Coherent oscillations and the evolution of the X-ray spectrum during thermonuclear X-ray bursts in accreting neutron-star X-ray binaries have been studied intensively but separately. We analysed all the X-ray bursts of the source 4U 1728-34 with the Rossi X-ray Timing Explorer. We found that the presence of burst oscillations can be used to predict the behaviour of the blackbody radius during the cooling phase of the bursts. If a burst shows oscillations, during the cooling phase the blackbody radius remains more or less constant for ~2 - ~8s, whereas in bursts that do not show oscillations the blackbody radius either remains constant for more than ~2 - ~8s or it shows a rapid (faster than ~2s) decrease and increase. Both the presence of burst oscillations and the time-dependent spectral behaviour of the bursts are affected by accretion rate. We also found that the rise time and convexity of the bursts' light curve are different in bursts with and without oscillations in 4U 1728--34. Bursts with oscillations have a short rise time (~0.5s) and show both positive and negative convexity, whereas bursts without oscillations have a long rise time (~1s) and mostly positive convexity. This is consistent with the idea that burst oscillations are associated with off-equator ignition.

Thermonuclear burning on the surface of accreting neutron stars is observed to stabilize at accretion rates almost an order of magnitude lower than theoretical models predict. One way to resolve this discrepancy is by including a base heating flux that can stabilize the layer. We focus our attention on pure helium accretion, for which we calculate the effect of a base heating flux on the critical accretion rate at which thermonuclear burning stabilizes. We use the MESA stellar evolution code to calculate $\dot m_{\rm crit}$ as a function of the base flux, and derive analytic fitting formulae for $\dot m_{\rm crit}$ and the burning temperature at that critical accretion rate, based on a one-zone model. We also investigate whether the critical accretion rate can be determined by examining steady-state models only, without time-dependent simulations. We examine the argument that the stability boundary coincides with the turning point $dy_{\rm burn}/d\dot m=0$ in the steady-state models, and find that it does not hold outside of the one-zone, zero base flux case. A linear stability analysis of a large suite of steady-state models is also carried out, which yields critical accretion rates a factor of $\sim3$ larger than the MESA result, but with a similar dependence on base flux. Lastly, we discuss the implications of our results for the ultracompact X-ray binary 4U~1820-30.

We investigate the constraints on neutron star mass and radius in GS 1826-24 from models of lightcurves and spectral evolution of type I X-ray bursts. This source shows remarkable agreement with theoretical calculations of burst energies, recurrence times, and lightcurves. We first exploit this agreement to set the overall luminosity scale of the observed bursts. When combined with a measured blackbody normalization, this leads to a distance and anisotropy independent measurement of the ratio between the redshift 1+z and color correction factor f_c. We find 1+z=1.19-1.28 for f_c=1.4-1.5. We then compare the evolution of the blackbody normalization with flux in the cooling tail of bursts with predictions from spectral models of Suleimanov et al. (2011b). The observations are well described by the models at luminosities greater than about one third of the peak luminosity, with deviations emerging at luminosities below that. We show that this comparison leads to distance independent upper limits on R_\infty and neutron star mass of R_\infty\lesssim 9.0-13.2 km and M<1.2-1.7 M_\odot, respectively, for solar abundance of hydrogen at the photosphere and a range of metallicity and surface gravity. The radius limits are low in comparison to previous measurements. This may be indicative of a subsolar hydrogen fraction in the GS 1826-24 photosphere, or of larger color corrections than that predicted by spetral models. Our analysis also gives an upper limit on the distance to GS 1826-24 of d<4.0-5.5 kpc \xi_b^{-1/2}, where \xi_b is the degree of anisotropy of the burst emission.