EDP Sciences
Free Access
Issue
A&A
Volume 532, August 2011
Article Number A153
Number of page(s) 8
Section Extragalactic astronomy
DOI https://doi.org/10.1051/0004-6361/201116997
Published online 10 August 2011

© ESO, 2011

1. Introduction

Many observations of the Small Magellanic Cloud (SMC) in the X-ray energy band have led to the discovery of a large number of high mass X-ray binaries (HMXRBs). The number of known HMXRBs in the SMC (about one hundred, Liu et al. 2005) is much larger than expected by scaling the number of these sources seen in the Milky Way according to the mass ratio of the two galaxies (MMW/MSMC  ~  50). This has been interpreted as evidence of a recent episode of star formation in the SMC (Majid et al. 2004; Shtykovskiy & Gilfanov 2007; Antoniou et al. 2010). It is remarkable that the SMC contains only one supergiant system, SMC X-1. All the other HMXRBs in the SMC consist of neutron stars accreting from Be type companions. Be X-ray binaries also constitute the largest class of HMXRBs in our Galaxy, but the relative number of supergiant systems is much higher than in the SMC, since 30 of the 114 Galactic HMXRBs (Liu et al. 2006) are confirmed or suspected supergiant systems.

Many new Galactic HMXRBs have been discovered in the past few years with the INTEGRAL satellite, thanks to high sensitivity in the hard X-ray range, coupled with an extensive monitoring of the Galactic plane. These observations led to the recognition of the new class of supergiant fast X-ray transients (SFXT) and to the discovery of persistent supergiant systems characterized by a high absorption (equivalent column density above a few 1023 cm-2), which escaped an earlier discovery because the high absorption severely suppresses their flux below 10 keV (see, e.g., Sidoli 2010, for a recent review).

To search for highly absorbed HMXRBs in the SMC, we carried out a dedicated analysis of the XMM-Newton data collected during the large program for the SMC survey performed by Haberl & Pietsch (2008) and other observations of SMC targets. In Sect. 2, we describe the X-ray observations and data processing and in Sect. 3 we describe the source detection and selection criteria that led us to identify a sample of 30 highly absorbed sources. Five of them are confirmed HMXRBs, including four that were already known as X-ray binaries (Sect. 4). In Sect. 5, we present optical spectroscopy of the new source, demonstrating that it is a Be system in the SMC. We finally discuss our results in Sect. 6.

2. X-ray observations and data processing

We analyzed the 62 XMM-Newton pointings of the SMC performed from 2001 May 31 to 2010 March 16 listed in Table 1. The three EPIC focal plane cameras (Turner et al. 2001; Strüder et al. 2001) were operated in standard full frame mode.

The data of each camera were processed independently using the standard XMM-Newton Science Analysis Software (SAS, v.10.0.0). We first used the tasks emproc and epproc to obtain the MOS and pn event files of each pointing. We then filtered out time intervals affected by high instrumental background induced by flares of soft protons (with energies below a few hundred keV). To avoid contributions from genuine X-ray source variability, this was done by examining light curves binned at 100 s for events with energies above 10 keV and PATTERN  ≤  4 or 0 for, respectively, the MOS peripheral CCDs and the whole pn camera; the good time intervals (GTI) were selected by adopting different count-rate thresholds for the different pointings and instruments. By selecting only events within GTI, we finally obtained three “clean” event lists for each observation (i.e. two for the MOS and one for the pn cameras), whose exposure times are reported in Table 1. The total net exposure times are respectively 1572 and 1598 ks for the MOS cameras, and 1361 ks for the pn one.

3. Source detection and selection criteria

We optimized our source detection strategy to select highly absorbed X-ray sources. To this aim, we simulated the expected EPIC count distributions for absorbed power-law spectra, with photon-index Γ in the range 1–2. We used two absorption components: the first one, with column density fixed at 6 × 1020 cm-2 and elemental abundances from Wilms et al. (2000), accounts for the foreground absorption in our Galaxy; for the second one, which accounts for the absorption in the SMC and is intrinsic to the source, we considered NH values in the range 1022–1024 cm-2, and metal abundances of 0.2, as is typical of the SMC (Russell & Dopita 1992). We defined the hardness ratio HR = (H − S)/(H + S)), based on the count rates in the soft (S, 1–3 keV) and hard (H, 3–10 keV) energy ranges. These simulations showed that, for Γ = 2 and NH > 3 × 1023 cm-2, the fraction of the detected source counts with energies above 3 keV is at least 70%, and that this fraction is even higher for lower values of Γ. In Fig. 1, we report the estimated HR values as a function of NH, for both the pn and MOS cameras and different values of Γ. On the basis of these simulations, we adopted a HR threshold value of 0.7, in order to select sources with NH > 3 × 1023 cm-2.

thumbnail Fig. 1

Expected hardness ratio (of the count rates in the energy ranges 1–3 to 3–10 keV) as a function of the column density. Solid and dashed lines refer to the pn and the MOS cameras, respectively. Red, green, and blue lines refer to power-law spectra with photon indices of 2, 1.5, and 1, respectively. The horizontal line at HR = 0.7 represents the threshold value used to select highly absorbed sources.

Open with DEXTER

For each observation in Table 1, we used the cleaned event file to produce merged images from the three cameras in the 1–3 and 3–10 keV energy ranges, according to the procedure described in detail in Baldi et al. (2002). For the 3–10 keV pn images, we excluded the energy range 7.8–8.2 keV, in order to reduce the instrumental background caused by the Cu line. For each image, we also produced the corresponding merged exposure map, which accounts for mirror vignetting and effective field of view. The total count-rate-to-flux conversion factors were obtained as a mean of the MOS1/MOS2/pn factors, weighted on the individual exposures of the three cameras. The source detection was based on a maximum likelihood technique, as described in detail in Novara et al. (2006), and used background maps produced with the correction algorithm described in Baldi et al. (2002). For each energy band, we only selected sources with a final detection likelihood -ln P  >  8.5 (where P is the probability of a false detection due to a Poissonian random fluctuation of the background). This likelihood threshold corresponds to a  ~3σ detection. After manually removing a few spurious detections caused, for instance, by events falling close to the CCD edges, we obtained a final master list containing  ≃ 1500 sources. For each source, the master list provides various parameters including the detector and sky coordinates, the effective exposure time, the total counts, count-rate and errors in the different energy ranges, and the detection likelihood.

4. Results

thumbnail Fig. 2

Hardness-ratio distribution of the detected X-ray sources.

Open with DEXTER

In Fig. 2, we report the HR distribution of all the detected sources. This figure illustrates that most sources have HR  <  0, with a peak around –0.5, hence they are characterized by non-absorbed soft spectra. On the other hand, only very few sources have high HR values. In our final source list, there are 30 objects with HR  >  0.7, which we consider as highly absorbed sources with NH > 3 × 1023 cm-2; given the HR distribution, we expect that, even if we take into account the HR errors, only a very few additional sources would have HR  >  0.7. The sky coordinates of these 30 objects, corrected for the satellite pointing uncertainty, are listed in Table 2. The astrometric correction of the X-ray images was done by cross-matching the brightest sources of each field with the Magellanic Clouds Photometric Survey optical catalogue (MCPS, Zaritsky et al. 2002), and selecting the X-ray sources with a single, clearly evident optical counterpart within a 3′′ error radius. We then used the IRAF task geomap to compute the linear transformation between the X-ray and optical coordinates of these reference sources.

Table 2

Main characteristics of the highly absorbed X-ray sources.

The calculated X-ray-to-optical coordinate transformations have rms fit residuals between 043 and 185 at 1σ c.l., with an average value of 11; they correspond to the (radial) systematic astrometric errors in the corrected observations. These transformations were then used by the IRAF task geoxytran to obtain the corrected X-ray coordinates given in Table 2. The total uncertainty on the X-ray position of each source is given by the sum in quadrature of the calculated systematic astrometric error in the observation and the measured statistical error of the source itself.

We also list in Table 2 the hardness ratios, count rates, and fluxes of the sources. The fluxes refer to the 3–10 keV energy range, and have been computed assuming a power-law with photon index Γ = 1.5 and NH = 3 × 1023 cm-2. The brightest source has a absorbed flux of 9 × 10-12 erg cm-2 s-1, while the faintest sources in our sample have fluxes of  ~4 × 10-14 erg cm-2 s-1 corresponding to  ~ 2 × 1034 erg s-1, assuming a SMC distance of 60 kpc (Hilditch et al. 2005).

Some of our selected sources were already reported in other studies of the SMC, in these cases we give in the table their original name and the relevant references. Four sources, including three pulsars, were already known as HMXRBs: SXP25.5, SMC34, SXP565, and SXP293. We searched for possible counterparts of our highly absorbed sources using the MCPS catalogue, which is based on observations performed in the U, B, V, and I filters between November 1996 and December 1999 (Zaritsky et al. 2002), and the InfraRed Survey Facility (IRSF) Magellanic Cloud catalogue (Kato et al. 2007) for the NIR (J, H, and K filters).

For each source, we report in Table 3 the optical and NIR properties of the brightest object in the K band present within a distance equal to two times the 1σ position error (with only five exceptions, marked in the table, the brightest object in K is also the brightest in V). For five sources, no catalogued optical or NIR counterparts were found to limits as faint as of V ~ 23 mag and K ~ 17 mag.

A plot of the NIR versus optical colors shows that the four already known HMXRBs of our sample are located in the region corresponding to J − K < 0.5 mag and B − V < 0.5 mag (see Fig. 3). As discussed in Sect. 5, the other source falling in this region (source #13) is a new Be binary system in the SMC. Three of these sources are well above the position of the main-sequence stars (whose colors are taken from Johnson 1966): if they are affected by an IR excess (which could be due to the decretion disk of the Be star), their optical/NIR colors are consistent with early-type stars with moderate optical reddenings (AV ~ 2–3 mag). On the other hand, HMXRBs in which the optical companion is also affected by a high local absorption would show much redder colors, as seen in several of the new HMXRBs discovered by INTEGRAL in our Galaxy. Therefore, we cannot exclude that some of the other sources plotted in Fig. 3, or sources for which no optical and NIR counterparts have been detected, are also highly absorbed HMXRBs. The five sources at the top of the plot are far from the main-sequence stars, therefore it is difficult to suggest a reliable classification of them; this is particularly true for the three sources shown as open circles, for which a different counterpart could also be suggested. These five sources could be early-type stars only if they are characterized by both a large IR excess and a high reddening. On the other hand, since for AGNs J − K > 0.5 (Kouzuma & Yamaoka 2010) and  −0.4 < B − V < 0.7 (Hatziminaoglou et al. 2002), these sources could be more probably classified as reddened AGNs.

Table 3

Candidate optical-IR counterparts of the highly absorbed X-ray sources.

thumbnail Fig. 3

Color indices of the candidate counterparts of the highly absorbed X-ray sources of our sample. For the sources reported with a filled circle, the brightest optical and infrared candidate counterparts are coincident, while the sources reported as an open circle correspond to those with the “a” flag in Table 3: in this case, there is also a possible alternative counterpart, with a slightly brighter V magnitude, to the selected infrared counterpart. We also indicate the main-sequence stars and a bar which shows the reddening corresponding to AV = 1 mag.

Open with DEXTER

To search for possible periodicities in any optical counterparts close to the XMM-Newton sources, data from OGLE III were acquired for all objects within 4′′ of the 30 XMM-Newton position. We applied a Lomb-Scargle analysis to a total of 129 lightcurves, searching for periods in the range 2–200 days. Periods shorter than 2 days were avoided because they are getting close to the average daily sampling of the OGLE data, and periods longer than 200 days approach the annual sampling patterns. In each case, the data were first de-trended using a polynomial of order three before being searched. In some cases, there were many optical counterparts; conversely, other sources had no optical objects within the search zone.

Significant periodicities were found for four sources (three of which were already known) and tentative periods for an additional three objects (see Table 4). That three previously known periods were “re-discovered” in this search reassures us that the techniques used are effective in identifying possible periods. The one new significant detection (source number 19) shows a modulation that is extremely sinusoidal in nature when the data are folded at the period of 199.6 days. This modulation shape is not normal for a long-period binary modulation, which tends to reveal short outbursts coincident with the periastron passage of the neutron star (see, for example, Coe & Edge 2004). Such smooth sinusoidal modulations tend to be associated with pulsations in stars. A likely possibility is that it could be an Non-Radial Pulsations (NRP) in a B-type star with a period very close to the one day sampling, with the result that the signal at 199.6 days is the beat frequency of the true pulse period and the sampling period.

Table 4

List of OGLE III objects within 4″ of an XMM-Newton position that show evidence of a periodic signal.

5. XMMU J005605.8–720012: a new Be HMXRB in the SMC

Source #13 was detected with a total of 68 counts in the hard energy range 3–10 keV and a HR value of 0.71, just above our threshold. It has a relatively bright (V = 16.71 mag) optical counterpart, with color B − V =  −0.12 mag, and with a quality flag in the MCPS indicating a successful fit with a stellar atmosphere model. It has an unabsorbed luminosity, in the energy range 3–10 keV, of 4.4 × 1034 erg s-1.

Optical spectral observations of source #13 were obtained in September 2010, using the 1.9-m telescope and the Cassegrain spectrograph at the South African Astronomical Observatory (SAAO) in Sutherland. We used grating number 7 (300 lines per mm) to obtain spectra between 3700 Å and 7700 Å at a resolution of 5 Å. For these, the slit size was 15  ×  15. Exposure times were limited to 900 s with a positional accuracy of  < 1′′. Data reduction included bias subtraction and flat-field correction using the IRAF software package. Extraction (task extractor), including background sky subtraction, of data allowed the creation of one-dimensional spectra, wavelength calibrated using standard lines from a Cu-Ar arc. Observing conditions were not photometric, seeing was limited to 2′′ at best but varied throughout the evening. We show in Fig. 4 the resulting spectrum: its moderate signal-to-noise ratio does not allow us to make a spectral classification, but a significant Hα emission line, with equivalent width EW = –37.44 Å, is clearly visible.

thumbnail Fig. 4

SAAO optical spectrum of source #13; note the presence of a strong Hα emission line.

Open with DEXTER

Despite no periodicities being found in the OGLE light curve of this star, these findings indicate that source #13 is a Be HMXRB. Since sources of this class typically have a transient nature, we searched for hard X-ray outbursts in all the public data obtained with the INTEGRAL satellite (Winkler et al. 2003), that is all pointings obtained with the IBIS instrument (Ubertini et al. 2003) in which the source position was within 12 degrees from the center of the field of view. This led to a total of 619 pointings, with different exposure times from about 1800 s to about 3600 s, spanning from July 2003 to June 2009. We used version 9.0 of the Off-line Scientific Analysis software to analyse the data. For each pointing we extracted an image in the 17–40 keV energy band, but in none of them was the source significantly detected. The typical flux upper limit of the individual images is on the order of  ~20 mCrab. We also produced a deep image by summing all the individal pointings, reaching a total exposure of about 2 Ms. The source was not detected, with a 5σ upper limit of about 1 mCrab in the 17–40 keV range, which is well above the extrapolation of the flux seen by EPIC.

6. Discussion

We have detected 30 sources with hardness ratios indicating a column density NH > 3 × 1023 cm-2. Five of these sources are HMXRBs, of which one was discovered in our search and spectroscopically identified as a new Be binary system. Most of the remaining 25 sources have very faint (V > 18) or no optical counterparts, with colors not compatible with reddened early type-stars (unless a significant IR excess is present). Therefore, they can hardly be considered as candidate absorbed HMXRBs.

Spectral information for a few of the brightest sources of our sample was previously reported in other works. The XMM-Newton spectrum of the 25 s pulsar XMMU J004814.1–731003 was closely fit with a power law of photon index 1.33 ± 0.27 and NH = 5 × 1022 cm-2, confirming our conclusion that it is a highly absorbed source. Our brightest source (#18) is the 293 s pulsar RX J0058.2–7231, for which Haberl et al. (2008a) reported an absorption NH = (1.35 ± 0.32) × 1022 cm-2, considerably smaller than the value we inferred from the hardness ratio. This might be due to the results of Haberl et al. (2008a) being obtained from the June 2007 XMM-Newton observation, while our results for this source are based on the SMC Survey pointing carried out in October 2009, when the source was about a factor of ten brighter. HMXRBs can display large variations in their intrinsic absorption, as shown for example in the case of the long period (1323 s) SMC pulsar RXJ0103.6–7201 (Eger & Haberl 2008).

We do not expect of course all the sources in our sample to be HMXRBs in the SMC. A large fraction of them will turn out to be background-absorbed AGNs. On the other, hand our sample should not contain foreground stars belonging to our Galaxy, which would show softer and less absorbed X-ray spectra.

A rough estimate of the number of absorbed AGNs in our sample can be derived using published number flux relations for extragalactic sources. For example, based on the Log   N − LogS relations for AGNs with NH = 3 × 1023 cm-2 by Gilli et al. (2007), we obtain a surface density of  ~4 AGNs per square degree at a flux limit of 4 × 10-14 erg cm-2 s-1. Our survey covered a sky area of  ~5 square degrees, although with a non-uniform sensitivity, because of telescope vignetting and different exposure times. Owing to the uncertainties in the LogN − LogS relations and the small number of objects, this prediction is consistent with the number of unidentified sources in our sample.

7. Conclusions

Motivated by the presence of several highly absorbed HMXRBs in our Galaxy, we have carried out a search for similar systems in the SMC  exploiting the nearly complete coverage of this galaxy obtained with XMM-Newton. Our selection criteria, corresponding to an absorption threshold of  ~3 × 1023 cm-2, were met by four known HMXRBs and led to the discovery of a new highly absorbed Be binary in the SMC. We also selected other 25 additional sources, among which other highly absorbed HMXRBs might be present. Since we also expect a good fraction of these sources to be AGNs, we can say that the SMC does not contain much more than  ~10 persistent HMXRBs with intrinsic NH > 3 × 1023 cm-2. We cannot exclude the presence of a much larger population of intrinsically absorbed binaries of transient nature, which could not be detected in our survey if their quiescent luminosity is below a few 1034 erg s-1.

Acknowledgments

We wish to thank Lara Sidoli for the useful discussion. This work is based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESA Member States and NASA. The XMM-Newton data analysis is supported by the Italian Space Agency (ASI) (ASI/INAF contract I/032/10/0). We were granted observation time at the South African Astronomical Observatory (SAAO) and wish to thank them for their kind help and accommodations. Travel to the SAAO was funded by Australian Government AINSTO AMNRF grant number 10/11-O-06. Partly based on observations with INTEGRAL, an ESA project with instruments and science data centre funded by ESA member states (especially the PI countries: Denmark, France, Germany, Italy, Spain, and Switzerland), Czech Republic and Poland, and with the participation of Russia and the USA. IRAF is the Image Reduction and Analysis Facility, a general purpose software system for the reduction and analysis of astronomical data. IRAF is written and supported by the IRAF programming group at the National Optical Astronomy Observatories (NOAO) in Tucson, Arizona. NOAO is operated by the Association of Universities for Research in Astronomy (AURA), Inc. under cooperative agreement with the National Science Foundation. Extractor is part of an IRAF package (PNDR) developed to assist with the efficient reduction of long-slit spectra of emission line objects. It was developed as part of the Macquarie/AAO/Strasbourg Halpha (MASH) Planetary Nebulae Catalogue.

References

Online material

Table 1

Log of the XMM-Newton observations of 62 SMC fields with the corresponding net good time interval (GTI) for the two MOS and the pn cameras.

All Tables

Table 2

Main characteristics of the highly absorbed X-ray sources.

Table 3

Candidate optical-IR counterparts of the highly absorbed X-ray sources.

Table 4

List of OGLE III objects within 4″ of an XMM-Newton position that show evidence of a periodic signal.

Table 1

Log of the XMM-Newton observations of 62 SMC fields with the corresponding net good time interval (GTI) for the two MOS and the pn cameras.

All Figures

thumbnail Fig. 1

Expected hardness ratio (of the count rates in the energy ranges 1–3 to 3–10 keV) as a function of the column density. Solid and dashed lines refer to the pn and the MOS cameras, respectively. Red, green, and blue lines refer to power-law spectra with photon indices of 2, 1.5, and 1, respectively. The horizontal line at HR = 0.7 represents the threshold value used to select highly absorbed sources.

Open with DEXTER
In the text
thumbnail Fig. 2

Hardness-ratio distribution of the detected X-ray sources.

Open with DEXTER
In the text
thumbnail Fig. 3

Color indices of the candidate counterparts of the highly absorbed X-ray sources of our sample. For the sources reported with a filled circle, the brightest optical and infrared candidate counterparts are coincident, while the sources reported as an open circle correspond to those with the “a” flag in Table 3: in this case, there is also a possible alternative counterpart, with a slightly brighter V magnitude, to the selected infrared counterpart. We also indicate the main-sequence stars and a bar which shows the reddening corresponding to AV = 1 mag.

Open with DEXTER
In the text
thumbnail Fig. 4

SAAO optical spectrum of source #13; note the presence of a strong Hα emission line.

Open with DEXTER
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.