Because of the high photon count rate of Gem, spectra can be measured for each individual observation listed in Table 2; hence some information on time variations can be derived. It is also possible to divide the data in 402 s bins (402 s is the known wobble period of ROSAT) and look for short time variations.
The spectra were extracted from a circle of 2' centered on the X-ray peak of the PSPC broad-band images. Since the integration time of each spectrum is from 600 s to 1900 s, the spectra comprise at least 5700 counts (see Table 2). As a rule, all bins with a signal-to-noise ratio less than 3 were discarded. Each bin contains therefore at least 15 photons. The number of degrees of freedom in the resulting spectra is about 150.
The data reduction (i.e. background subtraction) was performed in the IRAF/PROS software package, while the subsequent rebinning of pulse height channels occurred under FTOOLS. The final spectral fitting was carried out in the XSPEC (X-ray Spectral Fitting) package.
Due to the low spectral resolution of the PSPC, individual spectral lines are invisible. The resulting pulse height spectra can be modeled by a hot, optically thin plasma with line emission from prominent ions, convolved with the PSPC detector response matrix, to estimate source temperatures and emission measures. The Raymond-Smith plasma code (Raymond & Smith 1977) was employed for the spectral fitting. When using this code one should be aware that there may be systematic errros associated with errors in the underlying atomic physics (missing lines, inaccurate oscillator strengths). Furthermore, steady state ionization equilibria are assumed and photoionization is ignored. Therefore, if much of the plasma is in a flaring state, systmatic errors imply that temperatures and emission measures may be less accurate than appears from the stated random errors.
There is evidence of reduced metal content in Gem. A reduction of 50% as compared to Solar values has been adopted. This is in agreement with the observational results of Randich et al. (1994), regarding photospheric abundances. Coronal and photospheric abundances may be different especially for low first ionization potential elements like Fe (FIP effect). The first ionization potential of Fe is 7.87 eV, and it should have an enhanced abundance in the corona. The FIP effect has been studied in only a few stars and was not detected in Procyon, where coronal abundances were in agreement with their photospheric values (Drake et al., 1995).
All 11 spectra can be well fitted with a two temperature model. Five parameters were involved in the fitting procedure: the interstellar hydrogen column density , two temperatures and their corresponding emission measures. Since Gem is rather close, at a distance of 55.6 pc, and the PSPC is insensitive to the expected low value of the interstellar hydrogen column density, was kept fixed at an average value of obtained from the pre-fitting. Thus, the number of parameters to be fitted was reduced to four. Fig. 1 shows one of the PSPC spectra for Gem together with the curve which best fitted the data. The results for all observations are summarized in Table 3. Clearly, acceptable fits were derived in all cases (). One temperature models did not yield satisfactory fits to the PSPC spectra.
Table 3. Summary of spectral fitting.
X-ray observations of cool stars (RSCVn systems) are usually best explained using two-temperature models when 50% of Solar elemental abundances are assumed. But if the metal content is further reduced, satisfactory model fits can be obtained with one temperature (Kürster & Schmitt 1995). We therefore selected one of our ROSAT observations (no. 10 in Table 2), and tried to fit a one temperature model with reduced metal abundance to the data. Excellent agreement was obtained for T = 10 MK when the metallicity was reduced to 0.18 of the Solar value. But such a low metal content is in disagreement with the results of Randich et al. (1994) for Gem.
© European Southern Observatory (ESO) 1997
Online publication: July 3, 1998