A Chemo-dynamical Link between the Gjöll Stream and NGC 3201

Stellar streams, the remnants of accreted globular clusters (GCs) and dwarf galaxies, are excellent probes of galaxy formation and cosmology. They are natural consequences of hierarchical galaxy formation, forming as their progenitors fall into and are unraveled by tidal forces of their host galaxy. Observations of Milky Way streams can be used to constrain the formation of the stellar halo (Johnston 1998; Bullock & Johnston 2005), the total mass and shape of the Galactic potential (Koposov et al. 2010; Law & Majewski 2010; Bonaca et al. 2014; Price-Whelan et al. 2014; Bovy et al. 2016; Erkal et al. 2016; Malhan & Ibata 2019), and perturbations in the gravitational field due to known satellites (Gómez et al. 2015; Erkal et al. 2018, 2019) and low-mass dark matter substructures (Ibata et al. 2002; Johnston et al. 2002; Carlberg 2009, 2012; Erkal & Belokurov 2015; Banik et al. 2019). Wide-area digital photometric surveys have dramatically increased the number of known streams around the Milky Way (e.g., Shipp et al. 2018; see Riley & Strigari 2020 for a recent compilation) and other nearby galaxies (Martínez-Delgado et al. 2010), possibly leading to a new era of statistical modeling of the stream population (Bonaca & Hogg 2018).

The second public release of Gaia data (DR2; Gaia Collaboration et al. 2018) has also contributed to this renaissance of stream data. Several studies have used Gaia DR2 proper motions to remove foreground contamination around known streams and reveal new features (Malhan et al. 2018, 2019a; Bonaca et al. 2019, 2019, 2020) or characterize the stream's dynamics (Price-Whelan et al. 2019; Shipp et al. 2019; Ibata et al. 2020). In addition, over 15 new streams have been discovered utilizing this data set, mostly by using the STREAMFINDER algorithm (Ibata et al. 2018, 2019; Malhan et al. 2018, 2019b). Many of these streams have been kinematically confirmed with independent radial velocity measurements of the stream stars (Ibata et al. 2019; Malhan et al. 2019b; Roederer & Gnedin 2019), proving that Gaia proper motions can be used to uncover low surface brightness streams that are missed in photometric studies.

Ibata et al. (2019) linked one of these new streams, Fimbulthul, to the massive GC ω Cen. They showed that the two shared similar orbital energies and angular momenta. Ibata et al. (2019) expanded on this finding, using N-body simulations to show that ω Cen's expected tidal tail overlaps with Fimbulthul and high-resolution spectroscopy of five stars in the stream to show that the radial velocities and metallicities are consistent with the stars having originated in the cluster.

GCs are known to display very uniform metallicities and prominent element abundance correlations and anticorrelations. The most well-known of these is the Na–O anticorrelations detected in most of the GCs analyzed today (e.g., Carretta et al. 2009a). This abundance signature is believed to be the result of material being processed through the CNO and Ne–Na cycles in the first generation of stars in the cluster. However, the Mg–Al cycle was likely also active in the polluters of GC stars resulting in specific trends in the Mg, Al, and possibly Si abundances of the second generation stars in GCs (Yong et al. 2015). These distinct chemical signatures makes GC stars stripped from their original cluster easily identifiable. In fact, Simpson et al. (2020) used elemental abundances from the GALAH survey (de Silva et al. 2015; Buder et al. 2018) to chemically tag two Fimbulthul stars to ω Cen, strengthening the link between the two structures. But the Simpson et al. (2020) study also rejected two kinematically selected stars based on their detailed chemical abundances, highlighting the need for a combination of kinematic and detailed chemical analysis to tie streams to their parent system.

In this work, we aim to use these techniques to chemo-dynamically link another cluster-stream pairing: NGC 3201 and the Gjöll stream. Gjöll was discovered by Ibata et al. (2019), who applied the STREAMFINDER algorithm to Gaia DR2 data in the inner Galaxy. Both the stream (Ibata et al. 2019) and the cluster (Gaia Collaboration et al. 2018) are on strongly retrograde orbits (${L}_{z}\sim 2700$ kpc km s⁻¹) with similar pericenters (∼8 kpc) and apocenters (∼30 kpc). Noting this alignment in phase space, Bianchini et al. (2019) used Gaia DR2 data to identify tidal tails coming off of the cluster. Furthermore, orbit integrations of NGC 3201 pass through both endpoints of the Gjöll stream (Riley & Strigari 2020). Taken together, this evidence suggests that NGC 3201 and Gjöll are dynamically linked; establishing a chemical link between the two would solidify this association.

This work is structured as follows. In Section 2 we present the orbit integration evidence that dynamically links NGC 3201 to Gjöll and use this orbit to select target stars for high-resolution spectroscopic follow-up. Sections 3 and 4 describe the high-resolution observation and chemical analysis of the kinematically selected candidates. Using those results, we discuss the chemical membership of the stars in Section 5 and provide a summary in Section 6.

In an effort to associate Milky Way satellite galaxies and GCs with known stellar streams, Riley & Strigari (2020) integrated orbits for each satellite to see if they passed through both endpoints for any stream. They accounted for observational errors using Monte Carlo simulations and repeated the procedure for three different Milky Way potentials (Law & Majewski 2010; Bovy 2015; Price-Whelan 2017). Regardless of which potential was assumed, NGC 3201 had a high fraction of Monte Carlo orbits pass through both endpoints of the Gjöll stream (30%–60%, while most satellite-stream pairings had zero). NGC 3201 and Gjöll are also located relatively close to each other in physical space; integrating the cluster's orbit backwards ∼20 Myr overlaps with the stream.

Given this likely association, we targeted stars that were probable members of Gjöll for spectroscopic follow-up. We selected stars from Gaia DR2 based on the following criteria relative to NGC 3201's orbit in the potential from Price-Whelan (2017):

• 1.  
  Position on the sky within 3° of the orbit trajectory (excluding a region of 2° around the cluster's present location).
• 2.  
  Measured parallax $| \pi -1/D| \lt 3{\epsilon }_{\pi }$, where D is the heliocentric distance of the orbit and _π is the quoted parallax uncertainty.
• 3.  
  Measured proper motions $({\mu }_{\alpha \cos \delta },{\mu }_{\delta })$ each within 1.5 mas yr⁻¹ of the orbit's proper motion.

We also removed stars whose astrometric fits are potentially unreliable. As detailed in Lindegren et al. (2018), we remove sources that have astrometric_excess_noise greater than 1 or renormalized unit weight error $u\gt 1.2\,\times \max (1,\exp (-0.2(G-19.5))$. We also apply the following recommended cut to remove stars with significant color excess E (phot_bp_rp_excess_factor): 1.0 + 0.015 bp_rp² < E < 1.3 + 0.06 bp_rp². Finally, we include any stars that were identified as Gjöll members by Ibata et al. (2019) in their Tables 1 and 3.

Table 1.  Known and Suspected Gjöll Stream Stars

Gaia Source ID	RA	Dec	G0	${({G}_{\mathrm{BP}}-{G}_{\mathrm{RP}})}_{0}$	π	${\mu }_{\alpha \cos \delta }$	${\mu }_{\delta }$	${v}_{\mathrm{hel}}$	I + 19
	(deg)	(deg)	(mag)	(mag)	(mas)	(mas yr−1)	(mas yr−1)	(km s−1)
3258976074166599680	63.326	1.827	14.388	0.862	0.34 ± 0.04	18.04 ± 0.06	−19.95 ± 0.04	112.10 ± 3.00
3259158764894232192	63.739	2.568	13.458	0.594	0.23 ± 0.03	19.43 ± 0.06	−20.15 ± 0.03	−150.60 ± 3.50
3254112556278356608	65.582	−0.864	14.036	0.978	0.45 ± 0.04	15.64 ± 0.06	−16.47 ± 0.03	−79.50 ± 0.90	N
3229373063616887936	69.793	−1.536	15.758	0.843	0.64 ± 0.05	19.45 ± 0.09	−19.57 ± 0.06	−33.17 ± 0.65*	N
3200439105894310272	72.085	−5.176	18.732	0.865	0.29 ± 0.24	20.91 ± 0.41	−21.38 ± 0.34	−15.18 ± 4.38*	Y
3188058536245928576	72.848	−6.759	16.407	0.650	0.31 ± 0.06	22.79 ± 0.11	−23.69 ± 0.09	0.81 ± 3.18*	N
3187390548572555136	74.627	−6.423	13.145	1.036	0.23 ± 0.02	22.12 ± 0.04	−23.09 ± 0.03	−10.30 ± 0.60	Y
2990142148280216960	78.038	−10.897	13.100	0.528	0.23 ± 0.02	23.45 ± 0.03	−24.00 ± 0.03	7.20 ± 0.60
2985543956292701312	82.104	−13.340	15.748	0.771	0.24 ± 0.05	23.96 ± 0.08	−24.21 ± 0.08	74.41 ± 1.51*	Y
3278102525607689984	56.825	7.715	17.012	0.650	0.06 ± 0.12	15.86 ± 0.26	−15.34 ± 0.19
3277467660721519616	57.617	8.032	18.703	0.839	−0.39 ± 0.35	16.00 ± 0.68	−15.85 ± 0.54
3301522634377901952	57.851	8.542	17.992	0.698	0.14 ± 0.24	16.16 ± 0.47	−15.18 ± 0.35
3301425396318555008	58.323	7.947	17.815	0.634	0.39 ± 0.22	16.72 ± 0.41	−17.77 ± 0.28
3273949498390088448	58.377	6.254	16.420	0.646	0.23 ± 0.11	15.01 ± 0.19	−15.69 ± 0.14
⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮

Note. The table is separated into stars that have spectroscopic follow-up (above line) from either this work or Ibata et al. (2019) and those that do not (below line). The final column lists whether the star is part of the more conservative sample from Ibata et al. (2019). An asterisk in the ${\text{}}{v}_{\mathrm{hel}}$ column indicates the radial velocity is from Ibata et al. (2019). The complete version of Table 1 is available online only.

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

While we do not use photometry to select target stars, we did inspect the color–magnitude diagram (CMD) of our kinematically selected stars and compared to the CMD for stars selected to be NGC 3201 members by Gaia Collaboration et al. (2018). The CMD was extinction-corrected using Schlafly & Finkbeiner (2011) corrections to the Schlegel et al. (1998) extinction maps, assuming the extinction ratios A_G/A_V = 0.85926, ${A}_{{G}_{\mathrm{BP}}}/{A}_{V}=1.06794$, and ${A}_{{G}_{\mathrm{RP}}}/{A}_{V}=0.65199$, as listed on the web interface to the PARSEC isochrones (Bressan et al. 2012). We also convert to absolute magnitudes to account for the varying heliocentric distance along the orbit, assuming that the stars have a distance that matches perfectly with the orbit distance at that Galactic longitude rather than inverting the measured parallax (see the discussion in Bailer-Jones et al. 2018).

The results of these selections are illustrated in Figure 1. Kinematically selected target stars trace the orbit of NGC 3201 over 50° of the sky, overlapping with identified Gjöll members over 195 < l < 215. We note that there are no candidate members with $| b| \lt 10$, likely due to obscuration from the Milky Way. We compile the data of targets in Table 1. We conducted a thorough search to identify additional targets for which high-resolution spectra could be obtained, but even significantly loosening the above selection criteria did not produce suitable targets.

Zoom In Zoom Out Reset image size

While it is possible that some of our candidates are chance kinematic alignments and not former members of NGC 3201, photometric data largely support the association. The vast majority of stars that are kinematically selected to match Gjöll (using NGC 3201's orbit)—with no photometric cuts—also preferentially match NGC 3201's CMD (see the bottom right panel of Figure 1). Furthermore, all six stars that were spectroscopically identified as Gjöll members in Ibata et al. (2019) fall along the same CMD (four black points, black star, and blue star in Figure 1). Repeating this procedure with another potential stream-cluster pairing identified by Riley & Strigari (2020), NGC 4590 and PS1-E, resulted in far fewer kinematic candidates that did not cluster along NGC 4590's CMD. These results suggest that many, if not most, of the candidate stars identified here are part of this disrupting system.

The kinematically selected targets brighter than G = 15 mag were observed with the Harlan J. Smith 107 in (2.7 m) telescope and the TS23 echelle spectrograph (Tull et al. 1995) at McDonald Observatory. The spectra were obtained using the 18 slit and 1 x 1 binning yielding a resolving power of $R=\lambda /{\rm{\Delta }}\lambda \sim 35,000$, with full wavelength coverage over 3600 $\lt \lambda \lt 5800$ Å and partial wavelength coverage up to 10,000 Å. The five stars, Gaia 3254112556278356608 (Gaia 32541125), Gaia 3187390548572555136 (Gaia 31873905), Gaia 2990142148280216960 (Gaia 29901421), Gaia 3259158764894232192 (Gaia 32591587) and Gaia 3258976074166599680 (Gaia 32589760), were observed over four separate runs from 2019 November to 2020 February. Table 2 lists the observations of all stars. Two of the stars, Gaia 29901421 and Gaia 32591587, are RR-Lyrae (RRL) stars, and their phase and radial velocities from individual observations are listed in Table 3. The data were reduced using standard IRAF packages, including correction for bias, flatfield, and scattered light. Signal-to-noise ratios (S/Ns) measured at 5000 Å for the final spectra are listed in Table 2. Heliocentric radial velocities for the stars were determined via cross-correlation with spectra of two radial velocity standard stars, observed on the same nights as the target stars, HD 38230 (${V}_{\mathrm{helio}}=-29.07$ (Soubiran et al. 2018)) and HD 122563 (${V}_{\mathrm{helio}}=-26.17$ (Soubiran et al. 2018)). Between 3 and 20 orders in each spectrum were used for the correlation. Table 2 lists the mean heliocentric radial velocity and standard deviation for the stars. The measured radial velocities rejected Gaia 32589760 (red star in Figure 1) as a member while confirming the kinematic membership for the other four stars.

Stellar parameter and abundances were derived using the 2017 version of MOOG (Sneden 1973), making the assumption of local thermodynamic equilibrium (LTE), and including a Rayleigh scattering treatment as described by Sobeck et al. (2011).⁷ The stellar parameters for the three stars were determined spectroscopically from equivalent width (EW) measurements of Fe i and Fe ii lines. EWs were measured by fitting Gaussian profiles to the absorption lines in the continuum-normalized spectra. Uncertainties on the EWs were computed using ${\sigma }_{\mathrm{EW}}=1.5\sqrt{\mathrm{FWHM}* \delta x}/\mathrm{SNR}$ from Cayrel (1988), where S/N is the signal-to-noise per pixel and δx is the pixel size. The effective temperatures were determined from excitation equilibrium of Fe i lines, and surface gravities ($\mathrm{log}g$) were determined from ionization equilibrium between the Fe i and Fe ii lines. Finally, microturbulent velocities (ξ) were determined by removing any trend in line abundances with reduced EW for the Fe i lines. Final stellar parameters, along with estimated uncertainties, are presented in Table 4, and lines used for the parameter determination of each star are listed in Table 5. Uncertainties on T_eff are estimated by visually inspecting the trend of abundances with excitation potential at varying temperatures. The 100 and 150 K uncertainties correspond to trends resulting in 0.2 dex differences for the high and low excitation potential lines. While estimates of the uncertainties for $\mathrm{log}g$ and ξ were determined by examining the combined effect of the standard deviation of the Fe i abundances and the uncertainty in T_eff on these. For the two warm RRL stars significantly fewer Fe i lines were measurable in the spectra resulting in a higher uncertainty on the temperatures of these.

Table 4.  Stellar Parameters

ID	Teff	$\mathrm{log}g$	ξ	[Fe/H]
	(K)	(cgs)	(km s−1)	(dex)
Gaia 32541125	5250 ± 100	3.55 ± 0.3	0.50 ± 0.3	−1.03 ± 0.17
Gaia 31873905	5060 ± 100	2.48 ± 0.3	1.50 ± 0.3	−1.34 ± 0.11
Gaia 29901421	6300 ± 150	1.50 ± 0.3	2.65 ± 0.3	−1.68 ± 0.17
Gaia 32591587	6000 ± 150	2.15 ± 0.3	2.00 ± 0.3	−1.59 ± 0.21

Download table as:  ASCIITypeset image

Table 5.  EW and Atomic Data for Fe i and Fe ii Lines Used for Parameter Determination

Stellar ID	Species	λ	χ	log gf	EW	${\sigma }_{\mathrm{EW}}$	$\mathrm{log}\epsilon$
		(Å)	(eV)		(mÅ)	(mÅ)
Gaia 32541125	Fe i	4067.97	3.209	−0.53	97.1	3.5	6.44
Gaia 32541125	Fe i	4150.24	3.428	−1.19	61.5	3.9	6.66
Gaia 32541125	Fe i	4173.92	0.989	−3.29	68.6	3.4	6.62
Gaia 32541125	Fe i	4174.91	0.914	−2.94	83.7	3.3	6.57
Gaia 32541125	Fe i	4476.07	3.686	−0.34	85.5	4.0	6.64
Gaia 32541125	Fe i	4595.35	3.299	−1.73	44.1	4.7	6.69
Gaia 32541125	Fe i	4630.12	2.277	−2.58	41.7	3.6	6.33
⋮	⋯	⋯	⋯	⋯	⋯	⋯
Gaia 32541125	Fe ii	4489.18	2.828	−2.96	51.3	3.8	6.76
Gaia 32541125	Fe ii	4520.22	2.807	−2.65	48.5	3.7	6.34
Gaia 32541125	Fe ii	4522.63	2.840	−2.29	60.7	3.9	6.36
Gaia 32541125	Fe ii	4534.16	2.856	−3.28	34.9	3.4	6.61
Gaia 32541125	Fe ii	4541.52	2.856	−2.98	44.8	4.0	6.60

Note. The complete version of Table 5 is available online only. A short version is shown here to illustrate its form and content.

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

For Gaia 32541125 and Gaia 31873905 we also determine photometric temperatures using V–K colors and the temperature scale of Alonso et al. (1999). Photometric transformations from Evans et al. (2018) was used to convert the Gaia G₀, ${({G}_{\mathrm{BP}}-{G}_{\mathrm{RP}})}_{0}$ magnitudes to V₀ and K₀ magnitudes. This results in ${T}_{\mathrm{eff},\mathrm{photo}}=5218\pm 124\,{\rm{K}}$ and 5057 ± 118 K for Gaia 32541125 and Gaia 31873905, respectively. These are in good agreement with the temperatures derived from Fe i listed in Table 4.

Abundances are derived via equivalent width and spectral synthesis analysis using 1D LTE ATLAS9 model atmospheres (Castelli & Kurucz 2003) and the solar photosphere abundances from Asplund et al. (2009). Table 6 lists the abundances derived from individual lines in each star. Line lists were generated using the linemake package,⁸ including molecular lines for CH, C₂, and CN, and hyperfine structure information. Isotopic fractions for Ba are from Gallagher et al. (2010) and from Lawler et al. (2001) for Eu. Uncertainties on the derived abundances arising from stellar parameter uncertainties were determined including covariance terms following McWilliam et al. (2013) and Johnson (2002). The covariances were computed using the following equation

$$\begin{eqnarray}&&{\sigma }_{{XY}}=\displaystyle \frac{1}{N}\displaystyle \sum _{i}^{N}({X}_{i}-\bar{X})({Y}_{i}-\bar{Y}).\end{eqnarray} \tag{ 1 }$$

${\sigma }_{T\mathrm{log}g}$, σ_Tξ, and σ_T[M/H] were determined by generating 20 model atmospheres with effective temperatures drawn from a Gaussian distribution with a mean equal to the T_eff of the star and standard deviation equal to the uncertainty on T_eff. $\mathrm{log}g$ and ξ were then varied in turn until ionization equilibrium between the Fe i and Fe ii lines was obtained for ${\sigma }_{T\mathrm{log}g}$ and no trend was visible in line abundances with reduced EW of Fe i lines for σ_Tξ. For σ_T[M/H] the direct chance in [Fe/H] was used. Similarly to determine ${\sigma }_{\mathrm{log}g\xi }$, 20 model atmospheres with microturbulences drawn from a Gaussian distribution with a mean equal to the ξ of the star and standard deviation equal to the uncertainty on ξ were computed. The gravity was then again varied to obtain ionization equilibrium between the Fe i and Fe ii lines. The final covariances resulting from this process are ${\sigma }_{T\mathrm{log}g}=16$, σ_Tξ = 6, σ_T[M/H] = 7, and ${\sigma }_{\mathrm{log}g\xi }=-0.1$ for Gaia 31873905 and ${\sigma }_{T\mathrm{log}g}=43$, σ_Tξ = 0.2, σ_T[M/H] = 14, and ${\sigma }_{\mathrm{log}g\xi }=-0.04$ for Gaia 29901421. Tables 7 and 8 list uncertainties arising from stellar parameter uncertainties for Gaia 31873905 and Gaia 29901421, respectively. These were determined by deriving abundances for each star using different atmospheric models, each with one parameter varied by its uncertainty and added in quadrature including covariance terms to provide the systematic uncertainty on [X/H], σ_sys.

Table 6.  Individual Line Abundances

				Gaia 32541125		Gaia 31873905		Gaia 29901421		Gaia 32591587
Species	λ	χ	$\mathrm{log}{gf}$	EW	$\mathrm{log}\epsilon$	EW	$\mathrm{log}\epsilon$	EW	$\mathrm{log}\epsilon$	EW	$\mathrm{log}\epsilon$
	(Å)	(eV)		(mÅ)	(dex)	(mÅ)	(dex)	(mÅ)	(dex)	(mÅ)	(dex)
CH	4300.000	⋯	⋯	synth	7.40	synth	6.84	⋯	⋯	⋯	⋯
Na i	5682.633	2.101	−0.70	⋯	⋯	27.90	5.00	⋯	⋯	⋯	⋯
Na i	5688.205	2.103	−0.45	78.43	5.52	53.96	5.10	⋯	⋯	⋯	⋯
Na i	5895.924	0.000	−0.18	⋯	⋯	⋯	⋯	⋯	⋯	171.10	5.27
Mg i	4167.271	4.343	−1.00	83.20	6.31	⋯	⋯	27.80	5.84	⋯	⋯
Mg i	4571.096	0.000	−5.69	71.82	6.38	109.20	6.63	⋯	⋯	⋯	⋯
Mg i	4702.991	4.343	−0.67	128.80	6.27	136.40	6.46	94.41	6.42	67.60	6.01

Note. The complete version of Table 6 is available online only. A short version is shown here to illustrate its form and content.

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

Table 7.  Abundance Errors Arising from Stellar Parameter Uncertainties for Gaia 31873905

Element	ΔTeff	${\rm{\Delta }}\mathrm{log}g$	Δξ	Δ[M/H]	σsys
	(dex)	(dex)	(dex)	(dex)	(dex)
CH	0.14	0.06	0.01	0.05	0.21
Na i	0.06	0.02	0.02	0.00	0.07
Mg i	0.11	0.06	0.08	0.01	0.15
Al i	0.15	0.05	0.00	0.04	0.21
Si i	0.07	0.05	0.02	0.03	0.12
K i	0.12	0.05	0.13	0.01	0.18
Ca i	0.11	0.06	0.12	0.00	0.16
Sc i	0.02	0.13	0.06	0.04	0.10
Ti ii	0.03	0.11	0.15	0.02	0.07
V ii	0.02	0.11	0.09	0.02	0.05
Cr i	0.15	0.04	0.12	0.01	0.21
Mn i	0.11	0.02	0.05	0.01	0.14
Fe i	0.12	0.04	0.12	0.01	0.18
Fe ii	0.01	0.13	0.08	0.02	0.05
Co i	0.16	0.02	0.14	0.00	0.23
Ni i	0.10	0.01	0.08	0.00	0.14
Cu i	0.11	0.01	0.01	0.00	0.12
Zn i	0.03	0.08	0.10	0.01	0.06
Sr ii	0.14	0.08	0.07	0.04	0.21
Y ii	0.03	0.10	0.10	0.01	0.06
Zr ii	0.09	0.14	0.04	0.02	0.19
Ba ii	0.04	0.11	0.11	0.01	0.08
La ii	0.13	0.11	0.02	0.02	0.21
Ce ii	0.06	0.12	0.03	0.04	0.15
Nd ii	0.08	0.11	0.00	0.00	0.17
Eu iia	0.09	0.08	0.01	0.02	0.15

Note.

^aDerived for Gaia 32541125.

Download table as:  ASCIITypeset image

Table 8.  Abundance Errors Arising from Stellar Parameter Uncertainties for Gaia 29901421

Element	ΔTeff	${\rm{\Delta }}\mathrm{log}g$	Δξ	Δ[M/H]	σsys
	(dex)	(dex)	(dex)	(dex)	(dex)
Fe i	0.11	0.02	0.05	0.01	0.14
Na i	0.08	0.02	0.01	0.00	0.10
Mg i	0.09	0.01	0.09	0.01	0.14
Al i	0.18	0.04	0.12	0.06	0.27
Si i	0.23	0.03	0.00	0.04	0.28
K i	0.11	0.02	0.02	0.00	0.13
Ca i	0.10	0.02	0.04	0.00	0.12
Sc i	0.03	0.08	0.00	0.02	0.11
Ti ii	0.06	0.11	0.09	0.00	0.17
V ii	0.06	0.08	0.04	0.02	0.14
Cr i	0.14	0.01	0.09	0.00	0.17
Fe ii	0.02	0.10	0.08	0.00	0.12
Ni i	0.10	0.02	0.01	0.00	0.12
Cu i	0.12	0.04	0.00	0.01	0.16
Zn i	0.11	0.01	0.01	0.00	0.12
Sr ii	0.12	0.13	0.20	0.01	0.28
Y ii	0.01	0.11	0.04	0.01	0.11
Zr ii	0.08	0.07	0.02	0.01	0.15
Ba ii	0.11	0.06	0.04	0.04	0.19

Download table as:  ASCIITypeset image

Abundances or upper limits have been derived for 24 elements from C to Eu in Gaia 32541125, Gaia 31873905, Gaia 29901421, and Gaia 32591587. All abundances and upper limits are presented in Tables 9 and 10, listing the ${\mathrm{log}}_{\epsilon }$ (X) abundances, the number of lines used to derived the abundance and standard deviation (σ_stat) along with [X/H] and [X/Fe] and associated uncertainties on these. For elements where only one or two lines were used to derive the abundance we use an estimated ${\sigma }_{\mathrm{stat}}=0.2$ based on the standard deviation for elements with more lines available.

Table 9.  Derived Abundances

	Gaia 31873905							Gaia 29901421
X	$\mathrm{log}\epsilon$(X)	N	σstat	[X/H]	σ[X/H]	[X/Fe]	σ[X/Fe]	$\mathrm{log}\epsilon$(X)	N	σstat	[X/H]	σ[X/H]	[X/Fe]	σ[X/Fe]
	(dex)		(dex)	(dex)	(dex)	(dex)	(dex)	(dex)		(dex)	(dex)	(dex)	(dex)	(dex)
CH	+6.84	1	0.20	−1.59	0.29	−0.25	0.32	⋯		⋯	⋯	⋯	⋯	⋯
Na i	+5.00	2	0.14	−1.24	0.16	+0.10	0.27	+4.32		0.20	−1.92	0.22	−0.24	0.27
Mg i	+6.53	4	0.09	−1.07	0.18	+0.27	0.23	+6.25	5	0.25	−1.36	0.28	+0.33	0.31
Al i	+5.18	3	0.32	−1.27	0.38	+0.07	0.40	+4.77	1	0.20	−1.68	0.34	+0.00	0.29
Si i	+6.27	1	0.20	−1.24	0.23	+0.10	0.31	+6.13	1	0.20	−1.38	0.34	+0.30	0.30
K i	+4.26	1	0.20	−0.77	0.27	+0.57	0.28	+4.03	1	0.20	−1.00	0.24	+0.68	0.26
Ca i	+5.30	16	0.19	−1.05	0.25	+0.30	0.28	+4.93	13	0.23	−1.41	0.26	+0.27	0.29
Sc ii	+1.77	3	0.04	−1.38	0.11	−0.04	0.26	+1.18	5	0.15	−1.97	0.19	−0.29	0.24
Ti ii	+3.79	32	0.21	−1.16	0.22	+0.18	0.29	+3.51	19	0.24	−1.44	0.29	+0.24	0.30
V ii	+2.76	1	0.20	−1.17	0.21	+0.17	0.31	+2.59	1	0.20	−1.34	0.24	+0.34	0.26
Cr i	+4.22	18	0.15	−1.42	0.27	−0.08	0.26	+3.96	4	0.35	−1.68	0.39	+0.00	0.39
Mn i	+3.75	3	0.04	−1.68	0.14	−0.34	0.21	⋯		⋯	⋯	⋯	⋯	⋯
Fe i	+6.16	102	0.11	−1.34	0.27	⋯	⋯	+5.82	47	0.16	−1.68	0.22	⋯	⋯
Fe ii	+6.16	19	0.07	−1.34	0.09	⋯	⋯	+5.82	18	0.11	−1.68	0.16	⋯	⋯
Co i	+3.30	1	0.20	−1.69	0.30	−0.35	0.29	⋯		⋯	⋯	⋯	⋯	⋯
Ni i	+4.81	13	0.13	−1.41	0.19	−0.07	0.24	+4.46	1	0.20	−1.76	0.23	−0.08	0.27
Cu i	+1.92	1	0.20	−2.27	0.23	−0.93	0.29	+2.89	1	0.20	−1.30	0.26	+0.38	0.27
Zn i	+3.39	2	0.20	−1.17	0.21	+0.17	0.30	+3.40	1	0.20	−1.16	0.23	+0.52	0.27
Sr ii	+1.68	2	0.21	−1.19	0.30	+0.15	0.31	+1.22	2	0.20	−1.66	0.35	+0.03	0.30
Y ii	+0.63	4	0.11	−1.58	0.13	−0.24	0.25	+0.46	1	0.20	−1.75	0.23	−0.07	0.27
Zr ii	+1.24	1	0.20	−1.34	0.27	+0.00	0.34	+0.95	1	0.20	−1.63	0.25	+0.05	0.27
Ba ii	+0.79	2	0.20	−1.39	0.21	−0.05	0.30	+0.13	2	0.20	−2.05	0.27	−0.37	0.27
La ii	−0.39	1	0.20	−1.49	0.29	−0.15	0.33	⋯		⋯	⋯	⋯	⋯	⋯
Ce ii	+0.29	2	0.20	−1.29	0.25	+0.05	0.33	⋯		⋯	⋯	⋯	⋯	⋯
Nd ii	+0.28	2	0.20	−1.14	0.26	+0.20	0.35	⋯		⋯	⋯	⋯	⋯	⋯
Eu ii	⋯		⋯	⋯	⋯	⋯	⋯	⋯		⋯	⋯	⋯	⋯	⋯

Download table as:  ASCIITypeset image

Table 10.  Derived Abundances

	Gaia 32591587							Gaia 32541125
X	$\mathrm{log}\epsilon$(X)	N	σstat	[X/H]	σ[X/H]	[X/Fe]	σ[X/Fe]	$\mathrm{log}\epsilon$(X)	N	σstat	[X/H]	σ[X/H]	[X/Fe]	σ[X/Fe]
	(dex)		(dex)	(dex)	(dex)	(dex)	(dex)	(dex)		(dex)	(dex)	(dex)	(dex)	(dex)
CH	⋯		⋯	⋯	⋯	⋯	⋯	+7.40	1	0.20	−1.03	0.28	+0.00	0.30
Na i	+5.27	1	0.20	−0.97	0.22	+0.62	0.29	+5.52	1	0.20	−0.72	0.21	+0.31	0.29
Mg i	+5.87	4	0.23	−1.73	0.27	−0.14	0.32	+6.50	5	0.28	−1.10	0.32	−0.07	0.33
Al i	+4.86	1	0.20	−1.59	0.33	+0.00	0.32	<+4.42		⋯	<−2.03	⋯	<−1.00	⋯
Si i	+5.92	1	0.20	−1.59	0.34	+0.00	0.33		3	0.20	−0.65	0.23	+0.38	0.29
K i	+2.76	1	0.20	−2.27	0.24	−0.68	0.29	+4.70	2	0.20	−0.76	0.27	+0.70	0.26
Ca i	+4.60	9	0.31	−1.74	0.33	−0.15	0.37	+5.52	15	0.25	−0.82	0.30	+0.21	0.30
Sc ii	+1.62	4	0.08	−1.53	0.14	+0.06	0.24	+2.36	3	0.17	−0.79	0.19	+0.24	0.29
Ti ii	+3.45	20	0.27	−1.50	0.32	+0.09	0.34	+4.01	23	0.23	−0.94	0.24	+0.09	0.29
V ii	⋯		⋯	⋯	⋯	⋯	⋯	⋯		⋯	⋯	⋯	⋯	⋯
Cr i	+3.99	4	0.26	−1.68	0.31	−0.09	0.34	+4.30	8	0.18	−1.34	0.27	−0.31	0.25
Mn i	<+3.84		⋯	<−1.59	⋯	<+0.00	⋯	+3.94	3	0.38	−1.49	0.40	−0.46	0.42
Fe i	+5.91	27	0.21	−1.59	0.25	⋯	⋯	+6.47	77	0.17	−1.03	0.25	⋯	⋯
Fe ii	+5.92	14	0.18	−1.58	0.21	⋯	⋯	+6.46	15	0.17	−1.03	0.28	⋯	⋯
Co i	<+3.40		⋯	<−1.59	⋯	<+0.00	⋯	+3.80	1	0.20	−1.19	0.30	−0.16	0.27
Ni i	+4.45	7	0.23	−1.77	0.26	−0.18	0.31	+5.19	8	0.20	−1.03	0.24	+0.00	0.27
Cu i	+3.10		⋯	<−1.09	⋯	<+0.50	⋯	+2.94	1	0.20	−1.25	0.23	−0.22	0.27
Zn i	+3.47		⋯	<−1.09	⋯	<+0.50	⋯	+4.09	1	0.20	−0.47	0.21	+0.56	0.28
Sr ii	+0.89	2	0.30	−1.98	0.26	−0.39	0.31	<+1.84		⋯	<−1.03	⋯	<+0.00	⋯
Y ii	+0.82		⋯	<−1.39	⋯	<+0.20	⋯	+1.63	2	0.20	−0.58	0.21	+0.45	0.28
Zr ii	⋯		⋯	⋯	⋯	⋯	⋯	+2.05	1	0.20	−0.53	0.27	+0.50	0.32
Ba ii	+0.35	2	0.20	−1.83	0.27	−0.24	0.30	+1.79	2	0.20	−0.40	0.21	+0.64	0.28
La ii	⋯		⋯	⋯	⋯	⋯	⋯	⋯		⋯	⋯	⋯	⋯	⋯
Ce ii	⋯		⋯	⋯	⋯	⋯	⋯	+1.30	2	0.21	−0.28	0.29	+0.75	0.32
Nd ii	⋯		⋯	⋯	⋯	⋯	⋯	+1.19	1	0.20	−0.23	0.26	+0.80	0.33
Eu ii	<−0.57		⋯	<−1.09	⋯	<+0.50	⋯	−0.01	1	0.20	−0.53	0.25	+0.50	0.31

Download table as:  ASCIITypeset image

The four stars display somewhat different abundances. Most of them exhibit a mild enhancement in their α-element abundances, with the exception of Gaia 32591587, which is α-poor. Also, Gaia 32541125 displays an overall enhancement in neutron-capture elements, allowing us to derive abundances for a number of neutron-capture elements in this star. We find a barium to europium ratio of [Ba/Eu] = 0.14 in this star suggesting a mixed s- and r-process origin of the neutron-capture elements in this star (e.g., Frebel 2018).

5.1. Chemical Membership

As describe in the 1, the abundances for GC stars generally show distinct patterns, distinguishing them from other stellar populations. NGC 3201 is no exception: it has, like many other clusters, been found to have a very uniform metallicity. Carretta et al. (2009a) find $[\mathrm{Fe}/{\rm{H}}]=-1.50$ with a very small intrinsic scatter of σ = 0.05 for their sample of 149 stars, while Mészáros et al. (2020) find a somewhat higher value of [Fe/H] = −1.24 with a scatter of σ = 0.10 from their sample of 179 stars. Part of this discrepancy is due to the difference in solar Fe abundances used, Carretta et al. (2009a) used A(Fe)_⊙ = 7.54 (Gratton et al. 2003), while Mészáros et al. (2020) used A(Fe)_⊙ = 7.45 (Grevesse et al. 2007). However, an offset persists, even after correction for the different A(Fe)_⊙ values, which is likely due to a systematic difference in the effective temperatures used for the two studies. Both studies use photometric temperatures but with different scales (Mészáros et al. 2020). Other recent work analyzing smaller samples of stars in NGC 3201 find mean metallicities similar to Carretta et al. (2009a), e.g., [Fe/H] =−1.53 (Muñoz et al. 2013), [Fe/H] = −1.48 (Simmerer et al. 2013) , [Fe/H] = −1.42 (Mucciarelli et al. 2015), and [Fe/H] = −1.47 (Magurno et al. 2018). In this work we find a spread in the metallicities derived for the four Gjöll stars from [Fe/H] = −1.68 to −1.03. Our parameters are purely spectroscopically derived; however, for Gaia 32541125 and Gaia 31873905 we also derived photometric effective temperatures following a similar approach as Carretta et al. (2009b). Our photometric and spectroscopic temperatures are in good agreement, suggesting our temperature scale is more similar to Carretta et al. (2009b) than Mészáros et al. (2020). Furthermore, Carretta et al. (2009a) corrected their EWs measured from the intermediate resolution GIRAFFE spectra to a system defined by the high-resolution UVES spectra to account for unrecognized blends in the GIRAFFE EWs resulting in an optical EW analysis similar to the one presented in this paper. In any case, the large spread in our metallicities suggests that not all stars were stripped from NGC 3201, and highlights the need for a multielement abundance analysis of the stars, as presented here, to access the actual connection between a stream's stars and its parent object.

One of the most prominent features of stars born in GCs is the abundance correlations and anticorrelations found for specific element pairs. In Figure 2 we plot the abundances of the four Gjöll stars along with abundances derived for stars in NGC 3201 from Carretta et al. (2009b) and Mészáros et al. (2020). We inspect the specific abundance spaces of [Al/Fe] versus [Mg/Fe], [Al/Fe] versus [Si/Fe], [Mg/Fe] versus [Si/Fe], and [Al/Fe] versus [Na/Fe], where correlations or anticorrelations are known to exist for GCs and especially for NGC 3201. We also compare the mean α abundances of the Gjöll stars with the stars in NGC 3201 in Figure 3, in this plot we have included data from the study of Magurno et al. (2018) also. For the Gjöll stars, and for the stars from Mészáros et al. (2020), [α/Fe] is calculated as $[\langle \ \mathrm{Mg},\mathrm{Si},\mathrm{Ca}\rangle /\mathrm{Fe}]$. Carretta et al. (2009b) did not derive Ca abundances for their stars, so here only Mg and Si are used for the mean α abundance, and for Magurno et al. (2018) only Mg and Ca abundances are available. All abundances from the literature have been recalculated using the Asplund et al. (2009) solar abundance scale. From these figures, it can be seen that the abundances derived for Gaia 31873905 and Gaia 29901421 (green and blue star, respectively) both display a good match to the abundances derived for the NGC 3201 stars. The abundances derived for Gaia 32541125 (black star), however, display a fairly poor match to the GC star abundances. The mismatch is driven primarily by this star's low upper limit on the Al abundance. Thus from a chemical point of view, Gaia 31873905 and Gaia 29901421 were likely stripped from NGC 3201, while Gaia 32541125 is likely an unassociated field halo star. For Gaia 32591587, shown as a yellow star in Figures 2 and 3, we find a good match to NGC 3201 in the Al abundance, but the low α-element abundances and high Na abundance found in this star is less of a perfect match to the NGC 3201 stars displayed in the figures. It should be noted that Magurno et al. (2018) also find $[\mathrm{Mg}/\mathrm{Fe}]\lt 0$ and $[\mathrm{Ca}/\mathrm{Fe}]\lt 0$ for some of the NGC 3201 RRL stars they analyze (see Figure 3), thus despite the offset in α-element abundances from the sample of giant stars analyzed by Carretta et al. (2009b) and Mészáros et al. (2020), Gaia 32591587 is still likely to have been stripped from NGC 3201.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

A subset of GCs, including ω Cen, which was tied to the Fimbulthul stream, show enhancement in elements produced in the slow neutron-capture process (s-process). This particular abundance signature revealed by high [Y/Fe] and [Ba/Fe] ratios played a central role in strengthening the case of a chemical match between ω Cen and the Fimbulthul stream. There are only a few reports of neutron-capture element abundances for stars in NGC 3201. One of these is Magurno et al. (2018), who derived Y abundances for nine stars in NGC 3201 and found an average abundance of [Y/Fe] = 0.08 suggesting that NGC 3201 is not strongly enhanced in s-process elements. In Figure 4 we plot the [Y/Fe] for the Gjöll stars along with the data from Magurno et al. (2018). Again we see that the abundances of Gaia 31873905 and Gaia 29901421 display the best match to the abundances of NGC 3201.

Zoom In Zoom Out Reset image size

We present a chemo-dynamical analysis of stars in the northern stream Gjöll, which is likely the result of dynamical interaction between the GC NGC 3201 and the Milky Way (Riley & Strigari 2020). The orbit of NGC 3201 in a Milky Way potential (Price-Whelan 2017) overlaps with both endpoints of the stream. Using this orbit, we have identified 162 stars that are kinematic members of this system for spectroscopic follow-up, five of which were bright enough to obtain high-resolution spectroscopy for.

Based on our measured radial velocities four of the five stars remain good member candidates, while Gaia 32589760 is not a member. We further present a detailed chemical analysis for the four stars with high-resolution spectra Gaia 32541125, Gaia 31873905, Gaia 29901421, and Gaia 32591587, comparing the derived abundances to those of stars in NGC 3201. The abundances of Gaia 31873905 and Gaia 29901421 match well with the previously determined abundance patterns in NGC 3201, following the cluster's expected abundance correlations and anticorrelation, and showing low [Y/Fe] abundances. For Gaia 32591587, we find a partial match between the abundances of this star and the stars in NGC 3201. Finally, a poor match is found for Gaia 32541125, which displays an abundance pattern more similar to a halo field star, notably with a low upper limit on the Al abundance and higher [α/Fe] and neutron-capture element abundances. Based on this analysis, we conclude that Gaia31873905, Gaia 29901421, and Gaia 32591587 were all stripped from NGC 3201, demonstrating both a kinematic and chemical tie between NGC 3201 and the Gjöll stream. Our results highlight the power of and need for a combination of dynamical and detailed chemical analysis when linking streams to their parent object.

Spectroscopic follow-up of the fainter stars identified here as candidate members could further confirm this association. Additionally, we note that disrupting GCs typically form symmetric tidal tails. Searching along the projected future orbit of NGC 3201, or further along the trailing orbit than the l = 180° cutoff of the original Gjöll discovery, could also reveal more members and help further characterize this system.

We thank the referee for a timely report and insightful comments on the paper. A.H.R. acknowledges support from a Texas A&M University Merit Fellowship and an NSF Graduate Research Fellowship through grant DGE-1746932. T.T.H., P.S.F., J.L.M., and L.E.S. acknowledge generous support from the George P. and Cynthia Woods Institute for Fundamental Physics and Astronomy at Texas A&M University. This research made extensive use of the SIMBAD database operated at CDS, Strasbourg, France Wenger et al. (2000), arXiv.org, and NASA's Astrophysics Data System for bibliographic information.

Facility: Smith - McDonald Observatory's 2.7m Harlan J. Smith Telescope.

Software: Astropy (Astropy Collaboration et al. 2013, 2018), ATLAS9 (Castelli & Kurucz 2003), IRAF (Tody 1986, 1993), gala (Price-Whelan 2017; Price-Whelan et al. 2017), Jupyter (Kluyver et al. 2016), linemake (https://github.com/vmplacco/linemake), Matplotlib (Hunter 2007), MOOG (Sneden 1973; Sobeck et al. 2011), NumPy (van der Walt et al. 2011), Pandas (McKinney 2010), SciPy (Jones et al. 2001), sfdmap (https://github.com/kbarbary/sfdmap).