1: %9/23/99
2: %1st draft apj letter
3: %5/17/00
4: %revise for referee apj paper
5: %6/19/00
6: %revision--padi's comments, new gabor transforms
7:
8: %\documentstyle[12pt,aasms4]{article}
9:
10: \documentclass{aastex}
11: \usepackage{emulateapj5}
12:
13: \newcommand{\Msun}{\mbox{ M}_{\odot}}
14: \newcommand{\gae}{\mathrel{>\kern-1.0em\lower0.9ex\hbox{$\sim$}}}
15: \newcommand{\lae}{\mathrel{<\kern-1.0em\lower0.9ex\hbox{$\sim$}}}
16: \newcommand{\kms}{km~s$^{-1}$}
17: \newcommand{\NV}{\ion{N}{5}\,$\lambda\lambda\,1238.8,\,1242.8$\ }
18: \newcommand{\CIV}{\ion{C}{4}\,$\lambda\lambda\,1548,\,1550$\ }
19: \newcommand{\SiIV}{\ion{Si}{4}\,$\lambda\lambda\,1393.7,1402.8$\ }
20:
21: %\input{psfig}
22:
23: \lefthead{Boroson et al.}
24: \righthead{QPOs in Her~X-1}
25:
26: \slugcomment{Accepted for publication in the Astrophysical Journal, 2000,
27: v545}
28: \begin{document}
29:
30: \title{Discovery of mHz UV Quasiperiodic Oscillations in
31: Hercules~X-1}
32:
33: \author{Bram Boroson\altaffilmark{1}}
34: \affil{Goddard Space Flight Center, Greenbelt, MD 20771;
35: bboroson@falafel.gsfc.nasa.gov}
36:
37: \and
38:
39: \author{Kieran O'Brien, Keith Horne}
40: \affil{Physics and Astronomy, University of St. Andrews, North Haugh, St.
41: Anderws, Fife KY 16 988, UK;
42: kso@st-andrews.ac.uk, kdh1@st-and.ac.uk}
43:
44: \and
45:
46: \author{Timothy Kallman, Martin
47: Still\altaffilmark{2}, and Patricia T. Boyd\altaffilmark{2}}
48: \affil{
49: Goddard Space Flight Center;
50: tim@xstar.gsfc.nasa.gov,
51: still@chunky.gsfc.nasa.gov,
52: padi@dragons.gsfc.nasa.gov}
53:
54:
55: \and
56:
57: \author{Hannah Quaintrell}
58: \affil{The Open University, Walton Hall, Milton Keynes MK7 6AA,
59: UK; H.Quaintrell@open.ac.uk}
60:
61: \and
62:
63: \author{Saeqa Dil Vrtilek}
64: \affil{Harvard-Smithsonian Center for
65: Astrophysics, 60 Garden Street, Cambridge, MA 02138;
66: svrtilek@cfa.harvard.edu}
67:
68: \altaffiltext{1}{National Research Council Associate}
69: \altaffiltext{2}{Universities Space Research Association}
70:
71: %***
72:
73: \begin{abstract}
74:
75: Observations of the ultraviolet continuum of the X-ray binary
76: system Her~X-1/HZ~Herculis with the Space Telescope Imaging Spectrograph
77: on the Hubble Space Telescope show quasiperiodic
78: oscillations (QPOs) at frequencies of 8$\pm2$ and 43$\pm2$ mHz, with rms
79: amplitudes of
80: 2 and 4\%\ of the steady flux. Observations with the Keck telescope
81: confirm the presence of the higher frequency QPO in the optical
82: continuum, with a rms amplitude of $1.6\pm0.2$\%.
83: The QPOs are most prominent in the HST data near
84: $\phi=0.5$ (where $\phi=0$ is the middle of the X-ray eclipse), suggesting
85: that they arise not in the accretion disk but on the X-ray heated face of
86: the companion star. We discuss scenarios in which the companion star
87: reprocesses oscillations in the disk which are caused by either Keplerian
88: rotation or a beat frequency between the neutron star spin and Keplerian
89: rotation at some radius in the accretion disk.
90:
91: \end{abstract}
92:
93: \section{Introduction}
94:
95: Hercules~X-1/HZ~Herculis is an X-ray binary consisting of a 1.24 second
96: pulsar in an eclipsing 1.7 day orbit with a $\sim2\Msun$ mass normal
97: companion. As a result of its many periodicities, it is one of the most
98: frequently observed X-ray binaries. The X-rays vary over a 35~day cycle;
99: an $\approx11$ day ``Main-on'' state and $\approx8$ day ``Short-on'' state
100: (in which the observed X-ray flux is reduced by a factor of $\approx3$)
101: are separated by half of a 35-day phase (Scott \&\ Leahy 1999). Outside
102: of these states the
103: X-ray flux is several \%\ of that seen in the Main-on state. The X-ray
104: modulation is not due to a change in the total X-ray output, as the
105: optical magnitude continues to vary over the 1.7~day orbit due to X-ray
106: heating of the companion star. Instead, the 35-day variation probably
107: results from obscuration of the central source by an accretion disk which
108: wobbles over a 35-day period due to an unknown cause. X-ray absorption
109: dips occur at a period of 1.65~days, near to, but significantly greater
110: than, the 1.62~day beat period between the 1.7 and 35 day periods (Crosa
111: \&\ Boynton 1981; Scott \&\ Leahy 1999). Features in the X-ray spectrum
112: near $30-40$~keV have been interpreted as the result of cyclotron
113: absorption (Tr\"umper et al. 1978, Dal Fiume et al. 1998), implying a
114: magnetic field strength $B=2.9\pm0.3\times 10^{12}$~G (Mihara et al.
115: 1990). Recently, another model has been put forward for the cyclotron
116: lines, in which the electron momentum distribution is allowed to be
117: anisotropic. The magnetic field strength is then inferred to be
118: $B=4-6\times10^{10}$~G (Baushev \&\ Bisnovatyi-Kogan 1999).
119:
120: The X-ray power spectrum is dominated by the 1.24 second pulse and its
121: harmonics and by a continuum noise component (Belloni, Hasinger, \&\
122: Kahabka 1991). There has been a single report of 144~sec period X-ray
123: fluctuations during an X-ray turn-on (Leahy et al. 1992). The optical and
124: UV continuua both show evidence for reprocessed pulsations (Middleditch
125: \&\ Nelson 1977, Boroson et al. 1996) with fractional amplitudes of
126: several $\times10^{-3}$. The accretion disk and the atmosphere of HZ~Her
127: may both contribute to the reprocessed pulsations.
128:
129: Quasiperiodic oscillations (QPOs), observed in a variety of compact
130: accreting binary systems, have not previously been reported for Her~X-1.
131: QPOs in general are poorly understood. In the ``atoll'' and
132: ``Z-class'' sources that show kHz QPOs, spin periods are not observed
133: directly, cyclotron lines are not seen, and emission from the companion
134: star is often overwhelmed by the emission from the accretion disk. Thus
135: the discovery of QPOs in Her~X-1 may help to link QPO behavior to
136: such properties as the neutron star magnetic field and spin, and details
137: of the mass transfer.
138:
139: \section{Observations}
140:
141: We have been carrying out a multiwavelength study of Hercules~X-1
142: using the HST~Space Telescope Imaging Spectrograph (STIS), the
143: Rossi X-ray Timing Explorer (RXTE), the Extreme Ultraviolet Explorer
144: (EUVE), and ground-based observatories, including the Keck Telescope.
145: Empirical models for the emission lines observed during the first segment
146: of our campaign, in July of 1998, have been reported in Boroson et
147: al.
148: (2000). A second segment of the campaign took place during July of 1999.
149: The 1999 observations took place during an ``anomalous low'' state in
150: which the X-ray flux was two orders of magnitude lower than in the
151: expected ``main-on'' state, although accretion continued to take place
152: (Parmar et al. 1999). A summary of the multiwavelength campaign results
153: is presented in Vrtilek et al. (2000).
154:
155: %\begin{deluxetable}{cccccc}
156:
157: \vbox{
158: %\footnotesize
159: \scriptsize
160: \begin{center}
161: {\sc TABLE 1\\
162: The STIS observation log}
163: \vskip 4pt
164: \begin{tabular}{cccccc}
165: \hline
166: \hline
167: Root name & Start & Time & Orbital
168: & Count Rate & QPO$\rm ^b$\\
169: & (MJD) & (s) & Phase$\rm ^a$& (s$^{-1}$) & \\
170: \hline
171: O4V405010 & 51371.124 & 2227 & 0.512 & 624 & Y\\
172: O4V405020 & 51371.185 & 2636 & 0.549 & 605 & Y\\
173: O4V405030 & 51371.253 & 2636 & 0.589 & 550 & Y\\
174: O4V405040 & 51371.321 & 2636 & 0.629 & 491 & Y\\
175: O4V405050 & 51371.388 & 2620 & 0.669 & 391 & Y\\
176: O4V406010 & 51372.064 & 2227 & 0.065 & 24 & N\\
177: O4V406020 & 51372.124 & 2636 & 0.102 & 57 & N\\
178: O4V406030 & 51372.193 & 2636 & 0.142 & 70 & N\\
179: O4V406040 & 51372.260 & 2636 & 0.182 & 92 & N\\
180: O4V406050 & 51372.328 & 2620 & 0.221 & 131 & N\\
181: O4V407010 & 51373.004 & 2227 & 0.618 & 561 & Y\\
182: O4V407020 & 51373.065 & 2636 & 0.655 & 435 & Y\\
183: O4V407030 & 51373.133 & 2636 & 0.695 & 329 & Y\\
184: O4V407040 & 51373.200 & 2636 & 0.735 & 236 & ?\\
185: O4V407050 & 51373.268 & 2620 & 0.774 & 158 & ?\\
186: O4V408010 & 51374.011 & 2227 & 0.210 & 74 & N\\
187: O4V408020 & 51374.072 & 2636 & 0.247 & 128 & N\\
188: O4V408030 & 51384.140 & 2636 & 0.288 & 214 & S\\
189: O4V408040 & 51374.207 & 2620 & 0.327 & 325 & S\\
190: \hline
191: \vspace*{0.02in}
192: \end{tabular}
193: \end{center}
194: $\rm ^a${The orbital phase of the mid-exposure time, using the
195: ephemeris of Deeter et al. (1991)}\\
196: $\rm ^b${Y=QPOs detecteed, N=Definitely no QPOs detected,
197: ?=Borderline significance,
198: S=significant if a linear fit to the light curve is subtracted }
199: \normalsize
200: }
201: \vspace{0.02in}
202:
203: The UV observations used the HST STIS. The STIS instrument design is
204: described by Woodgate et al. (1998), and the in-orbit performance of the
205: STIS is described by Kimble et al. (1998).
206: We show a log of the 1999 STIS observations in Table~1.
207: All of these observations used the E140M grating for high
208: resolution echelle spectroscopy. This provides a spectral coverage of
209: $1150-1710$\AA\ with a resolving power of $R=45,800$ (6~\kms).
210:
211: \section{Detection of QPOs}
212:
213: We obtained the HST spectra using the TIME-TAG mode, which stamps each
214: photon with a time accurate to 125$\mu$sec and the position of the photon
215: along the two axes that define the two-dimensional echelle image. To
216: create light curves we first selected a region of the two-dimensional raw
217: spectral image corresponding either to the entire spectrum or the
218: brightest spectral lines (\ion{N}{5}$\lambda\lambda1238.8,1242.8$ and
219: \ion{C}{4}$\lambda\lambda 1548.2,1550.7$). We selected the region in the
220: two-dimensional echelle image using the task ``fselect'' in the FITS
221: utility package FTOOLS (version 4.2). Then we converted the resulting file
222: to a light curve with 0.1~second bins using the FTOOLS ``fcurve'' task.
223: We then formed power spectra from the discrete Fourier transform of the
224: data in each of the HST orbits, and for 400~second intervals within each
225: orbit.
226:
227: As a ``control'' on our method, we also examined STIS TIME-TAG data for
228: the X-ray binary LMC~X-4 using similar data reduction techniques. Analysis
229: of these observations, and the search for UV pulsations with the
230: 13.5~second LMC~X-4 pulsar period, will be presented in Kaper et al.
231: (2000). None of the 5~HST orbits in which LMC~X-4 was observed showed
232: significant power above the white noise due to counting
233: statistics in the 1-100 mHz region.
234:
235: We did not find any evidence for UV pulsations near the 1.24~s pulsar
236: period in the 1999 data. We would have detected (with 5$\sigma$
237: significance) pulsations of rms amplitude $0.2$\%\ near $\phi=0.5$,
238: amplitude $\approx0.3$\%\ near $\phi=0.75$, and amplitude $\approx0.4$\%\
239: near $\phi=0.2$ when the disk is emerging from eclipse. These values
240: should be compared with the UV rms amplitudes found with the GHRS in
241: Boroson et al. (1996): $0.4$\%\ at $\phi=0.56$ and $0.7$\%\ at
242: $\phi=0.83$. In the 1998 data, during the short-on state at $\phi=0.76$,
243: we found marginal evidence (false alarm probability $0.4$\%) for
244: pulsations with rms amplitude of $0.7$\%. With this sole exception, there
245: were no UV oscillations near the pulse period in the data from either 1998
246: or 1999. The fractional rms amplitude is defined and related to the power
247: spectrum in van der Klis (1989). For a constant plus pure sinusoid
248: signal, the fractional rms amplitude is $1/2 \sqrt 2$ times the
249: peak-to-peak difference, divided by the average value.
250:
251: It is useful to place the non-detection of UV pulsations in context by
252: examining the X-ray pulsations seen in 1998 and 1999. Our 1998
253: multiwavelength campaign overlapped with X-ray observations using Bepposax
254: (Oosterbroek et al. 2000). These observations found that the fractional
255: rms amplitude of the X-ray pulses during the July 1998 short-on state was
256: 10\%\ in the 4-10~keV band, compared with 20\%\ in the main-on state (Fig.
257: 4 of Oosterbroek et al. 2000). RXTE measured an X-ray flux of
258: $1.5\times10^{-9}$~erg~s$^{-1}$~cm$^{-2}$, steadily decreasing to
259: $5\times10^{-10}$~erg~s$^{-1}$~cm$^{-2}$ as the short-on state progressed.
260: The 1999 RXTE observations found that
261: although the direct component of the X-rays was obscured, X-rays were
262: observed at $\approx5$\%\ of the normal main-on level, and
263: continued to pulsate
264: with the 1.24~second period, with a fractional rms amplitude of
265: $1-3$\% (Still et al. 2000). We caution, however, that these low pulse
266: fractions
267: do not provide a full explanation of why UV pulses were not observed,
268: as the X-ray pulse fraction seen by the companion star and disk (which
269: cause the UV pulsations) can
270: be very different from that which we observe.
271:
272: %\begin{figure*}
273: %\epsscale{0.4}
274: \vbox{
275: \vspace{0.4in}
276: \epsscale{0.8}
277: \plotone{fig1.ps}
278: \vspace{0.5in}
279: \figcaption{Individual power spectra for the continuum light curves of
280: each HST orbit in which QPOs were detected. We label the mean orbital
281: phase and the observation root name for each orbit. The dotted vertical
282: lines mark 5 and 45 mHz.}
283: }
284: %\plotfiddle{fig1.ps}{1in}{90}{100}{100}{0}{0}
285: %\end{figure*}
286:
287: \vspace*{0.2in}
288:
289: In Figure~1 we show the average power spectrum for each orbit in which
290: QPOs were detected. In Figure~2 we show the average of 46~power spectra
291: of the light curves from all HST orbits from the 1999 campaign in
292: 400~second intervals. Near $\phi=0.5$ the count rate is sufficient to
293: detect individual QPO oscillations, which we show in Figure~3. No
294: significant QPOs were detected in the emission line flux. The lines
295: contribute about 20\%\ of the total counts in the spectrum. When we
296: subtract the light curve of the emission lines from the light curve of the
297: total spectrum and take a power spectrum, the QPOs remain unaffected.
298:
299: %\begin{figure*}
300: %\epsscale{0.5}
301: \vbox{
302: \epsscale{1.15}
303: \plotone{fig2.ps}
304: \figcaption{The sum of 46 power spectra of 400~second light curves of the
305: UV continuum of Her~X-1
306: observed in July 1999 with the STIS aboard HST. The noise level is
307: normalized to 2 (Leahy normalization). The dashed curve shows a Gaussian
308: fit to
309: the two peaks. The dotted vertical lines mark 5 and 45 mHz.)}
310: }
311: %\end{figure*}
312:
313: \vspace*{0.2in}
314:
315: We could not obtain accurate Gaussian representations of the power spectra
316: within each HST orbit. However, taking the average of all power spectra,
317: we could fit a power-law continuum and 2~Gaussians (Figure~2). We find
318: QPO centroids of $\nu_1=8\pm2$~mHz and $\nu_2=43\pm2$~mHz (we have
319: determined errors from a Monte Carlo bootstrap analysis, Press et al.
320: 1992). The gaussian widths of the QPOs were $4\pm1$~mHz and
321: $21\pm2$~mHz. Applying
322: the techniques outlined in van der Klis (1989), we find
323: total rms QPO fractions of 2.1$\pm0.7$\%\ and 3.7$\pm0.4$\%\ for the two
324: QPO peaks, respectively.
325:
326: We detect QPOs at $\phi=0.28-0.33$ if we remove a linear trend from the
327: UV continuum light curve (Table~1). It is possible that the
328: predominance of UV oscillations near $\phi=0.5$ results from the flatter
329: orbital light curve near $\phi=0.5$. However, removing linear trends from
330: the data for other HST orbits did not reveal mHz~QPOs.
331:
332: After finding QPOs in the 1999 STIS data, we examined optical observations
333: from our 1998 campaign and found evidence for a 35~mHz QPO peak there as
334: well (Figure~4).
335: Results from the optical campaign will be presented in more detail in
336: (O'Brien et al. 2000).
337: The HST~GHRS
338: observations of Boroson et al. (1996) during the Main-On state included
339: 1200~s of high count-rate (1100~cts~s$^{-1}$) data at $\phi=0.55$, which
340: show the 8~mHz QPO but not the 43~mHz QPO.
341:
342: %\begin{figure*}
343: \vbox{
344: \epsscale{1.15}
345: \plotone{fig3.ps}
346: \figcaption{A portion of the UV continuum light curve
347: showing individual QPO oscillations. The x-axis shows the time elapsed
348: from the start of observation 5030. A single error bar is shown; errors
349: on the other points are similar. Data are presented in 2~second
350: time bins. The rms variability for this portion of the data is 5.6\%,
351: while the expected noise rms is 3.0\%. This is consistent with a QPO rms
352: of 4.7\%.}
353: }
354:
355: %\vspace*{0.2in}
356:
357: %\begin{figure*}
358: \vbox{
359: \epsscale{0.66}
360: \plotone{fig4.ps}
361: \figcaption{Power spectrum of the optical continuum of Her~X-1 observed
362: in
363: July 1998 with the Keck Telescope.}
364: }
365: \vspace{0.1in}
366: %\end{figure*}
367:
368: We investigated the time dependence of the 45 mHz QPO feature using a
369: Gabor transform (Heil \&\ Walnut 1989, 1990). The Gabor transform is a
370: wavelet transform which decomposes the signal into the time-frequency
371: plane. The discrete implementation used here is equivalent to a
372: short-time, Gaussian-windowed Fourier transform (Boyd et al. 1995). We
373: chose a window size of 512 seconds, and restricted the frequency range to
374: lie between 2.5 - 90 mHz. Figure~\ref{fig:gt} displays power as a
375: function of frequency and time for the first five observations. The
376: palette runs from dark blue (low power) through deep red (high power).
377: The featureless dark blue vertical strips are times during which no data
378: was
379: collected. Significance was estimated by performing individual,
380: independent Fourier transforms on the data in windows fully containing
381: the features of highest power. The strongest of these features are found
382: to be significant at more than $\sim6\sigma$.
383:
384:
385: Figure~\ref{fig:gt} shows the frequency evolution of the 45~mHz QPO as a
386: function of time, and can be thought of as an
387: unfolding
388: of the single power spectrum, which shows a broad QPO feature centered
389: about a preferred frequency. The frequency evolution shown in the Gabor
390: transforms shows
391: that the QPO in the overall power spectrum
392: is composed of many individual short-time components that come and go
393: over the entire observation.
394:
395: \vbox{
396: %\vspace{0.1in}
397: \epsscale{0.65}
398: \plotone{fig5.ps}\\
399: \vspace{0.75in}
400: \epsscale{0.65}
401: \plotone{fig5b.ps}
402: \vspace{1.0in}
403: \figcaption{\label{fig:gt} The Gabor Transform, a dynamical power
404: spectrum, applied
405: to the observations for which QPOs were detected. Colors range from
406: dark blue (low power) through red (high power) according to the
407: palette shown. The frequency range covers both mHz QPO features.
408: Neither feature is constant in time, but made up
409: of shorter, apparently more coherent, signals. Features with the
410: highest power are significant at $\approx6\sigma$.
411: There is no obvious correlation between the instantaneous power or
412: frequency of one feature with the other. (Dark blue vertical bands
413: represent times when no data was obtained.) Panel (a) shows the 50x0
414: series of observations,
415: and (b) shows the 70x0 series.}
416: }
417:
418: %\vspace*{0.1in}
419:
420:
421: \section{Discussion}
422:
423: %qpos vs. phase
424:
425: The ``spikey'' power spectra could be caused by several processes.
426: Multiple overlapping exponential oscillating shots or a single sinusoid
427: whose phase undergoes a random walk could each reproduce the qualitative
428: power spectral shape observed. The Gabor transforms
429: suggest that the QPO width is due to high Q oscillations appearing and
430: disappearing at different frequencies.
431: Oscillating shot noise can cause both QPO
432: and a low-frequency red noise component together, and we have tried to
433: determine whether the 8~mHz feature could actually be red noise
434: associated
435: with the 45~mHz feature. Analysis of the dynamical power spectrum through
436: the Gabor transform shows no apparent correlation.
437:
438: Are the QPOs due to the accretion disk or the X-ray heated atmosphere of
439: HZ~Her? The QPOs were not detected at $\phi\lae0.33$ or $\phi\gae0.73$.
440: Models suggest that at $\phi<0.2$ and $\phi>0.8$, most of the optical and
441: UV continuum arises in the accretion disk, and that from $\phi=0.2$ to
442: $\phi=0.8$ the flux from the disk should be constant (Howarth \&\ Wilson
443: 1983). The X-ray heating of the atmosphere of HZ~Her causes the optical
444: continuum flux to reach a maximum near $\phi=0.5$ (Bahcall \&\ Bahcall
445: 1972), and at these phases, the UV emission arises predominantly on the
446: X-ray heated face HZ~Her (Vrtilek \&\ Cheng 1996). From Figure~6 we
447: conclude that at $\phi=0.5$, the disk continuum should contribute at most
448: $10$\%\ of the total continuum flux. Thus if the UV QPOs with 10\%\
449: amplitude of the total flux result from oscillations of the disk flux, the
450: disk flux must be modulated by an amplitude near 100\%.
451:
452:
453:
454: %\begin{figure*}
455: %\epsscale{0.5}
456: \vbox{
457: \epsscale{1.}
458: \plotone{fig6.ps}
459: \figcaption{Flux of the UV continuum (1260\AA\ to 1630\AA) as
460: observed with IUE, versus
461: orbital phase. The diamonds show the fluxes observed with the HST STIS
462: in July 1998, and the squares show the fluxes observed in July 1999.}
463: }
464: %\end{figure*}
465:
466: \vspace*{0.2in}
467:
468:
469: It would be simpler to attribute the QPOs to the heated atmosphere of
470: HZ~Her. The concentration of the QPO detections about $\phi=0.5$ could be
471: a result of the increased stellar flux; with a higher flux the QPOs would
472: be easier to detect. The light-travel time across the X-ray heated face of
473: the companion should be $<10$~seconds, as the separation of the neutron
474: star and the system center of mass is $a \sin i=13.186$~s (Deeter et al.
475: 1991). This allows variability on the observed timescales, while damping
476: the harmonics of the 45~mHz QPO (which are not observed).
477:
478: \vbox{
479: %\footnotesize
480: \scriptsize
481: \begin{center}
482: {\sc TABLE 2\\
483: X-ray Binary Pulsars with mHz QPOs}
484: \vskip 4pt
485: \begin{tabular}{lcccccl}
486: \hline
487: \hline
488: Name & Type$\rm ^a$ & $\nu_{\rm pulse}$ & $\nu_{\rm QPO}$ & B &
489: Reference\\
490: & & (mHz) & (mHz) & $10^{12}$~G & \\
491: \hline
492: Her X-1 & P & 807.9 & $5,45$ & 2.9 & this
493: work\\
494: & & & & $\pm0.3$ & \\
495: SMC X-1 & P & 1410 & $10,60$ & & Wojdowski\\
496: & & & & & et al. 1998\\
497: 4U 1627-67 & P & 132 & 48 & $\approx3$ & Chakrabarty\\
498: & & & & & 1998\\
499: Cen X-3 & P & 207 & 35 & & Takeshima \\
500: & & & & & et al. 1994\\
501: 4U 0115+63 & T & 277 & 62 & 1.0 & Soong \&\ \\
502: & & & & & Swank 1989\\
503: 4U 1907+09 & T & 2.27 & 55 & & in'tZand \\
504: & & & & & et al. 1998\\
505: XTEJ1858+034 & T & 4.5 & 110 & & Paul \&\ Rao \\
506: & & & & & 1998\\
507: V0332+53 & T & 229 & 51$\pm5$ & & Takeshima \\
508: & & & & & et al. 1994\\
509: EXO 2030+375 & T & 24 & 213 & & Angelini et al.\\
510: & & & & & 1989\\
511: A 0535+262 & T & 9.71 & 25-72 & & Finger et al.\\
512: & & & & & 1996\\
513: \hline
514: \vspace*{0.02in}
515: \end{tabular}
516: $\rm ^a${P: persistent X-ray source, T: transient X-ray source}\\
517: \end{center}
518: \normalsize
519: }
520:
521: Although we conclude that the UV QPOs are probably emitted by HZ~Her, they
522: may well be {\it reprocessed} from QPOs originating in the accretion disk,
523: either as a result of some structure in the disk {\it emitting} with the
524: QPO frequency, or {\it shadowing} the star from X-ray emission with the
525: same frequency. A comparison with optical and X-ray QPOs recently
526: discovered in other X-ray pulsars supports this conclusion (see Table~2
527: for a summary of these systems.)
528:
529: The KZ~TrA/4U~1627-67 system contains an X-ray pulsar with a 7.7~s period,
530: and shows 48 mHz QPOs in both X-ray and optical lightcurves (Chakrabarty
531: 1998). The optical QPOs, with an amplitude of $3-5$\%, are emitted by the
532: surface of the companion in response to X-ray illumination. The QPO
533: frequency is probably the orbital frequency of a structure revolving about
534: the neutron star (Kommers, Chakrabarty, \&\ Lewin 1998). The X-ray binary
535: pulsar SMC~X-1
536: shows X-ray power spectral turnover at $\sim10$~mHz (Angelini, Stella,
537: \&\ White 1991) and QPOs at 60~mHz (Wojdowski et al. 1998). Transient
538: X-ray pulsars also show QPOs; XTE J1858+034 has QPOs with $\nu=110$~mHz
539: with rms amplitude 3.7-7.8\%\ depending on the energy band (Paul \&\ Rao
540: 1998). The transient
541: Be system 4U~1907+09 shows very narrow QPO with $\nu\approx55$~mHz during
542: bursts (in'tZand, Baykal, \&\ Strohmayer 1998).
543:
544: These sources thus show QPOs with remarkably similar frequencies, and as
545: they are all X-ray pulsars they probably have similar magnetic field
546: strengths (for 4U~1627-67, BeppoSAX observations of a cyclotron feature
547: imply $B\approx3\times10^{12}$~G, Orlandini et al. 1998). It thus seems
548: natural to try to associate the QPOs with the region where the magnetic
549: field disrupts the disk. The Alfv\'en radius (where the magnetic energy
550: density equals the gas ram pressure) in Her~X-1 is probably
551: $r_A=2-7\times10^{8}$~cm (McCray \&\ Lamb 1976). Within this region, the
552: Keplerian frequency is $>100$~mHz, higher than the observed QPOs.
553: The QPO may be due to Keplerian rotation further out in the disk,
554: but a physical reason why this region is singled out is lacking.
555:
556: The 8 mHz QPO in Her~X-1 is even more problematic. If it occurs as the
557: result of Keplerian rotation, it arises at a radius in the disk of
558: $r=6\times10^{9}$~cm, even further from the Alfv\'en radius.
559:
560: We also consider a ``beat-frequency'' interpretation (Alpar \&\ Shaham
561: 1985), in which
562: the observed QPO frequency $\nu_{\rm QPO}=45\mbox{mHz}=\nu_{\rm ns}-
563: \nu_K$, where $\nu_{\rm ns}=808\mbox{mHz}$ is the neutron star
564: spin frequency and $\nu_K$ is the Keplerian frequency of some material
565: in the disk. In order for $\nu_{\rm QPO}$ to be a beat frequency,
566: the Keplerian frequency must be within 5\%\ of $\nu_{\rm ns}$.
567: This seems an unlikely coincidence, although we must have
568: $\nu_{\rm ns}<\nu_K$ or the gas will be expelled from the disk, spinning
569: down the pulsar via ``propellor effect'' (Illarionov \&\ Sunyaev 1975).
570: We note that Bepposax observations ending $<2$~days before the start
571: of the HST observations reported here (Parmar et al. 1999) show
572: that the pulsar has indeed spun-down, reversing $\approx5$ years of
573: spin-up in a period of months. The beat frequency interpretation
574: of the QPOs requires that the Alfv\'en radius be near the corotation
575: radius in Her~X-1 (at $2\times10^8$~cm), and this does appear to be the
576: case.
577:
578: The transition between Keplerian rotation and corotation occurs at a
579: radius $r_0\approx0.5 r_A$ (Ghosh \&\ Lamb 1979) over a radial zone of
580: extent $\Delta r=0.03r_0$. In this range, $\nu_K$ varies by
581: $\approx5$\%, naturally leading to $\nu_{QPO}\approx50$~mHz, given
582: $\nu_K=\nu(r_0)\approx\nu_{\rm ns}$. (However, this mechanism
583: would not be able to explain the 8~mHz QPO, which requires the
584: Keplerian frequency to be within $\approx0.5$\%\ of the pulse
585: frequency.)
586:
587: Further bolstering the beat-frequency case, we have found marginal
588: evidence for oscillations at 760~mHz=$808-45$~mHz in the X-ray data from
589: RXTE from our 1998 campaign. As an example of how this could arise in a
590: beat-frequency scenario, if a blob in orbit around the neutron star has an
591: orbital frequency of 760~mHz, it will cover the X-ray beam from the pulsar
592: (which rotates at 808~mHz) and prevent it from illuminating the companion
593: star with a frequency of $808-760$~mHz$=48$~mHz. The (marginally)
594: observed 760~mHz X-ray signal would then be a direct signal from this
595: blob, due either to emission or obscuration.
596:
597: It seems more likely that the reprocessed QPOs are due to material that
598: absorbs the X-rays and prevents them from heating HZ~Her than material
599: that emits X-rays. Whether the QPO frequency results from Keplerian
600: rotation or a beat-frequency, the accretion disk is too cold at the
601: implied radii to emit X-rays that significantly heat the companion.
602:
603: A third possible explanation for the QPOs, that may explain the presence
604: of {\bf both} QPOs, is provided by Titarchuk \&\ Osherovich (2000,
605: submitted). In this explanation, the higher frequency QPO results from a
606: global vertical oscillation of the disk, whereas the lower frequency QPO
607: results from Keplerian rotation.
608:
609: We detected UV continuum QPOs during an anomalous low state of Her X-1.
610: Although there may be a connection between the QPO and the low state,
611: time-resolved UV observations near $\phi=0.5$ have not been common, and
612: similar optical QPO are seen in observations using the Keck telescope
613: (O'Brien, et al. 2000) during the short-on state (Figure~4). The IUE
614: observations and
615: the HST~FOS observations of Anderson et al. (1994) did not have the
616: required time resolution for the detection of QPOs.
617:
618: The QPO phenomenon in Her~X-1 should help us understand the origin of QPOs
619: in other sources. The simultaneous detection of X-ray and
620: UV QPOs during an On state could test our suggestion that HZ~Her
621: reprocesses X-ray QPOs.
622: The UV QPOs should lag the X-ray QPOs by $\approx10$~s, in a manner
623: that varies with the orbital phase.
624: Simultaneous measurement of UV and optical QPOs could determine the
625: spectrum of the reprocessor. These analyses will test our proposal that
626: the QPOs originate on the star in response to disk QPOs, and may help us
627: distinguish between the possible origins in the disk that we have
628: suggested.
629:
630: \acknowledgements
631:
632: We would like to thank Vladimir Osherovich and Lev Titarchuk for
633: discussions and suggestions.
634: Based on observations with the NASA/ESA {\it Hubble Space Telescope},
635: obtained at the Space Telescope Science Institute, which is operated by
636: the Association of Universities for Research in Astronomy, Inc., under
637: NASA contract GO-05874.01-94A. BB and SDV supported in part by NASA
638: (NAG5-2532, NAGW-2685), and NSF (DGE-9350074). BB acknowledges an NRC
639: postdoctoral associateship. HQ is employed on PPARC grant L64621.
640: We would like to thank R. Kelley and D. Chakrabarty for suggestions.
641:
642: \begin{references}
643:
644: \reference{} Alpar, M.A. \&\ Shaham, J. 1985, Nature, 316, 239
645:
646: \reference{} Anderson, S.F., Wachter, S., Margon, B., Downes, R.A., Blair,
647: W.P., \&\ Halpern, J.P. 1994, ApJ, 436, 319
648:
649: \reference{} Angelini, L., Stella, L, \& White, N.E. 1991, ApJ, 371, 332
650:
651: \reference{} Bahcall, J.N., \&\ Bahcall, N.A. 1972, ApJ, 178, L1
652:
653: \reference{} Baushev, A.N., \&\ Bisnovatyi-Kogan, G.S. 1999, ARep, 43,
654: 241
655:
656: \reference{} Belloni, T., Hasinger, G., \&\ Kahabka, P. 1991, A\&A, 245,
657: L29
658:
659: \reference{} Boroson, B., Kallman, T., Vrtilek, S.D., Raymond, J., Still,
660: M., Bautista, M., \&\ Quaintrell, H. ApJ, 529, 414
661:
662: \reference{} Boroson, B., Vrtilek, S.D., McCray, R., Kallman, T., \&\
663: Nagase, F. 1996, ApJ, 473, 1079
664:
665: \reference{} Boyd, P.T., Carter, P.H., Gilmore, R., \&\ Dolan, J.F. 1995,
666: ApJ, 445, 861
667:
668: \reference{} Boyle, S.J., Howarth, I., Wilson, R., \&\ Raymond, J. 1986,
669: in New Insights in Astrophysics: 8 Years of UV Astronomy with IUE, ed.
670: E.J. Rolfe (Noordwijk: ESA), 471
671:
672: \reference{} Chakrabarty, D. 1998, ApJ, 492, 342
673:
674: \reference{} Crosa, L., \&\ Boynton, P.E. 1980, ApJ, 235, 999
675:
676: \reference{}Dal Fiume, D., Orlandini, M., Cusumano, G., Del Sordo, S.,
677: Feroci, M., Fronterra, F., Oosterbroek, T., Palazzi, E., Parmar, A.N.,
678: Santangelo, A., \&\ Segreto, A. 1998, A\&A, 329, 41
679:
680: \reference{} Deeter, J.E.,
681: Boynton, P.E.,
682: Miyamoto, S., Kitamoto, S., Nagase, F., \&\ Kawai, N. 1991, ApJ, 383, 324
683:
684: \reference{} Finger, M.H., Wilson, R.B., \&\ Harmon, B.A. 1996, ApJ, 459,
685: 288
686:
687: \reference{} Ghosh, P. \&\ Lamb, F.K. 1979, ApJ, 234, 296
688:
689: \reference{} Heil, C.E., \&\ Walnut, D.F. 1989, SIAM Review, 31, 4, 628
690:
691: \reference{} Heil, C.E., \&\ Walnut, D.F. 1990, in Recent Advances in
692: Fourier Analysis and its Applications, ed. J.S. Byrnes (Dordrecht:
693: Kluwer), 441
694:
695: \reference{} Howarth, I., \&\ Wilson, R. 1983, MNRAS, 202, 347
696:
697: \reference{} Illarionov, A.F. \&\ Sunyaev, R.A. 1975, A\&A, 39, 185
698:
699: \reference{} in'tZand, J.J.M., Baykal, A., Strohmayer, T.E. 1998, ApJ,
700: 496, 386
701:
702: \reference{} Kaper, L., et al. 2000, in preparation
703:
704: \reference{} Kimble, R.A., et al. 1998, ApJ, 492, L83
705:
706: \reference{} Kommers, J., Chakrabarty, D., \&\ Lewin, W.H.G. 1998, ApJ,
707: 497, L33
708:
709: \reference{} Leahy, D.A., Yoshida, A., Kawai, N., Matsuoka, M. 1992, In
710: NASA/Goddard
711: Space Flight Center, The Compton Observatory Science Workshop, p. 193
712:
713: \reference{} McCray, R., Lamb, F.K. 1976, ApJ, 204, 115
714:
715: \reference{} Middleditch, J., \&\ Nelson, J. 1976, ApJ, 208, 567.
716:
717: \reference{} Mihara, T., Makishima, K., Ohashi, T., Sakao, T., \&\
718: Tashiro, M. 1990,
719: Nature, 346, 250
720:
721: \reference{} O'Brien et al. 2000, in preparation
722:
723: \reference{} Orlandini, M., Fiume, D. D., Frontera, F., Del Sordo, S.,
724: Piraino, S., Santangelo, A., Sagreto, A., Oosterbroek, T., \&\ Parmar,
725: A.N. 1998, ApJ, 500, 1630
726:
727: \reference{} Parmar, A.N., Oosterbroek, T., Dal Fiume, D.,
728: Orlandini, M., Santangelo, A., Segreto, A., \&\ Del Sarda, S. 1999, A\&A,
729: 350, L5
730:
731: \reference{} Paul, B., \&\ Rao, A.R. 1998, A\&A, 337, 815
732:
733: \reference{} Scott, D.M. \&\ Leahy, D.A. 1999, ApJ, 510, 974
734:
735: \reference{} Still, M., O'Brien, K., Horne, K., Boroson, B., Titarchuk,
736: L.G., Vrtilek, S.D., Quaintrell, H., \&\ Fiedler, H. 2000, in preparation
737:
738: \reference{} Takeshima, T., Dotani, T., Mitsuda, K., \&\ Nagase, F. 1994,
739: ApJ, 436, 871
740:
741: \reference{} Takeshima, T., Dotani, T., Mitsuda, K., \&\ Nagase, F.
742: 1991, PASJ, 42, L42
743:
744: \reference{} Tr\"umper, J., Pietsch, W., Reppin, C., Voges, W., Staubert,
745: R., \&\ Kendziorra. 1978, ApJ, 219, L105
746:
747: \reference{} van der Klis, M. 1989, in Timing Neutron Stars, H. \"Ogelman
748: and E.P.J. Van den Heuvel (eds.), Kluwer, Dordrecht (NATO ASI Series C
749: 262), p. 27
750:
751: \reference{} Vrtilek, S.D., Quaintrell, H., Boroson, B., Still, M.,
752: Fielder, H.,
753: O'Brien, K., Raymond, J., Kallman, T., \&\ McCray, R. 2000, in
754: preparation
755:
756: \reference{} Vrtilek, S.D., \&\ Cheng, F.H. 1996, ApJ, 465, 915
757:
758: \reference{} Wojdowski, P., Clark, G.W., Levine, A.M., Woo, J.W., Zhang,
759: S.N. 1998, ApJ, 502, 253
760:
761: \reference{} Woodgate, B.E. et al. 1998, PASP, 110, 1183
762:
763: \end{references}
764:
765:
766: \end{document}
767: