2014ApJ...792....4C

Burst oscillations during the rising phases of thermonuclear X-ray bursts are usually believed to originate from flame spreading on the neutron star surface. However, the decrease of fractional oscillation amplitude with rise time, which provides a main observational support for the flame spreading model, have so far been reported from only a few bursts. Moreover, the non-detection and intermittent detections of rise oscillations from many bursts are not yet understood considering the flame spreading scenario. Here, we report the decreasing trend of fractional oscillation amplitude from an extensive analysis of a large sample of Rossi X-ray Timing Explorer Proportional Counter Array bursts from 10 neutron star low-mass X-ray binaries. This trend is 99.99% significant for the best case, which provides, to the best of our knowledge, by far the strongest evidence of such a trend. Moreover, it is important to note that an opposite trend is not found in any of the bursts. The concave shape of the fractional amplitude profiles for all the bursts suggests latitude-dependent flame speeds, possibly due to the effects of the Coriolis force. We also systematically study the roles of low fractional amplitude and low count rate for non-detection and intermittent detections of rise oscillations, and attempt to understand them within the flame spreading scenario. Our results support a weak turbulent viscosity for flame spreading, and imply that burst rise oscillations originate from an expanding hot spot, thus making these oscillations a more reliable tool to constrain the neutron star equations of state.