Helmi Stream: Dwarf Galaxy Debris

• Helmi Stream is a kinematically coherent halo substructure defined by clustering in angular momentum, orbital energy, and actions, representing debris from a disrupted dwarf galaxy.
• It exhibits a bimodal vertical velocity signature and a broad metallicity range, with precise dynamical selections used to differentiate its members from other substructures.
• Distinct chemical evolution patterns, including an inverted [Mg/Fe] knee and ultra metal-poor stars like HE 0144-4657, highlight its complex star formation and nucleosynthetic history.

The Helmi Stream is a kinematically coherent halo substructure in the Milky Way, interpreted as debris from a disrupted dwarf galaxy whose stars remain clustered in integrals-of-motion space even after substantial phase mixing. It was first identified by Helmi et al. (1999) as an overdensity in angular-momentum space among nearby low-metallicity stars, and subsequent work with Gaia established it as one of the canonical accreted components of the Galactic halo, with predominantly old stars, a broad metallicity distribution, associated globular clusters, and a dynamical history distinct from major merger remnants such as Gaia-Sausage-Enceladus (Koppelman et al., 2018).

1. Discovery, definition, and phase-space identity

The defining property of the Helmi Stream is not a narrow filament on the sky but clustering in phase space. In the modern literature, membership is typically expressed through angular momentum components $(L_x,L_y,L_z)$, the perpendicular angular momentum

$L_\perp = \sqrt{L_x^2 + L_y^2},$

orbital energy $E$, and in many studies the action variables $(J_R,L_z,J_z)$ or equivalent orbital frequencies. This identification strategy follows from the fact that accreted debris can be spatially diffuse while remaining coherent in approximately conserved dynamical quantities (Placco et al., 22 Jan 2026).

A longstanding kinematic signature is the bimodality in vertical velocity $V_Z$ in the solar neighborhood. Early work emphasized two local groups, one with $V_Z>0$ and one with $V_Z<0$, interpreted as different orbital wraps of the same disrupted progenitor. Gaia DR2 then expanded the known membership dramatically, identifying 523 candidate members within 5 kpc in a broad angular-momentum selection and nearly 600 new members overall, while confirming that the stream is prograde and highly inclined rather than strongly radial (Koppelman et al., 2018).

Several papers formalized the selection in angular-momentum space. A widely used box is

$$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$

with a tighter subset

$$1750 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 2600,\qquad 1000 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1500,$$

used for higher-confidence samples (Koppelman et al., 2018). In APOGEE-based work, an equivalent selection yielded 85 Helmi candidates and placed the stream as a prograde “pillar” in the $E$–$L_\perp = \sqrt{L_x^2 + L_y^2},$0 plane, centered around intermediate binding energies and $L_\perp = \sqrt{L_x^2 + L_y^2},$1 (Horta et al., 2022).

This dynamical definition also clarifies what the Helmi Stream is not. It is distinct from Gaia-Sausage-Enceladus, which occupies $L_\perp = \sqrt{L_x^2 + L_y^2},$2; from Sagittarius, which has much higher orbital energies and different angular momenta; and from Wukong/LMS-1, whose locus in inclination and angular momentum differs from the Helmi region (Placco et al., 22 Jan 2026).

2. Progenitor system and accretion history

The Helmi Stream is consistently interpreted as the debris of a dwarf-galaxy progenitor rather than a globular cluster. Estimates for the progenitor stellar mass differ across analyses but fall in the regime of a substantial dwarf system: one line of work favors $L_\perp = \sqrt{L_x^2 + L_y^2},$3, whereas another quotes $L_\perp = \sqrt{L_x^2 + L_y^2},$4, both placing it near the classical dwarf-spheroidal regime rather than the ultra-faint regime (Koppelman et al., 2018).

Its stellar population is old. Gaia-based HR-diagram work found a broad age range of approximately $L_\perp = \sqrt{L_x^2 + L_y^2},$5–$L_\perp = \sqrt{L_x^2 + L_y^2},$6 Gyr, with a metallicity distribution extending from $L_\perp = \sqrt{L_x^2 + L_y^2},$7 to $L_\perp = \sqrt{L_x^2 + L_y^2},$8 and peaking near $L_\perp = \sqrt{L_x^2 + L_y^2},$9, already implying extended star formation in the progenitor (Koppelman et al., 2018). Asteroseismic modeling later yielded precise ages for two bright Helmi red giants: $E$0 Gyr for HD 175305 and $E$1 Gyr for HD 128279, reinforcing the conclusion that the progenitor existed at least 12 Gyr ago and formed stars over an extended interval (Lindsay et al., 1 Jul 2025).

The time of accretion has been constrained by several independent methods. $E$2-body modeling and the asymmetry between positive- and negative-$E$3 wraps favored an accretion time of roughly $E$4–$E$5 Gyr ago (Koppelman et al., 2018). CMD-based star-formation-history reconstruction from Gaia EDR3 then found that star formation in the progenitor continued until $E$6–$E$7 Gyr ago and that half of its stellar mass was assembled about $E$8 Gyr later than in the local retrograde halo, with quenching near $E$9 Gyr ago interpreted as the likely accretion epoch (Ruiz-Lara et al., 2022).

A direct dynamical estimate is now available. Using orbital-frequency clumping and the Greedy Optimistic Clustering algorithm, the stream’s dynamical age was measured as

$(J_R,L_z,J_z)$0

with the observed $(J_R,L_z,J_z)$1 asymmetry—about two-thirds of the selected stars having negative $(J_R,L_z,J_z)$2—supporting a merger substantially later than Gaia-Sausage-Enceladus (Hattori, 4 Apr 2026).

3. Stellar populations, metallicity range, and associated clusters

The metallicity distribution of Helmi members has broadened as member samples improved. High-resolution spectroscopic compilations reached $(J_R,L_z,J_z)$3, extending the previously known range by about $(J_R,L_z,J_z)$4 dex toward higher metallicity and making the system look chemically more like a classical dwarf spheroidal than an ultra-faint dwarf (Limberg et al., 2021). More recent work on individual members pushed the metal-poor extreme much lower: HE 0144-4657, dynamically associated with the stream, has $(J_R,L_z,J_z)$5, making it the lowest-metallicity star yet found in a stellar stream (Placco et al., 22 Jan 2026).

This extreme star adds an important qualification to the standard metallicity picture. The bulk of the Helmi population still occupies the old, metal-poor, dSph-like regime centered around $(J_R,L_z,J_z)$6, but the presence of HE 0144-4657 indicates that the progenitor, or one of its pre-accreted subcomponents, also sampled the ultra metal-poor tail of star formation (Placco et al., 22 Jan 2026).

The stream is also associated with a globular-cluster system. Gaia-era integrals-of-motion analyses identified seven likely associated clusters: NGC 4590, NGC 5024, NGC 5053, NGC 5272, NGC 5634, NGC 5904, and NGC 6981. These clusters lie on a tight age–metallicity relation, consistent with formation in a relatively massive progenitor dwarf (Koppelman et al., 2018). NGC 5634 remains of particular interest: a 2025 dynamical analysis found that its orbit diverges from Sagittarius over long integrations, while its orbital parameters and $(J_R,L_z,J_z)$7–$(J_R,L_z,J_z)$8 coordinates overlap the GSE/Helmi regions, leaving a Helmi association plausible but not yet definitive (Wang et al., 5 Jun 2025).

These combined stellar and cluster populations imply that the Helmi progenitor was not a chemically simple one-shot system. A plausible implication is that it was a fairly massive dwarf with its own internal chemical evolution and cluster population, and perhaps also a merger tree of smaller contributors.

4. Chemical evolution and nucleosynthetic signatures

The chemistry of the Helmi Stream is one of the main reasons it is treated as a distinct accreted building block. A high-resolution sample of 22 vetted members showed that the stream follows a declining $(J_R,L_z,J_z)$9 sequence with increasing $V_Z$0, with a plateau near $V_Z$1–$V_Z$2 at low metallicity and a clear knee by $V_Z$3, a pattern typical of dwarf spheroidals with slower star formation than the Milky Way (Limberg et al., 2021).

Later line-by-line abundance work sharpened the picture. Compared to typical halo stars, Helmi members show systematically low [X/Fe] in elements produced by massive stars, especially Na and the $V_Z$4-elements, down to at least $V_Z$5. In that homogeneous analysis, $V_Z$6 stays almost constant but low over much of the sampled metallicity range, and $V_Z$7 becomes extremely low at low metallicity, reinforcing the view that light neutron-capture elements are deficient in the stream (Matsuno et al., 2022).

APOGEE-based comparisons across many halo substructures then showed that the Helmi abundance pattern is not merely generically “dwarf-like” but statistically unusual. In that work, the stream occupies the accreted locus in the $V_Z$8–$V_Z$9 plane, is Ni-poor and Al-poor like other accreted systems, but differs from all other halo substructures considered; a particularly noted feature is an “inverted knee” in [Mg/Fe] around $V_Z>0$0, interpreted as consistent with a burst-like change in star-formation history (Horta et al., 2022).

The neutron-capture record is equally distinctive. A 22-star MIKE study of nearby bright stream stars found that every Helmi debris and trail star with measurable heavy neutron-capture elements shows a scaled-solar $V_Z>0$1-process pattern, with $V_Z>0$2. Among stars with measurable Eu, the Helmi system contains $V_Z>0$3-0, $V_Z>0$4-I, and $V_Z>0$5-II stars, and several members also show a small $V_Z>0$6-process contribution superposed on the $V_Z>0$7-process pattern, implying AGB enrichment during continued chemical evolution (Gull et al., 2021). A complementary 22-star abundance compilation found a median $V_Z>0$8 dex with small dispersion and $V_Z>0$9-process-dominated $V_Z<0$0 for all analyzed members below $V_Z<0$1 (Limberg et al., 2021).

Another chemical thread concerns rare or peculiar iron-peak enrichment. The wide binary G112-43/44, whose kinematics place it firmly in the Helmi streams, is chemically low-$V_Z<0$2 but unusually enhanced in Mn, Ni, Cu, and Zn relative to comparison low-$V_Z<0$3 halo stars, prompting discussion of enrichment by helium-shell detonation Type Ia models, though those models overpredict Ca, Ti, and Cr (Nissen et al., 2021). This suggests chemical inhomogeneity inside the progenitor rather than a single perfectly mixed enrichment history.

5. Extremely metal-poor members and nested assembly

HE 0144-4657 has become central to discussions of the Helmi Stream because it extends the stream’s chemical record into the ultra metal-poor regime. The star has $V_Z<0$4, $V_Z<0$5, $V_Z<0$6, and $V_Z<0$7, making it a CNO-enhanced ultra metal-poor star dynamically consistent with the Helmi locus in $V_Z<0$8, $V_Z<0$9, and projected action space (Placco et al., 22 Jan 2026).

Its light-element pattern suggests mono-enrichment: the diagnostic $$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$0 places it in the mono-enriched regime, and a machine-learning classifier assigns an $$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$1–$$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$2 probability that it was enriched by a single supernova. Population III faint-supernova fitting then favored a $$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$3 progenitor with low explosion energy, with 99.2% of abundance realizations selecting that solution (Placco et al., 22 Jan 2026).

The astrophysical significance of this star is not merely that it is old and metal poor. One proposed scenario is hierarchical: it may have formed in an ultra-faint dwarf galaxy that was accreted by the Helmi progenitor before the Helmi progenitor itself merged with the Milky Way. This would make the Helmi Stream a record of nested assembly—Population III enrichment, then ultra-faint dwarfs, then a dSph-like progenitor, and finally the Milky Way halo (Placco et al., 22 Jan 2026).

A related strand comes from the S2 stream, which has been argued to be the negative-$$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$4 cold wrap of the broader Helmi debris system. Its chemistry shows a metallicity spread of about 1 dex, mildly decreasing $$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$5-element trends, UFD-like Sr deficiencies, unusual $$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$6, and a significant CEMP-no component, all consistent with a primitive dwarf progenitor and stochastic early enrichment. This suggests that some chemically primitive signatures associated with the broader Helmi system may survive most clearly in specific wraps or sub-streams (Aguado et al., 2020).

6. Helmi Stream in Galactic archaeology

The Helmi Stream occupies a special place in Galactic archaeology because it is simultaneously a dynamical benchmark, a chemical laboratory, and a chronometric test case. As a dynamical tracer, it helped establish that the inner halo contains merger debris identifiable in integrals-of-motion space, even when spatially mixed. As a chemical system, it shows dSph-like $$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$7-element evolution, coherent $$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$8-process enrichment, deficiencies in light neutron-capture elements, and star-to-star diversity in Zn and iron-peak chemistry, together distinguishing it from both in-situ halo stars and other accreted substructures (Matsuno et al., 2022).

It is also methodologically instructive. In Galactic potential fitting based on action-space clustering, the Helmi Stream proved to be a weak and noisy tracer relative to cold thin streams such as GD-1 and Pal 5 because its local sample is phase mixed, its progenitor was dynamically hot, and its stars span only a limited radial range in the currently observed sample. Including Helmi in a joint fit increased the uncertainty on the enclosed mass within 20 kpc from

$$1600 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 3200,\qquad 750 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1700,$$9

to

$$1750 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 2600,\qquad 1000 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1500,$$0

highlighting the difference between using cold streams and using hot dwarf-galaxy debris as precision probes of the Galactic potential (Reino et al., 2020).

At the same time, the stream remains one of the clearest examples of how multiple lines of evidence can be integrated. Dynamical selection in $$1750 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 2600,\qquad 1000 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1500,$$1, $$1750 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 2600,\qquad 1000 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1500,$$2, actions, and orbital frequencies; CMD-derived star-formation histories; high-resolution abundance analyses; asteroseismic ages; and now direct dynamical age estimates all converge on a picture of a substantial dwarf galaxy, old but not primordial in its accretion time, whose debris still preserves both its internal chemical evolution and the chronology of its merger with the Milky Way (Lindsay et al., 1 Jul 2025).

The Helmi Stream is therefore best understood not as a single narrow stream, but as the fossil remnant of an accreted dwarf-galaxy system whose stars, globular clusters, chemical subpopulations, and dynamical substructures encode a merger roughly $$1750 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 2600,\qquad 1000 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1500,$$3–$$1750 < L_\perp/(\mathrm{kpc\,km\,s^{-1}}) < 2600,\qquad 1000 < L_z/(\mathrm{kpc\,km\,s^{-1}}) < 1500,$$4 Gyr in the past and a star-formation history that began more than 12 Gyr ago.