2010A&A...522A..81A

We investigate the detectability of Zeeman broadening in optical Stokes I spectra of slowly rotating sun-like stars. To this end, we apply the LTE spectral line inversion package SPINOR to very-high quality CES data and explore how fit quality depends on the average magnetic field, Bf. One-component (OC) and two-component (TC) models are adopted. In OC models, the entire surface is assumed to be magnetic. Under this assumption, we determine formal 3σ upper limits on the average magnetic field of 200G for the Sun, and 150G for 61Vir (G6V). Evidence for an average magnetic field of ∼500G is found for 59Vir (G0V), and of ∼1000G for HD68456 (F6V). A distinction between magnetic and non-magnetic regions is made in TC models, while assuming a homogeneous distribution of both components. In our TC inversions of 59Vir, we investigate three cases: both components have equal temperatures; warm magnetic regions; cool magnetic regions. Our TC model with equal temperatures does not yield significant improvement over OC inversions for 59Vir. The resulting Bf values are consistent for both. Fit quality is significantly improved, however, by using two components of different temperatures. The inversions for 59Vir that assume different temperatures for the two components yield results consistent with 0-450G at the formal 3σ confidence level. We thus find a model dependence of our analysis and demonstrate that the influence of an additional temperature component can dominate over the Zeeman broadening signature, at least in optical data. Previous comparable analyses that neglected effects due to multiple temperature components may be prone to the same ambiguities.