A&A 467, 785-905 (2007)
DOI: 10.1051/0004-6361:20077084

New periodic variable stars coincident with ROSAT sources discovered using SuperWASP[*]

A. J. Norton1 - P. J. Wheatley2 - R. G. West3 - C. A. Haswell1 - R. A. Street4 - A. Collier Cameron5 - D. J. Christian4 - W. I. Clarkson6 - B. Enoch1 - M. Gallaway1 - C. Hellier7 - K. Horne5 - J. Irwin8 - S. R. Kane9 - T. A. Lister7 - J. P. Nicholas2 - N. Parley10 - D. Pollacco4 - R. Ryans4 - I. Skillen11 - D. M. Wilson7

1 - Department of Physics and Astronomy, The Open University, Walton Hall, Milton Keynes MK7 6AA, UK
2 - Department of Physics, University of Warwick, Coventry CV4 7AL, UK
3 - Department of Physics and Astronomy, University of Leicester, Leicester LE1 7RH, UK
4 - Astrophysics Research Centre, Main Physics Building, School of Mathematics & Physics, Queen's University, University Road, Belfast BT7 1NN, UK
5 - School of Physics and Astronomy, University of St. Andrews, North Haugh, St. Andrews, Fife KY16 9SS, UK
6 - Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA
7 - Astrophysics Group, School of Chemistry & Physics, Keele University, Staffordshire ST5 5BG, UK
8 - Institute of Astronomy, University of Cambridge, Madingly Road, Cambridge CB3 0HA, UK
9 - Department of Physics, University of Florida, Gainesville, FL 32611-8440, USA
10 - Planetary & Space Sciences Research Institute, The Open University, Walton Hall, Milton Keynes MK7 6AA, UK
11 - Isaac Newton Group of Telescopes, Apartado de Correos 321, 38700 Santa Cruz de la Palma, Tenerife, Spain

Received 11 January 2007 / Accepted 21 February 2007

We present optical lightcurves of 428 periodic variable stars coincident with ROSAT X-ray sources, detected using the first run of the SuperWASP photometric survey. Only 68 of these were previously recognised as periodic variables. A further 30 of these objects are previously known pre-main sequence stars, for which we detect a modulation period for the first time. Amongst the newly identified periodic variables, many appear to be close eclipsing binaries, their X-ray emission is presumably the result of RS CVn type behaviour. Others are probably BY Dra stars, pre-main sequence stars and other rapid rotators displaying enhanced coronal activity. A number of previously catalogued pulsating variables (RR Lyr stars and Cepheids) coincident with X-ray sources are also seen, but we show that these are likely to be misclassifications. We identify four objects which are probable low mass eclipsing binary stars, based on their very red colour and light curve morphology.

Key words: stars: binaries: eclipsing - stars: rotation - stars: variables: general

1 Introduction

The SuperWASP project (Pollacco et al. 2006) is a wide field photometric survey designed to search for transiting exoplanets and other signatures of variability on timescales from minutes to months. In its first run during 2004, SuperWASP-N on La Palma was operated between May and September with five cameras, each of which has a $7.8^{\circ} \times 7.8^{\circ}$ field of view. The fields surveyed in 2004 comprise a strip of sky centred at declination + $28^{\circ}$ and extending to all right ascensions (excluding the galactic plane). Coverage is not uniform though, with some regions of sky better sampled than others. The resulting area covered was $\sim$10 000 square degrees and comprised over 300 000 raw images. Photometry for all objects detected was extracted using a 2.5 pixel aperture ( $34^{\prime \prime}$ radius). As a result, the project produced unfiltered (white light) lightcurves of over 6.7 million stars in the magnitude range $V\sim 8$-15, totalling 12.9 billion data points. Because of the wide-field nature of the images, systematic errors in the lightcurves are present, but are largely removed using the SysREM algorithm from Tamuz et al. (2005). Plots showing the RMS precision of our data as a function of SuperWASP V magnitude, both before and after application of the SysREM algorithm, are shown in Collier Cameron et al. (2006a).

The 2004 survey allowed the identification of $\sim$100 transiting exoplanet candidates, which we have reported in a series of papers (Christian et al. 2006; Clarkson et al. 2007; Kane et al., in preparation; Lister et al. 2007; Street et al. 2007; West et al., in preparation). The first confirmed planets resulting from this are reported by Collier Cameron et al. (2006b). These data also uncovered huge numbers of variable stars, many of which are previously unidentified. In this paper we discuss a small subset of these, namely those that are coincident with ROSAT X-ray sources. Our reasons for concentrating on the subset of variable sources which are X-ray sources are twofold. First, variable objects coincident with X-ray sources are likely to yield significant numbers of astrophysically important and interesting objects. Secondly, this provides a manageable number of objects with which to demonstrate the efficiency of SuperWASP for detecting variable objects other than transiting exoplanets. SuperWASP provides a significant increase in time coverage as compared to many single object studies and smaller photometric campaigns.

2 Period searching

The positions of objects in the SuperWASP archive are derived from the USNO-B1 catalogue (Monet et al. 2003). These positions were cross correlated against the ROSAT sky survey (1RXS; Voges et al. 1999, 2000) and pointed phase (2RXP; ROSAT 2000) catalogues, taking the uncertainty in position of each ROSAT source as its 3$\sigma$ error radius, or 10 $^{\prime \prime}$, whichever was larger. This resulted in 4562 matches between the positions of SuperWASP objects from our 2004 Northern hemisphere run and the positions of ROSAT sources. From this set of cross identifications, 3558 SuperWASP lightcurves have more than 100 data points, and so were deemed to be suitable for period searching. A purpose written period search code was then run on these to identify periodic variables, coincident with X-ray sources.

The period search comprised two techniques. A CLEANed power spectrum was calculated, using the variable gain implementation of H. Lehto, and the strongest peaks within it identified. A period folding analysis was also performed, searching over periods from 20 min to half the lightcurve length in each case. Binning the data into 20 phase bins at each trial period, we calculated the reduced chi-squared of the folded lightcurve with respect to its mean flux and also the sum of the reduced chi-squareds of the data within each phase bin with respect to the mean flux in that bin. The most likely periods were then taken as those which maximised the difference between these two chi-squared values. This jointly minimised the dispersion within each phase bin and maximised the dispersion between the phase bins, in order to identify likely periods. Only those periods found in common between the CLEANed power spectrum and period folding technique (within a tolerance of 1%) were recorded. It was noted that a few systematic effects remain in the SuperWASP lightcurves, sometimes resulting in spurious periods being identified due to night-to-night variations. As a result, we ignored periods within 1% of one day and fractions thereof (i.e. 1d/2, 1d/3, 1d/4, etc.). We also rejected around 20 objects where several sources within a few arcminutes of each other each displayed similar lightcurves with similar long term (tens of days) periodicities. These are believed to be artefacts due to remaining systematic errors in the extracted data. After this there remained 516 SuperWASP sources in this set for which periodic variability was identified, with periods ranging from less than 3h to more than 50d.

88 of these variable sources were noted to be duplicates of other stars in the list, recorded with slightly different positions (typically within 10 arcseconds) due to the large pixel size of the SuperWASP cameras. These arise where there are multiple USNO-B1 objects within a few arcseconds of each other, and the SuperWASP data resulting from different images are variously assigned to one of this small set of objects. Encouragingly, we detected the same period in the multiple SuperWASP lightcurves in each case. Most of these duplicates consisted of just two SuperWASP lightcurves corresponding to the same object, but in a couple of cases there were as many as five SuperWASP lightcurves corresponding to the same object, each with slightly different positions but showing the same periodicity. After removing these duplicates, there remained 428 unique objects showing periodic variability in their SuperWASP lightcurves and coincident with ROSAT X-ray sources.

Table 1: SuperWASP objects showing periodic variability, coincident with ROSAT sources and previously classified as periodic variable stars.

3 Results

The positions of the 428 periodic SuperWASP objects were cross-correlated against the SIMBAD database. In order to account for the $34^{\prime \prime}$ radius photometry aperture used for our data, we searched for all objects in SIMBAD within twice this distance of the nominal SuperWASP position, and identified the most likely source of the variable signal in each case. As a result we identified 68 sources that have been previously recorded as periodic variable stars, of which 66 have periods given in the literature. These objects comprise 47 listed in the General Catalogue of Variable Stars, 17 discovered by the ROTSE (Robotic Optical Transient Search Experiment) survey (Akerlof et al. 2000), 2 discovered by the SAVS (Semi-Automatic Variability Search) survey (Maciejewski et al. 2004) and 2 objects (HD170451 and SAO46441) recently identified as W UMa type eclipsing binaries but not yet assigned GCVS designations. This set of 68 objects also includes two identified eclipsing binaries (KW Com and V1011 Her) whose periods appear not to have been previously published.

The details of these 68 objects are listed in Table 1; their folded SuperWASP lightcurves and power spectra are shown in Fig. 1. Phase zero for each of the folded lightcurves is set at 2004 January 1st 00:00UT (i.e. HJD 2453005.5). The columns of Table 1 are as follows: 1. the number of SuperWASP objects which are duplicates, recorded with slightly different positions but displaying the same period, where this number is greater than one; 2. the SuperWASP identifier in the form "1SWASP Jhhmmss.ss+ddmmss.s''; the position encoded in this identifier will be identical to the position of the corresponding object in the USNO B1 catalogue; 3. the ROSAT identifier either from the 1RXS or 2RXP catalogues. 4; the period in days as derived from the SuperWASP lightcurve; 5. the mean SuperWASP magnitude, defined as - $2.5\log_{10}(F/10^6)$ where F is the mean SuperWASP flux in microVegas; it is a pseudo-V magnitude which is comparable to the Tycho V magnitude; 6. and 7. the B1 and R1 magnitudes respectively from the USNO catalogue (Monet et al. 2003); 8. a previously recorded name of the object; 9. the astronomical classification of the object; 10. the previously recorded period of the object; 11. a reference to the previous period determination.

The 68 previously classified periodic variables consist of 13 pre-main sequence stars, 10 Algol type (EA) eclipsing binaries (4 of which are also RS CVn stars), 5 $\beta $ Lyrae type (EB) eclipsing binaries, 10 W UMa type (EW) eclipsing binaries, 6 BY Dra systems (5 of which also show UV Cet type behaviour, one of which has a white dwarf companion), 5 RS CVn systems (4 of which are also Algol type eclipsing binaries), 2 RR Lyrae stars, 15 Cepheid variables (13 of which were classified by the ROTSE project and one by the SAVS survey), one semi-regular pulsator, 3 cataclysmic variable stars, one supersoft source and one low mass X-ray binary.

The periods we have determined from our SuperWASP data for these known objects are generally in good agreement with previously published values (see Table 1). Where the periods differ significantly, we are confident that our determination is the more reliable measurement, owing to the better sampling of our data. In a couple of cases, for instance, we detect clear periodicities which are close to half that of objects discovered by ROTSE and claimed to be Cepheid variable stars. In another case, the cataclysmic variable PX And, the period we detect is the disc precession period rather than the binary orbital period of 0.14635d.

The remaining 360 objects comprise newly identified periodic variable stars which are also X-ray sources. Their details are listed in Table 2 and their SuperWASP lightcurves and power spectra are also shown in Fig. 1. The columns of Table 2 are essentially the same as those in Table 1, without the previously determined periods and references, but with the addition of the spectral type where this is recorded in SIMBAD. Many of the objects in Table 2 are anonymous stars, with only a Hubble Space Telescope Guide Star Catalog or Tycho Catalog designation. Where objects have a designation other than one of these catalogue numbers, that is listed in Table 2. The period distribution of all 428 objects is shown in Fig. 2.

In two cases (1SWASP J170033.82+200134.1 and 1SWASP J222229.09+281439.1) the SuperWASP lightcurves are double valued at all phases. In each case the objects are double stars (see Table 2) and the anomalous lightcurves are undoubtedly the result of one of the two stars (the non-variable one) sometimes appearing within the photometry aperture and sometimes not, so offsetting the mean brightness for a subset of the datapoints.

We also note that some folded lightcurves (e.g. that of 1SWASP J141630.88+265525.1) show regular "chunks'' of data (7 in this case) such that the measured period is close to the same integer number of days (i.e. the period is 6.9998d in the case of this object). However, although the phase coverage is uneven, these lightcurves cover many cycles of variation (121 days duration in the case of this object) and the period is reliable. The pattern seen is a result of the modulation period being close to an integer number of days and the sampling of the object repeating at the same time of night over many weeks. The longest period accepted for any object is less than half the data length in each case.

4 Discussion

4.1 Positional coincidence

As noted earlier, since the SuperWASP pixel size is relatively large (13.7 $^{\prime \prime}$ pixel-1), the 2.5 pixel extraction aperture for photometry corresponds to 34 $^{\prime \prime}$ in radius. Given that the ROSAT sources can have positional uncertainties of up to tens of arcseconds, there is clearly the likelihood of chance positional coincidences between SuperWASP objects and catalogued X-ray sources.

The sky area covered by our 2004 Northern hemisphere observations is about 104 square degrees, and we have lightcurves of 6 713 217 objects from this SuperWASP run, of which 5 271 091 have more than 100 data points. The mean separation between nearest neighbours is therefore about 140 $^{\prime \prime}$ for the total set of SuperWASP objects. There are 14 616 ROSAT sources that fall within the sky area covered by our 2004 Northern hemisphere SuperWASP run, including 3 826 from 1RXS and 10 790 from 2RXP. (Many of these will, however, be duplicates between the two catalogues.) The mean error radius of the ROSAT positions is 18 $^{\prime \prime}$. Hence, there is a 5.3% chance of a given ROSAT source coinciding with one of our SuperWASP sources, purely at random. The fact that we find 4 562 matches, rather than the $\sim$770 matches that would result from chance alone, suggests that at least (4562-770)/4562 =  83% of the positional coincidences are genuinely the result of X-ray emission from SuperWASP objects. At least 300 of the newly identified periodic variable stars identified here are therefore likely to be actual X-ray emitters.

Table 2: SuperWASP objects showing periodic variability, coincident with ROSAT sources, newly identified as periodic variable stars.

4.2 Previously known periodic variable stars

The list of known variable stars in the sample considered here contains five accreting binary stars - the X-ray binary Her X-1, the supersoft X-ray source QR And, and the three cataclysmic variables PX And, V795 Her and DQ Her. These latter three are all magnetic CVs to some extent, and so have enhanced X-ray emission as a result.

Apart from this, the X-ray emission seen in the other objects with previously known periods is generally a result of stellar coronal activity (e.g. Rosner, et al. 1985; Hartmann & Noyes 1987). The key feature linking their X-ray emission is rapid rotation causing enhanced magnetic fields through the dynamo mechanism. This is particularly evident in the 13 pre-main sequence stars which we detect whose periods have been previously determined (e.g. from the COYOTE campaigns of Bouvier et al. 1993, 1995, 1997). These are young, rapidly rotating stars, and all those detected here are in the Taurus-Auriga star forming complex. We note that the observed photometric modulation periods of pre-main sequence stars are due to the presence of star spots, and that these can therefore change with time as spots appear and disappear at different latitudes. This will give rise to different modulation periods at different epochs if the stars rotate differentially (Neuhauser et al. 1995).

Another manifestation of coronal X-ray emission due to rapid rotation is in RS CVn stars. These are detached binary systems in which the rotation of the two components is locked to the orbital period of typically just a few days (Hall 1976; Strassmeier et al. 1993). One of the stars is usually a K sub-giant, and it is this star with its deep convection zone which develops a strong magnetic field and enhanced stellar coronal activity. Through optical selection effects, many RS CVn systems are detected as eclipsing binaries, and indeed we see that four of the five known RS CVn systems in this sample are of the Algol type, i.e. detached eclipsing binaries.

BY Dra stars are also well-represented amongst the systems with previously known periods. These are dwarf K or M stars showing emission lines and believed to be rapid rotators (Alekseev 2000; Strassmeier et al. 1993). Many of them show flare star behaviour and so are classed as UV Cet type stars too (Gershberg et al. 1999). The periods we see in the previously known systems of these types are almost certainly the rotation periods of the star. One of the previously catalogued BY Dra stars is V1092 Tau, which is a wide binary system containing a rapidly rotating K2V star and a hot white dwarf (Jeffries, et al. 1996).

Finally, we apparently see a surprising number of pulsating variable stars as X-ray sources. This includes both RR Lyr type, which are A or F giant stars with periods typically less than a day (Smith 1995), and $\delta $ Cep type (Cepheids) which are super-giants with periods of around 1 to 100 days (e.g. Turner & Burke 2002). Most of the apparent Cepheids recovered here were discovered by the ROTSE project. This detected 201 Cepheids in 2000 square degrees of sky coverage (Akerlof et al. 2000), giving a mean separation between them of 3.15$^{\circ}$. There is therefore a chance of less than 0.001% that one of these ROTSE Cepheids would coincide with one of the ROSAT sources falling within the SuperWASP survey area. The fact that we find 13 ROTSE Cepheids coincident with ROSAT sources suggests that these objects are indeed X-ray emitters. However, whether they are actually Cepheid variables is not so certain.

\par\includegraphics[width=9cm,clip]{7084_fig2.eps}\end{figure} Figure 2: The period distribution of our sample of 428 periodic variable stars coincident with ROSAT sources.
Open with DEXTER

Akerlof et al. (2000) classified ROTSE objects as Cepheids on the basis of having sinusoidal lightcurves and periods in the range 1 to 50 days. However, there is no real evidence that these are likely to be Cepheids rather than other variables such as RS CVn systems. Objects such as ROTSE1 J172339.92+352759.3 and ROTSE1 J184633.30+485435.3 which we recover here are 11th magnitude stars with periods of $\sim$24 d and $\sim$5 d respectively. If they really were Cepheids, the period-luminosity relationship (Feast & Catchpole 1997) would place them at distances of 24.5 kpc and 11 kpc. Cepheids are Population I objects and therefore mostly lie in the Galactic disk. Since none of the objects we have considered here are in the Galactic plane, these distances are unrealistic. It is likely that none of the supposed ROTSE Cepheids we have detected as coincident with ROSAT sources are in fact correctly classified. The possibility of X-ray emission from pulsating stars was raised by Bejgman & Stepanov (1981) although there is little observational evidence in support of this other than a recent detection by Chandra of X-rays from the Cepheid Polaris (Evans et al. 2006). However, the observed X-ray to optical luminosity ratio of Polaris is many orders of magnitude smaller than the corresponding values of the supposed ROTSE Cepheids selected here (Engle, et al. 2006), providing further evidence of their misclassification.

In addition to discounting the ROTSE objects as true pulsating variables, the classification of SAVS J022708+342319 as a Cepheid is also uncertain (Maciejewski et al. 2004) and the RR Lyr variable HR Aur may be an active binary rather than a pulsating star (Loomis & Schmidt 1989). This leaves only V845 Her as a potential Cepheid with X-ray emission, but this appears to have H$\alpha$ in emission (Schmidt et al. 2004b). This too may be a sign of coronal activity and hence a misclassification.

4.3 Newly identified periodic variable stars

Amongst the newly identified periodic variable stars coincident with ROSAT sources, the majority are likely to be X-ray sources as a result of their rapid rotation. Indeed, 15 objects are previously identified as BY Dra stars, UV Cet stars, or other miscellaneous flare stars. We also see a further 30 previously catalogued pre-main sequence stars in this sample, which are likely to be rapid rotators too. We emphasise though that none of these 30 objects have previously been reported as showing coherent periodic variability. Within the set of 43 known young stars reported here (i.e. 13 with previously determined periods and 30 measured here for the first time), the distribution of periods found is quite broad: 7 have periods shorter than 1 d, 10 have periods between 1 d and 2 d, 13 between 2 d and 4 d, and 13 between 4 d and 10 d. We anticipate that the remaining, unclassified objects will contain further examples of these various classes of stars displaying rotational modulation.

Significant numbers of the newly identified objects are clearly eclipsing binaries with periods from a few hours to a few days. We note though that in many cases, the period we have identified will be half the binary period, particularly in the case of W UMa type variables which display two minima of comparable depth. The newly identified X-ray emitting eclipsing binaries appear to include Algol type (e.g. 1SWASP 175540.63+372516.0 and 1SWASP J180331.30+080836.3), $\beta $ Lyr type (e.g. 1SWASP J005101.78+200824.4 and 1SWASP J160248.22+252038.2) and W UMa type (e.g. 1SWASP J021208.77+270818.2 and 1SWASP J133538.39+491406.1) variables, as well as others which display more unusual morphology (e.g. 1SWASP J180207.45+183044.2). Many of these eclipsing binaries will contain tidally locked stars, so their components will also be rapid rotators displaying RS CVn type behaviour. This is likely to be the source of the X-ray emission in these cases too.

4.4 Flux ratios and modulation amplitudes

Given that X-ray activity in the majority of sources we have detected is expected to be linked to rapid rotation, one might expect to see an anti-correlation between X-ray emission and modulation period in our data. The modulation period may be the rotational period of a star, or a binary period, but tidal locking will make these periods identical for many close binaries, so preserving the correlation.

Previous studies have indeed shown strong correlations between X-ray emission and rotation period (e.g. Walter & Bowyer 1981; Pallavicini et al. 1981) but for very rapid rotation the X-ray luminosity is found to saturate at around 0.1% of the bolometric luminosity (e.g. Vilhu & Walter 1987; Wheatley 1998). Pizzolato et al. (2003) show that saturation occurs at rotation periods between 2 and 10 d depending on stellar mass.

\par\includegraphics[width=9cm,clip]{7084_fig3.eps}\end{figure} Figure 3: The X-ray to optical flux ratio (calculated as the mean ROSAT count rate divided by SuperWASP mean flux) plotted against measured period for our sample of 428 periodic variable stars. The line shows the best-fit correlation defined by $F_{\rm X}/F_{\rm O} \propto (P/{\rm day})^{-0.14}$. The symbols representing previously classified objects are shown in the inset key. "Binaries'' include Algol, $\beta $ Lyr and W UMa type eclipsing binaries, as well as RS CVn stars; "rotators'' include BY Dra, UV Cet and pre-main sequence stars; "pulsators'' include those stars catalogued as either RR Lyr or $\delta $ Cep variables; "accretors'' include X-ray binaries and cataclysmic variables.
Open with DEXTER

In Fig. 3 we plot ratios of X-ray to optical flux against our measured modulation periods. $F_{\rm X}$ is defined as the ROSAT count rate and $F_{\rm O}$ as the mean SuperWASP flux in microVegas. We do not attempt to calculate the more usual ratio of X-ray to bolometric luminosity because in most cases we do not have a sufficiently reliable measure of colour to estimate the spectral type and hence bolometric luminosity.

Figure 3 shows very little dependence of X-ray emission on modulation period. This indicates that, if the bulk of our sample are coronal emitters, they must be in the saturated regime. The lack of an obvious decrease in X-ray to optical flux even at periods longer than 10 d indicates that our sample is probably dominated by low mass stars ( $M < 0.6~M_\odot$, Pizzolato et al. 2003). We find a very weak anti-correlation with $(F_{\rm X}/F_{\rm O})\propto(P/{\rm day})^{-0.14}$ and a value for the Pearson correlation coefficient squared of R2=0.02. This may be due to the contribution of binaries containing giants and subgiants (Dempsey et al. 1993).

We further note that the known rapidly rotating isolated stars (BY Dra type, UV Cet type, pre-main sequence stars etc.) tend to lie at higher X-ray to optical flux ratios in this diagram, whilst the binary stars (eclipsing binaries and RS CVn stars) tend to lie at lower flux ratios. The relatively low observed X-ray to optical flux ratios of the RS CVn stars, when compared with the rapidly rotating single stars, may be due to the fact that we have not carried out any bolometric correction. The RS CVn stars will typically be K sub-giants, whilst the isolated rapid rotators are mostly M dwarfs, so their bolometric corrections will be different.

The five accreting binaries follow a different trend of increasing X-ray to optical flux ratio with period. In this case their X-ray emission is clearly not a result of enhanced coronal activity induced by rapid rotation. The cataclysmic variable PX And is detected here at its disc precession period (4.437d), if it were instead plotted at the position of its orbital period (0.146d) it would lie on the same trend as the other four accreting binaries in Fig. 3.

Figure 4 shows a histogram of the X-ray to optical flux ratios in this sample of 428 objects, compared with the flux ratios of the other SuperWASP objects which are coincident with ROSAT objects but did not yield a modulation period from the period searching. Interestingly, the non-periodic sample shows a spread to larger X-ray to optical flux ratios than the periodic sample. These appear to be mostly coincident with galaxies or AGN, which are not expected to be periodic variables. However this sample will also contain periodic sources that are not strongly modulated, and others where the modulation period is not well sampled by the SuperWASP observations.

\par\includegraphics[width=9cm,clip]{7084_fig4.eps}\end{figure} Figure 4: The distribution of X-ray to optical flux ratio for all SuperWASP objects coincident with ROSAT sources. Both the sample of 428 objects displaying a period, and the remaining objects for which no period was found, are shown. Percentages are shown with respect to the individual sample size in each case.
Open with DEXTER

Figure 5 shows the modulation amplitude for our sample of 428 objects plotted as a function of modulation period. Here we see no correlation, other than to note that eclipsing binaries and RS CVn stars generally appear with higher modulation amplitudes than do the isolated rapidly rotating stars. Since most of the newly identified objects lie at lower amplitudes, this might indicate that the majority of these are likewise rapid rotators, rather than eclipsing binaries. However, we should also be mindful of selection effects which will mean that most of the previously identified eclipsing binaries will tend to be those with the deepest eclipses.

\par\includegraphics[width=9cm,clip]{7084_fig5.eps}\end{figure} Figure 5: The fractional modulation amplitude (calculated as the maximum flux minus minimum flux, divided by the maximum flux, measured from the folded and binned lightcurve) plotted against measured period for our sample of 428 periodic variable stars. No correlation is apparent. Symbols are as for Fig. 3.
Open with DEXTER

4.5 Colours

It is also instructive to plot the colours of this sample of objects, as this yields several interesting candidates for further investigation. Figure 6 shows the SuperWASP V - 2MASS K colour versus the 2MASS J - H colour of the objects. The solid line shows the approximate locus of the main sequence from around A0 to M6, for zero reddening. Every magnitude of V band extinction will shift this line upwards by 0.12 in (J-H) and to the right by 0.92 in (V-K), according to the extinction law of Wegner (1994). It is likely that the slight offset between the locus of the data points and the plotted main sequence is a result of the non-standard V magnitude calculated from the SuperWASP unfiltered flux, and does not indicate any significant trend in these objects. The fact that most lie close to this zero-reddening main sequence suggests that the majority are relatively nearby objects. We note that four of the five accreting binaries (i.e. all except DQ Her) are amongst the bluest objects on this plot, whilst many of the pre-main sequence stars are amongst the reddest.

\par\includegraphics[width=9cm,clip]{7084_fig6.eps}\end{figure} Figure 6: The SuperWASP V - 2MASS K colour plotted as a function of the 2MASS J-K colour for our sample of 428 variable stars. The accreting binaries are the amongst the bluest objects in this diagram, whilst some of the pre-main sequence stars are amongst the reddest. Symbols are as for Fig. 3. The solid line shows the main sequence, for zero reddening, from A0 to late M.
Open with DEXTER

Looking first at the blue end of the colour-colour diagram, the two anonymous SuperWASP objects lying closest to the known accreting binaries are 1SWASP J132426.35+303314.2 (HD116635) with P=3.35 d and 1SWASP J153633.39+271029.2 (SAO83906) with P=1.34 d. Although neither have high X-ray to optical flux ratios, these objects warrant further investigation as potential accreting compact binaries. They are much bluer than their spectral classifications listed on SIMBAD (F and G respectively) would suggest.

The one previously classified rotational variable lying just above the accreting binaries with (V-K)= 0.45 and (J-H) = 0.34 is 1SWASP J034433.95+395948.0 which is positionally coincident with an anonymous pre-main sequence star listed on SIMBAD. The SuperWASP lightcurve clearly shows this object to be an eclipsing binary star with a period of 0.2888 d. Given its blue colour this is likely to be mis-classified and not a pre-main sequence star after all.

Moving to the red end of the colour-colour diagram, the one unclassified object lying above the main sequence amongst the very red pre-main sequence stars is 1SWASP J033025.95+310217.9 with (V-K)=4.63, (J -H)=0.84 and P=2.2308 d. As this lies close to the previously catalogued pre-main sequence stars in the Taurus-Auriga star forming complex, this is likely to be another example of this class of objects, but previously uncatalogued.

The reddest of the previously classified binary stars is KW Com, with (V-K)= 4.3, which is listed in the GCVS as an eclipsing binary, although no period has previously been published. Its SuperWASP lightcurve is indistinguishable from many of the rotational lightcurves of the various pre-main sequence stars we have detected, and we therefore suggest it has been mis-classified in the GCVS and is really a young star displaying rotational modulation.

Although most of the unclassified SuperWASP objects lying at the red end of the colour-colour plot appear to be further examples of young stars displaying rotational modulation, there are some exceptions. In particular we note that there are four unclassified SuperWASP objects with (V-K)>3.4, which therefore lie amongst the M stars, and which have lightcurves with a morphology that is suggestive of eclipsing binary stars. These are 1SWASP J142004.68+390301.5 with P=0.3693 d and (V-K)=4.02, 1SWASP J022050.85+332047.6 with P=0.1926 d and (V-K)=3.93, 1SWASP J220041.59+271513.5 with P=0.5235 d and (V-K)=3.61, and 1SWASP J224355.18+293647.6 with P=0.4443 d and (V-K)=3.49. These are likely to be low mass eclipsing binaries, and representatives of a previously very poorly sampled population.

5 Conclusions

We have demonstrated the effectiveness of the SuperWASP survey for detecting photometric modulation on timescales of hours to weeks, in objects within the magnitude range $\sim$8-15. As a result we have recovered the previously identified periodicities in 68 known variable stars coincident with ROSAT X-ray sources, and identified a modulation period for the first time in 360 more. By selecting on objects which are coincident with X-ray sources we have identified eclipsing binary stars and those showing rotational modulation, as well as picking out a few known accreting compact binary stars. We have shown that several previously catalogued pulsating variables coincident with ROSAT sources are likely to be misclassifications. Finally we have identified 4 objects as potential low mass eclipsing binaries on the basis of their lightcurve morphology and colours.

The WASP project is funded and operated by Queen's University Belfast, the Universities of Keele, St. Andrews and Leicester, the Open University, the Isaac Newton Group, the Instituto de Astrofisica de Canarias, the South African Astronomical Observatory and by PPARC.

This research has made extensive use of the SIMBAD database, operated at CDS, Strasbourg, France. We thank Harry Lehto for use of his implementation of the 1D  CLEAN algorithm.



\end{figure} Figure 1: ( left) SuperWASP lightcurves folded at the detected period. Phase zero corresponds to 2004 January 1st 00:00UT in each case. ( right) The associated power spectra, with the frequency corresponding to the detected period indicated by a dotted line. The SuperWASP identifier and the detected period are written above each pair of panels.
Open with DEXTER











































































































Copyright ESO 2007