Sub-subgiants: Binary Evolution & Anomalies

• Sub-subgiants are stars that appear redder than main-sequence stars yet underluminous compared to subgiants and giants, challenging standard stellar evolution.
• They frequently show signatures of binary interaction, including high X-ray luminosities, short-period variability, and extensive magnetic spot coverage similar to RS CVn systems.
• Multiple formation channels such as mass transfer, magnetically inhibited convection, and stellar collisions have been proposed, with cluster environment influencing their prevalence.

Sub-subgiants are stars observed to be redder than normal main-sequence stars and fainter than normal subgiant and giant stars in an optical color-magnitude diagram. Red stragglers, which lie redward of the red giant branch and are brighter than the normal subgiant branch, may be related and are often grouped together with sub-subgiants in the literature. Because this region of the color-magnitude diagram is not easily populated by standard single-star evolutionary pathways, sub-subgiants have become important tests for binary evolution, magnetic activity, and stellar collision models (Geller et al., 2017).

1. Definition, classification, and color-magnitude-diagram placement

In cluster work, sub-subgiants are defined empirically relative to cluster isochrones. In the formulation used by Geller et al., sub-subgiants occupy the region redward of the normal main sequence and fainter than the subgiant and giant branches, while red stragglers are redder than the normal red-giant branch but brighter than the normal subgiant branch. The boundaries are defined relative to PARSEC isochrones computed using each cluster’s age, $(m-M)_V$, $E(B-V)$, and $[\mathrm{Fe/H}]$, and the normal binary sequence is excluded from these regions (Geller et al., 2017).

The separation between the sub-subgiant and red-straggler domains is set by the magnitude at the base of the giant branch: objects redward of the isochrone and fainter than the giant-branch base are classified as sub-subgiants, whereas those redward and brighter than subgiants are classified as red stragglers. Because many of these stars are spotty and variable, some objects move between the sub-subgiant and red-straggler domains depending on filter choice; classification is therefore based on appearing in the sub-subgiant region in at least one optical color-magnitude combination (Geller et al., 2017).

A related photometric definition has also been applied to field giants. In the APOGEE-based study of rapidly rotating red giants, a very conservative criterion was adopted: any star lying to the red of a 14 Gyr, $[\mathrm{Fe/H}] = +0.5$ MIST isochrone in both infrared and optical color-magnitude diagrams was classified as a sub-subgiant, again excluding points that could be consistent with the binary sequence. That selection yielded 38 field sub-subgiants (Dixon et al., 7 Apr 2025).

The color-magnitude-diagram anomaly is central to the astrophysical significance of the class. Standard single-star evolution proceeds from the main sequence to the subgiant branch and red giant branch at higher luminosities for redder colors; being simultaneously redder than the main sequence and underluminous relative to the subgiant or red-giant loci implies an atypical structure. Proposed explanations in the cited literature include mass transfer, envelope stripping, stellar collisions, and reduced luminosity due to inhibited convection from strong magnetic fields and large spot coverage (Milliman et al., 2016, Gosnell et al., 2021).

2. Cluster demographics and membership

The first membership-vetted demographic synthesis compiled 65 sub-subgiants and red stragglers in 16 open and globular clusters from the literature: 56 sub-subgiants, 8 red stragglers, and 1 bright, flickering low-mass X-ray binary with uncertain color. Of these, 50 pass strict membership criteria, including 43 sub-subgiants and 7 red stragglers (Geller et al., 2017).

The clusters represented in that compilation comprise six open clusters—NGC 188, NGC 2158, NGC 2682/M67, NGC 6791, NGC 6819, and NGC 7142—and ten globular clusters—47 Tuc/NGC 104, $\omega$ Cen/NGC 5139, M4/NGC 6121, NGC 6218, NGC 6366, NGC 6397, NGC 6652, NGC 6752, M55/NGC 6809, and M71/NGC 6838 (Geller et al., 2017).

Strict membership selection combines kinematics, photometric position, and activity diagnostics. Stars with proper-motion or radial-velocity membership probabilities below 50%, or velocities more than $3\sigma$ from the cluster mean, are treated as non-members; objects marked with uncertain membership are excluded from the demographic analyses. Field contamination in the sub-subgiant region is estimated using the Besançon Galaxy model, together with estimates for short-period binaries and X-ray active binaries derived from binary period statistics and active-binary catalogs. The cumulative Poisson probability of observing at least $N_o$ stars when $N_e$ are expected is written as

$$P = 1 - e^{-N_e}\sum_{x=0}^{N_o-1}\frac{N_e^x}{x!}.$$

An object-level field-interloper probability is then formed as

$$P_\mathrm{field} = (1 - P_\mathrm{PM})(1 - P_\mathrm{RV}) \times \begin{cases} P_V(N_{oV}=1)P_X(N_{oX}=1), & \text{if var./X-ray},\ P(N_o=1), & \text{otherwise}. \end{cases}$$

Objects with $E(B-V)$0 are excluded from the strict member set (Geller et al., 2017).

The radial distribution supports the membership assignments. Ninety-three percent of strict-member sub-subgiants lie within 3.3 core radii, and approximately 60% lie within one core radius. A Kolmogorov-Smirnov test strongly rejects a uniform-in-surface-density field distribution, with $E(B-V)$1 (Geller et al., 2017).

A cluster-specific example is provided by NGC 6791, an old, metal-rich open cluster with age $E(B-V)$2 Gyr and $E(B-V)$3. Five candidate sub-subgiants were examined with WIYN/Hydra spectroscopy, and four were confirmed as likely cluster members from three-dimensional kinematic information. Among 58,901 stars, only 147 lie in the sub-subgiant color-magnitude-diagram area, and the expected number of masquerading three-dimensional members is $E(B-V)$4; the corresponding Poisson probabilities for misidentifying 1–4 field stars as members are 17%, 1.8%, 0.13%, and 0.007%, respectively (Milliman et al., 2016).

3. Activity phenomenology: X-rays, H$E(B-V)$5, variability, and binarity

The defining observational pattern of sub-subgiants is not confined to color-magnitude-diagram position. In the strict-member cluster sample, at least 58% of sub-subgiants, 25 of 43, are detected in X-rays, with typical $E(B-V)$6–$E(B-V)$7 keV luminosities of order $E(B-V)$8–$E(B-V)$9 erg s$[\mathrm{Fe/H}]$0. Their optical-to-X-ray flux ratios place them in the same locus as RS CVn-type active binaries and generally separate them from BY Dra active binaries, cataclysmic variables, and quiescent low-mass X-ray binaries (Geller et al., 2017).

The X-ray luminosity is computed from the absorption-corrected flux through

$[\mathrm{Fe/H}]$1

with $[\mathrm{Fe/H}]$2 corrected using line-of-sight $[\mathrm{Fe/H}]$3 and an assumed spectrum. For optical comparison, de-reddened fluxes are derived from magnitudes and bandpasses; in $[\mathrm{Fe/H}]$4 band,

$[\mathrm{Fe/H}]$5

with $[\mathrm{Fe/H}]$6 and $[\mathrm{Fe/H}]$7 (Geller et al., 2017).

Variability is comparably common. At least 65% of the strict-member sub-subgiants, 28 of 43, are photometric and/or radial-velocity variables, and among the 21 with measured photometric or radial-velocity periods, about 90% have $[\mathrm{Fe/H}]$8 d. The overall radial-velocity-binary fraction among strict sub-subgiants is 21 of 43, or $[\mathrm{Fe/H}]$9, and in open clusters with comprehensive radial-velocity monitoring—NGC 188, M67, NGC 6791, and NGC 6819—8 of 11 sub-subgiants are spectroscopic binaries, corresponding to $[\mathrm{Fe/H}] = +0.5$0 (Geller et al., 2017).

Orbital properties are also distinctive. Most open-cluster sub-subgiants with orbital solutions have short periods, with mean $[\mathrm{Fe/H}] = +0.5$1 d and dispersion $[\mathrm{Fe/H}] = +0.5$2 d, and nearly circular orbits. The longest-period solution in that set is M67 S1063, with $[\mathrm{Fe/H}] = +0.5$3 d and $[\mathrm{Fe/H}] = +0.5$4; most others have $[\mathrm{Fe/H}] = +0.5$5 (Geller et al., 2017). In NGC 6791, the three confirmed binaries are all single-lined spectroscopic binaries with nearly circular, short-period orbits: WOCS 130013 has $[\mathrm{Fe/H}] = +0.5$6 d and $[\mathrm{Fe/H}] = +0.5$7; WOCS 147014 has $[\mathrm{Fe/H}] = +0.5$8 d and $[\mathrm{Fe/H}] = +0.5$9; WOCS 170008 has $\omega$0 d and $\omega$1 (Milliman et al., 2016).

H$\omega$2 emission is also frequent. At least 33% of the strict-member sub-subgiants, 14 of 43, are H$\omega$3 emitters, and MUSE spectroscopy reports H$\omega$4 emission in multiple systems, including NGC 6397 U12 and U18 and several 47 Tuc sources (Geller et al., 2017). In NGC 6791, co-added Hydra spectra showed the strongest H$\omega$5 emission in WOCS 130013, with equivalent width $\omega$6 \AA, weak emission in WOCS 147014, mostly filled-in absorption in WOCS 170008, and ordinary absorption in WOCS 131020; the authors interpret the ensemble, together with the soft X-ray detections, as chromospheric and coronal activity rather than ongoing accretion (Milliman et al., 2016).

These demographics converge on the same empirical profile: short-period variability, frequent radial-velocity binarity, moderate X-ray luminosities, and chromospheric line emission. This suggests that a large fraction of sub-subgiants are chromospherically active, tidally synchronized binaries analogous to RS CVn systems (Geller et al., 2017).

4. Magnetic activity, rotation, and the RS CVn connection

Magnetic activity is increasingly treated not merely as a correlated property but as part of the physical explanation for at least some sub-subgiants. The M67 prototype S1063 provides the most detailed case study. S1063 is an SB1 with orbital period $\omega$7 d and eccentricity $\omega$8, is X-ray bright with $\omega$9 erg s$3\sigma$0, and shows a rotation period $3\sigma$1 d from K2 photometry (Gosnell et al., 2021).

A two-temperature spectral decomposition of high-resolution IGRINS H-band spectra, combined with K2 and ASAS-SN light curves, yields a projected spot filling factor of $3\sigma$2 and a spot temperature of $3\sigma$3 K, with the spot filling factor varying from approximately 20% to 45% over four years and averaging about 30%. The analysis models the disk-integrated spectrum as

$3\sigma$4

and the surface-averaged effective temperature as

$3\sigma$5

At the IGRINS epoch, the inferred surface-averaged effective temperature is $3\sigma$6 K (Gosnell et al., 2021).

The spot properties of S1063 are reported to be similar to those found in RS CVn systems. The cited study concludes that S1063, and likely other sub-subgiants, are magnetically active spotted stars, and that the derived spot fractions and temperatures support classifying sub-subgiants as another type of active giant-star binary system (Gosnell et al., 2021).

The 2025 APOGEE-GALEX-Gaia analysis extends the activity picture beyond clusters. Starting from 7,286 cool giants with APOGEE-derived $3\sigma$7 and GALEX NUV, the authors define a metallicity-corrected ultraviolet activity proxy

$3\sigma$8

where

$3\sigma$9

$N_o$0

and

$N_o$1

The preferred empirical rotation-activity relation is

$N_o$2

with $N_o$3 in km s$N_o$4 (Dixon et al., 7 Apr 2025).

Within that field sample, sub-subgiants are identified as the reddest, most active short-period binaries. None belong to the “single” category by the paper’s radial-velocity-variability criterion; roughly half overlap the “synchronized giants” category and most are radial-velocity variable. They cluster at $N_o$5 d with $N_o$6, strongly following the trend of synchronization, and 35 of 38 lie above the fitted rotation-activity relation. Twelve of the 38 are cataloged as RS CVn variables (Dixon et al., 7 Apr 2025).

The combined implication of the cluster and field studies is that sub-subgiants are not merely underluminous red anomalies. A plausible implication is that a substantial subset are “overactive RS CVn” systems: short-period synchronized giant binaries with unusually strong magnetic activity, large spot coverage, and color-magnitude-diagram displacements produced at least in part by magnetically modified stellar structure (Dixon et al., 7 Apr 2025, Gosnell et al., 2021).

5. Formation channels and evolutionary pathways

The formation-frequency synthesis identifies several viable channels, grouped into binary-evolution pathways, rapid stripping channels, and stellar-collision channels. The central result is that binary-evolution pathways are the most prevalent, while all channels appear viable routes to sub-subgiant creation, especially in higher-mass globular clusters (Geller et al., 2017).

One proposed channel is ongoing subgiant mass transfer, denoted SG MT. In this scenario, a main-sequence turnoff star in a sufficiently short-period binary overflows its Roche lobe shortly after leaving the main sequence; mass transfer reduces the donor’s envelope mass and luminosity and moves the system into the sub-subgiant region. The Roche-lobe condition is written using Eggleton’s approximation, and the formation timescale is parameterized as

$N_o$7

with $N_o$8 adopted for stable mass transfer under the assumed mass-ratio criterion (Geller et al., 2017).

A second binary-mediated pathway is SG Mag, in which magnetically inhibited convection on a tidally locked subgiant lowers $N_o$9, increases radius, and reduces luminosity at a given evolutionary stage. In the MESA calculations summarized by Geller et al., reduced mixing length produces tracks that enter the sub-subgiant region primarily after the subgiant downturn and on the lower red giant branch. The corresponding timescale is written as

$N_e$0

with $N_e$1 adopted as the empirical fraction of short-period subgiant binaries that appear as sub-subgiants in the best-observed open clusters (Geller et al., 2017).

Rapid envelope stripping of a subgiant is treated separately. In SG CE, common-envelope ejection removes the subgiant’s envelope; in SG Coll, a grazing dynamical encounter strips envelope material without merging. The paper also considers MS Coll, in which a main-sequence collision product contracts back toward thermal equilibrium and traverses the sub-subgiant region on a thermal timescale. These collision channels are viable but generally less consistent with the observed predominance of short-period spectroscopic binaries, because short-period binarity is disfavored immediately after typical collision outcomes (Geller et al., 2017).

Analytic formation-rate calculations use the cumulative Poisson probability

$N_e$2

where $N_e$3 is the duration that the product resides in the sub-subgiant region and $N_e$4 is the channel formation timescale. Across the observed cluster sample, the analytic aggregation predicts that about 67% of sub-subgiants arise from SG Mag, with the remainder split among SG MT and stripping/collision channels (Geller et al., 2017).

Monte Carlo globular-cluster models support the same hierarchy. In 327 models spanning the parameter space of Galactic globular clusters, more than 1100 SG MT sub-subgiants and more than 12000 additional sub-subgiants, primarily SG Mag, were identified. After correcting for the model’s flat-in-log period distribution, the Monte Carlo predictions align well with the analytic Poisson upper limits, with the largest discrepancy for the main-sequence collision channel (Geller et al., 2017).

The formation picture is therefore explicitly plural. Multiple channels likely operate simultaneously, and some of them connect sub-subgiants to red stragglers and blue stragglers as different stages along the same evolutionary sequence. SG MT and MS Coll can produce blue stragglers after the sub-subgiant or red-straggler phases, while SG Strip can move a star from the sub-subgiant region into the red-straggler region as the stripped star re-equilibrates (Geller et al., 2017).

6. Environmental scaling, cluster dependence, and field analogs

The number of sub-subgiants per unit cluster mass increases toward lower-mass clusters. The specific frequency is defined as

$N_e$5

where $N_e$6 counts strict-member sub-subgiants within the minimum completeness radius and $N_e$7 is the cluster mass. When the two most massive clusters, $N_e$8 Cen and 47 Tuc, are excluded, no significant correlation is found between the absolute number of sub-subgiants and cluster mass; Pearson’s $N_e$9, and $$P = 1 - e^{-N_e}\sum_{x=0}^{N_o-1}\frac{N_e^x}{x!}.$$0 is consistent with a flat mean $$P = 1 - e^{-N_e}\sum_{x=0}^{N_o-1}\frac{N_e^x}{x!}.$$1. By contrast, a $$P = 1 - e^{-N_e}\sum_{x=0}^{N_o-1}\frac{N_e^x}{x!}.$$2 test rejects a flat specific frequency at very high significance, with $$P = 1 - e^{-N_e}\sum_{x=0}^{N_o-1}\frac{N_e^x}{x!}.$$3, for the full sample and for open and globular clusters separately (Geller et al., 2017).

Open clusters have the highest specific frequencies, with $$P = 1 - e^{-N_e}\sum_{x=0}^{N_o-1}\frac{N_e^x}{x!}.$$4. Extrapolating that value to a typical globular-cluster mass of $$P = 1 - e^{-N_e}\sum_{x=0}^{N_o-1}\frac{N_e^x}{x!}.$$5 would predict roughly 100 sub-subgiants, far more than are observed; the maximum observed counts are 19 in $$P = 1 - e^{-N_e}\sum_{x=0}^{N_o-1}\frac{N_e^x}{x!}.$$6 Cen and 8 in 47 Tuc. This indicates that globular clusters are less efficient, per unit mass, at producing observable sub-subgiants (Geller et al., 2017).

The formation-frequency analysis provides an environmental interpretation of this trend. Binary-channel sub-subgiants are produced more efficiently in diffuse clusters with lower central density, larger core radius, and smaller $$P = 1 - e^{-N_e}\sum_{x=0}^{N_o-1}\frac{N_e^x}{x!}.$$7, whereas the collision channels increase in importance with cluster mass and encounter rate. SG MT is especially vulnerable to dynamical perturbations, exchanges, and hardening that disrupt mass transfer, while SG Mag is less affected. This suggests that strong dynamical interactions in dense massive globulars may inhibit some sub-subgiant progenitors even as they open additional collisional pathways (Geller et al., 2017).

The 2025 field-giant study broadens the context by showing that sub-subgiants are also identifiable outside clusters as a photometrically and dynamically defined subset of active red giants. In that sample, the sub-subgiants have relatively small extinction and mostly lie outside the Galactic disk, which the authors note makes confusion with young active stars unlikely. Their synchronization, short periods, large ultraviolet excesses, and strong Ca II infrared triplet excess indices place them at the most active end of the giant-binary sequence (Dixon et al., 7 Apr 2025).

This broader perspective suggests continuity between cluster sub-subgiants and field analogs. The cluster work establishes the color-magnitude-diagram anomaly, membership-vetted demographics, and binary incidence; the field work shows that the same phenomenology extends into a larger population of synchronized, ultraviolet-bright giant binaries. A plausible implication is that sub-subgiants are best understood not as a single evolutionary product but as a distinct observational regime produced by several binary- and activity-driven pathways whose relative importance varies with environment (Geller et al., 2017, Geller et al., 2017, Dixon et al., 7 Apr 2025).