Conclusions

1. Chi-square (d.o.f.) A systematic analysis of broad Fe lines in bright NS LMXBs with XMM-Newton M. Cadolle Bel1, C. Ng1, 2, M. Diaz Trigo3 and S. Migliari2 1 INTEGRAL Science Operation Centre, ESAC, Madrid, SPAIN 2 ESAC, Madrid, SPAIN 3 XMM-Newton Science Operation Centre, ESAC, Madrid, SPAIN Abstract We analysed the XMM-Newton archival observations of 16 neutron star (NS) low-mass X-ray binaries (LMXBs) to study the Fe K emission in these objects. The sample includes all the observations of NS LMXBs performed in the EPIC pn Timing mode and publicly available until September 30, 2009. We performed a detailed data analysis considering pile-up and background effects. The properties of the iron lines differed from previous published analyses due to either incorrect pile-up corrections or distinct continuum parameterization. 80% of the observations for which a spectrum can be extracted showed significant Fe line emission. We found an average line centroid of 6.67 keV and a finite width of 0.33 keV. The equivalent width of the lines varied between 17 and 189 eV, with an average weighted value of 42 eV. For sources where several observations were available, the Fe line parameters changed between observations whenever the continuum changed significantly. The line parameters did not show any correlation with luminosity. Most importantly, a simple Gaussian model could fit the Fe lines of our sample very well: the lines did not show the asymmetric profiles that were interpreted as an indication of relativistic effects in previous studies. Fig. 1: Ratio of the EPIC pn spectrum for Ser X-1 to its best fit continuum model for the spectrum free of pile-up (black). The red points show the ratio of the piled-up spectrum when the full PSF is used for the same observation: a significant hardening is seen above 25 keV, adding an artificial red-wing to the Fe line due to different spectrum curvature. Table 1: Reduced Chi-square (with d.of.) of our spectral fits for the sources for which asymmetric Fe lines, showing relativistic effects, have been reported. Columns (a) and (b) show the reduced Chi-square of the fit after including, respectively, a Gaussian and a LAOR component to the best-fit model. There is no need to claim for relativistic profiles in such NS LMXB. Data analysis and results We performed a systematic analysis of 26 XMM-Newton observations corresponding to all the NS LMXBs observed with the EPIC-pn Timing mode publicly available until September 30th, 2009 (using the SAS version 9.0) for the following sources: 4U 0614+09, Cen X-4, 4U 1543-62, 4U 1608-52, 4U 1636-536, GX 340+0, GX 349+2, 4U 1705-44, GX 9+9, 4U 1728-34, 4U 1753-44, Ser X-1, Aql X-1, IGR J00291+5934, XTE J1807-294 and SAX J1808.4-3658. We took special care in the treatment of the pile-up effects and of the correct analysis of the background (see Fig. 1 for differences when pile-up is carefully taken into account). We then produced light curves and X-ray spectra. We fitted the EPIC pn spectra with a model consisting of a black body and a disc black body modified by photo-electric absorption from neutral material, plus a Gaussian line for the Fe emission; once we added an edge (note that different continua have also been tested). We also tried to fit the Fe lines with the LAOR model. The results of our fits are presented in Table 1 and some examples of Fe line profiles are shown in Fig. 2. The Fe line has a weighted average energy of 6.67+/-0.02 keV, a width of 0.33+/-0.02 keV and an EW of 42+/-3 eV. As shown Table 1, fits with a LAOR or a Gaussian component are equally good. The width and EW have a well defined Gaussian-like distribution around the weighted average value. The outliers of the histograms in Fig. 3 have values with large errors and therefore do not contribute significantly to the weighted average. The energy distribution peaks at ∼6.7 keV, in agreement with the value of the weighted average and consistent with emission from Fe xxv. Clearly, the distribution has values consistent with emission from highly ionized species of iron, from Fe xxii to Fe xxvi, and it never shows a value consistent with neutral iron. There seems to be a correlation between the EW of the Fe line and its error: lines for which the EW can be determined with a small error, corresponding to the high quality spectra, show systematically smaller EWs. Fig. 2: Ratio of the data to the continuum model for some NS LMXBs analysed in our work for which significant Fe K emission is detected. The red line shows the Gaussian fit. Note that 4U 1636-536 has a large error for the line, probably due to bad spectral modelling. Conclusions We established the characteristics of the Fe K line emission in these sources. In 7 observations, we did not detect the source significantly. For the rest we extracted carefully (special attention paid to the effects of pile-up and background subtraction) one spectrum per observation. We detected the line in 80% of our studies. In contrast to previous analyses, we did not need to invoke relativistic effects to explain our results on the line width. The lines are well-fitted with a Gaussian model. Moreover, the profiles shown Fig. 2 do not present any asymmetric shape, as expected if the lines were emitted close to the NS and shaped by relativistic effects. The line profile is instead symmetric as for dipping sources (Diaz Trigo et al. 2006). The major difference between this work and others is the careful treatment of pile-up effects, and continuum modelling. We have shown that effects of pile-up have a strong influence on the breadth of the line (see also Done & Diaz Trigo 2010 for similar conclusion on the BH GX 339-4). Therefore we urge caution in using piled-up data for detailed spectral analysis. For more details, our paper will be on astro-ph soon! Fig. 3: Histogram of the Fe line parameters when fitted with a Gaussian component for all our sample. The white rectangles marked with an arrow correspond to the lines which are detected below a significance of 3 sigmas. The width (middle panel) and the EW (right panel) have a well defined “Gaussian-like” distribution around the weighted average values while the outliers have large error bars associated to their sigma and EW values.