1998A&A...338.1041L

High signal-to-noise spectra with spectral resolution of more than 10⁵ have been obtained of one normal B9.5V, one normal A1V, two Am stars, and two HgMn B stars having vsini less than 6km/s. These spectra are modeled with LTE line profile synthesis to test the extent to which the spectrum of each star can be modeled correctly with a single set of parameters T_e, logg, chemical abundances, v sin i, and (depth-independent) microturbulent velocity ξ. The answer to this question is important for abundance analysis of A and B stars; if conventional line synthesis does not reproduce the line profiles observed in stars of small v sin i, results obtained from such analysis are not likely to be very precise. The comparison of models with observations is then used to search for direct evidence of atmospheric motions, including line-strength dependent broadening, line core shape, and line asymmetries, in order to study how the microturbulence derived from abundance analysis is related to more direct evidence of atmospheric velocity fields. It is found for the three stars with 12,000≥T_e≥10,200K (the normal star 21 Peg and the two HgMn stars 53 Tau and HD 193452) that ξ is less than 1km/s, and line profiles are reproduced accurately by the synthesis with a single set of parameters. The slightly cooler (T_e≃9800K) star HD 72660 has only a slightly stronger surface convective layer than the hotter stars, but for this star ξ≃2.2km/s. Strong spectral lines all show significant asymmetry, with the blue line wing deeper than the red wing, and have line bisectors which have curvature towards the blue with a span of about 0.5 to 1.0km/s. A single model fits all lines satisfactorily. The two Am stars (HD 108642 and 32 Aqr), with T_e≃8000K, are found to have much larger values of ξ (4 to 5km/s). The strong spectral lines of these two stars are extremely asymmetric, with depressed blue wings, and the bisectors have spans of order 3km/s. No consistent fit to all lines can be found with a single model of the type used here. It is concluded (a) that classical LTE line synthesis is able to reproduce with considerable accuracy the line profiles of late B and early A stars with T_e above about 9500K, but that the LTE model with depth-independent microturbulence provides a very poor approximation for cooler A stars, (b) that curve-of-growth microturbulent velocities in A stars are related to directly detectable atmospheric velocity fields, and (c) that the discrepancies between calculated and observed line profiles in stars with temperatures in the vicinity of 8000K are so large that abundances derived mainly from saturated lines may well contain significant errors. As a by-product, laboratory gf values for Fe II between 3800 and 5300Å have been combined to form a set of data optimized for internal consistency of the gf values.