The absorption signatures of dwarf galaxies:: The z=1.04 multicloud weak MgII absorber toward PG 1634+706

We analyze high-resolution spectra of a multicloud weak [defined as W(r)(Mg II) < 0.3 angstrom] absorbing system along the line of sight to PG 1634+706. This system gives rise to a partial Lyman limit break and absorption in Mg II, Si II, C II, Si III, Si IV, C IV, and O VI. The lower ionization transitions arise in two kinematic subsystems with a separation of similar or equal to 150 km s(-1). Each subsystem is resolved into several narrow components having Doppler widths of 3-10 km s(-1). For both subsystems, the O VI absorption arises in a separate, higher ionization phase, in regions dominated by bulk motions in the range of 30-40 km s(-1). The two O VI absorption profiles are kinematically offset by similar or equal to 50 km s(-1) with respect to each of the two lower ionization subsystems. In the stronger subsystem, the Si III absorption is strong, with a distinctive, smooth profile shape, and may partially arise in shock-heated gas. Moreover, the kinematic substructure of Si IV traces that of the lower ionization Mg II, but may be offset by similar or equal to 3 km s(-1). On the basis of photoionization models constrained by the partial Lyman limit break, we infer a low metallicity of similar to 0.03 solar for the low-ionization gas in both subsystems. The broader O VI phases have a somewhat higher metallicity, and they are consistent with photoionization; the profiles are not broad enough to imply production of O VI through collisional ionization. Various models, including outer disks, dwarf galaxies, and superwinds, are discussed to account for the phase structure, metallicity, and kinematics of this absorption system. We favor an interpretation in which the two subsystems are produced by condensed clouds far out in the opposite extremes of a multilayer dwarf galaxy superwind.