Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Quasi-Per~odic Oscillations in Low-Mass X-ray Binaries W.H.G. LEWIN*, J. VAN PARADI3St, AND M. VAN DER KLIST Variability on short time scales in the X-ray flux (Lewin et al. 1968) Is a very general property of binary X-ray sources. Until recently, most efforts in studying such variability were spent on two types of intensity variations, periodic X-ray pulsations and X-ray bursts. Due to the lack of a direct and transparent interpretation, relatively little attention was paid to noise in X-ray intensity vanations. An outstanding exception has been the study of the very fast variability of the black-hole candidate Cyg X-1 and sources long thought to be of a similar nature such as Cir X-1 and GX 3394 (for a comprehensive review of the studies of noise in X-ray intensity variations of binary X-ray sources up to 1980, we refer to Bradt et at 1982~. Consequently the bright persistent X-ray sources in the central regions of the Galaxy (the galactic bulge sources) which showed neither pulsations nor bursts were somewhat neglected. Not until after the discovery (Van der Klis e' al. 1985) of intensiW-dependent quasi-penodic oscillations (QPO) and associated red noise from these luminous low-mass X-ray binaries (see Figure 1) were systematic studies of the shape of these power-spectral components made. In this note we give a brief account of the main developments since this discovery which have led to a new picture of the properties of LMXB. Since this is not intended to be a review paper, we will not give extensive references here but instead refer the reader to We *Center for Space Research and Department of Physics, Massachusetts Institute of Technology 37~27, Cambridge, MA 02139, USA t Astronomical Institute "Anton Pannekoek", University of Amsterdam, Roetersstraat 15, 1018 WB Amsterdam, The Netherlands and Center for High-Energy Astrophysics, NIKHEF-H, Am- sterdam, The Netherlands 251
252 AMERICAN AND SOVIET pERspEcTrvEs comprehensive reviews of Lewin et al. (1988), Lamb (1988, 1990), Van der Klis (1989), and a recent article by Hasinger and Van der Klis (1989). Slow QPO (frequencies In the range one to tens of milk-Hertz) had been reported prior to 1985 from several sources. It is likely that these QPO have a different origin from those which are the subject of this note. Short trains of fast oscillations during a number of type ~ X-ray bursts (i.e., runaway thermonuclear events) had been reported (for a review see Lewin and Joss 1983~. The association of these oscillations with X-ray bursts makes it difficult to compare them with the QPO seen in the persistent flux of galactic bulge X-ray sources. Sara et al. (1982) discovered quasi-periodic (A 2 Hz) oscillations in 2 out of 64 long type II X-ray bursts from the Rapid Burster. The frequency of these oscillations differed between the bursts and drifted within each burst, so Sara e! al. concluded that they could not be a direct manifestation of the rotation of the neutron star. There were at least two reasons why this discovery of QPO in the Rapid Burster received much less attention than the QPO discovery in GX 5-1 a few years later. (i) The Rapid Burster was (and still is) a very peculiar source (Lewin and Joss 1983), and the 2-Hz QPO were seen as "just another" strange phenomenon. (ii) The frequency of the QPO in GX 5-1 depended strongly on the source intensity; there were no obvious connections between the QPO in the Rapid Burster and other characteristics of this source. Alpar and Shaham (19$5) proposed that the intensity-dependent QPO discovered in GX 5-1 (Van der Klis et al. 1985) are caused by some interaction between the magnetic field of the rapidly rotating neutron star and the inner accretion disk (bounded by the magnetosphere) in which matter orbits the neutron star in appro~nateh,r Keplerian orbits. The QPO frequency is then the difference ("beat frequency'') between the Keplerian frequency at the inner disk edge and the neutron star spin frequency. Modulation of the accretion rate through magnetic gating was proposed as a feasible mechanism for causing the X-ray intensity variations, and it was shown that the red-noise component was a natural consequence of accretion-modulation mechanism (Lamb e! al. 1985~. From the observed intensity dependence of the QPO frequency, Alpar and Shaham derived a neutron star rotation period of 10 ms and a magnetic dipole field strength of a few 109 G. These parameters agree very well with the spin-up scenario for the binary msec radio pulsars. In this scenario the msec radio pulsars are the descendants of low-mass X-ray binaries (see, e.g., On den Heuvel 1986 for a renew of the evolution of X-ray binaries). For this reason the "beat frequenter" model of Alpar and Shaham was greeted with enthusiasm. However, if 10 me were the correct rotation period, it was somewhat purling why this period did not show up as coherent pulsations in the
HIGH-ENERGY ASTROPHYSICS 3.8 3.4 3.0 l 2.6 2.2 1.8 3.0 2.6 2.2 1.8 3.0 Con 2.6 3 o 2.2 1 .8 3.0 2.6 2.2 1.8 3.0 it t1 .: .1 ~';~1 t, ",' 1- ~ 'I J in 2277-2486 c/s ~ ~l it: U 1 2486-2695 C/S 2695-2904 c/s 2904-3113 c/s 1.8 3.0 2.6 2.2 r. 3113-3322 c/s 3322-3531 C/S 1.8 ~ ~ ~'111 u up v II I AL ~',-3 ~ ~ ~ - ~ I L an 11 0 10 20 30 40 50 60 70 80 90 100 FREQUENCY ( Hz ) 253 FIGURE 1 Display of average power spectra from GX 5-1 in six different source intensity integrals Indicated in the panels3. The lines drawn through the data indicate fits described lay a function which is the sum of a constant (Poisson noise), a line with a Lorentzian profile (the QPO peak), and an exponentially rising red noise component. This figure is Mom Van der Klis et al. 1985.
2~4 AMERICAN AND SOVIET PERSPECTIVES power spectrum since the existence of a magnetosphere implies that in the vicinity of the neuron star the magnetic field influences the accretion flow. QPO discovered subsequently in Sco X-1 (Middleditch and Priedhorsly 1986; Van der Klis et al. 1986) had quite different properties. Their frequency was sometimes low (near 6 Hz) and appro~natel!,r constant or even slightly anti-correlated with X-ray intensity, while at other times, it was high (10-20 Hz) and either positively correlated or varying erratically as a function of source intensity depending on the intensity of the source. Red noise was weak compared to the QPO. Both results seemed to pose a problem for the beat-frequency model. The "beat-frequenc~r" model was clearly unable to account for the complicated picture presented by the observations; alternative models were proposed and ways were worked out to broaden the range of phenomena that could be explained within the beat-frequency model. An underlying organization in the phenomena revealed itself when correlations were found between the spectral properties of the sources and their QPO characteristics. Three of the newly discovered QPO sources were known to exhibit two different X-ray spectral states, distinguishable as "branches" in an X-ray hardness versus intensity diagram. In OX 5-1 the 2040 Hz QPO were only and always observed when the source was found in the so-called "horizontal-branch" of this diagram (Van der Klis et al. 1987, see Figure 23. Its red noise was shown to consist of two components; one of these is the low-frequenc~r noise associated with the QPO and only present in the honzontal-branch state; the other ("very low frequency noise") is a power-law component that Is seen in all specual states, and dominates the power spectrum below ~ 0.1 HE A similar pattern was observed in Cyg X-2 (Hasinger 1987a). In Sco X-1 a strong correlation was also found between QPO and the spectral state; two-branched spectral behavior occurred with strongly intensity-dependent high-frequency QPO in one spectral branch and weakly intensity-dependent ~ 6 Hz QPO in the other. However, bow the morphology of the branches and the way in which the QPO frequency varied among them were quite different from that In OX 5-1 and Cyg X-2. A key contribution came from Hasinger (19~) who observed that the normal branch of Cyg X-2, known to be connected at its upper end to the horizontal branch, showed a sharp bend near its lower end suggesting a transition to yet another branch. Hasinger, therefore, proposed that there are three branches: the horizontal branch, the normal branch, and the flaring branch which form a Zshaped pattern in the X-ray hardness vs. intensity diagram and each of which has its characteristic power spectral behavior. In OX 5-1 and Cyg X-2 it was the upper part of the Z that had so far been observed; in Sco X-1 it was the lower part. Subsequent observations have beautifully confirmed Hasinger's proposal, and the class of "Z sources" is now well established (see Figure 3~.
HIGH-ENERGY ASTROPHYSICS 255 r I 0.70 o 060 z C) CY 0.50 ~: I 0.40 TYPI CAL STAT I STI CA L ERROR BAR · . . NB ° ° o o o o o o ° o o o o 8 o ° o ° ° 8 o ° o o o o o o . HB : ..~::~:.,.: ..· o o o o o o ° o o o o o o o o 80 100 120 140 160 180 200 220 I NTE NSITY (c/s) FIGURE 2 Spectral hardness (ratio of the 6-10 keV and 3~ keV count rate) vemus intensi~ (3-10 keV) diag~m for some EXOSAT data on GX 5-1. The solid dots in the "honzontal branch" (HB) represent data when strong 2040 Hz QPO were observed; the power spectra represented by open circles on the "normal branch" (NB) showed no significant high-frequengy QPO. Ihis figure is ~om Van der Klis et al. (1987~. W~th this clanfication, attempts to explain all obse~ved QPO behavior within the framework of the beat frequenc~y model were abandoned, and the model was proposed to be valid onl~r for the honzontal-branch high- frequengy QPO (and associated LFN). Models for the 6 Hz normal-branch QPO have been proposed in terms of radiation-dominated accretion Dows near the Eddington limit ~amb 198S, 1990, Hasinger 1987a). Millisecond time lags were discovered in the intensity-dependent hon
256 G.75 . : a_ V ~ O. AMERICAN AND SOVIET PERSPECTIVES Cyg x-2 0 . 7O 0.65 0.60. s5~ , .,, ., , . . . 11B - ..,, · ~ .\~ .. e ~' ~ ~ S. . I:I] . _ O.50 ; 0 75 C .80 0.85 0. SO 1 o2 - U2 Am ~10 1 Al 10 o 10 10 - 1 O ,., -. - --- --''''' '' ~ ~ Cyg X-Z ~ ~ ·1 ~ ~! L_ 1~' '. . . ~ 10-3 10-2 1o-l 10 ° 1ol 1o2 GX 17+2 0.70~ ~lII3` 0.60 a. 50 SOFT COLOR 1 1 ol 10 o 1 O 1 o-2 1 o-3 1 ~4 Or: ,~ FIN + 0. 35 C.~O 0.45 0.50 . . NLFX GX 17+2 l LEN \~ _1 W~0> At 1 :1 1~5 . 1' 10-3 10-2 io-i 10 ° 101 ~o2 FREQUENCY (Hz) FIGURE 3 Display of the X-ray color color diagrams and power spectra for the Z sources C;yg X-2 and GX 17 + 2 illustrating the Correlation between the QPO behavior and the location of the source on the Z-shaped track This figure has been adapted from Van der Klis (1989~. zontal-branch QPO between X-ray spectral bands (Hasinger 1987b). This suggested that X-rays originating from the near vicinity of the neutron star were Compton scattered in a surrounding hot plasma before they could escape. This Comptonization model had been proposed before to account for the absence of coherent millisecond X-ray pulsations through smearing and was consistent with some interpretations of the shape of the X-ray spectrum. The problem of accounting for the effect of scattering on QPO and beamed pulsar radiation stimulated a substantial theoretical effort. The idea that the magnetosphere in these systems is very small (a few neutron star radii, rather than hundreds as ~ X-ray pulsars) stimulated a reassess- ment of magnetospheric theories. It was pointed out that magnetosphenc
HIGH-ENERGY ASTROPHYSICS 2S7 formation might be qualitatively different in these systems from that in massive X-ray binaries since the boundaries of such small magnetospheres are located in the radiation-pressure dominated part of the disk (White and Stella 19884. Further observations showed that not all low-mass X-ray binaries con- formed to the above Z scheme. Some of them should have shown QPO, but they did not while others excited QPO with properties that did not fit into the Z scheme. Hasinger and Van der Klis (1989) distinguished, in addition to the Z sources, a second class of low-mass X-ray sources that shows a different pattern of correlated X-ray spectral and power-spectral behavior, many of these sources are X-ray bursters. In their X-ray color- color diagrams one can distinguish a "banana" shaped spectral branch and isolated "islands". Both spectral branches have a characteristic associated type of power spectrum (see Figure 4~. These so-called "atoll sources" are, on average, less luminous than the Z sources, and they do not exhibit the QPO found in the latter. Most of the persistently accreting low-mass binary X-ray sources can be classified as either a Z source or an atoll source; however, some (mostly transient) sources can not (e.g. the Rapid Burster, and Cir X-1~. Besides the fast-variability characteristics, other properties of low-mass X-ray binaries are correlated to their X-ray spectral states as well. It was recently discovered that the radio intensities of the Z sources OX 17 + 2, C:yg X-2, and Sco X-1 (Penninx et al. 1988; Hjellming et al. 1989, 1990) are strongly correlated with the location of these sources on the Z shaped track in the X-ray color-color diagram (see Figure 5~. When these sources are in the haring-branch state their radio brightness is relatively low, it increases when the source moves up the normal branch, becoming relatively high on the horizontal branch Since the radio emission is probably caused by relativistic electrons (synchrotron radiation), this correlation suggests that perhaps the required non-thermal processes and particle acceleration are connected with the boundary layer between the inner disk and the magnetosphere of the neutron star. A second, very recent, development is the discovery that for atoll sources there is a strong correlation between the spectral state of the source (i.e. "island" vs. "bananas state) and the properties of X-ray bursts (Van der Klis et al. 1990~. In particular, during the island state the duration of the bursts is much longer than during the banana state. This may be indicative of a difference in the hydrogen content of the layers in which the thermonuclear hash occurs that gives rise to the burst. Although it is likely that the accretion rate is the governing parameter for the properties of both the X-ray bursts and the persistent X-ray spectrum, the nature of the connection between the flashing layer and the (superficial) regions where the persistent (accretion) luminosity is emitted is presents unclear.
o 1.00 o o.so V 0.80 ¢ ~ c.60 258 AMERICAN AND SOVIET PERSPECTIVES GX 13+1 4U 1705-44 ....... , , , , i'''' ' ' 1120, art. I O.90 ~ 0.70 .~p! - _ 0.80 . 0.70 1 . i 0 1 . 8 0 1 . 9 0 2 . ~ 0 i . i 0 1 . 2 0 1 . 3 0 I; 4 0 i; 5 0 i . 6 0 SOFT COLOR 1 lo lo c lull ~3 1~4 ., ... . - _ - . . GX 13+1 '~\ 1~5 .. .. 1~3 1~2 1~1 10 ° 1 lol 10 o loll ~2 1~3 1~4 1~5 1~3 1~2 1~1 10 ° 1 ~1 4U 1705-44 ~- ~ l FREQUENCY (Hz) FIGURE 4 Display of the X-ray color color diagrams and power spectra of the atoll sources GX 13 ~ 1 and 4U 170544 illustrating the oo:Telation between power spectrum and the source spectral state ("island" versus "banana" state). This figure has been adapted Tom Van der Klis (1989~. In general, the rms variations of the QPO is only a few percent, and therefore most studies discussed so far have been based not on X- ray mtensibr curves but on power spectra of the intensity variations. An exception is the Rapid Burster. QPO rms variation up to ~ 30% have been observed from this source which made it possible with Toga to observe trains of individual oscillations (Dotani et al. 1990~. These observations showed that the finite width of the QPO peak in the power spectrum is caused by frequency modulation of the signal. Since the Rapid Burster is a very peculiar source, it is unclear whether the undertring mechanism holds for over sources as well. Finally, studies of the power spectra of some high-mass X-ray binaries
HIGH-ENER~ ~TROP~SICS in_ i: 0.06 1 2` ~ ~ 0.05 0 ~ . ~ ID a: 0 04 1 Go - - 0.03 0.02 259 I I I I I ~I I I I I I I I T S ~I 1 1 1 1 1 - Horizontal - I\ Branch ~I/, Normal I, ~ s, Bra a_ 0.8 1 1.2 0.8 1 1.2 soft colour (4.7-9.3 keV/1-4.7 key) FIGURE 5 Display of the correlation between the radio intensity and X-ray spectral properties of the Z source GX 17 + 2. In the left-hand panel the three branches are indicated in the X-ray color color diagram. In the nght-hand panel the size of the octagon is a measure of the radio intensity; its lo~tion indicates the spectral state of the source. (most of them pulsars) have shown that these power spectra also contain QPO and broad noise components, somewhat similar to those found in low-mass X-ray binaries (Ebisawa et al. 1989; Belloni and Hasinger 1989~. In the case of Me pulsating high-mass transient source EXO 2030 + 375 (Angelini et al. 1989) the variation of the QPO frequency with X-ray luminosity combined with the known spin frequenter of the neutron star was consistent ninth the beat frequency relation. However, recent results obtained by Norns et al. (1989) and by Mitsuda e! al. (19891 mav Dose problems for He beat frequency model ~, ~ In closing, the recent division of the low-mass X-ray binaries in Z sources and atoll sources has clarified matters a great deal. It seems that the beat frequency model is at present a promising (though not generally accepted) model for the honzontal-branch QPO in Z sources. The normal- branch and the flaring-branch QPO in Z sources probably have a common origin which is different from the origin of the horizontal-branch QPO.
260 AMERICAN AND SOVIET PERSPECTIVES REFERENCES Alpar, M A., and J. Shaham. 1985. Nature 316: 239. Angelini, L, L Stella, and ~N. Parmar. 1989. Astroph. J. In press. Belloni, 1:, and G. Hasinger. 1989. Astron. Astroph. In press Bradt, H.V., RL" Kelley, and LD. Petrol 1982. In: Sanford, P.W., et al. (eds). Galactice X-ray Sour~es. W~ley. 89. Dotani, T., K. Mitsuda, H. Inoue, Y. Tanaka, N. Kawai, Y. Tawara, K Makishima, J. Van Paradi~s, W. Penniox, M. Van der Klis, J. Tan, and W.H.G. Lewin. 1990. Astroph. J. 350: 395. Ebisawa, K, K Mitsuda, and H. Inoue. 1989. Publ. Astron. So~ Japan 41: 519. Hasinger, G. 1987a. Astron. Astroph. 186: 153. Hasinger, G. 1987b. In: White, N., and 3.-H. Huang (eds.~. The Origin and Evolution of Neutron Stam. IAU Symp. 125: 333. Hasinger, G. 1988. Adv. Space Res. Vol. 8, No. 2: 377. Hasinger, G., and M. Van der Klis. 1989. Astron. Astroph. In press. Hjellming, R.M., X.H. Han, F.A Cordova, and G. Hasinger. 1989. Astron. Astroph. In press. Hjellming, RM., R.l: Stewart, G. L White, R Strom, W.H.G. Lewin, P. Her~, K Wood, K Mitsuda, W. Penninx, and J. Van Paradijs. 1990. Astroph~ J. In press. Lamb, F.K 1988. Adv. Space Res. Vol. 8, No. 2: 421. Lamb, F.K 1990. Astroph. J. In press. Lamb, F.K., N. Shibazaki, M.A. Alpar, and J. Shaham. 1985. Nature 317: 681. Lewin, WH.G., G.W. Clark, and W.B. Smith. 1968. Astroph. J. 152: L55. Lewin, W.H.G., and P.C Joss. 1983. In: Lewin, W.H.G., and E.PJ. van den Heuvel (eds.~. Accretion-dnven Stellar X-ray Sources. Cambndge Unrversity Press. 41. Lewin, W.H.G., J. Van Paradijs, and M. Van der Klis. 1988. Space Sci. Rev. 46: 273. Middled itch, J., and W. C P riedhotsly. 1986. Astroph. J. . 306 : 230. Mitsuda, K, e' al. 1989. Proceedings of the Bologna conference, and paper submitted to Publ. Astron. Soc Japan. Noms, J., P. Hertz, K Wood, B. Vaughan, M. Michelson, K Mitsuda, and T. Dotani. 1989. Prooeedings of the Bologna Conference, and Astrophys J. Submitted. Penniox, W., W.H.G. Lewin, AN Zijlstra, K Mitsuda, J. Van Paradijs, and M. Van der Klis 1988. Nature 336: 146. ldwara, Y., S. Hayakawa, a al. 1982. Nature 299. 38. Van den Heuvel, E.PJ. 1986. In: ~mper, J., W.H.G. Lewin, W. Brinkman (ed~). Ihe Evolution of Galactic X-ray Binanes. Reidel Publ. Cry. 107. Van der Klis, M. 1989. Ann. Rev. Astron. Astroph. Z7: 517. Van der Klis, M., F. Jansen, J. Van Parad~js, W.H.G. Lewin, E.PJ. Van den Heuvel, J. l~mper, and M. Sztajno. 1985. Nature 316: 225. Van der Klis, M., F. Jansen, J. Van Paradijs, W.H.G. Lewin, M. Sztajno, and J. l~mper. 1987. Astroph. J. 319: L13. Van der K1LS, M., E. Damen, W. Penninx, J. van Paradijs, and W.H.G. Lewin. 1990. Astroph. J. (Lett) 360: Ll9. White, N.E., and L" Stella. 1988. Monthly Not. Roy. astron. Soc. 231: 325.