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.
The Cyclotron Absorption Line arid Eclipse Transition Phenomena of 4U 1538-52 GEORGE W. CLARK Massachusetts Institute of Technology ABSTRACT Observations of the eclipsing binary X-ray pulsar 4U 153~52 by the Japanese satellite Ginga have revealed a cyclotron absorption feature at 20 keV in the X-ray spectrum. The pulse-phase dependence of the intensity and spectrum can be mimicked by a model of X-ray emission from thin accretion-heated slabs at the magnetic poles of a rotating neutron star with its magnetic dipole axis inclined at 45 from the rotation axis. The observations also yielded data on the eclipse transitions which demonstrate that the radial density function at the base of the supersonic wind of the O-Wpe supergiant pnma~y has the forth of an exponential like that which characterizes the density run in the similar region of the O-bpe superg~ant primary of Cen X-3. As in the Cen X-3 system, the scale height of the exponential implies a temperature ~ the base region much greater than that of the supersonic wind. Recent observations of the eclipsing binary X-ray pulsar 4U 1538-52 with the Ginga satellite have yielded results bearing on two quite different topics. The first is the X-ray spectra and beaming pattern of the pulsar; the second is the density distribution in the winds of the early-trpe supergiant companions of this and other X-ray pulsars. Ginga was developed and launched by the Institute of Space and Astronautical Science of Japam It cames several detectors including a 4,000 cm2 Urge Area Counter ~AC) developed by the Leicester University group and specially suited to the measurement of the spectra and variability of compact X-ray sources like binary pulsars. The LAC is sensitive from 77
78 AMERICAN AND SOVIET PERSPECTIVES 1 to 38 keV and records data in 48 pulse-height channels with an energy resolution of 20% at 5.7 keV and a maximum time resolution of 1 msec. The Japanese group has made tune available to U.S. observers in a guest observer program supported on the U.S. side by NASA In this program an observation of 4U 153~52 was earned out over a complete 3.7~ay orbital cycle in March 1988 in a collaboration between myself and J. Woo of MIT and F. Nagase, K Makishima and T. Sakao at ISAS. The X-ray spectra of strong magnetized plasmas on neutron stars has been the subject of much theoretical work since the pioneering in- vestigations of Sunyaev and coworkers following the discovery of beady X-ray pulsars by Giacconi and colleagues with Uhuru in 1972. In 1975 Basko and Sunyaev suggested that the effects of cyclotron resonance in the Compton scattering cross section might be observed in the form of X-ray emission lines. The subsequent discovery in a balloon experiment by lumper et al. (1977) of the cyclotron feature near 50 keV in the Her X-1 spectrum focused attention on the problem of radiative diffusion in plasma with fields > 10~2 G. One other clear cyclotron line was found in 4U 0115+63 by Wheaton et al., and indications of a line in 4U 1626~7 by Pravdo (1979) and Koyama (1989~. Mazets e! al. (1981) found evidence of cyclotron features in gamma ray burst spectra, though other interpretations have been put forward. And just recently clear evidence of first and second harmonic absorption lines at 20 and 40 keV have been found with Ginga in two gamma-ray bursts. Our Ginga observation of 4U 153~52 adds a third definite example of cyclotron absorption line formation in an X-ray binary pulsar under what appear to be specially favorable circumstances for analysis (Clark al. 1989~. This object has a 530-second pulse period and an eclipse with a half-angle of about 30 . Considering first the data unaffected by the eclipse, we divided it into ten spectral intervals and plotted the counting rate as a function of pulse phase, as shown in Figure 1. We also divided the data into eight equal intervals of pulse phase and plotted the spectra of each portion, as shouts in Figure 2. Four salient properties of the spectra and vanabili~ are clearly evident: 1) The pulse profile has symmetrical primary and a secondary peaks of unequal amplitude and separated by 180 in phase. 2) The primary peak has a dimple at low energies. 3) There is an absorption line at 20 keV with a maximum equivalent width in the middle of the secondary peat Indeed, the line is so strong as to essentially blot out the secondary peak in the spectral internal centered on 20 keV. 4) The pulse fraction increases with energy. Properties 1), 2), and 4) have been seen in previous observations of
20 0 o 40 20 o US 60 ED as 40 o C: 20 o 80 JO 40 20 o 40 20 HIGH-ENER~ ASTROP~ICS 1 1 ~ 1 1 1 ' ' 1 n ~-12 6 keV 1 ~: ~ J 1 1 ~ 1 1 1 1 1 1 8.0-10.3 keV N-,~ 1 1 i 1 1 1 1 5.7 - 8.0 keV 1 1 1 1 1 1 1 1 1 3.4-5.7 keV _~ I_ _ _ 1 1 1 1 1 1 1 _ 1.1-3.4 keV ~- .5 .5 o 2 o 4 2 o 5 0 o 30 20 0 o o ~ ,~k _ 1 1 1 1 1 1 1 20.9 - 25.7 keV r ~ , 1 , 1 , 1 , ~ r ' I ' I 16.1 -20.9 keV _~ I_ 1 1 1 1 1 1 1 12.6 - 16.1 keV _~ ;~4' ~ _ , 1 , 1 , .5 ~ 1.5 2 PULSE PHASE PULSE PHASE FIGURE 1 X-ray counting rates of 4U 1538 52 plotted against pulse phase (pulse period = 530 s) for ten energy channels. 4U 1538-52. The absorption line at 20 keV is new and opens interestung possibilities for detailed comparisons between the observed phenomena and the results of recent theoretical treatments of the emissions of magnetized slabs and columns of accretion-heated plasmas at the poles of magnetized neutron stars.
80 1oo o ~ l o 2 10.3 10' 10° 10': ~ c' >. :. 1o 2 0 Y z In O z lo, o o ~ c' loo 10-1 10-2 10° 10-1 AMERICAN AND SOVIET PERSPECTIVES - i\ 3 ~ 10° 10't l o 2 101 10° ·o-7 l o 2 101 10° 10'1 1 o 2 101 5 L 6 me\ 'Gee 4 .~2 . , , , -. 10.3 1 0 20 30 CO 50 ENERGY (keV) 1 0 20 30 40 50 ENERGY (keV) FIGURE 2 Pulse height spectra and inferred inadent energy spectra of 4U 153~52 for eight intervals of pulse phase.
HIGH-ENERGY ASTROPHYSICS 81
82 AMERICAN AND SOVIET PERSPE~IIVES go 11 - 5 keV - If- ~- 20 keV - 25 keV :~""~ p = zoo ' 1 ' 1 1 1 , 25 keV I ~- L:i - W 1 1 1 1 1 1 1 1 1 1 1 1 - ~ -W ~20 keV 5 keV ~ TV , 1 , 1 , 1 , 0 .5 1 PULSE PHASE - v, go 180 ~ 5 7 10 20 30 50 ENERGY (keV) FIGURE 3 Calculations of the pulse light curve (middle panel) and pulse-phase resolved energy spectra (bottom panel) for a model X-ray pulsar with the radiation pattern illustrated in the top panel and the orientations of magnetic dipole axis and spin axis specified in the text.
HIGH-ENERGY ASTROPHYSICS 83 gradual, though one abrupt transition was apparently observed by Davison et al. (1977) with Ariel. A gradual eclipse immersion or emersion of an eclipsing X-ray binary, of which six are known, is obviously caused by absorption in an extended atmosphere of the primary, which opens a new direct way to a determination of the radial density distinction. The situation can of course be complicated by streams or blobs of matter flying around in the system as in Vela X-1 and sometimes in Cen X-3. But often in the case of Cen X-3, SMC X-1, EMC X4, Her X-1 and 4U 1538-52 the transitions are fairly clean and uniform, indicating that the measured column densities are fair measures of the radial density functions. The situation can be ideal because the size of the X-ray source, i.e. the neutron star, is negligible compared to the scale size of the atmosphere, and X-ray absorption from 1 to 10 keV is a relatively simple measure of column density along the line of sight. Compare this with the indirect arguments that must be used to interpret spectroscopic, radio and IR data in terms of atmospheric structure. The first remark about the significance of a gradual eclipse transition was made by Schreier et al. (1972) In the paper announcing the discovery of the binary nature of Cen X-3. They characterized the eclipse transition by ascribing a "scale height" of 5 x 10~° cm to the atmosphere of the primary implying thereby an exponential form of the density function Following the discovery of intense, cool (30,000-50,000 K), supersonic winds in O and B stars by Morton (1967) in rocket UV spectroscopy, theories of the wind acceleration process developed around the idea of radiation pressure arising from scattering of light from the Doppler-shifting UV lines of the metals in the wind. This works well in the supersonic regime, and since the radiation intensity varies as 1/r2, the resulting acceleration leads to velocity cones with a characteristic rapid initial rise and then a slow approach to an asymptotic terminal velocity with values in the range of 1000 to 3000 Ems, as observed in the P Cygni profiles of UV absorption lines. There has always been a problem, though, in how the winds get started because the passage from sub to supersonic is blocked by the sound barrier as expressed in the singularity in the hydrodynamic equation governing the how. Only by carefully tailoring the outward force by judicious combinations of thermal and radiation pressure can one achieve a steady flow from sub to supersonic with the requisite mass flux and cool temperature. Hearn suggested in 1975 that there may be a coronal layer at the base of the wind in which thermal pressure can cause the early acceleration, with sudden cooling and radiation pressure taking over when the velocity is high enough for Doppler shifting to prevent line saturation. Then Castor, Abbott and Klein (1975) showed He way to a theory of a radiation pressure driven cool wind from start to finish, and came up with velocity curies with the same rapid initial change, and then a slow approach to asymptotic terminal
84 AMERICAN AND SOVIET PERSPECTIVES velocity, which turns out to be, in general, about three times the escape velocity from the stellar surface (Abbott 1978~. Near the end of the SAS 3 mission we undertook a detailed study of Cen X-3 eclipses using data from a continuous two-week observation when Cen X-3 was in a high luminosity state (Clark et al. 1988~. The transitions are well accounted for by an exponential density function with a scale height of 6 x 10~° cm, and not by a density function implied by the conventional 1/r2 radiation force-driven wind theories. Here, as in the case of 4U 153~52, the opticaVUV luminosity of the companion is much greater than the X-ray luminosity of Cen X-3, so the X-ray heating effects are presumably negligible. Day et al. (19883, using data from several continuous observations of eclipse transitions by EXOSAT, drew the same conclusion, lie. that the density run is exponential Such a scale height, if interpreted as the characteristic of an isothermal, hydrostatic atmosphere of a 19 solar mass star, corresponds to a temperature of 106 K Following Hearn's idea that there may be a coronal layer at the base of the wind, heated by some mysterious process, in which the initial acceleration takes place, we constructed an ad hoc model with a 106 K base corona governed by the usual hydrodynamic flow equation with a gravity reduced by half by radiation pressure, and terminating at 1.4 R* at the high temperature sonic point where we assumed the heating mechanism turns off, the temperature drops to 50,000 K, and radiation driving takes over in the cool but supersonic regime. This hybrid model yields a density curve that can be fit to the data for reasonable mass loss rates. A major problem of the model is that its coronal layer has an enormous soft X-ray luminosity (about 10% of the optical luminosity of KRZ) which must be absorbed by the outer wind. Other ways to get more mass loss out of a cool subsonic flow Is to fine tune the effective gravity by allowing for centrifugal force (which may amount to 1/6 to 1/3 of the gravity) and radiation pressure without the benefits of the large Doppler shifts that keep it effective in the supersonic regime. But I am not aware of rigorous treatment of this subsonic regime that yields a density curve that explains the Cen X-3 eclipse transition data. Now, with the 4,000 cm2 sensitive area of Ginga, we have good data on 4U 1538-52, another massive binary pulsar, providing similar results. The variation with orbital phase of column density deduced by spectral analysis is displayed in Figure 4 along with the results of least squares fittings of three different functions-an exponential, an isothermal hydrostatic function, and a 1/r2 force-driven wind. The latter fails again because of its too rapid initial fall in density. The isothermal hydrostatic function fits well, allowing, to be sure, for the fluctuation that cannot be fitted by any reasonably smooth function. Of course, the situation is not hydrostatic, so a transition to a radiation~riven regime must occur, and one can undoubtedly construct another ad hoc hybrid model, combining an initial subsonic acceleration
HIGH-ENERGY ASTROPHYSICS Cot 2 A x - 3 z LU As ~ 2 o ~ I ' 1 ' 16 ' ISOTHERMAL HYDROSTATIC ~ 1 ~1 -.3 ORBITAL PHASE +-3 ~ · T i T ~ I · l EXPONENTIAL ~.1 ~ ::1 1_ Fit -. . . . . . . . . . . . ORBITAL PHASE +.3 4 0 -4.2 ~ . , ._ 1 /R2WIN D 1 ..2 . ORBITAL PHASE +.3 85 FIGURE 4 Measured column densities (H-atoms/cm2) along the lines of sight to 4U 1538 52 plotted against orbital phase during one eclipse ingress and egress. The solid curves are least-squares fits of column density curves derived from three trial atmospheric density functions of the primary BO star plus constant terms before and after eclipse. Deviations of the data from the fitted column densities are displayed below each plot. (In each case the line integration is terminated at two orbital radii).
86 AMERICAN AND SOVIET PERSPECTIVES with an exponential density run with a radiation-dr~ven supersonic regime. But what is needed now is theoretical attention to the facts of X-ray eclipse transitions and the direct information they provide about He mysterious subsonic acceleration phase of early star wind generation. REH:RENCES Abbott, D.C 1978. Ap. J. 225: 893. Basko, M.M., and R^. Sunyaev. 1975. Astr. Ap. 42: 311. Castor, J.I., D.C. Abbott, and R.I. Klein. 1975. Ap. J. 195: 157. Clark, G.W., J.W. Woo, F. Nagase, K Makishima, and 1: Sakao. 1990. Ap. J. 353-274. Clark, G.W., J.R. Minato, and G. Mi. 1988. Ap. J. 324: 974. Hampton, D., J.B. Hutchings, and A P. Cowley. 1978. Ap. J. (Letters3 Z5: L63. Day, ~ 1988. Thesis, Cambridge University. Davison, PJ.N., M.G. Watson, and J.P. Pye. 1977. M.N.R.\S. 181: 73P. Giacconi, R., H. Gursly, E. Kellogg, E. Schreier, 1: Matilsky, D. Koch, and H. Idnanbaum. 1971. Ap. J. Suppl. Z37(27): 37. Hearn, A.G. 1975. Astr. Ap. 40: Z77. Koyama, K et al. 1989. Pub. Astr. Soc Japan, in press. Mazets, E.P., S.V. Golonetskii, AL Aptekar', Y^. Gur~yan, V.N. Al'inskii. 1981. Nature 290: 378. Mesz~ros, P., and W. Nagel. 1985. Ap. J. 298: 147. Morton, D.C 1967. Ap. J. 147: 1017. Nagel, ~ 1981. Ap. J. 251: Z78. Pravda, S.H., N.E. White, EN Boldt, S.S. Halt, PA. Serlemitsos, J.H. Swank, and AK. Szymkowia~ 1979. Ap. J. =1: 912. Schreier, E., R. Levinson, H. Gursly, E. Kellogg, H. Tananbaum, and R Giacconi. 1972. Ap. J. (Letters) 172: L79. Hamper, J., W. Pietsch, ~ Reppin, W. Voges, R Staubert, and E. Kendziom. 1978. Ap. J. (Letters) 219: L105. Wheaton, Wm. ~ et al. 1979. Nature 282: 240.