Abstract

Context. Possible main formation sites of fluorine in the Universe include asymptotic giant branch (AGB) stars, the v-process in Type II supernova, and/or Wolf-Rayet stars. The importance of the Wolf-Rayet stars has theoretically been questioned and they are probably not needed in modeling the chemical evolution of fluorine in the solar neighborhood. It has, however, been suggested that Wolf-Rayet stars are indeed needed to explain the chemical evolution of fluorine in the bulge. The molecular spectral data, needed to determine the fluorine abundance, of the often used HF-molecule has not been presented in a complete and consistent way and has recently been debated in the literature. Aims. We intend to determine the trend of the fluorine-oxygen abundance ratio as a function of a metallicity indicator in the bulge to investigate the possible contribution from Wolf-Rayet stars. Additionally, we present here a consistent HF line list for the K- and L-bands including the often used 23 358.33 angstrom line. Methods. High-resolution near-infrared spectra of eight K giants were recorded using the spectrograph CRIRES mounted at the VLT. A standard setting was used that covered the HF molecular line at 23 358.33 angstrom. The fluorine abundances were determined using spectral fitting. We also re-analyzed five previously published bulge giants observed with the Phoenix spectrograph on Gemini using our new HF molecular data. Results. We find that the fluorine-oxygen abundance in the bulge probably cannot be explained with chemical evolution models that only include AGB stars and the v-process in supernovae Type II, that is a significant amount of fluorine production in Wolf-Rayet stars is most likely needed to explain the fluorine abundance in the bulge. For the HF line data, we find that a possible reason for the inconsistencies in the literature, where two different excitation energies were used, is two different definitions of the zero-point energy for the HF molecule and therefore also two accompanying different dissociation energies. Both line lists are correct as long as the corresponding consistent partition function is used in the spectral synthesis. However, we suspect this has not been the case in several earlier works, which led to fluorine abundances similar to 0.3 dex too high. We present a line list for the K- and L-bands and an accompanying partition function.