2012A&A...542A..71H

Line ratios in ``fir'' triplets of helium-like ions have proven to be a powerful diagnostic of conditions in X-ray emitting gas surrounding massive stars. Recent observations indicate that these ratios can be variable with time. The possibilities for causes of variation in line ratios are limited: changes in the radiation field or changes in density, which would result in radiational or collisional level-pumping, respectively; and changes in mass-loss or geometry, which could cause variations in X-ray absorption effects. In this paper, we explore the first of these potential causes of variability: changes in the radiation field. We therefore explore the conditions necessary to induce variability in the ratio R=f/i via this mechanism. To isolate the radiative effect, we use a heuristic model of temperature and radius changes in variable stars in the B and O range with low-density, steady-state winds. We then model the changes in emissivity of X-ray emitting gas close to the star due to differences in level-pumping owing to the availability of UV photons at the location of the gas. We find that under these conditions, variability in R is dominated by the stellar temperature. Although the relative amplitude of variability is roughly comparable for most lines at most temperatures, detectable variations are limited to a few lines for each spectral type. We predict that variable values in R due to stellar variability must follow predictable trends found in our simulations. Our heuristic model uses radial pulsations as a mode of stellar variability that maximizes the amplitude of variation in R. This model is robust enough to provide a guide to which ions will provide the best opportunity for observing variability in the f/i ratio at different stellar temperatures, and the correlation of that variability with other observable parameters. In real systems, however, the effects would be more complex than in our model, with differences in phase and suppressed amplitude in the presence of non-radial pulsations. This, combined with the range of amplitudes produced in our simulations, suggests that changes in R across many lines concurrently are not likely to be produced by a variable radiation field.