5 Thermonuclear X-ray Bursts
231
The effects of Poynting-Robertson drag arising from the increase in X-ray flux
during the burst have been predicted to lead to an increase in the accretion rate
(and hence the persistent flux component) during the burst ([191]; see also [111]).
This effect is potentially at odds with the conventional approach for time-resolved
spectroscopic analysis, which involves subtracting off the pre-burst emission (usually incorporating both the persistent emission and instrumental background).
This approach implicitly assumes that the burst and persistent component are
independent, even late in the tail of the burst, which cannot be completely true (e.g.
[108]). Nevertheless, the net burst spectrum is then usually well-fit with a blackbody
(e.g. [180]).
The burst sample accumulated by RXTE over its mission offers the most
stringent test of this conventional analysis approach. A study of 332 PRE bursts
observed from 40 sources found significant evidence for an increase in the persistent
contribution to the flux early in the bursts [203]. This increase, up to a factor of
20, was interpreted as the expected increase in the accretion rate in response to
Poynting-Robertson drag. It is somewhat surprising that this effect is seen in PRE
bursts, since the luminosity of those bursts should exceed the Eddington limit at
the peak, temporarily halting accretion. The effect has also been confirmed in an
even larger sample of 1759 non-PRE bursts from 56 sources [204], although at
with smaller increase factors. Additionally, a bright burst from SAX J1808.4−3658
observed simultaneously with Chandra and RXTE finds even more compelling
evidence for increases in the accretion component during the burst [82], from the
much broader band-pass (≈0.5–30 keV) available for the two instruments combined
(Fig. 5.7).
Unfortunately the available data is insufficient to test for possible changes in
the shape of the persistent spectrum accompanying the burst-induced increase in
accretion rate, which might be expected in parallel to the spectral shape variations
that are seen on longer timescales (see e.g. Sect. 5.1.2). Furthermore, while the
modified analysis approach improves the combined distribution of fit statistics for
the ensemble, it does not yet achieve the expected distribution for a well-fitting
model, suggesting that the net burst spectrum still deviates significantly from a
blackbody. As we have discussed, these deviations include the recombination edge
features observed in very energetic bursts (e.g. [76]; see also Sects. 5.7 and 5.3.2).
The residuals also appear inconsistent with the more modest deviations predicted
by spectral models (see Sect. 5.3.1). The question of the true shape of the burst
spectrum over the full range of burst types remains open, with the available data not
sufficiently sensitive to clearly measure deviations from a blackbody. However, with
the analysis of the RXTE sample, the key contributions of variations in the persistent
contribution, discrete features, as well as the effect of reflection from the accretion
disk (Sect. 5.4.1) are now well-established.
231
The effects of Poynting-Robertson drag arising from the increase in X-ray flux
during the burst have been predicted to lead to an increase in the accretion rate
(and hence the persistent flux component) during the burst ([191]; see also [111]).
This effect is potentially at odds with the conventional approach for time-resolved
spectroscopic analysis, which involves subtracting off the pre-burst emission (usually incorporating both the persistent emission and instrumental background).
This approach implicitly assumes that the burst and persistent component are
independent, even late in the tail of the burst, which cannot be completely true (e.g.
[108]). Nevertheless, the net burst spectrum is then usually well-fit with a blackbody
(e.g. [180]).
The burst sample accumulated by RXTE over its mission offers the most
stringent test of this conventional analysis approach. A study of 332 PRE bursts
observed from 40 sources found significant evidence for an increase in the persistent
contribution to the flux early in the bursts [203]. This increase, up to a factor of
20, was interpreted as the expected increase in the accretion rate in response to
Poynting-Robertson drag. It is somewhat surprising that this effect is seen in PRE
bursts, since the luminosity of those bursts should exceed the Eddington limit at
the peak, temporarily halting accretion. The effect has also been confirmed in an
even larger sample of 1759 non-PRE bursts from 56 sources [204], although at
with smaller increase factors. Additionally, a bright burst from SAX J1808.4−3658
observed simultaneously with Chandra and RXTE finds even more compelling
evidence for increases in the accretion component during the burst [82], from the
much broader band-pass (≈0.5–30 keV) available for the two instruments combined
(Fig. 5.7).
Unfortunately the available data is insufficient to test for possible changes in
the shape of the persistent spectrum accompanying the burst-induced increase in
accretion rate, which might be expected in parallel to the spectral shape variations
that are seen on longer timescales (see e.g. Sect. 5.1.2). Furthermore, while the
modified analysis approach improves the combined distribution of fit statistics for
the ensemble, it does not yet achieve the expected distribution for a well-fitting
model, suggesting that the net burst spectrum still deviates significantly from a
blackbody. As we have discussed, these deviations include the recombination edge
features observed in very energetic bursts (e.g. [76]; see also Sects. 5.7 and 5.3.2).
The residuals also appear inconsistent with the more modest deviations predicted
by spectral models (see Sect. 5.3.1). The question of the true shape of the burst
spectrum over the full range of burst types remains open, with the available data not
sufficiently sensitive to clearly measure deviations from a blackbody. However, with
the analysis of the RXTE sample, the key contributions of variations in the persistent
contribution, discrete features, as well as the effect of reflection from the accretion
disk (Sect. 5.4.1) are now well-established.