3 Determination of $T_{\mathrm{eff}}$, logg

For large samples of stars, the best method for $T_{\mathrm{eff}}$, logg determination is based on the use of calibrated photometric indices. For early-type stars the $uvby\beta$ and Geneva photometric systems are the most commonly used. In the $uvby\beta$ photometric system specific filters measure the strength of the H$\beta$ line.

We have determined $T_{\mathrm{eff}}$ and logg using calibrations of both photometric systems. A non systematic discrepancy between the the two sets of values will be used as an additional information to interpret the spectra. We adopted the Moon & Dworetsky (1985) [hereafter MD] and Künzli et al. (1997) calibrations for $uvby\beta$ and Geneva systems respectively.

The homogenized $uvby\beta$ colours are taken from the Hauck & Mermilliod (1990) Catalogue [hereafter HM] and are given in Cols. 2 to 5 of Table 2. In this catalogue the Strömgren indices are erroneous for HD 225200 (Geneva code = 409010021) and we adopted the values from the Simbad data base. The indices are missing for the 2 stars HD 27660 and HD 216931 and we checked that there is no update in the recent edition by Mermilliod et al. (1997). The reddening of these stars has been estimated from the UBV indices: it is negligible for the first star and E(B-V)=0.06 is derived for the second one. We also note that the secondary component of the broad visual double HD 87344, indicated as HD 87344(2) (V=8.0), in the Hauck and Mermilliod Catalogue is in reality HD 87330 (V=7.13, B9III-IV). Its spectrum clearly shows that the star HD 87330 is an SB2 with a well developed system of double lines at the epoch of our observations (May 13, 1990 and April 8, 1993).

For the Geneva photometry the version of the Catalogue available at CDS has been used. HD 129791 is not included in this Catalogue.

We recall that as a first step dereddened colours must be computed; for our sample of stars, dereddening is expected to be very low or negligible; a strong reddening is likely to indicate a flux distortion due to spectral peculiarities or undetected binarity.

3.1 Reddening and its correction

For these bright stars, mainly belonging to the BSC, standard methods of dereddening are assumed to be valid, these stars being expected to be slightly reddened. We used the programs by Moon (1985) to compute the colour excess. A similar determination cannot be performed from the Geneva photometric indices because no updated procedures are published.

  

Table 2: Photometric, astrometric data and atmospheric parameters

The A-type stars are in the domain where the Balmer lines reach their maximum. According to Strömgren (1966), the $uvby\beta$ calibration requires that each star should be assigned to one of the three groups: early ($\beta$ depends mainly on the luminosity, c₀ is mainly related to the $T_{\mathrm{eff}}$), late (the roles of $\beta$ and c₀ are reversed) or intermediate (for which a combination of various indices must be used). In the program UVBYBETA by Moon (1985) the choice of the group is based on photometric quantities and spectral classification. Since the boundaries between the groups overlap in some cases, the assignment to a group is sometimes ambiguous and uncertainty in group selection cannot be avoided. In this case, the computations have been made for both groups; the result is an average difference of 0.02 in the colour excess E(b-y). Details concerning this effect can be found in Gerbaldi et al. (1998a). The final choice of the group defined in the UVBYBETA program is given in Col. 6 of Table 2; this choice is made on the basis of the lower E(b-y) and by the analysis of the atmospheric parameters derived for both cases (see next subsection).

The determination of the amount of reddening, being based on the empirical calibration of the $uvby\beta$ system, is model independent. However, the intrinsic colours are slightly different according to different authors and this may lead to different values of the colour excess; a discussion on this effect can be found in Figueras et al. (1991) and in Jordi et al. (1997). According to these papers, for our sample of nearby stars, we can safely use the Moon program UVBYBETA without any further modification.

The value of E(b-y) is given in Col. 7 of Table 2. Only 6 stars have E(b-y) higher than 0.02; to this we can add the estimated value for HD 216931. The highest value, E(b-y)=0.05, is found only in one case, for HD 151527, which with HD 111786 [E(b-y)=0.00] are the only two stars to present the peculiarity of a (b-y) index very high for stars belonging to the A0 type. These two newly detected binaries are further discussed below. For the three stars HD 7916, HD 114570 and HD 129791 the colour excess is E(b-y)=0.04; for two others HD 67725, HD 104039 the value of the colour excess is E(b-y)=0.03.

For the remaining stars the E(b-y) is what expected, i.e. in the range -0.01, +0.02, with the exception of HD 60629, which has a slightly higher "blueing" E(b-y)=-0.02. These values of reddening and bluing are interpreted as possible sign of abnormality. These stars will be discussed in Sect. 7.

3.2 $T_{\mathrm{eff}}$ and logg from $uvby\beta$ photometry

The atmospheric parameters have been derived from the $uvby\beta$ photometry, using the calibration by Moon & Dworetsky (1985) (Table 2, Cols. 8 and 9) which has been tested by recent studies (Napiwotzki et al. 1993; Smalley & Dworetsky 1993, for stars cooler than A0). The corrections determined by Castelli (1991) and by Dworetsky & Moon (1986) have no influence on our range of parameters.

The internal errors on $T_{\mathrm{eff}}$ and logg due to the scatter of the individual observed colours have been estimated by Lemke (1989) to be approximately $\pm$100 K and $\pm$0.1 dex respectively. A different approach to the evaluation of these errors is found in Gerbaldi et al. (1998b), where it is shown that a difference of 0.015 in colour excess implies a difference of 200 K in $T_{\mathrm{eff}}$.

When the value of the colour excess has been found to be negative, no correction has been applied to the colour indices $T_{\mathrm{eff}}$ and logg.

3.3 $T_{\mathrm{eff}}$ and logg from Geneva photometry

We computed the atmospheric parameters from the Geneva photometric indices by using the recent calibration by Künzli et al. (1997).

This system provides an independent way to determine $T_{\mathrm{eff}}$ and logg since it does not include any filter centered on a Balmer line. We limited this computation to undereddened stars, i.e. to those which, according to $uvby\beta$ photometry, have a colour excess E(b-y) in the range $\pm$ 0.01. The computed $T_{\mathrm{eff}}$ and logg are given in Table 2, Cols. 10 and 11.

3.4 Comparison of $T_{\mathrm{eff}}$ and logg values

The atmospheric parameters derived from the two photometric system are directly comparable only for stars with E(b-y)=0.0. In order to have a larger sample, we consider as unreddened the stars with $E(b-y)\leq 0.01$ and, in order to have homogeneous data, the MD parameters have been recalculated for this comparison without taking into account the colour excess for stars with E(b-y)=0.01.

Figures 1a and 1b display the results for $T_{\mathrm{eff}}$ and logg respectively; Fig. 1a does not include the abnormally low values of $T_{\mathrm{eff}}$ derived from both photometric systems for the $\lambda$ Boo star HD 111786, which is, in fact, a binary, as demonstrated by Faraggiana et al. (1997).

We check the consistency of the two sets of $T_{\mathrm{eff}}$ and logg and we look for possible systematic differences. The values of $T_{\mathrm{eff}}$ from the $uvby\beta$are systematically higher than those from the Geneva photometry. This difference is not related to the stellar rotation as it appears from the absence of any systematic relation between the $\Delta$$T_{\mathrm{eff}}$ and the $v\,\sin\,i$ value. Nevertheless, the difference between these two sets of values is small, the average being 150 K. The logg comparison shows higher scatter than that of $T_{\mathrm{eff}}$; the largest difference refers, to the already cited HD 111786. The star HD 85504, is an intriguing object with peculiar spectrum and kinematics (see the Appendix).

  

Figure 1: Relation between $T_{\mathrm{eff}}$ (Fig. 1a) and logg (Fig. 1b) derived from MD and Geneva calibrations; no dereddening correction are applied to the colours in both systems, and only stars with $E(b-y)\leq 0.01$ are considered

For most of the stars, logg (MD) is higher than logg (Gen) for large logg values, while the opposite is true for logg lower than 4. The correction of logg (MD) proposed by Napiwotzki et al. (1993), being independent of the logg value, does not solve this discrepancy.

A discrepancy between logg values obtained from these two sets of photometric indices, such as for HD 85504 and HD 111786 can be interpreted as a sign of difference in their flux distribution compared to that used for calibration. The H$_\beta$ intensity plays an important role in the calibration of the Strömgren photometry, so we suggest that for these two stars, such a difference shows that their H$_\beta$ intensity is not coherent with their continuum character.

Independently of the adopted calibration, it is clear that the A0 dwarf stars occupy a broad domain of $T_{\mathrm{eff}}$ and logg which leads to a loose correlation with spectral type and luminosity class appears. We remind that the internal accuracy reached by different MK classifiers is of $\pm$0.7 in luminosity class (Jaschek & Valbousquet 1997); the logg parameter plotted in Fig. 1b, cannot be directly related to this luminosity class.

If we exclude the binary HD 111786, the $\lambda$ Boo star HD 31295 ($T_{\mathrm{eff}}$ (MD) = 8763 K) and the peculiarly reddened star HD 151527 ($T_{\mathrm{eff}}$(MD) = 7610 K), $T_{\mathrm{eff}}$(MD) spans from 9070 K (HD 21473) to 10930 K (HD 87344) and logg (MD), when only A0 V stars are considered, covers the range from 3.30 (HD 67725) to 4.50 (HD 80950).