«Discovery of the ﬁrst accretion-powered millisecond X-ray pulsar Rudy Wijnands∗ & Michiel van der Klis∗† ∗ Astronomical Institute “Anton ...»
Discovery of the ﬁrst accretion-powered millisecond X-ray pulsar
Rudy Wijnands∗ & Michiel van der Klis∗†
Astronomical Institute “Anton Pannekoek”, University of Amsterdam, & Center for High Energy
Astrophysics, Kruislaan 403, 1098 SJ Amsterdam, The Netherlands
Department of Astronomy, University of California, Berkeley, Berkeley, CA 94720
Sent to Nature on April 20, 1998
The precise origins of the millisecond radio pulsars, discovered in the early 1980s1, remain uncertain until this day. They plausibly evolve from accreting low magnetic-ﬁeld neutron stars in X-ray binary systems2,3. If so, these stars should spin at millisecond rates. In accordance with this idea, quasi-periodic oscillations discovered in X-ray binaries around 50 Hz4 and 1 kHz5,6, and drifting oscillations at several 100 Hz in X-ray bursts6 have all been interpreted in terms of millisecond spins of weakly magnetized neutron stars7,8. However, in 15 years of searching9,10,11,12, the expected coherent millisecond signals from X-ray binaries remained elusive. In this Letter, we report the discovery13 of the ﬁrst example of such a signal. Using the Rossi X-ray Timing Explorer we ﬁnd persistent 2.49 millisecond X-ray pulsations in an X-ray binary, which we interpret to come from an accretion-powered millisecond X-ray pulsar in the system. This is the ﬁrst known object of its kind. It is likely to switch on as a millisecond radio pulsar when the accretion turns oﬀ completely. The source is positionally coincident14,13 with the known transient X-ray burster SAX J1808.4–365815, which also makes this the ﬁrst X-ray pulsar which exhibits thermonuclear X-ray bursts.
The transient X-ray burster SAX J1808.4–3658 was discovered in September 1996 with the Wide Field Cameras onboard BeppoSAX15. Two thermonuclear X-ray bursts were observed, and the source was classiﬁed as a low-mass X-ray binary (LMXB) at a probable distance of 4 kpc15.
Recently, a transient source, positionally coincident with SAX J1808.4–3658 to within a few arcmin, was detected14 with the Proportional Counter Array (PCA) onboard the Rossi X-ray Timing Explorer (RXTE), showing the source was again in outburst. The 2–10 keV ﬂux on 1998 April 11 was 1.5 × 10−9 erg cm−2 s−1.
Data with a time resolution of 1/8192 s were taken with the PCA on 1998 April 11. The 2–60 –3– keV count rate varied between 892 and 1007 counts/sec, including a ∼130 c/s background; no X-ray bursts were observed. We computed a power spectrum of ∼3000 s of X-ray data. The spectrum clearly shows the X-ray pulsations near a frequency of 401 Hz (Fig. 1). Their amplitude was ∼4.1% (rms) in the 2–60 keV range, with little dependence on photon energy.
We then divided the data into 128 second segments and determined the pulsation frequency in each segment by an epoch folding technique. The signal was present in each segment. During our initial 2-hr observing span the pulsation frequency showed a smooth, approximately sinusoidal ∼110-min variation between 400.986 and 401.022 Hz in the satellite frame. We corrected for this smooth variation and found we could coherently fold the pulsations. Attributing the smooth frequency variation to Doppler shifts, in part due to the satellite motion, we concluded that the intrinsic coherence of the signal exceeded 106, leaving no other interpretation than the presence of a millisecond X-ray pulsar in the system13. We conﬁrmed the presence of the signal in later observations taken between 1998 April 13–18. The Doppler shifts clearly show up in a dynamical power spectrum (Fig. 2).
The folded pulse proﬁle (Fig. 3) is near-sinusoidal, with amplitudes of the harmonics at 2 and 3 times the 401 Hz frequency of ∼11 and ∼4% of the 401 Hz amplitude, respectively. This is similar to what is observed16 in the 0.5 sec pulsar GRO J1744–28 and might be related to both systems having a low inclination. It is possible that the spin period is 4.98 ms with two near-identical pulses per cycle. However, folding at this period indicates no signiﬁcant diﬀerence in shape between the odd and even pulses. The folded light curves in the 2–5.0, 5.0–8.6 and 8.6–60 keV photon energy bands are not signiﬁcantly diﬀerent.
pulsars as well as of low-luminosity LMXBs17. There is a strong (∼20% rms, 2–60 keV) broad noise component below 100 Hz, with a bump near 12 Hz that, when interpreted in terms of the magnetospheric beat-frequency model7 would suggest an inner disk radius of 30 km (just inside the corotation radius). An additional ∼17% (rms) noise component extends out to roughly the 0.4 kHz pulse frequency. This is not commonly seen in LMXBs, but is, at proportionally lower cutoﬀ frequencies, in other accreting pulsars17. No kilohertz quasi-periodic oscillations18 are detected.
However, upper limits (3–6% rms), when compared to observed amplitudes in other LMXBs, are as yet insuﬃcient to exclude their presence.
The detection of the pulsations from an X-ray transient that, moreover, likely showed thermonuclear X-ray bursts during an earlier outburst, identify this X-ray pulsar as one that is powered by accretion from a binary companion, and pulses because magnetically channeled accretion produces hot spots on the neutron star surface which spin around with the star’s spin.
In a subsequent analysis Chakrabarty and Morgan19,20 have found that, correcting for the satellite orbit, the pulsar’s Doppler shifts indicate it is in a 2-hr orbit around a low-mass star, with an intrinsic pulse frequency of 400.9753(1) Hz19.
For accretion to proceed uninhibited by the centrifugal force on the accreting matter corotating in the magnetosphere, the magnetic ﬁeld strength B of this pulsar must be much lower than the
where µ26 is the magnetic moment in 1026 Gauss cm3, L37 the bolometric luminosity in 1037 erg/s, m the neutron star mass in solar masses, R6 its radius in units of 106 cm, ξ a dimensionless factor that is probably between 0.5 and 121, and P1ms the spin period in ms. On 1998 April 11 the 2–10
hard22, and the bolometric luminosity may have been closer to 6 × 1036 erg s−1. With the above expressions this leads to an upper limit on B of ∼2–6×108 Gauss (4–14×108 Gauss if the pulse period is 4.98 ms). As the X-ray ﬂux decreases in the transient decay this upper limit will become less. When centrifugal inhibition sets in we might see a rapid drop in X-ray ﬂux23,24 ; this would allow to estimate the value of B.
Such a low magnetic ﬁeld strength is in the range of values inferred for millisecond radio pulsar25. The object appears to be located above the pulsar “death line”25, so when accretion ﬁnally turns oﬀ, this source would likely switch on as a radio pulsar. This might even happen between X-ray outbursts, if the density in the pulsar environment, at least out to the light cylinder at 120 km, drops suﬃciently. So, this low-mass X-ray binary could indeed be one of the progenitors of the millisecond radio pulsars, and it might be an intermittent radio pulsar now.
That this pulsar likely also bursts is in accordance with theoretical expectations. Normally, X-ray bursts do not occur in accreting pulsars, as due to the magnetic ﬁeld either the nuclear burning takes place steadily26, or nuclear burning fronts can not propagate rapidly27. However, if the ﬁeld is weak, these mechanisms are ineﬀective and bursts could occur21. This seems to be veriﬁed by SAX J1808.4–3658. This is the ﬁrst time pulsations and thermonuclear burst are seen in one source. The bursts in the 0.5 s pulsar GRO J1744–2816,28 are not thermonuclear but due to accretion instabilities28,29.
As the likely B ﬁeld of SAX J1808.4–3658 is not much diﬀerent from that inferred for other LMXBs from models for the rapid aperiodic variability7 and the X-ray spectra30, and its spin rate is similar to that inferred in other LMXBs from models for burst oscillations6 and kHz QPOs8 the
it can not be excluded that at higher luminosity SAX J1808.4–3658, like the other LMXBs, would not pulse, for LMXBs of similar luminosity the simplest explanation would seem to be that their B ﬁelds are considerably weaker than that of this pulsar. It would be of great interest to detect in SAX J1808.4–3658 the drifting oscillations during X-ray bursts interpreted in other LMXBs as due to the neutron star spin6, or the twin kHz QPO peaks whose frequency diﬀerence has also been interpreted as the neutron star spin frequency5,6,8, as the fact that the spin frequency of SAX J1808.4–3658 is known will provide a direct test of these interpretations.
We ﬁnally note that the high-precision timing measurements possible due to the high frequency of these pulsations will allow to study the torques of the accreting matter on the neutron star and the star’s response to these torques with unprecedented accuracy. Moreover, for the ﬁrst time this will be possible in an accretion ﬂow that extends freely down to only a few tens of kilometers from the neutron star surface. It is likely that this will provide new insights into the nature of the inner accretion ﬂows around weakly magnetized compact objects. It may also provide new information
1. Backer, D.C., Kulkarni, S.R., Heiles, C., Davis, M.M., & Goss, W.M. A millisecond pulsar.
Nature 300, 615 (1982)
2. Srinivasan, G. & van den Heuvel, E.P.J. Some constraints on the evolutionary history of the binary pulsar PSR 1913+16. Astr. Astrophys. 108 143 (1982)
3. Alpar, M.A., Cheng, A.F., Ruderman, M.A., & Shaham, J. A new class of radio pulsars. Nature 300, 728 (1982)
4. van der Klis, M., et al. Intensity-dependent quasi-periodic oscillations inthe X-ray ﬂux of GX 5−1. Nature 316, 225 (1985)
5. van der Klis, M. et al. Discovery of submillisecond quasi-periodic oscillations in the X-ray ﬂux of Scorpius X-1. Astrophys. J. Lett. 469, L1 (1996)
6. Strohmayer, T. E. et al. Millisecond X-ray variability from an accreting neutron star system.
Astrophys. J. Lett. 469, L9 (1996)
7. Alpar, M.A. & Shaham, J. Is GX 5–1 a millisecond pulsar? Nature 316, 239 (1985)
8. Miller, M.C., Lamb, F.K., & Psaltis, D. Sonic-point model of kilohertz QPOs in LMXBs.
Astrophys. J. Submitted (1998)
9. Leahy, D.A. et al. On searches for pulsed emission with application to four globular cluster X-ray sources: NGC 1851, 6441, 6624, and 6712. Astrophys. J. 266, 160 (1983)
10. Mereghetti, S. & Grindlay, J.E. A search for millisecond periodic and quasi-periodic pulsations in low-mass X-ray binaries. Astrophys. J. 312, 727 (1987)
11. Wood, K.S. et al. Searches for millisecond pulsations in low-mass X-ray binaries. Astrophys. J.
379, 295 (1991)
Astrophys. J. 435, 362 (1994)
13. Wijnands, R. & van der Klis, M. SAX J1808.4–3658 = XTE J1808-369. IAU Circ. No. 6876 (1998)
14. Marshall, F.E. SAX J1808.4–3658 = XTE J1808–369. IAU Circ. No. 6876 (1998)
15. in ’t Zand, J.J.M. et al. Discovery of the X-ray transient SAX J1808.4–3658, a likely low-mass X-ray binary. Astr. Astrophys. Lett. 331, L25 (1998)
16. Finger, M.H. et al. Discovery of hard X-ray pulsations from the transient source GRO J1744–28.
Nature 381, 291 (1996)
17. van der Klis, M. Rapid aperiodic variability in X-ray binaries. In X-ray binaries (eds. Lewin, W.H.G., van Paradijs, J., & van den Heuvel, E.P.J.) 252 (University Press, Cambridge) (1995)
18. van der Klis, M. Kilohertz quasi-periodic oscillations in low-mass X-ray binaries - a review.
In The many faces of neutron stars (eds. Buccheri, R., van Paradijs, J., & Alpar, M.A.) (Kluwer Academic Publisher) in press (1998)
19. Chakrabarty, D. & Morgan, E.H. SAX J1808.4–3658 = XTE J1808–369. IAU Circ. No. 6877 (1998)
20. Chakrabarty, D. & Morgan, E.H. The 2 hour orbit of the bursting millisecond X-ray pulsar SAX J1808.4–3658. Nature submitted (1998)
21. Bildsten, L. & Brown, E. F. Thermonuclear burning on the accreting X-ray pulsar GRO J1744–
28. Astrophys. J. 477, 897 (1997)
22. Heindl, W., Marsden, D. & Blanco, P. SAX J1808.4–3658 = XTE J1808–369. IAU Circ. No.
23. Zhang, S.N., Yu, W., & Zhang, W. Spectral state transition in Aquila X-1: evidence for
24. Campana, S. et al. Aquila X-1 from outburst to quiescence: the onset of the propeller eﬀect and signs of a turned-on rotation-powered pulsar. Astrophys. J. Lett. in press (1998)
25. Bhattacharya, D. & van den Heuvel, E.P.J. Formation and evolution of binary and millisecond radio pulsars. Physics Reports 203, 1 (1991)
26. Joss, P.C. & Li, F.K. Helium-burning ﬂashes on accreting neutron stars: Eﬀects of stellar mass radius, and magnetic ﬁeld. Astrophys. J. 238, 287 (1980)
27. Bildsten, L. Propagation of nuclear burning fronts on accreting neutron stars: X-ray bursts and sub-hertz noise. Astrophys. J. 438, 852 (1995)
28. Kouveliotou, C. et al. A new type of transient high-energy source in the direction of the Galactic Centre. Nature 379, 799 (1996)
29. Lewin, W.H.G., Rutledge, R.E., Kommers, J.M., van Paradijs, J., & Kouveliotou, C. A comparison between the Rapid Burster and GRO J1744–28. Astrophys. J. Lett. 462, L39 (1996)
30. Psaltis, D., Lamb, F.K., & Miller, G.S. X-ray spectra of Z sources. (Astrophys. J. Lett. 454, L137 (1995) Acknowledgements. We thank Lars Bildsten and Jan van Paradijs for carefully reading the manuscript, them and Jon Arons for stimulating discussions, and the RXTE team for promptly making the data available online. MK gratefully acknowledges the Visiting Miller Professor Program of the Miller Institute for Basic Research in Science (UCB). We thank the Netherlands Organization
Fig. 1.— Leahy9 normalized power spectrum of the 1998 April 11 20:38–21:21 UT persistent emission of SAX J1808.4–3658. The periodicity near 401 Hz is obvious. No harmonics or subharmonics are
Fig. 2.— Dynamical power spectrum of the 1998 April 11–18 data. The Doppler shifts in pulsation frequency due to the satellite and pulsar orbital motion are evident. Time axis is not contiguous;