We report the discovery, with NASA's Rossi X-ray Timing Explorer (RXTE), of the first sub-millisecond oscillation found in a celestial X-ray source. The quasi-periodic oscillations (QPO) come from Sco X-1 and have a frequency of approximately 1100 Hz, amplitudes of 0.6-1.2% (rms) and are relatively coherent, with Q up to ∼10 2 . The frequency of the QPO increases with accretion rate, rising from 1050 to 1130 Hz when the source moves from top to bottom along the normal branch in the X-ray color-color diagram, and shows a strong, approximately linear correlation with the frequency of the well-known 6-20 Hz normal/flaring branch QPO. We also report the discovery of QPO with a frequency near 800 Hz that occurs, simultaneously with the 1100 Hz QPO, in the upper normal branch. We discuss several possible interpretations, one involving a millisecond X-ray pulsar whose pulses we see reflected off accretion-flow inhomogeneities. Finally, we report the discovery of ∼45 Hz QPO, most prominent in the middle of the normal branch, which might be magnetospheric beat-frequency QPO.

Aquila X-1 is a soft X-ray transient source and emits type I X-ray bursts. A spectral state transition was observed with RXTE during its outburst decay in 1997 February and March. Its 10-30 keV and 5-10 keV count rate ratio increased suddenly when its luminosity was between 4 and ergs s Ϫ1 , assuming a 2.5 kpc 35 12 # 10 distance. Spectral fitting with a model composed of a blackbody and a power-law component showed that its blackbody component decreased and that the power-law component became much harder significantly and simultaneously. We interpret this transition to be caused by the centrifugal barrier, more commonly known as the "propeller" effect. We thus infer that the magnetic field strength of the neutron star is around G, if the 8 1 # 10 neutron star spin period is 1.8 ms. Similarly, we infer the neutron star magnetic field strength in another soft Xray transient Cen X-4 is about G. We also propose a unified scheme for spectral state transitions in soft 9 2 # 10 X-ray transients, from soft high state to hard low state and further to quiescent state. With this scheme accretion onto neutron star may take place even during the propeller regime.

We present results of RXTE observations of the low-mass X-ray binary and atoll source 4U 1608−52 made over 9 days during the decline of an X-ray intensity outburst in March 1996. A fast-timing analysis shows a strong and narrow quasi periodic oscillation (QPO) peak at frequencies between 850 and 890 Hz on March 3 and 6, and a broad peak around 690 Hz on March 9.Observations on March 12 show no significant signal. On March 3, the X-ray spectrum of the QPO is quite hard; its strength increases steadily from 5 % at ∼2 to ∼20 % at ∼12 keV. The QPO frequency varies between 850 and 890 Hz on that day, and the peak widens and its rms decreases with centroid frequency in a way very similar to the well-known horizontal branch oscillations (HBO) in Z-sources. We apply the HBO beat frequency model to atoll sources, and suggest that, whereas the model could produce QPOs at the observed frequencies, the lack of correlation we observe between QPO properties and X-ray count rate is hard to reconcile with this model. Subject headings: stars: individual (4U 1608−52) -stars: neutron -accretion, accretion disks However, in spite of these arguments, the absence of a correlation between X-ray intensity on the one hand and QPO frequency and rms amplitude on the other hand poses a major problem for a beat frequency model interpretation. The QPO frequency remains constant near 850 Hz between 3 and 6 March while the count rate drops by more than a factor 4, from ∼3200 to ∼600 counts/sec (Sect. 2). The beat frequency model predicts that if the QPO frequency remains constant, the mass flow through the inner edge of the disk should remain constant as well. Our data therefore are inconsistent with a beat frequency model interpretation if all accretion takes place via the inner edge of the disk. We note that the presence of a hypothetical additional (non-disk) mass flow component contributing more than 75% of the total flux on March 3 and much weaker on March 6 can not easily resolve this discrepancy. The rms amplitude of the QPO only increases by a factor 2 from March 3 to 6 while the X-ray intensity drops by a factor of more than 4, whereas a similar fractional change would be predicted in this explanation. Only by invoking rather large and entirely ad-hoc changes in the beaming or bolometric correction could the model be maintained. The 150 Hz drop in QPO from March 6 to March 9 without a change in count rate presents similar difficulties. Further observations of 4U 1608−52 during other outbursts, and study of its X-ray bursts are required to shed further light on the relation of the 800 Hz QPO in 4U 1608−52 with those in 4U 1728−34, and with the QPO in Sco X-1. Sztajno M. 1985, Nature 316, 225 van der Klis M., 1989. NATO ASI C262: Timing Neutron Stars,Ögelman and van den Heuvel (eds.), Kluwer, p. 27. van der Klis M. 1995, in X-ray Binaries, Van Paradijs and Van den Heuvel (eds.), Cambridge University Press, p. 252.

No abstract