Excitation of O I lines in the solar chromosphere

Observations of O I lines in the solar spectrum are examined to determine whether differences in behavior of lines of the quintet and triplet term systems are consistent with collisional excitation and/or photoexcitation of both quintets and triplets. Intensities, I_IR, in near-infrared emission lines observed above the limb at total eclipse decrease exponentially with height h. The inverse scale heights (d ln I_IR/dh) for the triplet lines at 844.6 nm and quintet lines at 777.2 nm are found to be in the ratio of 1.45. Ultraviolet O I emission-line intensities I_UV observed on the solar disk show strong variations, and the distributions of triplet (130.4 nm) and quintet line intensities about the means are different. Variances in In I_UV are found to have a triplet-to-quintet ratio of 1.50, in close agreement with the ratio of dlnI_IR/dh. It is shown that the simple assumption of collisional excitation of quintets and triplets coupled with collisional de-excitation of the quintets leads to the correct ratios for both the UV variances and dlnI_IR/dh. Also, under this assumption d ln I_IR/dh for the quintet lines is predicted to have the same value as dlnI/dh at the head of the hydrogen Balmer continuum, which, in fact, it does. On the other hand, Carlsson & Judge (1993) have shown that collision rates computed from the Vernazza, Avrett, & Loeser (1981, hereafter VAL) model chromosphere using current estimates of O I collision strengths are too low to produce the observed mean intensity in O I 130.4 nm. In a similar sense, we find that the predicted intensity of O I 130.4 nm is much too weak relative to O I 135.6 nm, and that the VAL mean models A-F cannot reproduce the observed behavior of these lines, even including photoexcitation by H Lyβ. These difficulties are removed by increasing specific electron-atom collision rates. Such increases could reflect (unacceptably?) large errors in atomic cross sections close to threshold and/or the inadequacy of the assumptions made by VAL for predicting line intensities. The nonlinear dependence of line intensities on temperature and density, especially for far-UV lines, makes the latter alternative a likely factor. We conclude that the O I UV lines are very sensitive to inhomogeneities, much more so than more traditional chromospheric lines (e.g., Mg II k) which are formed over similar regions of the chromosphere. Such lines could therefore provide valuable diagnostics of departures of the chromospheric plasma from mean models and thereby place constraints upon heating mechanisms, once accurate atomic data become available.