5.1. General Considerations
Observations reveal that more than half of the known stars are associated in binary or multiple systems. In many such systems, the stars are so well separated that there is negligible influence on each other's evolution. However, when the constituents of such a system are close, a range of interesting physical phenomena can ensue. Mass can be transferred from one star to the other when the donating star becomes so large that its atmosphere extends beyond the limiting region where the gravitational attraction of the two stars is equal. Material is then escaping from the gravitational well of the donor and flowing into the attractive field of the other object. It can be shown that there is a single point of gravitational equilibrium between the two objects, through which the mass transfer will occur. The volume around a star enclosed by the equipotential surface containing this point is called the Roche lobe. A star can fill or exceed its Roche lobe when it either increases its radius in the normal course of evolution (e.g., by becoming a Red Giant), or when the orbital separation of the two objects is decreasing due to the emission of gravitational waves. The latter situation corresponds to shrinking their respective Roche lobes. The inflow onto the surface of the companion can alter its surface properties, which is manifested, for instance, in the observed spectra. The further evolution might even lead to powerful explosions. A broad review of such phenomena is given by Warner (1995).
From the point of view of nucleosynthesis, explosive burning in such binary systems proves to be especially important as nuclides created in such an environment can be injected into the interstellar medium by the explosion. Furthermore, the nucleosynthetic signature will be different from stellar hydrostatic and explosive scenarios due to the different initial and burning conditions. Thus, different isotopic ratios can be obtained and, in some cases, it is possible to synthesize nuclides not accessible in other sites. In the following, therefore the main focus is put on outlining only such scenarios, which might have an impact on the abundances of elements observed in our Galaxy and the Universe. It has to be emphasized, however, that it is not yet possible to model all details of the explosive processes, due to the complexity of a multidimensional hydrodynamic problem including scales differing by many orders of magnitude. The problems are similar to those occurring in core-collapse supernova models. In scenarios where neutron stars are involved, additional uncertainties in the nuclear equation of state enter. As a result, a detailed, quantitative understanding of some of the processes is still lacking. Nevertheless, the basic features of all the processes can be described qualitatively in simplified models.
Historically, the Nova phenomenon is one of the oldest observed. Already astronomers of ancient civilizations were aware of sudden appearances of bright stars in the sky, at a position where no star had been observed before. This was termed "nova" as in "new star". Modern astronomy was able to identify nova progenitors. These are faint stars (mostly undetectable by the naked eye), which suddenly increase their brightness by several stellar magnitudes (i.e., their energy output increases by a factor of 100–100,000) within a few days. The star returns to its initial brightness after several months. In rare cases, the same star experiences another nova outburst within several years. This is called a recurring nova. In general, classical novae reappear only every 100 to 10,000 years.
The modern explanation of the Nova phenomenon invokes a binary system with a regular star in its early burning phases and a white dwarf (Fujimoto 1982; Truran 1984; Gehrz et al. 1998; Starrfield 1999). The latter is too faint to be directly observed. Such a system can have originated from two low-mass stars, one of them slightly more massive than the other. More massive stars evolve faster and thus the heavier star already underwent its complete stellar evolution and left a white dwarf as a remnant, while the other star has not yet even completed its early phases. From the atmosphere of the main sequence or red giant star, hydrogen-rich material is flowing across the Roche surface. It is captured by the gravitation of the white dwarf forming an accretion disk around it. Subsequently, the gas is accreted on the surface of the companion, forming a thin layer. Close to the surface, the gas becomes degenerate, delaying the onset of hydrogen burning. As soon as the necessary density and temperature is reached, hydrogen burning ignites explosively, pushing outward and ejecting part of the outer accreted layer. With the sudden expansion, temperature and density are dropping again and the burning ceases. It can reoccur as soon as a sufficient amount of new material has been accreted, which usually takes several thousand years at typical accretion rates of less than 1 Earth mass per 10 years.
In terms of nucleosynthesis it should be obvious that nova ejecta are mainly hydrogen and that they more or less retain the surface composition of the accompanying star. However, it was found observationally that they are also strongly enriched in C, N, and O. This has been a puzzle for over 2 decades (Rosner et al. 2001). In the thermonuclear runaway triggered by hydrogen ignition the main reaction sequences are those of the CNO-cycles. However, they can account neither for the observed enrichment nor for the energy production required for a fast nova burst. Currently, it is thought that material from the surface of the white dwarf might be dredged up and efficiently mixed into the burning layer and the outer zones. This would increase the CNO energy production and also provide a mechanism for the enrichment. However, to date one-dimensional and multidimensional simulations have not been able to account for the observations (Starrfield et al. 2000; Kercek et al. 1999). Nevertheless, it seems that novae are "digging up the ashes" of previous stars and distributing them throughout the Galaxy. They are considered to be the major sources of 15N, 17O, and 13C, and to have minor contributions to a number of additional species, mainly 7Li and 26Al. For current reviews on the nucleosynthetic contribution of novae, see Jordi and Hernanz (2007, 2008).
5.3. Type Ia Supernovae
Given a binary system consisting of a white dwarf and a companion star as in the nova case, the accretion rate on the surface of the white dwarf is essential for further development. Low accretion rates lead to a nova as described above. When the accretion rate exceeds about 10-8 solar masses per year, hydrogen can be quiescently burned during accretion and the burning products will sediment on the surface of the white dwarf, forming a He layer. The details on the further fate of the object are complicated and have not been fully simulated yet. Basically, two ways of explosion can be envisioned. In the first, the ignition of the He layer leads to a thermonuclear runaway, this time mainly burning via the triple-α reaction, not depending on CNO elements, contrary to novae. Nevertheless, the resulting explosion would lead to an expulsion of the outer layers, like in a nova, only much more powerful. The second type of explosion occurs when the accretion rate is even higher, about 10-6 solar masses per year. The energy released by the accretion and by burning layers heats the C/O core of the white dwarf sufficiently to ignite core C burning. This is followed by a complete disruption of the white dwarf because the nuclear energy exceeds the gravitational binding energy (Leibundgut 2001a, b).
The latter is the currently most widely accepted scenario explaining SN Ia (Nomoto et al. 1984; Hillebrandt and Niemeyer 2000). It is encouraging that observations can confirm also the basic nucleosynthesis features expected. In each explosive event, a quantity of material of the order of 0.6–0.8 solar masses is produced in the 56Ni region, which later decays to its stable isobars. Thus, the larger part of the Fe found in the solar system stems from SN Ia. In addition, some intermediate elements like Mg, Si, S, and Ca are also produced.
The details of the shockwave propagation and explosive burning are not fully understood yet. However, the total energy production remains robust due to the fact that the initial white dwarf mass is always close to the Chandrasekhar limit of 1.4 solar masses. This fact, in combination with the observational evidence that there are no compact central objects (i.e., neutron stars) found in SN Ia remnants, puts a strong constraint on the achieved explosion energy. This is why type Ia supernovae are thought to be excellent standard candles with very low variation in the effective energy output (Dominguez et al. 2001; Leibundgut 2001a, b). Type Ia supernovae as standard candles are important tools to measure astronomical distances although there is no firm theoretical grounding of the Phillips relation (see Sect. 2.1.4) yet.
5.4. X-ray Bursts and the rp-Process
Bursts have been observed not only in the optically visible frequency range of light. Powerful, brief bursts of X-rays are also observed throughout the Galaxy. A binary system is suggested to be responsible for one subclass of these bursts. Their duration is from several seconds up to minutes and they show a fast rise and a slowly decaying tail. In the assumed scenario, a main-sequence star and a neutron star are orbiting the common center of mass. As in the nova and SN Ia case, material is flowing from the atmosphere of the companion star and is accreted on the surface of the compact object. Due to the increased gravitational field of the neutron star in comparison with a white dwarf, the thermonuclear runaway after ignition of hydrogen burning can proceed differently from the one in novae and supernovae (Taam et al. 1996; Schatz et al. 1998; Wiescher and Schatz 2000; Boyd 2008). Hydrogen and subsequently also helium, burn explosively at higher temperatures as before. First, the so-called hot CNO-cycle (also found in certain massive stars) is established, followed by further CNO-type cycles beyond Ne. The energy production from these cycles leads to a breakout to further CNO-type cycles beyond Ne at temperatures surpassing 4 × 108 K. The additional cycles generate additional energy, further increasing the temperature. In the next stage of the ignition process, He is also burned in the triple-α reaction. The CNO-type cycles break up toward more proton-rich nuclides by (α, p) and (p, γ) reactions.
Finally, the rp-process (rapid proton capture) sets in. In the rp-process, similar to the r-process for neutrons, proton captures and photodisintegrations are in equilibrium. Thus, the abundances within an isotonic chain are only determined by temperature, density, and the proton separation energy of a nucleus. The timescales of flows from one chain into the next are given by β-decay half-lives. The reaction path basically follows the proton dripline. The processing of matter is hampered at nuclides with long half-lives, the so-called waiting points, which determine the processing timescales. A final endpoint of the rp-process path was found in a closed reaction cycle in the Sn–Sb–Te region, due to increasing instability against α decay of heavy proton-rich nuclides (Schatz et al. 2001).
Time-dependent calculations showed that the structure of type I X-ray bursts can be explained by the energy generation of the proposed processes. Regarding nucleosynthesis, it is being discussed whether a fraction of or all light p-nuclei (see Sect. 4.5.2) originate from X-ray bursters. Since the rp-process synthesizes very proton-rich nuclei, they would decay to p-nuclei after the burning ceased. Since it proved to be problematic to synthesize the light p-nuclei with mass numbers A < 110 in the γ-process by photodisintegration (Sect. 4.5.2), the rp-process provides a compelling alternative to such models, especially after the discovery of its endpoint, preventing the production of nuclides with mass numbers A> 110. Although it was shown that the isotopes in question can be produced in large quantities, it is still speculative whether any of the material is ejected. There could be a very small fraction of material lost from the outer atmosphere of the accreted layer on the neutron star surface but the rp-process burning takes place further down. Therefore, some convection has to be invoked to bring the freshly produced nuclei to the outer layers. Comparatively small amounts of ejecta would be sufficient to explain solar p-abundances but detailed hydrodynamic studies of the burning, convection, and possible ejection are still needed.
5.5. Neutron Star Mergers
Another interesting binary system is that of two neutron stars. Such systems are known to exist; four have been detected by now. They can be created when two massive stars complete their stellar evolution and both explode in a core-collapse supernova, each leaving a neutron star behind. In such a configuration, the system loses angular momentum by emission of gravitational radiation (Taylor 1994) and the two neutron stars spiral inward. At timescales of 108 years or less, the objects collide at their center of mass. Such a merger can lead to the ejection of neutron-rich material (Rosswog et al. 1999; Ruffert and Janka 2001). Since this material is even more neutron rich than the deep, high-entropy layers thought to be a possible site of the r-process in core-collapse supernovae, nucleosynthesis in decompressed neutron star matter could be a viable alternative site for the r-process (Freiburghaus et al. 1999). Detailed hydrodynamic calculations coupled to a complete r-process reaction network have not been undertaken yet, but parameterized r-process studies indicate the possibility that such mergers could even account for all heavy r-process matter in the Galaxy. However, detailed galactic chemical evolution models (Argast et al. 2004) show that neutron star mergers, occurring at late time in the life of a galaxy, cannot account for the r-process nuclei found in very old stars (Sneden et al. 2000; Cayrel et al. 2001; Frebel et al. 2005). Therefore, there may be several sites producing r-process nuclei, perhaps similar to the two components of the s-process (but occurring in different sites than the s-process, of course).