1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{natbib}
3: \usepackage{longtable}
4: \begin{document}
5: \title{Disk reflection signatures in the
6: spectrum of the bright Z-source GX 340+0
7: \footnote{Based on
8: observations obtained with XMM-Newton, an ESA science mission with
9: instruments and contributions directly funded by ESA Member States
10: and NASA}}
11:
12: \author{A. D'A\`i \altaffilmark{1}, R. Iaria\altaffilmark{1},
13: T. Di Salvo\altaffilmark{1}, G. Matt \altaffilmark{2}
14: N. R. Robba\altaffilmark{1}}
15:
16: \altaffiltext{1}{Dipartimento di Scienze Fisiche ed Astronomiche,
17: Universit\`a di Palermo, via Archirafi 36 - 90123 Palermo, Italy;
18: email:dai@fisica.unipa.it}
19: \altaffiltext{2}{Dipartimento di Fisica, Universit\`a degli Studi Roma Tre,
20: via della Vasca Navale 84, I-00146 Roma, Italy}
21:
22: \begin{abstract}
23:
24: We present the preliminary results of a 50 ks long XMM-Newton
25: observation of the bright Z-source GX 340+0. In this Letter we focus
26: on the study of a broad asymmetric emission line in the Fe K$\alpha$
27: energy band, whose shape is clearly resolved and compatible with a
28: relativistically smeared profile arising from reflection on a hot
29: accretion disk extending close to the central accreting neutron
30: star. By combining temporal and spectral analysis, we are able to
31: follow the evolution of the source along its Horizontal Branch.
32: However, despite a significant change in the continuum emission and
33: luminosity, the line profile does not show any strong correlated
34: variation. This broad line is produced by recombination of highly
35: ionized iron (Fe XXV) at an inferred inner radius close to 13 R$_g$,
36: while the fit requires a high value for the outer disk radius. The
37: inclination of the source is extremely well constrained at 35 deg,
38: while the emissivity index is -2.50.
39:
40: \end{abstract}
41: \keywords{line: identification -- line: formation -- stars: individual
42: GX 340+0 --- X-rays: binaries --- X-rays: general}
43: \maketitle
44:
45: \section{Introduction}
46:
47: GX 340+0 is a bright Low-Mass X-ray Binary (LMXB) belonging to the
48: class of the Z sources \citep{hasinger89}, and its inferred luminosity
49: is close to the Eddington limit (2 $\times$ 10$^{38}$ erg/s for a 1.4
50: M$_{\odot}$ NS). The source has also a radio counterpart, from which
51: a distance of 11 $\pm$ 3 kpc has been estimated \citep{penninx93};
52: radio emission from the counterpart is, however, highly variable, and
53: seems to be correlated with the X-ray flux when the source is in its
54: Horizontal Branch (HB), but anticorrelated on the other branches
55: \citep{oosterbroek94}; it can also show periods of radio quenching,
56: becoming extremely faint \citep{berendsen00}.
57:
58: Temporal analysis studies have shown a complex phenomenology, linked
59: to the accretion state, with characteristics typical of the Z-class
60: \citep{jonker00}. The power spectrum shows a low-frequency (tens of
61: Hz) quasi-periodic oscillation when the source resides on its HB
62: (called Horizontal Branch Oscillation, HBO) while at higher
63: frequencies the source shows, at the same time, twin kiloHz
64: Quasi-Periodic Oscillations (kHz QPOs), whose centroid frequencies are
65: correlated with the HBO peak frequency. Although no burst has been
66: yet observed, the very fast timing variability (at more than 800 Hz)
67: and the similarities with the other Z-sources indicate that the
68: compact object is a neutron star.
69:
70: The spectral properties of the source have not been fully investigated
71: so far; \citet{schulz93} studied the 2--12 keV spectrum using EXOSAT
72: data; the spectrum could be well described by a single component due
73: to thermal Comptonization of soft photons, emerging from the NS
74: surface, in a hot corona of moderately optical thickness ($\tau
75: \simeq$ 5--6);
76:
77: \citet{lavagetto04} presented the first broadband (0.1--200 keV)
78: spectrum of the source using data from a BeppoSAX observation. The
79: spectrum could be decomposed into the sum of a soft thermal component
80: of temperature of 0.5 keV, an optically thick Comptonized component,
81: and an excess at energies above 20 keV that they fitted with a simple
82: power law. A high resolution spectrum of the source was studied by
83: \citet{ueda05}, using a Chandra observation; Chandra data clearly
84: showed the presence of an emission line, fitted with a simple Gaussian
85: profile, at 6.57 keV with 40 eV equivalent width.
86:
87: In this Letter we report the results of an XMM-Newton observation of
88: GX 340+0. Thanks to the large collecting area of the EPIC-PN
89: instrument and its good spectral resolution, we find evidence of a
90: strong broad emission line at $\sim$ 6.7 keV; the line profile is very
91: well reproduce by a smeared disk-reflected profile; we find, moreover,
92: evidence ot other disk reflection signatures supporting the common
93: scenario, valid for AGNs, galactic black hole and NS systems, that
94: broad iron fluorescence lines are produced by reflection of photons
95: from a hot corona on the surface of an accretion disk that extends
96: very close to the compact object.
97:
98: \section{Observation, data reduction and spectral selection criteria}
99:
100: GX 340+0 was observed with XMM-Newton from 2007 September 2 at
101: 13:18:00 to September 3 at 02:32:05 UTC, for a total observing time of
102: 47640 s (Obs.ID 0505950101). However, because of the high count rate
103: reached in the EPIC-PN CCDs during most of the observation, the
104: effective exposure was reduced by data losses due to telemetry
105: saturation. The total good time intervals, thus, resulted in 40562 s
106: effective exposure. In this work we use data collected by the EPIC-PN
107: instrument in fast timing mode; the source region (the background
108: region) has been selected using CCD rows RAWX = 28-48 and RAWY = 0-200
109: (RAWX = 2-8 and RAWY = 0-200). No background flaring was present
110: during the observation. We selected only events with PATTERN $\leq$ 4
111: (singles and doubles) and FLAG = 0 and restricted our analysis to the
112: energy range 2.2–11.8 keV. We excluded data from PN below 2.2 keV and
113: rely on the Reflection Grating Spectrometers (RGS1 and RGS2, in the
114: 0.4--2.0 keV energy band) for the soft energy band, because PN data
115: present strong systematic residuals of instrumental origin which
116: affect the overall continuum determination. PN data have been
117: progressively rebinned to avoid data oversampling; RGS data are
118: rebinned in order to have at least 25 counts per energy channel.
119: Source, background spectra and response matrices were extracted using
120: the XMM-Newton Science Analysis Software (SAS, ver. 7.1.0). Spectral
121: analysis was performed using Xspec ver. 12.4.0. Temporal analysis
122: was performed using the FTOOLS ver. 6.4.1.
123:
124: In Figure \ref{fig1} we show the light curve obtained by the EPIC-PN
125: instrument in the 0.5--10 keV energy range. The source shows
126: significant X-ray variability, with a 40\% decrease in count rate
127: between the beginning and end of the observation.
128:
129: Besides the spectral analysis we also performed a study of the
130: temporal variability. We extracted power spectral densities (PSD) from
131: the EPIC-PN data in the frequency range 1/16 Hz to 512 Hz. We obtained
132: a single PSD after averaging every 128 PSD, thus having a total of 20
133: PSD for the entire observation. We fitted the averaged PSD using two
134: zero-centered Lorentzians, to account for the broadband low frequency
135: noise, a constant term to fit the white noise and a Lorentzian to fit
136: a broad quasi-periodic oscillation. We identify this broad QPO as the
137: HBO \citep{jonker00}; this feature, as clearly shown by
138: \citet{jonker00} is a good tracer of the overall temporal variability,
139: which in turn, can be used to track the accretion state of the source
140: in the X-ray Color-Color Diagram. With the exception of the first two
141: PSD, when the source was at the highest count rate, we clearly detect
142: the HBO in all the remaining PSD. The HBO centroids span a range of
143: frequencies from 17 to 37 Hz, and we noted this frequency to be
144: correlated with the source count rate. We used these results to
145: derive the good time intervals from which we extracted the
146: corresponding energy spectra. After some trials, we decided to slice
147: the observation into five segments; this choice allows us to trail the
148: spectral evolution of the source, still obtaining a high statistics
149: spectrum to constrain all the spectral parameters. We denote from
150: Spectrum 1 to Spectrum 5 the PN and RGS energy spectra extracted in
151: the time intervals when the HBO frequency was in the 17--22 Hz, 22--27
152: Hz, 27--32 Hz, 32--37 Hz, $>$ 37 Hz (or not detected) range,
153: respectively (see the time selected intervals shown in Figure
154: \ref{fig1}). In Figure \ref{fig2}, we show the time averaged power
155: spectra for the time intervals 1 to 4: the HBO evolved during the
156: observation, moving from lower frequencies, small widths and high
157: spectral power (Spectrum 1) to higher frequencies, larger widths and
158: less spectral power (Spectrum 4). The HBO is probably undetected in
159: Spectrum 5 because of the lower statistics, and because it is too
160: broad to be resolved.
161:
162: Given the high count rate of the source, we carefully investigated
163: pile-up related effects by extracting our spectra from rectangular
164: regions with and without the inclusion of the brightest central rows
165: \footnote{Strictly following the SAS thread as described in in
166: http://heasarc.nasa.gov/docs/xmm/sas/USG/node63.html}. Through
167: several different trials, and with the help of the $epatplot$ tool, we
168: concluded that spectra 1 to 4 are not significantly affected by
169: pile-up, and, as far as we are mostly concerned with the spectral
170: shape of the broad iron Fe K$\alpha$ line, the spectral parameters
171: are, within the relative uncertainties, self-consistent for region
172: masks with and without the inclusion of the central brightest rows.
173: On the other hand, the time-selected region 5 is affected by pile-up;
174: we extracted, therefore, the associated energy spectrum excluding the
175: central 2 brightest pixel rows, in order to avoid significant pile-up
176: dependent spectral distortions.
177:
178: \section{Spectral Analysis}
179:
180: We analyzed the time selected spectra independently, adopting a
181: continuum model consisting of soft thermal disk emission
182: (\texttt{diskbb} component, \citealt{makishima86}) and a thermal,
183: harder, blackbody emission (component \texttt{blackbody}). We used
184: for the interstellar absorption the \texttt{vphabs} model with
185: crss-sections of \citet{verner96} and table abundances of
186: \citet{wilms00}. We found significant residuals at the neutral iron
187: edge (7.11 keV); leaving the iron abundance free to vary resulted in a
188: significant improvement in the $\chi^2$ and in the residuals
189: scattering; iron is under-abundant with respect to the assumed
190: \cite{wilms00} solar value (see also \cite{ueda05}).
191:
192: Although broadband spectra of Z-sources reveal that the hard component
193: is better described by thermal Comptonization models \citep[see
194: e.g.][]{lavagetto04}, XMM data do not allow to discriminate among
195: different models of X-ray continuum, or to reliably constrain the
196: parameters of the Comptonization model. We tested, however, using a
197: variety of different spectral models, that the choice of the continuum
198: alter, only marginally the overall $\chi^2$ value of our fits and the
199: determination of the local features parameters, which is the main
200: objective of the present Letter.
201:
202: Superimposed to the continuum emission, a few discrete features are
203: always present: a broad emission feature in the Fe K$\alpha$ region, a
204: weaker but also broad emission line at $\sim$ 3.9 keV and an
205: absorption edge, whose energy threshold is in the 8.7--9.0 keV range.
206:
207: We preliminarly investigated the shape of the broad iron line, using
208: the total time averaged PN spectrum (energy range 3-12 keV), excluding
209: only the time interval 5 (see Figure 1) because pile-up affected this
210: part of the observation. Using a Gaussian profile to fit the line
211: emission, resulted in a line position of 6.76$\pm$0.02 keV,
212: 0.24$\pm$0.02 keV width and (2.0$\pm$0.2)$\times$10$^{-3}$ photons
213: cm$^{-2}$ s$^{-1}$ normalization value. Using a relativistically
214: smeared disk-reflected profile (modelled using the \texttt{diskline}
215: profile of \citealt{fabian89}), resulted in a rest frame energy of the
216: line at 6.69$\pm$0.02 keV, an inner radius R$_{in}$=13$\pm$3 R$_g$, an
217: inclination angle 34.6$\pm$1.3 deg and emissivity index
218: 2.50$^{+0.09}_{-0.13}$. The R$_{ext}$ has a best fitting value $\sim$
219: 10$^{4}$ R$_g$ but it is rather unconstrained with no upper limit and
220: a 90\% c.l. lower limit at 3000 R$_g$. The $\Delta \chi^2$ between
221: these two profiles is 25 (+3 dof in the \texttt{diskline} model)
222: indicating the relativistic profile as statistically more favorite
223: than the symmetric Gaussian profile.
224:
225: We assumed, thereafter, that the 3.9 keV emission line, which we
226: identify with the Ca XIX Ly$\alpha$ transition is also produced in the
227: disk reflecting plasma, and we fit it using another \texttt{diskline}
228: component, with R$_{in}$ \footnote{Inner and outer disk radii in the
229: \texttt{diskline} model are expressed in units of gravitational
230: radii (R$_{g}$), which for a 1.4 solar masses NS corresponds to 2.1
231: km.} , R$_{ext}$, emissivity index and inclination bound to the Fe
232: K$\alpha$ \texttt{diskline} component.
233:
234: We turned, then, to the analysis of the time selected spectra.
235: Because of the high statistics it is possible to well constrain all
236: the \texttt{diskline} spectral parameters, except the outer radius for
237: which we derived the largest uncertainties. We fixed this parameter to
238: the reference value of 10$^4$ R$_g$, after having tested that this
239: value lies extremely close ($\Delta \chi^2 \leq$ 1) to the $\chi^2$
240: minimum for each spectrum. We find that the line shape, within the
241: statistical uncertainties, does not significantly change during most
242: of the observation, despite a significant change of the continuum flux
243: and continuum parameters (see Table\ref{tab1}) with best fitting
244: values all consistent with the total time averaged values.
245:
246: For Spectra 4 and 5, corresponding to the highest luminosities reached
247: by the source during the observation, the temperature of the
248: \texttt{blackbody} component was rather unconstrained by the fit, so
249: that we kept it frozen in the fits to a reference value of 3.5 keV, in
250: order to avoid unphysical higher values due to the lack of high energy
251: spectral coverage. In Table 1 we comprehensively show the fitting
252: results for the five time selected spectra. We representatively show
253: in Figure \ref{fig3} the unfolded spectrum, with the contribution of
254: the additive components and a snapshot of the iron region, of Spectrum
255: 2.
256:
257: \section{Discussion and Conclusions}
258:
259: In this Letter we report on the first unambiguous determination of the
260: broad disk origin of the iron line in the Z-source GX 340+0. Broad
261: asymmetric iron lines present in the energy range 6.4--6.97 keV have
262: been discovered first in AGNs, and later, in galactic black hole
263: candidates, and very recently also in a NS system
264: \citep{disalvo05,bhattacharyya07,cackett08}. The common scenario for
265: the origin of such lines is disk reflection, where the line
266: deformation primarily arises from the high Keplerian velocity fields
267: of the disk reflecting matter and from general relativistic effects
268: produced by the intense gravitational well of the compact object. Our
269: analysis reveals, for the first time in a NS accreting at the
270: Eddington limit that the iron line shape in GX 340+0 is unambigously
271: described by a disk reflection profile, finding also evidence of other
272: reflection signatures (an absorption edge at $\sim$ 8.8 keV and a
273: broad Ca XIX emission line).
274:
275: During this XMM-Newton observation we can closely follow the spectral
276: evolution of GX 340+0 through all of its horizontal branch. At its
277: leftmost point, corresponding to the lowest count rate and to a
278: frequency of the HBO of $\sim$ 17 Hz, we derive the lowest
279: temperatures both for the disk (1.60 keV) and the hard emission (2.55
280: keV); as the source moves towards the hard apex, there is a continuum
281: rising of the temperatures and of the X-ray flux; assuming a distance
282: of 11$\pm$3 kpc and an inclination of the disk as inferred from the
283: diskline profile, we derive a change in the source luminosity from
284: (1.67$\pm$0.1) $\times$ 10$^{38}$ erg/s (Spectrum 1) to (2.57$\pm$0.2)
285: $\times$ 10$^{38}$ erg/s (Spectrum 5). The inner disk radius as
286: derived from the \texttt{diskbb} normalization (see Table 1) gives a
287: value in the 8--11 km range; considering that the inner disk radius
288: can be underestimated by a factor of two \citep{merloni00}, we find
289: that these values are in agreement with the ones derived from the
290: \texttt{diskline} line profile.
291:
292: Despite this remarkable continuum change, the profile of the broad
293: iron diskline does not sensibly change. The rest frame energy of the
294: \texttt{diskline} component present in the iron region indicates that
295: the line is most probably produced by He-like iron ions at a short
296: distance from the compact source. In Spectrum 5, because the
297: inclination angle and the emissivity of the line were bad constrained
298: by the fit, we kept them frozen to the best values of the time
299: averaged spectrum; we noted, therefore, a small shift of the rest
300: frame line energy towards higher energies, possibly indicating a
301: greater contribution from the H-like ions.
302:
303: The equivalent width of the line decreases from Spectrum 1 ($\sim$ 60
304: eV) to Spectrum 5 ($\sim$ 30 eV), indicating that the line does not
305: respond to the soft X-ray flux, which, on the contrary increases, but
306: presumably to the hard X-ray emission, above 12 keV, which falls out
307: from the observed spectral coverage of the EPIC/PN instrument. The
308: inclination angle of the disk and the emissivity index are very well
309: constrained by our fits, with relative uncertainties of only a few
310: percent. Contrary to what found in galactic black holes, the
311: emissivity index is not very steep, indicating a more extended
312: illuminating corona above the disk; we are not able to identify any
313: significant change in the value of this parameter during the
314: observation, despite a large increase in the temperatures of the
315: continuum components and their related fluxes.
316:
317: Recently \citet{laurent07} proposed an alternative scenario for the
318: formation of broad iron lines in LMXBs, where extensive red wings
319: could be formed by recoil of line photons in an optically thick medium
320: expanding, or converging, at relativistic velocities. Since a
321: spectral line model, adapted to this scenario, is not yet available
322: for fitting X-ray data, we are not able to test this scenario using
323: our data. However, as pointed in \citet{pandel08} for the case of 4U
324: 1636-536, we note that the lack of a narrow core and the presence of a
325: blue wing in the iron line profile of GX 340+0 is a strong indication
326: of the disk reflection origin.
327:
328: Most theoretical models on the QPO generation in NS LMXB agree to
329: identify the shortest dynamical timescales, i.e. the highest frequency
330: QPO at $\sim$ 1 kHz, with the frequency of the Keplerian motion of
331: matter at the inner rim of an accretion disk. From the HBO frequency
332: and the relation known to exist between this and the upper kHz QPO
333: frequency \citep{jonker00}, we can infer the upper kHz QPO to be in
334: the 550--750 Hz frequency range. For a typical 1.4 M$_{\odot}$ NS this
335: frequency range corresponds to an inner disk radius of 9.8--12.1
336: R$_g$. The \texttt{diskline} inferred inner radii for the time
337: selected spectra are all fully consistent within this range, and give
338: support to the identification of highest QPO frequency with the inner
339: disk Keplerian frequency. Our fit results seem also to indicate a
340: decrease in the value of the inner disk radius as the source moves
341: towards higher luminosities. The errors associated to this parameter
342: are, however, still too large to claim a strict correlation and longer
343: exposures are needed to assest the significance of such a relation.
344:
345: The spectra show, together with the broad Fe K$\alpha$ emission line,
346: other reflection signatures, like a broad Ca XIX resonance line and a
347: high-energy absorbtion edge, compatible with the Fe XXV K-edge (8.83
348: keV energy in the laboratory frame). The extraordinary good quality
349: of the EPIC-PN spectrum of this source will allow to test even more
350: recent theoretical line profiles and fully self-consistent disk
351: reflection models. We aim at expanding this analysis in a forthcoming
352: paper.
353:
354: \acknowledgements Authors express their gratitude to the EPIC/PN team
355: and, especially, to Matteo Guainazzi for the kind assistance
356: offered us in the process of data extraction and reduction of the PN
357: data. Authors would like to thank the anonymous referee for the carefull
358: reading of the manuscript and the many suggestions.
359: Authors also thank Giuseppe Lavagetto
360: for useful discussions.
361:
362: \bibliographystyle{apj}
363: \bibliography{refs}
364: \clearpage
365:
366: \begin{figure}
367: \centering
368: \includegraphics[angle=-90, width=12cm]{f1.eps}
369: \caption{Light curve of GX 340+0 from the EPIC-PN, starting from MJD 54345.554.
370: Bin time 100 s. Vertical lines and numbers inside
371: the boxes indicate the time selection for the corresponding energy spectra and the PSD shown in Figure 2.}
372: \label{fig1}
373: \end{figure}
374: \clearpage
375: \begin{figure}
376: \centering
377: \includegraphics[width=12cm]{f2.eps}
378: \caption{Lehay normalized power density spectra for four time-selected intervals, corresponding
379: to the time selections 1-4 used in the spectral analysis. Peak frequencies of the HBO are reported in Table 1.}
380: \label{fig2}
381: \end{figure}
382:
383: \begin{figure}
384: \centering
385: \begin{tabular}{c}
386: \includegraphics[angle=-90, width=12cm]{f3a.eps}\\
387: \includegraphics[angle=-90, width=12cm]{f3b.eps}\\
388: \end{tabular}
389: \caption{Top panel: unfolded spectrum (in the f(E)*E$^2$ representation, together with the best-fit model
390: in the 1.0-10.0 keV band for the representative Spectrum 2.
391: Bottom panel: snapshot of the top panel figure in the 5.5--7.5 restricted energy band.}
392: \label{fig3}
393: \end{figure}
394:
395: \clearpage
396:
397: \input{tab1}
398:
399: \end{document}