2005A&A...438..301B

We have analyzed the far-ultraviolet spectrum of two Galactic O4 stars, the O4If+ supergiant HD 190429A and the O4V((f)) dwarf HD 96715, using archival FUSE and IUE data. We have conducted a quantitative analysis using the two NLTE model atmosphere and wind codes, tlusty and cmfgen, which incorporate a detailed treatment of NLTE metal line blanketing. From the far-UV spectrum, we have derived the stellar and wind parameters and the surface composition of the two stars. The surface of HD 190429A has a composition typical of an evolved O supergiant (nitrogen-rich, carbon and oxygen-poor), while HD 96715 exhibits surface nitrogen enhancement similar to the enrichment found in SMC O dwarfs which has been attributed to rotationally-induced mixing. Following studies of Magellanic Cloud O stars, we find that homogeneous wind models could not match the observed profile of OVλ1371 and require very low phosphorus abundance to fit the PVλλ1118-1128 resonance lines. We show, on the other hand, that we are able to match the OV and PV lines using clumped wind models. In addition to these lines, we find that NIVλ1718 is also sensitive to wind clumping. For both stars, we have calculated clumped wind models that match well all these lines from different species and that remain consistent with Hα data. In particular, we have achieved an excellent match of the P v resonance doublet, indicating that our physical description of clumping is adequate. These fits therefore provide a coherent and thus much stronger evidence of wind clumping in O stars than earlier claims. We show that the success of the clumped wind models in matching these lines results from increased recombination in the clumps, hence from a better description of the wind ionization structure. We find that the wind of these two stars is highly clumped, as expressed by very small volume filling factors, namely f_∞=0.04 for HD 190429A and f_∞=0.02 for HD 96715. In agreement with our analysis of SMC stars, clumping starts deep in the wind, just above the sonic point. The most crucial consequence of our analysis is that the mass loss rates of O stars need to be revised downward significantly, by a factor of 3 and more. These lower mass loss rates will affect substantially the evolution of massive stars. Accounting for wind clumping is essential when determining the wind properties of O stars. Our study therefore calls for a fundamental revision in our understanding of mass loss and of O-type star stellar winds.