We present the results of a systematic study of the evolution of low-and intermediate-mass X-ray binaries (LMXBs and IMXBs). Using a standard Henyey-type stellar evolution code and a standard model for binary interactions, we have calculated 100 binary evolution sequences containing a neutron star and a normal-type companion star, where the initial mass of the secondary ranges from 0.6 to 7 M _ and the initial orbital period from D4 hr to D100 days. This range samples the entire range of parameters one is likely to encounter for LMXBs and IMXBs. The sequences show an enormous variety of evolutionary histories and outcomes, where di †erent mass transfer mechanisms dominate in di †erent phases. Very few sequences resemble the classical evolution of cataclysmic variables, where the evolution is driven by magnetic braking and gravitational radiation alone. Many systems experience a phase of mass transfer on a thermal timescale and may brieÑy become detached immediately after this phase (for the more massive secondaries). In agreement with previous results (Tauris & Savonije 1999), we Ðnd that all sequences with (sub)giant donors up to D2 are stable against dynamical mass transfer. Sequences M _ where the secondary has a radiative envelope are stable against dynamical mass transfer for initial masses up to D4 For higher initial masses, they experience a delayed dynamical instability after a M _ . stable phase of mass transfer lasting up to D106 yr. Systems where the initial orbital period is just below the bifurcation period of D18 hr evolve toward extremely short orbital periods (as short as D10 minutes). For a 1 secondary, the initial period range that leads to the formation of ultracompact M _ systems (with minimum periods less than D40 minutes) is 13È18 hr. Since systems that start mass transfer in this period range are naturally produced as a result of tidal capture, this may explain the large fraction of ultracompact LMXBs observed in globular clusters. The implications of this study for our understanding of the population of X-ray binaries and the formation of millisecond pulsars are also discussed.

We systematically examine how the presence in a binary affects the final core structure of a massive star and its consequences for the subsequent supernova explosion. Interactions with a companion star may change the final rate of rotation, the size of the helium core, the strength of carbon burning and the final iron core mass. Stars with initial masses larger than ∼ 11 M ⊙ that experience core collapse will generally have smaller iron cores at the point of explosion if they lost their envelopes due to a binary interaction during or soon after core hydrogen burning. Stars below ∼ 11M ⊙ , on the other hand, can end up with larger helium and metal cores if they have a close companion, since the second dredge-up phase which reduces the helium core mass dramatically in single stars does not occur once the hydrogen envelope is lost. We find that the initially more massive stars in binary systems with masses in the range 8 − 11M ⊙ are likely to undergo an electron-capture supernova, while single stars in the same mass range would end as ONeMg white dwarfs. We suggest that the core collapse in an electron-capture supernova (and possibly in the case of relatively small iron cores) leads to a prompt or fast explosion rather than a very slow, delayed neutrino-driven explosion and that this naturally produces neutron stars with low-velocity kicks. This leads to a dichotomous distribution of neutron star kicks, as inferred previously, where neutron stars in relatively close binaries attain low kick velocities. We illustrate the consequences of such a dichotomous kick scenario using binary population synthesis simulations and discuss its implications. This scenario has also important consequences for the minimum initial mass of a massive star that becomes a neutron star. For single stars the critical mass may be as high as 10 -12 M ⊙ , while for close binaries, it may be as low as 6 -8 M ⊙ . These critical masses depend on the treatment of convection, the amount of convective overshooting and the metallicity of the star and will generally be lower for larger amounts of convective overshooting and lower metallicity.

During 5 years of Chandra observations, we have identified seven X-ray transients located within 23 pc of Sgr A*. These sources each vary in luminosity by more than a factor of 10 and have peak X-ray luminosities greater than ergs s Ϫ1 , which strongly suggests that they are accreting black holes or neutron stars. The 33 5 # 10 peak luminosities of the transients are intermediate between those typically considered outburst and quiescence for X-ray binaries. Remarkably, four of these transients lie within only 1 pc of Sgr A*. This implies that, compared to the numbers of similar systems located between 1 and 23 pc, transients are overabundant by a factor of տ20 per unit stellar mass within 1 pc of Sgr A*. It is likely that the excess transient X-ray sources are low-mass Xray binaries that were produced, as in the cores of globular clusters, by three-body interactions between binary star systems and either black holes or neutron stars that have been concentrated in the central parsec through dynamical friction. Alternatively, they could be high-mass X-ray binaries that formed among the young stars that are present in the central parsec.

We investigate an interesting new class of high-mass X-ray binaries (HMXBs) with long orbital periods (P orb > 30 d) and low eccentricities (e 0.2). The orbital parameters suggest that the neutron stars in these systems did not receive a large impulse, or "kick," at the time of formation. After considering the statistical significance of these new binaries, we develop a self-consistent phenomenological picture wherein the neutron stars born in the observed wide HMXBs receive only a small kick ( 50 km s −1 ), while neutron stars born in isolation, in the majority of low-mass X-ray binaries, or in many of the wellknown HMXBs with P orb 30 d receive the conventional large kicks, with a mean speed of ∼ 300 km s −1 . Assuming that this basic scenario is correct, we discuss a physical process that lends support to our hypothesis, whereby the magnitude of the natal kick to a neutron star born in a binary system depends on the rotation rate of its immediate progenitor following mass transfer -the core of the initially more massive star in the binary. Specifically, the model predicts that rapidly rotating pre-collapse cores produce NSs with relatively small kicks, and vice versa for slowly rotating cores. If the envelope of the NS progenitor is removed before it has become deeply convective, then the exposed core is likely to be a rapid rotator. However, if the progenitor becomes highly evolved prior to mass transfer, then a strong magnetic torque, generated by differential rotation between the core and the convective envelope, may cause the core to spin down to the very slow rotation rate of the envelope. Our model, if basically correct, has important implications for the dynamics of stellar core collapse, the retention of neutron stars in globular clusters, and the formation of double neutron star systems in the Galaxy.

We present the first study that combines binary population synthesis in the Galactic disk and detailed evolutionary calculations of low-and intermediate-mass X-ray binaries (L/IMXBs). This approach allows us to follow completely the formation of L/IMXBs, and their evolution through the X-ray phase, to the point when they become binary millisecond pulsars (BMPs). We show that the formation probability of IMXBs with initial donor masses of 1.5-4 M ⊙ is typically 5 times higher than that of standard LMXBs with initial donor masses of <1.5 M ⊙ . Since IMXBs evolve to resemble observed LMXBs, we suggest that the majority of the observed systems may have descended from IMXBs. Distributions at the current epoch of the orbital periods, donor masses, and mass accretion rates of L/IMXBs have been computed, as have orbital-period distributions of BMPs. Several significant discrepancies between the theoretical and observed distributions are discussed. We find that the total number of luminous (L X > 10 36 ergs s −1 ) X-ray sources at the current epoch and the period distribution of BMPs are very sensitive to the parameters in analytic formula describing the common-envelope phase that precedes the formation of the neutron star. The orbital-period distribution of observed BMPs strongly favors cases where the common envelope is more easily ejected. However, this leads to a 100-fold overproduction of the theoretical number of luminous X-ray sources relative to the total observed number of LMXBs. X-ray irradiation of the donor star may result in a dramatic reduction in the X-ray active lifetime of L/IMXBs, thus possibly resolving the overproduction problem, as well as the long-standing BMP/LMXB birthrate problem.