The galactic chemical evolution of carbon: Implications for stellar nucleosynthesis

TL;DR

This study investigates the galactic chemical evolution of carbon to discern the relative contributions of core-collapse supernovae (CCSN) and asymptotic giant branch (AGB) stars in the Milky Way. Using multi-zone GCE models implemented in VICE and calibrating against APOGEE subgiant abundances, the authors find that the total C yield must increase with metallicity to offset the decline in AGB C production, consistent with rotating massive-star models. The [C/Mg]-[Mg/Fe] relation tightly constrains delayed C production, implying that AGB stars contribute roughly 10–40% of solar C, with best fits favoring FRUITY-like AGB yields scaled by about 1.5; the results highlight a degeneracy between yield scale and outflows. Gas-phase abundances reveal tensions at low metallicity, suggesting non-monotonic massive-star C production or revisions to Fe yields, and pointing to the need for refined stellar evolution models. Overall, the work demonstrates how empirical abundance trends can constrain nucleosynthesis and guide improvements in stellar evolution theory.

Abstract

Carbon (C) is thought to be produced by both core collapse supernovae (CCSN) and asymptotic giant branch (AGB) stars, but the relative contributions of these two sources are uncertain. We investigate the astrophysical origin of C using models of Galactic chemical evolution (GCE) appropriate for the Milky Way disk. We benchmark our results against APOGEE subgiant abundances. The trend between [C/Mg] and [Mg/H] is set by the total C yield as a function of metallicity. Observations indicate a gently rising [C/Mg] with [Mg/H], but AGB C production is predicted to decline with metallicity. Our sample therefore favours a scenario in which CCSN yields rise with metallicity to offset declining AGB C yields and drive a subtle increase in [C/Mg] with [Mg/H]. This result is consistent with massive star nucleosynthesis models incorporating rotation. The [C/Mg]-[Mg/Fe] trend is sensitive to delayed enrichment and therefore constrains the amount of AGB C production. Given the slope of this relation, we find that AGB stars likely account for 10-40 per cent of C at solar metallicity. Artificially shifting the AGB C yields towards lower mass stars with longer lifetimes also improves agreement with the observed [C/Mg]-[Mg/Fe] trend, possibly indicating a discrepancy with stellar evolution predictions or our assumed Fe production rate.

Figures (11)

• Figure 1: The [C/Mg] ratio versus [Mg/H] (left) and [Mg/Fe] (right) for the jack sample of APOGEE subgiants. Left: High- and low-$\alpha$ stars are shown in blue and orange, respectively, using the separation defined in Eq. \ref{['eq:high_alpha']}. The black points represent the median trend along the low-$\alpha$ sequence. Right: Stars are colour-coded by their [Mg/H] abundance. The black points show the median [C/Mg]-[Mg/Fe] sequence for stars in the $-0.15 \leq [{\rm Mg/H}] \leq -0.05$ range. We use these median trends as our empirical benchmark in subsequent figures.
• Figure 2: The net fractional C yield from AGB stars as a function of initial stellar mass and metallicity. Each panel represents yields from one of four AGB studies from the literature: \ref{['fruity']}, \ref{['aton']}, \ref{['monash']}, \ref{['nugrid']} (see Section \ref{['sec:agb']}). The black dotted line shows ${y_{\rm C}^{\rm AGB}}=0$ for reference.
• Figure 3: Left: The integrated C yield from AGB stars, ${y_{\rm C}^{\rm AGB}}$, as a function of metallicity for each yield table, calculated $10\,{\rm Gyr}$ after a stellar population forms. (Note that the shape of the integrated yields arises from our linear interpolation of yields in $Z$. Changing to linear interpolation in $\log Z$ does not substantially affect our results.) Right: Cumulative C production as a function of age for a single stellar population of metallicity $\log Z/ Z_{\sun}=-0.1$. The dashed black line shows the cumulative return fraction of type Ia supernovae ($\propto t^{-1.1}$) for comparison. The cumulative production of \ref{['aton']} reaches a minimum of -3 at a time of 0.3 Gyr.
• Figure 4: IMF-integrated C yields from massive stars plotted as a function of metallicity. The right axis provides the equivalent CCSN [C/Mg] ratio, assuming our fiducial $y_{\rm Mg}$ yield. The black line is our fiducial massive star yield (see Eq. \ref{['eq:y_cc']}). Yields are shown for tables from LC18, \ref{['nugrid']} (P16, orange hexagons), sukhbold+16, NKT13, and WW95. sukhbold+16 report yields for different black hole landscapes, while LC18 provide yields at different rotational velocities. The light blue line denotes ${y_{\rm C}^{\rm AGB}}$ from the \ref{['fruity']} AGB star model for comparison (see Fig. \ref{['fig:agb-ssp']}). All models include wind yields.
• Figure 5: Time evolution of gas-phase C abundances in our fiducial model for [C/Mg] versus [Mg/H] (left) and [Mg/Fe] (right). Each line represents a zone at a different Galactic radius, colour-coded by lookback time. We plot zones at 1 kpc intervals between 2 and 15 kpc. Evolution in [C/Mg] with [Mg/H] varies substantially between regions of the Galaxy, while evolution with [Mg/Fe] is more uniform.