Periodic Thermonuclear X‐Ray Bursts from GS 1826−24 and the Fuel Composition as a Function of Accretion Rate

We analyze 24 type I X-ray bursts from GS 1826-24 observed by the Rossi X-rayTiming Explorer between 1997 November and 2002 July. The bursts observedbetween 1997-98 were consistent with a stable recurrence time of 5.74 +/- 0.13hr. The persistent intensity of GS 1826-24 increased by 36% between 1997-2000,by which time the burst interval had decreased to 4.10 +/- 0.08 hr. In 2002July the recurrence time was shorter again, at 3.56 +/- 0.03 hr. The burstswithin each epoch had remarkably identical lightcurves over the full approx.150 s burst duration; both the initial decay timescale from the peak, and theburst fluence, increased slightly with the rise in persistent flux. Thedecrease in the burst recurrence time was proportional to Mdot^(-1.05+/-0.02)(where Mdot is assumed to be linearly proportional to the X-ray flux), so thatthe ratio alpha between the integrated persistent and burst fluxes wasinversely correlated with Mdot. The average value of alpha was 41.7 +/- 1.6.Both the alpha value, and the long burst durations indicate that the hydrogenis burning during the burst via the rapid-proton (rp) process. The variation inalpha with Mdot implies that hydrogen is burning stably between bursts,requiring solar metallicity (Z ~ 0.02) in the accreted layer. We show thatsolar metallicity ignition models naturally reproduce the observed burstenergies, but do not match the observed variations in recurrence time and burstfluence. Low metallicity models (Z ~ 0.001) reproduce the observed trends inrecurrence time and fluence, but are ruled out by the variation in alpha. Wediscuss possible explanations, including extra heating between bursts, or thatthe fraction of the neutron star covered by the accreted fuel increases withMdot.Comment: 9 pages, 6 figures, accepted by ApJ. Minor revisions following the referee's repor