On the Hunt for the Origins of the Orphan--Chenab Stream: Detailed Element Abundances with APOGEE and Gaia

Stellar streams in the Galactic halo are useful probes of the assembly of galaxies like the Milky Way. Many tidal stellar streams that have been found in recent years are accompanied by a known progenitor globular cluster or dwarf galaxy. However, the Orphan--Chenab (OC) stream is one case where a relatively narrow stream of stars has been found without a known progenitor. In an effort to find the parent of the OC stream, we use astrometry from the early third data release of ESA's Gaia mission (Gaia EDR3) and radial velocity information from the SDSS-IV APOGEE survey to find up to 13 stars that are likely members of the OC stream. We use the APOGEE survey to study the chemical nature (for up to 13 stars) of the OC stream in the $\alpha$ (O, Mg, Ca, Si, Ti, S), odd-Z (Al, K, V), Fe-peak (Fe, Ni, Mn, Co, Cr) and neutron capture (Ce) elemental groups. We find that the stars that make up the OC stream are not consistent with a mono-metallic population and have a median metallicity of --1.92~dex with a dispersion of 0.28 dex. Our results also indicate that the $\alpha$-elements are depleted compared to the known Milky Way populations and that its [Mg/Al] abundance ratio is not consistent with second generation stars from globular clusters. The detailed chemical pattern of these stars indicates that the OC stream progenitor is very likely to be a dwarf spheroidal galaxy with a mass of ~10$^6$ M$_\odot$.


INTRODUCTION
In Λ-cold dark matter (ΛCDM) cosmology, it is believed that galaxies in general are assembled in a hierarchical way through the accretion of small sub-Galactic systems to construct larger ones and eventually systems over a broad range in mass and size (e.g., Searle & Zinn 1978;Davis et al. 1985;Bullock & Johnston 2005). In this context, one would expect the stellar halo of the Milky Way to be built largely from the accretion of smaller objects, as well as a smaller component of material formed within its viral radius (so called in situ material, e.g., Helmi & de Zeeuw 2000;Cooper et al. keithhawkins@uteaxs.edu Corresponding author: Keith Hawkins 2015;Hawkins et al. 2015;Bonaca et al. 2017, and references therein).
The Galactic halo contains many relic substructures in the form of gravitationally-bound clusters and dwarf galaxies and disrupted analogs in the form of tidal stellar streams from both globular clusters and dwarf galaxies alike. The modern "field of streams" (Belokurov et al. 2006;Bonaca et al. 2012) in our stellar halo demonstrates the importance of accretion processes in the build up of stellar halos. Studying the amount and properties of substructure in the Milky Way's stellar halo has provided an important way of constraining the importance of accretion processes in forming stellar halos (e.g., Naidu et al. 2020), of measuring the mass and profile of dark matter in the Galaxy (e.g. Majewski et al. 2003;Law & Majewski 2010;Koposov et al. 2010;Gibbons et al. 2014), and of constraining the small-scale properties of dark matter within galaxies (e.g., Carlberg et al. 2012;Erkal et al. 2016;Price-Whelan & Bonaca 2018;Bonaca et al. 2019;Banik et al. 2021).
In this work, we focus on the Orphan-Chenab (OC) Stream: The Orphan stream was discovered independently by two separate teams (Grillmair 2006;Belokurov et al. 2007) and the Chenab Stream was discovered photometrically as a roughly 19 • stellar feature in the Dark Energy Survey DR1 (Shipp et al. 2018), in the same direction on the sky as the Southern Galactic component of the Orphan Stream and at an estimated photometric distance of 40 kpc. The spatial overlap between the Orphan and Chenab streams was shown in the all-sky view presented by Koposov et al. (2019). Additionally, both Koposov et al. (2019) and Shipp et al. (2019) illustrated that the proper motion signals of Orphan and Chenab streams are consistent. The equivalence of the Orphan and Chenab streams across multiple kinematic parameters is now understood as strong evidence that these streams are both remnants of the tidal disruption of the same progenitor. In particular, a misalignment in the northern/southern stream poles can be explained by a dynamic encounter of the Southern portion of the OC stream with the Large Magellanic Cloud (Erkal et al. 2019).
Despite being a large structure traced over ∼ 210 • on the sky (or ∼ 150 kpc in physical length; Grillmair et al. 2015;Fardal et al. 2019;Koposov et al. 2019), the OC stream still has no known progenitor system. The OC stream is also relatively narrow, just ∼ 2 • wide on the sky, but its width is significantly broader than known streams associated with globular cluster systems (e.g., the Palomar 5 stream, Odenkirchen et al. 2003;Erkal et al. 2017;Bonaca et al. 2020). However, the stellar mass of the OC stream implies that the system that it originated in could be on the massive end of classical globular clusters (e.g. Grillmair 2006;Belokurov et al. 2007). The stream is also relatively low in surface brightness and has a distance range from ∼19-55 kpc (there is a steep distance gradient along the stream; Sesar et al. 2013;Koposov et al. 2019. Recent orbital fitting of the stream indicates that it has a Galactocentric pericenter of ∼16 kpc and an apocenter of ∼90 kpc(e.g. Newberg et al. 2010), which implies a very eccentric orbit.
There are several theories for potential progenitors to the OC stream. The first is that the OC stream originates from a globular cluster system. Koposov et al. (2019) has found that as many as 7 globular clusters are within 7 • of the stream's great circle (see the top panel of their Fig. 14). Of these globular clusters, two have been discussed in the literature as potential parents for the Orphan stream, namely NGC 2419 (discussed in Brüns & Kroupa 2011) and Ruprecht 106 (discussed in Grillmair et al. 2015;Koposov et al. 2019). However, both are thought implausible based on chemo-kinematic arguments (e.g. Casey et al. 2014;Grillmair et al. 2015). Re-cently, Li et al. (2022) noted that the globular cluster system Laevens 3 (Laevens et al. 2015), may be a potential progenitor of OC stream. The chemical exploration of the OC stream in this study will enable us to further test this idea. Another possibility is that the OC stream has originated from an ultra faint dwarf galaxy, such as Segue 1 (e.g., Gilmore et al. 2013;Casey et al. 2014) or Ursa Major II (e.g., Fellhauer et al. 2007) or Grus II (e.g., Koposov et al. 2019). Distinguishing between these theories of the long lost parent of the OC stream is, in some sense, the great challenge. The kinematic, spatial and chemical nature of the stream is important to quantify in order to better understand where it came from (e.g., Li et al. 2022;Naidu et al. 2022). Specifically, the chemical fingerprint of the OC stream, which has been relatively less explored than its kinematic signature, will offer a powerful tool in our hunt for the parent system (as it has for other systems, e.g., Aguado et al. 2021;Carrillo et al. 2022;Matsuno et al. 2022).
Therefore, in this work, we revisit the chemical properties of the Orphan-Chenab stream with the newest data release (DR17) of the Sloan Digital Sky Survey's (Gunn et al. 2006) Apache Point Observatory Galactic Evolution Experiment DR17 data (APOGEE, Nidever et al. 2015;Abdurro'uf et al. 2022;Majewski et al. 2017;Blanton et al. 2017;Wilson et al. 2019) survey in order to ascertain a possible parent population and provide chemical abundances for the largest sample of OC stars to-date. To do this, we begin in section 2, with a description of the exquisite data from the early third data release (EDR3) from the Gaia mission (Gaia Collaboration et al. 2021), which we use in order to select probable members of the OC stream (section 2.1). We also describe the spectroscopic data from the APOGEE survey, where the stellar parameter and chemical abundance information for the OC stars are sourced from (section 2.2). In section 3, we describe the chemical abundance patterns for stars in the OC stream and discuss these results in the context of probable parent populations in section 4. We end by summarizing in section 5.

DATA
We use data from the early Data Release 3 (EDR3) of the European Space Agency's Gaia mission (Gaia Collaboration et al. 2021 cross-matched to Data Release 17 (DR17) of the APOGEE survey (Abdurro'uf et al. 2022;Majewski et al. 2017;Blanton et al. 2017, using the cross-match provided by the APOGEE team and included in the DR17 data files). We use the astrometric data from Gaia to select members of the OC stream based on their kinematics, and use the element abundance measurements from the APOGEE survey to characterize the chemical properties of the stream.

Selection of Probable Orphan Stream Stars
Though candidate members of the OC stream were explicitly targeted in certain fields in the APOGEE-2 survey (ORPHAN-1 through ORPHAN-5 Zasowski et al. 2017), we search all APOGEE fields to identify candidate members with the hope of finding serendipitously-targeted stream members. We start by selecting all APOGEE sources that match a set of expected stellar parameters for red giant branch (RGB) stars in the stellar halo. Based on the known distance trend of the stream from precise RR Lyrae (RRL) distance measurements (Koposov et al. 2019), we expect that only mid-to upper-RGB stars will be luminous enough to have been observed in the H-band by APOGEE, so we first select only stars with surface gravities between −0.5 < log g < 2 (with the lower limit primarily to remove missing values of −9999). We then transform the sky positions of the remaining APOGEE sources from equatorial coordinates to the OC stream coordinate frame defined in Koposov et al. (2019) and implemented in the Gala package (Price-Whelan 2017) to utilize the astropy.coordinates transformation framework (Astropy Collaboration et al. 2018).
To define criteria for selecting stream stars in sky position, proper motion components, and radial velocity, we use the candidate RRL and RGB star members of the OC stream from Koposov et al. (2019). In detail, as a function of OC stream longitude φ 1 , we fit 5th-order polynomials to the sky position in OC stream latitude φ 2 , the proper motion in stream longitude µ φ * 1 (including the cos φ 2 term), and the proper motion in stream latitude µ φ2 using the OC RRL stars. Using these polynomial tracks, we select all APOGEE stars within ±2.5 • of the RRL track in φ 2 , within ±0.75 mas yr −1 of the RRL track in µ φ * 1 , and within ±0.5 mas yr −1 of the RRL track in µ φ2 , where these widths were chosen to span the maximum dispersion of the RRL stars in each dimension. We then cross-match the Koposov et al. (2019) RGB stream member stars to the SEGUE survey Yanny et al. (2009) and fit a 3rd-order polynomial to RGB stars with v helio > 100 km s −1 (to remove contamination; see upper right panel of Figure 1) as a function of φ 1 . We finally select APOGEE stars with velocities within ±20 km s −1 (about 5σ, using the velocity dispersion of the stream as measured in Li et al. 2022) of the RGB track in radial velocity. Figure 1 summarizes our kinematic selection of OC stream members along with our final sample of APOGEE DR17 RGB stars. The styled markers in the figure show the three different datasets used here: The RRL and RGB candidate stream members from Koposov et al. (2019) are shown as blue (square) and red (triangle) markers, and the final sample of 13 APOGEE stars that are likely members of the OC stream are shown with the black (circle) markers.

Chemical Abundances from the APOGEE Survey
The APOGEE spectroscopic survey has collected more than 700,000 moderate resolution (R = λ/∆λ ∼ 22, 500) near infrared (H-band, 1.51-1.60 µm) spectra of mostly red giant stars across the Milky Way. The primary goal of the APOGEE survey is to study the kinematic and chemical properties of stars across the Milky Way to better understand its structure and nature. In recent years, the survey has also expanded to include red giant stars deep in the stellar halo and Galactic bulge in combination with the many stars it has already observed in the Galactic disk. The survey uses both the 2.5m SDSS telescope at the Apache Point Observatory in the Northern hemisphere and the 2.5m du Pont telescope in the Southern hemisphere.
The spectra obtained by the APOGEE survey have enabled the exploration of the detailed chemical properties of the Milky Way and its substructures. The APOGEE Stellar Parameter and Abundance Pipeline (ASPCAP, García Pérez et al. 2016) is the primary tool that has been used to derive both the atmospheric parameters (T eff , log g, [Fe/H], microturblent velocity, ξ) and chemical abundances for up to 20 elements (e.g. Holtzman et al. 2015Holtzman et al. , 2018Abdurro'uf et al. 2022). These elements span the α (O, Mg, Ca, Si, Ti, S), odd-Z (Al, K V) Fe-peak (Fe, Ni, Mn, Co, Cr), and neutron capture (Ce) groups. Currently, ASPCAP also allows for derivation of heavier elements including s-process elements Nd, Ce, Rb, and Yb (e.g. Hawkins et al. 2016;Hasselquist et al. 2016;Cunha et al. 2017), however only a relatively small fraction contain measurements of these elements from the current DR17 analysis Abdurro'uf et al. 2022;Majewski et al. 2017;Blanton et al. 2017). While we provide a short description here, we refer the reader to García Pérez et al. (2016) and Jonsson et al. (in prep.), for a thorough description of ASPCAP. In order to measure the stellar properties, ASPCAP (as setup for APOGEE DR17 and for these data presented here) begins by first deriving the atmospheric parameters (T eff , log g, [Fe/H]) of the stars along with the [C/H] and [N/H] [α/Fe] values. This is done finding a best matching spectrum (via χ 2 minimization) within an interpolated grid of synthetic spectra using the FERRE code (Allende Prieto et al. 2006;Zamora et al. 2015). The ξ parameter was determined through an empirically determine relationship with log g.
Once this is complete, the ASPCAP pipeline derives stellar abundances by fitting models to the observed spectra in specific windows. Each window is weighted by its sensitivity to a given abundance and the how well the line modeled in Arcturus can be reproduced. Uncertainties in the stellar abundances are estimated using the scatter of each abundance within (assumed) chemically homogeneous open clusters. We refer the reader to the APOGEE data release papers  for more details on the exact procedures. proper motion components µ φ * 1 , µ φ 2 are all taken from cross-matches with Gaia EDR3. Radial velocity (RV) measurements come from either APOGEE or SEGUE. The dashed lines show 5th-or 3rd-order polynomial fits to the RRL (for the φ2, µ φ * 1 , and µ φ 2 panels) or RGB (for the RV panel) stars as a function of stream longitude φ1. The black circle markers show the 13 stars identified as stream members from the APOGEE DR17 catalog as described in Section 2.1.
In recent years, the OC stream was specifically targeted within the SDSS-IV APOGEE-2 survey. In order to obtain reliable stellar parameters and chemical abundances, we take the probable OC stream stars observed in APOGEE (described in sec. 2.1) and apply several quality control cuts. Following other studies of distant systems (e.g., Hasselquist et al. 2021), we begin by selecting stars that have spectra with signal-to-noise ratios (SNR) of at least 70. We also remove stars that had the STAR_BAD bit set in the AS-PCAPFLAG. These two criteria were employed to remove problematic or suspect stellar parameters and derived abundances, while attempting to keep the largest number of measurements. As noted in the APOGEE data release papers (e.g. García Pérez et al. 2016;Holtzman et al. 2015Holtzman et al. , 2018Abdurro'uf et al. 2022), this is done in order ensure that the ASPCAP pipeline converged with no major warnings (e.g., if there were known issues with the T eff , log g, χ 2 , rotation, SNR, or if the derived parameters were near a grid edge, etc., the STAR_BAD flag would be set) indicating that the reported parameters and abundances may not be reliable. This cut reduced the sample from 13 probably OC members to 10. We also require that 3500 < T eff < 5500 K, log g > 0.5 dex, but all remaining OC member stars pass this selection. We note here that an additional 4 stars would be removed if we applied the more strict quality control cut that requires ASP-CAPFLAG to equal 0 (i.e. absolutely no issues were raised in the ASPCAPFLAG pipeline). Even though the sample would be reduced, the results would remain unchanged.
In addition to the OC stream stars, for comparison, we also source the chemical abundance information of several globular clusters (specifically, we take members for the M107, M13, M71, M92, and M15 clusters from Mészáros et al. 2020) and dwarf galaxies (we take members of the Sagittarius dwarf galaxy from Hasselquist et al. 2021). These comparisons are on the same abundance scale as the OC stream stars and therefore give us a way to determine which chemical patterns are most similar. The ratio α elements with respect to iron, i.e. [O, Mg, Si, Ca, S, Ti/Fe], are widely used in Galactic archaeological studies because of their usefulness as a discriminator of the star-formation history and past enrichment of the gas from which stars formed (e.g., Nomoto et al. 2013, and references therein). Additionally the [α/Fe] abundance ratio is thought to be sensitive to the ages of stars. This is because the α elements are largely dispersed into the interstellar medium on short time scales by core-collapse Type II supernovae from massive stars. On the other hand, Fe and Fe-peak (Mn, Cr, Co, etc.) elements are dispersed by Type Ia supernovae requiring a white dwarf. These occur on longer timescales (e.g. Gilmore Hawkins et al. 2016;Jönsson et al. 2018). In this figure (as well as Fig. 4), the probable OC stream stars observed in the APOGEE survey (this study) are noted as black filled circles, while those observed in Casey et al. (2014) are shown as black open circles and those from Ji et al. (2020) are shown as black open squares. As reference, we also show several comparison samples observed and analyzed homogeneously within the APOGEE survey. These include a Milky Way sample 1 (shown as a blue-scaled background), several globular clusters (closed orange symbols), and candidate stars from the Sagittarius dwarf galaxy. The stellar clusters included are M107 (downward pointed orange triangles), M13 (upward pointed orange triangles), M71 (rightward pointed orange triangles), M92 (leftward pointed orange triangles), M15 (orange diamonds).
It is clear from Fig. 2 that the OC stream is not monometallic, as one might expect if it originates from a chemically homogeneous cluster of stars, and instead spans metallicities from −2.10 < [Fe/H] < −1.50 dex. This result is similar to Casey et al. (2014) and Ji et al. (2020), who both study up to 3 stars each in the OC stream. In addition, we find that generally the ratio of α elements, specifically O, Mg, Si, Ca, Ti, and S, with respect to iron are lower in the OC stream stars compared to homogeneously analyzed Milky Way sample. However, we note here that there is significant scatter in the [Ca/Fe] abundance of OC stream stars in the APOGEE survey, which is not seen in the other datasets (Casey et al. 2014;Ji et al. 2020). This former point indicates that the stars in the OC stream were formed in an environment similar to a dwarf galaxy and not the Milky Way. In fact, taking the stars from this study and those of Casey et al. (2014) and Ji et al. (2020) Fig. 3) above [Fe/H] −2.0 dex and a metallicity distribution the is peaked at [Fe/H] = -1.92 dex and a dispersion of 0.28 dex implies that the OC stream stars may originate in a system below the mass of Sculptor (M 1.7 × 10 7 M ) (e.g., Lianou & Cole 2013). This metallicity distribution is similar, but slightly metal-poorer, compared to earlier works (e.g. Li et al. 2022;Naidu et al. 2022). Additionally, the mean metallicity of the OC stream stars imply a progenitor mass of ∼ 8×10 5 M using the mass-metallicity relation of dwarf galaxies (Kirby et al. 2013). This value is 1 As a simple Milky Way comparison sample, we select APOGEE stars with the same quality control cuts as in Section 2.2. We make an additional criteria to select Milky Way stars, namely that the absolute Galactic latitude, |b|, of the stars had to be less than 10 • .  The Hex ratio and its importance is described in the text and in Carlin et al. (2018). In all panels, the symbols are the same as in Fig. 2 not far, but slightly lower than the predicted upper limit of 9.3 × 10 6 M from Grillmair et al. (2015) and the value of 4 × 10 6 M predicted from Koposov et al. (2019). More recently, Mendelsohn et al. (2022) estimated a total mass within 300 pc of M = 1.1 × 10 6 M , with a mass-to-light ratio of 74, for the OC stream progenitor. This mass value suggests a system much less massive than Sculptor, perhaps as low as the ultra-faint dwarf Leo IV (Strigari et al. 2008). However, we point out that while the current data implies this, we would recommend a more comprehensive study of the chemical properties of OC stream stars specifically in the metallicity range −2.5 < [Fe/H] < −1.7 dex. Additional data in the regime would help pin down the exact location of any 'knee' in the [α/Fe]-[Fe/H] diagram. We also note that while we assume here that the stream has the same metallicity of progenitor, if the OC stars were stripped from a dwarf galaxy they may be more metal-poor than the actual parent system. Carlin et al. (2018), the α elements can actually be separated into those formed during hydrostatic carbon and neon burning (Mg, O) and those formed in the explosion event of the Type II supernova (Si, Ca, Ti). Those authors propose a useful abundance ratio constructed by taking the average abundances of the hydrostatic (Mg, O) α elements relative to their explosive counterparts (HEx ratio, Carlin et al. 2018). In the bottom panel of Fig. 3, we further study the α elements by inspecting this HEx ratio ([α h/ex ]) as a function of [Mg/H] (similar to Fig. 4 of Carlin et al. 2018). We find that the Sgr stars in APOGEE from Hasselquist et al. (2021), which are a mix of stars in the main body and stream, show more scatter in [Mg/H] than the Sgr stream stars from Carlin et al. (2018), but similarly show a collective trend of HEx ratios less than zero; this trend supports a scenario in which Sgr initially had very few massive stars. The HEx ratio for the OC stream stars also decrease with [Mg/H] and show a downturn at a roughly constant value of [Mg/H] ∼ −2, which would correspond to a decrease in stars formed during hydrostatic burning at that (very low) level of Mg enrichment.

Odd-Z elements (Al, K, V)
The H-band spectra observed within the APOGEE survey allow for the measurement of several odd-Z elements including Al, K, and V. We note that there are Na lines found in the H-band spectra, however they were not well measured for each of the OC stream stars, so we choose to not report or discuss it here. While the various odd-Z elements have different production sites, they still provide useful diagnostics of the environment where the OC stream has originated from.
For example, Al, often known as a 'mild α element', is thought to be produced (similarly to the hydrostatic α elements) during carbon and neon burning in massive stars, however it is also produced in the MgAl cycle in AGB stars (e.g., Samland 1998;Doherty et al. 2014;Smiljanic et al. 2016). Upon production, Al is dispersed into the interstellar medium through both Type II supernovae and through the winds of AGB stars. Interestingly, Al seems to be a key element for distinguishing stars originating in the second generation of stars formed in globular clusters through the Mg-Al anti-correlation (e.g., Carretta et al. 2009b,a;Mészáros et al. 2015;Pancino et al. 2017 Fig. 2 and Fig. 4, respectively. However it is more useful to view the [Mg/Fe] as a function of [Al/Fe] for the purpose of looking for potential globular clusters (e.g. see Fig. 3 of Pan-cino et al. 2017). This can be found in the middle panel of Fig. 3. It clearly shows that the OC stars are not enhanced in [Al/Fe] and are, in fact, significantly depleted. Additionally the [Al/Fe] is lower than expected for the various globular clusters observed in the APOGEE survey.
K is probably produced in explosive oxygen burning in Type II supernovae. It is observationally expected to increase with decreasing metallicity, similar to α elements also dispersed in similar explosions (Hawkins et al. 2016). The [K/Fe] abundance of the OC stream stars (Fig. 4) observed in APOGEE seem to have significant scatter. This large scatter is not seen in the sample observed in Casey et al. (2014) with optical spectra.
The production site for V is poorly modeled in supernovae yields. Currently, it is thought that V is produced in explosive oxygen, silicon, and neon burning in both Type II and Type Ia supernovae (e.g. Samland 1998). Theoretically, [V/Fe] is expected to increase with decreasing metallicity. This is seen within the Milky Way sample as well as the OC stream stars. However, we find in Fig. 4 that the OC stream stars have lower [V/Fe] compared to the Milky Way or known clusters at a given metallicity .

Iron-Peak elements (Mn, Co, Cr, Ni)
The Fe-peak elements observed within the APOGEE survey include Mn, Co, Cr, and Ni. While these elements are thought to be produced and dispersed largely through Type Ia explosions, many of these Fe-peak elements are also produced in Type II supernova expositions (Iwamoto et al. 1999;Kobayashi et al. 2006;Kobayashi & Nomoto 2009;Nomoto et al. 2013, and references therein). The [Mn, Co, Cr, Ni/Fe] abundance ratios of the Fe-peak elements as a function of metallicity for the OC stream (and comparison sample) stars can be found in Fig. 4.
Mn is thought to be produced in significantly higher amounts in Type Ia supernovae compared to Fe and is therefore a good tracer of such explosions. However, unlike many of the other Fe-peak elements, it is expected (and observed) that the abundance ratio [Mn/Fe] decrease with decreasing metallicity in the Milky Way. This pattern could be due to a metallically dependence on the yields of Mn or a delay in enrichment from Type Ia supernovae (e.g. Kobayashi et al. 2006;Feltzing et al. 2007). As expected, the Milky Way (blue scale hexagonal bins in Fig. 4) the [Mn/Fe] does decrease with decreasing metallicity. We find a relatively constant sub-solar [Mn/Fe] for most OC stream stars although with significant scatter. This is in contrast to the results of Casey et al. (2014) and Ji et al. (2020) . This is observed for both the OC stream stars and the comparison samples, albeit with moderate dispersion. We also find that [Ni/Fe] for the OC stars is slight lower compared to Milky Way and SGR stars. Recent studies of Mn and Ni abundance patterns in dwarf galaxies (e.g., Sanders et al. 2021), have shown that the sub-Chandrasekhar mass systems are a significant contribution to Type Ia supernova in metal-poor, dwarf galaxy-like environments. We find abundance patters (e.g., sub-solar [Ni/Fe] and [Mn/Fe]) that are consistent with this.
Finally, the production of Co, while similar to Cr and Ni, is through both Type Ia and Type II supernovae. However unlike Cr and Ni, the amount of Co produced in the explosions is thought to depend on both mass and metallicity (Kobayashi et al. 2006) and therefore the [Co/Fe] ratio increases with decreasing metallicity for subsolar [Fe/H]. This abundance pattern is seen in the comparison samples as well the OC stream stars, which is not seen in the stars where Co could be measured in Casey et al. (2014). We also note in all of the Fe-peak elements the OC stream stars could not really be distinguished from the comparison Milky Way sample in mean or dispersion.

S-Process elements (Ce)
It was noted in Hawkins et al. (2016) that the H-band spectra obtain by APOGEE contain s-process information in addition to the α, Fe-peak, and odd-Z elements that had been known before (e.g. Holtzman et al. 2015). They identified Rb, Nd, and Yb but noted that each were difficult to measure at the resolution. Following this, Hasselquist et al. (2016Hasselquist et al. ( , 2021 identified and measured Nd in the core of the Sagittarius dwarf galaxy and Cunha et al. (2017) identified several Ce lines within the APOGEE spectra. We checked whether Ce and Nd were measured in any of the OC stream stars with the ASPCAP pipeline. We found in each case the Nd could not be measured in such low metallicity stars given the very weak line strength. Ce, on the other hand, could be measured in four stars. In the bottom right panel of Figure 4, we show the [Ce/Fe] abundance ratio as a function of [Fe/H]for OC stream stars (red circles), compared to the other Milky Way stars (blue-scaled background), globular clusters (orange symbols), and Sagittarius dwarf galaxy (magenta stars) stars. The [Ce/Fe] for values are on the lower end (all showing negative in [Ce/Fe]) compared to those found in halo field stars of similar metallicity (e.g. see Fig. 7 of Cunha et al. 2017). We note however that spectra with higher signal-to-noise for many more OC stream stars will be required to draw any conclusions.

DISCUSSION
In this section, we contextualize the chemical abundance patterns for OC stream stars and what it can tell us about what the parent population may (or may not) be. Namely, in section 4.1 we evaluate the hypothesis that the OC stream was created by a fully-disrupted globular cluster. In section 4.2, we contrast that scenario with the idea that the OC stream was created by a disrupted dwarf galaxy.

Ruling Out Globular Cluster Origin
Globular clusters are old, metal-poor relics that contain many (10 4 − 10 6 ) stars. It has been shown that several globular clusters in the Milky Way halo are being tidally disrupted by the Milky Way (e.g., the tidal tails around Palomar 5 or Palomar 13 Odenkirchen et al. 2003;Shipp et al. 2020). It is therefore possible that the OC stream could be produced by a relic globular cluster. Koposov et al. (2019) notes that 7 globular clusters are within 7 • of the stream's great circle.
There are a handful of globular clusters that have been mentioned in the literature that could be associated with the OC stream. These include several clusters that have already been discussed in the literature as possible parents for the OC stream, namely NGC 2419 (discussed in Brüns & Kroupa 2011) and Ruprecht 106 (discussed in Grillmair et al. 2015), and Laevens 3 (Li et al. 2022). In the first two cases the authors conclude that while these clusters are close by to the stream (Ruprecht 106 being the closest, Grillmair et al. 2015;Koposov et al. 2019), they are not likely to be associated with either, which we confirm here.
From the perspective of chemical abundance patterns, we are in a position to evaluate the claim that the OC stream could originate from a globular cluster. First, from section 3, we show that the OC stream is not mono-metallic and has considerable scatter in many abundances. This rules out a chemically homogeneous globular cluster formation scenario. Given that Ruprecht 106 has been shown to likely be a single population, chemically-homogeneous globular cluster (Villanova et al. 2013), we rule out this cluster as a potential parent.
However, we know that not all globular clusters are monometallic and those which are not have second generation stars which often display anti-correlations between Mg-Al and Na-O (e.g Carretta et al. 2009b,a;Mészáros et al. 2015;Pancino et al. 2017;Bastian & Lardo 2018). In these clusters it is expected [Al/Fe] is enhanced while [Mg/Fe] is depleted in the second generation stars. In Fig. 3, we clearly show that none of the OC stream stars display such chemical signatures.
Furthermore, NGC 2419, a cluster said to be a potential parent to the OC stream (Brüns & Kroupa 2011), has been  (Cohen et al. 2011;Mucciarelli et al. 2012). These authors found that NGC 2419 has stars which have simultaneously high K (with [K/Fe] > 1) and low Mg (with [Mg/Fe] < 0). Similar to Casey et al. (2014), we find that the [K/Fe] is too low to be consistent with this globular cluster. We therefore rule out this cluster as a potential parent.
Finally, Laevens 3 was proposed by Li et al. (2022) as a potential parent progenitor of the OC stream. Only one detailed chemical study of the Laevens 3 cluster exists (Longeard et al. 2019), which explore the metallicity and [α/Fe] of a total of 3 member stars. The metallicity they derive for the cluster is [Fe/H] = -1.8 ±0.10 dex and a metallicity dispersion of σ[Fe/H] < 0.50 dex. Additionally, they find a [α/Fe] = 0.00 ± 0.20 dex. The APOGEE OC stream stars have a median metallicity of -1.92 with a dispersion of ∼0.28 dex. While the metallicity is consistent with those of Laevens 3 the mean [α/Fe] of Laevens 3 is 0.00 dex, while the OC stream stars have a significantly higher mean [α/Fe] of 0.15 dex. Given this we note that Laevens 3 is not a likely parent of the OC stream. However we note that a more detailed chemical char-acterization of Laevens 3 and the OC stream would be helpful to further illustrate this.
Given that none of the globular clusters that are close to the OC stream match its phase-space tracks (Koposov et al. 2019) as well as the chemical abundance information presented here and in Casey et al. (2014), the lack of an Mg-Al or K-Mg anti-correlation, and the spread in [Fe/H], we determine that it is unlikely that the parent of OC stream is a globular cluster.

Dwarf Galaxy Origins for the Orphan Stream
In the last section, consistent with results on a smaller sample of stars from Sesar et al. (2013) and Casey et al. (2014), we conclude that the OC stream could not have originated from a disrupting globular cluster. Therefore, in this section, we evaluate the possibility that the OC stream has originated from a dwarf galaxy system (including ultra faint and classical dwarfs). In Koposov et al. (2019), it is noted that there are 3 ultra faint dwarf galaxies (Segue 1, Ursa Major II, and Grus II) and 1 classical dwarf spheroidal galaxy (Leo I) in the vicinity of the OC stream. Of these Leo I is automati-cally ruled out because it is much farther away (∼250 kpc) compared to the stream and thus they are not likely physically associated (Koposov et al. 2019). We do not discuss Leo I further for this reason.
In the recent works (e.g., Casey et al. 2014), it was discussed that the OC stream progenitor could be the Segue 1 dwarf spheroidal. With the chemical characterization in this work, we are in a position to assess this as a possibility. The chemical patterns of the Segue 1 dwarf spheroidal were studied in Frebel et al. (2014). In that work, they indicated that Segue 1 is one of the only dwarf galaxy systems where the ratio of [Mg, Si, Ca/Fe] are actually enhanced and show no evidence for a 'knee'. This is inconsistent with the abundance patterns seen in section 3.1, where we find that the [Mg, Si, Ca/Fe] abundance ratios are not only depleted relative to the rest of the Milky Way halo but they also increase with decreasing metallicity indicating there is likely a 'knee' below [Fe/H] < −2 dex. Assuming that the chemical pattern observed in the OC stream should be drawn from the chemical make-up of its progenitor, our results rule out Segue 1 a possible progenitor to the OC stream.
Ursa Major II has also been discussed as a potential parent of the OC stream (Fellhauer et al. 2007). Kirby et al. (2008) was among the first to derive chemical abundance information, though only [Fe/H]. They found in that work that Ursa Major II has a mean metallicity of [Fe/H] ∼ −2.44 dex with a spread of 0.57 dex. The first detailed chemical abundance study of Ursa Major II was done by Frebel et al. (2010); However, in that work they only obtained chemical abundances for three stars and each of those stars happened to have metallicities [Fe/H] < −2.3 dex (significantly lower than the [Fe/H] of the OC stream stars discussed in this work).
Another dwarf galaxy has been recently presented (e.g. Koposov et al. 2019;Erkal et al. 2019) as a potential parent of the OC stream, namely the Grus II system. The Grus II system is an ultra faint dwarf galaxy at distance of ∼49 kpc from the Galactic center and an estimated physical size of ∼93 pc (Drlica-Wagner et al. 2015). To date, the chemical pattern of Grus II is still poorly constrained. Hansen et al. (2020) completed the first and only detailed chemical study of three of the brightest members of Grus II, which were in the metallicity range −2.95 < [Fe/H] < −2.49 dex. If this metallicity range is representative of Grus II, it would indicate that the either Grus II has a very large metallicity spread or that the OC stream is likely too metal-rich to have originated within Grus II. Further chemical characterization of Grus II as a more metal-rich (i.e., [Fe/H] > −2.10 dex) dwarf galaxy would be a necessary step to determine if it is connected to the OC stream. With limited chemical abundance information from Grus II, Ursa Major II, and the OC stream, we recommend detailed chemical abundance studies in overlapping metallicity regimes for each in order to distinguish which could be the parent. We here that all of the dwarf galaxies listed here are likely offset from the OC stream in at least one component of its track (e.g., Koposov et al. 2019) or the chemical pattern does not match, which illustrates that the hunt for a progenitor is still ongoing. The chemical patterns presented in this work will help identify a progenitor.

SUMMARY
The Orphan-Chenab (OC) stream is an relatively-old, fairly narrow stream of stars found around the Milky Way (e.g. Grillmair 2006;Belokurov et al. 2007) with no known progenitor system. Recently, the kinematics of the OC stream has been studied and modeled using data from the second data release of the all-sky Gaia survey (Koposov et al. 2019;Erkal et al. 2019;Fardal et al. 2019). These studies have furthered our knowledge of the spatial and dynamical properties of this interesting. An interesting but still rather unexplored avenue to help find the parent of the OC stream lies in matching its detailed chemical abundance pattern to possible progenitor systems.
Only two studies currently exists on the detail abundances of at most 6 probable OC stream members (Casey et al. 2014;Ji et al. 2020). In this work, we use the updated kinematic and spatial properties of the OC stream to find 13 probable OC stream stars within the APOGEE survey. The APOGEE survey contains moderate resolution, H-band spectra for more than a few times 10 5 stars. These spectra enable the atmospheric (T eff , log g) and chemical characterization of the stars in various elemental families including the α (O, Mg, Ca, Si, Ti, S), odd-Z (Al, K, V), and Fe-peak (Fe, Ni, Mn, Co, Cr) families. Of the 14 probable OC stream stars, 5 have measured stellar parameters and chemical abundances in the APOGEE survey. This represents the largest study of the chemical abundance pattern of the OC stream and the first with a large spectroscopic survey in the infrared.
Our results can be summarized as follows: 1. The metallicity of the OC stream ranges at least from −2.10 < [Fe/H] < −1.55 dex indicating that is not mono-metallic, consistent with other findings (Casey et al. 2014). We also find that in many of the elements studied, that the OC stream is not a mono-abundance population.
2. The α elements are largely depleted compared to the Milky Way reference sample at similar metallicities ( Fig. 2 and top panel of Fig. 3). This result is particularly important because it indicates that the stars that make up the OC stream have not likely originated in the Milky Way. Instead, they have come from an environment with lower star formation rates (e.g., dwarf spheroidal or ultra faint dwarf galaxies.

The distribution of OC stars in the [α/Fe]-[Fe/H]
abundance plane seems to indicate that the OC stream has no reasonable 'knee' at [Fe/H] larger than −2.0 dex, implying an upper limit on the mass of its progenitor of M 1.4 × 10 7 M ). We emphasize, however, that many more stars with [Fe/H] < −2 dex are needed to pin down if and at what metallicity a 'knee' can be found. The metallicity distribution combined with the [α/Fe] abundances suggest the OC progenitor should be between 8 × 10 5 M 1 × 10 7 M 4. Studying the abundance patterns of Mg and Al (in Fig. 3) indicates that stars making up the OC stream stars are not consistent with originating in a known globular cluster consistent with other findings (e.g. Casey et al. 2014;Ji et al. 2020).
5. We find that the dispersion in many Fe-peak elements (Mn, Cr, Ni) is significantly larger than what is found with 3 OC stream candidate stars from Casey et al. (2014) and those of Ji et al. (2020). This discrepancy underscores the need for significantly larger samples of stars within the OC stream in order to further constrain the nature of its chemical pattern elements. Interestingly, we also find that the OC stream is offset in some Fe-peak elements (e.g., Ni, Mn) compared to the Milky Way.
We are currently working on two paths forward based on these results. Firstly, we need to further evaluate the claim the the Grus II dwarf spheroidal could be a potential parent for the OC stream. In order to properly test this, we are working to obtain high-resolution spectra of tens of stars within Grus II to determine if its metallicity and chemical abundance patterns are consistent with the OC stream. An initial chemical study of 3 stars in Grus II was carried out by Hansen et al. (2020). These authors found that Grus II is very metal-poor (i.e., −2.49 < [Fe/H] < −2.94 dex) and has a chemical pattern (elevated [Mg/Ca] ratios) that suggest it likely had a top-heavy initial mass function. It is clear that a larger study of the system focusing on candidates at higher metallicity (one that overlaps with the OC) stream is required. Further, kinematic and dynamical studies of all potential progenitors that include the impact of the Large Magellanic Cloud (e.g., Lilleengen et al. 2022, , Koposov et al., in preparation), should be done in order to confirm any potential progenitor of the OC stream. While in this work, we further advance our understanding of the chemical nature of the OC stream by increasing the sample size by a factor of 2, we still have only relatively small numbers. Therefore, we also recommend exploring the chemical pattern of 24 OC stream candidate found in the LAMOST survey (e.g. Li et al. 2017) using advanced spectral analysis techniques such as The Payne (Ting et al. 2017).