The Astrophysical Journal Supplement Series, 92:189-218, 1994 May .189B © 1994. The American Astronomical Society. All rights reserved. Printed in U.S.A. .92. 94ApJS. VLA POSITIONS AND DISTRIBUTIONS OF H20 MASERS ASSOCIATED WITH 15 MIRA 19 AND SEMIREGULAR VARIABLES P. F. Bowers1 SEA, Inc., 1401 McCormick Drive, Lando ver, MD 20785 AND K. J. Johnston Remote Sensing Division, Naval Research Laboratory, Washington, DC 20375 Received 1993 May 27; accepted 1993 October 11 ABSTRACT Results of VLA observations of 22 GHz H20 masers associated with 15 Mira and semiregular variables are reported. The combination of angular resolution (^70 mas), spectral resolution (0.3 km s_1), rms sensitivity -1 (±35 mJy beam ), and hour-angle coverage is the best yet obtained for H20 masers from a significant sample of stars. Comparison of accurate optical positions of the stars with estimates from the masers yields total differences typically <0''15—an improvement over previous comparisons by a factor of about 2. The homogeneous, high-quality data set provides several new insights into the time-varying structure of cir- cumstellar H20 profiles and angular distributions. Profile structures can exhibit dramatic, short-term (less than 1 yr) changes but also may exhibit stability of features over timescales up to 15 yr after allowance for phase-depen- -7 -6 -1 dent effects. For this sample of stars (rates of mass loss from about 10 to 10 M0 yr ), the average profile -1 shape shows strongest emission generally within ±2 km s of the stellar velocity F0, regardless of the stellar light-curve phase 0, but there is a higher probability of detecting multiple (particularly blueshifted) features and a -1 larger velocity range (V0±9 km s ) at 0.2 < 0 < 0.4, when the integrated H20 luminosity is largest. Based on observations of different stars and on repeated VLA observations of specific stars, the shell size may vary typically by about a factor <2 during the light cycle, with statistically largest radius ((p) ^ 25 AU) at 0.2 < 0 < 0.4. The morphology of the angular distribution can change strongly with 0, but similar morphologies are sometimes observed at comparable phases of different light curves, indicating that short-term changes (less than a few years) are likely to reflect changes in the pumping conditions. The shell radii range from >5 to about 50 AU, and the region of maximal H20 intensity at a given epoch appears to occur over a radial interval less than or equal to one-third of the shell radius. The angular distributions at a given epoch frequently are elongated and sometimes are strongly asymmetric relative to the estimated stellar position. For three stars (RT Vir, U Her, UX Cyg), thin, asymmetric loop structures are found. For U Ori, strongest components near V0 are confined northeast and southwest of the star, suggestive of axial symmetry. Comparison of OH and H20 distributions shows points of similarity for several stars. Plots of shell radius as a function of radial velocity exhibit two forms of curves: singly peaked curves, where the radius increases rapidly within a few kilometers per second of F0, and doubly peaked curves, where the maximum radius occurs at velocities symmetric to but displaced from V0. The latter type of curve also tends to be double-valued (X-shaped) with two distinct map maxima at a given velocity over significant velocity intervals. There are indications of some combination of nonuniformities in the density distribution, deviations from sphericity, and anisotropies in the velocity field, but it is difficult to disentangle the effects of these phenomena and achieve unique interpretation. A plausible model is clumps or filaments distributed at radii which vary with direction from the star. From the angular distributions and velocity ranges, there is good evidence that at least some parcels of gas are accelerated to the terminal outflow velocity at p > 20 AU and that there is a component of outflow at p ^ 10-15 AU; some gas probably is fully accelerated at radii less than 10 AU. To illustrate the problems of interpretation, the well-known case of VX Sgr is reexamined. The data do not support a simple model of increasing outflow velocity with increasing maser shell radius. It is proposed that the outflow is weakly bipolar. Subject headings: ISM: jets and outflows — masers — stars: variables: other (long-period, semiregular) 1. INTRODUCTION ing in (2) a spherical shell with (3) an isotropic velocity field. Data pertaining to circumstellar outflows of long-period vari- For masers, amplification occurs as long as the fine width is able stars typically are interpreted in the context of an expand- smaller than the line-of-sight velocity gradient. If the outflow ing-shell model. In its simplest form, this model assumes ( 1 ) a velocity Vp is constant throughout the shell and if the shell is globally averaged, uniform density distribution of gas expand- not extremely thin (Bowers 1993a), a doubly peaked profile is formed with the peaks occurring at K = L0 ± Lp, where F0 is 1 Work performed at NRL under contract number N00014-89-C-2398. the stellar radial velocity. This model predicts circular rings of 189 © American Astronomical Society • Provided by the NASA Astrophysics Data System 190 .189B BOWERS & JOHNSTON Vol. 92 emission centered about the star with an angular radius 6 at Bowers, Claussen, & Johnston 1993; § 5 below), placing them .92. velocity V which is related to the shell radius 6S at V0 by in a complex but interesting region of the envelope. At these distances (^5-30 stellar radii) some dynamical models of 2 1/2 e = es{\-[(v- Vq)/v^[ } . (i) mass loss predict that the gas should essentially have attained 94ApJS. its terminal outflow velocity (Bowen 1988), while other mod- 19 The standard model usually provides a good, first-order ap- els allow for continued acceleration (Pijpers & Habing 1989). proximation to the large (p > 1000 AU) 1612 MHz OH maser The velocity fields also may be influenced by outwardly propa- shells of asymptotic giant branch (AGB) OH/IR stars with gating shock waves (Rudnitskii & Chuprikov 1990; Fleischer, -5 -1 rates of mass loss Af > 10 M0 yr , indicating that the OH is Gauger, & Sedlmayr 1991), by anisotropies in the outflow essentially distributed in all directions relative to the star and which introduce latitudinal velocity gradients and possibly that the expansion velocity derived from the half-velocity sepa- nonradial motions (e.g., Collison & Fix 1992; Pascoh 1992), ration of the peaks (or half-profile width) is a good measure of or by incomplete momentum coupling between the dust and the terminal outflow velocity (Bowers 1993b). The brightness gas for stars with low rates of mass loss ( Justtanont & Tielens distributions are always clumpy, however, and sometimes 1992; MacGregor & Stencel 1992; Netzer & Elitzur 1993). asymmetric relative to the stellar position (Bowers & Johnston The profile shapes and angular distributions of water masers 1990) , indicating departures from the assumptionsare known to of change the stan- strongly with time, complicating inter- dard model. Evidence for asymmetries also is seen from infra- pretation of single-epoch measurements (Johnston, Spencer, red polarization measurements, suggesting clumpy and/or & Bowers 1985; Bowers et al. 1993). From comparison of aspherical dust distributions (Jones & Gehrz 1990). Theoreti- published profiles for Mira variables, Engels, Schmid-Burgk, & cal efforts demonstrate that asymmetries in dust or cold gas Walmsley (1988) conclude that individual spectral features distributions often must be fairly strong in order to be convinc- may be stable for a period of a few years. This is consistent with ingly demonstrated by the data (Efstathiou & Rowan-Robin- interferometric data for IK Tau (Bowers et al. 1993) and for son 1990; Collison & Fix 1991; Bowers 1991 ). RX Boo (Engels et al. 1993), each of which shows a relatively -7 1 5 For classical Mira variables ( 10 M0 yr“ <M< 10“ MQ stable angular distribution at epochs separated by about 1 year yr“1 ), the approximations of the standard outflow model gen- or less. Over a longer timescale (more than a few years), the erally are less adequate. Distortions from spherical symmetry angular distributions may show little correspondence. The are known to occur close to the stars (Karovska et al. 1991; H20 distribution published by Reid & Menten ( 1990) for W Haniff et al. 1992 ) and in the dust shells ( e.g., Johnson & Jones Hya is quite different from that obtained 5 years earlier by 1991) . Observations of gaseous distributionsBowers etwhich al. ( 1993)trace theat the same phase of the light curve. Be- envelope structures to about 1000-2000 AU also show many cause the H20 masers are located in the inner envelope, these complexities. Sharply peaked or Gaussian-like profiles of (u = results may suggest that there are changes in the density or 0) SiO and CO have been found from a number of Mira vari- velocity fields over a timescale of a few years. However, dra- ables and have been interpreted to arise from outflows with matic variations of the profile structures can sometimes occur complex structure or kinematics (Bowers 1990; Margulis et al. over a much shorter timescale ( a few months), possibly reflect- 1990) or from an extended inner envelope (p ^ 100 AU) in ing changes of the temperature structure and pumping condi- which grains are not entirely formed and the wind has not tions in the maser region (Lewis & Engels 1991 ). attained its terminal velocity (Bujarrabal et al. 1986; Bujarra- Possible evidence for a component of outflow in the H20 bal, Gómez-González, & Planesas 1989; Lucas et al. 1992). maser shell has been found for a few cases (IK Tau, RT Vir, For cases where the thermal distributions are spatially re- RX Boo) based on a larger shell radius near the stellar velocity solved, there is sometimes evidence for asymmetric outflow than at velocities near the edges of the profiles (Lane et al.