4 Accreting Millisecond X-ray Pulsars
193
J1900.1-2455 has shown burst oscillations in only one burst, when the source
entered an unusually soft state [354]. By contrast five of the six persistent pulsars
with bursts have had burst oscillations in all of their bursts, even though these
sources tend to be in the hard (low accretion rate) state. For the fifth, IGR J174982921, the data are of much poorer quality: oscillations are seen only in the brightest
bursts but the upper limits on the presence of detections in the weaker bursts are
comparable to the amplitudes detected in the brighter bursts [49]. The duration of
burst oscillation trains also appears to differ. For SAX J1808.4-3658, XTE 1814338 and the mildly recycled pulsar IGR J17480-2466, burst oscillations persist
throughout the bursts (except during episodes of Photospheric Radius Expansion
in the bright bursts from SAX 1808.4-3658). For the non-pulsars, although burst
oscillations are sometimes detected throughout bursts they are more commonly
detected in the tails [103], and this is also the case for the intermittent accretionpowered pulsars. In this regard the persistent pulsar IGR J17511-3057 seems to be a
transitional object: at lower accretion rates oscillations are detectable throughout
bursts, but as accretion rate rises the burst oscillation signal vanishes from the
burst rise [4]. Burst oscillations from the pulsars have higher harmonic content
than burst oscillations from the intermittent pulsars and non-pulsars [4, 46, 51, 225,
324, 351, 354]. They also have rather different amplitude-energy relations. Burst
oscillations from the persistent pulsars show the same amplitude-energy relationship
as their accretion-powered pulsations, irrespective of whether the latter rise or fall
with energy but with the latter being more common ([4, 46, 49, 51, 239, 324, 354]
and see also Sect. 4.5.1). The intermittent pulsars and non-pulsars, by contrast,
have burst oscillation amplitudes that rise with energy [226, 353]. The cause of
these various differences in burst oscillation properties between the pulsars and
non-pulsars remains as yet unclear, but the most obvious hypothesis is that the
differences are due to the effects of a dynamically important magnetic field.
One can also use the pulsars to compare the properties of the accretion-powered
pulsations (where the peak of the emission is presumed to be centered on the
magnetic poles) and their burst oscillations. Burst oscillation amplitudes vary
substantially from source to source, but are in most cases within a few percent of
(and most lower than) the amplitude of the accretion-powered pulsations at the time
of the burst [4, 46, 51, 239, 324, 351, 354]. The transitional AMXP/radio pulsar
IGR J18245-2452 is an exception: for the one burst detected, the burst oscillation
amplitude was much higher than the amplitude of accretion-powered pulsations
immediately before the burst [247].
Pulsar burst oscillations show no statistically significant phase lags as a function
of energy [226, 350, 354], unlike accretion-powered pulsations which show soft lags
([67, 99, 110, 111, 163, 243]; see also Sect. 4.5.1). Of particular interest is the strong
phase-locking between accretion-powered pulsations and burst oscillations seen in
XTE J1814-338 and IGR J17480-2466 [46, 352], with very small phase offset (<3 ◦
and 10 ◦ respectively), implying that the hot spots responsible for the two sets of
pulsations are longitudinally coincident in these two sources.
So what do these results tell us about the burst oscillation mechanism ? Although
the mechanism is not yet understood, substantial progress has been made and the
193
J1900.1-2455 has shown burst oscillations in only one burst, when the source
entered an unusually soft state [354]. By contrast five of the six persistent pulsars
with bursts have had burst oscillations in all of their bursts, even though these
sources tend to be in the hard (low accretion rate) state. For the fifth, IGR J174982921, the data are of much poorer quality: oscillations are seen only in the brightest
bursts but the upper limits on the presence of detections in the weaker bursts are
comparable to the amplitudes detected in the brighter bursts [49]. The duration of
burst oscillation trains also appears to differ. For SAX J1808.4-3658, XTE 1814338 and the mildly recycled pulsar IGR J17480-2466, burst oscillations persist
throughout the bursts (except during episodes of Photospheric Radius Expansion
in the bright bursts from SAX 1808.4-3658). For the non-pulsars, although burst
oscillations are sometimes detected throughout bursts they are more commonly
detected in the tails [103], and this is also the case for the intermittent accretionpowered pulsars. In this regard the persistent pulsar IGR J17511-3057 seems to be a
transitional object: at lower accretion rates oscillations are detectable throughout
bursts, but as accretion rate rises the burst oscillation signal vanishes from the
burst rise [4]. Burst oscillations from the pulsars have higher harmonic content
than burst oscillations from the intermittent pulsars and non-pulsars [4, 46, 51, 225,
324, 351, 354]. They also have rather different amplitude-energy relations. Burst
oscillations from the persistent pulsars show the same amplitude-energy relationship
as their accretion-powered pulsations, irrespective of whether the latter rise or fall
with energy but with the latter being more common ([4, 46, 49, 51, 239, 324, 354]
and see also Sect. 4.5.1). The intermittent pulsars and non-pulsars, by contrast,
have burst oscillation amplitudes that rise with energy [226, 353]. The cause of
these various differences in burst oscillation properties between the pulsars and
non-pulsars remains as yet unclear, but the most obvious hypothesis is that the
differences are due to the effects of a dynamically important magnetic field.
One can also use the pulsars to compare the properties of the accretion-powered
pulsations (where the peak of the emission is presumed to be centered on the
magnetic poles) and their burst oscillations. Burst oscillation amplitudes vary
substantially from source to source, but are in most cases within a few percent of
(and most lower than) the amplitude of the accretion-powered pulsations at the time
of the burst [4, 46, 51, 239, 324, 351, 354]. The transitional AMXP/radio pulsar
IGR J18245-2452 is an exception: for the one burst detected, the burst oscillation
amplitude was much higher than the amplitude of accretion-powered pulsations
immediately before the burst [247].
Pulsar burst oscillations show no statistically significant phase lags as a function
of energy [226, 350, 354], unlike accretion-powered pulsations which show soft lags
([67, 99, 110, 111, 163, 243]; see also Sect. 4.5.1). Of particular interest is the strong
phase-locking between accretion-powered pulsations and burst oscillations seen in
XTE J1814-338 and IGR J17480-2466 [46, 352], with very small phase offset (<3 ◦
and 10 ◦ respectively), implying that the hot spots responsible for the two sets of
pulsations are longitudinally coincident in these two sources.
So what do these results tell us about the burst oscillation mechanism ? Although
the mechanism is not yet understood, substantial progress has been made and the