We explore the extent to which current titanium oxide (TiO) line data and M dwarf model atmospheres can be used to reproduce an R = 120,000 optical spectrum of the relatively inactive star Gliese 725B (M3.5 V). We find that tabulated TiO wavelengths have errors large enough to complicate line identification, especially for transitions involving higher vibrational states. We determine empirical wavelength corrections for 12 strong γ-bands near 6680 and 7090 ?. For the sequence of orbital quantum numbers, J, within any one of these bands, our observations confirm the predicted line spacing, thereby validating the rotational constants for low vibrational levels. However, the predicted wavelengths have zero-point errors that differ for each overlapping band. Next, we compare observed and synthetic spectra near 8463 ?, where an Q3 0-0 band head is expected, demonstrating that the electronic oscillator strength of 0.014 advocated by J?rgensen is too large by at least a factor of 5. This has a minor effect on the structure of theoretical model atmospheres. Using our empirically corrected TiO wavelengths, we compute a grid of synthetic spectra for Allard & Hauschildt models spanning a range in effective temperature (Teff), surface gravity (log g), and metallicity ([M/H]). Interpolating in this grid of synthetic spectra, we simultaneously fit observations of the TiO band head region near 7088 ? and five Ti I and Fe I lines near 8683 ?. For Gl 725B, we find Teff = 3170 ± 71 K, log g = 4.77 ± 0.14, [M/H] = -0.92 ± 0.07, and vmac = 1.1 ± 0.7 km s-1. We show that by using both atomic and molecular lines as constraints, systematic uncertainties in derived stellar parameters can be reduced. These parameters are consistent with published values obtained by other means, but more stringent tests would be useful. In the Appendix, we tabulate wavelengths, identifications, relative line strengths, and other properties of the strongest band heads in the α, β, γ, γ', δ, , and electronic systems of TiO.
展开▼
