The Astrophysical Journal, 966:174 (30pp), 2024 May 10 https://doi.org/10.

3847/1538-4357/ad31a6
© 2024. The Author(s). Published by the American Astronomical Society.

Four-hundred Very Metal-poor Stars Studied with LAMOST and Subaru. III.
Dynamically Tagged Groups and Chemodynamical Properties
Ruizhi Zhang1,2 , Tadafumi Matsuno3 , Haining Li1 , Wako Aoki4,5 , Xiang-Xiang Xue1 , Takuma Suda6,7 ,
1
Gang Zhao1,2 , Yuqin Chen1,2 , Miho N. Ishigaki4 , Jianrong Shi1 , Qianfan Xing1 , and Jingkun Zhao1
CAS Key Laboratory of Optical Astronomy, National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100101, Peopleʼs Republic of China;
[email protected], [email protected]
2
School of Astronomy and Space Science, University of Chinese Academy of Sciences, No.19(A) Yuquan Road, Shijingshan District, Beijing 100049, Peopleʼs
Republic of China
3
Astronomisches Rechen-Institut, Zentrum für Astronomie der Universität Heidelberg, Mönchhofstraße 12-14, 69120 Heidelberg, Germany
4
National Astronomical Observatory of Japan, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan; [email protected]
5
Astronomical Science Program, The Graduate University for Advanced Studies, SOKENDAI, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan
6
Department of Liberal Arts, Tokyo University of Technology, Nishi Kamata 5-23-22, Ota-ku, Tokyo 144-8535, Japan
7
Research Center for the Early Universe, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
Received 2024 January 17; revised 2024 February 17; accepted 2024 February 21; published 2024 May 3

Abstract
Very metal-poor (VMP) stars record the signatures of early accreted galaxies, making them essential tools for
unraveling the early stages of Galaxy formation. Understanding the origin of VMP stars requires comprehensive
studies of their chemical compositions and kinematics, which are currently lacking. Hence, we conduct a
chemodynamical analysis of 352 VMP stars selected from one of the largest uniform high-resolution VMP star
samples, jointly obtained from LAMOST and Subaru. We apply a friends-of-friends clustering algorithm to the
master catalog of this high-resolution sample, which consists of 5778 VMP stars. It results in 131 dynamically
tagged groups with 89 associated with known substructures in the Milky Way, including Gaia-Sausage-Enceladus
(GSE), Thamnos, Helmi streams, Sequoia, Wukong, Pontus, and the very metal-poor disk (VMPD). Our findings
are: (i) the VMPD shows lower Zn abundances than the rest, which indicates that it could be a relic of small stellar
systems; (ii) Sequoia shows moderately high r-process abundances; (iii) Helmi streams show deficiencies in carbon
and light neutron-capture elements; (iv) the fraction of carbon-enhanced metal-poor stars with no enhancement in
heavy elements (CEMP-no stars) seems low in the VMPD and the Helmi streams; and (v) a subgroup in GSE
exhibits a very high fraction of r-process enhanced stars, with four out of five showing [Eu/Fe]> +1.0. The
abundance patterns of other elements in VMP substructures largely match the whole VMP sample. We also study
large-scale correlations between abundance ratios and kinematics without classifying stars into substructures, but it
does not yield significant correlations once the overall chemical evolution is considered for most elements.
Unified Astronomy Thesaurus concepts: Milky Way dynamics (1051); Milky Way evolution (1052); Milky Way
Galaxy (1054); Galaxy chemical evolution (580); Galaxy abundances (574); Galactic archaeology (2178);
Population II stars (1284)
Supporting material: machine-readable table

1. Introduction abundances in the atmospheres of stars can be used to trace

Our Milky Way has gone through frequent mergers with the chemical evolution history of their progenitor stellar
dwarf galaxies during its formation, and these merging events systems (Kobayashi et al. 2020). For example, the distribution
have left abundant stellar substructures, mainly in the Galactic of [α/Fe] with [Fe/H] reflects the star formation efficiency
halo (Bullock et al. 2001; Zolotov et al. 2009; Purcell et al. (Tinsley 1980) and has been widely used to distinguish stars
2010). Though relatively recent merging events can still be that have been born in situ from those accreted from smaller
identified through spatial density in the form of streams or systems (Nissen & Schuster 2010; Helmi et al. 2018; Matsuno
shells of stars (Ibata et al. 1994; Belokurov et al. 2006; Zhao et al. 2019, 2022a, 2022b; Fernandes et al. 2023; Horta et al.
et al. 2015; Ibata et al. 2021), debris left from more ancient 2023).8 Therefore, chemodynamical studies that combine
accretions has already lost its spatial coherence. Fortunately, dynamical properties and chemical abundances of halo stars
for relic stars that have originated from the same stellar system, enable us to identify various accretion components in the
Galactic halo and uncover the nature of their progenitor
the integrals of motion of their orbits within the Milky Way
systems.
remain coherent for a much longer time (Helmi & White 1999;
Thanks to the combination of the Gaia mission (Gaia
Helmi 2020), and hence we can excavate fossils of building
Collaboration et al. 2016, 2018, 2021) and various spectroscopic
blocks accreted a long time ago by identifying dynamical
surveys including SEGUE from the Sloan Digital Sky Survey
groups among Galactic halo stars. Moreover, elemental
(Yanny et al. 2009), LAMOST (Zhao et al. 2006, 2012), GALAH
(De Silva et al. 2015), APOGEE (Majewski et al. 2017), etc., the
Original content from this work may be used under the terms
of the Creative Commons Attribution 4.0 licence. Any further
8
distribution of this work must maintain attribution to the author(s) and the title The standard notations [X/Y] = log (NX NY ) - log (NX NY ) and
of the work, journal citation and DOI. A(X) = log (NX NH ) + 12 for elements X and Y are adopted in this work.

1
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

chemodynamical studies of the Milky Way halo stars have differences in the chemical enrichments among different
significantly advanced in recent years (e.g., Myeong et al. 2018a; galaxies (Ishimaru et al. 2015; Ojima et al. 2018). Therefore,
Belokurov et al. 2018; Helmi et al. 2018; Koppelman et al. 2018; separating stars according to their progenitor galaxy whenever
Helmi 2020; Naidu et al. 2020; Lövdal et al. 2022; Dodd et al. possible is highly beneficial in observationally disentangling
2023). The combination of astrometry and radial-velocity the two mechanisms. Moreover, if one can associate an
measurements has enabled the identification of kinematic abundance difference between two galaxies to their different
substructures, which are promising candidates for accreted properties, such as mass or star formation timescale, it allows
galaxies. It is now clear that the Milky Way’s inner stellar halo us to constrain the property of the nucleosynthesis process
is dominated by debris of the last major merger Gaia-Sausage- producing the elements.
Enceladus (GSE; Belokurov et al. 2018; Helmi et al. 2018) and We surpass previous chemodynamical analyses of VMP
those heated by this accretion from the disk present at that time stars by investigating a large, homogeneous sample selected
(Splash; Belokurov et al. 2020). The nearby halo stars contain from the LAMOST survey, leveraging the high-resolution
smaller but still quite significant kinematic substructures, such as spectroscopy using the High Dispersion Spectrograph (HDS;
the Helmi streams9 (Helmi et al. 1999), Thamnos (Koppelman Noguchi et al. 2002) equipped on the Subaru telescope. This
et al. 2019a), and Sequoia (Matsuno et al. 2019; Myeong et al. allows us to identify distinct groups of stars sharing similar
2019), and there are also signatures of past accretion events in dynamics within the Galactic halo with a technique known as
the outer part of the Galaxy as well, such as Cetus (Newberg dynamical tagging. This paper is the third one of a series.
et al. 2009) and Wukong/LMS-1 (Yuan et al. 2020a; Naidu Detailed descriptions of target selection and observations of
et al. 2020). Although each kinematic substructure shows this LAMOST/Subaru VMP sample are provided in (Aoki
coherent chemical abundance trends among its member stars, et al. (2022, hereafter Paper I), which also reports radial
the general trends can be different among different substruc- velocities and interstellar reddening of the program stars. A
tures (Aguado et al. 2021a; Matsuno et al. 2021, 2022a, 2022b; homogeneous chemical abundance analysis for this LAMOST/
Horta et al. 2023). Thus, these differences enable detailed Subaru VMP star sample is presented in Li et al. (2022,
characterization of corresponding progenitor galaxies. hereafter Paper II), where stellar parameters and abundances for
Very metal-poor (VMP) stars with [Fe/H]< −2.0 are low- more than 20 species have been determined for 385 stars in the
mass old stars that preserve the record of the chemical LAMOST/Subaru VMP sample.
composition and dynamics of the early Milky Way. Given This paper is organized as follows. In Section 2, we
their dominance in small dwarf galaxies (Simon 2019) and their summarize the properties of the sample and the determination
potential association with disrupted stellar systems, VMP stars of their basic kinematic parameters. The method of dynamical
are expected to harbor a significantly higher fraction of analysis and the corresponding result of dynamical clustering
accreted remnants from low-mass stellar systems compared to of the sample are respectively presented in Sections 3.1 and
more metal-rich populations. Therefore, studying the chemo- 3.2. We discuss chemodynamical properties of identified VMP
dynamical properties of VMP stars is of great importance to substructures in Section 4.1, kinematics of stars with peculiar
fully reproduce the early history of our Galaxy, which involves abundance ratios in Section 4.2, and large-scale chemodyna-
the formation of in situ Galactic stars, accretions of low-mass mical correlations in Section 4.3. We present the conclusion in
dwarf galaxies such as the ultra-faint dwarf (UFD) galaxies, Section 5.
and those that have merged with the Milky Way in its very
early stages. 2. Data
However, despite recent progress, we still lack comprehen-
sive chemodynamical studies on the VMP region of the Although the VMP sample provided by Paper II allows us to
Galactic halo. On the one hand, various chemodynamical investigate the chemodynamics of the Milky Way halo in a
studies on Galactic halo stars either focus on relatively higher very low metallicity region, its size is not sufficiently large
metallicity regions or involve only a very limited number of compared to those used to define kinematic substructures or
VMP stars (e.g., Monty et al. 2020; Aguado et al. 2021a, dynamically tagged groups (DTGs). Hence, we make use of the
2021b; Limberg et al. 2021b; Gull et al. 2021; Matsuno et al. LAMOST 10,000 VMP star catalog by Li et al. (2018, hereafter
2022a, 2022b; Carrillo et al. 2022; da Silva & Smiljanic 2023; L18) to construct a larger VMP sample, which can be regarded
Horta et al. 2023; Limberg et al. 2024). On the other hand, a as the master catalog for the high-resolution VMP sample. This
few studies have examined the dynamical properties of VMP large sample allows us to more accurately define DTGs, which
stars but have not provided detailed information on elemental we can then use to associate VMP stars from Paper II to
abundances (e.g., Yuan et al. 2020b, hereafter Y20; Limberg corresponding DTGs. In the following discussion, we refer to
et al. 2021a; Carollo et al. 2023). the high-resolution VMP sample from the LAMOST/Subaru
Chemodynamical analyses of VMP stars hold the key to survey as the “HR sample” and the larger VMP master sample
unraveling early nucleosynthesis and the origins of diverse from LAMOST as the “LR sample.”
elements, thanks to their minimal enrichment from subsequent
generations of stars. Low-metallicity stars in the halo show 2.1. Kinematics of the HR Sample
diverse chemical abundance ratios, especially in neutron-
capture elements (e.g., McWilliam et al. 1995; Honda et al. Candidates for the HR sample were originally selected from
2004; Barklem et al. 2005). Such a dispersion can be due to LAMOST DR3 through DR5 and were then follow-up
inhomogeneous mixing of the ejecta from nucleosynthesis observed using the HDS on the Subaru telescope. We have
events (e.g., Argast et al. 2004; Hirai et al. 2015) and/or measured 1D local thermodynamic equilibrium (LTE) abun-
dances of Li, C, Na, Mg, Si, Ca, Sc, Ti, V, Cr, Mn, Co, Ni, Zn,
9
Because of the bimodal component in vertical velocity, we refer to this Sr, Y, Zr, Ba, La, and Eu based on the high-resolution spectra
structure as streams (see Helmi 2020 and references therein). and applied a non-LTE correction to Na abundances. Details

2
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

dispersion values are small, indicating that the two velocities

are consistent.
However, some outliers need further discussion. We treated six
stars with ∣rvSubaru - rvGaia - mDrv∣ > 3 s 2Subaru + s 2Gaia + s D
2
rv as
outliers. We cross-matched these stars with LAMOST DR9 to
check whether they have variable radial velocities. One star (J0701
+1813) has similar radial velocities between Subaru and
LAMOST measurements at different observation dates; thus, we
consider the Subaru measurement reliable. The other five stars
have different radial velocities among Subaru spectra, Gaia, and
LAMOST, and three of the stars show clear radial-velocity
variation among the multiple observations from LAMOST. We
removed these five stars since they are likely part of binary systems
and we do not have reliable radial velocities of the system’s
barycenters. In total, the final HR sample selected for chemody-
namical analysis consists of 352 stars with reliable astrometry and
Figure 1. The distributions of the differences between Gaia and Subaru radial- radial-velocity measurements.
velocity measurements. The red dashed line and the gray shaded region are, We computed the positions and velocities of the stars adopting
respectively, the median value and 3σ scale of the residuals. The red open dot
is J0701 + 1813, which has consistent radial velocities between Subaru and
the photogeometric distance, coordinates, and proper motion
LAMOST DR9, and the red solid dots are sources with inconsistent radial from the Gaia DR3, and the radial velocity derived from Subaru
velocities between Subaru spectra and Gaia DR3. The median and dispersion of spectra. We assumed that the Sun is located at 8.21 kpc from the
the differences are shown in the lower right of this panel, and the typical error Galactic Center (McMillan 2017) and 20.8 pc above the Galactic
bar of the whole sample is shown in the lower left. plane (Bennett & Bovy 2019). We took the local standard of rest
(LSR) velocity as VLSR = 233.1 km s−1 from the measurement
about the target selection, follow-up strategy, and abundance by McMillan (2017) and adopted the solar peculiar motion
analysis can be referred to in Papers I and II. An advantage of (Ue, Ve, We) = (11.1, 12.24, 7.25) km s−1 as measured by
our sample is that the spectra have been analyzed homo- Schönrich et al. (2010). We used a right-hand Galactocentric
geneously. Therefore, our chemodynamical investigation pre- Cartesian coordinate system (X, Y, Z) and a Galactocentric
sented in the present paper is not subject to significant cylindrical coordinate system (R, f, z), where the Sun is at
systematic uncertainties. (Xe, Ye, Ze) = ( −8.21, 0.0, 0.0208) kpc, and vf is negative for
We cross-matched this HR sample with the Gaia DR3 stars on the disk rotation.
catalogs (Gaia Collaboration et al. 2023) and the distance catalog Using the software Agama (Vasiliev 2019), we further
of Bailer-Jones et al. (2021, hereafter BJ21). All of our 385 stars computed the following quantities that characterize the orbits of
have parallax, proper-motion, and Gaia magnitude measure- the stars and integrated five times the period of the circular
ments, and photogeometric distances from BJ21. In this paper, orbit corresponding to the orbital energy for each star to derive
we primarily use the photogeometric distances when computing the orbital parameters:
stellar kinematics. Of these photogeometric distances, only ten
stars have a relative distance uncertainty larger than 20%, which 1. E = 1 2 (vR2 + vf2 + vz2 )+ ΨMW(R, z),
is defined as (dhi,photogeo − dlo,photogeo)/(2 × dmed,photogeo), where 2. L^ = (Rvz - zvR )2 + (zvf )2 ,
dhi,photogeo, dlo,photogeo, and dmed,photogeo are the 84th and 16th 3. Jf = −Lz = −Rvf, where a positive azimuthal action
percentiles, and median of BJ21 photogeometric distance, indicates a prograde orbit, which is opposite for the
respectively. We excluded these ten stars from the kinematic z-component of angular momentum,
analysis, and the median relative distance uncertainty of the rest 4. radial action (Jr) and vertical action (Jz),
of the HR sample is about 3%. We further removed 18 stars with 5. total action Jtot = Jr + Jz + |Jf|,
the renormalized unit weighted error (RUWE) larger than 1.4 6. circularity h = Jf ∣Lz,max (E )∣, where Lz,max is the angular
since their astrometry is less reliable (Lindegren et al. 2021). momentum of the circular orbit with orbital energy E,
In addition to astrometry, we also needed radial-velocity 7. Rmax and ∣z∣max ,
measurements to compute kinematics. In Figure 1, we compare 8. Galactocentric distance at pericenter (rperi) and apocenter
the radial velocities reported in Paper I based on Subaru (rapo),
spectra10 with the radial velocities provided by Gaia DR3 for r - rperi
9. eccentricity (ec = rapo
r +r
) and semimajor axis (a = apo 2 peri )
apo + rperi
273 stars whose Gaia radial velocities are available. While the of the orbit.
median uncertainty in the Gaia radial velocity (σGaia) is
2.1 km s−1, we adopt 1.0 km s−1 as the typical uncertainty in We assumed the axisymmetric Milky Way potential of
the Subaru radial velocity (σSubaru), considering the long-term McMillan (2017) for ΨMW(R, z). For each star, we ran a 10,000
stability of HDS. The median velocity difference (μΔrv) and times Monte Carlo (MC) simulation assuming Gaussian uncer-
half the difference between the 16th and 84th percentiles (σΔrv) tainties for the observables to estimate uncertainties for these
are 0.063 km s−1 and 1.875 km s−1, respectively. Given the quantities. We note that, although the distribution of the BJ21
velocity uncertainties, we conclude that these median offset and photogeometric distance does not follow a Gaussian distribution,
over 90% of the stars in the HR sample show almost symmetric
confidence intervals with |(σhi − σlo)/(σhi + σlo)| < 20%, where
10
We note that the radial velocity for a few objects from Subaru spectra
σhi = dhi,photogeo −dmed,photogeo and σlo = dmed,photogeo − dlo,photogeo.
reported in Paper I needs further corrections. We use the corrected values in We note here that the interpretation of the observed
this work, for which readers can refer to the erratum of Paper I. correlations between kinematics and chemical abundances

3
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure 2. The relations between Gaia G-band magnitude, photogeometric distance (Bailer-Jones et al. 2021), and metallicity from Paper II of the stars. Red open
squares, blue open triangles, and green open diamonds, respectively, show stars having a relative distance uncertainty larger than 20%, stars having RUWE larger than
1.4, and those having an inconsistent radial velocity between Subaru spectra, Gaia DR3, and LAMOST DR9. These figures show a clear correlation between distance
and [Fe/H] because of our high-resolution observation strategy described in Paper I.

might need caution because of our sample selection (see a procedure similar to that of Xue et al. (2014). Since distances
Section 2.2 of Paper I). While we focused on bright stars from Z23 and BJ21 might not be on the same scale, we applied a
at [Fe/H] > −3, we extended the sample to fainter stars at correction to Z23. We first selected giants with log g < 3.5 and
[Fe/H] < −3. This observing strategy introduces a correlation cross-matched them with the K-giant catalog from Z23, and then
between metallicity and heliocentric distance (Figure 2), which fit a linear relation between Z23 and BJ21 distances, assuming that
can then introduce a correlation between metallicity and stars follow a normal distribution around the relation. We applied a
kinematic quantities. Since abundance ratios evolve with Markov Chain Monte Carlo (MCMC) method to fit the relation to
metallicity, the metallicity–kinematics correlation would the 111 VMP giants within 4 kpc (ϖ > 0.25 mas). Figure 3 shows
further lead to an apparent correlation between chemical the best-fitting result and we calibrated Z23 distances using this fit
abundance ratios and kinematics. We try to mitigate this effect relation.
throughout the paper. A criterion is needed for different distance adoptions. Since
giants with lower surface gravity are usually more luminous,
2.2. Kinematics of the LR Sample stars with lower log g would have smaller apparent magnitudes
We combined the L18 catalog with our HR sample to at the same distance. We present all the stars with calibrated
construct a larger VMP master sample for clustering. To avoid Z23 distances greater than 5 kpc on the log g − GGaia panel as
the influence of the relatively metal-rich stars in clustering, we shown in Figure 4, where log g are from LAMOST DR9, and
cross-matched the L18 catalog with LAMOST DR9 and GGaia is the G-band apparent magnitude from Gaia DR3. We
removed stars with signal-to-noise ratios (S/N) less than 15 or found that these distant giants tend to be above the black line in
[Fe/H]> −1.5, to eliminate the stars that are incorrectly Figure 4 and took this black line as the division of different
classified as VMP in L18. Note that the official LAMOST distance estimates. We used calibrated Z23 distances for stars
pipeline does not provide metallicity estimation at [Fe/H] within the upper-right area, while, out of this area, the BJ21
< −2.5, and thus we did not exclude stars with no parameter photogeometric distances are adopted.
estimation in LAMOST DR9, where no metallicity estimation We then removed 545 stars with large distance uncertainties
may also imply a relatively low metallicity. (derr/d > 20%). Among the remaining 5675 stars, we adopted
Subsequently, we cross-matched the L18 sample with Gaia BJ21 distances for 5444 objects and Z23 distances for 231
DR3 and the distance catalog of BJ21, and obtained a sample objects. The typical uncertainties are 6% and 14% for distances
of 7741 VMP stars with Gaia astrometric data. We excluded from BJ21 and Z23, respectively. This reduced large VMP
239 stars with RUWE > 1.4. Additionally, we removed two sample has a similar distance distribution to the HR sample as
stars located in a crowded field, where there is more than one shown in Figure A1.
star within a 5″ radius, which may cause incorrect cross- We used the same coordinate systems and computed the
matching. Using the Gaia source id, we also found 249 same parameters for the selected L18 catalog as for the HR
duplicate sources, and we only kept the data with the highest sample. For radial velocity, we prioritized the measurement
LAMOST S/N for each of them and eliminated 260 records. In from LAMOST DR9; otherwise, we adopted the value from
addition, 1020 stars with a reliable radial velocity from neither Gaia DR3. We found two unbound stars with positive energy
LAMOST DR9 nor Gaia DR3 were also removed. and removed them from the following analysis. Unlike other
There are ∼500 faint and distant giants in the reduced sample. works that aim to explore the clustering substructures of VMP
For some of these giants, the photogeometric distances from BJ21 stars in the Galactic halo (e.g., Y20; Limberg et al. 2021a), we
may be underestimated (see, e.g., Section 5.5 in BJ21), and the did not make any additional restrictions in the Toomre diagram,
parallaxes may not be accurate enough for distance determinations. which allows us to investigate the VMP disk-like substructures.
Therefore, we introduced distances from Zhang et al. (2023, For 247 common stars between the HR sample and the L18
hereafter Z23) for such giants, who derived the distances of catalog, we replaced the kinematic parameters of these stars in
∼20,000 K giants in the Galactic halo from photometry following the L18 catalog with those computed in Section 2.1 and

4
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

gravitational potential, implying that stars with similar IoMs may

share a common origin. An axisymmetric system usually admits
three isolating IoMs, and under the action-angle coordinates the
three actions (Jr, Jf, Jz) are useful isolating integrals to characterize
the stellar orbits in axisymmetric galaxies in dynamical equilibrium
(Ollongren 1965; Binney & Tremaine 2008; Binney 2012). Helmi
& White (1999) first proposed that disrupted satellites could be
identified as clusters in the action-angle phase space. In the Gaia
era, many substructures and DTGs are found both in action phase
space and in energy–action combined phase space (Myeong et al.
2018b; Roederer et al. 2018; Myeong et al. 2019; Limberg et al.
2021a; Shank et al. 2022a, 2022b; Lövdal et al. 2022; Cabrera
Garcia et al. 2024; Dodd et al. 2023), which demonstrates the
feasibility of searching for substructures in action phase space.
In this work, we used the friends-of-friends (FoF) clustering
algorithm to identify substructures in the (Jr, Jf, Jz) phase space.
This algorithm links two stars with a separation δ smaller than a
given threshold (linking length) into one group, and this group
continues to grow until the δ between any star in this group and
all the other nongroup stars is greater than the linking length.
We adopted the following definition of δ, which is similar to
that of Wang et al. (2022) but in a different phase space.
d 2ij = wD Jr ( Jr , i - Jr , j )2 + wDJf (Jf, i - Jf, j )2
+ wD Jz ( Jz , i - Jz , j )2 . (1 )
Figure 3. The comparison of BJ21 and Z23 distances for the VMP giants with For radial and vertical actions, we considered separations in
dZ23,error,calib/dZ23,calib < 0.2 and (dhi,photogeo − dlo,photogeo)/(2 × dmed,photogeo) < 0.2.
The red line is the fitting result using MCMC. their square roots. This helped us identify clusters on polar or
eccentric orbits, as Jz and Jr of stars on such orbits can vary
significantly when their orbits are changed slightly.
The wD Jr , wDJf , wD Jz are weights used to normalize the
separations in the three different actions. We normalized each
action so that the median deviations of all the pairs are 1, i.e.,
1
wDX = (X = Jr , Jf, Jz ). (2 )
median [(DX )2 ]
In our sample, the ωΔX are 1/8.34 s1/2 kpc−1/2 km−1/2,
1/821.74 s kpc−1 km−1, and 1/6.98 s1/2 kpc−1/2 km−1/2 for
Jr , Jf, and Jz , respectively.
To ensure that we selected reliable groups, for each group
identified by the FoF algorithm we chose a suitable linking
length, as shown in Appendix B. Only groups with no less than
five members were kept for dynamic analysis.
Figure 4. The distribution of G-band apparent magnitude and surface gravity of
stars in the L18 catalog. The colored dots are stars with calibrated Z23 distances
greater than 5 kpc. The red dashed line is log g = 3.5, and the black line is the
3.2. Results
criterion we adopted to select distant giants. The upper-right region beyond these Based on the clustering results of the LR sample, we identified
two lines is the area where calibrated Z23 distances are used for stars.
286 DTGs; for 131 of them, members can be found in the HR
sample. The result of the clustering analysis is shown in Table 1,
combined these two tables. Finally, we obtained the LR sample where we also show the median dynamical parameters for each
of 5778 VMP stars with kinematic parameters for clustering. group. We found that 89 out of the 131 DTGs can be associated to
To summarize, we constructed two samples with reliable known substructures in the Milky Way halo, namely, the GSE
dynamical parameters based on the LAMOST/Subaru high- (Belokurov et al. 2018; Helmi et al. 2018), Thamnos (Koppelman
resolution survey and the low-resolution sample from et al. 2019a), Sequoia/I’itoi (Myeong et al. 2019; Naidu et al.
LAMOST, respectively. We emphasize that we only used the 2020), Helmi streams (Helmi et al. 1999), Pontus (Malhan 2022),
kinematics of the LR sample for clustering and the chemody- and Wukong/LMS-1 (Yuan et al. 2020a; Naidu et al. 2020).
namical analysis will focus on the HR sample. Table 2 lists such associations of our DTGs with the substructures,
mostly following the criteria shown in Table C1.11 We note that
3. Clustering Analysis we relaxed the criteria for these associations to have a sufficient
3.1. Method
Integrals of motion (IoM) of stars originating from the same 11
For groups that are close to substructures but do not meet the criteria, we
progenitor system remain coherent over time in a slowly varying decided whether they should be assigned to these substructures manually.

5
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Table 1
Dynamical Parameters of DTGs Clustered by FoF Algorithm

Group Linking Length NLR a NHR b E (Jr, Jf, Jz) (vR, vf, vz) ec ∣z∣max
(σE) (sJr , sJf, sJz ) (svR, svf, svz ) σec s∣ z ∣max
(105 km2 s−2) (kpc km s−1) (km s−1) (kpc)
DTG-1 0.098 212 6 −1.738 (501.9, 427.3, 96.0) (−3.59, −49.74, −1.93) 0.789 3.518
0.045 (93.4, 125.3, 25.9) (87.79, 17.62, 66.91) 0.073 0.644
DTG-2 0.112 105 3 −1.659 (251.3, 1087.4, 61.8) (31.60, −127.37, −11.01) 0.494 2.047
0.041 (68.9, 158.7, 12.1) (83.11, 21.07, 53.26) 0.067 0.268
DTG-3 0.098 93 3 −1.733 (704.8, -23.0, 74.6) (−27.98, 2.70, −6.38) 0.893 3.913
0.043 (52.6, 126.2, 28.1) (85.16, 14.40, 51.95) 0.048 0.431
DTG-4 0.100 82 5 −1.736 (603.2, 377.3, 28.7) (−36.10, −44.63, −3.44) 0.817 1.352
0.062 (62.6, 124.8, 9.1) (101.08, 13.38, 36.08) 0.037 1.371
DTG-5 0.134 79 3 −1.702 (589.5, 25.9, 301.4) (−40.2, −4.46, 57.15) 0.891 7.175
0.060 (82.5, 135.4, 50.2) (86.76, 17.17, 104.63) 0.042 1.150

Notes. The full table in machine-readable form is available. The value and uncertainty of each parameter is the median value and half the difference between the 16th
and 84th quantiles of each group.
a
NLR is the number of stars from the LR sample in the DTG.
b
NHR is the number of stars from the HR sample in the DTG.
(This table is available in its entirety in machine-readable form.)

number of HR stars for each group for later investigations on medium from which they have formed. Any stars that have
chemical properties. experienced external contamination to their surface chemical
We also found some DTGs with thick-disk-like kinematics abundance should be excluded from our analysis. For example,
with ∣z∣max < 3 kpc and 0.25 < ec < 0.60. Although they occupy the chemical compositions of carbon-enhanced metal-poor
a similar region as the thick-disk stars in the Toomre diagram, (CEMP) stars with large excesses of s-process elements
their rotational velocities peak at roughly −130 km s−1, slower (CEMP-s) are changed due to mass transfer from their
than the typical rotational velocity of the canonical thick disk companions (Abate et al. 2015; Jorissen et al. 2016), and thus
(vf ∼ −180 km s−1; Carollo et al. 2010). Thus, we call these we excluded ten CEMP-s stars and a potential CEMP-s star
DTGs the very metal-poor disk (VMPD) throughout this paper. J0446 + 2124 (see Section 4.2.1 for details) when discussing
Figure 5 shows the energy–action distribution of the the abundance distribution.
identified substructures. We present the distributions of the To make sure that our studies are based on the most reliable
LR sample and HR sample in the upper and lower panels of chemical abundances, we excluded stars with uncertainties
Figure 5, respectively. In addition to these substructures, we larger than 0.2 dex in [Fe/H] and [X/Fe] during discussions on
also found smaller DTGs, which we could not associate with corresponding element X. Two stars with metallicity larger
known substructures. Among these new DTGs, we discuss the than −1.5 were also eliminated as their metallicities are much
four groups containing more than ten members and at least higher than the rest of the sample.
three stars in the HR sample. The dynamical properties of these Using the selected HR sample, we present the results for
four groups are also shown in Figure 6 and Table 2. The GSE, VMPD, retrograde substructures, polar substructures, and
dynamical parameters and clustering results of the HR sample other DTGs, respectively.
are available in Zenodo and can be accessed via doi:10.5281/ We also utilize the elemental abundances from the literature
zenodo.10780897. and the GALAH survey to compare the chemical distribution
Again, we stress that we allowed some contamination by between our results and previous studies in a wider metallicity
adopting relatively loose criteria so that we have sufficient range. Readers may refer to Appendix D for details.
numbers of stars for chemical investigation with the HR sample.
Although we did not consider that several outliers would alter the
interpretation of chemical abundance, caution is needed when the 4.1.1. GSE
interpretation is based on a small number of stars. GSE is the remnant of a major merger event of the ancient
Milky Way, which was revealed by the “sausage”-like
4. Chemodynamics of VMP Stars distribution in vR − vf, extremely radial/eccentric orbits, and
low α-element abundances of its members (Belokurov et al.
4.1. Early Chemical Evolution of Substructures and DTGs
2018; Haywood et al. 2018; Helmi et al. 2018). Its progenitor
In this section, we combine the chemical abundances of 352 merged with the Milky Way approximately 10 Gyr ago (Helmi
stars in the HR sample as introduced in Section 2.1 with the et al. 2018; Gallart et al. 2019; Helmi 2020; Montalbán et al.
clustering results from Section 3.2 to present the chemical 2021) and dominates the inner Galactic halo (e.g., see
patterns of carbon, α-elements (Mg, Si, Ca, Ti), light odd-Z Lancaster et al. 2019; Naidu et al. 2020; Bird et al. 2021; Wu
elements (Na, Sc), iron-peak elements (V, Cr, Mn, Co, Ni, Zn), et al. 2022).
and heavy elements (Sr, Y, Zr, Ba, La, Eu) for each GSE is the largest substructure we recovered, taking up one-
substructure. third of the LR sample and one-fourth of the HR sample. This
The purpose of our study is to investigate the chemical result is similar to previous studies on the VMP halo using
evolution of the progenitors of the substructures, assuming that clustering algorithms (Limberg et al. 2021a; Shank et al.
VMP stars retain the chemical properties of the interstellar 2022b, 2022a) and Gaussian mixture models (An & Beers 2021).

6
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure 5. The distributions of groups associated with known substructures in energy and action phase space. The upper and lower panels are the results for the LR
sample and the HR sample, respectively. Different colors represent different substructures: GSE (red), Thamnos (green), Sequoia (orange), Helmi streams (brown),
Pontus (purple), Wukong/LMS-1 (blue), and VMPD (sky blue). The red stars stand for the r-process-enhanced subgroup in GSE; see Section 4.2.3 for details.

Figure 6. The distributions of four large DTGs in energy and action phase space. The open diamond symbols are group members in the LR sample, and the star
symbols with black borders are stars in the HR sample. Different colors represent different DTGs: DTG-8 (pink), DTG-9 (green), DTG-20 (blue) and DTG-49
(orange).

The chemical distribution of GSE is shown in Figure 7. There is GSE is the large scatter in most elements (e.g., C, Mg, Ca, Cr, and
no significant difference between the distribution of GSE and that heavy elements). This property suggests that the GSE possesses a
of the entire sample. This may be due to the fact that the chemical complicated chemical composition, which is also reported in
evolution of the massive progenitor of GSE (Må ∼ 108–109Me; several previous works (e.g., Zhao & Chen 2021; Donlon &
see Mackereth et al. 2019; Vincenzo et al. 2019; Helmi 2020; Newberg 2023). This may indicate a series of merging processes
Naidu et al. 2020, 2021 and references therein) and the Milky between the massive progenitor of GSE and other dwarf galaxies
Way are similar at [Fe/H] < −2. Another noticeable feature of before it merged with our Galaxy.

7
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Table 2
Dynamical Parameters of Known Substructures and Large New DTGs

Substructure Linking Length NLR NHR E (Jr, Jf, Jz) (vR, vf, vz) ec ∣z∣max
(σE) (sJr , sJf, sJz ) (svR, svf, svz ) σec s∣ z ∣max
(105 km2 s−2) (kpc km s−1) (km s−1) (kpc)
GSE [0.056, 0.350] 1940 89 −1.684 (649.9, 113.8, 89.1) (−18.25, −13.84, −1.34) 0.857 4.202
0.131 (263.4, 376.9, 120.1) (146.03, 43.56, 69.37) 0.093 3.013
VMPD [0.072, 0.214] 336 18 −1.671 (235.6, 1087.8, 55.0) (19.98, −129.49, −3.33) 0.490 1.876
0.047 (95.2, 253.4, 32.4) (78.43, 30.78, 46.17) 0.117 0.787
Thamnos [0.118, 0.306] 282 23 −1.693 (266.6, −881.1, 75.9) (−3.15, 107.24, −1.06) 0.544 2.462
0.060 (107.2, 270.5, 82.3) (74.75, 32.05, 58.71) 0.126 1.445
Sequoia [0.202, 0.682] 136 7 −1.283 (727.8, −1687.0, 388.5) (40.94, 199.52, −32.96) 0.590 9.484
0.169 (640.7, 666.5, 330.2) (187.35, 80.51, 146.77) 0.163 4.465
Helmi streams [0.290, 0.628] 112 15 −1.329 (258.7, 1331.4, 991.4) (10.97, −153.11, −186.19) 0.379 14.308
0.081 (230.8, 260.3, 381.2) (114.98, 34.97, 234.00) 0.152 4.060
Pontus [0.124, 0.158] 62 3 −1.702 (429.8, −435.1, 251.6) (−47.1, 54.03, −71.25) 0.734 5.401
0.044 (77.7, 88.0, 62.9) (85.88, 10.79, 94.65) 0.053 0.825
Wukong [0.308, 0.366] 30 4 −1.431 (98.5, 432.0, 1597.3) (29.7, −43.60, 171.83) 0.263 12.397
(LMS-1) 0.043 (58.1, 275.0, 322.3) (103.24, 46.79, 202.93) 0.065 1.265

DTG-8 0.114 60 7 −1.665 (130.3, 1150.4, 132.8) (24.38, −140.66, −8.98) 0.368 3.124
0.024 (29.1, 90.9, 47.8) (62.94, 11.43, 77.68) 0.038 0.775
DTG-9 0.212 53 5 −1.611 (270.2, 343.1, 790.5) (9.22, −40.51, −126.87) 0.541 9.284
0.040 (80.3, 340.8, 150.8) (85.23, 39.03, 154.08) 0.091 0.972
DTG-20 0.126 32 4 −1.756 (348.8, 756.8, 14.3) (15.45, −89.52, 6.02) 0.627 0.803
0.042 (29.8, 124.3, 9.8) (53.80, 9.83, 28.79) 0.022 0.275
DTG-49 0.476 13 3 −1.197 (1400.2, 1829.9, 107.2) (219.55, −206.61, 37.17) 0.729 7.053
0.037 (171.4, 317.2, 46.0) (259.27, 58.05, 55.11) 0.039 2.062

Note. We present the minimum and maximum linking length of each substructure, and the meaning of the other columns is the same as in Table 1.

However, this scatter could also be due to contamination galaxies of the GSE progenitor; see more detailed discussions
from other substructures. To have a large sample of VMP stars in Section 4.2.2.
in GSE, we adopted an eccentricity-based selection as this Previous studies have revealed the r-process enhancement of
method is shown to have the highest completeness (Carrillo GSE at [Fe/H]> −2 using the GALAH Survey catalog
et al. 2024). While this sample has a lower purity, we do not (Matsuno et al. 2021; Myeong et al. 2022; da Silva &
consider the contamination to be significant. Two major Smiljanic 2023) and elemental abundances derived from other
sources of the contamination using the eccentricity criterion high-resolution spectroscopic studies (Aguado et al. 2021a;
are the smooth components in the Galactic halo and the metal- Koch-Hansen et al. 2021; Carrillo et al. 2022; Naidu et al.
weak tail of the kinematically heated high-α disk (Feuillet et al. 2022). Despite different selection methods to define GSE, those
2020; Carrillo et al. 2024). The clustering algorithm we GSE samples all show enhancements of Eu and relatively low
[Ba/Eu] ratios. In our sample, certain members of GSE are also
adopted should reduce the contamination from the smooth
r-process enhanced. The median Eu abundance of the GSE
components, and the contamination of the high-α disk should
members with Eu measurements is 0.59, which is similar to the
not be significant at [Fe/H]< −2 since the number of high-α results from Aguado et al. (2021a; [Eu/Fe] = 0.59) and da
disk members drops sharply toward low metallicity. In Silva & Smiljanic (2023; [Eu/Fe] = 0.52). Specifically, GSE in
addition, the distribution of our GSE stars in the Jr−Lz plane our HR sample contains 16 r-process-enhanced stars, of which
( Jr = 25.5 kpc1 2 km1 2 s-1 2 and Lz = 113.8 kpc km s−1) is 11 are r-I stars12 and five are r-II stars.13 Two of the five r-II
similar to chemically selected GSE stars in Buder et al. (2022; stars have also been associated with GSE by Y20.
Jr = 26 kpc1 2 km1 2 s-1 2 and Lz = 100 kpc km s−1). However, when compared to our whole HR sample, GSE stars
GSE members do not have a clear decreasing trend in do not display higher Eu abundances and exhibit a scatter as large
α-elements with metallicity, which is in agreement with the as the other sample stars. This may be due to the low metallicity of
previous studies; that is, the [α/Fe] ratio begins to decrease at our sample since the large scatter of neutron-capture elements
[Fe/H] −2.0 (e.g., −2.0 in Matsuno et al. 2019; −1.6 in at [Fe/H]< −2 has been acknowledged for decades (e.g.,
Monty et al. 2020; −1.3 in Mackereth et al. 2019; and −1.1 in McWilliam 1998; Johnson & Bolte 2002; Aoki et al. 2005; Sneden
Horta et al. 2023). We notice that a large proportion of the Mg- et al. 2008 and references therein) and is also clearly shown in
poor stars ([Mg/Fe] < 0, as defined in Paper II) are associated Paper II. The scatter among halo stars can be explained by
with GSE, and this Mg-poor population in GSE is also found stochastic enrichments of r-process elements as a result of the
by Horta et al. (2023) at −1.8 < [Fe/H] < −0.8 based on data hierarchical formation of galaxies and the inhomogeneity in the
from APOGEE DR17. Among our HR sample, 50% (two out interstellar medium (Hirai et al. 2015; Ishimaru et al. 2015). It is
of four) of the Mg-poor stars belong to GSE, and this fraction is 12
The r-I stars are defined as those with 0.3 „ [Eu/Fe] „ 1.0, [Ba/Eu] < 0,
higher than the proportion of GSE to the whole sample (25%). following Sakari et al. (2018). This is slightly different from the definition we
These Mg-poor stars at all metallicity ranges could originate used in Paper II.
13
from dwarf galaxies accreted to GSE or from satellite dwarf [Eu/Fe] > 1.0, [Ba/Eu] < 0, which is the same as that used in Paper II.

8
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure 7. The chemical distribution of GSE. The element species is shown in the upper right of each subplot, and the average uncertainty is in the lower-left
region. Red dots represent GSE member stars and gray dots represent all stars with reliable measurements. The red stars with black borders are members of the r-
process-enhanced subgroup in GSE; see Section 4.2.3 for details.

plausible that the interstellar medium of the GSE was inhomoge- Chiba & Beers 2000; Beers et al. 2014; Li & Zhao 2017; Carollo
neous at its early phase of evolution, i.e., at low metallicity, and et al. 2019; Naidu et al. 2020). Recently, as stellar samples expand
GSE itself might be a product of mergers of smaller galaxies. We toward low metallicities, VMP stars with disk-like kinematics have
also note that our Eu abundance distribution might be biased been gradually revealed (e.g., Sestito et al. 2019, 2020, 2021; Di
because of the detection limit at low metallicity, where Eu lines Matteo et al. 2020; Venn et al. 2020; Carter et al. 2021; Cordoni
cannot be detected if a star has an intrinsically low Eu abundance. et al. 2021; Mardini et al. 2022; Bellazzini et al. 2024; Carollo
et al. 2023; Dovgal et al. 2024). However, the origin of these disk-
like VMP stars is still debated. For example, using the NIHAO-
4.1.2. VMPD
UHD simulations (Wang et al. 2015; Buck 2020), Sestito et al.
A thick-disk-like component of metal-poor stars, which is (2021) propose two scenarios to explain the origins of these stars:
known as the metal-weak thick disk, has been noticed for a the dominant part of them have formed in the building blocks of
long time and studied using various data sets (Norris et al. the proto-Galaxy, and the remaining stars have originated from
1985; Morrison et al. 1990; Beers & Sommer-Larsen 1995; later merger events.

9
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Several stars in our sample also demonstrate planar orbits with (Koppelman et al. 2019a; Matsuno et al. 2019; Myeong et al.
moderate eccentricity, providing us an opportunity to study the 2019). Naidu et al. (2020) argued that the high-energy
chemical properties of disk-like VMP stars. Based on the retrograde halo should be further divided into three, Arjuna,
distribution in the ec - ∣z∣max panel of our sample (Figure C1), Sequoia, and I’itoi, according to their different metallicity
we defined the disk-like VMP component as VMPD using distributions, peaked at −1.2, −1.6, and <−2, respectively. In
0.25 < ec < 0.60 and ∣z∣max < 3 kpc. In addition, we also used the following discussion, we still refer to this high-energy
rotational velocities and velocity distributions in the Toomre retrograde halo as Sequoia since the presence of the three
diagram to constrain VMPD with vf > −200 km s−1 and subsubstructures has not been established yet. The chemical
(vy - 233.1)2 + vx2 + vz2 < 180 km s-1. The kinematic proper- distributions of Thamnos and Sequoia are shown in Figure 9.
ties and metallicity distribution of our VMPD are similar to the We assigned the largest number of retrograde stars to Thamnos.
low-eccentricity low-∣z∣max prograde VMP component as identi- This is in agreement with a previous study by da Silva &
fied in, e.g., Cordoni et al. (2021). Smiljanic (2023), whose most-retrograde groups, dominated by
The chemical distribution of VMPD is shown in Figure 8. Thamnos, contain more VMP stars than other retrograde and
VMPD displays a smaller scatter among most elements prograde clusters. There is no obvious slope in Mg of Thamnos.
compared to the whole sample, except for J0705 + 2552, Horta et al. (2023) showed that Thamnos presents no α-knee
which is a CEMP star with a noticeable enhancement in Ba feature at [Fe/H] > −2 using elemental abundances from
([Ba/Fe] = 0.64) represented by the star symbol. Although this APOGEE DR17, which suggests that Thamnos likely quenched
star does not meet our criterion of CEMP-s stars (see star formation before the onset of Type Ia supernovae (SNe Ia).
Section 4.2.1), the nonnegligible enhancement of α-elements Our results do not contradict their conclusion.
and s-process elements (e.g., C, Mg, Y, Ba) is similar to the Monty et al. (2020) found that for iron-peak elements (Mn,
chemical pattern of CEMP-s stars, which indicates that it could Zn) and neutron-capture elements (Y, Ba), the distribution of
have been contaminated by an asymptotic giant branch (AGB) Thamnos has no obvious differences compared with other halo
companion. We thus exclude this star from the following stars, which is also confirmed in our analysis. Da Silva &
discussions on VMPD. Smiljanic (2023) proposed that two [Eu/Mg] sequences are
As shown in Figure 8, one clear feature of VMPD is its found in their Thamnos-dominated most-retrograde groups.
relatively low Zn abundance. The median [Zn/Fe] of VMPD is One sequence has lower metallicity and [Eu/Mg] ∼ 0.4, and
0.16 dex, while that of the entire HR sample is 0.25 dex. We the other has [Fe/H] > −1.5 and [Eu/Mg] ∼ 0.0, which may
conduct a Kolmogorov–Smirnov (KS) test on the distribution of be the contaminant from in situ components. All four Thamnos
[Zn/Fe] at [Fe/H] −3 between VMPD and the whole HR stars with Eu measurements in our sample show
sample. The result shows D = 0.47 with p = 0.02, indicating a [Eu/Mg] > 0.15, and the average value is 0.31, which is
significant difference. One suggested main production site of Zn
similar to the low-metallicity r-process-enhanced sequence in
for VMP stars is hypernovae (HNe), whose explosion energy is
da Silva & Smiljanic (2023).
one order of magnitude higher than regular supernovae (Umeda &
There are six Sequoia stars with robust [Fe/H] measurements.
Nomoto 2002; Kobayashi et al. 2006; Tominaga et al. 2007). In
Although low α-element abundances (e.g., [Mg/Fe] < 0.2) have
this scenario, a high [Zn/Fe] ratio would indicate either a high HN
been reported for Sequoia at [Fe/H] > −2 (Matsuno et al. 2019;
fraction or a high HN upper mass limit (e.g., Grimmett et al.
Monty et al. 2020; Matsuno et al. 2022b; Horta et al. 2023), our
2020), and hence it would correspond to the production of Zn
from relatively more massive stellar systems. For example, VMP Sequoia stars do not show such low Mg abundances. This
Mucciarelli et al. (2021) have reported that a globular cluster (GC) indicates that there is a decreasing trend in these elemental
in the Large Magellanic Cloud (LMC), NGC2005, shows abundances of Sequoia, supporting the idea that the low α-element
significantly lower [Zn/Fe] than the other GCs in the LMC or abundances at [Fe/H] > −2 are due to the large contribution from
those in the Milky Way. They argued that the low Zn abundance SNe Ia to the chemical enrichment in Sequoia. Unlike what was
could be explained by a chemical evolution model for a small demonstrated in Matsuno et al. (2022b), the Na and Zn of Sequoia
dwarf spheroidal galaxy that has merged with the LMC a long are not lower than those of other stars in our HR sample. In
time ago. addition, the low Y abundance of Sequoia indicated by Matsuno
Following the same reasoning, the low Zn abundance of et al. (2022b) is not seen in most of our Sequoia stars except for
VMPD suggests that its progenitor systems might have been one star with [Y/Fe] = −0.62 and [Y/Mg] = −0.99. These
small and experienced slow star formation. Our result supports differences are likely due to the different metallicity coverage;
the first scenario by Sestito et al. (2021) that the disk-like low- our sample has [Fe/H] < −2, while Matsuno et al. (2022b)
metallicity stars could have originated from low-mass building covers −1.8 < [Fe/H] < −1.4, where SNe Ia have started to
blocks that merged in the very early Universe and initiated the contribute to the chemical enrichment of the progenitor of Sequoia.
formation of the Milky Way. There are four stars with robust Eu abundance measurements
(σ[Eu/Fe] „ 0.2) in our six Sequoia stars, three of which are r-I
stars. These fractions are higher than the proportion in the entire
4.1.3. Retrograde Substructures: Thamnos and Sequoia/I’itoi sample, where ∼22% of stars have reliable Eu abundance
An excess of stars in the retrograde halo is considered to be measurements and ∼20% of stars are r-process enhanced. This
one of the signatures of early accretion events (Helmi et al. implies that Sequoia tends to have higher abundances of r-process
2017; Myeong et al. 2018a, 2018b). Helmi et al. (2018) elements on average. Aguado et al. (2021a) suggested that Sequoia
proposed that these retrograde stars belong to Gaia-Enceladus. members are r-process enhanced and show a tight and flat
However, the differences in chemical properties suggest that sequence with [Ba/Eu] ∼ −0.7. Considering that Aguado et al.
the retrograde components and GSE may come from different (2021a) selected Sequoia members with higher energy and that
merger events, which were subsequently identified as Sequoia Koppelman et al. (2019a) proposed a more strict selection method
and Thamnos, distinguished by their higher and lower energies of Sequoia with E > −1.35 × 105 km2 s−2, we divided the Sequoia

10
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure 8. The chemical distribution of VMPD. Blue dots represent VMPD member stars, and the blue star with black border is J0705 + 2552, which is a CEMP star
that might have been affected by mass transfer from an AGB companion. The meanings of other symbols are the same as in Figure 7.

into two groups with higher (E > −1.35 × 105 km2 s−2) and lower been identified after Gaia releases by Yuan et al. (2020a) and
energy (E < −1.35 × 105 km2 s−2) parts. All the three stars in the Naidu et al. (2020), LMS-1 and Wukong, respectively. Due to
high-energy Sequoia are r-I stars with [Eu/Fe] ∼ 0.7 and higher Sr their similar dynamics and the same GC memberships, LMS-1
and Ba abundances than the entire sample of the present work, and Wukong are considered to be the same substructure. We
which is similar to the finding of Aguado et al. (2021a) but at will refer to this substructure as Wukong in the following
lower metallicity. discussion. The chemical distributions of Helmi streams and
Wukong are shown in Figure 10.
We do not find CEMP stars in Helmi streams, and all stars
4.1.4. Polar Substructures: Helmi Streams and Wukong/LMS-1
with reliable carbon measurements (σ[C/Fe] „ 0.2) have
Helmi streams are characterized by their large perpendicular relatively low carbon abundances with [C/Fe] „ 0.11. This
component of angular momentum and the polar ring-like orbits fraction of CEMP stars in Helmi streams is significantly lower
of their members in the solar vicinity (Helmi et al. 1999; than the previous estimates (∼20%; see Aguado et al. 2021b;
Koppelman et al. 2018, 2019b). Another polar substructure has Gull et al. 2021). One possible reason for this result is that

11
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure 9. The chemical distribution of Thamnos and Sequoia. Colored symbols represent member stars in Thamnos (green dots) and Sequoia (orange inverted
triangles). The Sequoia members are divided into two parts, presented by inverted triangles with black borders (E > −1.35 × 105 km2 s−2) and with white borders
(E < −1.35 × 105 km2 s−2). The meanings of other symbols are the same as in Figure 7.

members of Helmi streams in our sample are mainly giants CEMP stars are found. The fraction of CEMP stars will be
(11/15), while in Aguado et al. (2021b) they are mainly main- discussed in detail in Section 4.2.1.
sequence/turnoff stars (54/62).14 As discussed in Section 3.4 The low CEMP fraction based on our HR sample does not
of Paper II, the fraction of CEMP stars is higher in turnoff stars contradict the previous findings that Helmi streams have a
(22%) than in giants (7.8%), so a higher giant fraction may relatively low C abundance. For example, Roederer et al. (2010)
indicate a lower fraction of CEMP stars. However, most of the reports subsolar carbon abundances of Helmi stream stars, and
stars in Gull et al. (2021) are also giants (10/12), and two Gull et al. (2021) also show that the fraction of CEMP stars is
lower in Helmi streams than in other dwarf galaxies such as
14
Sculptor and the progenitor of ω Cen, which agrees with our result.
We use same criteria as in Paper II to define turnoff stars Previous studies have found low Mg abundances of Helmi
(Teff  5500 K & log g  3.0 ), red giant branch (RGB) stars (Teff < 5500 K),
and horizontal branch (HB) stars (Teff  5500 K & log g < 3.0 ). Both RGB streams at [Fe/H] > −2.0 (e.g., [Mg/Fe] ∼ 0.2 in Aguado et al.
stars and HB stars are considered giants in this work. 2021b and [Mg/Fe] ∼ 0.1 in Matsuno et al. 2022a), while most

12
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure 10. The chemical distribution of Helmi streams and Wukong/LMS-1. Colored dots represent member stars in Helmi streams (brown) and Wukong/LMS-1
(blue). The meanings of other symbols are the same as in Figure 7.

of our Helmi stream members have [Fe/H]< −2.0 with a of Helmi streams is also ∼0.1 dex lower than the entire sample,
median Mg abundance of [Mg/Fe] = 0.4. This implies that while another light neutron-capture element, Y, does not have
Helmi streams should have a decreasing trend in Mg such a feature. Although low light neutron-capture element
abundance as found in other substructures (e.g., GSE and abundances of Helmi streams are also noticed by Aguado et al.
Sequoia). (2021b) and Matsuno et al. (2022a), they report a more
The members of Helmi streams show lower light neutron- significant deficiency of Sr and Y in the Helmi stream
capture element abundances, such as Sr and Zr, compared with (>0.5 dex) than our findings.
the entire sample. For [Fe/H] > −3, where the scatter of Sr is Unlike the light neutron-capture elements, the distribution of
relatively small, the median Sr abundance of Helmi streams is heavier neutron-capture elements does not show significant
0.14, which is ∼0.1 dex lower than the whole sample, and for differences between Helmi streams and the other stars. Among
[Fe/H] < −3 three Helmi stream members have extremely low the four stars with robust Eu measurements in Helmi streams,
Sr abundances with [Sr/Fe] ∼ −1. The median Zr abundance three are r-process enhanced, of which two are r-I stars and one

13
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

is an r-II star. This indicates that the r-process dominates the identified a new substructure named Pontus in the retrograde
nucleosynthesis of the heavier neutron-capture elements in halo using the clustering algorithm, which overlaps with GSE
Helmi streams. This is consistent with previous studies and Thamnos in the energy–action phase space. Two DTGs in
(Limberg et al. 2021b; Gull et al. 2021). our sample might be associated with Pontus. Given that the
The deficiency of light neutron-capture elements in Helmi chemical and dynamical properties of Pontus remain unclear,
streams can be attributed to the lack of production sites of these we discuss it together with new DTGs in this paper. The
elements in the early Universe within its progenitor, such as chemical distributions of these four DTGs and Pontus are
rotating massive stars (Frischknecht et al. 2012; Choplin et al. shown in Figure 11.
2018), collapsars (Siegel et al. 2019), and electron-capture DTG-8 is a prograde group with low-eccentricity orbits, but
supernovae (Wanajo et al. 2011). However, the distribution of it is not associated with VMPD because of its relatively large
heavier neutron-capture elements in Helmi streams is similar to ∣z∣max . DTG-8 also shows different chemical distributions from
that of other stars, hence the progenitor of Helmi streams likely VMPD. Specifically, the members of DTG-8 show heavy
lacked rotating massive stars, which was also proposed by
element enhancements and no significant deficiency in Zn. Two
Matsuno et al. (2022a). Rotating massive stars might also
stars in DTG-8 with Eu abundance measurements are both r-II
contribute to the formation of some CEMP-no stars at
stars, and the [Ba/Eu] of these two stars are −0.59 and −0.80,
extremely low metallicity (e.g., see Maeder & Meynet 2015;
Maeder et al. 2015; Liu et al. 2021; Jeena et al. 2023, and implying an r-process-dominated nucleosynthesis for heavy
references therein).15 A smaller number of rotating massive elements in DTG-8. The [Eu/Mg] of these two stars are 0.63
stars also explains the lower carbon abundances of Helmi and 1.08, which requires a significant r-process element
streams to some extent. production in the progenitor system of DTG-8 through several
We associated DTGs with Helmi streams using the r-process events other than core-collapse supernovae (CCSNe),
z-component and the perpendicular component of angular such as neutron star mergers. Although the other five stars do
momentum, following the criterion proposed by Koppelman not have Eu measurements, the enhancement of other heavy
et al. (2019b). The member stars in Helmi streams show two elements, such as Ba and La, still indicates that DTG-8 is likely
sequences in radial action phase space, which is also noticed enhanced in heavy elements in general.
by Y20 and studied in detail by Dodd et al. (2022). Compared DTG-9 is a polar component with small net rotational
with the low radial action component, stars with higher radial velocity, which has been noticed by several works (e.g., Y20;
action have higher energy but the same vertical motion and Limberg et al. 2021a; Shank et al. 2022a) and labeled as ZY20:
action. As discussed in Y20, this high radial action part may be DTG-35, GL21:DTG-8, and DS22:DTG-6, respectively.16
stripped at an earlier pericentric passage from the progenitor of DTG-20 is a planar substructure with moderate eccentricity,
Helmi streams or come from a different origin. Matsuno et al. whose median ∣z∣max is lower than 1 kpc. The chemical
(2022a) suggested that the stars with high energy (E ∼ −1.0 × distributions of DTG-9 and DTG-20 are similar to those of
105 km2 s−2) may not come from the same progenitor as Helmi the entire sample. However, the number of stars in these two
streams, because the high-energy star Gaia EDR3 groups is not large enough to draw any conclusions about their
2447968154259005952 has different chemical abundance progenitors.
ratios in α-elements and neutron-capture elements from the DTG-49 is a high-energy prograde group that has also been
other stars in their sample. However, in our HR sample, there clustered by Y20 as ZY20:DTG-2. DTG-49 shows a significant
are no significant differences in the chemical distribution deficiency in Na and Mg, and two of the stars also have low Ca
between the high radial action sequence and the other members and Ti abundances, suggesting a low star formation efficiency in
of Helmi streams, so both of these two sequences are the progenitor of this group. Only three stars are assigned to
considered Helmi streams. this group in the HR sample, so it is difficult to obtain any reliable
The chemodynamical studies of the Wukong substructure are constraints on its progenitor. However, such low Mg and
still limited, and its selection criteria are still unclear. In this Na abundances (e.g., [Mg/Fe] = 0.05 and [Na/Fe] = −0.51) at
work, we refer to a relatively strict criterion from Limberg et al. [Fe/H] ∼ −3 imply that it is likely to be from an accreted low-
(2024) to associate stars in our sample with Wukong. We have mass dwarf galaxy, which provides us with a unique perspective
only four Wukong members from the HR sample.
to investigate the chemical properties of low-mass dwarf galaxies
Wukong stars have relatively higher α-element abundances and
with solar neighborhood stars. Therefore, a high-resolution
do not show a significant decreasing trend at [Fe/H] < −2.2,
which is consistent with results from Limberg et al. (2024), follow-up of other DTG-49 members in the LR sample is highly
indicating a relatively massive progenitor of Wukong. For other desired.
elements, there are too few stars in Wukong to discuss their Pontus has three stars with elemental abundances, one of
chemical properties, so we only provide the distribution of these which has a metallicity uncertainty greater than 0.2 and should
stars for reference. be excluded from the chemical discussion. Therefore, only two
Pontus members are available, which is too few to draw any
conclusions from. We simply provide the chemical distribution
4.1.5. Other DTGs of these two stars in Figure 11 for reference. We notice that
As discussed in Section 3.2 and presented in Figure 2, there only two stars are associated with Pontus in Horta et al. (2023),
are four DTGs, namely DTG-8, DTG-9, DTG-20, and DTG-49, too. A larger sample of Pontus is needed for the further study of
which contain at least three HR sample stars but are not 16
associated with known substructures. Recently, Malhan (2022) Here, we follow the nomenclature proposed by Y20 and use XXYY:DTG-x
to refer to DTGs identified in previous work, where XX are the initials of the
names of the first author, YY are the last two digits of the year of publication,
15
CEMP-no stars are CEMP stars that do not present enhancements in heavy and DTG-x is the number of the DTG. For example, our DTG-1 will be
elements. referenced as RZ24:DTG-1.

14
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure 11. The chemical distribution of DTG-8, DTG-9, DTG-20, DTG-49, and Pontus. Colored diamonds represent member stars in DTG-8 (pink), DTG-9 (green),
DTG-20 (blue), DTG-49 (orange), and Pontus (purple). The meanings of other symbols are the same as in Figure 7.

its chemical properties and its relationship with GSE and 4.2.1. CEMP Stars
Thamnos in detail.
The distribution of CEMP (especially CEMP-no) stars could
be rather sensitive to the environments of their progenitor
4.2. Kinematics of Stars with Particular Abundance Features systems. It is thus of paramount interest to investigate the
The chemical properties within each individual substructure fraction and the C abundance distribution of CEMP stars in our
and DTG have been discussed in Section 4.1. In the following HR sample.
section, we will compare the kinematic properties of CEMP Adopting the same criterion of CEMP stars as in Paper II,
stars, α-peculiar stars, and neutron-capture-element peculiar i.e., [C/Fe] + 0.7 for stars with log (L L )  2.3 and [C/Fe]
stars. Note that we do not exclude stars with relatively large 3.0 − log (L /L ) for more luminous objects (Aoki et al.
abundance uncertainties in this section, and thus the number of 2007), there are 47 CEMP stars in the HR sample. Note that
stars having these particular abundance features is slightly adopting the correction on the stellar evolution by Placco et al.
different from those in Section 4.1. (2014) would not make any difference in the selected CEMP

15
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Table 3
CEMP Star Fraction of Known Substructures, DTGs, and Nonclustered Stars

Substructure CEMP CEMP-s CEMP-no C-normala Members CEMP-no ratiob

Turnoff Giants Turnoff Giants Turnoff Giants Turnoff Giants Turnoff Giants
GSE 12 2 0 7 3 20 35 46 43 25.9% 7.9%
VMPD 2 1 1 0 0 8 7 10 8 0 0
Thamnos 4 1 0 3 0 7 8 15 8 30.0% 0
Helmi streams 0 0 0 0 0 2 8 4 11 0 0
Sequoia 1 0 0 0 1 0 5 1 6 0 16.7%
Wukong 0 0 0 0 0 1 1 2 2 0 0
Pontus 1 0 0 1 0 1 1 2 1 50.0% 0
other DTGs 9 1 1 6 1 12 38 27 41 33.3% 2.6%
Nonclustered 19 3 2 9 5 14 61 47 78 39.1% 7.6%

Total 48 8 4 26 10 65 164 154 198 28.6% 5.7%

Notes.
a
C-normal refers to stars with a [C/Fe] value or upper limit less than 0.7.
b
CEMP-no ratio = CEMP-no/(CEMP-no + C-normal), if the numbers of CEMP-no and C-normal stars are both equal to 0, this ratio is 0.

stars from the HR sample, as pointed out by Paper II. As

discussed at the beginning of Section 4, ten of them with
[Ba/Fe] > 1.0 are classified as CEMP-s stars. Additionally,
based on noticeable enhancement in Ba and other s-process
elements, another CEMP star J0705 + 2552 (discussed in
Section 4.1.2) and a mildly C-enhanced ([C/Fe] = 0.56) giant
J0446 + 2124 are also treated as (potential) CEMP-s stars in
this study. In summary, there are 12 CEMP-s stars and 36
CEMP-no stars.
To investigate whether the fraction of CEMP-no stars varies
with their birth environments, we examined numbers and ratios
of CEMP-no stars in different substructures, as shown in
Table 3. For DTGs not associated with known substructures,
including the four large DTGs discussed in Section 4.1.5, they
are referred to as other DTGs. All of the nonclustered stars are
considered as a mixture of smooth components and small
accreted components that have too few member stars to be
clustered in our sample. The CEMP-no ratios are computed by
CEMP-no/(CEMP-no + C-normal), where C-normal refers to
stars with a [C/Fe] value or upper limit less than 0.7. In
general, for the clustered stars, the fraction of CEMP-no stars Figure 12. The C abundance distribution of all CEMP stars in the HR sample.
The solid dots refer to CEMP-no stars and open squares refer to CEMP-s stars.
tends to be higher among turnoff stars than that among giants, Red and gray symbols represent the clustered and nonclustered CEMP stars,
which is consistent with the whole HR sample. For GSE and respectively. The average uncertainties are shown in the upper left of this
Thamnos, the fractions of CEMP-no stars in both turnoff stars figure. The morphological classification from the Yoon–Beers diagram (see
and giants are similar to those in nonclustered stars. However, Figure 1 in Yoon et al. 2016) has been overplotted.
for VMPD and the Helmi streams, which have statistically
meaningful numbers of objects in the HR sample, there are no sample in the A(C)−[Fe/H] diagram, as shown in Figure 12.
CEMP-no stars detected. This may be related to the properties There are quite a number of CEMP stars belonging to Group II
of their progenitor systems. For example, as shown in the and a few to Group III. For both groups, there are more clustered
merger-tree model by Salvadori & Skúladóttir (2015), CEMP- CEMP stars than the nonclustered ones, which is consistent with
no stars should exist in all dwarf galaxies within the observed the hypothesis that these two groups are more closely related to
luminosity range; however, as the galaxy luminosity increases, dwarf and/or UFD galaxies (e.g., Yoon et al. 2019).
the overall probability of observing CEMP-no stars decreases. Zepeda et al. (2023), analyzing CEMP stars from various
Such a theoretical prediction has been recently confirmed by abundance studies, identified 99 CEMP stars in GSE, 15
the discovery of an extremely metal-poor CEMP-no star in CEMP stars each in both Wukong and Thamnos, and none in
Sculptor (Skúladóttir et al. 2024). Therefore, a larger sample of the Helmi streams. The relative number of CEMP stars found
VMPD and Helmi stream stars is required to explore whether in their compiled sample is in general agreement with our
the negative detection of CEMP-no stars in these two results; that is, compared with GSE, Wukong, and Thamnos,
substructures reflects the properties of their progenitor systems Helmi streams tend to contain fewer CEMP stars. Furthermore,
or is caused by the limited size of our HR sample. the distribution of their sample in the A(C)−[Fe/H] diagram
Yoon et al. (2016) classifies CEMP stars into three groups in favors a scenario in which the Group II CEMP stars are more
the A(C)−[Fe/H] diagram that would reflect the primary sources likely to be formed in low-mass dwarf galaxies that do not
of their carbon enhancements. We placed all CEMP stars in our support extended star formation environments, which is

16
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

enrichment of the interstellar matter, while PISNe would also

lead to a subsolar [Mg/Ca], but lower [C/Fe] and [Na/Fe] are
also expected. As discussed in Paper II, most of the Mg-poor
stars show subsolar [Mg/Ca] ratios, but no Mg-poor stars fit
the chemical pattern of PISNe in our sample, which is
consistent with the increased contribution of SNe Ia. Therefore,
these Mg-poor stars may have been born in dwarf galaxies with
extremely low star formation rates, where the SNe Ia began to
contribute to the chemical enrichment below [Fe/H]< −2.5.
These dwarf galaxies were then accreted by a massive system,
such as the Milky Way or GSE.

4.2.3. Neutron-capture-element Peculiar Stars

As discussed in Sections 4.1.1 through 4.1.5, abundances of
neutron-capture elements generally present rather large scatter
in most substructures. Based on the chemodynamical analysis
of our HR sample, there is no clear evidence indicating whether
stars that have been accreted are more likely to be enhanced in
neutron-capture elements. This may indicate that neutron-
Figure 13. The Mg abundance distribution of the HR sample. All Mg-normal capture elements were synthesized by various production sites
stars are represented by gray open dots and the average uncertainty of the Mg- in the early times, which led to the diversity in the level of
normal stars is shown in the lower left of this panel. Mg-poor stars are heavy elements enrichment in the low-metallicity region.
represented by solid dots and Mg-rich stars by squares. The symbols with black However, there is one interesting dynamical subgroup of
borders stand for CEMP-s stars. Red, sky blue, and gray stand for members of
GSE, VMPD, and nonclustered stars, respectively. The dashed and dotted lines
r-process-enhanced stars in GSE.
are [Mg/Fe] = 0 and [Mg/Fe] = 1. Among our GSE stars five are on particularly eccentric orbits,
having high radial actions. Because of the slight differences in the
consistent with what we have found based on our VMP star vertical actions, they do not belong to a single DTG and, instead,
sample. belong to three DTGs. Interestingly, this dynamically defined
Moreover, we note that there is a peculiar ultra metal-poor group contains a high fraction of r-II stars (see stellar symbols in
CEMP-no star J2217 + 2104 ([Fe/H] = −4.12, A(C) = 5.23) Figure 7), with four out of five stars being r-II stars; the remaining
showing [C/Mg]∼ −0.5, which has been studied in detail by one is still an r-I star. All the five stars are giants with low
Aoki et al. (2018). Based on the dynamical analysis in this [Ba/Eu] (Figure 14), indicating that their neutron-capture
work, J2217 + 2104 turns out to be an accreted star, which is elements are mostly produced through the r-process nucleosynth-
associated with GSE. We also note that one CEMP-no star with esis. Moreover, they display similar abundance patterns in α-
[C/Fe] > 2, which is found in the area of Group I in Figure 12, elements, iron-peak elements (e.g., Cr, Mn, Ni), and neutron-
is J0150 + 2149 ([Fe/H] = −2.95, A(C) = 7.81; Paper II). This capture elements, suggesting that the five stars might share a
star is also associated with GSE. common origin. These stars show a wide metallicity distribution
from ∼ −3.5 to ∼ −2.0, which differs greatly from GCs.
We suggest that this r-process-enhanced subgroup of GSE may
4.2.2. α-peculiar Stars have originated from a dwarf galaxy similar to the UFD galaxy
We follow the criteria in Paper II to divide the HR sample stars Ret II (Ji et al. 2016). It is not clear at the moment whether the
into three parts according to their Mg abundances, namely Mg- progenitor galaxy has been accreted before or after the merger
poor stars ([Mg/Fe] < 0), Mg-rich stars ([Mg/Fe] > 1) and Mg- between GSE and the Milky Way or happens to have similar
normal stars (0 „ [Mg/Fe] „ 1). As shown in Figure 13, there are kinematics as GSE. A detailed study of this subgroup with a larger
ten Mg-poor stars and four Mg-rich stars in the HR sample. sample and theoretical simulations of galaxy mergers and
Among the four Mg-rich stars, two are from GSE, one is chemical enrichments would be necessary to completely unravel
from VMPD, and one is not clustered. All of the Mg-rich stars its origin. It would also be of interest to study the abundance
are CEMP stars. Three of them are CEMP-s stars whose patterns for more elements, especially actinides.
abundances are influenced by their AGB companion, and the This group is different from what has been studied in
remaining one is the ultra metal-poor CEMP-no star Roederer et al. (2018) and Hattori et al. (2023) in the sense that
J2217 + 2104, which is mentioned in Section 4.2.1. We do while they studied clustering in kinematics among r-rich stars,
not discuss the kinematic properties of these Mg-rich stars we found a group of stars clustering in kinematics that turned
because there is only one star if the CEMP-s stars are excluded. out to contain a high fraction of r-rich stars. It is thus the first
Four Mg-poor stars are from GSE, and the remaining six are example of a compact dynamical group with extreme r-process
not clustered. Nonclustered stars may form in situ or come enhancement in the GSE. The stars are relatively bright (Gaia
from dynamically relaxed accretion events. The deficiency in G < 15 mag) and nearby, providing us with a valuable
Mg can be explained by the contribution of SNe Ia (Ivans et al. laboratory to obtain full r-process element patterns of r-rich
2003; Caffau et al. 2013) or pair-instability supernovae (PISNe, stars that were likely born in a dwarf galaxy. Further studies on
Aoki et al. 2014; Xing et al. 2023). SNe Ia can produce Ca but a more complete abundance pattern of heavy elements, e.g.,
hardly produce Mg; therefore, a subsolar [Mg/Ca] ratio may from Sr through Th, will enable us to reveal the origin of the
indicate that the SNe Ia have contributed to the chemical r-process elements in this subgroup. Such detailed abundance

17
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

constraints on the role of the halo mass of the accreted dwarf

galaxies in their chemical enrichment and tidal interaction with the
Milky Way. Since massive galaxies suffer from stronger tidal
friction during the accretion onto the Milky Way, they are
expected to deposit stars deeper into the gravitational potential of
the Milky Way (Amorisco 2017). Therefore, if the chemical
evolution of dwarf galaxies primarily depends on their halo
masses, we might be able to observe correlations between
chemical abundances and kinematics among halo stars in the
present Milky Way. While, for example, Gratton et al. (2003),
Venn et al. (2004), Ishigaki et al. (2012), and Monty et al. (2020)
studied chemodynamical correlations among halo stars, including
the very metal-poor range, we can now study such correlations
among a large sample of VMP stars using precise kinematics from
the Gaia mission.
We examine correlations between chemical abundance ratios
and kinematics in Figures 15–17. We choose Jf, Jr , and Jz as
the variables representing kinematics of stars to maintain the
consistency with our clustering analysis described earlier. The
three quantities respectively enable us to investigate how the
chemical properties vary with the orbital angular momentum in
the disk rotation direction of the Milky Way, the radial excursion,
and the vertical excursion. More massive accreted galaxies are
expected to show small absolute values in the actions. We include
in each panel Pearson’s correlation coefficient and partial
correlation coefficient, and the corresponding p-values with
p < 0.01 are considered to be significant.17 The partial correla-
Figure 14. The distribution of the r-process-enhanced subgroup in GSE in the
tions are adopted to minimize the combined effect of our
Jr -[Ba/Eu] panel (top) and the HR diagram (bottom). The gray dots and the sample selection and chemical evolution. For the panels
black error bars in the upper left of each panel are the value and typical showing correlations between [Fe/H] and kinematics, we take
uncertainties of the whole sample (median uncertainty for action and average the distances to the stars as the controlling variable to test if the
uncertainties for other quantities), and the red dots in the top panel are all GSE observed correlations can be attributed to the correlation
member stars. The subgroup members are presented by stars colored in red
(top) and colored by their metallicity (bottom). In the top panel, the pure-r between [Fe/H] and distance (Figure 2). For the other panels,
production is from Frebel & Ji (2023). In the bottom panel, the red error bars we take [Fe/H] as the controlling variable since [Fe/H] may
represent the average uncertainties of the subgroup members. show correlations with kinematics due to the sample selection
and because [X/Fe] ratios evolve with metallicity due to
chemical evolution (see Figures 15, 18, and 20 in Paper II),
patterns would also allow us to confirm the common origin of which can then produce an artificial correlation between
these stars if they turn out to be coherent within this group. [X/Fe] and kinematics even when [X/Fe] evolves similarly
We also study the dynamical origin of 25 neutron-capture- with metallicity regardless of kinematics.
element-poor VMP stars whose [Sr/Fe]< −0.5 and [Ba/Fe] We observe a significant [Fe/H] gradient as a function of
< −0.5. Twelve of these stars are not clustered in our sample, Jz in the sense that lower-metallicity stars tend to show large
while five are not associated with the known substructures. vertical actions. However, if we take the distance as the control
Among the remaining eight stars, four are associated with GSE, variable and compute the partial correlation coefficient between
two with Helmi streams, and one each from Thamnos and [Fe/H] and Jz , the correlation becomes insignificant. Hence,
Wukong. There is no significant connection between the it is necessary to conduct a study of the metallicity gradient
distribution of neutron-capture-element-poor VMP stars and using a volume-complete sample that is unbiased in metallicity,
the different substructures. such as in Youakim et al. (2020).
Regardless of the origin of the metallicity gradients, we can still
4.3. Global View determine if there are intrinsic correlations between [X/Fe] and
kinematics since our sample selection does not bias [X/Fe] at a
Here, we discuss the global trend and correlation among given metallicity. Figures 15–17 show that there are significant
chemodynamics of halo stars without classifying stars into correlations between C, Na, V, Cr, Zn, Sr, and Ba abundances and
substructures or DTGs. Such global trends could potentially Jz , and between Sr abundance and Jf. Once conditioned on
enable us to observe signatures of accreted galaxies that are yet to [Fe/H], all but the one between [C/Fe] and Jz become
be discovered from kinematics. For example, Venn et al. (2004) insignificant. Therefore, most of the significant correlations
found that there is a significant trend between α-element observed can be explained by the combination of the [Fe/H]
abundances and velocities of halo stars in the solar neighborhood gradient and the evolution of [X/Fe] with metallicity. Even
in the sense that stars on extremely retrograde orbits have low though there is a metallicity gradient in the halo as a function of
[α/Fe] ratios. Thanks to recent advances in the field, the
extremely retrograde component is now considered to correspond 17
We compute the p-values for partial correlation (r) by converting r to
to a kinematic substructure, Sequoia (Matsuno et al. 2019; Monty t = r (N - 3) (1 - r 2 ) , which follows the t-distribution with the N − 3
et al. 2020). In addition, global trends might also provide degree of freedom, where N is the number of data points.

18
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure 15. Correlations between chemical abundance ratios and actions. Except for the panels in the first row, we color code stars according to the metallicity. Partial
correlation coefficients are computed taking the distance and [Fe/H] as the controlling variable for [Fe/H] and the other panels, respectively. We highlight correlations
that are highly significant with the p-value smaller than 0.01.

kinematics, abundance ratios at a fixed [Fe/H] do not seem to be related to the varying fraction of CEMP stars in different
show gradients, indicating that there is not much variation in substructures as we discussed in Section 4.2.1, hinting at a
chemical evolution in the very metal-poor regime. correlation between [C/Fe] in a galaxy and its halo mass.
The negative correlation between [C/Fe] and Jz needs However, C is one of the elements whose surface abundance
attention. The correlation seems highly significant: the p-value is decreases as low-mass stars evolve, which complicates the
0.0002, which is clearly below our threshold of p = 0.01. While interpretation. The correlation might be explained by different
we have 60 combinations of chemical abundance and kinematics distributions of turnoff stars and red giants in kinematics.
in the figures, the probability of all 60 pairs of random variables While we have shown that most of the elemental abundance
showing p > 0.0002 is 1 − (1 − 0.0002)60 = 98.8%. This might ratios [X/Fe] do not show significant correlations with

19
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure 16. Same as Figure 15, but for iron-group elements.

kinematics, the above investigations are limited to comparisons 1 1 1

between one-dimensional chemical information and one-
Bij = bij - å
n k
bik - å blj + 2 å bk, l .
n l n k,l
(6 )
dimensional kinematic information. We know that both
chemical and kinematics spaces are multidimensional; for The distance correlation Rd is then defined as the square root of
example, kinematic substructures cannot be found in a one- 1
dimensional investigation. We thus examined a correlation å A B
n 2 i, j ij ij
Rd2 (X , Y) = . (7 )
between multidimensional chemical information and three- 1 1
dimensional kinematics using the distance correlation defined å A2
n 2 i, j ij
å B2
n 2 i, j ij
in Székely et al. (2007).
The distance correlation takes a value in 0 „ Rd „ 1 and
We first provide a brief summary of distance correlation
based on Székely et al. (2007). We considered a distance satisfies Rd = 0 if and only if X and Y are independent. When
correlation of N data points between X and Y spaces, which can Rd = 1, there is a relation between X and Y such that
be multidimensional. Using distance matrices in the X and Y Y = A + bXC, where A is a vector, b is a scalar, and C is an
spaces, orthonormal matrix. The null hypothesis that X and Y are
independent can be tested by computing the Rd distribution for
aij = ∣Xi - Xj∣ (3 ) samples with the order of Y shuffled.18
For the chemical space, we considered eight elemental
bij = ∣Yi - Yj∣ , (4 ) abundance ratios ([X/Fe], where X are Na, Mg, Ca, Sc, Ti, Cr,
where distances are computed in the Euclidean geometry, 18
The distance correlation can approach 1 even if X and Y are independent
Székely et al. (2007) defined the following two matrices: when their dimensions are high. In such case, we could use the modified
distance correlation by Székely & Rizzo (2013), for which the null hypothesis
1 1 1 can be tested by using a t-distribution if the dimensions of X and Y are large.
Aij = aij - å aik - n å alj + n2 å a k,l
n k
(5 ) Since our data do not have such large dimensions, we stuck to the original
l k,l distance correlation in Székely et al. (2007).

20
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure 17. Same as Figure 15, but for neutron-capture elements.

Sr, and Ba) since these elemental abundances are available for a dynamical evolution of the Galaxy. While massive accreted
large number of stars. By requiring stars to have all of these galaxies are expected to reside closer to the Milky Way’s center
ratios, we were left with 273 stars. For the kinematics, we due to tidal interactions, our analysis revealed no statistically
considered the space defined by Jf, Jz, and Jr. We further significant correlation between their chemical composition and
normalized these chemical abundances and kinematics so that kinematics. This suggests that mass may not be the primary driver
the median value became zero and the difference between the of early chemical enrichment in these dwarf galaxies.
16th and 84th percentiles became 2.
We obtained the distance correlation coefficient of 0.29 with
p < 0.001, which seems significant. However, we again note 5. Conclusion
that this might simply reflect the metallicity gradient and In this work, we have conducted a chemodynamical analysis
chemical evolution. Therefore, we ran the same test after on one of the largest uniform high-resolution VMP star
removing a linear trend between [X/Fe] and [Fe/H], which samples, which has been obtained from a joint project between
yielded the correlation coefficient of 0.23 with p = 0.07, LAMOST and Subaru, i.e., the HR sample. To ensure a robust
suggesting again that the observed significant correlation can clustering, the identification of dynamical groups has first been
be explained by the chemical evolution. made on the master catalog of the HR sample, which consists
Notably, the absence of a significant correlation here contrasts of 5778 VMP stars mostly based on the LAMOST low-
with the small abundance variations among substructures reported resolution VMP catalog and is referred to as the LR sample.
in Section 4.1, such as the lower Zn abundances observed in Dynamical parameters have been estimated using Agama, and
VMPD. Unlike Section 4.1, which delves into specific kinematic the FoF algorithm has been adopted for the clustering. The
substructures through targeted chemical and kinematic analyses, analysis has resulted in 131 DTGs, including 227 stars from the
this section investigates the broader picture across the Milky Way. HR sample, 89 of which have been later associated with several
Here, we focus on identifying global trends in element distributions known Galactic substructures, including GSE, Helmi streams,
and stellar motions to understand the overall chemical and Thamnos, Sequoia, Wukong, and Pontus, as well as a very

21
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

metal-poor disk-like substructure called VMPD. It thus enabled Chemodynamics of a large VMP star sample is crucial to
us to systematically explore the chemical properties of these understanding the origin of these ancient stars and could provide
substructures and dynamical groups down to such a low- important observational constraints on the properties of the early
metallicity region with [Fe/H] < −2.0. The major results Milky Way and ancient accretions of small dwarf galaxies.
include: Though there have been great advances made in relevant fields,
there is still a lack of comprehensive chemodynamical studies on
1. About one-fourth of the HR sample belongs to GSE. In the VMP region of the Galactic halo. Therefore, this paper has
general, the distribution of elemental abundances of GSE made such an effort to conduct a comprehensive chemodynamical
is similar to that of the whole HR sample and the study on a large high-resolution spectroscopically analyzed VMP
nonclustered VMP stars, accompanied by large scatters, star sample, which provides uniform chemical abundances for
indicating a massive progenitor system with a compli- over 20 species and thus enables an investigation of the internal
cated enrichment history of GSE. chemical properties of various substructures.
2. The VMPD shares similar distributions in kinematics and We have found a number of interesting features for some
metallicities as the low-eccentricity, prograde, very substructures; however, in general, we find little significant
metal-poor disk components identified in previous difference in the abundance trends for different elements
studies. We identify a systematically lower [Zn/Fe] between VMP substructures and nonclustered VMP halo stars.
compared with the whole HR sample, indicating that the This may partly be related to the fact that at such early times
VMPD may have originated from small stellar systems. with very low metallicities, the chemical enrichment of most
Such a feature supports the scenario that stars associated galaxies is dominated by nucleosynthesis through CCSNe.
with the VMPD have probably originated from low-mass Even though our HR sample is one of the largest VMP star
building blocks of the proto-Galaxy. catalogs with homogeneously derived abundances for over 350
3. All of the stars in high-energy Sequoia (E > −1.35 × objects and we utilized a much larger LR sample with about
105 km2 s−2) are r-I stars with [Eu/Fe] ∼ 0.7, revealing the 5800 VMP stars for the clustering, the number of VMP stars in
r-process-enhanced feature of Sequoia at [Fe/H] < −2. our sample of the smallest galaxies such as UFDs is insufficient
4. Helmi streams show deficiencies in carbon and light for robust clustering analysis or to observe chemical signatures
neutron-capture elements. This can be explained by the of individual accreted small galaxies. Future spectroscopic
lack of rotating massive stars in the progenitor of Helmi surveys, such as 4MOST, WEAVE, and PFS, will significantly
streams. The chemical distribution of other substructures increase the number of VMP stars with abundances of various
and DTGs is rather similar to the whole HR sample. elements, which shall allow us to ultimately upscale this study.
5. The fraction of CEMP-no stars seems to vary among
different substructures in the HR sample. We find that, in
general, the Group II and III CEMP stars in the A(C)− Acknowledgments
[Fe/H] diagram are more likely to be associated with We thank the anonymous referee for the very efficient and
accreted small dwarf galaxies, which is consistent with helpful report to improve the draft. This work was supported by
previous studies. For GSE and Thamnos, the fractions of the National Natural Science Foundation of China grant Nos.
CEMP-no stars in both turnoff stars and giants are quite 11988101, 12222305, and 11973049, the National Key R&D
similar to that in nonclustered stars, while for VMPD and Program of China No.2019YFA0405500, the JSPS–CAS Joint
the Helmi streams, there are no CEMP-no stars detected Research Program, the CAS Project for Young Scientists in Basic
in the HR sample. Although such differences can be Research (No. YSBR-092), and the science research grants from
related to the properties of their progenitor dwarf the China Manned Space Project. This work was also supported
galaxies, larger samples of VMP stars would be necessary by JSPS KAKENHI grant Nos. 20H05855, 21H04499,
to unravel their origins. 22K03688, and 23HP8014. H.L. and Q.X. acknowledge support
6. There is no clear evidence of whether neutron-capture- from the Strategic Priority Research Program of Chinese
element peculiar stars are more likely to be accreted. Academy of Sciences grant No. XDB34020205 and the Youth
However, we do find a very interesting subgroup in GSE Innovation Promotion Association of the CAS (id. Y202017 and
that exhibits a significant excess of r-process material. 2020058). Q.X. acknowledges support from the Beijing Munici-
Stars in this subgroup provide a great opportunity to pal Natural Science Foundation grant No. 1242031. T.M. is
explore the complete r-process pattern and, thus, the supported by a Gliese Fellowship at the Zentrum für Astronomie,
r-process nucleosynthesis in GSE. University of Heidelberg, Germany.
7. It is expected that more massive galaxies sink deeper in This research is based on data collected at Subaru Telescope,
the Milky Way potential, which would create a gradient which is operated by the National Astronomical Observatory of
in the progenitor galaxy mass as a function of kinematics. Japan. We are honored and grateful for the opportunity of
We studied if we can observe the signature of this observing the Universe from Maunakea, which has cultural,
gradient in abundance ratios by studying large-scale historical, and natural significance in Hawaii. Guoshoujing
correlations between kinematics and chemical abun- Telescope (the Large Sky Area Multi-Object Fiber Spectro-
dances among VMP stars. For most elements, there is scopic Telescope, LAMOST) is a National Major Scientific
no significant correlation, indicating that accreted Project built by the Chinese Academy of Sciences. Funding for
galaxies of the Milky Way experienced similar chemical the project has been provided by the National Development and
evolutions at early times. The only exception is C, which Reform Commission. It is operated and managed by the
shows a significant correlation with vertical action. National Astronomical Observatories, Chinese Academy of
However, since the C abundance changes during the Sciences. This work has made use of data from the European
evolution of low-mass stars, it is needed to confirm the Space Agency (ESA) mission Gaia (https://www.cosmos.esa.
correlation with a better-controlled, larger sample. int/gaia), processed by the Gaia Data Processing and Analysis

22
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/ Appendix B

dpac/consortium). Funding for the DPAC has been provided Determination of the Linking Length
by national institutions, in particular the institutions participat- We use the group to which J2216 + 0246 belongs as an
ing in the Gaia Multilateral Agreement. example (hereafter example group) to clarify the choice of
Facilities: LAMOST, Subaru linking length in the FoF algorithm. Figure B1 shows how the
Software: Agama (Vasiliev 2019), TOPCAT (Taylor 2005). size of the example group changes as the linking length
increases. In most cases, the group size initially grows slowly
and then jumps up at a certain linking length (e.g., 0.094 and
Appendix A 0.100 for the example group). Since the rapid increase could
LR and HR Sample Distance Distribution correspond to a merger of two or more groups, we checked if
The distance distributions of stars from the reduced L18 the group shows a multimodal distribution after the jump. We
catalog and the HR sample are shown in Figure A1. adopted the linking length before the jump, which leads to the
formation of a multimodal DTG. As shown in Figure B2, the
example group clusters well at a linking length of 0.098 but
shows three subcomponents at 0.100, indicating that the merger
at 0.100 connects multiple DTGs. Therefore, we took 0.098 as
the linking length of this DTG.

Figure A1. The normalized distance distribution fraction of stars from the
reduced L18 catalog and the HR sample. The bin size of these two samples is
1 kpc. Stars in the HR sample show a larger fraction of stars located farther
than 5 kpc; this may be caused by the selection bias from the selection strategy
(see Section 2.2 of Paper I). Figure B1. The size of the example group at different linking lengths.

23
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure B2. Distribution in clustering phase space of the example group at linking length 0.098 (orange) and 0.100 (blue).

Appendix C members based on the overdensity within the red circle in

Substructure Criteria Figure C1. Table C1 shows the criteria we adopted to associate
The distribution of eccentricity and |z|max for prograde LR DTGs with different substructures.
sample stars is shown in Figure C1. We selected the VMPD

Table C1
Criteria Adopted for Each Substructure

Substructures Criteria References

−1
GSE ec > 0.7; |vf| < 100 (km s ) (1, 2)
Thamnos η < −0.4; −1.8 < E < −1.6 (×105 km2 s−2) (1, 3)
Sequoia η < −0.15; Jf < −0.7 (×103 kpc km s−1); E > −1.5 (×105 km2 s−2) (1, 4)
Helmi streams 0.75 < Jf < 1.7 (×103 kpc km s−1); 1.6 < L⊥ < 3.2 (×103 kpc km s−1) (1, 5)
Wukong (LMS-1) 0 < Jf < 1.0 (×103 kpc km s−1); E < −1.15 (×105 km2 s−2); (Jz − Jr)/Jtot > 0.3; (6, 7)
90° < arccos (Lz/L) < 120°
Pontus −1.72 < E < −1.56 (×105 km2 s−2); −470 < Jf < 5 (kpc km s−1); 245 < Jr < 725 (kpc km s−1); (8, 9)
115 < Jz < 545 (kpc km s−1); 390 < L⊥ < 865 (kpc km s−1); 0.5 < ec < 0.8; 1 < rperi < 3 (kpc);
8 < rapo < 13 (kpc)
VMPD 0.25 < ec < 0.6; ∣z∣max < 3 (kpc); (vy - 233.1)2 + vx2 + vz2 < 180 (km s-1); vf > −200 (km s−1) (10)

References. (1) Naidu et al. (2020); (2) Belokurov et al. (2018); (3) Koppelman et al. (2019a); (4) Myeong et al. (2019); (5) Koppelman et al. (2019b); (6) Limberg
et al. (2024); (7) Yuan et al. (2020a); (8) Horta et al. (2023); (9) Malhan (2022); (10) see Section 4.1.2 and Figure C1.

24
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure C1. The distribution of eccentricity and ∣z∣max for prograde LR sample stars. We can find an overdensity within the red circle with 0.25 < ec < 0.60 and
∣z∣max < 3 (kpc) on this panel. The red dotted line is ∣z∣max = 3 (kpc) and blue dashed lines are ec = 0.25 and ec = 0.60.

Appendix D then applied the following cut for the GALAH DR3 nearby sample
Comparison with Literature Results to remove stars with unreliable measurements and exclude the
To compare our results with literature and high-resolution observations of GCs and the bulge (Buder et al. 2021, 2022):
sky survey catalogs, we collected the published chemical 1. flag_sp=0 and flag_fe_h=0,
abundances for GSE, Thamnos, Sequoia, Helmi streams, and 2. survey_name ≠ “other”,
Wukong and summarize the results in Table D1. In addition, 3. 3500 K < Teff < 6250 K.
we used the stars from the high-α sequence in Nissen &
Schuster (2010) and stars not classified as GSE in Reggiani Subsequently, we used |v − vLSR| > 180 km s−1 and [Fe/H]
et al. (2017) as background stars. For each work, we converted < −0.8 to select the halo stars, and applied the criteria
the chemical abundances to that under the solar abundances in Table C1 to select Thamnos, Sequoia, Helmi streams,
from Asplund et al. (2009). and Wukong. For GSE, we added the criterion Jr >
We also made use of the GALAH DR3 catalog (Buder et al. 30 kpc km s-1 to avoid the contamination of the heated disk
2021), which could provide us with 16 common elements to (see, e.g., Feuillet et al. 2020; Matsuno et al. 2021; Carrillo et al.
compare with our results. Following the same procedure as 2024). We also identified the prograde stars with Jr <
described in Section 2.2, we reduced the GALAH DR3 catalog 15 kpc km s-1 as in situ components (see Matsuno et al. 2021)
and computed the dynamic parameters for stars within 4 kpc. We for comparison with the accreted substructures. We note that this

Table D1
Chemical Results from Literature for Each Substructure

Substructures Literature Common Elements NStars

a,b
GSE NS10&11 Na I, Mg I, Si I, Ca I, Ti II, Cr I, Mn I, Ni I, Zn I, Y II, Ba II 16
Reggiani et al. (2017)b Na I, Mg I, Si I, Ca I, Ti II, Cr I, Mn I, Ni I, Zn I, Y II, Ba II 6
Monty et al. (2020) Na I, Mg I, Si I, Ca I, Ti II, Cr I, Ni I, Y II, Ba II 5
Aguado et al. (2021a) C, Sr II, Y II, Ba II, Eu II 4
Carrillo et al. (2022) Na I, Mg I, Si I, Ca I, Sc II, V I, Cr I, Mn I, Co I, Ni I, Zn I, 62
Y II, Zr II, Ba II, La II, Eu II
Thamnos NS10&11a,b Na I, Mg I, Si I, Ca I, Ti II, Cr I, Mn I, Ni I, Zn I, Y II, Ba II 7
Monty et al. (2020)c Na I, Mg I, Si I, Ca I, Ti II, Cr I, Ni I, Y II, Ba II 3
Sequoia Monty et al. (2020)c Na I, Mg I, Si I, Ca I, Ti II, Cr I, Ni I, Y II, Ba II 3
Aguado et al. (2021a) C, Sr II, Y II, Ba II, Eu II 5
Matsuno et al. (2022b) Na I, Mg I, Si I, Ca I, Ti II, Cr I, Mn I, Ni I, Zn I, Y II, Ba II 12
Helmi streams Roederer et al. (2010) C, Mg I, Si I, Ca I, Sc II, Ti II, V I, Cr I, Mn I, Co I, Ni I, Zn I 12
Sr II, Y II, Zr II, Ba II, La II, Eu II
Aguado et al. (2021b) C, Na I, Mg I, Si I, Ca I, Sc II, Ti II, V I, Cr I, Mn I, Co I, Ni I, Sr II, Ba II 7
Gull et al. (2021) C, Na I, Mg I, Si I, Ca I, Sc II, Ti II, Cr I, Mn I, Co I, Ni I, Zn I, 12
Sr II, Y II, Zr II, Ba II, La II, Eu II
Matsuno et al. (2022a) Na I, Mg I, Si I, Ca I, Ti II, Cr I, Mn I, Ni I, Zn I, Y II, Ba II 11
Wukong Limberg et al. (2024) C, Na I, Mg I, Si I, Ca I, Sc II, Ti II, V I, Cr I, Mn I, Co I, Ni I, Zn I 14
Sr II, Y II, Zr II, Ba II, La II, Eu II

Notes.
a
NS10&11 refers to the abundances from Nissen & Schuster (2010) and Nissen & Schuster (2011).
b
We used the homogenized abundances from Matsuno et al. (2022b) for these two samples. The GSE sample is kinematically selected by Matsuno et al. (2022b), and
the Thamnos sample comprises the low-[Fe/Mg] stars in Nissen et al. (2024), which are suggested to belong to Thamnos.
c
The “SeqG1” and “SeqG2” in Monty et al. (2020) are considered as Thamnos and Sequoia in this work, respectively.

25
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure D1. The combined chemical distributions of GSE. The colors show different substructure memberships, and the shapes show different sources of abundances.
The gray symbols are background stars, combining all stars with reliable abundances in this work (circles), high-α-sequence stars in Nissen & Schuster (2010) and
Nissen & Schuster (2011; squares), stars not classified as GSE in Reggiani et al. (2017; diamonds), and in situ components selected from GALAH DR3 (triangles with
error bars). The red symbols are GSE members. Results from this paper are presented as circles, and the other symbols represent results from Nissen & Schuster (2010)
and Nissen & Schuster (2011; squares), Reggiani et al. (2017; diamonds), Monty et al. (2020; left triangles), Aguado et al. (2021a; filled Xs), and Carrillo et al. (2022;
filled pluses). The GALAH data (triangles) are binned in metallicity, and the median values are plotted. The upper and lower error bars are the 84th and 16th
percentiles in each bin. For elements with fewer than three stars in each of these bins, we plot them directly in the corresponding panels. The meaning of the rest of the
symbols is the same as in Figure 7.

simple selection method would contaminate the substructure polar substructures (Helmi streams and Wukong) are shown in
samples, so we primarily focused on trends in the distribution of Figures D1, D2, and D3, respectively. The colors show
the GALAH sample rather than on the abundance properties of different substructure membership, and the shapes show
individual stars. different sources of abundance.
Similar to Section 4.1, we only kept stars with uncertainties In general, the chemical distribution of these substructures is
less than 0.2 dex in [Fe/H] and [X/Fe] during discussions for similar to that in the discussion in Section 4.1 and previous
corresponding element X. The combined chemical distribution studies. For example, we can confirm the moderate r-process
of GSE, retrograde substructures (Thamnos and Sequoia), and enhancement of Sequoia members in Figure D2, and the light

26
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure D2. The combined chemical distributions of Thamnos and Sequoia. The green and yellow symbols are members of Thamnos and Sequoia, respectively. The
meaning of the gray symbols and triangle symbols (GALAH data) is the same as in Figure D1. Results from this paper are presented as circles, and the Sequoia
members are divided into two parts, presented by circles with black borders (E > −1.35 × 105 km2 s−2) and with white borders (E < −1.35 × 105 km2 s−2). The
other symbols represent results from Nissen & Schuster (2010) and Nissen & Schuster (2011; squares), Monty et al. (2020; left triangles), Aguado et al. (2021a; filled
Xs), and Matsuno et al. (2022b; inverted triangles). The meaning of the rest of the symbols is the same as in Figure 9.

neutron-capture elements (e.g., Sr, Zr) deficiency of Helmi differences between the GALAH DR3 in situ sample and
stream members in Figure D3. Nissen & Schuster (2010) in situ members for several elements
However, we note that directly combining the chemical (e.g., Mg, Ba, etc.) show the effect of systematic uncertainties
abundances from the literature without homogenization would in our combined sample. Therefore, chemodynamical analyses
introduce systematic uncertainties. The systematic uncertainties based on large, uniform, high-resolution samples analyzed
can be significant and affect our interpretation of the chemical using a homogeneous method could provide more reliable
properties (e.g., see Section 4.2 in Matsuno et al. 2022a). The chemical properties of the substructures.

27
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Figure D3. The combined chemical distributions of Helmi streams and Wukong. The brown and blue symbols are members from Helmi streams and Wukong,
respectively. The meaning of the gray symbols and triangle symbols (GALAH data) is the same as in Figure D1. Results from this paper are presented as circles, and
the other symbols represent results from Roederer et al. (2010; pentagons), Aguado et al. (2021b; filled Xs), Gull et al. (2021; filled pluses), Matsuno et al. (2022a;
inverted triangles), and Limberg et al. (2024; stars). The meaning of the rest of the symbols is the same as in Figure 10.

ORCID iDs Jianrong Shi https://orcid.org/0000-0002-0349-7839

Qianfan Xing https://orcid.org/0000-0003-0663-3100
Ruizhi Zhang https://orcid.org/0009-0008-1319-1084
Jingkun Zhao https://orcid.org/0000-0003-2868-8276
Tadafumi Matsuno https://orcid.org/0000-0002-8077-4617
Haining Li https://orcid.org/0000-0002-0389-9264
Wako Aoki https://orcid.org/0000-0002-8975-6829 References
Xiang-Xiang Xue https://orcid.org/0000-0002-0642-5689
Takuma Suda https://orcid.org/0000-0002-4318-8715 Abate, C., Pols, O. R., Stancliffe, R. J., et al. 2015, A&A, 581, A62
Aguado, D. S., Belokurov, V., Myeong, G. C., et al. 2021a, ApJL, 908, L8
Gang Zhao https://orcid.org/0000-0002-8980-945X Aguado, D. S., Myeong, G. C., Belokurov, V., et al. 2021b, MNRAS, 500, 889
Yuqin Chen https://orcid.org/0000-0002-8442-901X Amorisco, N. C. 2017, MNRAS, 464, 2882
Miho N. Ishigaki https://orcid.org/0000-0003-4656-0241 An, D., & Beers, T. C. 2021, ApJ, 918, 74

28
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Aoki, W., Beers, T. C., Christlieb, N., et al. 2007, ApJ, 655, 492 Ishimaru, Y., Wanajo, S., & Prantzos, N. 2015, ApJL, 804, L35
Aoki, W., Honda, S., Beers, T. C., et al. 2005, ApJ, 632, 611 Ivans, I. I., Sneden, C., James, C. R., et al. 2003, ApJ, 592, 906
Aoki, W., Li, H., Matsuno, T., et al. 2022, ApJ, 931, 146 Jeena, S. K., Banerjee, P., Chiaki, G., & Heger, A. 2023, MNRAS, 526, 4467
Aoki, W., Matsuno, T., Honda, S., et al. 2018, PASJ, 70, 94 Ji, A. P., Frebel, A., Chiti, A., & Simon, J. D. 2016, Natur, 531, 610
Aoki, W., Tominaga, N., Beers, T. C., Honda, S., & Lee, Y. S. 2014, Sci, Johnson, J. A., & Bolte, M. 2002, ApJ, 579, 616
345, 912 Jorissen, A., Van Eck, S., Van Winckel, H., et al. 2016, A&A, 586, A158
Argast, D., Samland, M., Thielemann, F. K., & Qian, Y. Z. 2004, A&A, Kobayashi, C., Karakas, A. I., & Lugaro, M. 2020, ApJ, 900, 179
416, 997 Kobayashi, C., Umeda, H., Nomoto, K., Tominaga, N., & Ohkubo, T. 2006,
Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, ARA&A, 47, 481 ApJ, 653, 1145
Bailer-Jones, C. A. L., Rybizki, J., Fouesneau, M., Demleitner, M., & Koch-Hansen, A. J., Hansen, C. J., & McWilliam, A. 2021, A&A, 653, A2
Andrae, R. 2021, AJ, 161, 147 Koppelman, H., Helmi, A., & Veljanoski, J. 2018, ApJL, 860, L11
Barklem, P. S., Christlieb, N., Beers, T. C., et al. 2005, A&A, 439, 129 Koppelman, H. H., Helmi, A., Massari, D., Price-Whelan, A. M., &
Beers, T. C., Norris, J. E., Placco, V. M., et al. 2014, ApJ, 794, 58 Starkenburg, T. K. 2019a, A&A, 631, L9
Beers, T. C., & Sommer-Larsen, J. 1995, ApJS, 96, 175 Koppelman, H. H., Helmi, A., Massari, D., Roelenga, S., & Bastian, U. 2019b,
Bellazzini, M., Massari, D., Ceccarelli, E., et al. 2024, A&A, 683, A136 A&A, 625, A5
Belokurov, V., Erkal, D., Evans, N. W., Koposov, S. E., & Deason, A. J. 2018, Lancaster, L., Koposov, S. E., Belokurov, V., Evans, N. W., & Deason, A. J.
MNRAS, 478, 611 2019, MNRAS, 486, 378
Belokurov, V., Sanders, J. L., Fattahi, A., et al. 2020, MNRAS, 494, 3880 Li, C., & Zhao, G. 2017, ApJ, 850, 25
Belokurov, V., Zucker, D. B., Evans, N. W., et al. 2006, ApJL, 642, L137 Li, H., Aoki, W., Matsuno, T., et al. 2022, ApJ, 931, 147
Bennett, M., & Bovy, J. 2019, MNRAS, 482, 1417 Li, H., Tan, K., & Zhao, G. 2018, ApJS, 238, 16
Binney, J. 2012, MNRAS, 426, 1324 Limberg, G., Ji, A. P., Naidu, R. P., et al. 2024, MNRAS, in press
Binney, J., & Tremaine, S. 2008, Galactic Dynamics (2nd ed.; Princeton, NJ: Limberg, G., Rossi, S., Beers, T. C., et al. 2021a, ApJ, 907, 10
Princeton Univ. Press) Limberg, G., Santucci, R. M., Rossi, S., et al. 2021b, ApJL, 913, L28
Bird, S. A., Xue, X.-X., Liu, C., et al. 2021, ApJ, 919, 66 Lindegren, L., Bastian, U., Biermann, M., et al. 2021, A&A, 649, A4
Buck, T. 2020, MNRAS, 491, 5435 Liu, B., Sibony, Y., Meynet, G., & Bromm, V. 2021, MNRAS, 506, 5247
Buder, S., Lind, K., Ness, M. K., et al. 2022, MNRAS, 510, 2407 Lövdal, S. S., Ruiz-Lara, T., Koppelman, H. H., et al. 2022, A&A, 665, A57
Buder, S., Sharma, S., Kos, J., et al. 2021, MNRAS, 506, 150 Mackereth, J. T., Schiavon, R. P., Pfeffer, J., et al. 2019, MNRAS,
Bullock, J. S., Kravtsov, A. V., & Weinberg, D. H. 2001, ApJ, 548, 33 482, 3426
Cabrera Garcia, J., Beers, T. C., Huang, Y., et al. 2024, MNRAS, 527, 8973 Maeder, A., & Meynet, G. 2015, A&A, 580, A32
Caffau, E., Bonifacio, P., François, P., et al. 2013, A&A, 560, A15 Maeder, A., Meynet, G., & Chiappini, C. 2015, A&A, 576, A56
Carollo, D., Beers, T. C., Chiba, M., et al. 2010, ApJ, 712, 692 Majewski, S. R., Schiavon, R. P., Frinchaboy, P. M., et al. 2017, AJ, 154, 94
Carollo, D., Chiba, M., Ishigaki, M., et al. 2019, ApJ, 887, 22 Malhan, K. 2022, ApJL, 930, L9
Carollo, D., Christlieb, N., Tissera, P. B., & Sillero, E. 2023, ApJ, 946, 99 Mardini, M. K., Frebel, A., Ezzeddine, R., et al. 2022, MNRAS, 517, 3993
Carrillo, A., Deason, A. J., Fattahi, A., Callingham, T. M., & Grand, R. J. J. Matsuno, T., Aoki, W., & Suda, T. 2019, ApJL, 874, L35
2024, MNRAS, 527, 2165 Matsuno, T., Dodd, E., Koppelman, H. H., et al. 2022a, A&A, 665, A46
Carrillo, A., Hawkins, K., Jofré, P., et al. 2022, MNRAS, 513, 1557 Matsuno, T., Hirai, Y., Tarumi, Y., et al. 2021, A&A, 650, A110
Carter, C., Conroy, C., Zaritsky, D., et al. 2021, ApJ, 908, 208 Matsuno, T., Koppelman, H. H., Helmi, A., et al. 2022b, A&A, 661, A103
Chiba, M., & Beers, T. C. 2000, AJ, 119, 2843 McMillan, P. J. 2017, MNRAS, 465, 76
Choplin, A., Hirschi, R., Meynet, G., et al. 2018, A&A, 618, A133 McWilliam, A. 1998, AJ, 115, 1640
Cordoni, G., Da Costa, G. S., Yong, D., et al. 2021, MNRAS, 503, 2539 McWilliam, A., Preston, G. W., Sneden, C., & Searle, L. 1995, AJ, 109, 2757
da Silva, A. R., & Smiljanic, R. 2023, A&A, 677, A74 Montalbán, J., Mackereth, J. T., Miglio, A., et al. 2021, NatAs, 5, 640
De Silva, G. M., Freeman, K. C., Bland-Hawthorn, J., et al. 2015, MNRAS, Monty, S., Venn, K. A., Lane, J. M. M., Lokhorst, D., & Yong, D. 2020,
449, 2604 MNRAS, 497, 1236
Di Matteo, P., Spite, M., Haywood, M., et al. 2020, A&A, 636, A115 Morrison, H. L., Flynn, C., & Freeman, K. C. 1990, AJ, 100, 1191
Dodd, E., Callingham, T. M., Helmi, A., et al. 2023, A&A, 670, L2 Mucciarelli, A., Massari, D., Minelli, A., et al. 2021, NatAs, 5, 1247
Dodd, E., Helmi, A., & Koppelman, H. H. 2022, A&A, 659, A61 Myeong, G. C., Belokurov, V., Aguado, D. S., et al. 2022, ApJ, 938, 21
Donlon, T., & Newberg, H. J. 2023, ApJ, 944, 169 Myeong, G. C., Evans, N. W., Belokurov, V., Sanders, J. L., & Koposov, S. E.
Dovgal, A., Venn, K. A., Sestito, F., et al. 2024, MNRAS, 527, 7810 2018a, ApJL, 856, L26
Fernandes, L., Mason, A. C., Horta, D., et al. 2023, MNRAS, 519, 3611 Myeong, G. C., Evans, N. W., Belokurov, V., Sanders, J. L., & Koposov, S. E.
Feuillet, D. K., Feltzing, S., Sahlholdt, C. L., & Casagrande, L. 2020, MNRAS, 2018b, MNRAS, 478, 5449
497, 109 Myeong, G. C., Vasiliev, E., Iorio, G., Evans, N. W., & Belokurov, V. 2019,
Frebel, A., & Ji, A. P. 2023, arXiv:2302.09188 MNRAS, 488, 1235
Frischknecht, U., Hirschi, R., & Thielemann, F. K. 2012, A&A, 538, L2 Naidu, R. P., Conroy, C., Bonaca, A., et al. 2020, ApJ, 901, 48
Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2018, A&A, 616, A1 Naidu, R. P., Conroy, C., Bonaca, A., et al. 2021, ApJ, 923, 92
Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2021, A&A, 649, A1 Naidu, R. P., Ji, A. P., Conroy, C., et al. 2022, ApJL, 926, L36
Gaia Collaboration, Prusti, T., de Bruijne, J. H. J., et al. 2016, A&A, 595, A1 Newberg, H. J., Yanny, B., & Willett, B. A. 2009, ApJL, 700, L61
Gaia Collaboration, Vallenari, A., Brown, A. G. A., et al. 2023, A&A, 674, A1 Nissen, P. E., Amarsi, A. M., Skúladóttir, Á., & Schuster, W. J. 2024, A&A,
Gallart, C., Bernard, E. J., Brook, C. B., et al. 2019, NatAs, 3, 932 682, A116
Gratton, R. G., Carretta, E., Desidera, S., et al. 2003, A&A, 406, 131 Nissen, P. E., & Schuster, W. J. 2010, A&A, 511, L10
Grimmett, J. J., Karakas, A. I., Heger, A., Müller, B., & West, C. 2020, Nissen, P. E., & Schuster, W. J. 2011, A&A, 530, A15
MNRAS, 496, 4987 Noguchi, K., Aoki, W., Kawanomoto, S., et al. 2002, PASJ, 54, 855
Gull, M., Frebel, A., Hinojosa, K., et al. 2021, ApJ, 912, 52 Norris, J., Bessell, M. S., & Pickles, A. J. 1985, ApJS, 58, 463
Hattori, K., Okuno, A., & Roederer, I. U. 2023, ApJ, 946, 48 Ojima, T., Ishimaru, Y., Wanajo, S., Prantzos, N., & François, P. 2018, ApJ,
Haywood, M., Di Matteo, P., Lehnert, M. D., et al. 2018, ApJ, 863, 113 865, 87
Helmi, A. 2020, ARA&A, 58, 205 Ollongren, A. 1965, ARA&A, 3, 113
Helmi, A., Babusiaux, C., Koppelman, H. H., et al. 2018, Natur, 563, 85 Placco, V. M., Frebel, A., Beers, T. C., & Stancliffe, R. J. 2014, ApJ, 797, 21
Helmi, A., Veljanoski, J., Breddels, M. A., Tian, H., & Sales, L. V. 2017, Purcell, C. W., Bullock, J. S., & Kazantzidis, S. 2010, MNRAS, 404, 1711
A&A, 598, A58 Reggiani, H., Meléndez, J., Kobayashi, C., Karakas, A., & Placco, V. 2017,
Helmi, A., & White, S. D. M. 1999, MNRAS, 307, 495 A&A, 608, A46
Helmi, A., White, S. D. M., de Zeeuw, P. T., & Zhao, H. 1999, Natur, 402, 53 Roederer, I. U., Hattori, K., & Valluri, M. 2018, AJ, 156, 179
Hirai, Y., Ishimaru, Y., Saitoh, T. R., et al. 2015, ApJ, 814, 41 Roederer, I. U., Sneden, C., Thompson, I. B., Preston, G. W., &
Honda, S., Aoki, W., Kajino, T., et al. 2004, ApJ, 607, 474 Shectman, S. A. 2010, ApJ, 711, 573
Horta, D., Schiavon, R. P., Mackereth, J. T., et al. 2023, MNRAS, 520, 5671 Sakari, C. M., Placco, V. M., Farrell, E. M., et al. 2018, ApJ, 868, 110
Ibata, R., Malhan, K., Martin, N., et al. 2021, ApJ, 914, 123 Salvadori, S., Skúladóttir, Á., & Tolstoy, E. 2015, MNRAS, 454, 1320
Ibata, R. A., Gilmore, G., & Irwin, M. J. 1994, Natur, 370, 194 Schönrich, R., Binney, J., & Dehnen, W. 2010, MNRAS, 403, 1829
Ishigaki, M. N., Chiba, M., & Aoki, W. 2012, ApJ, 753, 64 Sestito, F., Buck, T., Starkenburg, E., et al. 2021, MNRAS, 500, 3750

29
The Astrophysical Journal, 966:174 (30pp), 2024 May 10 Zhang et al.

Sestito, F., Longeard, N., Martin, N. F., et al. 2019, MNRAS, 484, 2166 Wanajo, S., Janka, H.-T., & Müller, B. 2011, ApJL, 726, L15
Sestito, F., Martin, N. F., Starkenburg, E., et al. 2020, MNRAS, 497, L7 Wang, F., Zhang, H. W., Xue, X. X., et al. 2022, MNRAS, 513, 1958
Shank, D., Beers, T. C., Placco, V. M., et al. 2022a, ApJ, 926, 26 Wang, L., Dutton, A. A., Stinson, G. S., et al. 2015, MNRAS, 454, 83
Shank, D., Komater, D., Beers, T. C., Placco, V. M., & Huang, Y. 2022b, Wu, W., Zhao, G., Xue, X.-X., Bird, S. A., & Yang, C. 2022, ApJ, 924, 23
ApJS, 261, 19 Xing, Q.-F., Zhao, G., Liu, Z.-W., et al. 2023, Natur, 618, 712
Skúladóttir, Á., Vanni, I., Salvadori, S., & Lucchesi, R. 2024, A&A, 681, A44 Xue, X.-X., Ma, Z., Rix, H.-W., et al. 2014, ApJ, 784, 170
Siegel, D. M., Barnes, J., & Metzger, B. D. 2019, Natur, 569, 241 Yanny, B., Rockosi, C., Newberg, H. J., et al. 2009, AJ, 137, 4377
Simon, J. D. 2019, ARA&A, 57, 375 Yoon, J., Beers, T. C., Placco, V. M., et al. 2016, ApJ, 833, 20
Sneden, C., Cowan, J. J., & Gallino, R. 2008, ARA&A, 46, 241 Yoon, J., Beers, T. C., Tian, D., & Whitten, D. D. 2019, ApJ, 878, 97
Székely, G. J., & Rizzo, M. L. 2013, J. Multivar. Anal., 117, 193 Youakim, K., Starkenburg, E., Martin, N. F., et al. 2020, MNRAS,
Székely, G. J., Rizzo, M. L., & Bakirov, N. K. 2007, AnSta, 35, 2769 492, 4986
Taylor, M. B. 2005, in ASP Conf. Ser. 347, Astronomical Data Analysis Yuan, Z., Chang, J., Beers, T. C., & Huang, Y. 2020a, ApJL, 898, L37
Software and Systems XIV, ed. P. Shopbell, M. Britton, & R. Ebert (San Yuan, Z., Myeong, G. C., Beers, T. C., et al. 2020b, ApJ, 891, 39
Francisco, CA: ASP), 29 Zepeda, J., Beers, T. C., Placco, V. M., et al. 2023, ApJ, 947, 23
Tinsley, B. M. 1980, FCPh, 5, 287 Zhang, L., Xue, X.-X., Yang, C., et al. 2023, AJ, 165, 224
Tominaga, N., Umeda, H., & Nomoto, K. 2007, ApJ, 660, 516 Zhao, G., & Chen, Y. 2021, SCPMA, 64, 239562
Umeda, H., & Nomoto, K. 2002, ApJ, 565, 385 Zhao, G., Chen, Y.-Q., Shi, J.-R., et al. 2006, ChJAA, 6, 265
Vasiliev, E. 2019, MNRAS, 482, 1525 Zhao, G., Zhao, Y.-H., Chu, Y.-Q., Jing, Y.-P., & Deng, L.-C. 2012, RAA,
Venn, K. A., Irwin, M., Shetrone, M. D., et al. 2004, AJ, 128, 1177 12, 723
Venn, K. A., Kielty, C. L., Sestito, F., et al. 2020, MNRAS, 492, 3241 Zhao, J.-K., Zhao, G., Chen, Y.-Q., et al. 2015, RAA, 15, 1378
Vincenzo, F., Spitoni, E., Calura, F., et al. 2019, MNRAS, 487, L47 Zolotov, A., Willman, B., Brooks, A. M., et al. 2009, ApJ, 702, 1058

30