5. Discussion and conclusions
Using a simple but crucial model of the thermonuclear flash on accreting neutron stars, we have shown that the nuclear data near the proton drip line affect the rp-process significantly: the energy generation rate (peak temperature) as shown in Figs. 2, 3 and 4. The amounts of the fuel left after the flash (final products) are shown in Figs. 5 and 6. The final products are summarized in Tables 3 and 4. As a consequence, the profile of the light curve must be affected also as shown in Fig. 7. Compared with HSH, break out from the HCNO cycle is so rapid that iron peak elements have been produced before the flash. Contrary to the large differences in the rp-process attributed to the various nuclear data, the differences are quantitatively very small between independent calculations of the explosive nucleosynthesis in supernovae (e.g. Hashimoto et al. 1989, Thielemann et al. 1990) because the supernova nucleosynthesis proceeds under the condition of approximately nuclear statistical equilibrium.
It is noted that, since our network is limited to Kr isotopes, we should extend it to include the nuclei of for the high temperature cases with as pointed out by Schatz et al. (1998) though the flow beyond Kr depends on the Q-value of waiting nucleus 68Se; only in the deepest accreted layer the rp-process beyond Kr would proceed appreciably (see Fig. 5 in Fujimoto et al. 1987). Therefore, it is highly needed to determine the proton drip line from Ge to Kr isotopes (Mohar et al. 1991, Henncheck et al. 1994, Blank 1995).
While our one zone model represents rather well the flash phase, convection must be actually taken into account. Fujimoto et al. (1987) demonstrated the importance of the mixing process to explain bursts of very short intervals like 10 minutes X-ray bursts (e.g. Murakami et al. 1980). They used spherically symmetric evolutionary code, and the helium abundances averaged in the burning shell are assumed with the use of an approximate network of the previous version of HSH. It should be noted that the mixing process is still very uncertain. Therefore, one zone approach is still very useful to see how the rp-process will be affected by the change of the nuclear data and other physical inputs. Though the detailed multi-dimensional hydrodynamical calculation would be desirable, even the spherical calculation of stellar evolution should be highly necessary to see the effects of uncertainties of nuclear data. Then, approximate network which simulates nuclear energy generation rates should be employed in realistic calculations as is done by Fujimoto et al. (1987).
Recent observations of quasi-periodic oscillations may constrain the mass and radius of neutron stars. For example, analyzing the LMXB 4U 1636-536, Kaaret et al. (1997) suggested that the mass and the radius of the neutron star are around 2.0 and 9.0 km, respectively, which is nearly compatible with the maximum state derived from the equation of state by Friedman & Pandharipande (1981). Then, we have log = 14.75 with . A neutron star of and km has log = 14.25 with which state is reproduced by the equation of state AV14+UVII by Wiringa et al. (1988). Taam et al. (1996) adopted a star of and km; log = 14.5 with . Therefore, our parameters are within the range of more realistic X-ray burst models.
Related to the rp-process, Chakrabarti et al. (1987) pointed out the possibility inside a thick accretion disk around a stellar mass black hole. Recent calculations of nucleosynthesis inside thick accretion disks have suggested that significant nuclear processing would occur if a viscosity is low (e.g. Arai & Hashimoto 1992, 1995); it may be promising for the rp-process to occur if hydrogen is mixed into the hot interior material. Though the disk model is still rather uncertain, it is worth while to investigate the nucleosynthesis using another model like advection dominated disk; if it occurs, some accretion disks could give an important site of nucleosynthesis even in the early stage of the universe.
Finally, we want to stress the difference of the available reaction rates. We employed REACLIB and compared the results with the old rates, many of which seem to be still surviving. The database of another reaction rates (HW92) 2 in which reaction rates are tabulated in specified temperature grids has been mostly adopted in our cases A and B except for some reactions in Caughlan & Fowler (1988). It must be noted that if we compare individual rates in two databases, many of them have significant deviations in the temperature range of which is relevant in the present investigations. In particular, as the temperature becomes higher, say, , differences become larger; in part due to the different treatment of the level density. Let us evaluate the ratio f of the reaction rate given by REACLIB to HW92. For example, one has for in , and for in . Even for the lower temperature, for and for in . However, it must be noted that for European Compilation instead of REACLIB for (Rayet 1998, see also the experimental value by Kiener et al. 1993). Though the effects of these differences on X-ray burst phenomena are not always clear, we should be careful of the uncertainty of reaction rates.
© European Southern Observatory (ESO) 1999
Online publication: February 22, 1999