We have assembled a sample of 1187 thermonuclear (type I) X-ray bursts from observations of 48 accreting neutron stars by the Rossi X-ray Timing Explorer, spanning more than 10 years. The sample contains examples of two of the three theoretical ignition regimes (confirmed via comparisons with numerical models) and likely examples of the third. We present a detailed analysis of the variation of the burst profiles, energetics, recurrence times, presence of photospheric radius expansion, and presence of burst oscillations, as a function of accretion rate. We estimated the distance for 35 sources exhibiting radius-expansion bursts, and found that the peak flux of such bursts varies typically by 13%. We classified sources into two main groups based on the burst properties: (1) both long and short bursts (indicating mixed H/ He accretion), and (2) consistently short bursts (primarily He accretion), and we calculated the mean burst rate as a function of accretion rate for the two groups. The decrease in burst rate observed at >0:06Ṁ Edd (k2 ; 10 37 ergs s À1) is associated with a transition in the persistent spectral state and (as has been suggested previously) may be related to the increasing role of steady He burning. We found many examples of bursts with recurrence times <30 minutes, including burst triplets and even quadruplets. We describe the oscillation amplitudes for 13 of the 16 burst oscillation sources, as well as the stages and properties of the bursts in which the oscillations are detected. The burst properties are correlated with the burst oscillation frequency; sources spinning at <400 Hz generally have consistently short bursts, while the more rapidly spinning systems have both long and short bursts. This correlation suggests either that shear-mediated mixing dominates the burst properties, or alternatively that the nature of the mass donor (and hence the evolutionary history) has an influence on the long-term spin evolution.

Millisecond pulsars are neutron stars that are thought to have been spun-up by mass accretion from a stellar companion.1 It is unknown whether there is a natural brake for this process, or if it continues until the centrifugal breakup limit is reached at submillisecond periods. Many neutron stars that are accreting mass from a companion star exhibit thermonuclear X-ray bursts that last tens of seconds, caused by unstable nuclear burning on their surfaces.2 Millisecond-period brightness oscillations during bursts from ten neutron stars (as distinct from other rapid X-ray variability that is also observed 3,4 ) are thought to measure the stellar spin, 2,5 but direct proof of a rotational origin has been lacking. Here, we report the detection of burst oscillations at the known spin frequency of an accreting millisecond pulsar, and we show that these oscillations always have the same rotational phase. This firmly establishes burst oscillations as nuclear-powered pulsations tracing the spin of accreting neutron stars, corroborating earlier evidence.5,6 The distribution of spin frequencies of the 11 nuclear-powered pulsars cuts off well below the breakup frequency for most neutron star models, supporting theoretical predictions that gravitational radiation losses can limit accretion torques in spinning up millisecond pulsars. 7-9The millisecond oscillations observed during X-ray bursts are not perfectly coherent, but usually drift in frequency by several hertz over the course of a burst, generally reaching an asymptotic maximium frequency that is repeatable in a given neutron star.2 This frequency drift has been interepreted as arising from angular momentum conservation in a decoupled surface burning layer that expands and contracts during the burst, so that the asymptotic frequency is the stellar spin frequency.10,11 A puzzle in this picture is why the oscillation persists late in the burst, well after the nuclear burning has spread over the entire star. Also, in most of these neutron stars, unexplained pairs of kilohertz quasi-periodic oscillations (kHz QPOs) are also observed in the non-burst X-ray emission, with the QPO separation frequency approximately equal to either the burst oscillation frequency or half this value, We have observed the transient X-ray source SAX J1808.4−3658, which has been detected in four outbursts since its discovery

We have identified the third known accretion-powered millisecond pulsar, XTE J0929−314, with the Rossi X-Ray Timing Explorer. The source is a faint, high-Galactic-latitude X-ray transient (d 5 kpc) that was in outburst during 2002 April-June. The 185 Hz (5.4 ms) pulsation had a fractional rms amplitude of 3-7% and was generally broad and sinusoidal, although occasionally double-peaked. The hard X-ray pulses arrived up to 770 µs earlier than the soft X-ray pulses. The pulsar was spinning down at an average rate ofν = (−9.2 ± 0.4) × 10 −14 Hz s −1 ; the spin-down torque may arise from magnetic coupling to the accretion disk, a magnetohydrodynamic wind, or gravitational radiation from the rapidly spinning pulsar. The pulsations were modulated by a 43.6 min ultracompact binary orbit, yielding the smallest measured mass function (2.7 × 10 −7 M ⊙ ) of any stellar binary. The binary parameters imply a ≃ 0.01M ⊙ white dwarf donor and a moderately high inclination. We note that all three known accreting millisecond pulsars are X-ray transients in very close binaries with extremely low mass transfer rates. This is an important clue to the physics governing whether or not persistent millisecond pulsations are detected in low-mass X-ray binaries.

The standard approach for time-resolved X-ray spectral analysis of thermonuclear bursts involves subtraction of the pre-burst emission as background. This approach implicitly assumes that the persistent flux remains constant throughout the burst. We reanalyzed 332 photospheric radius expansion bursts observed from 40 sources by the Rossi X-ray Timing Explorer, introducing a multiplicative factor f a to the persistent emission contribution in our spectral fits. We found that for the majority of spectra the best-fit value of f a is significantly greater than 1, suggesting that the persistent emission typically increases during a burst. Elevated f a values were not found solely during the radius expansion interval of the burst, but were also measured in the cooling tail. The modified model results in a lower average value of the χ 2 fit statistic, indicating superior spectral fits, but not yet to the level of formal statistical consistency for all the spectra. We interpret the elevated f a values as an increase of the mass accretion rate onto the neutron star during the burst, likely arising from the effects of Poynting-Robertson drag on the disk material. We measured an inverse correlation of f a with the persistent flux, consistent with theoretical models of the disc response. We suggest that this modified approach may provide more accurate burst spectral parameters, as well as offering a probe of the accretion disk structure.

We analysed Rossi X-ray Timing Explorer observations of the accretion-powered 401 Hz pulsar SAX J1808.4−3658, in order to precisely determine the source distance. While the fluences for the five transient outbursts observed from 1996 were constant to within the uncertainties, the outburst interval varied signficantly, so that the time-averaged flux (and accretion rate) decreased by around 40%. By equating the time-averaged X-ray flux with the expected mass transfer rate from gravitational radiation, we derived a lower limit on the distance of 3.4 kpc. Combined with an upper limit from assuming that the four radius-expansion thermonuclear bursts observed during the 2002 October outburst reached at most the Eddington limit for a pure He atmosphere, we found that the probable distance range for the source is 3.4-3.6 kpc. The implied inclination, based on the optical/IR properties of the counterpart, is i 30 • .We compared the properties of the bursts with an ignition model. The time between bursts was long enough for hot CNO burning to significantly deplete the accreted hydrogen, so that ignition occurred in a pure helium layer underlying a stable hydrogen burning shell. This is the first time that this burning regime has been securely observationally identified. The observed energetics of the bursts give a mean hydrogen fraction at ignition of X ≈ 0.1, and require that the accreted hydrogen fraction X 0 and the CNO metallicity Z CNO are related by Z CNO ≈ 0.03 (X 0 /0.7) 2 . We show that in this burning regime, a measurement of the burst recurrence time and energetics allows the local accretion rate onto the star to be determined independently of the accreted composition, giving a new method for estimating the source distance which is in good agreement with our other estimates.

The X-ray burster GS 1826-24 shows extremely regular Type I X-ray bursts whose energetics and recurrence times agree well with thermonuclear ignition models. We present calculations of sequences of burst lightcurves using multizone models which follow the nucleosynthesis (αp and rp-processes) with an extensive nuclear reaction network. The theoretical and observed burst lightcurves show remarkable agreement. The models naturally explain the slow rise (duration ≈ 5 s) and long tails (≈ 100 s) of these bursts, as well as their dependence on mass accretion rate. This comparison provides further evidence for solar metallicity in the accreted material in this source, and gives a distance to the source of (6.07 ± 0.18) kpc ξ −1/2 b , where ξ b is the burst emission anisotropy factor. The main difference is that the observed lightcurves do not show the distinct two-stage rise of the models. This may reflect the time for burning to spread over the stellar surface, or may indicate that our treatment of heat transport or nuclear physics needs to be revised. The trends in burst properties with accretion rate are well-reproduced by our spherically symmetric models which include chemical and thermal inertia from the ashes of previous bursts. Changes in the covering fraction of the accreted fuel are not required.

Spectral measurements of thermonuclear (type-I) X-ray bursts from low mass X-ray binaries have been used to measure neutron star (NS) masses and radii. A number of systematic issues affect such measurements and have raised concerns as to the robustness of the methods. We present analysis of the X-ray emission from bursts observed from 4U 1608-52 at various persistent fluxes. We find a strong dependence of the burst properties on the flux and spectral hardness of the persistent emission before burst. Bursts occurring during the low-accretion rate (hard) state exhibit evolution of the black body normalisation consistent with the theoretical predictions of NS atmosphere models. However, bursts occurring during the high-accretion rate (soft) state show roughly constant normalisation, which is inconsistent with the NS atmosphere models and therefore these bursts cannot be easily used to determine NS parameters. We analyse the hard-state burst to put the lower limit on the neutron star radius in 4U 1608-52 of 13 km (for masses 1.2-2.4 M ⊙ ). The best agreement with the theoretical NS mass-radius relations is achieved for source distances in the range 3.1-3.7 kpc. We expect that the radius limit will be 10 per cent lower if spectral models including rapid rotation are used instead.

We present the largest sample of type I (thermonuclear) X-ray bursts yet assembled, comprising 7083 bursts from 85 bursting sources. The sample is drawn from observations with Xenon-filled proportional counters on the long-duration satellites RXTE, BeppoSAX, and International Gamma-Ray Astrophysics Laboratory between 1996 February 8 and 2012 May 3. The burst sources were drawn from a comprehensive catalog of 115 burst sources, assembled from earlier catalogs and the literature. We carried out a consistent analysis for each burst light curve (normalized to the relative instrumental effective area) and provide measurements of rise time, peak intensity, burst timescale, and fluence. For bursts observed with the RXTE/PCA and BeppoSAX/Wide Field Camera we also provide time-resolved spectroscopy, including estimates of bolometric peak flux and fluence, and spectral parameters at the peak of the burst. For 950 bursts observed with the PCA from sources with previously detected burst oscillations, we include an analysis of the high time resolution data, providing information on the detectability and amplitude of the oscillations, as well as where in the burst they are found. We also present analysis of 118,848 observations of the burst sources within the sample time frame. We extracted 3–25 keV X-ray spectra from most observations, and (for observations meeting our signal-to-noise criterion) we provide measurements of the flux, spectral colors, and, for selected sources, the position on the color–color diagram, for the best-fit spectral model. We present a description of the sample, a summary of the science investigations completed to date, and suggestions for further studies.