5 Thermonuclear X-ray Bursts
235
Unfortunately, the effects of anisotropy are often neglected, because the system
inclination is difficult to measure accurately.
5.5 Burst Oscillations and the Neutron Star Spin
The evolutionary role for accretion-powered neutron stars in low-mass binaries as
precursors to rotation-powered millisecond pulsars [1] has been well established
with the discovery of several types of coherent millisecond oscillations, by which
the neutron-star spin frequencies may be measured. Most compelling has been
the discovery of persistent X-ray pulsations at frequencies typically in the range
200–600 Hz (e.g. [148] and Patruno and Watts, this volume). Those sources also
confirmed the earlier discoveries of burst oscillations as an alternative tracer of the
neutron star spin [24].
Burst oscillations are coherent, periodic variations in the X-ray intensity of
bursting sources that are observed during a thermonuclear burst (see [193] for
a review). These oscillations were first detected in the well-studied burst source
4U 1728−34 by RXTE shortly after its launch, in 1996 [175]. To date oscillations
have been detected in 17 sources, with tentative (low significance, or single-burst
detections, or both) claims for a further 9 sources. The oscillations are detected at
a typical fractional amplitude of 0.05–0.2, and are within a few Hz of a frequency
value characteristic for each source. The oscillation frequency tends to drift upwards
by 1–2 Hz while it is present. For those sources which also show pulsations in
the persistent (accretion-powered) X-ray emission, the frequencies of the persistent
pulsations and burst oscillations are within a few Hz of each other, confirming the
burst oscillation frequency as (roughly) the neutron star spin [24]. Oscillations are
not observed in every burst, but instead are preferentially seen at high accretion rates
[137, 142], when the spectrum is in the “soft” state (see Sect. 5.1.2). Even in those
bursts where they are present, the oscillations are not found throughout the burst,
instead tending to occur most often in the burst rise, and less frequently through to
the burst tail [53].
The mechanism giving rise to the oscillations remains unknown; candidates
include the spreading hotspot model, the related “cooling wake” that follows,
and models involving surface modes of various kinds. The rise in fractional burst
oscillation amplitude in the early phases of the burst generally supports the hotspot
model [25, 176], although once the burning has spread to the entire surface it is
difficult to understand how oscillations can persist. If the cooling timescale is similar
to the spreading timescale, it seems possible that the temperature contrast can also
produce oscillations in the tail [123], although such models have difficulty achieving
the high amplitudes that are observed. Models invoking rotation modes [73] suffer
from difficulties reproducing the observed frequency drifts [13].
In the absence of certainty as to the mechanism, much effort has been focussed
on understanding why oscillations are not seen in every burst. Studies of the
assembled RXTE sample have confirmed that oscillations preferentially occur at
235
Unfortunately, the effects of anisotropy are often neglected, because the system
inclination is difficult to measure accurately.
5.5 Burst Oscillations and the Neutron Star Spin
The evolutionary role for accretion-powered neutron stars in low-mass binaries as
precursors to rotation-powered millisecond pulsars [1] has been well established
with the discovery of several types of coherent millisecond oscillations, by which
the neutron-star spin frequencies may be measured. Most compelling has been
the discovery of persistent X-ray pulsations at frequencies typically in the range
200–600 Hz (e.g. [148] and Patruno and Watts, this volume). Those sources also
confirmed the earlier discoveries of burst oscillations as an alternative tracer of the
neutron star spin [24].
Burst oscillations are coherent, periodic variations in the X-ray intensity of
bursting sources that are observed during a thermonuclear burst (see [193] for
a review). These oscillations were first detected in the well-studied burst source
4U 1728−34 by RXTE shortly after its launch, in 1996 [175]. To date oscillations
have been detected in 17 sources, with tentative (low significance, or single-burst
detections, or both) claims for a further 9 sources. The oscillations are detected at
a typical fractional amplitude of 0.05–0.2, and are within a few Hz of a frequency
value characteristic for each source. The oscillation frequency tends to drift upwards
by 1–2 Hz while it is present. For those sources which also show pulsations in
the persistent (accretion-powered) X-ray emission, the frequencies of the persistent
pulsations and burst oscillations are within a few Hz of each other, confirming the
burst oscillation frequency as (roughly) the neutron star spin [24]. Oscillations are
not observed in every burst, but instead are preferentially seen at high accretion rates
[137, 142], when the spectrum is in the “soft” state (see Sect. 5.1.2). Even in those
bursts where they are present, the oscillations are not found throughout the burst,
instead tending to occur most often in the burst rise, and less frequently through to
the burst tail [53].
The mechanism giving rise to the oscillations remains unknown; candidates
include the spreading hotspot model, the related “cooling wake” that follows,
and models involving surface modes of various kinds. The rise in fractional burst
oscillation amplitude in the early phases of the burst generally supports the hotspot
model [25, 176], although once the burning has spread to the entire surface it is
difficult to understand how oscillations can persist. If the cooling timescale is similar
to the spreading timescale, it seems possible that the temperature contrast can also
produce oscillations in the tail [123], although such models have difficulty achieving
the high amplitudes that are observed. Models invoking rotation modes [73] suffer
from difficulties reproducing the observed frequency drifts [13].
In the absence of certainty as to the mechanism, much effort has been focussed
on understanding why oscillations are not seen in every burst. Studies of the
assembled RXTE sample have confirmed that oscillations preferentially occur at