NuGrid stellar data set I. Stellar yields from H to Bi for stars with metallicities Z = 0.02 and Z = 0.01

We provide a set of stellar evolution and nucleosynthesis calculations that applies established physics assumptions simultaneously to low- and intermediate-mass and massive star models. Our goal is to provide an internally consistent and comprehensive nuclear production and yield data base for applications in areas such as pre-solar grain studies. Our non-rotating models assume convective boundary mixing where it has been adopted before. We include 8 (12) initial masses for $Z = 0.01$ ($0.02$). Models are followed either until the end of the asymptotic giant branch phase or the end of Si burning, complemented by a simple analytic core-collapse supernova models with two options for fallback and shock velocities. The explosions show which pre-supernova yields will most strongly be effected by the explosive nucleosynthesis. We discuss how these two explosion parameters impacts the light elements and the $s$ and $p$ process. For low- and intermediate-mass models our stellar yields from H to Bi include the effect of convective boundary mixing at the He-intershell boundaries and the stellar evolution feedback of the mixing process that produces the $^{13}$C pocket. All post-processing nucleosynthesis calculations use the same nuclear reaction rate network and nuclear physics input. We provide a discussion of the nuclear production across the entire mass range organized by element group. All our stellar nucleosynthesis profile and time evolution output is available electronically, and tools to explore the data on the NuGrid VOspace hosted by the Canadian Astronomical Data Centre are introduced.


Introduction
All the elements heavier than H can be formed in stars and their outbursts. Understanding the processes that have lead to the abundance distribution in the solar system is one of the fundamental goals of stellar nucleosynthesis and galactic astronomy. The solar system abundance distribution has been formed through nucleosynthesis in several generations of different stars. Despite significant progress, details regarding the chemical evolution of the Galaxy remain poorly understood (e.g., Tinsley 1980;Timmes et al. 1995;Goswami & Prantzos 2000;Travaglio et al. 2004;Gibson et al. 2006;Kobayashi et al. 2006). This makes the analysis of the solar abundances challenging, since not only stellar yields need to be known for different metallicities, but the respective contribution from different stellar sources depends on the dynamical evolution of the Galaxy. The analysis of spectroscopic observations of unevolved stars in the local disk of the Galaxy carries a similar degeneracy in the analysis of stellar nucleosynthesis. The observation of evolved low-and intermediate mass stars (e.g., Busso et al. 2001;García-Hernández et al. 2006;Hernandez et al. 2012;Abia et al. 2010Abia et al. , 2012 and of the ejecta of core-collapse supernova (CCSN) (e.g., Kjaer et al. 2010;Isensee et al. 2010Isensee et al. , 2012Hwang & Laming 2012) may provide information about intrinsic nucleosynthesis in these objects.
A closer source of information about stellar nucleosynthesis processes is hidden in primitive meteorites. Small particles of dust grains of presolar origin, which were produced in ancient stars that finished their life before the formation of our solar system, can be found preserved in meteorites (Lewis et al. 1987;Bernatowicz et al. 1987;Amari et al. 1990;Bernatowicz et al. 1991;Huss et al. 1994;Nittler et al. 1995;Choi et al. 1999). These carry a relatively unmodified nucleosynthesis signature from the environment of their parent stars (e.g., Zinner 2003;Clayton & Nittler 2004).
Stars with different initial masses and metallicities contribute to the production of elements.
-3 -Low-and intermediate-mass stars contribute over longer time scales to the chemical evolution of the interstellar medium than massive stars, firstly during the advanced hydrostatic phases via a stellar wind, then mainly in the asymptotic giant branch phase (AGB e.g., Iben & Renzini 1983;Busso et al. 1999; and possibly later as progenitors of Type Ia supernovae (SNIa, e.g., Nomoto 1984;Timmes et al. 1995;Hillebrandt & Niemeyer 2000;Domínguez et al. 2001;Thielemann et al. 2004;Travaglio et al. 2011;Pakmor et al. 2012;Seitenzahl et al. 2013;Hillebrandt et al. 2013). During the AGB phase, light elements like carbon, nitrogen and fluorine can be significantly produced, dependent on the initial mass range, in addition to heavy s-process elements (e.g., Herwig 2004;Karakas et al. 2010;Cristallo et al. 2011;Bisterzo et al. 2011). In particular, low-mass AGB stars are responsible for the production of the main s-process component in the solar system, explaining the s-process abundances between strontium and lead, and for the strong s-process component, mainly contributing to the solar lead inventory (e.g., Travaglio et al. 2001;Sneden et al. 2008).
Massive stars (M > 8 M ) provide the first contribution to the elemental chemical evolution, because of their short lifetimes. They produce metals during both their evolution and their final explosion. During their evolution, massive stars contribute to the chemical enrichment of the interstellar medium via stellar winds, mainly for light elements up to silicon (for instance H and He-burning products like carbon and nitrogen, see e.g., Meynet et al. 2006). Most α-elements up to the iron group are produced during the advanced evolutionary stages (e.g.,  and/or by the final CCSN (CCSN, e.g., Thielemann et al. 1996;Rauscher et al. 2002). Massive stars are also the main site for the weak s process(e.g., Käppeler et al. 2011). The weak s-process component (forming most of the s-process abundances in the solar system between iron and strontium, e.g., Travaglio et al. 2004) is produced during convective core He-burning and convective shell C-burning stages (e.g., Raiteri et al. 1991b,a;The et al. 2007;Pignatari et al. 2010). Since the s-process yields from massive stars are mostly ejected during the SN explosion, partial or more extreme modifications triggered by explosive nucleosynthesis need to be considered for these elements (e.g., Thielemann et al. 1996;Rauscher et al. 2002). One example is the classical p process (also known as the γ process) which forms proton-rich nuclei due to the photo-disintegration of s-process products in deep s-process-rich layers (Arnould & Goriely 2003).
The s process is responsible for about half of the abundances heavier than iron in the solar system. The r process is responsible for the production of most of the remaining part, with some relevant distinction between the predictions from the r-process residual method (e.g. Arlandini et al. 1999) and direct observations of elemental abundances of metal-poor, r-process-rich stars (Sneden et al. 2008). The astrophysical source of the r process is not clearly defined yet, besides the fact that it is related to CCSN events or merging of their remnants. Proposed scenarios are the neutrinoinduced winds from the CCSNe, before the formation of the reverse shock (e.g., Woosley et al. 1994;Wanajo et al. 2001;Farouqi et al. 2010) or after the fallback started (e.g., Fryer et al. 2006;Arcones et al. 2007); the polar jets from rotating magneto-hydrodynamical explosions of CCSN (Nishimura et al. 2006); the neutron-rich matter ejected from merging neutron stars (Freiburghaus et al. 1999) -4and neutron-star-black hole mergers (Surman et al. 2008). For a review of the different scenarios and recent r-process results see Thielemann et al. (2011); Winteler et al. (2012) and Korobkin et al. (2012).
Many applications in astronomy and meteoritics require stellar yield and nuclear production data. Presently, one may use for AGB stars the yields of  which are available for a suitable metallicity and mass coverage, but they are limited to the light elements. Heavy element predictions for elemental compositions based on the parameterized post-processing method are available from Bisterzo et al. (2010). s-process yields from stellar evolution models are available for a wide range of metallicities from the FRUITY database , however only for low-mass stars (M ≤ 3 M ). For super-AGB stars one may use the models of Siess (2010) which do not include heavy elements. Several choices are available of massive star yields (e.g. Chieffi & Limongi 2004;Nomoto et al. 2006). These different investigators have used different assumptions on stellar micro physics such as opacities and nuclear reaction rates, macro physics such as mixing assumptions and mass loss, as well as numerical implementation. Thus, yield tables stiched together from this range of sources do not only suffer from the inevitable uncertainty in many of the ingredients of such calculations, but also from a significant internal inconsistency, limiting the feedback that the confrontation of stellar models with observations can provide to the modeling.
The NuGrid research platform aims to address this issue by providing different sets of stellar yields to be used for Galactic Chemical Evolution (GCE), nuclear sensitivity and uncertainty studies, and direct comparison with stellar observations. Each set represents a comprehensive coverage of low-and intermediate-mass and massive star models for a given set of physics assumptions and using the same modeling codes for all masses. In this work, we present our first step toward achieving these goals. The first set of stellar models and their yields in the NuGrid production flow (Set 1) includes a grid of stellar masses from 1.65 to 25 M at 2 metallicities: Z = 0.02 and 0.01.
Massive star simulations include one-dimensional simplified CCSN models, to qualitatively study explosive nucleosynthesis. Other studies have adopted a more realistic approach, however the uncertainties and limits of our simulation capabilities remain a significant obstacle (see, e.g. discussion in Roberts et al. 2010). Here our goal is simply to provide an estimate of the explosive contribution to stellar yields, including a general understanding on how pre-explosive abundances are modified by the explosion (e.g., Limongi et al. 2000;Rauscher et al. 2002;Nomoto et al. 2006).
The stellar models combined represent the Stellar Evolution and Explosion (SEE) library and all of these models are then post-processed using mppnp to calculate the nucleosynthesis during the evolution of each model, which comprises the post-processing data (PPD) library. The SEE and PPD libraries associated with Set 1 and with this work are available online (see Appendix A).
Simulations for super-AGB stars (e.g., Siess 2007;Poelarends et al. 2008;Doherty et al. 2010;Ventura & D'Antona 2011), electron-capture SN (e.g., Nomoto 1984;Hoffman et al. 2008;Wanajo -5et al. 2009), SNIa (e.g., Hillebrandt & Niemeyer 2000) and r process (Thielemann et al. 2011) are not included in this work. The paper is organized as follows: in Section 2 the stellar codes and stellar explosion simulations are described and in Section 3 we present the post-processing calculations and the stellar yields of Set 1. Finally, in Section 4, we summarize the main conclusions of this work and discuss future prospects. Details regarding the physics assumption and published data can be found in Appendices A.2 and B.

Stellar Evolution Calculations
The stellar evolution models for Set 1 were calculated with two stellar evolution codes, MESA and GENEC. MESA (described in detail in Paxton et al. 2011), revision 3372, was used for low-and intermediate-mass stars. The GENEC code (Eggenberger et al. 2008;Bennett et al. 2012;Pignatari et al. 2013) was used for massive stars. In particular, GENEC is a well established research and production code for massive stars, but is not able to simulate in detail the AGB evolution and nucleosynthesis. On the other hand, MESA calculations showed quantitative consistent results with established AGB stellar codes (e.g., EVOL, Herwig 2004;Paxton et al. 2011). While MESA nonrotating massive star models provide quite comparable results with other codes, including GENEC (Paxton et al. 2011, the detailed analysis of the differences and the impact on the following nucleosynthesis still need to be fully considered. Set 1 includes models at two metallicities Z = 0.02 (Set 1.2) and Z = 0.01 (Set 1.1). Set 1.2 includes models with initial masses, M = 1. 65,2,3,5,15,20,25,32, 60 M and Set 1.1 includes models with initial masses, M = 1. 65,2,3,5,15,20,25 M . The main input physics used in the models is described below. Note that the models do not include the effects of rotation and magnetic fields.

Input Physics
GENEC massive star models were calculated with the same input physics as the MESA low-and intermediate-mass models when possible. The main differences in input physics between the two codes are described below (treatment of convective boundary mixing and prescriptions for mass loss).

Initial Composition and Opacities
The initial abundances are scaled to Z = 0.01 and Z = 0.02 from Grevesse & Noels (1993) and the isotopic percentage for each element is given by Lodders (2003). The initial composition corresponds directly to the OPAL Type 2 opacity tables used in both MESA and GENEC for the present work (Rogers et al. 1996). For lower temperatures, the corresponding opacities from Ferguson et al. -6 -(2005) are used.

Nuclear Reaction Network and Rates
In MESA, the agb.net nuclear reaction network is used, which includes the p-p chains, the CNO cycles, the triple-α reaction and the following α-capture reactions: 12 C(α, γ) 16 O, 14 N(α,γ) 56 Ni. Note that additional isotopes are included implicitly to follow the p-p chains, CNO tri-cycles and the combined (α,p)-(p,γ) reactions in the advanced stages.
In both codes, most reaction rates were taken from the NACRE (Angulo et al. 1999) reaction rate compilation for the experimental rates and from their website 15 for the theoretical rates. However, there are a few exceptions. In GENEC, the rate of Mukhamedzhanov et al. (2003) was used for 14 N(p ,γ) 15 O below 0.1 GK and the lower limit NACRE rate was used for temperatures above 0.1 GK. This combined rate is very similar to the more recent LUNA rate (Imbriani et al. 2004) at relevant temperatures, which was used in MESA. In both codes, the Fynbo et al. (2005) rate was used for the triple-α reaction and the Kunz et al. (2002) rate was used for 12 C(α, γ) 16 O. In GENEC, the 22 Ne(α, n) 25 Mg rate was taken from Jaeger et al. (2001) and used for T ≤ 1 GK. The NACRE rate was used for higher temperatures. The 22 Ne(α, n) 25 Mg rate competes with 22 Ne(α, γ) 26 Mg, where the NACRE rate was used. The key reaction rates responsible for the energy generation are therefore the same for the high-(GENEC) and intermediate-and low-mass (MESA) stellar models.

Mass Loss
For low-and intermediate-mass star models, we adopted in MESA the Reimers mass loss formula (Reimers 1975) with η R = 0.5 for the RGB phase. For the AGB phase we used the mass loss formula by Blöcker (1995) with η B = 0.01 for the O-rich phase. After the thermal pulse at which the surface C/O ratio exceeds ∼ 1.15, because of the third dredge-up, we increase the mass loss to η B = 0.04 for the 1.65 and 2 M tracks and to η B = 0.08 for the 3 M tracks. This choice is motivated by observational constraints, as for example the maximum level of C enhancement seen in Crich stars and planetary nebulae , as well as by recent hydrodynamics simulations investigating mass loss rates in C-rich giants (e.g., Mattsson et al. 2010;Mattsson & Höfner 2011) and preliminary evaluations of how they would influence AGB stellar evolution (Mattsson et al. ,. This choice is also motivated by stellar observations, including counts of C-and O-rich stars in the Magellanic Clouds (e.g. Marigo & Girardi 2007), which together indicate that the C-rich phase cannot last for more than at most a dozen thermal pulses. The 5 M tracks are dominated by hot-bottom burning and do not become C-rich for most of their TP-AGB evolution. We adopt η B = 0.05 from the beginning of the AGB phase for the two tracks with this mass.
For massive star models, several mass loss rates are used depending on the effective temperature, T eff , and the evolutionary stage of the star in GENEC. For main sequence massive stars, where log T eff > 3.9, mass loss rates are taken from Vink et al. (2001). Otherwise the rates are taken from de Jager et al. (1988). However, for lower temperatures (log T eff < 3.7), a scaling law of the formṀ is used, whereṀ is the mass loss rate in M yr −1 , L is the stellar luminosity. During the Wolf-Rayet (W-R) phase, mass loss rates by Nugis & Lamers (2000) are used.

Convective boundary mixing
The MESA code allows for the exponential diffusive convective boundary mixing (CBM) or overshooting introduced by Herwig et al. (1997) based on hydrodynamic simulations by Freytag et al. (1996). More recent hydrodynamic simulations of He-shell flash convection zone also show convection-induced mixing at convective boundaries (Herwig et al. 2007;Herwig et al. 2006). However, the nature of the instabilities observed in the deep interior is different then the Rayleigh-Taylor type overshooting situation found in shallow surface convection studies by Freytag et al. (1996). We therefore refer to our exponential overshooting method rather as a CBM algorithm. An efficiency of f ov = 0.014 was used at all boundaries, except during the dredge-up, when f DUP = 0.126 was used to generate a 13 C-pocket for the s-process according to Herwig et al. (2003a), and f PDCZ = 0.008 (where PDCZ stands for pulse-driven convective zone) was used at the bottom of the He-shell flash convection zone (Herwig et al. 2007;Werner & Herwig 2006;Karakas et al. 2010). The core overshooting value for the 1.65 M case is 1/2 of the value appropriate for higher masses, as motivated by the investigation of VandenBerg et al. (2006) using star cluster data on low-mass stars.
The Schwarzschild criterion for convection is used. In GENEC convective mixing is treated as instantaneous from hydrogen up to neon burning. In MESA and in GENEC from oxygen burning onwards (since the evolutionary timescale is becoming too small to justify the instantaneous mixing assumption) convective mixing is treated as a diffusive process. In GENEC overshooting is only included for hydrogen-and helium-burning cores, where an overshooting parameter of α = 0.2H P is used as in previous non-rotating models grids (Schaller et al. 1992).

Additional MESA Code Information
The low-and intermediate-mass models (1.65, 2, 3 and 5 M ) have been calculated with the MESA code (rev. 3372), for which a comprehensive code description and comparison (including GENEC for massive stars) is provided by Paxton et al. (2011). Concerning AGB stellar evolution, MESA has been compared in detail to results obtained with the EVOL stellar evolution code (e.g. Blöcker 1995;Herwig 2000Herwig , 2004. In particular, the 2 M , Z = 0.01 MESA AGB stellar model has been compared in detail to the corresponding track of Herwig & Austin (2004), which shares extremely similar features, e.g., in terms of occurrence and efficiency of third dredge-up as well as subsequent C-star formation. Intermediate-mass MESA models of correspondingly lower metallicity have been confirmed to be very similar to the EVOL stellar evolution tracks published by Herwig (2004).
We used the following settings in MESA: • structure, nuclear burning and time-dependent mixing operators were always solved together using a joint operator method; • in addition to the default MESA mesh refinement, we refined on H, 4 He, 13 C and 14 N also, to resolve the 13 C pocket during the entire interpulse time. This is needed to accurately follow s-process nucleosynthesis; • the mixing-length parameter used is 1.73H p , as calibrated for a solar model; • additional time step controls are used to allow for sufficient resolution of the He-shell flashes as well as the evolution of the thin H-burning shell during the interpulse evolution; • the OPAL Type 2 opacity tables (Rogers et al. 1996), and • the atmosphere option simple photosphere.

Stellar evolution tracks
The H-R diagram for low mass and intermediate mass tracks is given in Fig. 1, and the evolution of central temperature and density in Fig. 2. In Fig. 3 we also show, as an example, the Kippenhahn diagram for the 3 M , Z = 0.02 model. The final core masses for all low mass and intermediate mass star models are shown in Table 1. Convective boundary mixing during the thermal pulse phase is important for nucleosynthesis in two locations, the bottom of the He-shell flash convection zone during the thermal pulse and the bottom of the convective envelope during the third dredge-up phase. It also influences the efficiency of the third dredge-up which is responsible for mixing C and O from the intershell to the surface, which eventually is responsible for the formation of C-stars (Fig. 4). Many investigations of the s-process (e.g., Goriely & Mowlavi 2000;Busso et al. 1999) have shown that the models agree with many abundance observables if at the -9end of the third dredge-up a partial mixing zone of H and 12 C leaves behind the conditions for the formation of a 13 C-enriched layer (Fig. 5) that can subsequently release neutrons mostly under radiative conditions during the interpulse phase. In our low-mass AGB stellar models we achieve this partial mixing zone through the exponential CBM algorithm (cf. Section 2.1).
Full details regarding the Set 1.2 massive stars can be found in Bennett et al. (2012). Here the stellar evolution data is extended to include Set 1.1 models. The H-R diagram for all models in Set 1 are shown in Fig. 6 and the evolutionary tracks in the T c -ρ c plane are shown in Fig. 7 and 8, which are consistent with previous results (see e.g., Hirschi et al. 2004). In particular, models with masses M ≤ 25 M end as red super giants (RSGs), and the Set 1.2 32 and 60 M models end as W-R stars. In Fig. 9 and 10 the Kippenhahn diagrams of massive stars are shown. The final core masses of these models are comparable to other grids of models calculated with GENEC (Hirschi et al. 2004;Ekström et al. 2012). The choice of 0.2H p for core H-and He-burning overshooting implies that core masses are slightly larger than in the other grids using 0.1H p for core overshooting. The final masses at both Z = 0.01 and 0.02 are typically lower than the models obtained using other stellar evolution codes, due to the mass loss prescriptions used in the RSG phase, which is based on observational contrains (see §2.1).
The core masses for all of the massive star models are shown in Table 2. The core masses are determined at the end of silicon burning and are defined as the position in mass coordinate where a criterion for the core mass is satisfied. The helium-core mass, M 75% α , is defined by the mass coordinate where 4 He abundance becomes lower than 0.75 in mass (note that the 32 and 60 M stars become W-R stars and have lost their entire H-rich envelope). For the CO-core mass, M CO , the position corresponds to the mass coordinate where the 4 He abundance falls below 0.001 toward the center of the star. For the silicon-core mass, M Si , the position corresponds to a mass coordinate where the sum of Si, S, Ar, Ca and Ti mass fraction abundances, for all isotopes, is 0.5. The coreburning lifetimes for hydrostatic-burning stages are presented in Table 3 for the Set 1.2 and Set 1.1 massive star models. The lifetimes are defined for each stage as the difference in age from the point where the principal fuel for that stage ( 1 H for hydrogen burning, 4 He for helium burning, etc.) is depleted by 0.3% from its maximum value to the age where the mass fraction of that fuel depletes below a mass fraction of 10 −5 . However, there are exceptions for carbon burning and neon burning, where this value is 10 −3 , and oxygen burning, where this value is 10 −2 . These criteria are necessary to ensure that a lifetime is calculated in those cases where residual fuel is unburnt and to ensure that the burning stages are correctly separated (for example, the mass fraction abundance of 12 C at neon ignition for the Set 1.2 60 M model is 4.123 × 10 −5 ). The lifetime of the advanced stages is relatively sensitive to the mass fractions of isotopes defining the lifetime, particularly for stages following carbon burning.

The approximations of CCSN explosion
Particularly for the light elements carbon and nitrogen, winds play a role dispersing nuclides into the circumstellar medium. But the bulk of the nucleosynthetic yields from massive stars are ejected by the supernova explosion. Especially in the deeper layers (e.g., silicon and oxygen), the supernova shock drives further nuclear burning, but the ultimate yield is complex (e.g., Chieffi et al. 1998;Limongi et al. 2000;Woosley et al. 2002;Nomoto et al. 2006;Tominaga et al. 2007;Thielemann et al. 2011) and specific discussion is needed for different species (see for example Rauscher et al. 2002;Tur et al. 2009). In this paper, our stellar models follow the evolution of the star through silicon burning, but not to collapse. Instead of forcing a collapse, we model the explosive nucleosynthesis by a semi-analytic description for the shock heating and subsequent evolution of the matter to produce a qualitative picture of explosive nuclear burning.
The first step in our semi-analytic prescription is the determination of the mass-cut defining the line between matter ejected and matter falling back onto the compact remnant . We use the prescription outlined in Fryer et al. (2012) for the final remnant mass as a function of mass and metallicity (Table 4). In agreement with Fryer et al. (2012), two explosion models are considered for each SN model, labeled as delayed and rapid. Based on the convective engine, the explosion time is related to the explosion energy and different explosion energies produce different remnant masses. We include the two models here to give a range of remnant masses. In general, the rapid explosion produces smaller remnant masses than the delayed explosion. For more massive stars, the rapid explosion model fails, producing large remnants. Comparing our remnant masses to the core masses in Table 2, we note that a direct correspondence between core mass and remnant mass does not exist with the new remnant-mass prescription in Fryer et al. (2012) that includes both supernova engine and fallback effects. Beyond the mass cut, our stellar structure is in very good agreement with pre-collapse stellar models (Limongi & Chieffi 2006;Woosley et al. 2002;Young & Fryer 2007). Hence, our semi-analytic prescription for the shock will produce the same yield with a pre-collapse star as it does with our end-of-silicon-burning models.
For our shock, we determine the shock velocity using the Sedov blastwave solution (Sedov 1946) through the stellar structure. The density and temperature of each zone are assumed to spike suddenly following the shock jump conditions in the strong shock limit (Chevalier 1989). Pressure (P ) is given by where γ is the pre-shock adiabatic index determined from our stellar models, ρ is the pre-shock density, and v shock is the shock velocity. After being shocked, the pressure is radiation dominated, allowing us to calculate the post-shock temperature (T shock ), where a is the radiation constant. The post-shock density (ρ shock ) is given by -11 -After the material is shocked to its peak explosive temperature and density, it cools. For these models, we use a variant of the adiabatic exponential decay , and where t is the time after the the material is shocked, τ = 446/ρ 1/2 shock s, and ρ shock is the post-shock density in g cm −3 .
The details of the explosion for our Set 1.2 model with a delayed explosion are shown in Fig. 11. The lower mass cut is determined using the prescription in Fryer et al. (2012). Beyond this mass cut, there is no difference between our implementation of the rapid and delayed explosions (we implement the same shock velocities). In this manner, our delayed/rapid comparisons highlight the effect of the mass cut on the yield. We use an initial velocity of 2×10 9 cm s −1 for this standard model (on par with reasonably strong velocities at the launch of a shock in core-collapse calculations). For comparison, the explosion characteristics for a model assuming an initial shock velocity of 5 × 10 8 cm s −1 is also shown in Fig. 11.
The strong shocks in our standard model produce higher shock temperatures than common onedimensional models of CCSN (e.g., , affecting the explosive nucleosynthesis. In particular, because of the higher temperatures present nucleosynthesis calculations may show many similarities with hypernovae or the high energetic components of asymmetric supernovae (e.g., Nomoto et al. 2009). At the elemental boundary layers, the shock can accelerate a small amount of material to high velocities as it travels down the density gradient. In most explosion calculations (Young & Fryer 2007), viscous forces limit this acceleration, and we artificially cap our maximum velocity to v shock = 5 × 10 9 cm s −1 .
With these analytic explosion models, we are able to understand the trends in explosive burning. To compare in detail post-explosive and pre-explosive abundances, we refer to the production factors presented in Section 3, and to the complete yields tables provided online.

Nucleosynthesis code and calculated data
The nucleosynthesis simulations presented in this paper are provided by the multizone frame mppnp of the NuGrid post-processing code. A detailed description of the code and the postprocessing method is available in Appendix A.
Thermodynamic and structure information regarding the stellar models and SN explosion simulations are described in Section 2. The nuclear network size increases dynamically as needed, up to -12a limit of 5234 isotopes during the SN explosion with 74313 reactions. The NuGrid physics package uses nuclear data from a wide range of sources, including the major nuclear physics compilations and many other individual rates (Section A.2, Herwig et al. 2008b). As explained in Section A the post-processing code must adopt the same rates as the underlying stellar evolution calculations for charged particle reactions relevant for energy generation (Section 2). These include triple-α and 12 C(α,γ) 16 O reactions from Fynbo et al. (2005) and Kunz et al. (2002), respectively, as well as the 14 N(p,γ) 15 O reaction (Imbriani et al. 2005). The neutron source reaction 13 C(α,n) 16 O is taken from Heil et al. (2008) and the competing 22 Ne(α,n) 25 Mg and 22 Ne(α,γ) 26 Mg reactions are taken from Jaeger et al. (2001) and Angulo et al. (1999), respectively. Experimental neutron capture reaction rates are taken, when available, from the KADoNIS compilation (Dillmann et al. 2006). For neutron capture rates not included in KADoNIS, we adopt data from the Basel REACLIB database, revision 20090121 (Rauscher & Thielemann 2000). The β−decay rates are from Oda et al. (1994) or Fuller et al. (1985) for light species and from Langanke & Martínez-Pinedo (2000) and Aikawa et al. (2005) for the iron group and for species heavier than iron. Exceptions are the isomers of 26 Al, 85 Kr, 115 Cd, 176 Lu, and 180 Ta. For isomers below the thermalization temperature the isomeric state and the ground state are considered as separate species and terrestrial β−decay rates are used (e.g., Ward et al. 1976).
In Table 5 the isotopic overproduction factors-the final products normalized to their initial abundances-are given for Set 1.2. In Tables 6 and 7 the pre-explosive and explosive overproduction factors are given for massive stars at the same metallicity. The overproduction factors, OP im , for a given model of initial mass, M , for element/isotope i is given by where EM im is the total ejected mass of element/isotope i, M ini is the initial mass of the model, and X 0 i is the initial mass fraction of element/isotope i.
The total ejected masses include the contributions from both stellar winds and the SN explosion for massive stars and from the wind for low-and intermediate-mass stars.
The wind contribution is given by: where τ (m) is the final age of the star,Ṁ (m, t) is the mass loss rate, X S i is the surface mass-fraction abundance.
The SN contribution is given by: where m τ is the total mass of the star at τ (m), M rem,m is the remnant mass and X i (m r ) is the mass fraction abundance of element/isotope i at mass coordinate m r .
-13 - The same data are given in Tables 8, 9, and 10 for the elemental abundances. The radiogenic contribution is included. Similar information is provided for Set 1.1 in Tables 11, 12 and 13 for  isotopes, and in Tables 14, 15, and 16 for elements, respectively. Complete tables are provided online, together with the analogous production factors and stellar yields in form of ejected masses (given in solar masses; see for example Bennett et al. 2012, for details).
The analysis of nucleosynthesis in one-dimensional explosions provides fundamental information to understand how species are formed or modified in these conditions (e.g., . We note again that the SN explosions are treated in a simplistic way (see Section 2.3). The primary goal of the SN yields calculations is to estimate which elements and isotopes would be strongly affected by explosive nucleosynthesis. An overview of this information is available in Fig. 12. Our simple explosions feature explosion temperatures larger than usual CCSN, and our models therefore give some insight into the yields of such explosions. Complete tables with preexplosive and post-explosive abundances, as overproduction factors, production factors and yields in solar masses, as well as the thermodynamic histories from these models, are available online (Appendix B).
Based on our calculations we present in the following a brief discussion of the different element groups and their production in different mass regimes and evolution phases. We refer to e.g., Woosley et al. (1973); ; ; ; Thielemann et al. (1996); Chieffi et al. (1998); Limongi et al. (2000); Rauscher et al. (2002); Woosley et al. (2002) for similar analysis and discussion but focused only on massive stars, while we consider also the nucleosynthesis in low-mass and intermediate-mass stars. The discussion will follow the yield plots Fig. 13 to 18 for Set 1.2. Similar plots are available online for all stable isotopes and elements for both metallicities. The yield plots show the weighted stellar yields in the following sense. For each initial mass the ejected amount (during the wind as well as during the final SN or wind ejection as appropriate) in solar masses is weighed by a Salpeter IMF (α exponent = 2.35) sampled by non-uniform initial mass intervals, normalized to 1 M , and represented by a dashed black line in the yield plots. The initial mass intervals are chosen to represent initial masses with the similar nucleosynthetic production mechanisms compared to the available stellar evolution track. The dashed line corresponds to zero yields, i.e. to the return of the material that was present into the star from the initial abundance distribution. A yield line above or below the dashed line corresponds to production and destruction respectively. These plots therefore allow to compare the contribution from different masses through their production factors (the ratio of the yield line with the IMF line) as well as the relative importance of the contributing mass range (via the difference of the yield line and the IMF line) under the assumption that stars of all masses have enough time to return their winds and ejecta. While low-and intermediate mass stars eject all their yields during the wind phase (into which even a rapid superwind phase at the end is included), we distinguish for the massive stars between contributions from different processes. The wind yields are the ejecta returned during the pre-SN stellar evolution mass loss. The pre-SN contribution is an imaginary component that represents the ejecta that the SN would -14mechanically expel without any explosive nucleosynthesis. It is basically the integral of the to be ejected layers just before the explosion. For the explosive SN contribution including the explosive nucleosynthesis different options are shown, reflecting some of the uncertainties in modeling the explosions. Notice that in this work we do not include models representative for the mass range 7 − 11 M . In such a range there are super-AGB stars, electron-capture supernovae and the lowest mass iron-core collapse supernovae . Therefore, in Fig. 13 to 18 this mass range is shaded.
Notice also that the production of Li, Be, and B is not fully available in this release, since our stellar models miss some important physics processes that contribute to their their nucleosynthesis. Li production from intermediate mass stars through Hot Bottom Burning (HBB) during the AGB phase (initial mass higher than ∼ 4 M , e.g., Lattanzio & Forestini 1999)) is present in the 5 M models 16 . However a finer mass grid is required for a thorough characterization of HBB Li yields. Li may also be produced as a result of extra-mixing (the so-called cool bottom process) in AGB (and RGB) stars with lower initial masses (Sackmann & Boothroyd 1999;Nollett et al. 2003;Denissenkov & Merryfield 2011;Palmerini et al. 2011). Such non-standard mixing processes are not included in this model generation.
Production of Be and B in stars is mostly due to neutrino irradiation on 4 He and 12 C respectively, during CCSN (e.g., Woosley et al. 2002;Nakamura et al. 2010;Banerjee et al. 2013) and hypernovae (Fields et al. 2002). In the present stellar models we do not include neutrino nucleosynthesis.

C, N, and O
C is efficiently produced by low-mass and massive stars (e.g., Goswami & Prantzos 2000;Woosley et al. 2002) in He shell burning. In massive stars 12 C can originate from the portion of He-core ashes which is ejected by the SN explosion. A non-negligible contribution from Wolf-Rayet stars with masses larger than 25-30 M has been suggested in order to reproduce carbon abundances in the Galactic disk (e.g., Gustafsson et al. 1999). In low mass stars, 12 C comes from the triple-α reaction in the He-shell flash and is brought to the surface in the third dredge-up mixing following the thermal pulse (e.g., , and references therein).
In our calculations (Fig. 13, Tables 8 and 14 for wind contributions, Tables 16 and 10 for explosive contributions) the production factors of low-mass stars and massive stars are similar. The 12 C yields similar for both metallicities corresponding to the primary nature of C production. The weighted yield from massive stars is a factor of ∼ 5-10 lower than from the low-mass stars 16 Model predictions for Li have to be taken from the MESA profile output which was computed with coupled mixing and nuclear burning operators. The mppnp post-processing output employs an operator split which does not completely resolve the Cameron-Fowler transport mechanism with the present time stepping algorithm. regime, and comes mostly from (pre-)SN ejecta. Only the 60 M model has a dominant wind contribution, while the lesser massive star models are dominated by C formed during the pre-SN evolution and ejected in the explosion. An exception is the 25 M , Z = 0.01 case with rapid explosion, where the fall-back mass is larger compared to other models of the same mass and the amount of carbon ejected is insignificant. In general our models are confirming previous results that the overall production factor of carbon tends to increase with the initial stellar mass.
For low-mass stars the C production increases with the initial mass, peaking at 3 M model and then decreasing again for the 5 M models by a factor of ≈ 2 due to HBB (e.g., Lattanzio & Forestini 1999;Herwig 2004). We do not include possible effects due to binary evolution, which may reduce the C contribution from AGB stars (by ∼ 15 %, according to e.g., Tout et al. 1999).
N in the solar system is mostly produced by intermediate-mass stars. The amount of N from winds is similar to the SN explosion ejecta for the 25 M model (Tables 8 and 10) due to the enhanced mass loss efficiency for more massive stars. The N production only weakly depends on the SN explosion or the initial mass and is mostly located in the more external He-rich stellar regions that have not yet been processed by He burning 14 N is converted to 22 Ne in helium burning conditions (e.g., Peters 1968). In low-and intermediate-mass stars N lost by stellar winds increases with initial mass (Table 8). The higher nitrogen production in intermediate mass stars (i.e., 5 M ) is due to HBB (e.g, Lattanzio & Forestini 1999). Again, as with C, overall production factors of lowand intermediate-, and high-mass stars are similar but, in terms of weighted yields, intermediatemass stars dominate N production for both metallicities (Fig. 13).
After H and He, O is the most abundant element in the solar system. Most of it is considered to be produced in massive stars, and possibly from low-mass AGB stars, according to the O enrichment in the He intershell (Herwig 2000). Most of the O from massive stars is ejected by the SN explosion, but of pre-SN origin. Therefore, according to standard one-dimensional SN models, the amount of ejected oxygen increases with initial mass (see e.g., Thielemann et al. 1996). Our models take into account fallback and, as a result, the 20 M model ejects more 16 O than the 15 M and the 25 M cases (Table 10) Our low-mass AGB models are producing O due to the CBM applied at the bottom of the He-shell flash convection zone (see Section 2). O is then brought to the envelope along with C during the third dredge-up. This source of O may contribution to the total O inventory (Fig. 13).

F, Ne, and Na
F is produced in massive stars during CCSN, mostly via neutrino spallation on 20 Ne (e.g., Woosley & Haxton 1988), the Wolf-Rayet (WR) wind phase (Meynet & Maeder 2000), and low-mass AGB stars (e.g., Jorissen et al. 1992;Lugaro et al. 2004;Cristallo et al. 2007;Stancliffe et al. 2007;Karakas et al. 2008). No relevant contribution is expected from massive AGB stars, since HBB in the envelope destroys 19 F via proton capture (Smith et al. 2005;Karakas & Lattanzio 2007). F enhancement has been confirmed spectroscopically only in AGB stars (Abia et al. 2010;Lucatello et al. 2011), but chemical evolution studies seems to indicate that all the sources above are required in order to explain the abundance evolution of this element in the galaxy (Renda et al. 2004). Our simulations have no contributions from neutrino spallation during SNe or rotationally induced mixing and identify AGB stars with M ZAMS ≤ 3 M as the most productive source of F (Fig. 13). However, contributions from WR stars (see the 60 M star) or from CCSN (15 M case). All wind contributions for massive star Set 1.1 models are negative, and only the 15 M explosion leads overall to a small positive massive star production factor.
Our models (Fig. 14) confirm that Ne is produced as 20 Ne in massive stars. 20 Ne is efficiently produced already during the pre-explosive evolution of massive stars, mostly in C-burning layers. During the CCSN, 20 Ne in deeper ejecta is processed and destroyed by the SN shock wave, whereas more external parts of C-burning Ne-rich layers are ejected almost unchanged. Notice that some production of Ne is obtained at the bottom of the explosive He shell, depending on the SN shock temperatures. A similar effect can be be observed for the α-elements Mg, Si, S, Ar, and Ca. Due to similar high explosion temperatures, hypernova models or the high energy component of asymmetric CCSN explosion models show such a production for 28 Si (e.g., Nomoto et al. 2009). Those specific signatures identify a stellar region at the bottom of the He shell called C/Si zone, which provide a suitable location for carbide grains condensation in the ejecta. Furthermore, the existence of the C/Si zone may find some agreement with observations of CasA and SN1987A objects . 21 Ne is produced efficiently in massive AGB stars, via the proton capture on 20 Ne during HBB and in massive stars (Fig. 14). Finally, 22 Ne is mostly produced in low-mass AGB stars, where part of it may be primary based on third dredge-up of 12 C that is returned as 14 N to the next thermal pulse He-shell flash convection zone. It has an additional contribution from CCSN.
23 Na is mostly made in hydrostatic carbon burning in massive stars, just like 20 Ne Na is also partially destroyed by CCSN, compared to the pre-SN contribution (Fig. 14). Na may be ejected during the WR phase of more massive stars (e.g., the 32 and 60 M models) via proton capture on 22 N e . In the same way Na originates in both low-mass AGB and massive AGB stars (e.g., Cristallo et al. 2006;Lucatello et al. 2011).

Mg, Al, and Si
Our models (Fig. 14) confirm that most of the Mg is produced in massive stars. Individual Mg isotopes show a more complex behavior. 24 Mg is only produced in massive stars. In the 15 and 20 M , Z = 0.02 models 24 Mg is produced mostly during the pre-explosive phase, with a partial -17depletion due to nucleosynthesis during CCSN On the other hand, for larger masses, explosive nucleosynthesis provides an additional contribution to 24 Mg. The dependence on the initial mass is due to the large amount of material falling back on the SN remnant in the 25 and 32 M models, where most of the pre-explosive 24 Mg will not be ejected and the explosive He shell component dominates the final abundance. 25 Mg and 26 Mg are produced also in He-shell flash convection zone of AGB stars from more massive AGB stars due to α-capture reactions of 22 Ne (e.g., Karakas & Lattanzio 2007).
Al is efficiently produced in massive stars, mainly in C-burning zones, with no contribution from AGB stars (Fig. 13). 27 Al shares nuclear production conditions with 25 Mg and 26 Mg in the 15 and 20 M stars, and the same dependence as 24 Mg on the amount of material falling back after the SN explosion.
Si is also produced in massive stars (Fig. 16). However, here the origin from explosive nucleosynthesis component is always larger than the pre-explosive contribution. The 15 M case shows an increase of Si yield with decreasing explosion energy. In order to account for all of the Si inventory observed in the solar system, a contribution from SNIa (not considered here) is needed (e.g., Seitenzahl et al. 2013).

From P to Sc
Most P is made in massive stars (Fig. 16). The amount of 31 P made by the s process during the pre-explosive evolution is further increased during the SN explosion. The dominant contribution is given by explosive C-burning and explosive He-burning.
S is mainly composed of 32 S, while 36 S is the least abundant stable sulfur isotope (0.01% in the solar system). S comes from massive stars, with the exception of 36 S, which can have a small contribution from AGB stars (Fig. 16). Again, the contribution from SNIa (e.g., Thielemann et al. 2004) are not considered in our models. S is made during explosive C-burning and O-burning, whereas the pre-explosive production is marginal for 32 S (except for the Set 1.1 15 M rapid case), but more relevant for 33,34 S produced mainly via neutron captures on 32 S (Fig. 16). 36 S has a different origin from the weak s process(e.g., Woosley et al. 2002;Mauersberger et al. 2004, and references therein), mainly via the production channel 35 Cl(n,γ) 36 Cl(n,p) 36 S, where the initial 35 Cl is the main seed (Mauersberger et al. 2004;Pignatari et al. 2010). In our models 36 S is mainly produced in explosive C-and He-burning, in the latter case via direct neutron capture on 34 S.
Cl is made in the explosion of massive stars with a small pre-SN contribution for 37 Cl (Fig. 16). 35 Cl may also come from neutrino interactions with stellar material not considered here. The yields correlate in a non-linear way with the SN explosion energy. Comparing results for the 25 M star with lower masses shows that the yields strongly depend on fallback. The s process produces 37 Cl efficiently in massive stars (see also Woosley et al. 2002;Rauscher et al. 2002). In some cases (15 and 20 M models) explosive nucleosynthesis significantly increases the 37 Cl yield.
-18 -Ar is made in explosive O-burning (Fig. 15). Some pre-explosive production of 38 Ar is obtained for the 15 M model in the convective O-burning shell. Larger masses do not show such a component, since the O shell region is below the fallback coordinate. 40 Ar with a much smaller solar system abundance is efficiently produced by the s processin all models. An additional contribution originates in the explosive He-burning shell during the SN explosion due to the n process (Blake & Schramm 1976;Thielemann et al. 1979;Meyer et al. 2000).
K has 2 stable isotopes, 39,41 K, and a long-lived isotope, 40 K, decaying in part to 40 Ca and in part to 40 Ar. 41 K and 39 K are efficiently produced in CCSN, along with a small s-process production of 41 K during the pre-explosive phase (Fig. 16). A small production of 41 K in low-mass AGB stars may be relevant (electronic table, 3 M stellar model, Set 1.2). 40 K shows a strong production in AGB and massive stars. In agreement with the solar system distribution, 40 K stellar yields are about 2 orders of magnitude smaller than the total K yields. In massive stars, 40 K is made by the s process before the explosion and during the SN event by explosive He burning.
Most 40 Ca (and therefore most of the calcium) originates in explosive O-burning, with a minor contribution from models with an α-rich freezout component (Fig. 13). In particular, the large difference between the 32 M models with rapid and delayed explosion is due to the different amount of fallback material. 44 Ca can be efficiently produced as 44 Ti in α−rich freezout conditions (e.g., Magkotsios et al. 2010). A small amount of 44 Ca may also be produced in more external explosive regions, mainly as 44 Ti in explosive O-and C-burning or as 44 Ca and its neutron rich unstable isobars in explosive He-burning conditions. 46 Ca is the only Ca isotope with a clear contribution from AGB stars, in particular from massive AGB stars where high neutron densities during the TP allows the s-process path to open a branching at the unstable isotope 45 Ca. In a similar way 46 Ca can be produced by the s process in the convective C-burning shell in massive stars. The explosive contribution is mainly due to the n process in the explosive He-burning. 48 Ca originates in the n-process in massive stars with a small contribution from the 15 and 20 M stellar models (see full tables online), but weak compared to the similar 46 Ca production. 48 Ca may originate in special conditions in CCSN with a high neutron excess (Hartmann et al. 1985). Alternatively 48 Ca production is predicted in i-process-conditions with characteristic neutron densities of N n ∼ 10 15 cm −3 ).
Mono-isotopic Sc is among the least abundant elements in the solar system among light and intermediate elements. Because of its low abundance Sc can be efficiently produced from adjacent Ca at high neutron densities obtained in low mass stars (e.g., by the i process, Cowan & Rose 1977;Herwig et al. 2011). Besides the pre-explosive production by the s process, in massive stars we find a strong Sc production mainly in the explosive He-burning (Fig. 17). In the 15 M models with α-rich freezout Sc production is even larger (as previously reported, e.g., by Umeda & Nomoto 2005). Sc production may also be increased by neutrino-driven nucleosynthesis Yoshida et al. 2008).

From Ti to Ni
Ti is produced in CCSN and in SNIa (e.g., Rauscher et al. 2002;Seitenzahl et al. 2013). Most production comes from the mass range 15-20 M (Fig. 17). For larger masses part of the Ti-rich material is falling back. However, looking specifically at the production of individual isotopes of Ti, the situation is more complex. For example, 50 Ti is underproduced compared to the other Ti isotopes in several SN models (e.g., Thielemann et al. 1996). In our calculations, most of 50 Ti is made during the pre-explosive evolution by the s process in the convective He-burning core, in the following convective C-burning shell (e.g., Woosley et al. 2002;The et al. 2007) and via neutron captures during explosive He and C burning, which partially compensates for the destruction of 50 Ti at high temperatures deeper in the star. The difference in the final 50 Ti yields for the two 32 M explosion cases is due to the larger amount of fallback material in the delay model. Since recent SNIa models are not producing 50 Ti efficiently (e.g., Travaglio et al. 2011;Kusakabe et al. 2011), it is possible that most of the solar 50 Ti is made in massive stars. In principle, the final 50 Ti abundance in the SN ejecta would be a good indicator of the amount of fallback and explosion energy, keeping into account the uncertainties of its s-process production.
V is produced in massive stars during the CCSN. The contribution to the V inventory from SNIa is quite uncertain (Travaglio et al. 2011;Seitenzahl et al. 2013). 50 V does not receive a radiogenic contribution and therefore its abundance is a direct indicator of its production, which is mostly during explosive O-burning conditions The bulk of the 51 V is synthesized by the decay of 51 Cr and 51 Mn during freezout, both of which are produced in deeper regions and at higher temperature than 50 V in the explosion. Since most of 51 V is made in extreme conditions, its total abundance in the ejecta is severely reduced with increasing fallback (Fig. 17). Therefore V is underproduced in the 25 and 32 M models (SN rapid).
Cr is efficiently produced in massive stars (e.g., Woosley et al. 2002) and in SNIa (e.g., Thielemann et al. 2004;Seitenzahl et al. 2013). The most abundant Cr stable species ( 52,53 Cr) are made mostly as 52,53 Fe. Therefore, Cr is mostly produced in the 15-20 M star models, whereas for larger masses fallback is limiting the ejection of Cr-rich material (Fig. 17). 54 Cr (2.365 % of solar Cr) originates in the s process or via neutron capture in the explosive He-burning shell, and is destroyed in explosive O-or Si-burning conditions.
Mn is mostly produced during CCSN as 55 Co and 55 Fe. 55 Mn is efficiently produced only in the 15-20 M stars, whereas it is underproduced in higher mass models (Fig. 17) because the yield strongly decreases with increasing fallback efficiency. Mn production also shows a weak dependence on the explosion energy in the 15 M models.
The dominant Fe isotope 56 Fe is produced in CCSN and in SNIa mostly as 56 Ni (Fig. 18) Because of fallback, only the 15 M star efficiently produces 57 Fe, mostly as 57 Ni. Like 56 Fe 57 Fe also has a strong dependence on the explosion energy. Contrary to that the 54 Fe is increasing with decreasing energy in the 15 M models because 54 Fe is mostly made without radiogenic con--20tributions and for larger explosion energies 58 Ni is made instead of 54 Fe. 58 Fe is produced over the whole stellar range (Fig. 18). In massive stars, it is produced mostly by the s process during the pre-explosive phase and is partially destroyed by the SN explosion. In our models a relevant amount of 58 Fe is also produced in the explosive He-burning shell, via neutron captures. This contribution is particularly important for the lower-mass CCSN, like the 15 M case, where most of the pre-explosive abundances are strongly affected by the SN explosion. For the Z = 0.02 models AGB stars provide the largest contribution to the 58 Fe inventory, via thes process (Section 3.7).
Besides a small positive contribution to Co from the AGB star s process the strongest production happens in massive stars (Fig. 17). The 15 M SN models show a correlation of the Co yields with the explosion energy. In the most energetic SN models, most of 59 Co is made as 59 Cu, with a smaller contribution from 59 Ni, 59 Co itself and 59 Fe from the explosive He-burning shell. At lower explosion energy the 59 Cu and 59 Ni production is reduced. In this case 59 Co comes from direct production and from 59 Fe decay. This makes Co a possible nucleosynthesis signature of highly energetic SN (see e.g., Nomoto et al. 2009). For larger masses, where the fallback contribution in our models is stronger, the explosive contribution of the radiogenic 59 Fe in the He shell becomes more relevant for the Co final yields. For weaker fallback (e.g., the 25 M SN rapid model or the 60 M models) most Co originates comes from the s process.
The most abundant Ni species, 58,60 Ni, are produced efficiently in CCSN at high temperatures, with a strong contribution also from SNIa (e.g., Thielemann et al. 2004;Bravo et al. 2010). The production in massive stars depends on fallback and explosion energy. For example, the 25 M SN models do not efficiently contribute to the bulk of Ni inventory because of the strong fallback (Fig. 18). The other stable Ni isotopes, 61,62,64 Ni, can have a contribution from AGB stars. The 64 Ni yield from the s process in massive AGB stars is comparable to the massive star yield. For models with less fallback and with less energetic explosions, the 64 Ni abundance is mostly given by the pre-explosive s-process contribution. The explosive contribution via neutron capture in the explosive He shell becomes more relevant for models with large fallback and/or high SN energy, where 64 Ni is produced directly or via the radiogenic decay of neutron rich isobars of the lighter iron group elements (e.g., unstable 64 Co and 64 Fe, Section 3.7).

Trans-iron elements
Trans-iron elements are made during the quiescent stellar evolution by the s process (Meyer 1994;Busso et al. 1999). Our models contain contributions from AGB and massive stars. In addition there is a high-neutron density contribution during explosive nucleosynthesis according to our SN explosion approximation Section 2.3. Because the site of the rapid neutron capture process (r process) is still unknown our models do not contain any r-process component.
The mix of astrophysical sources of heavy elements in different astrophysical sites because of the range of time scales of different contributions to chemical evolution. Heavy elements in old -21stars (e.g., Sneden et al. 2008), the the solar system distribution (Asplund et al. 2009), or stellar abundances of field stars in the Milky Way or in other environments (e.g., for Dwarf Spheroidal Galaxies, Tolstoy et al. 2009, and references therein) of different contributions of heavy elements. Considering the isotopic distribution of nuclides in the solar system, approximately half of the abundances beyond iron are produced by the s process and the other half by the r process. A small contribution is provided by the so called p process (Arnould & Goriely 2003). The reproduction of the total s-process distribution in the solar system is then assigned to three different components: a component between iron and strontium (60 A 90), where most of the s-abundances are produced in massive stars (the weak s-process component, see for example Käppeler et al. 1989;Beer et al. 1992, and references therein); a component comprising mostly of nuclides with A 90, mostly produced in AGB stars (the main s-process component, see for example Arlandini et al. 1999;Bisterzo et al. 2011); and finally a component which forms approximately 50% of the solar 208 Pb, which is produced by low metallicity AGB stars (the strong s-process component, ).
Due to GCE simulations including s-process yields (e.g., Travaglio et al. 2004;Serminato et al. 2009), it is now known that the demarcation of the different components in atomic mass is not strict. For example, low-mass and massive AGB stars also contribute significantly to the inventory of s-process abundances below the Sr neutron magic peak. Furthermore, Travaglio et al. (2004) showed that according to present models the weak s-process component and the main s-process component do not fully reproduce the s-abundances between Sr and Ba, proposing the existence of another unknown s-process component called the lighter element primary process (LEPP). The nature of this component is still under debate and it may be related to a similar component observed in old metal-poor halo stars (e.g., Truran et al. 2002;Travaglio et al. 2004;Hansen & Primas 2011;Cescutti et al. 2013;Hansen et al. 2013), and in old Galactic bulge globular clusters (Chiappini et al. 2011). Despite the consistency between the two elemental signatures in the solar system and in the early Galaxy (Montes et al. 2007), it is still unclear if the LEPP is an s process and how many nucleosynthesis mechanisms contribute to this mass range. Different scenarios have been proposed as astrophysical sources of the LEPP abundances, among them the s process in fast rotating massive stars at low metallicity Frischknecht et al. 2012), the incomplete α−rich freezout in CCSN within the high-entropy neutrino winds produced during the formation of a neutron star (e.g., Hoffman et al. 1996;Kratz et al. 2008;Qian & Wasserburg 2008;Farouqi et al. 2010) and the νp process in proton-rich neutrino winds (Fröhlich et al. 2006;Arcones & Montes 2011).
We consider the s process nucleosynthesis in both massive stars and AGB stars. A detailed discussion of the isotopic distribution in the different models is beyond the scope of this work, but the production of typical s process elements in stars of different masses is discussed.

The s process in massive stars
In massive stars, the main neutron source for the s process is the 22 Ne(α,n) 25 Mg reaction (Peters 1968;Couch et al. 1974;Lamb et al. 1977). Depending on the initial mass of the star (e.g., Prantzos et al. 1990) and on the 22 Ne+α rates (e.g., Käppeler et al. 1994), some 22 Ne is left in the He-burning ashes, which is activated later in the subsequent C-burning conditions (e.g., Raiteri et al. 1991b). The elements produced most efficiently are copper, gallium and germanium (Pignatari et al. 2010, and references therein).
The pre-SN of the s-process elements in our models has been discussed in the context of an analysis of the 12 C+ 12 Cnuclear reaction rate uncertainty by Bennett et al. (2012) and Pignatari et al. (2013). Another contribution to s-process elements in our models comes from the SN explosion. The SN shock wave may partially deplete or change the original pre-explosive s-process abundances (e.g., Rauscher et al. 2002;Tur et al. 2009). In this case, the resulting explosive stellar yields would still be secondary, since the newly built nuclides are produced from secondary s-process seeds. However, the relevance of explosion feedback on the pre-explosive s-process signature depends on many details of the SN mechanism. In a 25 M star, the bulk of the pre-explosive s-process abundances lies in the convective C-burning shell and in the ashes of the He core material located between the C shell and the He shell (e.g., The et al. 2007;Pignatari et al. 2010, and references therein). For standard CCSN models, with a SN explosion energy in the order of ∼ 10 51 erg and a "masscut" located near to the bottom of the convective C shell, most of the 25 M star s-process-rich material is ejected unchanged by the explosion (e.g., Rauscher et al. 2002;Limongi et al. 2000). In the 25 M stellar models discussed here, most or all of the s-process-rich material falls back forming a BH (the star in this case ends as a failed SN, see e.g., Woosley et al. 2002, and references therein), according to Fryer et al. (2012). In particular, the central 5.71 M and 6.05 M is not ejected for the delay explosive calculations of Set 1.1 and Set 1.2, respectively. For the rapid explosive models, at Z = 0.02 no material is ejected (complete fallback), and for Z = 0.01 only the material external to the mass coordinate 7.91 M (see Table 4) is ejected. Furthermore, the remaining s-process-rich material will be significantly modified by the sudden increase of temperature and density related to the SN shock wave. Notice also that pre-explosive C shell material could be modified by shell merging in the last ∼ day before the core collapse starts. This does not happen in our simulations, but it has been obtained, for instance, in the 25 M star by Rauscher et al. (2002) and Tur et al. (2009).
In Fig. 19, we show the abundance profile before and after the SN shock wave, for the two s-only species 70 Ge and 76 Se, and we compare the 25 M and 60 M models, with Z = 0.02 and delay. In the 25 M model only about 0.3 M of the s-process-rich material from the convective C shell is ejected, including small modifications from the explosion. The s-process abundances are strongly modified in the He core window and at the bottom of the He shell by neutron captures, where stellar conditions and fuel are suitable to trigger explosive He-burning and the efficient neutron production by the 22 Ne(α,n) 25 Mg is possible. In the 60 M model, the amount of fall-back material is smaller than in the previous case (3 M , see Table 4). However, the pre-explosive 70 Ge -23and 76 Se made by the s process in the regions between about 3 and 6 M and between 10.5 M and the surface of the star are modified by photodisintegration during explosive O-and C-burning and by neutron captures due to explosive He burning respectively. The external part of the C shell material (between 6 M and 10.5 M ) is only modified slightly.
In Fig. 20, the final isotopic production factors are given for the same models discussed in Fig. 19. The abundance distributions are given compared to the 16 O production factor, which is mainly produced in massive stars in the same zones where the s-process yields are made. However, 16 O is a primary isotope and its yields therefore do not change with the initial metallicity of the star. Unlike primary isotopes, s-process yields in massive stars (or more in general heavy nuclides produced starting from s-process seeds) show a direct dependence on the initial stellar metal content, which is closer to a secondary-like nucleosynthesis (e.g., Tinsley 1980). According to Tinsley (1980), secondary-like isotopes produced in massive stars are expected to show an overabundance with a factor of 2 higher than 16 O at solar metallicity, to be mostly made by the weak s process. Concerning the 25 M star, fallback reduces the s-process and 16 O yields in a similar way. Therefore, the tendency to have abundances laying above the 16 O×2 line in the Cu-As region (see Fig. 20) is conserved, in agreement with models using different fallback treatment (e.g., Rauscher et al. 2002).
The footprint of the s process in producing different elements of the weak s-process component with different efficiencies is maintained in the final yields, beside the uncertainties related to the nucleosynthesis triggered by the SN explosion. For this reason, the abundances start decreasing in the Se region, and become marginal above the Sr-Y-Zr peak, in agreement with the s-process preexplosive distribution. Concerning the 60 M star, the larger yields between Fe and Nb compared to the 25 M star are due to a stronger activation of the 22 Ne(α,n) 25 Mg in the convective Heburning core (Prantzos et al. 1990). Indeed, the central He-burning temperature tends to increase with the initial mass of the star, leading to a more efficient s process in these conditions. Above the Sr neutron magic peak, where the pre-explosive contribution from the C shell and the He core window material is less significant, the explosive nucleosynthesis signature in different parts of the star (including from the explosive He-burning shell) becomes easier to identify in the total ejecta. For instance, the isotopic signature of Mo in both masses (but more in the 60 M star) shows a clear 95,97 Mo enrichment compared to other Mo isotopes, in agreement with the signature measured in SiC-X presolar grains (Meyer et al. 2000) due to the n-process activation (e.g., Blake & Schramm 1976;Thielemann et al. 1979).
In general, the present sets of CCSN models may be used to qualitatively study the impact of fallback and CCSN explosions with high shock velocities on the weak s-process distribution.

AGB contribution
The s-process species above the Sr peak are mostly produced in low mass AGB stars. The neutrons are mainly produced in radiative conditions in the so-called 13 C-pocket  via the 13 C(α,n) 16 O reaction. The AGB thermal pulses induce recurrent, interacting, mixing episodes, including mixing for the formation of the 13 C-pocket and the third dredge-up mixing that brings nuclear processed material from the core to the envelope , and references therein).
The efficiency of mixing-processed material from the core to the envelope is correlated with the dredge-up parameter where ∆M DU P is the dredged up mass and ∆M H is the hydrogen free core growth during the last interpulse phase. The calculated dredge-up parameter evolution is shown in Fig. 21. The parameter reflects the evolutionary behavior of the core and envelope mass, and it decreases for higher metallicities (Lattanzio 1989).
Properties of the 13 C-pocket can be obtained, for example, from comparison with isotopic information from pre-solar grains (Lugaro et al. 2003b). Rotation-induced mixing may have a prohibiting effect on the s process (Herwig et al. 2003a;Siess et al. 2004). It is therefore likely that a convection-induced instability, such as a combination of Kelvin-Helmholz instabilities and internal gravity wave mixing, will lead to convective boundary mixing that generates the 13 C-pocket. We model this mixing with the exponentially decaying diffusion algorithm (Section 2.1.4).
A smaller contribution comes from the partial activation of the 22 Ne(α,n) 25 Mg reaction in He-shell flash convection zone at T > ∼ 2.5 × 10 8 K and ρ ∼ 10 3 cm −3 with a higher neutron density ( 10 10 cm −3 ) for up to a few years. This exposure causes local isotopic shifts in the s-process distribution as evident in presolar grains (e.g., Pignatari et al. 2006;Lugaro et al. 2003a;Käppeler et al. 2011).

In massive AGB stars (intermediate mass stars) 22
Ne becomes the dominant neutron source and may produce elements that are usually assigned to the weak s-process component (Travaglio et al. 2004). However, stellar models can presently not explain a very high [Rb/Zr] observed in galactic and LMC massive AGB stars (García-Hernández et al. 2006van Raai et al. 2012).
-25 -In Fig. 22, we report for the models of Set 1.1 the [ls/Fe] surface evolution, where ls is an average of Sr, Y, and Zr, as a function of the s-process index [hs/ls] (Luck & Bond 1991). The term hs includes elements from the second neutron magic peak Ba, La, Nd, and Sm. The [ls/Fe] ratio indicates the efficiency in producing elements at the first neutron magic peak. The [hs/ls] ratio provides an indication of the average neutron exposure in the 13 C-pocket. The 1.65 M model (not reported in the figure) shows only a negligible s-process enrichment in the envelope. The 2 and 3 M stars show an [ls/Fe] lower than 0.4 dex. This enrichment is lower than the maximum observed in AGB stars, by about 1.0 dex (e.g, Busso et al. 2001).
In general, the size of a typical 13 C-pocket in the present models is 2 − 3 × 10 −5 M , similar to the value obtained by Lugaro et al. (2003b). According to the simple estimate made by Herwig et al. (2003b), the 13 C-pocket should be about 3-4 times larger in order to reproduce the largest [ls/F e] ∼ 1 observed at solar-like metallicity in MS-S stars ).
In Fig. 22, the [hs/ls] of low mass AGB models tends to become positive, in contrast to observations of AGB stars at metallicity close to solar. This has already been noticed and discussed by Lugaro et al. (2003b) and Herwig et al. (2003b) for models applying CBM at the bottom of the convective TP and is even more severe for models at Z = 0.01. Indeed, these models are characterized by a higher 12 C abundance in the He intershell, which in turn causes a larger neutron exposure in the 13 C-pocket. Therefore, independently from the total s-process enrichment in the AGB envelope, the larger 12 C concentration in the He intershell makes the production ratio between the different neutron magic peaks to increase favoring more the hs elements.
The [Rb/Sr] ratio is affected by the branching point at 85 Kr. It increases with the neutron density and therefore with the relative importance of the 22 Ne(α,n) 25 Mg reaction compared to the 13 C(α,n) 16 O reaction. Our AGB models in general show a mild negative [Rb/Sr] (Fig. 22), in agreement with observations (Lambert et al. 1995;Abia et al. 2001).
Another signature of AGB models including CBM at the bottom of the He-shell flash convection zone is the overproduction of 25 Mg and 26 Mg due to a higher temperature at the bottom of the convective TP and consequently a more efficient activation of α-captures on 22 Ne (Lugaro et al. 2003b). In our models the isotopic ratios 25 Mg/ 24 Mg and 26 Mg/ 24 Mg increase with the initial mass of the star (Fig. 23). For instance, for the 3 M model, they are 0.23 and 0.40 respectively. The 1.65 M case shows final isotopic ratios of 0.13 and 0.16. These latter values are more in agreement with observations in AGB stars at solar-like metallicity, which do not show any significant increase (e.g., Smith & Lambert 1986) According to Zr measurements in presolar mainstream SiC grains, 96 Zr is not efficiently produced in low mass AGB stars Zinner 2003). This is in agreement with the signature of the 2 M star (Fig. 23). However, the 3 M star shows a final 96 Zr/ 94 Zr higher than solar. The same trend was observed by (Lugaro et al. 2003b) for models with CBM due to an excessively large 22 Ne(α,n) 25 Mg efficiency during the TP (similar to the Mg isotopic ratios). Furthermore, for the present models, both the high [hs/ls] and the weak contribution from the 13 C--26pocket to the total neutron exposure affects the production of 94 Zr, increasing the final 96 Zr/ 94 Zr in the AGB envelope. In our models the 152 Gd/ 154 Gd ratio is lower than solar for low mass AGB models (Fig. 23). The anomalously high isotopic ratio observed by Lugaro et al. (2003b) (Abia et al. 2001). The high neutron density is also responsible for a large 96 Zr/ 94 Zr ratio, whereas the 152 Gd/ 154 Gd ratio, after an initial increase, tends to decrease to the solar ratio.
Generally, the present AGB stellar calculations confirm the main features of AGB models with the present CBM prescription described by Lugaro et al. (2003b) and Herwig et al. (2003b). They are able to reproduce the large C and O abundances observed in H-deficient stars. On the other hand, the 13 C-pocket in Set 1 is too small and the 22 Ne+α activation during the TP is too strong. At least the former issue can be solved by an alternative parameterization of the CBM mixing at the bottom of the convective envelope where the partial mixing zone of the 13 C-pocket originates.

The trans-iron p-process isotopes
In the solar system distribution, 35 stable nuclides 17 beyond iron have been identified to be produced by the classical p process (or γ process, e.g., Woosley & Howard 1978), which is driven by chains of photodisintegration reactions on pre-existing local and heavier nuclides (e.g., Arnould & Goriely 2003, and references therein). At present, the most well-established astrophysical site for p-process nucleosynthesis is the O/Ne-rich layers of massive stars, during the CCSN explosion (e.g., Arnould 1976;Woosley & Howard 1978). With the relevant exception of 92,94 Mo and 96,98 Ru (14.84, 9.25% and 5.52, 1.88% of the Mo and Ru solar abundance, respectively), the abundances of p-process nuclei are 2-3 orders of magnitude lower than other stable nuclides. Such isotopes were defined as p-only and, by definition, cannot receive a significant contribution from other processes such as the s process or the r process.
152 Gd and and 164 Er receive a dominant s-process contribution from low-mass AGB stars (Bisterzo et al. 2011) and 113 In and 115 Sn are likely not of p-process origin, even if their origin is still unclear (see Dillmann et al. 2008, and references therein). Therefore, they cannot be indicated as p-only nuclides. Furthermore, 138 La and 180 Ta could not be produced only by the p process. Indeed, 138 La might receive a significant contribution from neutrino capture on 138 Ba (Goriely & Siess 2001) and the long-lived 180 Ta isomer (half-life larger than 1.2×10 15 yr, Cumming & Alburger 1985) may be efficiently produced by the s process in low-mass AGB stars (see for different and -27controversial predictions Arlandini et al. 1999;Goriely & Mowlavi 2000;Bisterzo et al. 2011) and in massive stars (e.g., Rauscher et al. 2002). Finally, in all CCSN calculations using realistic massive stars models, 92,94 Mo and 96,98 Ru are systematically underproduced by more than an order of magnitude compared to the other p-process species (Arnould & Goriely 2003), taking into account present nuclear uncertainties (Rapp et al. 2006;Rauscher 2006, and references therein). Recently, Pignatari et al. (2013) showed that assuming an enhanced (compared to Caughlan & Fowler 1988) 12 C+ 12 C fusion reaction rate may lead to a Mo and Ru p-nuclide production up to the level of other p-nuclei (cp-component).
Besides problems in reproducing single isotopes, the average p-process massive star yields are underproduced by about a factor of three compared to the amount required to explain the solar system distribution (e.g., Rayet et al. 1995), or the secondary nature of the classical p process (see Pignatari et al. 2013). Alternative astrophysical sources have been proposed that reproduce, at least in part, to the abundances of p-process nuclides in the solar system (see for example from SNIa ejecta, Howard et al. 1991;Howard & Meyer 1993;Travaglio et al. 2011;Kusakabe et al. 2011) or for the abundances of nuclides in the Mo-Ru region only (in α-rich freezout conditions during the CCSN explosion, or the νp process in proton-rich neutrino-wind conditions, e.g., Hoffman et al. 1996;Fröhlich et al. 2006;Farouqi et al. 2010). The most promising scenario may be SNIa (Travaglio et al. 2011).
Our models represent the p-process contribution in CCSN with high shock velocities and including fallback. Similar results are expected for p-process yields from hypernova or from high energy component of asymmetric CCSN. Among all the CCSN models presented in this work, we discuss the p-process distribution of a 15 M star and a 25 M star, Z = 0.02 (SN model delay, Set 1.2). In Fig. 25, left panel, the 15 M star shows the strongest explosive contribution at the lightest p-process species 74 Se, 78 Kr and 84 Sr, with a production factor 10. No p-process contribution is seen in the Mo-Ru region, in agreement with standard CCSN calculations (e.g., Rauscher et al. 2002). In this specific model, there is no relevant production of 102 Pd either. The p-process contribution becomes positive again from 106,108 Cd to 196 Hg (with a production factor of ∼ 1.5-4), with the exception of 156,158 Dy and 190 Pt, which are not efficiently produced. Among those species, the most produced are 180 Ta and 180 W, with a production factor of about 4. The high SN explosion energy causes a larger contribution to the lightest p-process species, in disagreement with the classical flat p-process distribution. However, the oxygen production factor of this model is reduced to about 2.1, since the high energy of the explosion depletes O in a large region of the ejecta. Therefore, the p-process production factor in CCSN characterized by high shock velocities and/or high explosion energies is similar or larger than O, one of the fundamental requirements in order to reproduce the solar system p-process abundances.
In Fig. 25, right panel, the 25 M star shows a dominant p-process signature starting from Ba. Indeed, besides 108 Cd (with a mild production of ∼ 1.5) and 126 Te (∼ 1.4), p-nuclides lighter than Ba are not ejected. Above Ba, the production factors range between ∼ 1.1 ( 144 Sm) and 9 ( 196 Hg). The reason for this behavior is that the hotter material carrying the lighter p-nuclides falls back -28onto the forming BH and only the colder p-process component is ejected. The oxygen production factor is about 3.4. Compared to standard CCSN models (e.g., Rauscher et al. 2002), the oxygen yields are also smaller. This is due to the large amount of mass falling back (for this model the central 5.7 M are not ejected).
In summary, we have seen that the 15 M model with Z = 0.02 shows a stronger p-process contribution at the lightest nuclei, but with no production of Mo-Ru p-only isotopes. Notice that the 20 M model with the same metallicity and the same explosion energy shows a more standard p-process distribution (see Fig. 12). Therefore, the dependence on the initial stellar mass should be considered before comparing quantitatively the p-process yields with the solar system distribution. On the other hand, a strong fallback (see the 25 M star case discussed here) potentially favors heavier p-process ejecta. In both of the two cases discussed here, however, the oxygen production tends to be reduced compared to standard SN models of the same mass.

Summary and final remarks
In this work we present a set of stellar models and their chemical yields (Set 1). We define 1.65, 2, 3, 5, 15, 20, 25 M models. We also calculated 32 and 60 M models at Z = 0.02. Massive star models are calculated using the stellar evolution code GENEC and lower mass models are calculated using MESA. For low-and intermediate-mass stars, wind yields are provided in the form of production factors and absolute yields in solar mass units. For massive stars, the yields are given for the stellar wind, pre-explosive and post-explosive contributions. Two sets of explosion models are considered, each with a different fallback prescription. The NuGrid post-processing code mppnp is used to perform all nucleosynthesis calculations, for AGB stars, massive stars as well as the SN explosions.
Core collapse SN models are performed in 1D semi-analytic way. The shock velocity profiles and fallback prescriptions used are motivated by multi-dimensional hydrodynamic simulations. Due to their simplified nature they are foremost meant to indicate species that will be affected by explosive nucleosynthesis in any significant way, rather than provide the basis of detailed quantitative analysis of explosive yields for complex studies such as galactic chemical evolution. However, the explosive yields nevertheless provide important insights on the main features of explosive nucleosynthesis. Furthermore, the Set 1 SN models represent an example of explosive nucleosynthesis at high shock velocity (high temperature or high energy), and with a fallback signature.
The 16 O isotope is the most abundant product of massive stars. Considering both the CCSN ejecta and the winds (see the yields available online), the models 15, 20 and 25 M and Z = 0.02, delay explosion, produce 0.30, 1.27 and 0.82 M of 16 O. For instance, for the same masses and metallicity Thielemann et al. (1996) provides 0.42, 1.48 and2.99 M , Rauscher et al. (2002) 0.85, 2.20 and 3.32 M (models S15, S20 and S25), Chieffi & Limongi (2004) 0.52, 1.38 and 2.44 M . For the 15 M star, the results change by almost a factor of three, and by a factor of 1.7 for the 20 M star. The large fallback included in our simulations causes lower 16 O yield for the 25 M star, which is e.g., about a factor of four smaller than Rauscher et al. (2002). Differences can be even larger if we compare the yields of 44 Ti and 56 Ni, which critically depend on the explosion parameters applied in the simulations. For these two species, we obtain for the same models considered before 1.62×10 −4 and 0.33 M , 1.44×10 −5 and 0.0087 M , 1.05×10 −7 M and no 56 Ni ejected, respectively. The extended fallback in the 25 M model does not allow to eject any relevant amount of 44 Ti or 56 Ni. For the same models and isotopes, Thielemann et al. (1996)  For the first time we present a grid of full yields for s-process and p-process species for SN models with strong shocks, and with consequent explosion temperatures larger than standard CCSN. In particular, models with a large fallback have reduced s-process yields, which are modified significantly in models with higher explosion energies. For most cases the s-process distribution is affected by local abundance redistribution. In particular, in the He shell the n-process from explosive He burning may have a relevant impact. The weight of the n-process component on the final yields increases with increasing fallback. The final yield of the neutron-magic 88 Sr is a good indicator of the neutron capture efficiency in our massive star models. In particular, the models 15, 20 and 25 M and Z = 0.02, delay explosion, produce 6.78×10 −7 , 2.38×10 −6 and 1.54×10 −6 M of 88 Sr. For comparison, Rauscher et al. (2002) predicts respectively 1.08×10 −6 , 4.69×10 −6 and 1.14×10 −5 M , and Chieffi & Limongi (2004) 7.88×10 −7 , 1.98×10 −6 and 3.98×10 −6 M . Beside the impact of different physics and explosion choices made in these different models, the differences in this case are also due to the different nuclear reaction rates used in the simulations, e.g., for the 22 Ne(α,n) 25 Mg and 22 Ne(α,γ) 26 Mg reactions.
For the p-process, the main effect of a higher energy SN explosion is to move outward the p-process rich region, without dramatic modification of the p-process efficiency. On the other hand, the O production tends to decrease with increasing explosion temperatures (and fallback), which is used as a reference for p-process efficiency. Furthermore, different models show local differences in the p-process distribution, but in no case do we obtain a significant p-process production of the p-rich isotopes of Mo and Ru. In general, the present yields could potentially relieve the p-process underproduction relative to O.
Low-and intermediate mass stars are evolved until the end of the AGB evolution, with the exception of the 5 M AGB models for which one-dimensional modeling assumptions are violated before all mass is lost (e.g., Lau et al. 2012). The remaining envelope mass is assumed to be ejected without any further processing. All AGB models include convective boundary mixing (overshooting) prescriptions. In agreement with previous work, this causes a larger amount of carbon and oxygen in the He intershell compared to AGB models without overshooting. The sprocess carries the known signature of overshooting applied at the bottom of the envelope, with -30large neutron exposures in the 13 C-pocket. On average, the low-mass AGB models of Set 1 have 13 C-pockets that are too small, with s-process enrichment in the envelope about 3-4 times weaker than the abundance observations in AGB stars. Despite this, the most efficient producers of the first peak elements (Y, Sr, Rb, Zr) are the 5 M intermediate mass star models. We are in the process of updating this area of our model parameterization for the next data release. The present work comprises for the first time stellar yields from low mass stars, intermediate mass stars and massive stars calculated using the same nuclear reaction network. We estimate the contribution from different stars to the nuclides, but note that a more quantitative study would require the use of a galactic chemical evolution model. For instance, we show that although massive stars are generally the dominant source of α-elements beyond carbon, intermediate mass stars do have a non-negligible contribution to oxygen. In addition, our yields have at this point no contribution for the r process or from SN type Ia, which again is something we would like to improve upon in the future. Stellar yields of Set 1 provide stellar abundance data covering both low mass and massive star models. This data release, however, is based on simplifications, such as the use of rather basic semi-analytic explosion assumptions as well as a rather simplistic treatment of mixing related to convective boundaries, which in fact we assume to present in low-and intermediate mass stars at all times at some level, while no overshooting is assumed during post-He core burning in the massive star models. We also use two different stellar codes for high-mass and low-mass stars, which introduces a small amount of inconsistency, although efforts have been made to minimize these. Our predictions presently exclude super-AGB stars. Our goal is to remove such limitations in future data release. In addition we will provide data sets for lower initial metal content, and such simulations are well underway.

A. NuGrid codes
The NuGrid nucleosynthesis codes provide a framework for performing both single-zone (sppn) and multi-zone parallel (mppnp) simulations for given thermodynamic conditions (Herwig et al. 2008a;. Both, the sppn and mppnp drivers use the same solver (Section A.3) and physics (Section A.2) packages. The single-zone driver is used, for example, for simplified simulations of trajectories for reaction rate sensitivity studies. The yields presented in this paper have been obtained with the multi-zone driver mppnp.
The stellar structure evolution is calculated with a small network, just large enough to accurately account for the nuclear energy generation. For the MESA AGB simulations the network (MESA agb.net) contains 14 isotopes, while the GENEC network contains 8 to 15 isotopes. The stellar structure evolution data for all zones at all time steps are written to disk using the NuGrid se format, a data structure based on HDF5 18 . All zones at all timesteps are then processed with the mppnp code using a dynamic network that includes all relevant reactions automatically.
In order for this post-processing approach to work the stellar evolution code has to include a large enough network to reproduce the energy generation in the same way the post-processing network would, which implies that for important reactions like 14 N(p, γ) 15 O and 12 C(α, γ) 16 O the same nuclear physics has to be adopted in both cases. The quality of the stellar evolution and post-processing network consistency is checked by comparing abundance profiles for key species from both cases, and shows in general good agreement (Fig. 27). The 12 C abundance agrees well both in the He-intershell and the H-burning ashes, indicating that both He-burning and H-burning are treated consistently between the stellar evolution and post-processing approaches. The 14 N abundance agrees for the two cases in the H-burning ashes. This reflects the consistent treatment of CNO burning in the stellar evolution and the post-processing, where 14 N is the most important isotope due to its small p-capture cross section. 14 N does not contribute in significant ways to the -32energy generation in He-burning, and therefore the difference between 14 N in MESA and mppnp in this isotope in the He-burning layers (in the mass region 0.540 < m r / M < 0.567) reflects the more complete nuclear network (including n-capture reactions) in the post-processing simulation. The latter is the more realistic solution in that case.
The advantages of the post-processing approach over a complete inline network include larger flexibility and shorter computing time. One of the reasons for the superior numerical behavior of the MESA code during the advanced phases of stellar evolution is the simultaneous solution of the structure, network and mixing operators. It would be numerically too time consuming to perform such a joint operator solve for a full s-process network with up to 1000 isotopes.
However, the implementation of a fully coupled solver in MESAis a source of inconsistency with the post-processing approach, since mppnp solves the mixing and nucleosynthesis in separate steps. There is little that can be done about this, except monitor the difference (Fig. 27) and, in case they get unacceptably large, force sub-time stepping in mppnp. So far this was not necessary.
Further, the post-process approach allows easy and rapid post-processing of the same stellar evolution track with modified input nuclear physics, provided the reactions are not important for energy generation. Realistic sensitivity studies can be performed in this way for many application. Finally, it was straight-forward to adopt a distributed parallelized computing model for the postprocessing simulations (Section A.1).

A.1. Parallel-programming implementation -mppnp
The implementation of parallelism in mppnp frame is a simple master-worker (or Workqueue) routine that assigns a single process (normally a single processor) to be the 'master' with the rest as 'workers', which is coded using the Message-Passing Interface (MPI). The main advantage gained by using MPI is the ability to use mppnp over distributed memory resources, such as cluster networks. The master performs all the serial computations, such as initialization, input/output and simple tasks, and coordinates the assignment of work to the workers using a first-in first-out (FIFO) scheduler. The worker calculates the work and then returns the result to the master. For mppnp, the unit of work is the network calculation for a single spherical shell (or 'zone') at a single time step, which is assigned by passing a message containing the temperature, density and chemical composition in the shell to the worker. We choose this definition of 'work' because network calculations for individual zones do not depend on each other and therefore no communication is required between workers. This allows for an embarrassingly parallel implementation, which simplifies the parallel implementation and reduces significantly the communication overhead. Load balancing in mppnp is simple in that zones are allocated spatially, in order, from the centre of the star, through the interior towards the surface. The reason for this is that the dynamic network typically assigns larger networks to regions of higher temperature, so the zones with the most work are allocated first.
-33 - The general operation of mppnp can be described as follows. First, the initialization is performed by the master, which includes the loading into memory of reaction rates, input parameters and initial stellar model data. The reaction rate data are then passed to all workers using broadcasts, which provide each processor with a private copy of the data required to calculate the nuclear reaction network. Following the broadcasts, the master invokes the scheduler for the first timestep. It assigns work to all available workers and then waits for a reply. Upon completion of a network calculation, the worker returns the modified abundances to the master, which it stores in an array. If there is more work to be assigned, the master assigns further work to the worker and waits for further messages. If no more work is to be assigned, the worker returns a message indicating that it is to be terminated. Once all workers respond with a termination message, all work has been completed for a single timestep and the master performs some additional serial tasks, such as a mixing step (in case a specific zone of the star has mixing coefficient larger than zero, according to the stellar structure input) and output. When the next timestep is calculated, the master invokes the scheduler again and the process is repeated.
The parallel performance of the scheduler can be estimated using a scaling curve, which is a plot of the speed-up factor as a function of the number of processors. The scaling of mppnp for a test run with 2500 timesteps of a 15 M massive star model with approximately 250 zones per timestep is shown in Fig. 28. Fig. 28 also shows the curves for Amdahl's law and Gustafson's law with a serial fraction of 1%. Since the amount of work was fixed during the test run, it is unsurprising that the curve in the strong-scaling test follows that of Amdahl's law, but it otherwise indicates that the communication overhead is negligible and that load balancing is reasonably close to optimal.

A.2. Physics package
The physics package provide to the post-processing code the list of isotopes and the nuclear reaction network to use in the calculations, and for every stellar evolution time step and stellar zone the new set of reaction rates given at the correct temperature, density and electron fraction Y e .
The species included in the network are defined by a list in a database file and by two parameters, giving the maximum allowed number of species (N N N ) and the lower limit of half life of unstable species by β-decay (tbetamin). The parameter tbetamin regulates the width of the network departing from the valley of stability. In other words, all the unstable isotopes with an half life shorter than tbetamin are not included in the network. For Set 1 the non-explosive calculations the isotopic list contains 1095 species (N N N = 1095 and tbetamin = 0.5 s). For explosive simulations the network is increased up to 5200 species (N N N = 5200 and tbetamin = 10 −5 s).
For temperatures above 6 × 10 9 K, network calculations are switched to Nuclear Statistical Equilibrium (NSE). Temperature-dependent partition function and mass excess are given by the REACLIB revision used for the simulations. Coulomb screening correction is applied according to Calder et al. (2007). The NSE module is included into a loop where feedback to the Y e from weak interactions is checked, and considered for following NSE steps.
The isomers considered are 26 Al m , 85 Kr m , 115 Cd m , 176 Lu m , and 180 Ta m . Long-lived nonthermalized isomers and ground states are considered as separated species. For temperatures lower than a given thermalization temperature, both the ground state and the isomeric state are produced. In case they are unstable, we use terrestrial β-decay rates (e.g., Ward & Fowler 1980). For temperatures higher than the thermalization temperature, the production channels to the considered isomer are neglected, and only the thermalized specie is fed. In case the isotope is unstable, above thermalization temperature the stellar β-decay rate is used. Such a simple implementation is going to be upgraded in the near future, to keep into account properly the transition phase to thermalization.

A.3. Solver Package
The solver package used to perform nucleosynthesis post-processing calculations relies on a Newton-Raphson implicit implementation, which is controlled on full precision, mass conservation and maximum size of negative yields. In case convergence criteria are not satisfied, adaptive subtime stepping is allowed. A recursive, dynamic network generation has been integrated into the solver, i.e., the size of the network automatically adapts to the conditions given. If, for example, a neutron source is activated the network will be automatically enlarged to include all heavy and unstable isotopes as needed according to the network fluxes. This dynamic network feature ensures that the network calculation never misses any production/depletion of different species or reaction chains.
Different numerical solvers based on the fully implicit method are included, and may be selected according to the architecture of the machine where the calculations are performed. At present, the available solvers are ludcmp/lubksb (Press et al. 1992), leqs (solves a linear system of equations a x = b via Gauss Jordan elimination), and standard LAPACK dgesv (double-precision general solver). The LAPACK solvers are provided from ACML or MKL libraries, which are optimized respectively for AMD Opteron and Intel processors. These LAPACK solvers can invert even rather -35large matrices (650 elements) rapidly (∼ 0.01s).

B. NuGrid data products
Although we have provided the most commonly requested derived data sets, such as yield tables, the calculations hold much more information than we can report in this paper. We are therefore making the entire computed raw data sets available via CADC 19 or the NuGrid website 20 . The data consists of two libraries. The Stellar Evolution and Explosion (SEE) library contains, for each time step, profile data needed for nucleosynthesis post-processing as well as a few abundance profiles (to check the accuracy of the post-processing) for each grid point and some scalar data (like T eff , L, etc.). The Post-Processing Data (PPD) library contains the post-processing nucleosynthesis data of the SEE library data. Data is provided in the se-flavour of HDF5. These files are normal HDF5 files but follow a certain structure suitable for the purpose. Software libraries for writing and reading se files with Fortran, C and Python, as well as detailed instructions on how to access the data are available at the NuGrid project website.
The provided data is structured in the following way. NuGrid data comes in sets. Each set corresponds to a model generation, which is defined by a common (or similar enough) set of modeling assumptions. The data provided in this paper belong to Set 1, which are meant to be standard models and which will serve as a baseline for future, improved sets. In this paper we provide two subsets, containing models with Z = 0.01, which are Set 1.1, and Z = 0.02, which are Set 1.2. This (and the following) structure is reflected in the directory tree on the CADC data server. In each of the subset directories (set1.1 and set1.2) are four directories. For both, the SEE and the PPD libraries there are pre-supernova data (i.e., the stellar evolution output, * wind) and the explosion data for the massive stars (* exp) directories. Each of these four directories is populated with one directory for each of the relevant masses. In the stellar evolution directories see wind output files with the ending .se.h5 can be found. The directories for low-and intermediate-mass star directories in see wind are the actual MESA run directories, and the se.h5 are found in a subdirectory. The time evolution of the approximated one-dimensional explosion profiles ( §2.3) are provided in .se.h5 files in the see exp directories.
Likewise, the ppd * directories contain the mppnp run directories for each mass with three types of output directories in each of them. H5 out contains se-type hdf files with the ending .out.h5. These contain complete profiles for all stable and a number of longer-lived unstable (like 14 C) species for every 20 th time step. The H5 restart directory contains restart files with all species that are considered in these calculations, every 500 time steps. The H5 surf directory contains surface elemental and isotopic, decayed and undecayed abundance evolutions at each time -36step in the .surf.h5 files.
-74 - Fig. 12.-Final isotopic distribution between C and Ni after the explosion are compared to preexplosive abundances for the models in Fig. 11: two 15 M models with delayed SN explosion and rapid/4 (where the shock velocity from explosion rapid is reduced by a factor 4), a 20 M and a 25 M models with delayed SN explosion. For a detailed comparison for all the species and for all the models we refer to the complete online tables 7 and 6 for Set 1.2, and tables 13 and 12 for Set 1.1. The ∆M limits used for the IMF weight are shown. No models representative of the mass range M = 7 − 11 M are considered (see text Section 3.1). Big red crosses are the contribution by stellar winds. Small green and blue circles are the pre-supernova abundances, between the remnant mass and the surface of the star when core collapse starts, associated to the SN fallback prescription delay and rapid (cf. Section 2.3). Green and blue large circles are the abundances including the explosive contributions according to the two fallback assumptions. Black diamonds show the yields including rapid SN with reduced explosion energy. In order to clarify if a model has a positive contribution to the chemical enrichment of e.g., carbon, we report the initial content for comparison (black-dashed line). A positive production requires that the yield is larger than the value given by the dashed line.  Fig. 13 for Si, P, S, Cl, and K and some of their stable isotopes.  Fig. 13 for Sc, V, Ti, Cr, Mn, Ca, and stable isotopes.    (Luck & Bond 1991). The 1.65 M model is not included, since the envelope material is only marginally enriched in s-process material. The ls term includes the average of Sr, Y, and Zr production. The hs term includes the average production of the elements Ba, La, Nd and Sm, according to Busso et al. (2001). Right P anel: The evolution of the [Rb/Sr] with respect to [hs/ls] for the same models in the Lef t P anel.