2007A&A...469.1083K

Historically, stellar model atmospheres with scaled solar abundances (for all elements heavier then helium) have been widely used for analysis of the atmospheres of chemically peculiar (CP) stars. However, in reality, atmospheres of CP stars demonstrate a variety of abundances, not necessarily scaled to the solar composition. We study the effects of individual abundance patterns on the model atmospheres of CP stars. The main purpose is to conduct a systematic homogenous study to explore the abundance parameter space occupied by these stars. We calculated a grid of the model atmospheres of A and B stars (logg=4.0) for different effective temperatures (T_eff=8000, 9500, 11000, 13000, 15000, 20000K) and chemical compositions. We used the LLmodels code to compute model atmospheres with individual abundance patterns, varying the following elements: C, Mg, Si, Ca, Ti, Cr, Mn, Fe, Ni, Sr, Eu and He. We compared the computational results for these peculiar model atmospheres with those of reference model atmospheres of the solar chemical composition. We present a homogeneous study of model atmosphere temperature structure, energy distribution, photometric indices in the uυbyβ and Δa systems, hydrogen line profiles, and the abundance determination procedure as it applies to CP stars. In particular, we found that Si, Cr and Fe are the main elements to influence model atmospheres of CP stars, and thus to be considered in order to assess the adequacy of model atmospheres with scaled solar abundances in application to CP stars. We provide a theoretical explanation of the robust property of the Δa photometric system to recognize CP stars with peculiar Fe content. Also, the results of our numerical tests using model atmospheres with one or several elements overabundant (Si and Fe by +1dex, Cr by +2dex) suggest that the uncertainty of abundance analysis in the atmospheres of CP stars using models with scaled abundances is less than ±0.25dex. If the same homogeneous models are used for the abundance stratification analysis then we find that the uncertainty of the value of the vertical abundance gradient is within an 0.4dex error bar. Model atmospheres with individual abundance patterns should be used in order to match the actual anomalies of CP stars and minimize analysis errors.