipeline. Section 4 con-
licity threshold directly impacts when the earliest small tains a description of our forward model software. We
planets were formed in the Galaxy − the oldest stars discuss our results in Section 5. In Section 6, we com-
have metallicities significantly below the predicted range pare our results to previous studies and their implica-
of values. Therefore, they would not be expected to form tions for super-Earth formation before summarizing our
these planets. The super-Earth-metallicity correlation conclusions in Section 7
measured to date has been limited to host stars with
metallicities greater than [Fe/H] = −0.4, due to the fo- 2. STELLAR SAMPLE
cus of Kepler on Sun-like stars, and the relative rarity Using the TESS Input Catalog (TIC; Ricker et al.
of metal-poor stars in the local solar neighborhood. 2015a), we constructed a sample of metal-poor stars.
To date, there has been no large, systematic search for We included TESS data from Sectors 1–55, and apply
small planets orbiting metal-poor host stars. However, the following cuts:
the Transiting Exoplanet Survey Satellite (TESS, Ricker
1. Spectroscopically-derived metallicities: The
et al. 2015a) enables consideration of the long-standing
TIC compiled data from many surveys into a cata-
question: what is the critical metallicity required for
log of ∼1.8 billion sources; thus, the spectroscopic
small planet core formation? Given that TESS is an all-
metallicities are heterogeneous. The vast major-
sky survey providing high-precision photometry on mil-
ity of these sources have parameters derived solely
lions of stars, the number of metal-poor stars surveyed
from photometric surveys, but the TIC also com-
is orders of magnitude greater than its predecessors, K2
piled data from the available spectroscopic sur-
and Kepler, enabling large population studies. Here, we
veys. To test existing planet formation theories,
perform the first analysis of the occurrence rates of small
we included all stars with spectroscopically de-
planets at metallicities approaching the predicted cut-
rived parameters having [Fe/H] < −0.5, hereafter
offs for planet formation. We empirically test whether a
known as the main sample. In addition, we in-
metallicity cut-off exists and quantitatively address the
cluded a subset of stars with metallicities between
 3

−0.5 and −0.25 to compare and cross-check our Parameter Range Median Units
TESS planet occurrence rates with previous re- Mass 0.71 − 1.24 1.05 M⊙
sults from Kepler and K2, referred to as the con- Radius 0.71 − 1.65 1.17 R⊙
trol sample hereafter. From this initial cut, we log(g) 4.04 − 4.63 4.31 cgs
began with a main sample of 654,675 stars and TESS Magnitude 10.58 − 13.91 13.09 mag
control sample of 987,924 stars. Temperature 4646 − 6376 5819 K

2. Stellar effective temperature: We limited our Table 1. Stellar Sample Parameters: We show the 5% to
sample to stars with effective temperatures rang- 95% quantiles for each parameter range of our sample from
the TIC (Stassun et al. 2018a)
ing from 4000–6500 K. This requirement selects
FGK spectral types and allows for direct compar-
ison with previous higher metallicity Kepler and
K2 results. From this requirement, we excluded
32,609 main sample stars and 93,694 control sam-
ple stars.

3. Surface gravity: We excluded low surface grav-

ity red giants by requiring log g > 4.0 and removed
291,095 main sample stars and 377,648 control
sample stars.

4. TESS magnitude (Tmag ): We selected stars

brighter than Tmag = 14; at fainter magnitudes,
our detection efficiency for small planets declines
rapidly. From this cut, we excluded 232,312 main
sample stars and 271,130 control sample stars.

5. Galactic Plane: We excluded stars with |b| <

5◦ to mitigate the effect of false positives due to
Figure 2. The probability density function (PDF, %) of the
crowding near the Galactic plane, removing 2,265 metallicities within our main sample (−1 ≤[Fe/H]≤ −0.5) by
main sample stars and 33,869 control sample stars. survey. 90% of the stars within our sample have metallicities
from LAMOST.
6. Removal of Binary Stars: Similar to previous
works, we relied on the Gaia Renormalized Unit
Weight Error (RUWE) metric to minimize this po- sample cuts, resulting in a total of 84,809 metal-poor
tential source of contamination and only included stars with metallicities below [Fe/H] ≤ −0.5.
targets with RUWE<1.4 (Stassun & Torres 2021; To enable a comparison with previous studies (e.g. Pe-
Lindegren et al. 2018). Additionally, Gaia DR3 tigura et al. 2018; Zink et al. 2023), we also included a
provides a flag, “non single star”, which denotes control-sample of stars within the metallicity range of
sources that provide evidence of a binary (Creevey −0.5 ≤ [Fe/H] ≤ −0.25. In doing so, we account for any
et al. 2023). Removing targets with Gaia binary systematic difference between the combined Kelper and
flags and high RUWE values, we excluded 20,867 K2 occurrence rates and TESS. From the criteria listed
potential binary targets from that main sample above, our control sample consisted of 205,832 stars.
and 31,429 from the control sample. Given the significant number of stars within this metal-
licity range, we included an additional cut to create our
From these criteria, our main sample consisted of control sample, as the main objective of this study is to
75,527 metal-poor stars that are below [Fe/H] ≤ -0.5 determine planet occurrence rates below [Fe/H] = −0.5.
directly from the TIC. To increase the sample below We randomly selected 23,000 stars that produce a sim-
[Fe/H] ≤ −0.5 and better constrain the occurrence rate, ilar distribution of stellar properties as the metal-poor
we cross-matched the TIC with LAMOST DR8 (Luo sample (see Figure 1). With the addition of the mod-
et al. 2015) (the TIC is only complete to LAMOST DR3) erately more metal-rich sub-sample, our total sample
to identify stars with spectroscopic parameters derived included 107,809 metal-poor stars. In Table 1, we show
after the creation of the TIC. From LAMOST DR8, we the ranges and medians of the stellar properties for the
gained an additional 9,282 stars that pass our stellar total sample.
4

additions and adjustments for this sample. For each star

in our sample, we extract the light curves from TESS
full-frame images (FFI). Before processing each light
curve, we separated the data into the primary mission,
30-minute cadence observations (Sectors < 27), and ex-
tended mission 1, 10-minute cadence (Sectors 27-55) ob-
servations. We do this to increase the detectability of
planet candidates, as the primary and extended mission
data do not have the same noise levels or systematics.
Given that the primary and extended mission data are
separated, we preferentially chose the 10-minute cadence
data when they were available for a given target. The
Figure 3. We show a histogram of the stellar metallicities primary mission data were processed for targets without
for our sample with bins indicated by black dashed lines. extended mission data.
The sharp peak at [Fe/H]= -0.5 is due to the reduction of To optimize the light curves to be searched for planet
the control sample as the main objective of this study is to
candidates, we relied on the detrending software wotan
determine super-Earth occurrence rates below [Fe/H]= -0.5
. (Hippke et al. 2019). Within wotan, we employed
Tukey’s bi-weight method, which is indicated to be the
most robust detrending method (Hippke et al. 2019).
From the TIC, we found that approximately 90% Since the TESS data have sharp peaks at the beginning
of the stars within the final sample have metallicities and ends of each observation sequence, we performed
from LAMOST, with the following surveys account- sigma-clipping to remove any systematics that wotan
ing for the remaining∼10%: PASTEL (Soubiran et al. was unable to remove at the 5-σ level.
2010), Hermes-TESS (Sharma et al. 2018), and Geneva- We used Transit Least Squares (TLS, Hippke & Heller
Copenhagen (Holmberg et al. 2009) (see Figure 2). From 2019) as our transit search algorithm as it generally per-
these surveys, we determined the mean uncertainty for forms exceptionally well for smaller planets. Within
the metallicities to be ∼ 0.04dex. Given that the ma- TLS, we specified stellar mass, stellar radius, and limb-
jority of the stars have metallicities derived from LAM- darkening coefficients, the latter of which were deter-
OST spectroscopic data, the impact of systematic bi- mined by interpolating the (Claret 2017) limb-darkening
ases from other surveys is limited. However, for con- tables using the stellar effective temperature, surface
sistency, we transformed metallicities from PASTEL, gravity, and stellar metallicity from the TESS Input
Hermes-TESS, and Geneva-Copenhagen to LAMOST Catalog (Stassun et al. 2018a). TLS produced a set
following the methodology in (Soubiran et al. 2022). In of TCEs with a minimum of 3 transits that were then
brief, we determined the offsets between LAMOST and vetted and classified as planet candidates or false posi-
each survey using targets that overlap with LAMOST. tives.
We found the overlaps between LAMOST to be 102, 61,
and 24 stars for PASTEL, Hermes-TESS, and Geneva-
3.1. Survey Completeness
Copenhagen, respectively. A linear fit was produced us-
ing the offsets for each survey and LAMOST, and the Using the framework in Boley et al. (2021), we con-
[Fe/H] measurements were then calibrated to the LAM- ducted injection and recovery tests to determine the de-
OST system using the linear fit. We found offsets to tection efficiency of our pipeline and vetting process.
be 0.12, 0.07, and 0.09 dex for PASTEL, Hermes-TESS, For each star in our sample, we uniformly sampled for
and Geneva-Copenhagen, respectively. periods between 1–10 days and radii between 1–3 R⊕ .
In Figure 3, we show the stellar metallicities. The We choose to focus on periods between 1–10 days to
sharp peak at [Fe/H]= -0.5 is due to requiring spec- ensure a one-to-one comparison with previous studies
troscopically derived metallicities [Fe/H]≤ -0.5. Within (Petigura et al. 2018; Zink et al. 2023). We assumed
the -0.25 to -0.5 bin, we note that the peak at approx- zero eccentricity, as planets on close-in orbits are subject
imately [Fe/H]= -0.4 results from the random cut per- to strong orbital circularization (Alvarado-Montes &
formed when reducing the sample. Garcı́a-Carmona 2019), and calculated limb-darkening
coefficients as described above.
3. PLANET DETECTION The simulated planets were injected into the light
While Boley et al. (2021) describes our planet detec- curves, which were then processed through our planet
tion pipeline in detail, we briefly describe it here, noting detection pipeline described above. We considered a
 5

2. Determine detected planets: From the syn-

thetic planet population, the number of transiting
planets that would be observed by TESS was de-
termined by simulating the instrumental and geo-
metric selection effects;

3. Calculate planet population likelihood: Us-

ing a modified Poisson likelihood (Zink et al.
2020b), the detectable planet population was com-
pared to the observed planet candidates to deter-
mine the goodness of fit;

4. Repeat: The likelihood continued to be drawn

until the distribution of parameters was well sam-
pled, typically 500,000 iterations.
Figure 4. We show the detection efficiency of our pipeline,
which is calculated via injection-recovery test. Purple indi-
To generate the planet population, we modeled the
cates a lower detection efficiency, whereas yellow indicates a planet population distribution function as a joint power-
higher detection efficiency. law in planet radius (q(r)) and orbital period (g(p)) fol-
lowing the methodology outlined in (Zink et al. 2019).
We assumed that the planet radius and orbital period
simulated planet successfully recovered if the signal-to- distributions are independent, similar to previous stud-
noise ratio (S/N) was above 6σ, the false alarm proba- ies, and modeled with broken power laws.
bility was ≤ 0.0001, there were a minimum of 3 transits,
and the period was within 1% of the injected period. d2 N
= f q(r)g(p) (2)
Using the injection and recovery tests, a detection ef- drdp
ficiency grid was created in orbital period and planet
radius space, shown in Figure 4. The survey complete- 
ness was then calculated by multiplying the detection r α 1 if r < Rbr
q(r) ∝ (3)
efficiency and the geometric transit probability, which is r α 2 if r ≥ Rbr
the ratio of the combined host star and planet radius to
the planet’s orbital radius,

pβ1 if p < Pbr
Rp + R⋆
Pt (Rp , P ) = (1) g(p) ∝ (4)
a pβ2 if p ≥ Pbr
where R⋆ is the stellar radius and a the orbital semi-
major axis of the planet, determined from Kepler’s third where f is the number of planets per star within our
law. occurrence model, α1 , α2 , β1 , and β2 values are scal-
ing model parameters. Pbr and Rbr represent the corre-
4. FORWARD MODEL sponding break in the period and radius power laws.
Using the derived survey completeness, we can in-
fer the intrinsic planet population by forward modeling. 4.1. Model Optimization
Given a population of planets, the forward model sub- Using the simulated planet populations, we measured
jects the population to the selection effects of the detec- the Bayesian posterior for the seven model parameters
tion pipeline, producing a simulated planet sample for by employing the emcee affine-invariant sampler (Good-
comparison to the observed sample. Following previous man & Weare 2010; Foreman-Mackey et al. 2013) with
studies (Mulders et al. 2018; Zink et al. 2019, 2020b), 50 semi-independent walkers, 5000 burn-in steps, and
our forward modeling followed the procedure : 10,000 sample steps. We used uniform priors for all
scaling parameters. Specifically, the priors for α1 , α2 ,
1. Generate planet population: A synthetic β1 , and β2 range from -30 to +30. For Pbr , we only con-
planet population was generated using a joint sidered the range from 1–10 days similar to the range of
power-law in period-radius space and assigned a our sample. For Rbr , the priors ranged from the mini-
random orbital orientation; mum and maximum values for each planet type. Given
6

Figure 5. Super-Earth occurrence for Kepler (black circles) and K2 (gray circles), TESS data assuming all candidates are
real (teal square), and 99.7% confidence intervals to calculate the upper limits for TESS data assuming all candidates are false
positives (teal triangles) as a function of metallicity. The [-0.25,-0.5] bin occurrence rates are offset horizontally for visual clarity.
The best-fit power-law trend line for the Kepler and K2 data (yellow) is displayed and extrapolated to [Fe/H]= -0.75 showing
the 1-σ uncertainties (Zink et al. 2023). We show the combined best-fit exponential trend line for Kepler, K2, and TESS (purple)
(eq. 6) including the 1-σ uncertainties. Each metallicity bin is indicated by gray dashed lines. Within the [-0.75,-1] bin, there is
insufficient data to further constrain the super-Earth occurrence rate as a function of metallicity (denoted by the gray hatched
region).

the co-variance in radius between super-Earths and sub- largest difference of 1.2% between the [−0.25, −0.5] and
Neptunes, we separated these planet populations along [−0.5,−0.75] bins. Therefore, the average detection ef-
the population valley (Fulton et al. 2017; Van Eylen ficiency being ∼10% in the [−0.25, −0.5] bin would be
et al. 2018). We used an empirically derived equation for ∼11.2% in the [−0.5,−0.75].
the radius valley from Ho & VanEylen (2023) (equation Figure 5 displays the occurrence rates derived for
4): our TESS sample, compared to previous analyses us-
    ing Kepler and K2 (Zink et al. 2023). We analyzed the
Rp P −0.5 < [Fe/H] ≤ −0.25 bin to overlap with those previ-
log10 = m log10 +c (5)
R⊕ days ous studies, to test for any systematic offsets from our
where m = −0.11 and c = 0.37. TESS analysis. Within this bin, we found three planet
candidates. We do not have any constraints on the false
5. RESULTS positive rate (i.e., reliability) of our TESS planet candi-
dates, so we test the most optimistic (all planet candi-
Within our sample, we find a minimal increase in
dates are real) and conservative (all planet candidates
the detection efficiency as a function of metallicity.
are false positives) scenarios. Assuming all candidates
Stars at low metallicities are less active Amard & Matt
are real, we find freal,[−0.25,−0.5] = 2.1+6.6 1
−0.99 % , consis-
(2020). They also have decreased opacities, resulting in
smaller radii for a given temperature (Xin et al. 2022).
Therefore, the difference in super-Earth detection ef- 1 The provided uncertainty is a direct measure of the 99.7% con-
ficiency between each metallicity bin ([−0.25, −0.5], fidence interval from the sampled occurrence rate (f ) posterior
distribution.
[−0.5,−0.75], [−0.75,−1.0]) is ∼1% on average, with the
 7

tent with the previous analyses. Assuming all the planet 5 shows their results for Kepler and K2, and an extrap-
candidates are false positives, we place a 99.7% con- olation of their combined trend from both datasets to
fidence interval upper limit of ff p,[−0.25,−0.5] = 9.32%. lower metallicities. Below [Fe/H] ≤ -0.5, the occurrence
Our result is consistent with previous studies, which find rate from the Kepler and K2 extrapolation combined
the occurrence rates to be 4.78 ± 1.1% 4.88 ± 3.45% with our measured survey completeness should yield a
for Kepler and K2, respectively (Zink et al. 2023). detection of 68 super-Earth candidates in our sample
Therefore, we do not find any statistically significant of TESS stars with a total of ∼ 54 candidates in the
systematic offsets between TESS and the Kepler and [−0.5,−0.75] bin and ∼ 14 candidates in the [−0.75,−1].
K2 samples using our detection pipeline. Within the To calculate the expected value of super-Earth candi-
[−0.5,−0.75] bin, we detect no super-Earth candidates, dates within our below [Fe/H] ≤ -0.5, we multiply the
and place a stringent 99.7% confidence interval upper search completeness of our pipeline (∼ 1.4%) by the total
limit of f[−0.5,−0.75] = 1.67%, shown in Figure 5. sample (∼ 85,000 stars). We then multiply that value by
Similarly, no super-Earth candidates are detected in the average extrapolated Kepler and K2 occurrence rate
the [−0.75,−1.0] bin. We find a 99.7% confidence inter- (∼ 5.7%) within that bin. Our new upper limit (1.67 %)
val upper limit of f[−0.75,−1] = 4.24%. However, given is well below the extrapolated trend (∼ 6.06%) within
the limited sample size of 20,148 stars within this bin, the [−0.5,−0.75] bin, and is statistically discrepant with
we cannot further constrain super-Earth occurrence rate a p-value = 0.000685. We calculate the p-value by cal-
compared to Kepler and K2 extrapolations (Figure 5). culating the overlap of the TESS and Kepler and K2
To determine whether in super-Earth occurrence rate distributions. Our pipeline has a detection threshold of
would continue to decrease in the [−0.75,−1.0] bin would 6-σ, which is similar to detection thresholds used in pre-
require a sample ∼ 4 times larger within that metallicity vious TESS analyses (e.g., Thuillier et al. 2022; Ment &
range based on our detection efficiency. Charbonneau 2023). Adopting a considerably stricter
requirement of 10-σ for a TCE, reducing the parameter
5.1. Exponential fit space in which we would detect signals but increasing
To determine an updated occurrence rate (fp ) trend the robustness of those signals, we find a lower but still
as a function of metallicity, we combined the data from strong discrepancy of with a p-value = 0.0293. This is
Kepler, K2, and TESS. We assume a simple exponential strong evidence that the trend in Zink et al. (2023) can-
to model a cut-off in planet formation without introduc- not be extended to lower metallicities, and that instead
ing additional parameters above a power-law fit, of the we are potentially seeing the onset of the expected criti-
form: cal metallicity cut-off for the formation of small planets.
This discovery has implications for planet formation.
First, it represents an incredibly useful, observational
h i
fp = f0 exp −102([F e/H]break −[F e/H]) (6)
constraint on models of planet formation. The critical
where f0 is the initial occurrence rate at [F e/H] = 0.4 metallicity threshold implied here may already be in ten-
and [F e/H]break is the metallicity at which the slope sion with, or rule out, a number of proposed models. By
of the exponential changes. To determine the best-fit comparing the dust settling timescale to the metallicity-
parameters, we employed scipy.optimize.curve fit. dependent disk lifetime (e.g., Ercolano & Clarke 2010),
This software relies on a nonlinear least squares method Johnson & Li (2012) derived the critical metallicity re-
to fit the exponential (Vugrin et al. 2007). We find quired to have enough solid material in the disk mid-
the best-fit parameters to be f0 = 17.33 ± 0.7 and plane before the dispersal of the disk (see their equa-
[F e/H]break = −0.31 ± 0.02 using 1σ uncertainties. tion 10). At the orbital distances relevant for our sam-
ple, their critical metallicity corresponds to [Fe/H] =
6. DISCUSSION -2.5 which is two orders of magnitude lower than our
[Fe/H]break . Population synthesis models that require
A number of studies have considered planet occur-
initial core formation near the ice line and large-scale
rence rates as a function of metallicity (see, e.g. Petigura
migration predict a range of critical metallicity for small
et al. 2018; Johnson et al. 2010; Udry & Santos 2007;
planet formation from [Fe/H] > −1.8 (Hasegawa & Hi-
Fischer & Valenti 2005). Recently, Zink et al. (2023)
rashita 2014) to [Fe/H] > −0.6 (Andama et al. 2024),
used Kepler and K2 data to investigate the correla-
subject to disk parameters such as the level of turbu-
tion between planet occurrence rates and metallicities
lence. Similarly, using the pebble flux model of Lam-
for short-period (1–10 days) planets from 1–20 R⊕ , di-
brechts & Johansen (2014), Lin et al. (2018) computed
rectly comparable to our parameter space that also con-
the core mass growth as a function of disk turbulence
siders short-period planet on orbits of 1–10 days . Figure
8

and metallicity (see their Figures 15 an 16). We have re- tion theories that involve pebble accretion (e.g.,
peated their calculation to find that the formation of an Lin et al. 2018), planetesimal formation (e.g., An-
Earth-mass core becomes difficult subject to disk tur- dama et al. 2024) and more fundamentally the ini-
bulence when the metallicity falls below [Fe/H] ∼-0.7. tial solid clumping by streaming instability (e.g.,
From direct numerical simulations, Li & Youdin (2021) Li & Youdin 2021) (§5).
find the critical metallicity to trigger the initial clump-
ing of solids via streaming instability to be sensitively • We find that planet occurrence rate trends above
determined by the particle Stokes number, the disk ra- [Fe/H] ≳ -0.5 likely cannot be extended to more
dial pressure gradient, and turbulence. A solar or sub- metal-poor environments(§5).
solar critical metallicity would generally require a large • We provide a functional form for super-Earth oc-
Stokes number reaching ∼0.1, which is near or above currence rates as a function of metallicity (eq. 6,
the maximum value expected in the analysis of nearby §5.1), and determine the metallicity cut-off to be-
protoplanetary disks with concentric rings (e.g., Rosotti gin at [F e/H]break = −0.31 ± 0.02 using 1σ uncer-
et al. 2020). tainties.
Second, it could imply that short-period super-Earths
do not form early in the history of the universe. The vast Theory predicts that planet formation for all planet
majority of stars older than ∼7 billion years, nearly half populations should become suppressed with decreasing
the lifetime of the Galaxy, have metallicities below −0.5 metallicity (e.g., Johnson & Li 2012; Lee et al. 2014;
(Feuillet et al. 2019). If small planet formation must Lee & Chiang 2015). However, this study is the first to
wait until the Galaxy has been enriched to a third of so- discover empirical evidence suggesting that super-Earth
lar metallicity or more (Soubiran et al. 2008), after the formation may become significantly more difficult. Our
death of the initial generations of stars enriching the in- study acts as an initial investigation into super-Earth
terstellar medium (Abel et al. 2002; Beers & Christlieb formation in the metal-poor regime, but more studies
2005; Frebel et al. 2007; Clark et al. 2008), this thresh- are necessary to determine whether this metallicity cut-
old could directly inform the galactic inventory of small off may be as steep for longer period planets. Given that
planets. metal-poor stars have shorter disk lifetimes and smaller
Extending our analysis to stars of even lower metal- disk masses (e.g., Yasui et al. 2010), long-period planet
licities could improve our understanding of the nature formation may be suppressed as well. However, tran-
of the metallicity cut-off. Even constructing a larger sit studies with longer baselines would be required to
sample within the [-0.5, -0.75] bin within this study probe the metallicity-correlation for longer period super-
would be instrumental in providing a strong constraint Earths. These observations will likely be feasible with
on the critical metallicity for planet formation and per- Roman and PLATO launch in the coming decade.
haps constraining the properties of the protoplanetary
disk. Therefore, subsequent studies at metallicities be-
low [Fe/H]=-0.75 will reveal the complete picture of
planet formation across Galactic space and time.

7. SUMMARY & CONCLUSIONS

In this paper, we analyzed ∼110,000 stars with
spectroscopically-derived metallicities within the metal-
poor regime (-1 ≤ [Fe/H] ≤ -0.25). The objective of this
study was to detect super-Earths at periods of 1–10 days
and determine their occurrence rates in order to test the-
oretical predictions of the critical metallicity threshold
for small planet formation (Johnson & Li 2012; Lin et al.
2018; Li & Youdin 2021). From this study, we present
our main conclusions:

• We find a distinct metallicity “cliff”in for super-

Earths at low metallicities (-0.75 ≤ [Fe/H] ≤ -
0.5). This result suggests that super-Earths may
be difficult to form within this regime, which, un-
der certain disk conditions, is in line with forma-
 9

ACKNOWLEDGMENTS
We would like to thank Scott Gaudi for his insight-
ful conversations and suggestions that have improved
this work. K.M.B. acknowledges support from the NSF
Graduate Research Fellowship Program under Grant
No. (DGE1343012). This research has made use of
the NASA Exoplanet Archive, which is operated by the
California Institute of Technology, under contract with
the National Aeronautics and Space Administration un-
der the Exoplanet Exploration Program. This paper
includes data collected by the TESS mission. Funding
for the TESS mission is provided by the NASA’s Sci-
ence Mission Directorate. The results reported herein
benefited from collaborations and/or information ex-
change within NASA’s Nexus for Exoplanet System
Science (NExSS) research coordination network spon-
sored by NASA’s Science Mission Directorate. This
material is based upon work supported by the National
Aeronautics and Space Administration under Agree-
ment No. 80NSSC21K0593 for the program “Alien
Earths”. K.M.B. thanks the LSSTC Data Science Fel-
lowship Program, which is funded by LSSTC, NSF Cy-
bertraining Grant No. 1829740, the Brinson Founda-
tion, and the Moore Foundation; her participation in
the program has benefited this work.

Software: emcee (Foreman-Mackey

et al. 2013),eleanor (Feinstein et al. 2019),
pylightcurve (Tsiaras et al. 2016), transit least
quares algorithm (Hippke & Heller 2019), and
astropy (Astropy Collaboration et al. 2013)
10

REFERENCES
Abel, T., Bryan, G. L., & Norman, M. L. 2002, Science, Choi, J.-H., & Nagamine, K. 2009, MNRAS, 393, 1595,
295, 93, doi: 10.1126/science.295.5552.93 doi: 10.1111/j.1365-2966.2008.14297.x
Adibekyan, V. Z., Delgado Mena, E., Sousa, S. G., et al. Claret, A. 2017, A&A, 600, A30,
2012a, A&A, 547, A36, doi: 10.1051/0004-6361/201629705
doi: 10.1051/0004-6361/201220167 Claret, A., & Bloemen, S. 2011, A&A, 529, A75,
Adibekyan, V. Z., Santos, N. C., Sousa, S. G., et al. 2012b, doi: 10.1051/0004-6361/201116451
A&A, 543, A89, doi: 10.1051/0004-6361/201219564 Clark, P. C., Glover, S. C. O., & Klessen, R. S. 2008, ApJ,
Akeson, R. L., Chen, X., Ciardi, D., et al. 2013, PASP, 125, 672, 757, doi: 10.1086/524187
989, doi: 10.1086/672273 Creevey, O. L., Sordo, R., Pailler, F., et al. 2023, A&A,
Alvarado-Montes, J. A., & Garcı́a-Carmona, C. 2019, 674, A26, doi: 10.1051/0004-6361/202243688
MNRAS, 486, 3963, doi: 10.1093/mnras/stz1081 Cumming, A., Marcy, G. W., & Butler, R. P. 1999, ApJ,
Amard, L., & Matt, S. P. 2020, ApJ, 889, 108, 526, 890, doi: 10.1086/308020
doi: 10.3847/1538-4357/ab6173 Dressing, C. D., & Charbonneau, D. 2015, The
Andama, G., Mah, J., & Bitsch, B. 2024, arXiv e-prints, Astrophysical Journal, 807, 45,
arXiv:2401.16155, doi: 10.48550/arXiv.2401.16155 doi: 10.1088/0004-637x/807/1/45
Anderson, T. W., & Darling, D. A. 1952, The Annals of El-Badry, K., & Rix, H.-W. 2019, MNRAS, 482, L139,
Mathematical Statistics, 23, 193. doi: 10.1093/mnrasl/sly206
http://www.jstor.org/stable/2236446 Endl, M., Cochran, W. D., Kürster, M., et al. 2006a, ApJ,
Angelov, T. 1996, Bulletin Astronomique de Belgrade, 154, 649, 436, doi: 10.1086/506465
13 —. 2006b, ApJ, 649, 436, doi: 10.1086/506465
Arriagada, P. 2011, ApJ, 734, 70, Ercolano, B., & Clarke, C. J. 2010, MNRAS, 402, 2735,
doi: 10.1088/0004-637X/734/1/70 doi: 10.1111/j.1365-2966.2009.16094.x
Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., Feinstein, A. D., Montet, B. T., Foreman-Mackey, D., et al.
et al. 2013, A&A, 558, A33, 2019, PASP, 131, 094502, doi: 10.1088/1538-3873/ab291c
doi: 10.1051/0004-6361/201322068 Feuillet, D. K., Frankel, N., Lind, K., et al. 2019, MNRAS,
Bai, X.-N., & Stone, J. M. 2010, ApJ, 722, 1437, 489, 1742, doi: 10.1093/mnras/stz2221
doi: 10.1088/0004-637X/722/2/1437 Fischer, D. A., & Valenti, J. 2005, ApJ, 622, 1102,
Bailey, E., & Batygin, K. 2018, ApJL, 866, L2, doi: 10.1086/428383
doi: 10.3847/2041-8213/aade90 Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman,
Batygin, K., & Brown, M. E. 2016, AJ, 151, 22, J. 2013, PASP, 125, 306, doi: 10.1086/670067
doi: 10.3847/0004-6256/151/2/22 Forgan, D. H., Rowlands, K., Gomez, H. L., et al. 2017,
Batygin, K., & Morbidelli, A. 2023, Nature Astronomy, 7, MNRAS, 472, 2289, doi: 10.1093/mnras/stx2162
330, doi: 10.1038/s41550-022-01850-5 Frebel, A., Johnson, J. L., & Bromm, V. 2007, MNRAS,
Beers, T. C., & Christlieb, N. 2005, AAR&A, 43, 531, 380, L40, doi: 10.1111/j.1745-3933.2007.00344.x
doi: 10.1146/annurev.astro.42.053102.134057 Fressin, F., Torres, G., Charbonneau, D., et al. 2013, ApJ,
Bernstein, R., Shectman, S. A., Gunnels, S. M., Mochnacki, 766, 81, doi: 10.1088/0004-637X/766/2/81
S., & Athey, A. E. 2003, Proc. SPIE, 4841, 1694, Fulton, B. J., & Petigura, E. A. 2018, AJ, 156, 264,
doi: 10.1117/12.461502 doi: 10.3847/1538-3881/aae828
Boley, K. M., Wang, J., Zinn, J. C., et al. 2021, AJ, 162, Fulton, B. J., Petigura, E. A., Howard, A. W., et al. 2017,
85, doi: 10.3847/1538-3881/ac0e2d AJ, 154, 109, doi: 10.3847/1538-3881/aa80eb
Bromm, V., Ferrara, A., Coppi, P. S., & Larson, R. B. 2001, Fung, J., & Lee, E. J. 2018, ApJ, 859, 126,
MNRAS, 328, 969, doi: 10.1046/j.1365-8711.2001.04915.x doi: 10.3847/1538-4357/aabaf7
Burgasser, A. J., Kirkpatrick, J. D., Reid, I. N., et al. 2003, Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al.
ApJ, 586, 512, doi: 10.1086/346263 2018, A&A, 616, A1, doi: 10.1051/0004-6361/201833051
Burrows, A., Hubbard, W. B., Saumon, D., & Lunine, J. I. Goldreich, P., Lithwick, Y., & Sari, R. 2004, ARA&A, 42,
1993, ApJ, 406, 158, doi: 10.1086/172427 549, doi: 10.1146/annurev.astro.42.053102.134004
Chen, J., & Kipping, D. 2017, ApJ, 834, 17, Gonzalez, G., Laws, C., Tyagi, S., & Reddy, B. E. 2001,
doi: 10.3847/1538-4357/834/1/17 AJ, 121, 432, doi: 10.1086/318048
 11

Goodman, J., & Weare, J. 2010, Communications in Kovács, G., Zucker, S., & Mazeh, T. 2002, A&A, 391, 369,
Applied Mathematics and Computational Science, 5, 65, doi: 10.1051/0004-6361:20020802
doi: 10.2140/camcos.2010.5.65 Kratter, K., & Lodato, G. 2016, ARA&A, 54, 271,
Guo, X., Johnson, J. A., Mann, A. W., et al. 2017, ApJ, doi: 10.1146/annurev-astro-081915-023307
838, 25, doi: 10.3847/1538-4357/aa6004 Kunimoto, M., & Bryson, S. 2020, Research Notes of the
Hartman, J. D., & Bakos, G. Á. 2016, Astronomy and American Astronomical Society, 4, 83,
Computing, 17, 1, doi: 10.1016/j.ascom.2016.05.006 doi: 10.3847/2515-5172/ab9a3c
Hasegawa, Y., & Hirashita, H. 2014, ApJ, 788, 62, Kutra, T., Wu, Y., & Qian, Y. 2021, AJ, 162, 69,
doi: 10.1088/0004-637X/788/1/62 doi: 10.3847/1538-3881/ac0431
Haworth, T. J., Ilee, J. D., Forgan, D. H., et al. 2016, Lambrechts, M., & Johansen, A. 2014, A&A, 572, A107,
PASA, 33, e053, doi: 10.1017/pasa.2016.45 doi: 10.1051/0004-6361/201424343
Haywood, M. 2008, A&A, 482, 673, Lee, E. J. 2019, ApJ, 878, 36,
doi: 10.1051/0004-6361:20079141 doi: 10.3847/1538-4357/ab1b40
Hellier, C., Anderson, D. R., Collier Cameron, A., et al. Lee, E. J., & Chiang, E. 2015, ApJ, 811, 41,
2014, MNRAS, 440, 1982, doi: 10.1093/mnras/stu410 doi: 10.1088/0004-637X/811/1/41
Hippke, M., David, T. J., Mulders, G. D., & Heller, R. Lee, E. J., Chiang, E., & Ormel, C. W. 2014, ApJ, 797, 95,
2019, AJ, 158, 143, doi: 10.3847/1538-3881/ab3984 doi: 10.1088/0004-637X/797/2/95
Hippke, M., & Heller, R. 2019, A&A, 623, A39, Li, R., & Youdin, A. N. 2021, ApJ, 919, 107,
doi: 10.1051/0004-6361/201834672 doi: 10.3847/1538-4357/ac0e9f
Ho, C. S. K., & VanEylen, V. 2023, Monthly Notices of the Lin, J. W., Lee, E. J., & Chiang, E. 2018, MNRAS, 480,
Royal Astronomical Society, 519, 4056, 4338, doi: 10.1093/mnras/sty2159
doi: 10.1093/mnras/stac3802 Lindegren, L., Hernández, J., Bombrun, A., et al. 2018,
Holmberg, J., Nordström, B., & Andersen, J. 2009, A&A, A&A, 616, A2, doi: 10.1051/0004-6361/201832727
501, 941, doi: 10.1051/0004-6361/200811191 Lodders, K. 2003, ApJ, 591, 1220, doi: 10.1086/375492
Howard, A. W., Marcy, G. W., Bryson, S. T., et al. 2012, Lodders, K. 2010, in Astrophysics and Space Science
ApJs, 201, 15, doi: 10.1088/0067-0049/201/2/15 Proceedings, Vol. 16, Principles and Perspectives in
Hui-Bon-Hoa, A. 2021, arXiv e-prints, arXiv:2101.08510, Cosmochemistry, 379, doi: 10.1007/978-3-642-10352-0 8
doi: 10.48550/arXiv.2101.08510 Lu, C. X., Schlaufman, K. C., & Cheng, S. 2020, AJ, 160,
Ida, S., & Lin, D. N. C. 2004, ApJ, 616, 567, 253, doi: 10.3847/1538-3881/abb773
doi: 10.1086/424830 Luo, A. L., Zhao, Y.-H., Zhao, G., et al. 2015, Research in
Johansen, A., Oishi, J. S., Mac Low, M.-M., et al. 2007, Astronomy and Astrophysics, 15, 1095,
Nature, 448, 1022, doi: 10.1038/nature06086 doi: 10.1088/1674-4527/15/8/002
Johnson, J. A., Aller, K. M., Howard, A. W., & Crepp, Marcy, G., Butler, R. P., Fischer, D., et al. 2005, Progress
J. R. 2010, PASP, 122, 905, doi: 10.1086/655775 of Theoretical Physics Supplement, 158, 24,
Johnson, J. A., & Apps, K. 2009, ApJ, 699, 933, doi: 10.1143/PTPS.158.24
doi: 10.1088/0004-637X/699/2/933 Masuda, K., & Winn, J. N. 2017, AJ, 153, 187,
Johnson, J. L., & Li, H. 2012, The Astrophysical Journal, doi: 10.3847/1538-3881/aa647c
751, 81, doi: 10.1088/0004-637x/751/2/81 Mayor, M., Marmier, M., Lovis, C., et al. 2011, arXiv
Johnson, J. L., & Li, H. 2013, MNRAS, 431, 972, e-prints, arXiv:1109.2497.
doi: 10.1093/mnras/stt229 https://arxiv.org/abs/1109.2497
Jorge, D. M., Kamp, I. E. E., Waters, L. B. F. M., Woitke, McCarthy, C., & Zuckerman, B. 2004, AJ, 127, 2871,
P., & Spaargaren, R. J. 2022, A&A, 660, A85, doi: 10.1086/383559
doi: 10.1051/0004-6361/202142738 Ment, K., & Charbonneau, D. 2023, AJ, 165, 265,
Kokubo, E., & Ida, S. 1998, Icarus, 131, 171, doi: 10.3847/1538-3881/acd175
doi: 10.1006/icar.1997.5840 Miller, N., & Fortney, J. J. 2011, ApJL, 736, L29,
Koppelman, H., Helmi, A., & Veljanoski, J. 2018, ApJL, doi: 10.1088/2041-8205/736/2/L29
860, L11, doi: 10.3847/2041-8213/aac882 Moe, M., & Kratter, K. M. 2019, arXiv e-prints,
Kornet, K., Bodenheimer, P., Różyczka, M., & Stepinski, arXiv:1912.01699. https://arxiv.org/abs/1912.01699
T. F. 2005, A&A, 430, 1133, Moe, M., Kratter, K. M., & Badenes, C. 2019, ApJ, 875,
doi: 10.1051/0004-6361:20041692 61, doi: 10.3847/1538-4357/ab0d88
12

Mordasini, C., Alibert, Y., Benz, W., Klahr, H., & Soubiran, C., Bienaymé, O., Mishenina, T. V., &
Henning, T. 2012, A&A, 541, A97, Kovtyukh, V. V. 2008, A&A, 480, 91,
doi: 10.1051/0004-6361/201117350 doi: 10.1051/0004-6361:20078788
Mortier, A., Santos, N. C., Sousa, S., et al. 2013, A&A, Soubiran, C., Brouillet, N., & Casamiquela, L. 2022, A&A,
551, A112, doi: 10.1051/0004-6361/201220707 663, A4, doi: 10.1051/0004-6361/202142409
Mortier, A., Santos, N. C., Sozzetti, A., et al. 2012, A&A, Soubiran, C., Le Campion, J. F., Cayrel de Strobel, G., &
543, A45, doi: 10.1051/0004-6361/201118651 Caillo, A. 2010, A&A, 515, A111,
Mulders, G. D., Pascucci, I., & Apai, D. 2015, ApJ, 798, doi: 10.1051/0004-6361/201014247
112, doi: 10.1088/0004-637X/798/2/112 Sozzetti, A., Torres, G., Latham, D. W., et al. 2009, ApJ,
Mulders, G. D., Pascucci, I., Apai, D., & Ciesla, F. J. 2018, 697, 544, doi: 10.1088/0004-637X/697/1/544
AJ, 156, 24, doi: 10.3847/1538-3881/aac5ea Squire, J., & Hopkins, P. F. 2018, MNRAS, 477, 5011,
Omukai, K. 2000, ApJ, 534, 809, doi: 10.1086/308776 doi: 10.1093/mnras/sty854
Ormel, C. W., & Klahr, H. H. 2010, A&A, 520, A43, Stassun, K. G., & Torres, G. 2021, ApJL, 907, L33,
doi: 10.1051/0004-6361/201014903 doi: 10.3847/2041-8213/abdaad
Petigura, E. A., Marcy, G. W., Winn, J. N., et al. 2018, Stassun, K. G., Oelkers, R. J., Pepper, J., et al. 2018a, AJ,
AJ, 155, 89, doi: 10.3847/1538-3881/aaa54c 156, 102, doi: 10.3847/1538-3881/aad050
—. 2018b, AJ, 156, 102, doi: 10.3847/1538-3881/aad050
Pettitt, A. N. 1976, Biometrika, 63, 161.
Stevenson, D. J. 1982, Planet. Space Sci., 30, 755,
http://www.jstor.org/stable/2335097
doi: 10.1016/0032-0633(82)90108-8
Piso, A.-M. A., & Youdin, A. N. 2014, ApJ, 786, 21,
Thorngren, D. P., Marley, M. S., & Fortney, J. J. 2019,
doi: 10.1088/0004-637X/786/1/21
Research Notes of the American Astronomical Society, 3,
Pollack, J. B., Hubickyj, O., Bodenheimer, P., et al. 1996,
128, doi: 10.3847/2515-5172/ab4353
Icarus, 124, 62, doi: 10.1006/icar.1996.0190
Thuillier, A., Van Grootel, V., Dévora-Pajares, M., et al.
Pont, F., Zucker, S., & Queloz, D. 2006, MNRAS, 373, 231,
2022, A&A, 664, A113,
doi: 10.1111/j.1365-2966.2006.11012.x
doi: 10.1051/0004-6361/202243554
Rafikov, R. R. 2006, ApJ, 648, 666, doi: 10.1086/505695
Tokovinin, A., Fischer, D. A., Bonati, M., et al. 2013,
Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2015a,
PASP, 125, 1336, doi: 10.1086/674012
Journal of Astronomical Telescopes, Instruments, and
Tsiaras, A., Waldmann, I. P., Rocchetto, M., et al. 2016,
Systems, 1, 014003, doi: 10.1117/1.JATIS.1.1.014003
pylightcurve: Exoplanet lightcurve model.
—. 2015b, Journal of Astronomical Telescopes, Instruments,
http://ascl.net/1612.018
and Systems, 1, 014003, doi: 10.1117/1.JATIS.1.1.014003
Udry, S., & Santos, N. C. 2007, ARA&A, 45, 397,
Rosotti, G. P., Teague, R., Dullemond, C., Booth, R. A., &
doi: 10.1146/annurev.astro.45.051806.110529
Clarke, C. J. 2020, MNRAS, 495, 173,
Van Eylen, V., Agentoft, C., Lundkvist, M. S., et al. 2018,
doi: 10.1093/mnras/staa1170 MNRAS, 479, 4786, doi: 10.1093/mnras/sty1783
Sandford, E., & Kipping, D. 2017, AJ, 154, 228, Venturini, J., Guilera, O. M., Ronco, M. P., & Mordasini,
doi: 10.3847/1538-3881/aa94bf C. 2020, A&A, 644, A174,
Santerne, A., Moutou, C., Tsantaki, M., et al. 2016, A&A, doi: 10.1051/0004-6361/202039140
587, A64, doi: 10.1051/0004-6361/201527329 Vugrin, K. W., Swiler, L. P., Roberts, R. M., Stucky-Mack,
Santos, N. C., Israelian, G., & Mayor, M. 2004, A&A, 415, N. J., & Sullivan, S. P. 2007, Water Resources Research,
1153, doi: 10.1051/0004-6361:20034469 43, doi: https://doi.org/10.1029/2005WR004804
Schneider, R., Ferrara, A., Natarajan, P., & Omukai, K. Wang, J., Fischer, D. A., Horch, E. P., & Huang, X. 2015,
2002, ApJ, 571, 30, doi: 10.1086/339917 ApJ, 799, 229, doi: 10.1088/0004-637X/799/2/229
Seager, S., & Mallén-Ornelas, G. 2003, ApJ, 585, 1038, Wright, J. T., Marcy, G. W., Howard, A. W., et al. 2012,
doi: 10.1086/346105 ApJ, 753, 160, doi: 10.1088/0004-637X/753/2/160
Sharma, S., Bland-Hawthorn, J., Johnston, K. V., & Wyatt, M. C., Clarke, C. J., & Greaves, J. S. 2007,
Binney, J. 2011, ApJ, 730, 3, MNRAS, 380, 1737,
doi: 10.1088/0004-637X/730/1/3 doi: 10.1111/j.1365-2966.2007.12244.x
Sharma, S., Stello, D., Buder, S., et al. 2018, MNRAS, 473, Xin, C., Renzo, M., & Metzger, B. D. 2022, MNRAS, 516,
2004, doi: 10.1093/mnras/stx2582 5816, doi: 10.1093/mnras/stac2551
 13

Yasui, C., Kobayashi, N., Tokunaga, A. T., Saito, M., & Zhu, W. 2019, ApJ, 873, 8, doi: 10.3847/1538-4357/ab0205
Tokoku, C. 2010, ApJL, 723, L113,
Zink, J. K., Christiansen, J. L., & Hansen, B. M. S. 2019,
doi: 10.1088/2041-8205/723/1/L113
Youdin, A. N., & Goodman, J. 2005, ApJ, 620, 459, MNRAS, 483, 4479, doi: 10.1093/mnras/sty3463
doi: 10.1086/426895 Zink, J. K., Hardegree-Ullman, K. K., Christiansen, J. L.,
Zechmeister, M., & Kürster, M. 2009, A&A, 496, 577,
et al. 2020a, AJ, 159, 154, doi: 10.3847/1538-3881/ab7448
doi: 10.1051/0004-6361:200811296
Zhou, G., Bakos, G. Á., Hartman, J. D., et al. 2017, AJ, —. 2020b, AJ, 160, 94, doi: 10.3847/1538-3881/aba123
153, 211, doi: 10.3847/1538-3881/aa674a
—. 2023, AJ, 165, 262, doi: 10.3847/1538-3881/acd24c
Zhou, G., Huang, C. X., Bakos, G. Á., et al. 2019, AJ, 158,
141, doi: 10.3847/1538-3881/ab36b5