A&A 661, A86 (2022)

[Link] Astronomy
c ESO 2022 &
Astrophysics

The p-process in exploding rotating massive stars

A. Choplin1, S. Goriely1, R. Hirschi2,5,6 , N. Tominaga3,4,5,7 , and G. Meynet8

1
Institut d’Astronomie et d’Astrophysique, Université Libre de Bruxelles, CP 226, 1050 Brussels, Belgium
e-mail: [Link]@[Link]
2
Astrophysics Group, Lennard-Jones Labs 2.09, Keele University, ST5 5BG Staffordshire, UK
3
National Astronomical Observatory of Japan, National Institutes of Natural Sciences, 2-21-1 Osawa, Mitaka, Tokyo 181-8588,
Japan
4
Department of Astronomical Science, School of Physical Sciences, The Graduate University of Advanced Studies (SOKENDAI),
2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan
5
Kavli Institute for the Physics and Mathematics of the Universe (WPI), The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa,
Chiba 277-8583, Japan
6
UK Network for Bridging the Disciplines of Galactic Chemical Evolution (BRIDGCE), UK
7
Department of Physics, Faculty of Science and Engineering, Konan University, 8-9-1 Okamoto, Kobe, Hyogo 658-8501, Japan
8
Geneva Observatory, University of Geneva, Chemin Pegasi 51, 1290 Sauverny, Switzerland

Received 15 February 2022 / Accepted 25 March 2022

ABSTRACT

Context. The p-process nucleosynthesis can explain proton-rich isotopes that are heavier than iron, which are observed in the Solar
System, but discrepancies still persist (e.g. for the Mo and Ru p-isotopes), and some important questions concerning the astrophysical
site(s) of the p-process remain unanswered.
Aims. We investigate how the p-process operates in exploding rotating massive stars that have experienced an enhanced s-process
nucleosynthesis during their life through rotational mixing.
Methods. With the Geneva stellar evolution code, we computed 25 M stellar models at a metallicity of Z = 10−3 with different
initial rotation velocities and rates for the still largely uncertain 17 O(α,γ)21 Ne reaction. The nucleosynthesis calculation, followed
with a network of 737 isotopes, was coupled to stellar evolution, and the p-process nucleosynthesis was calculated in post-processing
during both the final evolutionary stages and spherical explosions of various energies. The explosions were modelled with a relativistic
hydrodynamical code.
Results. In our models, the p-nuclides are mainly synthesized during the explosion, but not much during the ultimate hydrostatic
burning stages. The p-process yields mostly depend on the initial number of trans-iron seeds, which in turn depend on the initial
rotation rate. We found that the impact of rotation on the p-process is comparable to the impact of rotation on the s-process. From no
to fast rotation, the s-process yields of nuclides with mass number A < 140 increase by 3−4 dex, and so do the p-process yields. Fast
rotation with a lower 17 O(α, γ) rate significantly produces s- and p-nuclides with A ≥ 140. The dependence of the p-process yields
on the explosion energy is very weak.
Conclusions. Our results suggest that the contribution of core-collapse supernovae from massive stars to the solar (and Galactic) p-
nuclei has been underestimated in the past, and more specifically, that the contribution from massive stars with sub-solar metallicities
may even dominate. A more detailed study including stellar models with a wide range of masses and metallicities remains to be
performed, together with a quantitative analysis that is based on the chemical evolution of the Galaxy.
Key words. stars: massive – stars: rotation – stars: interiors – stars: abundances – nuclear reactions, nucleosynthesis, abundances

1. Introduction Just et al. 2015), magnetorotational supernovae (Winteler et al.

2012; Nishimura et al. 2015), or collapsars (Siegel et al. 2019).
Despite tremendous progress during the past decades, the ori- At neutron densities in between the s- and r-processes, the exis-
gin of the trans-iron chemical elements is still debated and not tence of an intermediate (i) neutron capture process (first named
yet fully understood (e.g. Arnould & Goriely 2020). The slow by Cowan & Rose 1977) is expected. Its astrophysical site(s) is
(s) and rapid (r) neutron capture processes are the two main (are) actively debated (see Sect. 1 of Choplin et al. 2021, for a
processes that have each forged about the half of the trans-iron list of possible sites). Other nuclear processes also include short
nuclides. The s-process (e.g. the review of Käppeler et al. 2011) but possibly intense neutron bursts taking place in the helium
operates during the late life of asymptotic giant branch (AGB) shell of exploding massive stars (Blake & Schramm 1976;
stars (main s-process; e.g. Gallino et al. 1998; Herwig 2005; Thielemann et al. 1979; Meyer et al. 2004; Choplin et al. 2020).
Cristallo et al. 2011; Karakas & Lattanzio 2014) and during the This process is thought to cause the anomalous isotopic sig-
core helium-burning and shell carbon-burning stages of massive natures found in meteorites (Meyer et al. 2000; Pignatari et al.
stars (weak s-process; e.g. Langer et al. 1989; Prantzos et al. 2015, 2018), and possibly the abundances of some metal-poor
1990; Raiteri et al. 1991; The et al. 2007). The r-process is asso- r/s-stars (Choplin et al. 2020).
ciated with explosive events such as neutron star mergers (e.g. Finally, the p-process (Arnould 1976; Woosley & Howard
Arnould et al. 2007; Goriely et al. 2011; Wanajo et al. 2014; 1978; Meyer 1994; Arnould & Goriely 2003) is thought to

Article published by EDP Sciences

A86, page 1 of 11
 A&A 661, A86 (2022)

take place in the hydrostatic oxygen- and neon-burning shells was rotating with1 υini /υcrit = 0.4. All details of the input physics
of massive stars (Arnould 1976; Rauscher et al. 2002), dur- can be found in Choplin et al. (2018). We recall the main physi-
ing core-collapse supernovae (CCSNe; Rayet et al. 1995), or cal ingredients here.
during type Ia supernovae (Travaglio et al. 2015). The p- During stellar evolution, a nuclear reaction network of
process occurs through combinations of (γ,n), (γ,p), and 737 isotopes (from hydrogen to polonium), coupled to the
(γ,α) reactions affecting pre-existing s- or r-nuclides. The p- structure equations, was used. Opacity tables were computed
process provides a possible explanation for the neutron-deficient with the OPAL tool2 and complemented at low temperature
nuclides in Solar System abundances (e.g. Arnould & Goriely by opacities from Ferguson et al. (2005). Radiative mass-loss
2003; Lugaro et al. 2016). Some discrepancies remain, how- rates were taken from Vink et al. (2001) if log T eff ≥ 3.9
ever. In particular, 92,94 Mo and 96,98 Ru p-isotopes are system- and from de Jager et al. (1988) otherwise. For convection, the
atically underproduced (Rayet et al. 1995; Arnould & Goriely Schwarzschild criterion was used. During the H- and He-
2003, 2020, their Fig. 22). This fact motivates the search for burning phases, the size of the convective core is extended by
alternative or additional ways to produce these nuclides. In dover = αHP , with HP the pressure scale height estimated at
particular, the so-called pn process, i.e. a proton-poor neutron- the Schwarzschild boundary, and α = 0.1. Rotation was treated
boosted rp-process, which is encountered during He detona- according to the shellular scheme (Zahn 1992; Chaboyer & Zahn
tion, has been suggested as a promising nucleosynthesis source 1992; Maeder & Zahn 1998). The diffusion coefficients for hor-
(Goriely et al. 2005). Such an object is made of a carbon-oxygen izontal and vertical shear were taken from Zahn (1992) and
white dwarf with a sub-Chandrasekhar mass (M < 1.4 M ) Talon & Zahn (1997), respectively. The rates of 22 Ne(α, n)25 Mg
that accumulates a He-rich layer at its surface. An alternative and 22 Ne(α, γ)26 Mg were taken from Longland et al. (2012).
site proposed to explain the origin of the Mo and Ru p-nuclei The rates of 17 O(α, n)20 Ne and 17 O(α, γ)21 Ne were taken from
is the p-rich neutrino-driven wind in CCSNe, where antineu- Best et al. (2013) if not stated otherwise.
trino absorptions in the proton-rich environment produce neu- Choplin et al. (2018) evolved the stellar models up to core
trons that are immediately captured by neutron-deficient nuclei O-burning ignition with the 737 isotopes network. At core O-
(the νp-process; Fröhlich et al. 2006; Ghosh et al. 2022). Recent burning ignition, most of the stellar layers ejected at the time
results of Bliss et al. (2018) suggest that these proton-rich winds of the supernova have reached their final state and will not be
can make dominant contributions to the solar abundance of 98 Ru, impacted during O- and Si-burning phases. However, the inner-
but that additional astrophysical sources are likely required to most layers (below a mass coordinate of about Mr = 3 M for
account for 92,94 Mo and 96 Ru. The proton richness of the wind the non-rotating 25 M model; Fig. 1, left panel) are impacted
also remains a highly debated question. during these short ultimate stages. The p-process occurs in rel-
Costa et al. (2000) have shown that an increase in atively deep layers into the star, and for the present work, it is
22
Ne(α,n)25 Mg reaction during stellar evolution leads to a pre- therefore preferable to go further in the evolution to have a more
supernova seed distribution that can ultimately provide enough reliable pre-supernova structure of the inner layers. We thus con-
92,94
Mo and 96,98 Ru p-isotopes during the explosion. Interest- tinued the evolution until the end of the core Si-burning stage.
ingly, the 22 Ne(α,n)25 Mg reaction can be naturally boosted During these ultimate stages, the effect of rotation was switched
in rotating massive stars through additional production of off. This was shown to be a good approximation because the
22
Ne during core-helium burning (Meynet et al. 2006; Hirschi evolutionary timescale is far shorter than the rotational mixing
2007; Hirschi et al. 2008). The efficiency of the weak s-process timescale (Choplin et al. 2017, their Fig. 2). We also used the
is thus significantly boosted if the massive star is rotating standard minimum network of the Geneva code (e.g. Hirschi
(Pignatari et al. 2008; Frischknecht et al. 2016; Choplin et al. 2007; Ekström et al. 2012) instead of the full 737 isotope net-
2018; Limongi & Chieffi 2018; Banerjee et al. 2019). In rotat- work. This minimum network ensures a proper description of
ing massive stars, the pre-supernova distribution of trans-iron the energetics and keeps track of the main isotopes. The pre-
elements is therefore different from that in non-rotating massive supernova abundance profiles of some specific nuclei are illus-
stars. This will impact the p-process that can take place during trated in Fig. 1.
the explosion of such stars, and which largely depends on the In addition to the two reference 25 M models (with
trans-iron seeds before the explosion. υini /υcrit = 0 and 0.4), we investigated the case of a model with
In this paper we explore how rotation impacts the p-process the same initial properties that was a faster rotator, however, with
nucleosynthesis in exploding rotating massive stars and whether υini /υcrit = 0.7. Unlike the first two models, the evolution of
these stars can account for the abundances of solar p-nuclides. this model was not followed after core O-burning ignition. We
Section 2 presents the input physics relevant to the present work, adopted the same explosion model as for the υini /υcrit = 0.4
Sect. 3 discusses the results obtained in terms of s-process during model, but with the important difference that the initial abun-
the hydrostatic burning stages and of p-process during both the dances of trans-iron elements prior to the explosion were not the
hydrostatic and explosive evolutions. Section 4 investigates the same. This means that the same mass cut3 as for the υini /υcrit =
contribution of massive rotating stars to the Galactic p-nuclei. 0.4 model was adopted and the temperature and density histories
Conclusions are given in Sect. 5. of the stellar layers during the explosion were the same as for
the υini /υcrit = 0.4 model (cf. Sect. 2.2 for more details of the

2. Input physics 1
The critical velocity υcrit is reached when gravitational acceleration
is counterbalanced by centrifugal force. In the Roche approximation, it
2.1. Stellar evolution models q
is expressed as υcrit = 23 GMRp,c
, with Rp,c the polar radius at the critical
We started from two 25 M models at a metallicity of Z = 10−3 limit.
in mass fraction, which were computed with the Geneva stel- 2
[Link]
lar evolution code (Eggenberger et al. 2008) and published in 3
At the time of the explosion, the mass cut is the mass coordinate that
Choplin et al. (2018, models labelled 25S0 and 25S4 in their delimits the part of the star that is expelled from the part that is locked
Table 1). The first model was non-rotating, and the second model into the remnant.

A86, page 2 of 11
 A. Choplin et al.: The p-process in exploding rotating massive stars

1H 1H
10 1 10 1
Abund. Profile [mass frac.]

Abund. Profile [mass frac.]

4He 4He
12C 109 12C 109
16O 16O
10 3 56Ni 10 3 56Ni
108
88Sr 88Sr
108
10 5 138Ba
10 5 138Ba

T [K]

T [K]
208Pb 208Pb 107
T [K] T [K]
10 7 107 10 7
106
10 9 10 9
105
106
10 11 10 11
0 5 10 15 20 25 0 5 10 15 20 25
Mr [M ] Mr [M ]
Fig. 1. Pre-supernova abundance profiles for the M25Zm3V0 (non-rotating, left panel) and M25Zm3V4 (rotating, υini /υcrit = 0.4, right panel)
models computed with the rate of 17 O(α,γ)21 Ne from Best et al. (2013). The abundances of trans-iron elements are not explicitly followed from
core O-burning ignition (see text for details). The dashed grey line shows the pre-supernova temperature profile, the scale of which is indicated
on the right axis. The shaded grey area represents the remnant. The cyan area represents the zone in which the maximum temperature during a
spherical 1051 erg explosion ranges between 1.8 and 3.7 GK (i.e. the zone in which the explosive p-process takes place).

the s-process. The uncertainty on this rate was shown to dra-

matically affect the s-process efficiency in rotating massive stars
100 (e.g. Taggart et al. 2019). A low 17 O(α,γ)21 Ne rate enhances
the s-process efficiency because in this case, the competing
17
O(α,n)20 Ne reaction becomes dominant and recycles neutrons
10 1
v / v ref

(more details in Sect. 3.1, also in Sect. 3.5 of Choplin et al.

2018). Only a theoretical estimate of 17 O(α,γ)21 Ne is available
at low temperature (Caughlan & Fowler 1988; Best et al. 2013,
black and red lines in Fig. 2). Taggart et al. (2019) constrained its
10 2
lower limit experimentally. In the temperature range of interest
cf88 (ref.)
bg13 for the s-process, it is 10 to 1000 times lower than the recom-
bg13 / 10 mended rate of Best et al. (2013, green line in Fig. 2).
10 3 ta19 As mentioned above, in comparison with the standard mod-
els, the sets of models computed with the Best et al. (2013)
10 1 100 rate divided by 10 or the Taggart et al. (2019) rate experience
Temperature [GK] a more efficient s-process during the evolution because the
Fig. 2. Ratio of 17 O(α,γ)21 Ne reaction rates to the rate of
17
O(α,γ)21 Ne rate is lower (Fig. 2). Like for the υini /υcrit = 0.7
Caughlan & Fowler (1988, cf88 label). The red curve (label bg13) is the model with the rate of Best et al. (2013), the evolution after core
recommended rate of Best et al. (2013). The green curve (label ta19) is O-burning ignition is not followed for these six additional mod-
the experimental lower limit from Taggart et al. (2019). The shaded area els. We rely on the explosion of the corresponding models com-
indicates the approximate range of temperatures in the helium-burning puted with the Best et al. (2013) rate. Here again, this is a good
core of massive stars. approximation because for a given υini /υcrit ratio, the models
with different 17 O(α,γ)21 Ne rates behaves similarly. The pre-
supernova abundance distributions for the different models are
explosive modelling). This is a good approximation because the compared in Figs. 3 and 4 and are further discussed in Sect. 3.1.
υini /υcrit = 0.4 and 0.7 models have very similar characteristics Table 1 summarizes the main characteristics of the nine models
at core oxygen-burning ignition (in particular, similar tempera- considered in this work.
ture and density profiles, and similar CO-core masses). The only
important difference between the υini /υcrit = 0.4 and 0.7 mod- 2.2. Explosion models
els is the abundance of trans-iron elements prior to the explosion
(cf. Sect. 3.1). Two common and simple approximations for simulat-
For each of our three model stars, we also considered two ing the explosion of massive stars are the piston model
alternative models for which the rate of the 17 O(α,γ)21 Ne reac- (e.g. Woosley & Weaver 1995; Limongi et al. 2003;
tion was changed, as illustrated in Fig. 2. In one case (models B), Heger & Woosley 2010) and the thermal bomb (e.g.
we used the rate from Best et al. (2013) divided by 10 instead of Thielemann et al. 1996; Tominaga et al. 2007). The piston
the original rate from Best et al. (2013). In a second case (mod- model usually imitates the collapse and bounce of the inner
els C) we adopted the lower limit from Taggart et al. (2019) boundary. The radius of inner boundary is first reduced and then
instead of Best et al. (2013). Although theoretical and experi- increased. In the second method, the energy is deposited at the
mental works were carried out to study this rate (Descouvemont mass cut of the exploding star in the form of internal energy.
1993; Best et al. 2011, 2013; Taggart et al. 2019), significant More recent methods are the so-called PUSH (e.g. Perego et al.
uncertainties still remain in the temperature range of interest for 2015; Ebinger et al. 2019; Curtis et al. 2019) or P-HOTB (e.g.

A86, page 3 of 11
 A&A 661, A86 (2022)

10 5
M25Zm3V0 10 5
M25Zm3V0C
M25Zm3V4 M25Zm3V4C
10 6
M25Zm3V7 10 6
M25Zm3V7C
initial abundances initial abundances
10 7 10 7
Mass fraction

Mass fraction
10 8 10 8

10 9 10 9

10 10 10 10

10 11 10 11

10 12 10 12

10 13 10 13
60 80 100 120 140 160 180 200 60 80 100 120 140 160 180 200
Mass number (A) Mass number (A)
Fig. 3. Pre-supernova mass fractions of elements heavier than iron as a function of the mass number A at a mass coordinate of 4 M for the models
computed with the rate of 17 O(α,γ)21 Ne from Best et al. (2013, left panel) and Taggart et al. (2019, right panel). The four vertical dashed lines
show the location of the s-nuclides 56 Fe, 88 Sr, 138 Ba, and 208 Pb.

104 M25Zm3V0B Table 1. Characteristics of our 25 M , Z = 10−3 stellar models at the

M25Zm3V0C end of their evolution.
M25Zm3V4
103 M25Zm3V4B
M25Zm3V4C
M25Zm3V7 Model label υini /υcrit 17
O(α,γ)21 Ne MCO MNi
M25Zm3V7B
102 M25Zm3V7C [M ] [M ]
X/Xref

101 M25Zm3V0 0 Best et al. (2013) 5.90 1.11

M25Zm3V4 0.4 Best et al. (2013) 7.61 1.31
100 M25Zm3V7 0.7 Best et al. (2013) 7.56 –
M25Zm3V0B 0 Best et al. (2013)/10 5.88 –
10 1 M25Zm3V4B 0.4 Best et al. (2013)/10 7.64 –
M25Zm3V7B 0.7 Best et al. (2013)/10 7.62 –
10 2
M25Zm3V0C 0 Taggart et al. (2019) 5.88 –
10 20 30 40 50 60 70 80
Atomic number (Z) M25Zm3V4C 0.4 Taggart et al. (2019) 7.65 –
M25Zm3V7C 0.7 Taggart et al. (2019) 7.57 –
Fig. 4. Elemental mass fractions of all models normalized by the mass
fractions of the reference M25Zm3V0 model. Like in Fig. 3, the abun- Notes. MCO (Col. 6) is the mass of the carbon-oxygen core, defined
dances are shown at a mass coordinate of 4 M . The four vertical dashed where the mass fraction of 4 He has dropped below 10−2 . MNi (Col. 7) is
lines show the location of Fe (Z = 26), Sr (Z = 38), Ba (Z = 56), and the mass of the Ni core, defined when the mass fraction of 56 Ni drops
Pb (Z = 82). below 10−2 (the models without a value for MNi were not computed
until the very end of the evolution; see text for details).

Sukhbold et al. 2016; Ertl et al. 2020) approaches, in which the ticles representing Lagrangian mass elements of the stellar
mass cut and explosion energy can emerge from the simulation mantle.
instead of being assumed.
In this work, the explosion energy is deposited as kinetic 2.3. p-process nucleosynthesis
energy at the chosen mass cut. We used the relativistic hydrody-
namical code from Tominaga et al. (2007) and Tominaga (2009). During the last hydrostatic burning phases of massive stars and
Although this code allows the computation of two dimensional in the most inner layers, the trans-iron elements progressively
explosions, we considered one-dimensional spherical explo- photodisintegrate through a combination of (γ, n), (γ, α), and
sions. A code like this was already used in Choplin et al. (2020) (γ, p) reactions. It was shown that some p-nuclides could form
to investigate the effect of a jet-like explosion hitting the helium- before the supernova explosion in the oxygen-burning shell and
burning shell of rotating models computed with the Geneva that this process is very sensitive to initial mass or to the con-
code. vection model (Arnould 1976; Rauscher et al. 2002). The pro-
In the present work, a spherical explosion was triggered duction of p-nuclides during the explosion takes place at higher
in the non-rotating and the υini /υcrit = 0.4 models with the radii than during the hydrostatic evolution (in the layers that
Best et al. (2013) rate (M25Zm3V0 and M25Zm3V4 in Table 1). are sufficiently heated by the supernova wave). We considered
The total energy deposited was Etot = 1051 erg in the standard both production channels (during the evolution and during the
case. Different explosion energies are considered in Sect. 3.5. explosion). The p-process nucleosynthesis was treated in post-
The energy was deposited at a mass coordinate of 1.2 and processing calculations during the late hydrostatic evolution
1.4 M for the non-rotating and rotating model, respectively. (oxygen-burning onwards) and during the explosion.
These mass coordinates correspond to the top of the 56 Ni The nucleosynthesis was calculated with a code that was
core (Fig. 1). The hydrodynamics was followed for 500 s. The especially designed to follow all reactions of relevance dur-
temperature and density histories are recorded by mass par- ing the p-process nucleosynthesis (Arnould & Goriely 2003).
A86, page 4 of 11
 A. Choplin et al.: The p-process in exploding rotating massive stars

Changes in composition were followed by a full network cal- 1010

culation including all 2200 species from protons up to Z = 84 X(88Sr)
that lie between the proton-drip line and the neutron-rich region
that may be populated. All neutron and charged-particle fusion 3.7 GK 10 6
reactions as well as their reverse reactions on elements up to

Mass fraction
Po isotopes were included. The reaction rates on light species
were taken from the NETGEN library, which includes all the 1.8 GK
1.5 GK

T [K]
latest compilations of experimentally determined reaction rates
(Xu et al. 2013). Experimentally unknown reaction rates were 109 10 7
estimated with the TALYS code (Koning & Rochman 2012;
Goriely et al. 2008) on the basis of the Skyrme Hartree-Fock- end core Ne-b
Bogolyubov (HFB) nuclear mass model, HFB-24 (Goriely et al. end core O-b
2013), when they were not available experimentally. In addi- end core Si-b
tion to these reactions, electron captures and β-decays were also Tex, max
included. The corresponding rates were taken from experimen- 10 8
0 2 4 6 8 10 12
tal data (Kondev et al. 2021) when available, and from the gross
theory (Tachibana et al. 1990), otherwise. Mr [M ]
p-process nuclei are produced exclusively in layers heated Fig. 5. Temperature profiles at different stages of the evolution
at temperatures ranging between typically 1.8 and 3.5 × 109 K (blue lines) and maximum temperature reached during the explosion
(Rayet et al. 1990, 1995; Arnould & Goriely 2003). Below (magenta line) for the M25Zm3V4 model. The abundance profile of the
1.8 GK, the heavy seeds do not efficiently photodisintegrate dur- s-nuclide 88 Sr is shown in black with the scale on the right. The blue and
ing the evolutionary timescale. Above 3.5 GK, all seeds are pho- magenta rectangle show where the p-process can take place (in terms of
todisintegrated into iron-peak elements. For this reason, only temperature range and mass coordinate range) during the hydrostatic
layers with 1.8 . T max . 3.7 GK (where T max is the maximum and explosive burning, respectively (see text for details). The dark grey
temperature reached during the hydrostatic evolution and/or area shows the extent of the remnant (corresponding to the size of the
56
Ni core; cf. Sect. 2.2). The three horizontal dashed lines correspond
explosion) were post-processed in the present study. These lay- to temperatures of 1.5, 1.8, and 3.7 GK.
ers are referred to as p-process layers (PPLs).
In this work, the production of a p-nucleus i is described by
hFi i, which is the overproduction factor of this p-nucleus aver-
aged over the PPLs, i.e.
22
Ne through the 14 N(α, γ)18 F(β+ )18 O(α, γ)22 Ne chain. The neu-
M2 tron source 22 Ne(α,n)25 Mg is then boosted, and so is the s-
Z
1
hFi i = Xi (Mr ) dMr , (1) process (e.g. Pignatari et al. 2008; Frischknecht et al. 2012).
M p Xi,ini M1
The s-process also becomes more efficient with the
17
where M p = M2 − M1 is the total mass of the PPLs (these layers O(α,γ)21 Ne reaction rate from Taggart et al. (2019, right panel
are delimited by the lower Lagrangian mass coordinate M1 and in Fig. 3) compared to the rate of Best et al. (2013, left panel).
the upper one, M2 ), Xi,ini the initial abundance of the p-nucleus The reason is that the 16 O(n,γ)17 O(α,γ)21 Ne chain is weaker than
i and Xi (Mr ) the mass fraction of p-nucleus i at mass coordinate the competing 16 O(n,γ)17 O(α,n)20 Ne neutron recycling chain.
Mr (after the explosion and beta-decays, if not stated otherwise). The latter gives the neutrons back to the medium and hence
favours neutron captures by heavy seeds.
In the rotating models computed with the rate of Best et al.
3. Results (2013), rotation boosts the s-process by up to a factor of 10−103
for 30 < Z < 60 with a peak at Z ∼ 45 (solid lines in
Figure 1 shows that the non-rotating model (left panel) only lost Fig. 4). For a given rotation, the uncertainties associated with the
17
∼0.5 M through winds during the evolution, while the rotat- O(α,γ)21 Ne rate change the production by a factor of typically
ing model ejected about ∼8 M . The main reason for this is that 10, except when fast rotation and heavy elements with Z > 55
rotation produces larger helium-burning cores, which boosts the are considered, where the differences reach a factor of about 100
stellar luminosity and hence the mass loss. A more luminous (the solid and dashed red patterns in Fig. 4).
star is also more likely to enter the supra-Eddington regime (cf.
Sect. 3.2 in Choplin et al. 2018, for more details of this model).
3.2. p-process nucleosynthesis during hydrostatic burning

3.1. s-process nucleosynthesis during hydrostatic burning At solar metallicity, some p-nuclides can already form during
the ultimate hydrostatic burning stages in the oxygen-burning
s-process nuclei represent the seed from which p-process ele- shell, as initially suggested by Arnould (1976). Due to the rel-
ments are made. The s-process has been shown to be affected atively long evolution of the ultimate hydrostatic stages (with
by rotation during core helium-burning. Consequently, the p- respect to the explosion timescale), the p-process during hydro-
process will also be impacted later in the evolution or during static burning already takes place at a minimum temperature of
the explosion. about 1.5 GK (Rayet et al. 1990). In our M25Zm3V4 model, this
As shown in Fig. 3, the s-process becomes more efficient temperature was reached at some point during the evolution in
with increasing initial rotation. This is due to the stronger oper- the layers with Mr < 3.9 M . This is illustrated by the light blue
ation of the rotational mixing during the core-helium burning area in Fig. 5. We evaluated the p-process contribution during
phase. It first transports 12 C and 16 O from the He-core to the H- hydrostatic burning by post-processing the M25Zm3V4 model.
shell, which creates primary 14 N. The 14 N diffuses backward and We selected the layers that reached a temperature of at least 1 GK
penetrates the growing convective He-core. This makes primary before the end of the hydrostatic evolution. These layers were
A86, page 5 of 11
 A&A 661, A86 (2022)

10 5 burning phase is seen to be significant and smaller than the

explosive contribution by a factor of 2−5 for most nuclides.
This matter is further processed by the explosive nucleosynthe-
sis. Considering only the explosive nucleosynthesis is seen to be
10 7 an excellent approximation to the total hydrostatic plus explo-
Mass fraction

sive processing. It shows that at least in this model, p-nuclides

are mostly synthesized during the explosion. In other condi-
tions, and more particularly, in very massive (M & 100 M )
stars leading to pair creation supernovae, hydrostatic Ne/O burn-
10 9 ing has been shown to provide a dominant component to the
total production of p-nuclei (Rayet et al. 1993). As proposed by
88Sr Arnould & Goriely (2003), the p-process may also be found to
74Se develop in multi-dimensionally simulated pre-supernova O-rich
10 11 shells where additional convective mixing is induced by grav-
2 3 4 5 ity waves (Arnett 2001). In this case, this extended mixing may
Mr [M ] also move the p-nuclides that are synthesized before the explo-
Fig. 6. Abundance profiles of the s-nuclide 88 Sr (black) and the p-
sion to external stellar regions, in which they might survive the
nuclide 74 Se (red) in the region of the star that is relevant for the p- supernova explosion, in contrast to the pattern found in one-
process. The abundance profiles are shown at four different stages: at the dimensional simulations.
birth of the star (or zero-age main-sequence, dotted lines), at the start of
core oxygen burning (dash-dot lines, at this point, the 74 Se abundance
is about zero, hence not visible on the plot), at the end of the hydro- 3.3. p-process nucleosynthesis during the explosion
static post-processing calculation (dashed lines) and after the supernova We considered explosions of total energy Etot = 1051 erg (the
explosion (solid lines). impact of the explosion energy is discussed in Sect. 3.5). The
energy was deposited at a mass cut located at the top of the iron
10
4 (56 Ni) core. During the explosion, the p-process took place in
PPLs reaching a peak temperature between ∼1.8 and ∼3.5 GK
Overproduction factors <F >

M25Zm3V4
3
and corresponding to mass coordinates of 2.59 < Mr < 5.44 M
i

10
in our M25Zm3V4 model (magenta area in Fig. 5). The tempera-
2
10 ture and density evolution of several PPLs verifying this criteria
is shown in Fig. 8. The relatively high abundance of heavy s-
1
10 process seeds in this region (represented by the s-nucleus 88 Sr,
0
black line in Fig. 5) gives rise to an efficient p-process during
10 the explosion. This is visible in Fig. 6 for the specific cases
-1 Hydrostatic only of 88 Sr and 74 Se. As other s-nuclides, the pre-supernova 88 Sr
10 abundance (dashed black line) is high above Mr = 2.5 M .
Explosive only
-2 Hydrostatic+Explosive This enables an abundant production of p-nuclides such as 74 Se
10
80 100 120 140 160 180 200 during the explosion (solid red line). We note that the zone in
A the star in which the explosive p-process takes place is above
the zone in which the hydrostatic p-process takes place (com-
Fig. 7. Overproduction factors hFi i (Eq. (1)) of p-nuclei mass-averaged pare the blue and magenta area in Fig. 5) because the maxi-
over the 4.25 M PPLs (corresponding to the union of the blue and
mum temperature reached during the explosion is higher than in
magenta boxes in Fig. 5) in the M25Zm3V4 model at the end of the
hydrostatic burning pre-supernova phase (green diamonds), only due the hydrostatic evolution. The overlap between these regions is
to the explosive burning (blue) or including both the hydrostatic and about 1.3 M . This region contains p-nuclides produced by both
explosive burning phases (red; see text for more details). hydrostatic and explosive burning. In the inner blue zone (around
Mr = 2 M for instance), the p-nuclides built during hydro-
static burning are all destroyed by photodisintegration due to the
post-processed with our p-process nucleosynthetic code from high temperatures (T ex,max > 3.7 GK) encountered during the
the onset of core oxygen-burning to the pre-supernova stage. explosion.
Figure 6 shows the abundance profiles of the s-nuclide 88 Sr The s-process efficiency during the evolution increases with
(black) and p-nuclide 74 Se (red) at different burning stages. At initial rotation and with decreasing 17 O(α, γ) rate. Consequently,
the start of core oxygen burning, only 88 Sr is abundant because the p-process efficiency, which very strongly depends on the ini-
74 tial trans-iron seeds content, also increases with initial rotation
Se was fully destroyed by the s-process during core helium-
burning (dot-dashed black line in Fig. 6). During the ultimate and when the 17 O(α, γ) rate is lowered (Fig. 9). For p-nuclides
hydrostatic phases, 88 Sr is partially photodisintegrated at Mr < with 90 . A . 150, the overproduction differences between the
2.4 M (dashed black line) and 74 Se is synthesized around Mr ∼ non-rotating and the fast-rotating models reach about 3 dex at
2.2 M (dashed red line) through the photodisintegration of s- most (Fig. 9, black and red curves). This is similar for the dif-
nuclides. ferences shown in Figs. 3 and 4 (black and red curves) for the
Figure 7 shows the results of p-process nucleosynthesis s-process. For p-nuclides with A . 90 and A & 150, the dif-
when only the hydrostatic evolution (green), only the explosion ferences decrease, as expected from the pre-supernova s-process
(blue, see Sect. 3.3 for details; in this case, the abundances of yields shown in Figs. 3 and 4: they reach about 1 dex at most. A
heavy seeds are those at the start of the core oxygen-burning lower 17 O(α, γ) rate increases the s-process efficiency (Figs. 3
phase), or both the hydrostatic plus explosive burning (red) and 4), and thus also the p-process (Fig. 9). In particular, it
are considered. The contribution stemming from the hydrostatic leads to the production of a substantial amount of p-nuclides

A86, page 6 of 11
 A. Choplin et al.: The p-process in exploding rotating massive stars

4.0 Mr, ini = 2.61 M

Mr, ini = 2.98 M
3.5 Mr, ini = 3.4 M 106
Mr, ini = 3.95 M
3.0 Mr, ini = 4.51 M

[g cm 3]
Mr, ini = 5.36 M
2.5
T [GK]

2.0 105
1.5
1.0
104
10 1 100 101 10 1 100 101
time [s] time [s]
Fig. 8. Temperature (left panel) and density (right panel) evolution of several tracer particles coming from the M25Zm3V4 explosion model. The
maximum temperature of these particles during the explosion verifies 1.8 < T ex,max < 3.7 GK. These particles are initially located in the cyan area
in Fig. 1 (right panel).

with A & 100, as expected from the s-process yields shown in 10

4

Figs. 3 and 4. Overproduction factors <F >

3
i
Figure 10 illustrates the overproduction factors in the rotat- 10
ing 25 M models with υini /υcrit = 0.4 and 0.7, not only for 2
10
p-nuclei, but also for all stable nuclei. The p-nuclei appear to
be rather well co-produced with the α-elements such as 16 O, 10
1
20
Ne, 24 Mg, or 28 Si. Although the overabundances shown in 0
Fig. 10 only represent those characterizing the PPLs, the integra- 10
M25Zm3V0
tion over the entire stellar mass gives an overproduction of light -1 M25Zm3V4
10 M25Zm3V7
p-elements (A = 74−84 for M25Zm3V4 and A = 74−132 for <M25Zm3>
5
M25Zm3V7) similar to those of the light α-elements. By con- 10
Overproduction factors <F >

trast, the non-rotating model gives an overproduction factor of 10

4
i

about 100 in the PPLs, i.e. about ten times smaller than the factor 3
obtained in such stars for 16 O (black symbols in Fig. 10). This 10
difference is due to the low initial metallicity of about Z /10 10
2

adopted here. 10
1
Finally, we note that the emitted flux of (anti)neutrinos is not 0
followed in our explosion simulation. For this reason, neutrino- 10 M25Zm3V0B
M25Zm3V4B
induced nucleosynthesis that can boost the production of some 10
-1
M25Zm3V7B
rare p-nuclei, such as the odd-odd 138 La and 180 Ta (Goriely et al. 5 <M25Zm3B>
10
2001; Sieverding et al. 2018), is not described in the present
Overproduction factors <F >

4
work. 10
i

3
10
2
3.4. 92,94
Mo and 96,98 Ru p-nuclides 10
1
10
As shown in Fig. 9 and found systematically by previous cal-
0
culations (see e.g. Rayet et al. 1995; Arnould & Goriely 2003; 10 M25Zm3V0C
Travaglio et al. 2018), the p-nuclides 92,94 Mo and 96,98 Ru are -1 M25Zm3V4C
10 M25Zm3V7C
always underproduced by non-rotating models. In particular, the -2 <M25Zm3C>
overproduction factors of Mo-Ru p-isotopes are smaller by a fac- 10
80 100 120 140 160 180 200
tor of about 100 than the overproduction factors of the light p- A
nuclei 74 Se, 78 Kr, and 84 Sr. For rotating models, this deviation
is seen to be substantially reduced to a ratio of about 20 and Fig. 9. Overproduction factors hFi i (Eq. (1)) mass-averaged over all
3 for the models with υini /υcrit = 0.4 and 0.7, respectively. In PPLs for all p-nuclides when considering the 17 O(α,γ)21 Ne rate of
the M25Zm3V7B model with a lower 17 O(α,γ)21 Ne rate (Fig. 9, Best et al. (2013, top panel), Best et al. (2013) divided by 10 (middle),
and Taggart et al. (2019, bottom panel). The filled symbols highlight
middle panel, red pattern), the Mo-Ru isotopes are overproduced the n p p-nuclides (Table 2) that have the highest overabundances, i.e.
at the same level as the light p-nuclei, but the 100 . A . 132 p- those with hFi i higher than the highest value divided by an arbitrary
nuclides are now overproduced with respect to the light nuclides factor of 20. The green curves show the velocity-averaged yields over
by a factor of about 10. The increased production of Mo-Ru p- the three 25 M models assuming they follow the rotational distribution
nuclides is directly linked to the s-enrichment in heavy seeds of observed young B stars from Huang et al. (2010, their Fig. 6). See
such as Ba during the evolution of rotating models (cf. Sect. 3.1). text for more details.

A86, page 7 of 11
 A&A 661, A86 (2022)

1e 6
10
4
24
Mg
12
16
Overproduction factors <F >

3 O 28
Si
i

10 40
Ca 10 2.0
2
10
8 1.5
1
10

Mr [M ]

X(88Sr)
0
6
10 1.0
10
-1
M25Zm3V0 4
M25Zm3V4
M25Zm3V7 0.5
10
-2
2
0 50 100 150 200
A
0 50
Fig. 10. Overproduction factors hFi i (Eq. (1)) mass-averaged over all 10 1051 1052 1053
PPLs for all stable nuclei (open symbols) and p-nuclides (full symbols) E [erg]
for the 25 M models with the 17 O(α,γ)21 Ne rate from Best et al. (2013)
Fig. 11. Mass coordinate as a function of explosion energy for the
and with υini /υcrit = 0 (black diamonds), 0.4 (blue squares) and 0.7 (red
M25Zm3V4 model. The color code indicates the mass fraction of
circles). The overproduction factors of some α-elements, including 16 O,
the s-nuclide 88 Sr prior to the explosion. The two black lines delimit
are highlighted.
the zone in which the maximum temperature during the explosion veri-
fies 1.8 < T ex,max < 3.7 GK. The dark grey rectangular zone shows the
extent of the remnant. The dashed line shows the mass coordinate below
3.5. Impact of explosion energy on the p-process which some photodisentegration of trans-iron seeds can have occurred
during the last hydrostatic burning stages (zone in which T > 1.5 GK at
To obtain an efficient explosive p-process, a region in the star some point during hydrostatic burning).
must be enriched in trans-iron seeds and experience temper-
atures between 1.8 and 3.7 GK during the explosion. In the
4
M25Zm3V4 model, the region enriched in trans-iron seeds 10
extends to about Mr = 8 M (Figs. 5 and 11). Above this, the
Overproduction factors <F >
i

abundances of trans-iron elements are very similar to their ini- 3

10
tial abundances. The extension of the 1.8 < T ex,max < 3.7 GK
region is shown in Fig. 11 for various explosion energies. As 2
10
expected, this zone shifts upward in mass for more energetic
explosions. Nevertheless, for explosion energies 0.3 × 1051 < 1
Etot < 5 × 1052 erg, this zone stays in the stellar region that 10
is enriched in trans-iron seeds. This implies that an efficient
0 51
p-process takes place for a whole range of explosion energies 10 E =10 erg
tot
in this model. Moreover, the yields will be weakly impacted M25Zm3V4 52
E =10 erg
by the explosion energy because the pre-supernova trans-iron -1 tot
10
seed abundances vary weakly (by less than a factor of 2) up to 80 100 120 140 160 180 200
Mr = 8 M . A
The resulting abundance distribution of p-nuclides assuming
Fig. 12. Overproduction factors hFi i (Eq. (1)) mass-averaged over all
two different values of the explosion energy, namely Etot = 1051 PPLs for all p-nuclides for the M25Zm3V4 model with two values of
and 1052 erg, are illustrated in Fig. 12. Both distributions are the total explosion energy Etot = 1051 and 1052 erg.
globally identical, except for the lightest 74 Se and 78 Kr and
the A = 180 Ta and W p-isotopes. As shown in Fig. 11, for
Etot = 1052 erg, the PPLs extend up to the region in which s- young main-sequence B-type stars. We divided their υini /υcrit
process overabundances are highest. For these outer layers, the probability density function into three intervals corresponding
peak temperature remains relatively low (T ex,max ' 1.8 GK), and to our three 25 M models with υini /υcrit = 0, 0.4, and 0.7. The
this consequently essentially boosts the production of the heavi- first interval extends from υini /υcrit = 0 to 0.2, the second from
est p-nuclei. 0.2 to 0.55 (midway between 0.4 and 0.7), and the third from
We note that non-spherical explosions can lead to a wide 0.55 to 1. We then integrated their probability density function
variety of temperature and density conditions compared to the over these three intervals to obtain the relative weights of 0.14,
spherical explosions considered in this work. This could ulti- 0.45, and 0.41 for the υini /υcrit = 0, 0.4, and 0.7 models, respec-
mately impact the yields of p-nuclides. We plan to investigate tively. The green curves in Fig. 9 (also reported in Fig. 13) show
the effect of the explosion geometry on the p-process nucleosyn- the resulting overproduction factors when weighting the yields
thesis in a future work. of our three models by the derived coefficients. In each of the
three panels of Fig. 9, the final yields are mainly influenced by
3.6. Velocity-averaged yields for a 25 M star the fast-rotating υini /υcrit = 0.7 model, which indeed has (i) high
overproduction factors and (ii) a relatively high weight of 0.41.
To make a first rough estimate of the integrated p-nuclide yields Overall, at this mass and metallicity, the p-nuclide yields of a
of a population of rotating massive stars, we can use the velocity fast-rotating (typically υini /υcrit = 0.7) massive star are likely
distribution derived by Huang et al. (2010, their Fig. 6). Their well representative of a population of massive stars with vari-
velocity distribution was obtained from the observation of 220 ous initial rotation rates. A more detailed estimation including a
A86, page 8 of 11
 A. Choplin et al.: The p-process in exploding rotating massive stars

is the initial mass. The supernova ejecta enrich the interstellar

medium with oxygen and p-nucleus i in solar proportion if
yO XO,
= , (5)
yi Xi,
where XO, and Xi, are the solar mass fraction of oxygen and p-
nuclide i, respectively. Combining Eq. (5) with Eqs. (3) and (4)
gives the (O/p)0 ratio for a given averaged p-enrichment, given
by F0 as
yO Xi, YO /XO, − fZ Mini
(O/p)0 ≡ = , (6)
yi XO, fZ F0 M p − fZ MCO
where fZ = Z/Z = 0.001/0.014 = 0.071 in our case. We note
that considering a simple average overproduction factor F0 may
not be optimal in our case because of the relatively large disper-
Fig. 13. Same as Fig. 9, where only the overproduction factors for the sion of the overproduction factors hFi i in some of our models
three velocity-averaged models are compared. (Fig. 9). Therefore we also define an averaged overproduction
factor F1 , which in contrast to F0 only considers the most abun-
dant p-nuclei. This new factor F1 is defined by an equation simi-
finer initial rotational grid, different masses, and metallicities, is lar to Eq. (2), but the sum runs only over the n p p-nuclides (given
deferred to a future work. in Table 2), the mass fraction of which is greater than Xmax /20,
where Xmax is the mass fraction of the most abundant p-nucleus.
4. Contribution of massive rotating stars to Galactic These n p p-nuclides are highlighted with filled symbols in Fig. 9.
The factor of 20 considered here is purely arbitrary and aims
p-nuclei at considering the various uncertainties associated with both the
Since oxygen is known to be essentially produced by CCSNe, astrophysical modelling (including those affecting the s-process
the contribution of massive stars to the Galactic p-nuclei enrich- in massive stars) and the nuclear ingredients (see in particular
ment can be estimated through the oxygen to p-nuclide yield Fig. 35 of Arnould & Goriely 2003). Therefore, we define two
ratio normalized to the solar abundances, (O/p), as defined by ratios (O/p)0 and (O/p)1 depending on whether the (O/p) ratio in
Rayet et al. (1995). The calculation of this quantity requires the Eq. (6) is calculated with F0 or F1 (Table 2). We also computed
estimate of the average overproduction factor F0 over the 35 p- the (O/p) ratios for the velocity-averaged models, as reported in
nuclides given by the fourth row of each series of models in Table 2.
X If (O/p) = 1, the p-nuclides are produced in solar propor-
F0 = hFi i/35, (2) tion with oxygen on average. However, if our models predict
i (O/p) > 1, p-nuclides are predicted to be underproduced com-
pared to oxygen. This would imply that our models alone can-
where hFi i is the overproduction factor of p-nuclide i averaged not fully account for the p-nuclide enrichment in the Solar Sys-
over the PPLs (Eq. (1)). We note that we considered only the tem, and, more generally, in the Galaxy. As shown in Table 2,
explosive contribution of the p-process to estimate the (O/p) 0.17 < (O/p)0 < 61.3 and 0.12 < (O/p)1 < 36.9. The (O/p)
ratios (which is a good approximation, as shown in Fig. 7). ratios rapidly drop with increasing rotation and with decreasing
Therefore, the PPLs mass M p used in Eq. (1) and given in Table 2 17
O(α, γ) rate, meaning that more p-nuclei are produced com-
corresponds to the zone in the star that experiences p-process pared to oxygen if massive stars are rotating or if the 17 O(α, γ)
only during the explosion (magenta zone in Fig. 5). rate is lower (as already discussed in Sect. 3).
The net yield yi of a p-nuclide i is defined as the difference Rayet et al. (1995) considered solar metallicity non-rotating
between the mass of a p-nuclide i returned by the star to the inter- massive stars with initial masses ranging between 13 and 25 M .
stellar medium and the mass of that same p-nuclide engulfed at They found 1.8 < (O/p)0 < 8.4 and a ratio of 4.2 when they inte-
the birth of the star. In our case, it can be simplified as follows: grated over the initial mass function from Kroupa et al. (1993).
In our case, the (O/p)0 ratios for non-rotating models are sig-
yi = Xi,ini F0 M p − Xi,ini MCO , (3) nificantly higher, with values ranging between 6.7 and 61.3
where MCO is the carbon-oxygen core mass, which also cor- (Table 2). Nevertheless, their results are compatible with ours
responds to the helium-free core (Table 2). As in Rayet et al. in view of the different metallicities considered. Oxygen is a
(1995), Eq. (3) assumes that the p-nuclei initially present in the primary product, in contrast to s- and p-nuclei, which are sec-
helium-burning core have been destroyed by the s-process and ondary in non-rotating models and are hence metallicity depen-
that their production in the PPLs can be averaged by the same dent. Rayet et al. (1995) concluded that about one quarter of the
overproduction factor F0 . solar system p-nuclei could be attributed to supernovae from
To derive the (O/p) ratio, a similar estimate of the net oxygen massive stars (assuming that the entire solar oxygen is coming
yield can be derived by from such events). Similarly, on the basis of a Galactic chemical
evolution model, Travaglio et al. (2018) estimated that CCSNe
yO = YO − XO,ini Mini , (4) from populations of non-rotating massive stars with different ini-
tial masses and metallicities could contribute no more than 10%
where YO is the absolute oxygen yield (i.e. the total mass of to the Galactic p-enrichment, with only a few exceptions (e.g.
oxygen ejected through stellar winds and supernova, as reported the light p-nuclides 74 Se, 78 Kr, and 84 Sr). The contribution stem-
in Table 2), XO,ini is the initial oxygen mass fraction, and Mini ming from models with sub-solar metallicities was also found
A86, page 9 of 11
 A&A 661, A86 (2022)

Table 2. Quantities required to estimate the (O/p) ratio corresponding to the oxygen to p-nuclide yield ratio normalized to the solar abundances
(Eq. (6)).

Model label Mini υini /υcrit 17

O(α,γ)21 Ne MCO Mp YO F0 F1 np (O/p)0 (O/p)1
[M ] [M ] [M ] [M ]
M25Zm3V0 25 0 Best et al. (2013) 5.90 2.37 2.75 30.2 50.8 20 62.0 37.1
M25Zm3V4 25 0.4 Best et al. (2013) 7.61 2.93 4.56 312 908 10 7.30 2.50
M25Zm3V7 25 0.7 Best et al. (2013) 7.56 2.93 5.30 1229 1814 23 2.13 1.39
hM25Zm3i − − Best et al. (2013) 7.35 2.93 4.61 649 952 23 3.55 2.41
M25Zm3V0B 25 0 Best et al. (2013)/10 5.88 2.37 2.91 62.3 154 13 30.0 11.8
M25Zm3V4B 25 0.4 Best et al. (2013)/10 7.64 2.93 5.23 1047 1544 23 2.49 1.69
M25Zm3V7B 25 0.7 Best et al. (2013)/10 7.62 2.93 5.27 8373 11346 25 0.31 0.23
hM25Zm3Bi − − Best et al. (2013)/10 7.39 2.93 4.92 3914 4963 27 0.63 0.49
M25Zm3V0C 25 0 Taggart et al. (2019) 5.88 2.37 2.91 86.3 271 10 21.4 6.68
M25Zm3V4C 25 0.4 Taggart et al. (2019) 7.65 2.93 5.21 2353 3346 24 1.10 0.78
M25Zm3V7C 25 0.7 Taggart et al. (2019) 7.57 2.93 5.28 15008 21084 24 0.18 0.12
hM25Zm3Ci − − Taggart et al. (2019) 7.37 2.93 4.91 7229 9734 25 0.34 0.25
Notes. MCO (Col. 5) is the mass of the carbon-oxygen core (or helium-free core), M p (Col. 6) is the mass of the layers undergoing p-process
nucleosynthesis during the explosion, YO (Col. 7) is the yield (total ejected mass) of oxygen, F0 (Col. 8) is the average overproduction factor
(Eq. (2)), F1 (Col. 9) is like F0 , but the sum runs only over the n p (Col. 10) most abundant p-nuclides, (O/p)0 (Col. 11) and (O/p)1 (Col. 12) are
computed with F0 and F1 , respectively. The hM25Zm3i, hM25Zm3Bi, and hM25Zm3Ci rows correspond to the velocity-averaged yield of the
three models above (cf. Sect. 3.6 for details).

to be negligible due to the secondary nature of the s- and p- evolution due to the effect of rotational mixing. Consequently,
processes. they are enriched in trans-iron seeds in the last evolutionary
Our results with (O/p) ratios close to or even below one stages and during the explosion. Because of the higher abun-
(Table 2) suggest that sub-solar-metallicity rotating massive stars dance in heavy seeds, they are expected to experience an efficient
can co-produce or even overproduce p-nuclei with respect to p-process nucleosynthesis during their explosion and possibly
oxygen. The impact of rotation on the s-process was found to be during their last hydrostatic burning stages.
so large in stars with a metallicity Z ' 0.001 (Frischknecht et al. We computed 25 M stellar models at a metallicity of Z =
2016; Choplin et al. 2018, 2020) that it gives rise to large 10−3 with different initial rotation velocities and 17 O(α,γ)21 Ne
overproduction factors of p-nuclei (see Table 2 and Fig. 9) rates. We found that most of the p-nuclides are synthesized dur-
that can compensate for their sub-solar metallicity. Table 5 in ing the explosion. The impact of rotation on the p-process fol-
Frischknecht et al. (2016) shows that the overproduction fac- lows the effect of rotation on the s-process. From no to fast
tors of light s-nuclides for their solar metallicity models are rotation, both the s- and p-process efficiencies are boosted by
∼50 at most and that rotation increases them by no more than about 3−4 dex for nuclides with A < 140. The impact remains
a factor of about 2. The overproduction factors of p-nuclides small for nuclides with A ≥ 140, unless both fast rotation and a
for these models therefore is not significantly affected by rota- lower 17 O(α,γ)21 Ne reaction rate are considered. In this case,
tion and is characterized by averaged values of F0 ' 100 (see a significant number of heavy p-nuclides (with A ≥ 140) is
e.g. Table 3 in Rayet et al. 1995). The averaged overproduc- produced. The impact of the explosion energy on the p-process
tion factors of our models with respect to the solar abundances yields remains weak because a different explosion energy just
F0, and F1, can also be estimated from our Table 2 by sim- shifts the zone in the star in which the p-process takes place.
ply dividing the F0 and F1 factors by the Z /Z = 14 ratio. Because this zone is similarly enriched in s-nuclides (built dur-
Considering the velocity-averaged yields of p-nuclides, overpro- ing the previous evolutionary stages), the p-process yields are
duction factors of F0, ' 50 and F1, ' 70 are found when weakly impacted.
considering the rates of Best et al. (2013). When using the rate By considering a population of solar-metallicity non-rotating
of Best et al. (2013) divided by 10, we find F0, ' 280 and massive stars, Rayet et al. (1995) and Travaglio et al. (2018)
F1, ' 350. Finally, F0, ' 520 and F1, ' 700 if using the rate concluded that CCSNe from populations of non-rotating mas-
of Taggart et al. (2019). These value are similar to or higher than sive stars could contribute no more than 10−25% to the Galactic
those expected from solar metallicity non-rotating and rotating p-enrichment, and as in previous works, they reported that sub-
massive stars. On this basis, sub-solar-metallicity rotating stars solar-metallicity stars play a minor role. However, their stud-
may even be dominant contributors to the Galactic enrichment of ies only considered non-rotating models and may consequently
p-nuclei (and s-nuclei) compared to solar metallicity (rotating) have underestimated the role of CCSNe in the Galactic enrich-
stars. Detailed Galactic evolution models are needed to quantify ment of p-nuclei. Our present results suggest that rotating mas-
their global contribution. Such a quantitative study is postponed sive stars with sub-solar metallicity may substantially contribute
to a future work. to the Galactic p-enrichment, and that the global contribution
stemming from CCSNe needs to be revisited. In particular, rotat-
5. Summary and conclusions ing massive stars with sub-solar metallicity may be the dom-
inant contributors compared to solar metallicity massive stars
We studied the p-process in rotating massive stars during their because the effect of rotational mixing at sub-solar metallic-
ultimate hydrostatic burning stages and during their explosions. ity is stronger. However, before drawing quantitative conclu-
These stars can experience an enhanced s-process during their sions, more detailed studies are required that use a larger grid

A86, page 10 of 11
 A. Choplin et al.: The p-process in exploding rotating massive stars

of masses, metallicities, and rotation velocities, supplemented Fröhlich, C., Martínez-Pinedo, G., Liebendörfer, M., et al. 2006, Phys. Rev. Lett.,
by Galactic chemical evolution simulations. 96, 142502
The p-process efficiency in rotating massive stars also remains Gallino, R., Arlandini, C., Busso, M., et al. 1998, ApJ, 497, 388
Ghosh, S., Wolfe, N., & Fröhlich, C. 2022, ApJ, 929, 43
very strongly impacted by the uncertainty associated with the Goriely, S., Arnould, M., Borzov, I., & Rayet, M. 2001, A&A, 375, L35
17
O(α,γ)21 Ne reaction rate and its effects on the s-process seed Goriely, S., Garcia-Senz, D., Bravo, E., & José, J. 2005, A&A, 444, L1
distribution. An accurate determination of this rate will greatly Goriely, S., Hilaire, S., & Koning, A. J. 2008, A&A, 487, 767
help constrain the contribution of massive rotating stars to the Goriely, S., Bauswein, A., & Janka, H.-T. 2011, ApJ, 738, L32
Goriely, S., Chamel, N., & Pearson, J. 2013, Phys. Rev. C, 88, 024308
Galactic content not only in s-nuclei, but also in p-nuclei. Heger, A., & Woosley, S. E. 2010, ApJ, 724, 341
As mentioned at the end of Sect. 3.5, the p-process Herwig, F. 2005, ARA&A, 43, 435
nucleosynthesis in self-consistent multi-dimensional explosions Hirschi, R. 2007, A&A, 461, 571
from rotating progenitors also remains to be assessed. These Hirschi, R., Chiappini, C., Meynet, G., Maeder, A., & Ekström, S. 2008, in
models are thought to experience r-process nucleosynthesis Massive Stars as Cosmic Engines, eds. F. Bresolin, P. A. Crowther, & J. Puls,
250, 217
(Winteler et al. 2012; Nishimura et al. 2015; Mösta et al. 2018; Huang, W., Gies, D. R., & McSwain, M. V. 2010, ApJ, 722, 605
Reichert et al. 2021). Our simulations suggest that exploding Just, O., Bauswein, A., Ardevol Pulpillo, R., Goriely, S., & Janka, H.-T. 2015,
rotating massive stars can be rich nucleosynthesis astrophysical MNRAS, 448, 541
sites for the production of elements heavier than iron, in particu- Käppeler, F., Gallino, R., Bisterzo, S., & Aoki, W. 2011, Rev. Mod. Phys., 83,
157
lar by enriching the interstellar medium in (1) s-process material Karakas, A. I., & Lattanzio, J. C. 2014, PASA, 31, e030
from their outer He-rich layers, (2) p-process material from deep Kondev, F., Wang, M., Huang, W., Naimi, S., & Audi, G. 2021, Chin. Phys. C,
O-rich layers, and (3) some r-process material from the inner- 45, 030001
most ejecta. Koning, A. J., & Rochman, D. 2012, Nucl. Data Sheets, 113, 2841
Finally, our study once again highlights the important impact Kroupa, P., Tout, C. A., & Gilmore, G. 1993, MNRAS, 262, 545
Langer, N., Arcoragi, J.-P., & Arnould, M. 1989, A&A, 210, 187
of mixing mechanisms in deep stellar interiors on the nucleosyn- Limongi, M., & Chieffi, A. 2018, ApJS, 237, 13
thesis in general, and of the p-process in particular. The only Limongi, M., Chieffi, A., & Bonifacio, P. 2003, ApJ, 594, L123
way to substantially improve the current situation in this respect Longland, R., Iliadis, C., & Karakas, A. I. 2012, Phys. Rev. C, 85, 065809
is to master multi-dimensional star simulations in future devel- Lugaro, M., Pignatari, M., Ott, U., et al. 2016, Proc. Natl. Acad. Sci., 113, 907
Maeder, A., & Zahn, J.-P. 1998, A&A, 334, 1000
opments. Meyer, B. S. 1994, ARA&A, 32, 153
Meyer, B. S., Clayton, D. D., & The, L.-S. 2000, ApJ, 540, L49
Acknowledgements. We wish to thank M. Arnould for his careful reading and Meyer, B. S., The, L. S., Clayton, D. D., & El Eid, M. F. 2004, in Lunar and
valuable remarks. This work was supported by the Fonds de la Recherche Planetary Science Conference, eds. S. Mackwell, & E. Stansbery, 1908
Scientifique-FNRS under Grant No IISN 4.4502.19. S.G. is FRS-F.N.R.S. Meynet, G., Ekström, S., & Maeder, A. 2006, A&A, 447, 623
research associate. R.H. acknowledges support from the World Premier Interna- Mösta, P., Roberts, L. F., Halevi, G., et al. 2018, ApJ, 864, 171
tional Research Centre Initiative (WPI Initiative, MEXT, Japan), STFC UK, the Nishimura, N., Takiwaki, T., & Thielemann, F.-K. 2015, ApJ, 810, 109
European Union’s Horizon 2020 research and innovation programme under grant Perego, A., Hempel, M., Fröhlich, C., et al. 2015, ApJ, 806, 275
agreement No 101008324 (ChETEC-INFRA) and the IReNA AccelNet Net- Pignatari, M., Gallino, R., Meynet, G., et al. 2008, ApJ, 687, L95
work of Networks, supported by the National Science Foundation under Grant Pignatari, M., Zinner, E., Hoppe, P., et al. 2015, ApJ, 808, L43
No. OISE-1927130. R.H. also acknowledges support from the ChETEC COST Pignatari, M., Hoppe, P., Trappitsch, R., et al. 2018, Geochim. Cosmochim. Acta,
Action (CA16117), supported by COST (European Cooperation in Science and 221, 37
Technology). G.M. has received funding from the European Research Council Prantzos, N., Hashimoto, M., & Nomoto, K. 1990, A&A, 234, 211
(ERC) under the European Union’s Horizon 2020 research and innovation pro- Raiteri, C. M., Busso, M., Picchio, G., & Gallino, R. 1991, ApJ, 371, 665
gramme (grant agreement No 833925, project STAREX). Rauscher, T., Heger, A., Hoffman, R. D., & Woosley, S. E. 2002, ApJ, 576, 323
Rayet, M., Arnould, M., & Prantzos, N. 1990, A&A, 227, 271
Rayet, M., Eid, M. E., & Arnould, M. 1993, in Nuclei in the Cosmos, eds. F.
References Käppeler, & K. Wisshak (Bristol: Institute of Physics Publishing), 613
Arnett, W. 2001, Nucl. Phys. A, 688, 177 Rayet, M., Arnould, M., Hashimoto, M., Prantzos, N., & Nomoto, K. 1995,
Arnould, M. 1976, A&A, 46, 117 A&A, 298, 517
Arnould, M., & Goriely, S. 2003, Phys. Rep., 384, 1 Reichert, M., Obergaulinger, M., Eichler, M., Aloy, M. Á., & Arcones, A. 2021,
Arnould, M., & Goriely, S. 2020, Progr. Part. Nucl. Phys., 112, 103766 MNRAS, 501, 5733
Arnould, M., Goriely, S., & Takahashi, K. 2007, Phys. Rep., 450, 97 Siegel, D. M., Barnes, J., & Metzger, B. D. 2019, Nature, 569, 241
Banerjee, P., Heger, A., & Qian, Y.-Z. 2019, ApJ, 887, 187 Sieverding, A., Martinez-Pinedo, G., Huther, L., Langanke, K., & Heger, A.
Best, A., Görres, J., Couder, M., et al. 2011, Phys. Rev. C, 83, 052802 2018, ApJ, 865, 143
Best, A., Beard, M., Görres, J., et al. 2013, Phys. Rev. C, 87, 045805 Sukhbold, T., Ertl, T., Woosley, S. E., Brown, J. M., & Janka, H. T. 2016, ApJ,
Blake, J. B., & Schramm, D. N. 1976, ApJ, 209, 846 821, 38
Bliss, J., Arcones, A., & Qian, Y. Z. 2018, ApJ, 866, 105 Tachibana, T., Yamada, M., & Yoshida, Y. 1990, Prog. Theor. Phys., 84, 641
Caughlan, G. R., & Fowler, W. A. 1988, At. Data Nucl. Data Tab., 40, 283 Taggart, M. P., Akers, C., Laird, A. M., et al. 2019, Phys. Lett. B, 798, 134894
Chaboyer, B., & Zahn, J.-P. 1992, A&A, 253, 173 Talon, S., & Zahn, J.-P. 1997, A&A, 317, 749
Choplin, A., Ekström, S., Meynet, G., et al. 2017, A&A, 605, A63 The, L.-S., El Eid, M. F., & Meyer, B. S. 2007, ApJ, 655, 1058
Choplin, A., Hirschi, R., Meynet, G., et al. 2018, A&A, 618, A133 Thielemann, F.-K., Arnould, M., & Hillebrandt, W. 1979, A&A, 74, 175
Choplin, A., Tominaga, N., & Meyer, B. S. 2020, A&A, 639, A126 Thielemann, F.-K., Nomoto, K., & Hashimoto, M.-A. 1996, ApJ, 460, 408
Choplin, A., Siess, L., & Goriely, S. 2021, A&A, 648, A119 Tominaga, N. 2009, ApJ, 690, 526
Costa, V., Rayet, M., Zappalà, R. A., & Arnould, M. 2000, A&A, 358, L67 Tominaga, N., Maeda, K., Umeda, H., et al. 2007, ApJ, 657, L77
Cowan, J. J., & Rose, W. K. 1977, ApJ, 212, 149 Travaglio, C., Gallino, R., Rauscher, T., Röpke, F. K., & Hillebrandt, W. 2015,
Cristallo, S., Piersanti, L., Straniero, O., et al. 2011, ApJS, 197, 17 ApJ, 799, 54
Curtis, S., Ebinger, K., Fröhlich, C., et al. 2019, ApJ, 870, 2 Travaglio, C., Rauscher, T., Heger, A., Pignatari, M., & West, C. 2018, ApJ, 854,
de Jager, C., Nieuwenhuijzen, H., & van der Hucht, K. A. 1988, A&AS, 72, 259 18
Descouvemont, P. 1993, Phys. Rev. C, 48, 2746 Vink, J. S., de Koter, A., & Lamers, H. J. G. L. M. 2001, A&A, 369, 574
Ebinger, K., Curtis, S., Fröhlich, C., et al. 2019, ApJ, 870, 1 Wanajo, S., Sekiguchi, Y., Nishimura, N., et al. 2014, ApJ, 789, L39
Eggenberger, P., Meynet, G., Maeder, A., et al. 2008, Ap&SS, 316, 43 Winteler, C., Käppeli, R., Perego, A., et al. 2012, ApJ, 750, L22
Ekström, S., Georgy, C., Eggenberger, P., et al. 2012, A&A, 537, A146 Woosley, S. E., & Howard, W. M. 1978, ApJS, 36, 285
Ertl, T., Woosley, S. E., Sukhbold, T., & Janka, H. T. 2020, ApJ, 890, 51 Woosley, S. E., & Weaver, T. A. 1995, ApJS, 101, 181
Ferguson, J. W., Alexander, D. R., Allard, F., et al. 2005, ApJ, 623, 585 Xu, Y., Goriely, S., Jorissen, A., Chen, G., & Arnould, M. 2013, A&A, 549,
Frischknecht, U., Hirschi, R., & Thielemann, F.-K. 2012, A&A, 538, L2 A106
Frischknecht, U., Hirschi, R., Pignatari, M., et al. 2016, MNRAS, 456, 1803 Zahn, J.-P. 1992, A&A, 265, 115

A86, page 11 of 11