Astron. Astrophys. 357, L21-L24 (2000)

4. Discussion

In addition to the thermonuclear X-ray bursts, also called type I bursts, low-mass X-ray binaries show other sudden enhancements in X-ray flux. Type II bursts are different from type I bursts in that type II bursts do not show cooling of the characteristic temperature of the X-ray spectrum during the decline. X-ray flares have an irregular flux evolution. Type II bursts are thought to be accretion events; the nature of flares is unknown.

The flux enhancement of 4U 1735-44 shows a smooth exponential decay of the countrate and of the characteristic temperature. Its rise must have been shorter than the decline. A black body gives a good fit to the observed spectrum, for a radius as expected from a neutron star, similar to earlier, ordinary bursts of 4U 1735-44. All these properties indicate a type I burst. The only special property of the new burst is its duration, which when expressed as the ratio of fluence and peak flux : s, is more than 300 times longer than the longest burst observed previously from this source (see Lewin et al. 1980). This duration also translates in a fluence which is several orders of magnitude larger than the previous record holder for 4U 1735-44, because the peak flux is similar to those of normal type I bursts. The fluence of a type I burst which burns all matter deposited onto a neutron star since the previous burst must be 1% of the accretion energy released by deposition of this matter. We do not have a measurement to the previous burst, but in seven days following the burst no other burst was observed. Multiplying this time by the persistent luminosity we obtain erg, or about 50 times the energy of the burst, well in the range of previously observed ratios for type I bursts.

The presence of clear cooling argues against a type II burst; this and the smooth decay argues against a flare. If the flux enhancement were due to an accretion event, the amount of matter dropped extra onto the neutron star (assuming a mass of and a radius of 10 km) must have been g, which may be compared to the average accretion rate of g s^-1 derived for the persistent flux. If the inner part of the accretion disk would have depleted itself onto the neutron star during the flux enhancement, one would expect the accretion rate immediately after to be lower than before. The observations suggest the opposite.

We conclude that a type I X-ray burst is the best explanation for the enhanced flux event. We consider it significant that the occurrence of this burst is accompanied by the absence of any ordinary - i.e. short - burst throughout our 9-day observation, whereas all previous observations of 4U 1735-44 did detect ordinary bursts (see Introduction).

Searching the literature for long bursts we find that the longest type I burst published previously is a radius expansion burst observed with SAS-3, probably in 4U 1708-23 (Hoffman et al. 1978; see also Lewin et al. 1995). The ratio of fluence and peak flux for that burst was s, so that the BeppoSAX WFC burst of 4U 1735-44 lasted at least six times longer. Other events published as long bursts from Aql X-1 (Czerny et al. 1987) and from X 1905+000 (Chevalier and Ilovaisky 1990) are in fact relatively short bursts followed by an enhanced constant flux level which persisted for several hours: in both cases the flux declined to 1/e of the peak level within 20 s. These events are clearly different from the long exponential bursts seen in 4U 1708-23 and 4U 1735-44.

From the theoretical point of view, a long interval between bursts would allow hydrogen to burn completely before the onset of the burst, so that the energetics of the burst is dominated by pure helium burning. If matter accreted at a rate of g s^-1 during one week, the energy released by helium burning is compatible with the energy of the observed burst. The problem with this model is that theory predicts for this accretion rate that the burst initiates well before hydrogen burning is completed, i.e. that bursts are more frequent and less energetic, in accordance with those previously observed of 4U 1735-44. Indeed, Fujimoto et al. (1987) find that a burst of s duration occurs only for accretion rates . The persistent flux during the BeppoSAX observation is a factor higher than this limit; observations previous to ours have consistently found 4U 1735-44 at a similar luminosity.

An alternative model for bursts with a duration of s is accretion of pure helium at an accretion rate in excess of the Eddington limit (, Brown & Bildsten 1998). The orbital period and optical spectrum indicate a main-sequence, i.e. hydrogen-rich, donor star (Augusteijn et al. 1998).

Perhaps the main challenge for any theoretical explanation is that the properties of the persistent flux during our nine day long observation, during which a single very long X-ray burst was observed, are not different from those during earlier observations with EXOSAT when more frequent ordinary bursts were found.

© European Southern Observatory (ESO) 2000

Online publication: May 3, 2000
helpdesk.link@springer.de