O–C Shell Mergers in Massive Stars

• O–C shell mergers are convective-reactive events in massive stars where the oxygen-burning shell ingests adjacent carbon/neon material, creating a large mixed region.
• Diagnostic criteria based on CO core mass and central 12C abundance indicate nearly a 50% likelihood of mergers in 12–25 M☉ models.
• These mergers drive enhanced odd-Z nucleosynthesis and modify pre-supernova structure, impacting explosion dynamics and observable supernova yields.

O–C shell mergers, also termed C–O shell mergers and, in some contexts, O–Ne–C shell interactions, are late-stage convective-reactive events in massive stars in which the convective oxygen-burning shell ingests material from adjacent carbon- and neon-rich layers and the two shells effectively merge into a single extended mixed region. They occur after central He burning, during advanced C/Ne/O shell burning, typically hours to days before core collapse, and they can strongly modify pre-supernova structure, odd-$Z$ nucleosynthesis, p-nucleus production, and the radioactive inventory ejected by the subsequent core-collapse supernova (Roberti et al., 26 Apr 2025).

1. Definition and stellar setting

In the relevant massive-star regime, successive H, He, C, Ne, O, and Si burning stages build an onion-like interior consisting of an Fe core surrounded by distinct burning shells. Near collapse, the star may contain an inner O-burning shell at temperatures $T \sim 2.6{-}2.8$ GK and an overlying C-burning shell at $T \sim 1$ GK. An O–C shell merger occurs when the convective O-burning shell penetrates into the overlying C- and Ne-rich layers, ingests that material, and forms a large mixed convective region in which C-rich fuel is exposed to O-burning temperatures (Roberti et al., 26 Apr 2025).

The physical trigger is weakening of the entropy and composition barrier between the shells. Once that barrier is sufficiently eroded, convection in the O shell entrains C/Ne-rich material upward and downward across the interface, and burning and mixing become strongly coupled. In the broader literature, these events are variously described as “C–O shell mergers,” “O–C shell mergers,” and “O–Ne–C shell interactions,” reflecting the fact that the O shell often interacts not only with C-rich material but also with Ne-rich layers in the same region (Roberti et al., 17 Sep 2025).

The phenomenon is intrinsically convective-reactive. A standard diagnostic is the Damköhler number,

$$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$

with $Da \sim 1$ indicating that convective transport and nuclear burning operate on comparable timescales. In this regime, the assumption of passive diffusive mixing is no longer adequate, because the flow changes the burning and the burning feeds back on the flow (Ritter et al., 2017).

2. Occurrence and early diagnostics

A systematic survey of 209 published 1D stellar models from FRANEC, MESA, KEPLER, and GENEC identified 41 models with a C–O shell merger. Those models span $9.2 \leq M_{\rm ini}/M_\odot \leq 25$, $0 \leq Z_{\rm ini} \leq 0.05$, and $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$. The key result is that merger occurrence can be approximated from the outcome of central He burning using two quantities: the CO core mass $M_{\rm CO}$, defined as the helium-exhausted core mass, and the central $^{12}\mathrm C$ mass fraction $T \sim 2.6{-}2.8$0 at the end of He-core burning (Roberti et al., 26 Apr 2025).

The empirical diagnostic is

$T \sim 2.6{-}2.8$1

Ninety percent of merger models satisfy these bounds, with average values

$T \sim 2.6{-}2.8$2

These conditions are necessary but not sufficient. Within that region of the $T \sim 2.6{-}2.8$3 plane, 79 models are present, of which 41 merge and 38 do not, implying a merger likelihood of about 50% once a model enters the favored region (Roberti et al., 26 Apr 2025).

The physical interpretation is twofold. A smaller CO core places later C, Ne, O, and Si burning shells closer together in mass coordinate, which makes interaction more likely. A lower central carbon abundance reduces subsequent Ne production and weakens the entropy jumps at the C–Ne–O interfaces, thereby making upward penetration by the O shell easier. Rotation and metallicity modify $T \sim 2.6{-}2.8$4 and $T \sim 2.6{-}2.8$5 indirectly, but rotation is not a direct trigger, and mergers occur at both solar and very low metallicity once the two structural diagnostics fall in the critical range (Roberti et al., 26 Apr 2025).

3. Hydrodynamics, convection, and modeling

In 1D stellar evolution, O–C shell mergers are identified by following convective boundaries, burning rates, and composition profiles through the final stages of evolution. A merger is registered when the O-burning convective shell overtakes the C/Ne shell and a single mixed convective zone forms. Detailed nucleosynthesis is then usually computed with post-processing networks that include heavy-ion fusion, proton captures, $T \sim 2.6{-}2.8$6 captures, and photodisintegrations, while the final explosion is often represented by parameterized or semi-analytic prescriptions rather than full neutrino-radiation hydrodynamics (Roberti et al., 26 Apr 2025).

Three-dimensional simulations show that the multi-D dynamics differ substantially from mixing-length theory. Shell merging is driven by turbulent entrainment and erosion of the stable layer between shells, convective velocities are much faster than MLT predicts, and merged shells develop strongly asymmetric, often dipolar, composition distributions. In a detailed 3D study of late shell merging in a $T \sim 2.6{-}2.8$7 model, entrainment and erosion of stable regions were identified as the main drivers of the merger, and multiple burning phases were observed within the same merged shell (Rizzuti et al., 2024).

For O-shell convection with C ingestion, 3D simulations find robust scaling laws: $T \sim 2.6{-}2.8$8 and show that carbon burning can provide $T \sim 2.6{-}2.8$9–$T \sim 1$0 of the total luminosity, depending on the adopted reaction set. Most runs settle into a quasi-stationary state dominated by a few large-scale convective cells, but an experimental high-feedback case develops global oscillations with Mach numbers in excess of $T \sim 1$1, representing a violent regime akin to a full O–C shell merger (Andrassy et al., 2018).

This multi-D behavior motivated later “3D-inspired” post-processing studies that replace MLT mixing with diffusion profiles featuring a downturn near the boundary, boosted convective velocities, quenched regions, and explicit C-shell ingestion rates. Those studies demonstrate that merger nucleosynthesis depends sensitively on the assumed macro-physics of entrainment and boundary mixing, not only on the underlying stellar structure (Issa et al., 19 Dec 2025).

4. Energetics and nucleosynthetic products

When an O–C shell merger occurs, the O shell expands by ingesting C/Ne-rich material, the former O and C/Ne shells are replaced by a single extended convective region, and composition profiles are flattened by convection. In the affected layers, C-burning products such as $T \sim 1$2, $T \sim 1$3, and $T \sim 1$4 are depleted, while O-burning products such as $T \sim 1$5, $T \sim 1$6, and $T \sim 1$7 are enhanced. Odd-$T \sim 1$8 isotopes including Cl, K, and Sc are produced efficiently in the same zone (Roberti et al., 26 Apr 2025).

A major revision of the standard picture is that, under typical merger conditions in 1D models, the dominant energy source is not classical O fusion but proton captures on S, P, and Ar. This set of reactions is termed the SPAr process and consists chiefly of

$T \sim 1$9

plus the corresponding reverse photodisintegration channels. In a representative $$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$0 model, SPAr proton captures produce approximately 400 times more energy than classical O fusion under the same conditions, and about 330 times more than the sum of C, O, and C+O fusion channels (Roberti et al., 17 Sep 2025).

The merger environment strongly enhances light-particle abundances. At the bottom of the O shell, the proton mass fraction rises from $$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$1 in classical O-shell burning to $$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$2 at the beginning of the merger and $$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$3 at the end; $$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$4-particle and neutron abundances increase by similar factors, and the neutron density reaches $$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$5. These light particles drive the odd-$$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$6 reaction flows. Examples include $$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$7, $$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$8, and the Sc-producing sequence $$Da=\frac{\tau_{\mathrm{mix}}}{\tau_{\mathrm{reac}}},$$9 (Roberti et al., 17 Sep 2025).

This nucleosynthesis is largely primary, because the seeds are generated within the O shell itself rather than inherited from the initial metallicity. In yield space, merger models occupy the high-$Da \sim 1$0, high-$Da \sim 1$1 region: $Da \sim 1$2 and $Da \sim 1$3, whereas non-merger models, especially at low metallicity, remain sub-solar in both ratios. This is the basis for the claim that C–O shell mergers resolve the longstanding underproduction of K and Sc in standard massive-star yields (Roberti et al., 26 Apr 2025).

O–C shell mergers are also pre-explosive $Da \sim 1$4-process sites. In a $Da \sim 1$5, $Da \sim 1$6 post-processing study, the average spread in p-nucleus overproduction across MLT and 3D-inspired mixing prescriptions is $Da \sim 1$7 dex, ingestion-rate uncertainties produce spreads of $Da \sim 1$8–$Da \sim 1$9 dex depending on the mixing profile, and nuclear-rate variations for unstable neutron-deficient isotopes from Se to Po contribute $9.2 \leq M_{\rm ini}/M_\odot \leq 25$0–$9.2 \leq M_{\rm ini}/M_\odot \leq 25$1 dex. This places macro-physics and nuclear-physics uncertainties on a comparable footing for quantitative p-nucleus predictions (Issa et al., 22 Jul 2025).

5. Structural consequences and the approach to explosion

Beyond nucleosynthesis, O–C shell mergers alter the pre-supernova structure itself. They weaken entropy and density discontinuities between shells, can create a density drop at the Si/O interface, and introduce non-radial asymmetries that are expected to be larger in reality than in 1D models. Because the merged region often sits at relatively large radius, the supernova shock modifies but does not erase the hydrostatic signature; in simplified shock arguments, the peak shock temperature scales as $9.2 \leq M_{\rm ini}/M_\odot \leq 25$2, so only moderate explosive reprocessing occurs in much of the merged shell (Roberti et al., 26 Apr 2025).

These structural changes are directly relevant to explodability. Closely related studies of strong C-shell burning adjacent to the O shell show that extra heating in the C shell can prevent contraction of outer layers, decrease the compactness parameter at $9.2 \leq M_{\rm ini}/M_\odot \leq 25$3, and steepen the density and pressure gradients at the O-burning shell. In that framework, the pressure gradient is described by $9.2 \leq M_{\rm ini}/M_\odot \leq 25$4 and the density drop by $9.2 \leq M_{\rm ini}/M_\odot \leq 25$5, where

$9.2 \leq M_{\rm ini}/M_\odot \leq 25$6

The proposed critical values are $9.2 \leq M_{\rm ini}/M_\odot \leq 25$7 and $9.2 \leq M_{\rm ini}/M_\odot \leq 25$8, above and below which, respectively, models are more likely to explode and form neutron stars rather than black holes (Xin et al., 16 Feb 2025).

Three-dimensional shell-merger simulations support the same qualitative picture. In a $9.2 \leq M_{\rm ini}/M_\odot \leq 25$9 simulation of a violent oxygen–neon shell merger just before collapse, the merged shell attained convective Mach numbers of $0 \leq Z_{\rm ini} \leq 0.05$0, erased the interface between the Ne layer and the Si-enriched O layer, and developed large-scale $0 \leq Z_{\rm ini} \leq 0.05$1 and $0 \leq Z_{\rm ini} \leq 0.05$2 asymmetries. Although that case is an O–Ne merger with partial C-shell involvement rather than a pure O–C merger, it demonstrates that shell mergers can seed the large-scale perturbations known to favor shock revival in core-collapse simulations (Yadav et al., 2019).

6. Observational and chemical-evolution evidence

Stellar archaeology supplies one line of evidence. A comparison to JINAbase abundances selected 121 metal-poor stars with K and Sc measurements in the range $0 \leq Z_{\rm ini} \leq 0.05$3, including 37 extremely metal-poor stars with $0 \leq Z_{\rm ini} \leq 0.05$4. The observations occupy a junction region between three model populations: non-merger CCSNe with low K and Sc, C–O merger CCSNe with high K and Sc, and neutrino-enhanced CCSNe with elevated Sc but not K. No single progenitor type reproduces the full observed distribution, but a significant contribution from C–O shell mergers is required to explain near-solar $0 \leq Z_{\rm ini} \leq 0.05$5 and moderately sub-solar $0 \leq Z_{\rm ini} \leq 0.05$6 at very low metallicity (Roberti et al., 26 Apr 2025).

Galactic chemical-evolution modeling reaches a similar conclusion. Earlier work showed that, if merger-like O–C shell interaction yields are taken as representative and such events occur in more than 50% of all stars, one-zone chemical-evolution models reproduce the observed Galactic trends of the odd-$0 \leq Z_{\rm ini} \leq 0.05$7 elements (Ritter et al., 2017). A later analysis linked the improvement at low metallicity specifically to rotating massive-star yields that frequently undergo C–O shell mergers, reinforcing the idea that the mergers are chemically important rather than exceptional curiosities (Roberti et al., 26 Apr 2025).

Young supernova remnants now provide a second, more direct observational route. A 2026 comparison of eight CCSN model sets to Cassiopeia A identified large $0 \leq Z_{\rm ini} \leq 0.05$8, $0 \leq Z_{\rm ini} \leq 0.05$9, and $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$0 as robust diagnostics of a C–O shell merger. Merger models produce $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$1, $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$2, $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$3, and typically $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$4, whereas non-merger models remain at lower ratios. For Cas A, only models with a C–O shell merger match the combined X-ray and infrared abundance constraints (Boccioli et al., 25 Mar 2026).

The same remnant constrains merger-produced $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$5. Merger-compatible models predict $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$6 to $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$7 of $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$8 synthesized in the C–O shell and eventually ejected, corresponding to roughly $0 \leq v_{\rm ini} \leq 800\,\mathrm{km\,s^{-1}}$9–30% of the total Cas A $M_{\rm CO}$0. Because this component originates above the Si/Si–O interface, it should lie outside the reverse shock and remain below current NuSTAR and COSI detection limits, though it may be accessible to future instruments such as ASCENT. By contrast, for SN 1987A the observed redshift of the $M_{\rm CO}$1 line argues against a dominant C–O-merger origin for the measured $M_{\rm CO}$2 (Boccioli et al., 25 Mar 2026).

7. Uncertainties and future directions

The frequency of O–C shell mergers remains uncertain. The sample of 41 mergers among 209 models is not weighted by realistic distributions in mass, metallicity, and rotation, so it is not itself a cosmic merger fraction. At the same time, the presence of mergers across FRANEC, MESA, KEPLER, and GENEC suggests that they are a generic possible feature of $M_{\rm CO}$3–$M_{\rm CO}$4 massive-star evolution rather than a code-specific artifact (Roberti et al., 26 Apr 2025).

The dominant theoretical uncertainties are convective-boundary physics, dimensionality, and nuclear rates. One-dimensional models treat convection with MLT plus diffusive mixing, whereas 3D calculations show more efficient entrainment, higher convective velocities, and possible quenching or global oscillatory states. This suggests that mergers may be more common in nature than in current 1D grids, but also that the detailed extent of mixing and burning in any given event may differ substantially from the 1D prediction (Roberti et al., 26 Apr 2025).

Nuclear uncertainties remain central. The $M_{\rm CO}$5 rate is relatively well constrained, with an uncertainty of about 20%, but shifts in that rate change the post-He-burning $M_{\rm CO}$6 and therefore the location of models in merger-favored parameter space. The heavy-ion fusion rates $M_{\rm CO}$7 and $M_{\rm CO}$8 are more uncertain, and they directly affect shell energetics and the onset of convective-reactive behavior. Future measurements at LUNA, JUNA, and LNS-INFN are expected to tighten those constraints (Roberti et al., 26 Apr 2025).

Quantitatively, uncertainties in 3D macro-physics are already as large as the nuclear uncertainties in some channels. For p nuclei, the spread from different mixing prescriptions rivals the spread from varying photodisintegration rates; for light odd-$M_{\rm CO}$9 products in 24 mixing cases, the pre-explosive yields of $^{12}\mathrm C$0, $^{12}\mathrm C$1, $^{12}\mathrm C$2, and $^{12}\mathrm C$3 vary over the intervals $^{12}\mathrm C$4, $^{12}\mathrm C$5, $^{12}\mathrm C$6, and $^{12}\mathrm C$7 dex, respectively, and the inferred $^{12}\mathrm C$8 yield varies by more than three orders of magnitude (Issa et al., 22 Jul 2025, Issa et al., 23 Sep 2025).

The immediate research program implied by these results is well defined: larger systematic grids that vary single stellar-physics inputs, more multi-D hydrodynamic simulations of late burning shells with realistic nuclear energetics, improved low-metallicity K and Sc observations, and chemical-evolution models that disentangle the roles of rotation, neutrino physics, and O–C shell mergers. A plausible implication is that the field is moving from asking whether O–C shell mergers can occur to asking how often they occur, under what structural conditions, and how strongly their multi-D dynamics reshape the final hours of massive-star evolution.