Nucleosynthesis of Cobalt and Copper

Abstract

Chemical abundances of cobalt (Co; Z = 27) and copper (Cu; Z = 29) in bulge and halo stars are presented and compared with chemical evolution models. The aim is to distinguish if Co and Cu are dominantly produced by neutron-capture or the alpha-rich freeze-out processes. Neutron-capture can be identified by a secondary behaviour in the [X/Fe] vs. [Fe/H] plot, and alpha-rich freeze-out would give rather a primary behaviour.

1. Introduction

Ref. [1], hereafter WW95 report on the formation, in massive stars, of the lower iron group, that include titanium (Ti; Z = 22), vanadium (V; Z = 23), chromium (Cr; Z = 24), manganese (Mn; Z = 25), and iron (Fe; Z = 26), mainly due to explosive burning of O and Si.

The upper iron group includes cobalt (Co; Z = 27), nickel (Ni; Z = 28), copper (Cu; Z = 29), zinc (Zn; Z = 30), gallium (Ga; Z = 31), and germanium (Ge; Z = 32). These elements are produced, according to WW95, through two processes: neutron capture on iron group nuclei during helium burning and later burning stages, and the alpha-rich freezeout at temperature T > 5 × 10⁹ K. Calculations are also given in [2,3]. There is some questioning of if Ge is the last of the iron-peak elements, as suggested by e.g., [4], or a neutron-capture element, with a similar behavior to As and Se, as found by [5]. Ref. [3] presented updated nucleosynthesis calculations on the weak s-process in massive stars, and adopting new reaction cross sections for isotopes beyond Fe, is able to produce higher amounts of Ge, As, and Se.

Another nucleosynthesis channel is the neutrino-interaction process ($\nu$-process), which enhances significantly the yields of odd-Z elements (see [6]), especially at very low metallicities.

From the observational point of view, it is verified that the elements V, Cr., and Ni vary in lockstep with Fe. On the other hand, Ti tends to be enhanced, like the alpha-elements, and it is often considered as an alpha-element (e.g., [7]). Zn tends to be enhanced as well (e.g., [8]). Mn instead shows a secondary behaviour, being deficient in metal-poor stars.

In the present paper, we compare data on Co and Cu in bulge and halo stars, in order to study their behaviour compared with the nucleosynthesis predictions.

2. Production of Co and Cu

This work was suggested by Stan Woosley, having pointed out that a verification with observational data would be useful, to check if the production of Co and Cu are mainly due to the alpha freeze-out or to neutron capture.

According to [9,10], in alpha-rich freezout ejecta, the density of alpha particles is so high that the end of equilibrium is caused by the “freezing out” of the forward reactions which involve alpha-particles. It favours the production of many nickel, copper, and zinc isotopes.

Ref. [11] confirmed that for the models of supernovae of 12–30 M⊙ Co, Ni, and Cu are made in the alpha-rich freeze-out, roughly in solar proportions. In their 9–120 M⊙ models, cobalt, copper, and nickel isotopes are well produced.

The other option to produce Co and Cu is the s-process (neutron-capture), that in massive stars takes place during helium burning.

Note that, for the sake of completeness, it has to be mentioned that at metallicities above [Fe/H]∼−1.0, or above 1 Gyr after SNII, Co and Cu are also produced in type Ia supernovae (SN Ia).

Figure 1 shows, for the fiducial model with specific star formation $\nu$ = 1 Gyr⁻¹ (see Section 4), the relative contribution of SN Ia, $f_{SN Ia}$, to the enrichment of Fe, Co, and Cu as a function of time. For t > 2 Gyr, corresponding approximately to [Fe/H] > −1, Co behaves as a typical iron-peak element, and SN Ia become its dominant source, with the $f_{SN Ia}$ of Co approaching rapidly that of iron. In contrast, SN Ia contribute less to Cu and does not reach a level of 60% of the total enrichment.

Our nucleosynthesis prescriptions for CCSNe with nearly-solar abundances are quite canonical. They do not include special prescriptions for weak s-process [12] nor yields from rotating massive stars [13]. When these processes are included, the contribution of SN Ia becomes less important in comparison to SNII, also at high metallicities, for cobalt and especially for copper. However, our conclusions for the bulge remain unaffected, since they refer to [Fe/H] $\lesssim -1$, recalling that our aim in the present work is the enrichment in Co and Cu at the metal-poor end of the Galactic chemical enrichment.

3. The Data

Our main sample consists of a sample of 58 stars that appear to represent the stellar population of an early bulge spheroid, based on chemical, kinematical, and dynamical criteria. These old bulge stars were selected to have metallicities of [Fe/H] $<-$0.8, chosen to exclude most of bulge stars that are metal-rich, and to include stars from a small metallicity peak at [Fe/H]∼−1.0 for bulge stars (e.g., [14,15]).

The sample was first described in [16]: it was based on the reduced proper motion (RPM) stars from [17]. The selection identified stars with a distance to the Galactic Center of d_GC < 4 kpc, a maximum height of ${\left|Z\right|}_{\max }<3$ kpc, eccentricity $>0.7$, and with orbits not supporting the bar, and imposing a metallicity cut of [Fe/H] $<-0.8$. By cross-matching this sample with stars having spectra from the Apache Point Observatory Galactic Evolution Experiment (APOGEE-2; [18]) a sample of 58 stars was built. In [16], the abundances of C, N, O, Mg, Si, Ca, and Ce were studied. Refs. [19,20] derived Na and Al, and iron-peak elements V, Cr, Mn, Co, Ni, and Cu, respectively, and interpreted these abundances in terms of their chemical-evolution models.

For the present work, we adopted the results from [20], and extended the study to the halo with results from [21,22] for Co, and [21,23] for Cu. For Cu, the lines are not measurable for very metal-poor stars. For cobalt, literature data on Co and Cu of bulge field-stars are from the following sources: [24] analysed 156 stars observed by [25] with the GIRAFFE spectrograph at the Very Large Telescope. Ref. [26] analysed a sample of bulge stars using the APOGEE Data Release 13. Refs. [27,28] analysed the same sample, originally the bulge stars from [29], plus some other stars previously studied by [30]. For this same sample of bulge stars, Mn and Zn were derived by [8,31] heavy elements Ba, La, Ce, Nd, and Eu by [32], and [33] reanalyzed C, N, O abundances. Ref. [34] analysed bulge stars based on data observed with the IGRINS spectrograph at the Gemini-South telescope, in the H and K-bands. For copper, the bulge field-star abundances are from [24,27,28,34,35], and also globular cluster data from [36]. Note that [35] present the NLTE copper abundances of the same sample analysed by [24], and that were originally observed by [25,37]—this shows how useful the bulge star sample of UVES and GIRAFFE data by [29], and [25,37] are for bulge studies. A review on iron-peak elements in the Galactic bulge was given by [38].

4. Chemical Evolution Models

The chemical-evolution model for the Galactic bulge is derived from the chemical-evolution model for elliptical galaxies of [39]. The application to the Galactic bulge was first presented by [33], then applied to alpha-elements, and consists of a multi-zone chemical evolution coupled with a hydrodynamical code. The gas receives feedback from the stellar population via heating, ionization, mechanical pressure and chemical enrichment. For the Galactic bulge, a classical spheroid with a baryonic mass of 2 × ${10}^9$ ${\mathrm{M}}_{\odot }$ and a dark halo mass of 1.3 × 10¹⁰ ${\mathrm{M}}_{\odot }$ are assumed. Cosmological parameters from the [40] are adopted, namely ${\mathrm{\Omega }}_m$ = 0.31, ${\mathrm{\Omega }}_{\mathrm{\Lambda }}$ = 0.69, Hubble constant ${\mathrm{H}}_0$ = 68 km ${\mathrm{s}}^{-1}$${\mathrm{Mpc}}^{-1}$, and an age of the Universe of 13.801 ± 0.024 Gyr.

For the nucleosynthesis yields, we adopt: (i) for massive stars, the metallicity-dependent yields from CCSNe/SNe II from WW95, and for low metallicities (Z < 0.01 ${\mathrm{Z}}_{\odot }$, or [Fe/H] $<-$2.5), the yields are from [41], which includes high explosion energy hypernovae (HNe); (ii) Type Ia Supernovae (SN Ia) yields are from [42]—their models W7 (progenitor star of initial metallicity Z = ${\mathrm{Z}}_{\odot }$) and W70 (zero initial metallicity); and (iii) for intermediate-mass stars (0.8–8 ${\mathrm{M}}_{\odot }$) with initial Z = 0.001, 0.004, 0.008, 0.02, and 0.4, we adopt yields from [43] with variable $\eta$ (AGB case), noting that the contribution of AGBs to Co and Cu is negligible.

Models of CCSNe from progenitors with low metallicities tend to underestimate the production of odd-Z elements (see e.g., [44]). Ref. [45], using a Galactic chemical evolution model with WW95 yields, have already observed this effect and recommended multiplying factors for the WW95 yields of some odd-Z elements, namely, three for V, two for Mn, and two for Co (Cu does not need this correction). These factors have been used in the fiducial model of [20].

For CCSNe, our nucleosynthesis prescriptions also include the enhancement of the yields of odd-Z elements by $\nu$-process at Z < 0.01 ${\mathrm{Z}}_{\odot }$ [6], or [Fe/H] <−2.5. The fiducial model of [20] includes $\nu$-process, assuming the case of [6] with total neutrino energy of the supernovae explosion ${\mathrm{E}}_{\nu }=3\times {10}^{53}$ erg. Note that the $\nu$-process is negligible for HNe with $M>25M_{\odot }$ of the progenitor star.

Models are computed for radii of r $<0.5$, $0.5<$ r $<1$, $1<$ r $<2$, and $2<$ r $<3$ kpc from the Galactic centre, and for specific star-formation rate values of $\nu =1$ and 3 ${\mathrm{Gyr}}^{-1}$.

Figure 2 shows the results of our models at different radii from the Galactic centre, for two specific star-formation rates, $\nu$ = 1 ${\mathrm{Gyr}}^{-1}$ and 3 ${\mathrm{Gyr}}^{-1}$. Also shown is the [Co/Fe] evolution of the fiducial model (Co WW95 yield multiplied by two) and using the original Co WW95 yield. In addition, Figure 2 displays a model without neutrino-interactions, thus allowing to assess the impact of this process on the Co/Fe and Cu/Fe ratios.

The new data allows us to extend the analysis of [20] to metallicities lower than [Fe/H] = −1.5. Now, the model predictions seem to be more consistent with the Co WW95 yields multiplied by two, but also suggest that WW95 may not underestimate so severely the Co yields, which could be multiplied by a factor smaller than 2 (the value recommended by [45]. This factor is not a mere fine-tuning parameter of the model, but rather it is critical to weight the relative importance of alpha-rich freezeout and s-process in the nucleosynthesis of Co in massive stars. A larger or smaller multiplication factor implies that alpha-rich freezeout contributes more or less to the Co yields.

Our models consider hypernovae [41] at metallicities [Fe/H] < −2.5. The hypernovae are responsible of the rise of [Co/Fe] at very low metallicities.

Co is in principle synthesized through alpha-rich freezeout, with a smaller contribution from the s-process. For Cu, we note that the s-process production between A = 60 and 88 is sensitive to the choice of key reaction rates, in particular to ²²Ne($\alpha$,n)²⁵Mg [2,46,47].

The data and our models indicate that Co is mainly synthesized through alpha-rich freezeout, with a smaller contribution from the s-process. This conclusion is less clear in the metallicity range −1.8 < [Fe/H] < −0.8, where the thick disk data from [21] tend do be flat, and the bulge data from [20] tend to drop. In this metallicity range the yields are from WW95. Cu has a clear secondary behaviour and should be built from the s-process.

Figure 2. [Co/Fe] vs. [Fe/H] and [Cu/Fe] vs. [Fe/H] for the present sample of bulge and halo stars. Data on Co are from [24,26,27,28,34], and [20] for bulge field stars, [48] for stars from bulge globular clusters, [21] for halo and thick disk stars, and [22] for halo stars. For Cu, additional results are shown from [35] for bulge field stars, and from [23] for halo stars. Different model lines correspond to the outputs of models computed for radii r < 0.5, 0.5 < r < 1, 1 < r < 2, and 2 < r < 3 kpc from the Galactic centre. Black lines correspond to specific star formation $\nu$ = 1 ${\mathrm{Gyr}}^{-1}$ plus $\nu$-process; red is the same without $\nu$-process; blue lines to $\nu$ = 3 ${\mathrm{Gyr}}^{-1}$ with $\nu$-process; magenta lines for $\nu$ = 1 ${\mathrm{Gyr}}^{-1}$, with $\nu$-process but without the factor of 2 for Co.

Figure 2. [Co/Fe] vs. [Fe/H] and [Cu/Fe] vs. [Fe/H] for the present sample of bulge and halo stars. Data on Co are from [24,26,27,28,34], and [20] for bulge field stars, [48] for stars from bulge globular clusters, [21] for halo and thick disk stars, and [22] for halo stars. For Cu, additional results are shown from [35] for bulge field stars, and from [23] for halo stars. Different model lines correspond to the outputs of models computed for radii r < 0.5, 0.5 < r < 1, 1 < r < 2, and 2 < r < 3 kpc from the Galactic centre. Black lines correspond to specific star formation $\nu$ = 1 ${\mathrm{Gyr}}^{-1}$ plus $\nu$-process; red is the same without $\nu$-process; blue lines to $\nu$ = 3 ${\mathrm{Gyr}}^{-1}$ with $\nu$-process; magenta lines for $\nu$ = 1 ${\mathrm{Gyr}}^{-1}$, with $\nu$-process but without the factor of 2 for Co.

5. Conclusions

A comparison of model predictions with the gathered data suggests that WW95 do not underestimate so severely the Co yields, and that a multiplication factor could be smaller than 2.

Our models consider hypernovae [41] at metallicities [Fe/H] < −2.5. The hypernovae are responsible for the rise of [Co/Fe] at very low metallicities.

The data and our models indicate that Co is mainly synthesized through alpha-rich freezeout, with a smaller contribution from the s-process.

Cu has a clear secondary behaviour and should be built from the s-process. For Cu, we note that the s-process production between A = 60 and 88 is sensitive to the choice of key reaction rates, in particular to ²²Ne($\alpha$,n)²⁵Mg [2,46,47].

The $\nu$-process is an important contributor to the Co nucleosynthesis at [Fe/H] < −2.5 and it is clearly needed to justify the relatively high abundances of Co at these metallicities. On the other hand, the contribution of neutrino interaction to Cu yields does not have a clear signature, and it cannot be tested since there are no Cu abundances for [Fe/H] < −2.5. The models using the yields down to [Fe/H] ≥ −2.5 are from WW95, plus [6] for the $\nu$-process, are consistent with the [21,23] data for halo and thick disk stars.