ForumSpringerAstron. Astrophys.
ForumWhats NewSearchOrders

Astron. Astrophys. 326, 662-668 (1997)

Previous Section Next Section Title Page Table of Contents

2. Observations

2.1. Soft X-rays

RX J0720.4-3125 was discovered as soft X-ray source in the ROSAT all-sky survey data in the course of the Galactic Plane Survey. The aim of this project is to optically identify ROSAT sources in the galactic plane between [FORMULA] galactic latitude (Motch et al. 1991). A description of the satellite and the position sensitive proportional counter (PSPC, 0.1 - 2.4 keV) may be found in Trümper (1983) and Pfeffermann et al. (1986). The high resolution imager, HRI, is described by David et al. (1993).


Table 1. Soft X-ray detections of RX J0720.4-3125 (0.1-2.4keV)

The source was re-observed in dedicated pointings with the PSPC and HRI detectors. A summary of the observations is presented in Table 1 together with previous soft X-ray detections (see below). The light curve obtained from the PSPC observation is consistent with a constant count rate at a mean compatible to the survey intensity. The average PSPC spectrum (Fig. 1) is very soft with practically no emission above 1 keV. Among single component models thermal bremsstrahlung and blackbody models acceptably fit the spectrum and the derived parameters are summarized in Table 2. Thermal emission from an optically thin plasma with a temperature of 150 eV is however bound to show emission lines, but e.g. a Raymond-Smith type model yields a reduced [FORMULA] of 1.6 favouring the blackbody model. For the latter a bolometric luminosity of Lbol = 2.6 1031 (d/100 pc)2 erg s-1 and an emission area of 65 (d/100 pc)2 km2 is derived.

[FIGURE] Fig. 1. The PSPC spectrum of RX J0720.4-3125 fitted by a blackbody model. The residuals are shown in the lower panel


Table 2. Bremsstrahlung and blackbody fits to the PSPC spectrum of RX J0720.4-3125. The errors are 90% confidence for two parameters of interest. The observed flux is in the range 0.1-2.4 keV.

A folding analysis of the PSPC data using a Rayleigh Z2 test (Buccheri et al. 1983) reveals a periodic modulation at 8.3914 s. However the uncertainty in cycle counts between the observation intervals causes several peaks with offsets of multiples of about 0.01 s around this period which can formally not be rejected. The power spectrum is shown in Fig. 2 with a peak near 0.12 Hz. To determine the uncertainty of the pulse period a pulse phase timing analysis was used. The data was divided into five intervals, between 500 s and 800 s long, and for each the phase shift of the folded light curve relative to the total was derived by cross-correlation of the pulse profiles. This yields a formally best period of 8.3914 [FORMULA] 2 10-4 s (90% confidence), but periods offset by multiples of 0.01 s can not be excluded up to [FORMULA] 0.03 s. The light curve folded by the pulse period is plotted in Fig. 3 and shows a sinusoidal semi-amplitude modulation of 11 [FORMULA] 2%.

[FIGURE] Fig. 2. Power spectrum from the PSPC observation from Sep. 27, 1993. The peak near 0.12 Hz indicates the period of 8.3914 s

[FIGURE] Fig. 3. The light curve from Sep. 27, 1993, folded with a period of 8.3914 s, for clarity data points are repeated for a second period

In the light curves of the HRI observations no statistically significant variations are seen (an example is shown in Fig. 4) and a Kolmogorov-Smirnov test yields in all cases a probability for variability of below 2.5 sigma.

[FIGURE] Fig. 4. The HRI light curve of RX J0720.4-3125 on Apr. 25, 1996 (15:19:32 - 17:34:29 UT) with a time resolution of 20 s

The timing analysis was performed in the same way as for the PSPC data. The pulse folding reveals two peaks in the Z2 test of the first HRI observation at periods of 8.3797 s and 8.3917 s with the peak at the shorter period only insignificantly higher. From pulse timing analysis of three observation intervals the period is derived to 8.3917 [FORMULA] 8 10-4 s (taking the peak which is most significant in the PSPC observation). However, like for the PSPC data, periods away by multiples of 0.01 s can not be excluded and in particular the period of 8.3797 s is equally prominent.

During the second HRI observation when the source was 10´ off-axis the statistics was too low to derive useful constraints on the pulse period.

The deep 33 ksec HRI observation performed in Nov. 1996 was distributed over more than 18 hours and allows to uniquely determine the pulse period to 8.39115 [FORMULA] 2 10-5 s with a probability for chance detection of 4 [FORMULA] derived from the [FORMULA] test (Fig. 5). The folded light curve with a sinusoidal modulation of 12 [FORMULA] 2% is plotted in Fig. 6 and is consistent with that from the PSPC observation. A linear fit to the 3 measured periods yields a P_of -2.6 [FORMULA] with a 90% confidence range of -6.0 [FORMULA] to 0.8 [FORMULA], compatible with no period change.

[FIGURE] Fig. 5. [FORMULA] test for the HRI observation from Nov. 3-4, 1996

[FIGURE] Fig. 6. The HRI light curve from Nov. 3-4, 1996, folded with a period of 8.39115 s and repeated for two cycles

The best determined PSPC position was derived from the survey data utilizing the X-ray detection of the nearby, optically identified source EUVE0720-317, for bore-sight correction (only about 2" were required). The remaining 90% confidence error of 9.6" is mainly determined by the statistical uncertainties in the survey data. Unfortunately the EUVE source is partly hidden by the detector window support structure in the pointed PSPC observation, resulting in a 15" error, dominated by bore-sight uncertainties. To derive the most accurate position of RX J0720.4-3125 the two HRI observations from May and Nov. 1996 can be used. The first of the two pointings was directed to include both RX J0720.4-3125 and EUVE0720-317 at an off-axis angle of 10´, while six other faint objects were detected in the HRI image of the latter observation. In the first case an interactive analysis using the maximum likelihood technique of EXSAS (Zimmermann et al. 1994) gives RA (2000) = 07h 20m 25:s00 Dec = -31o 25´ 46:003 with a 90% confidence error of 3" (statistical errors of the X-ray positions and 2" error for the optical position of EUVE0720-317). The HRI pixel size has recently been determined to 99.57% of the originally assumed value (see discussion in Neuhäuser et al. 1997) which shifts the HRI position by about 5" towards EUVE0720-317 to RA (2000) = 07h 20m 25:s09 Dec = -31o 25´ 51:008.

During the deep HRI observation from Nov. 1996 EUVE0720-317 was outside the field of view of the detector, but six other X-ray sources were detected with likelihoods exceeding 10. For five of the X-ray sources close objects can be found in the Digitized Sky Survey (Laidler et al. 1996) with distances between 0.7" and 7.4" (for only one object the distance exceeds 3.1") and R magnitudes between 9.4 and 16.0. The HRI positions of the five X-ray sources were again corrected for the HRI pixel size which improved the coincidence with the optical positions by 1" on average. A firm identification of the X-ray sources requires detailed optical observations, but on a statistical basis they can be used to reduce the systematic bore-sight uncertainty. A correction of only -0.8" was found by minimizing the distances between X-ray and optical position. Taking into account the statistical errors on the X-ray positions and 2" uncertainty for the optical positions an uncertainty of 3" remains. The position of RX J0720.4-3125 was determined to RA (2000) = 07h 20m 24:s90 Dec = -31o 25´ 51:003. The two obtained confidence circles for the position of RX J0720.4-3125 overlap and are shown in Fig. 7 (see below).

[FIGURE] Fig. 7. CCD R-band image with 300 s exposure around the X-ray position of RX J0720.4-3125. The circles represent the 90% confidence regions from the X-ray positions of the HRI observations with available bore-sight correction. From the three faint objects A, B, and C optical spectra were taken. The size of the image is about 2.2´ x 2.2´, north is to the top and east to the left

RX J0720.4-3125 was also in the field of view of EXOSAT LE observations on March 12, 1984, using the 3000 Lexan and the Aluminum-Parylene filters. The observed intensities are consistent with the ROSAT values. All the soft X-ray observations are summarized in Table 1. Assuming the blackbody model derived from the PSPC spectrum of the pointed observation, the expected intensities for the other detectors are calculated. The count rates differ by less than 5% from the expected count rates calculated for the assumed spectral model. The EXOSAT position derived from the observation using the 3000 Lexan filter is 10" away from the best HRI positions but has a large error radius of 15".

2.2. Optical

CCD images of the RX J0720.4-3125 field were obtained using the South African Astronomical Observatory's 1.0-m telescope in November 1995. The Tek4 camera, utilizing a Tektronics 512 x 512 CCD, was used to obtain B, V, R, I and H-alpha images. The CCD frames were cleaned, bias subtracted, and analyzed using the DoPHOT analysis program (Mateo & Schechter 1989) to derive PSF and aperture magnitudes of all stars. Colour-colour and colour-magnitude plots were produced (e.g. U-B vs V) to identify unusual objects in the frame. No such object, nor indeed any stellar image, was found in or near the X-ray error circles. All stars in the vicinity had colours consistent with normal field stars. Only one unusual object, exhibiting possible H-alpha emission (from a comparison of H-alpha versus R magnitudes), was seen in the entire 9´ x 9´ field. However it was 5´ from the X-ray position, clearly unassociated with the X-ray source. In Fig. 7 the R-band image around the position of RX J0720.4-3125 is shown together with the positional error circles from the ROSAT HRI observations.

No objects fainter than B = 21.0, V = 20.7 and R = 21.4 magnitudes are seen in the images. The limiting V magnitude yields a f [FORMULA] /f [FORMULA] [FORMULA] 510 for RX J0720.4-3125 (using log(f [FORMULA] /f [FORMULA] ) = log(f [FORMULA] ) + (m [FORMULA] /2.5) + 5.37 (Maccacaro et al. 1988)).

In March 1996 we observed the three faint objects about 6-10" away from the best HRI position (Fig. 7) spectroscopically at the ESO/MPIA 2.2 m telescope. Their V magnitudes range between 19.9 and 20.1. The observations were obtained with the EFOSC2 spectrometer which was equipped with a Thomson 1024 [FORMULA] pixel CCD chip (ESO CCD #19). The spectral resolution obtained with grism #1 and the 1" slit (cf. ESO Users Manual) was [FORMULA] Å. The spectra are displayed in Fig. 8. They are consistent with those of normal stars.

[FIGURE] Fig. 8. Spectra of the three objects marked in Fig. 9. Spectra #B and #C are vertically shifted by 0.5 units in order to avoid overlap. "x"denotes residuals of a night sky line. Some characteristic spectral features are indicated. The spectra of objects #A and #B are consistent with [FORMULA] F-G type stars, although in #B the absorption features are relatively faint. Object #C is an M-type dwarf
Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1997

Online publication: October 15, 1997