Astron. Astrophys. 341, 751-767 (1999)

2. HRI observations and data analysis

We present data from eight ROSAT High Resolution Imager (HRI, Zombeck et al. 1990) observations targeted toward the Pleiades. These pointing directions were selected to resolve source-confused regions in images obtained with the ROSAT Position Sensitive Proportional Counter (PSPC, Briel & Pfeffermann 1995), and to narrow the error circles of previously unidentified PSPC sources. By targeting regions of the cluster previously covered only by the less sensitive external part of the PSPC field, these HRI exposures of about 30 ksec reached limiting sensitivities comparable (within a factor two) to those achieved in the central part of existing PSPC observations. In addition, one of the pointings (ROR 202060) was designed to determine the X-ray luminosities of HII-2341 and HII-2284, two of the mostly slowly rotating K dwarfs in the Pleiades.

A journal of observations is presented in Table 1. Note that the two segments of ROR 201412 were also analyzed separately since they were obtained nearly a year apart. For a subsample of stars, this circumstance and the partial overlaps between fields permit a study of X-ray variability on the time scale of a year.

Table 1. Characteristics of the HRI observations: ROSAT observation Request (ROR) number, field center coordinates, nominal exposure times, dates of observation, and number of detections obtained using a Wavelet-based algorithm (cf. Sect. 2.1). For ROR 201412, we list separately two segments obtained nearly a year apart.

In Fig. 1 we show the sky coverage of the HRI fields together with the optical positions of stars from the Pleiades membership catalog we have adopted. ¹ Also indicated in the figure are the fields of view of previous PSPC observations (Paper I; Paper II). Note that the HRI fields targeted regions previously seen only in the outer, less sensitive regions of the PSPC fields of view.

Fig. 1. Targeting of HRI (small solid circles) and previous PSPC (large dotted circles) observations, superposed on a map of known Pleiades stars.

The combined fields of view of the eight HRI images contain 260 stars with astrometric and/or photometric indications of membership; optical characteristics of these stars are summarized in Table 2. Photometry is photoelectric when such data are available, and photographic otherwise.

2.1. Source detections

For the detection of sources, we applied the Wavelet transform method of Damiani et al. (1997). While this algorithm does not require an exact knowledge of the detector point spread function (PSF), it was nevertheless specifically "tuned" for our HRI application (Damiani et al., in preparation). Over all fields, we found 177 distinct X-ray detections, some of which were detected in more than one field.

In this particular application of the algorithm, a flat fielded image was constructed from both an exposure map (which models the vignetted cosmic X-ray background and the intrinsic detector nonuniformities) and a particle map (which models particle-induced background that increases at large off-axis angles and contributes a substantial fraction of the total observed background, (Snowden 1998, Snowden et al. 1994).

Our detection thresholds were set to correspond roughly to a gaussian equivalent of 4.5 . For typical livetimes of 30 ksec, such thresholds yield 1 spurious source per field. Due to its high spatial resolution, the typical number of photons per HRI resolution element is small, even for rather long exposures. With photon statistics in the HRI images therefore far from the gaussian limit (cf. Damiani et al. 1997), a large number of simulations are required to determine proper acceptance thresholds. At the end of the detection process, we removed instrumental hot spots known to be present in a fraction of fields (cf. David et al. 1993).

Count rates were computed using source effective exposure time as function of position in the field of view (exposure map) computed according to the recipe of Snowden et al. (1994).

2.2. Identifications

Our search for catalogued counterparts of Pleiades X-ray detections used a coincidence circle of 20" radius. Following the initial identification process, we deduced the presence of a systematic offset between the optical and X-ray positions of up to 6 arcsec (median of measured values in an individual image) in some of the observations. Such shifts are due to limitations of the ROSAT aspect corrections and are of the same order as the spatial resolution of the HRI. In order to improve our identifications, we corrected the X-ray source positions for the mean offset measured in each field and then repeated the identification process. After this second iteration, we identified 117 X-ray sources with 120 Pleiades members and 24 with field stars from SIMBAD catalogs or with stars known as non members, either from the Hertzsprung catalog (Hertzsprung 1947) or from HCG catalog (Stauffer et al. 1991). A remaining 36 detections could not be identified with any cataloged objects. An inspection of finding charts, reported in appendix B, shows that in a large fraction of the cases there is an "obvious" stellar counterpart inside the error circle. We identify most of the Pleiades stars with an offset smaller than 8 arcsec ( 87%), all the other have an offset smaller than 15" but one (HII-153 ²) that has an offset of 19 arcsec. In some cases a large offset can be due to position uncertainty in our optical catalog (such as in the case of TS-51x which has a position different by 8 arcsec from the corresponding star in the Guide Star Catalog, GSC). The probability of identifying an X-ray source with a Pleiades member uncorrelated with the emission is very small and the number of expected spurious identifications in the entire survey is less than 3.

Even the high angular resolution of the HRI was unable to resolve ambiguities in five cases: (1) HII-298/299, (2) HII-883/879, (3) HII-1794/1805, (4) HII-1392/1397, and (5) HII-2500/2507. For the purpose of computing luminosities, in the first three cases we apportioned the count rate equally between two late-type counterparts; in the remaining two cases, we attributed the emission to a late-type star rather than its A-type neighbor.

For purposes of consistency, we used the same Wavelet algorithm to determine count rate upper limits for all undetected members in our Pleiades catalog.

In seeking an explanation for the large number of unidentified sources, we computed the number of sources unrelated with the Pleiades expected in our fields, using sensitivity maps obtained from the Wavelet algorithm at a spatial resolution of 10"10". When a given area of sky was observed more than once, we retained the observation of highest sensitivity. For each 10"10" pixel we estimated the number of expected sources, assuming a hydrogen column N_H = 10²¹ cm^{- 2} obtained by interpolation from data of Stark et al. (1992) and power law spectra with photon indexes between 1 and 2. For our full 2.2 square degree region, we estimate a total of 26 to 27 field sources, using the log(N)-log(S) of Branduardi et al. (1994), or between 32 and 34 sources, if we adopt the log(N)-log(S) of Hasinger et al. (1993). These numbers suggest that most of our unidentified sources are of origin unrelated with the Pleiades.

2.3. X-ray fluxes and luminosities

We employed the detailed shape of the HRI PSF (David et al. 1993) in determining fluxes. We computed a conversion factor from count rate to flux in the 0.1-2.4 keV band, assuming a single-temperature Raymond spectrum with kT=0.8 keV and an average value of log cm^-2. Since Gagné et al. (1995) found typical Pleiades coronal temperatures in the range of 0.8-1.4 keV, our temperature assumption introduces a systematic conversion factor uncertainty of less than 20%. For stars with known anomalous reddening (Stauffer & Hartmann 1987), we have computed the conversion factors assuming kT=0.8 keV and the hydrogen column derived from the individual E(B-V). For consistency with previous X-ray measurements, luminosities were computed assuming the "standard" cluster distance of 127 pc instead of the new Hipparcos distance (116 pc, Mermilliod et al. 1997).

X-ray characteristics of the 260 Pleiades stars falling in our fields of view are summarized in Table 3, where the first and second columns list a running number and star name, and columns 3 and 4 report offsets between X-ray and optical positions in arc seconds of right ascension and declination, respectively. In column 5 we report the detection significance in equivalent (i.e., same probability as the normal distribution with this value). Columns 6 and 7 contain the X-ray count rate and its error, and columns 8 and 9, the value of and , respectively. The bolometric correction has been computed using Johnson's (1966) data for stars with B - V 1.34, Monet et al.'s (1992) data for stars redder than B - V 1.34 or V - I 1.6 and the conversion from M_V to M_BOL by Lang (1992) for the stars without measured colors. For B stars, the values of and are placed in parenthesis because we believe they suffer UV leak contamination (cf. Sect. 2.4). We show in Figs. 2 and 3 scatter plots of and versus B - V, respectively. Vertical line segments connect different measurements for the same star. The distribution for HRI sources may be compared with the log(L_x/L distribution based on PSPC sources in the Pleiades as shown by Giampapa et al. (1998, Fig. 5)

Fig. 2. Distribution of versus B - V. Vertical line segments connect different measurements of the same star.

Fig. 3. Scatter plot of versus B - V. Vertical line segments connect different measurements of the same star.

Table 3. X-Ray properties of observed Pleiades stars

Table 3. (continued)

Table 3. (continued)

Table 3. (continued)

Table 3. (continued)

2.4. Indeterminate luminosities for B-type stars

An investigation of both the UV susceptibility of the HRI detector and the B-star count rates inferred in this study has led us to the conclusion that the HRI cannot determine accurate X-ray luminosities for B-type stars. Whereas the HRI detected 8 of 11 B stars present in the observed Pleiades region, the PSPC detected only 3 of these stars. Moreover, the HRI X-ray luminosities naively inferred for the three common B star detections are significantly higher than those measured with the PSPC.

Recent laboratory measurements have shown that the HRI UV/ion shield has a "UV leak" - enhanced transmission for wavelengths longward of 2000 Å (Barbera et al. 1997, Zombeck et al. 1997), where early-type stars with photospheric temperatures of the order of 10000-15000 K have the bulk of their emission. In spite of its much lower sensitivity to UV emission, however, the PSPC still detected some B stars. Hence we cannot arbitrarily attribute all detected HRI counts to UV contamination.

In Fig. 4 we report the measured luminosity excess (the difference between luminosity inferred from the HRI and that inferred from the PSPC) versus the absolute bolometric magnitude for the B stars in our survey. Also included in this figure is the excess for the star Vega (square+cross, from Zombeck et al. 1997), plotted by applying the conversion factor used here to its measured HRI count rate. The clear dependence of observed excess on absolute stellar magnitude is entirely attributable to the UV leak.

Fig. 4. Measured luminosity excess (difference between luminosities inferred from the HRI and the PSPC) versus absolute bolometric magnitude for B stars in our survey. Open squares indicate the three stars detected both in the HRI and PSPC, the arrows indicate the stars detected with the HRI and undetected with the PSPC. The square with plus indicate Vega (see text).

© European Southern Observatory (ESO) 1999

Online publication: December 16, 1998
helpdesk.link@springer.de