EDP Sciences logo
Subscriber Authentication Point
Search
Menu
Home All issues Volume 542 (June 2012) A&A, 542 (2012) L27 Full HTML
Free Access
Issue
A&A
Volume 542, June 2012
Article Number L27
Number of page(s) 4
Section Letters
DOI https://doi.org/10.1051/0004-6361/201219283
Published online 30 May 2012

© ESO, 2012

1. Introduction

Spectra of accreting compact objects, such as active galactic nuclei and X-ray binary systems containing stellar mass black holes or weakly magnetized neutron stars (NS), often exhibit broad emission lines (FWHM up to  ~1 keV) at E ~ 6.4−6.97 keV, identified with Kα radiative transitions of iron at different ionization states. These broad lines are thought to originate from reprocessed emission that is relativistically smeared, in the inner parts of the accretion disk illuminated by the primary Comptonized spectrum. This process produces different spectral signatures, including a reflection hump that peaks between 20 − 30 keV, emission lines, and absorption edges (see Fabian et al. 2000, for a review). Owing to their relatively high abundance and fluorescence yield, the iron lines are the best suited features to diagnose the accretion flows close to the compact object, because the predicted line profile depends on the geometry – inclination and inner radius – of the system (Fabian et al. 1989). For NS systems, the inner disk radius sets an upper limit to the radius of the star, which can constrain the NS equation of state.

Piraino et al. (2000) first suggested a relativistic origin of the broad iron line observed in the NS system 4U 1728-34. Recently, thanks to the improved spectral capability of XMM-Newton and Suzaku, asymmetric and smeared broad Fe K lines have been observed in about 10 NS low-mass X-ray binaries (Di Salvo et al. 2009; D’Aì et al. 2009; Iaria et al. 2009; Pandel et al. 2008; Bhattacharyya & Strohmayer 2007; Cackett et al. 2008, 2010) and were interpreted as relativistic reflection features from the inner part of the accretion disk. As shown in the self-consistent approach of D’Aì et al. (2010), the broad features other than that of iron in the spectrum also support the “relativistic origin” scenario. However, the robustness of these results with respect to pile-up effects has been questioned. Ng et al. (2010), based on archival XMM observations, studied the Fe K emission of 16 NS systems, and concluded that the line profiles show no evidence of asymmetry when spectra are extracted from pile-up free regions. Conversely, Miller et al. (2010) conducted simulations for CCD based detectors to assess the impact of photon pile-up on relativistic disk line and continuum spectra, concluding that relativistic disk spectroscopy is generally robust against pile-up when the effect is modest.

Although the “relativistic origin” is a widely accepted interpretation, alternative physical mechanisms to explain the line broadening are still debated. A broad iron line can originate from thermal Comptonization of line photons in the so-called accretion disk corona (ADC; Brandt & Matt 1994), or can be caused by non-thermal Comptonization in strong outflow winds (Laurent & Titarchuk 2007). The predicted line profiles differ significantly from the relativistic disk line profiles.

In this Letter, we report detection of a broad and asymmetric iron line in a spectrum of the neutron star LMXB GX 3+1 with high statistics and excellent energy resolution obtained with the EPIC-pn camera on board XMM-Newton. We discuss in detail instrumental pile-up effects and show that the relativistic profile of the line is robust against biases from photon pile-up.

2. Observation and data analysis

GX 3+1 is a typical bright (F2−10   keV ~ 150−400 mCrab) atoll source with peculiar bursting behavior. It exhibits type-I bursts (Makishima et al. 1983; Kuulkers & van der Klis 2000) and superbursts, thought to arise from carbon shell flashes in the layers below the surface (Kuulkers 2002). The spectrum of the source observed with the BeppoSAX narrow field instruments has been modeled by Oosterbroek et al. (2001) with a disk-blackbody, a Comptonized component, and absorption by interstellar medium. Interestingly, a broad iron line was clearly detected but only marginally discussed by the authors.

GX 3+1 was observed on September 1, 2010 by XMM-Newton and simultaneously with RXTE for about 56 ks.

The XMM-Newton Observatory (Jansen et al. 2001) includes three 1500 cm2 X-ray telescopes each with a European Photon Imaging Camera (EPIC, 0.1−15 keV) at the focus. Two of the EPIC imaging spectrometers use MOS CCDs (Turner et al. 2001) and one uses pn CCDs (Strüder et al. 2001). Data products were reduced using the science analysis software (SAS) version 11.0. Owing to the expected high count rate from the source, the telemetry of EPIC MOS cameras was allocated to the EPIC pn camera, therefore no data from the MOS cameras are available. We did not analyze the RGS data since they do not cover the Fe K energy band.

The EPIC pn was operated in timing mode for the entire observation. In this mode only one CCD chip is operated and the data are collapsed into a one-dimensional row (44) and read out at high speed, the second dimension being replaced by timing information. This allows a time resolution of 30 μs. We used the SAS task epfast on the event files to correct for the effect of charge transfer inefficiency (CTI), which has been seen in the EPIC pn timing mode at high count rates1. Ancillary response files were generated using the SAS task arfgen following the recommendations in the XMM-Newton SAS user guide2 for piled-up observations in timing mode whenever applicable. Response matrices were generated using the SAS task rmfgen.

Since the average EPIC pn count rate was close to the 600 counts s-1 level at which pile-up effects start to be important, we investigated in detail if there was any pile-up before extracting the spectra. We used the SAS task epatplot, which uses the relative ratios of single- and double-pixel events as a diagnostic tool. These ratios deviate from standard values when there is significant pile-up in the pn camera timing mode.

The source coordinates fell in the center of Col. 37 of the CCD. Source events were first extracted from a 53.3′′ (13 columns) wide box centered on the source position; we call this Region 1. We also defined regions excluding the central columns. Region 2 has the central column (Col. 37) removed. Region 3 has the three central columns removed (Cols. 36 − 38). The relative ratios of single- and double-pixel events exhibits an energy-dependent deviation between the data and the model at E  ≳  4 keV when Region 1, including all source events, is used. The effects of pile-up are reduced for Region 2, with the central column removed. After removal of the three central columns, Region 3, any deviation in the ratios disappears, showing that pile-up has no significant effect. Spectra were extracted for each of the three regions, selecting only events with PATTERN  ≤  4 (singles and doubles) and FLAG = 0. The energy channels were grouped by a factor of four to avoid oversampling the instrumental energy resolution.

3. Results

To model the spectrum, we used the spectral analysis package XSPEC v.12.7.0. To avoid the systematic effects discussed in D’Aì et al. (2010) that were recently addressed in the EPIC-pn calibration note3 we removed the energy bins below 2.4 keV and those above 9 keV. We then fitted the continuum with a model widely used for atoll LMXBs (Piraino et al. 2007; Di Salvo et al. 2009; Barret & Olive 2002): a photoelectric absorbed two-component model (Model 1) containing a blackbody and a thermal Comptonized component described by the model comptt (Titarchuk 1994). For the photoelectric absorption, we used the XSPEC model phabs.

This model gave an unacceptable fit (χ2/d.o.f. = 2621/322) because there were several unmodeled features across the entire energy range, the most prominent being a broad profile with residuals from 5.5 to 7.3 keV and a maximum deviation of  ~14σ from the best-fit continuum (see Fig. 1, black points). Following the approach of Di Salvo et al. (2009), we added to the continuum three Gaussian emission lines at  ~3.3 keV (Ar XVIII),  ~3.9 keV (Ca XIX), and  ~6.6 keV (Fe XXIV and/or Fe XXV), respectively. This greatly improved the fit to χ2/d.o.f. = 557/313. All these lines are significantly broader than the pn energy resolution, with Gaussian σ of 140 eV for Ar, 180 eV for Ca, and 290 eV for Fe. The fit was additionally improved by adding an absorption edge at 8.40 keV resulting in χ2/d.o.f. = 521/311.

Because the residuals of the Fe line profile, shown in Fig. 1, suggest an asymmetric shape, we replaced the Gaussian at 6.6 keV with a diskline profile (Fabian et al. 1989), obtaining again a significant improvement of the fit with χ2/d.o.f. = 451/311. Adding two parameters improved the fit by Δχ2 = 100, corresponding to an F-test probability of chance improvement of 3  ×  10-15. However, the comptt parameters are poorly constrained and the high value of the optical depth, near 10, suggests that the Comptonized component is close to a Wien spectrum. Therefore, we used a second two-component continuum model containing a diskbb component and a simple blackbody component (Model 2). For this model all continuum parameters are well constrained.

thumbnail Fig. 1

Residuals in units of σ in the pn energy range used with respect to the best-fitting continuum model, composed of an absorbed (phabs) diskbb and a blackbody component for the spectrum extracted in Region 1 (black), Region 2 (red) and Region 3 (blue).

As for Model 1, using a diskline profile for the iron emission line added to Model 2 significantly improves the χ2/d.o.f. with respect to a Gaussian from 549/315 to 441/313. The Δχ2 = 108 for the addition of two parameters corresponds to an F-test probability of chance improvement of 1.3  ×  10-15. The diskline parameters are well constrained, except for the outer disk radius, which we fixed after some trials at Rout = 4500 (GM/c2). The best-fit parameters are reported in Table 1 and the unfolded spectrum and residuals are shown in the upper panel of Fig. 2.

Table 1

Best-fit parameters for spectra of GX 3+1 observed with the XMM-Newton EPIC-pn using different models and data extraction.

thumbnail Fig. 2

Top: unfolded pn spectrum and individual model components for Model 2 using data from Region 1 which includes piled-up events. Bottom: the same for Model 2, with NH fixed as described in the text, for Region 3, which excludes piled-up events. Residuals in units of σ are shown in each panel.

The best-fit value obtained for the energy of the iron line, Eline, suggests an ionization parameter of log ξ ≃ 2.5 − 2.7, with ξ defined as LX/(nr2), where LX is the ionizing X-ray luminosity, n the density in the reflector, and r2 is the distance from the central source of emission.

In Col. 2 of Table 1 we report the best-fit parameters of the spectrum of GX 3+1 for data extracted in Region 2 (without the central Col. 37). All spectral parameters are compatible with those obtained for data extracted from Region 1. The diskline gives again a better fit, with a Δχ2 = 73 for the addition of two parameters, corresponding to an F-test probability of chance improvement of 2  ×  10-12.

In the bottom panel of the Fig. 2, we show the unfolded pn spectrum of GX 3+1 with individual components and residuals in units of σ for data extracted from Region 3 (without the central Cols. 36 − 38). The best-fit parameters are reported in Col. 3 of Table 1. Leaving all parameters free and using the 2.4 − 9 keV energy range, the NH value and the diskbb normalization slightly differ from those obtained for Regions 1 and 2, while the line parameters remain consistent. However, if we use data up to 11 keV, the NH value and the diskbb normalization agree with those obtained for the other regions, since this spectrum is not affected by pile-up even at high energies. Therefore, we fixed NH to the value obtained with data up to 11 keV, and again fitted the spectrum in the energy range 2.4 − 9 keV. All parameters are now consistent with those obtained for Region 1, as shown in Table 1, Col. 3. The diskline model improves the fit, with a Δχ2 = 36 for the addition of two parameters and an F-test chance improvement probability of 2  ×  10-7.

As shown in Fig. 1, where we show the residuals in units of σ with respect to the best-fitting continuum model, for the spectrum extracted in Region 1 (black), Region 2 (red) and Region 3 (blue), the line shape does not strongly depend on the photon pile-up fraction. To check if the low-energy lines are affected by Doppler and relativistic distortion, we replaced the Gaussian lines in Model 2 with disk lines, with all the parameters fixed at those of the iron diskline except the line centroid energy and the normalization (Model 3). We obtained an equally good fit with the elimination of two parameters, chi2/d.o.f. = 369/316 (see the rightmost column of Table 1).

4. Discussion

We reported the detection of broad emission lines at  ~3.3 keV,  ~3.9 keV and  ~6.6 keV, corresponding to Ar XVIII, Ca XIX, and most likely a blend of Fe XXIV and Fe XXV, in high-quality XMM-Newton X-ray spectra of the LMXB and atoll source GX 3+1 obtained when the source was in a relatively low-intensity state. A broad, σ  ~  1.0 keV, iron line feature was previously reported by Oosterbroek et al. (2001), who simply mentioned that the line might originate from reflection from an accretion disk or from Compton broadening in the external parts of a  ~3 keV corona. We showed that the disk-reflection scenario is consistent with the high-quality data from the XMM-Newton EPIC-PN. The iron line’s broadening and shape are therefore affected by Doppler and relativistic effects at the inner edge of the accretion disk. In addition, Compton broadening is expected due to the hot ionized disk (Ross & Fabian 2007). Such a component does not affect the skewness of the line, however, and indeed we have verified that even including Compton broadening, relativistic smearing is required with high statistical significance.

In our study of the spectral properties, we carefully investigated effects caused by instrumental photon pile-up. We conclude that the line shape does not significantly depend on the photon pile-up fraction. Our results agree with the conclusion of Miller et al. (2010) that relativistic disk spectroscopy is generally robust against pile-up when this effect is modest.

We measured an iron line rest-frame energy, E ~ 6.61 keV that is compatible with line emission from Fe XXIV and Fe XXV, showing that the reflecting matter is ionized and leading to an estimate for the ionization parameter of log ξ ≃ 2.4 − 2.7. The line appears to be produced at radii extending inward to Rin  =  (25  ±  4)Rg, where Rg = GM/c2 is the gravitational radius. Because Rg ~ 1.5 km for a 1   M⊙ neutron star, the inner radius is then Rin  =  (50  ±  8)m1.4 km, where m1.4 is the neutron star mass in units of 1.4   M⊙. For reasonable values of the neutron star mass, our results indicates that the disk is truncated quite far from the NS surface. The apparent inner radius of the soft disk blackbody emission gives a radius of 8  ±  2 km, calculated for a distance to the source of 4.5 kpc (Kuulkers & van der Klis 2000). Note that this value is thought to underestimate the true inner radius by a factor of more than 2 (Merloni et al. 2000). The power law dependence of the disk emissivity, rbetor, is parametrized by the index betor, which is  − (2.61  ±  0.03), indicating that the disk emission is dominated by irradiation by a central source (see Fabian et al. 1989). Finally, the disk inclination with respect to the line of sight is also well determined, i  =  (37  ±  2) degrees. This agrees well with the intermediate inclinations, between 20 and 65 degrees, found in other sources, as indeed expected from selection criteria.

Signatures of a disk reflection are also the discrete features present at soft X-rays from Ar XVIII, and Ca XIX, respectively. We note that the scenario emerging from this work on GX 3+1 is coherent with results obtained for other sources and is in particular very similar to the case of 4U 1705-44 (Di Salvo et al. 2009; D’Aì et al. 2010).

We focused on the robust detection of the relativistic iron line in the source, taking into account systematic instrumental effects caused by pile-up. Additional studies based on more physical self-consistent reflection models are in progress. We also aim to present more extensive spectral-timing studies based on the simultaneous RXTE observation and on the XMM RGS in the near future.

Acknowledgments

We thank the XMM and RXTE teams for the rapid scheduling of the observations of GX 3+1. The Tübingen group was supported by DLR (grant DLR 50 OR 0702) and EE by Initial Training Network ITN 215212: Black Hole Universe funded by the European Community.

References

  1. Barret, D., & Olive, J.-F. 2002, ApJ, 576, 391 [NASA ADS] [CrossRef] [Google Scholar]
  2. Bhattacharyya, S., & Strohmayer, T. E. 2007, ApJ, 664, L103 [NASA ADS] [CrossRef] [Google Scholar]
  3. Brandt, W. M., & Matt, G. 1994, MNRAS, 268, 1051 [NASA ADS] [CrossRef] [Google Scholar]
  4. Cackett, E. M., Miller, J. M., Bhattacharyya, S., et al. 2008, ApJ, 674, 415 [NASA ADS] [CrossRef] [Google Scholar]
  5. Cackett, E. M., Miller, J. M., Ballantyne, D. R., et al. 2010, ApJ, 720, 205 [NASA ADS] [CrossRef] [Google Scholar]
  6. D’Aì, A., Iaria, R., Di Salvo, T., Matt, G., & Robba, N. R. 2009, ApJ, 693, L1 [NASA ADS] [CrossRef] [Google Scholar]
  7. D’Aì, A., di Salvo, T., Ballantyne, D., et al. 2010, A&A, 516, A36 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  8. Di Salvo, T., D’Aí, A., Iaria, R., et al. 2009, MNRAS, 398, 2022 [NASA ADS] [CrossRef] [Google Scholar]
  9. Fabian, A. C., Rees, M. J., Stella, L., & White, N. E. 1989, MNRAS, 238, 729 [NASA ADS] [CrossRef] [Google Scholar]
  10. Fabian, A. C., Iwasawa, K., Reynolds, C. S., & Young, A. J. 2000, PASP, 112, 1145 [NASA ADS] [CrossRef] [Google Scholar]
  11. Iaria, R., D’Aí, A., di Salvo, T., et al. 2009, A&A, 505, 1143 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  12. Jansen, F., Lumb, D., Altieri, B., et al. 2001, A&A, 365, L1 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  13. Kuulkers, E. 2002, A&A, 383, L5 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  14. Kuulkers, E., & van der Klis, M. 2000, A&A, 356, L45 [NASA ADS] [Google Scholar]
  15. Laurent, P., & Titarchuk, L. 2007, ApJ, 656, 1056 [NASA ADS] [CrossRef] [Google Scholar]
  16. Makishima, K., Mitsuda, K., Inoue, H., et al. 1983, ApJ, 267, 310 [NASA ADS] [CrossRef] [Google Scholar]
  17. Merloni, A., Fabian, A. C., & Ross, R. R. 2000, MNRAS, 313, 193 [NASA ADS] [CrossRef] [Google Scholar]
  18. Miller, J. M., D’Aì, A., Bautz, M. W., et al. 2010, ApJ, 724, 1441 [NASA ADS] [CrossRef] [Google Scholar]
  19. Ng, C., Díaz Trigo, M., Cadolle Bel, M., & Migliari, S. 2010, A&A, 522, A96 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  20. Oosterbroek, T., Barret, D., Guainazzi, M., & Ford, E. C. 2001, A&A, 366, 138 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  21. Pandel, D., Kaaret, P., & Corbel, S. 2008, ApJ, 688, 1288 [NASA ADS] [CrossRef] [Google Scholar]
  22. Piraino, S., Santangelo, A., & Kaaret, P. 2000, A&A, 360, L35 [NASA ADS] [Google Scholar]
  23. Piraino, S., Santangelo, A., di Salvo, T., et al. 2007, A&A, 471, L17 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  24. Ross, R. R., & Fabian, A. C. 2007, MNRAS, 381, 1697 [NASA ADS] [CrossRef] [Google Scholar]
  25. Strüder, L., Briel, U., Dennerl, K., et al. 2001, A&A, 365, L18 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  26. Titarchuk, L. 1994, ApJ, 434, 570 [NASA ADS] [CrossRef] [Google Scholar]
  27. Turner, M. J. L., Abbey, A., Arnaud, M., et al. 2001, A&A, 365, L27 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]

All Tables

Table 1

Best-fit parameters for spectra of GX 3+1 observed with the XMM-Newton EPIC-pn using different models and data extraction.

In the text

All Figures

thumbnail Fig. 1

Residuals in units of σ in the pn energy range used with respect to the best-fitting continuum model, composed of an absorbed (phabs) diskbb and a blackbody component for the spectrum extracted in Region 1 (black), Region 2 (red) and Region 3 (blue).
In the text
thumbnail Fig. 2

Top: unfolded pn spectrum and individual model components for Model 2 using data from Region 1 which includes piled-up events. Bottom: the same for Model 2, with NH fixed as described in the text, for Region 3, which excludes piled-up events. Residuals in units of σ are shown in each panel.
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.