A high spectral resolution VLBI study of the 12 GHz methanol masers in W3(OH): Their submilliarcsecond structure and clues on saturation

This paper reports on the results of high velocity resolution (0.02 km s(-1)) VLBA observations of the 12 GHz methanol masers toward the ultracompact (UC) H II region W3(OH). About 50 maser features are detected, and their emission is resolved both in velocity and in space (with an angular resolution of approximate to1 mas). The maser feature spectra are very well reproduced by single Gaussian profiles, and a narrow distribution for the FWHM line widths is found from 0.14 to 0.32 km s(-1). The measured brightness temperatures distribute over the range 10(10) to 2 x 10(12) K. By performing elliptical Gaussian fits, the spot parameters ( peak position, intensity, and FWHM sizes) in the high velocity resolution maps were derived. The highest brightness temperatures occur in correspondence with the smallest spots, with FWHM size less than or equal to4 AU. For all the maser features, it is found that the channel peak positions vary smoothly with the velocity, delineating straight or arclike distributions projected onto the sky. In most cases, the channel (line-of-sight) velocities vary linearly with the peak positional shifts, with measured gradients in the range 0.05-0.65 km s(-1) mas(-1) (0.02-0.30 km s(-1) AU(-1)). Most of the maser spot spectra have very small deviations from a Gaussian profile, less than or equal to2 x 10(-3), and a trend for decreasing deviations with increasing intensities is noted. In most cases, the kurtosis of the spectral profiles is positive, with the most accurate values concentrated in the range 0.1-0.5. Comparison of these results with recent calculations of the deviations from a Gaussian pro le that occur across the line pro le of a linear maser ( as a result of the amplification process) suggests that the 12 GHz CH3OH masers are below the threshold, I-0, at which line rebroadening occurs. However, the model predicts deviations from a Gaussian pro le significantly smaller than the observed values, and this discrepancy needs to be investigated. In particular, the calculations of the deviations from a Gaussian pro le should take into account the effects of a velocity gradient through the maser features ( evidenced by our observations) that will influence the spectral pro le. The maximum output given by the pump rate of saturated masers may account for the observational result that the highest intensities are found in correspondence with the smallest sky-projected spot areas.