Symbiotics are interacting binary stars composed of an evolved red giant and a hot companion star. The hot companion is usually a white dwarf, sometimes a main sequence star, and rarely a neutron star. The hot component generates energy by accreting material lost by the red giant. Starting with the work for my Ph. D. thesis, I have been interested in the physical structure of symbiotic stars (see Proga, Kenyon , Raymond & Mikolajewska 1996 and Proga et at. 1998 ). In particular, I have worked of illumination effects in those stars. As in many other close binary systems, symbiotic stars have light curve that display the ``reflection effect'' in which the hot component star heats up the facing hemisphere of the red giant. The higher effective temperature of the heated hemisphere produces a characteristic sinusoidal light variation. Aside from this simple photospheric display, illumination can have a significant impact on spectroscopic analyses. For example, radiation from a hot secondary can distort absorption line profiles and thus cause errors in effective temperature or gravity estimates and in radial velocity curve used for orbits. In some cases, this extra radiation might cause the heated atmosphere to expand. Extra mass-loss from the extended atmospheres of illuminated red dwarfs is important in low mass X-ray binary systems (LMXB's), CVs, some symbiotic stars, because it can significantly affect the evolution of the binary system. Illumination effects can also be very important in studies of extrasolar giant planets if the radiation of their parent star is intense.

To tackle the problem of illumination I have constructed a non-LTE photoionization code which handles both low and high ionization state conditions and calculates a spectrum for a wide wavelength range. I included many opacity sources and forbidden line subroutines that are important in a red giant atmosphere. The model assumes radiative, and statistical equilibria for the red giant photosphere or wind and solves the radiative transfer equation with a local escape probability method. I computed non-LTE level populations for a variety of ions and predict the variation of emission line fluxes as functions of the temperature and luminosity of the hot component. My models generally match observations of the symbiotic stars EG And and AG Peg. The optically thick cross-section of the red giant wind as viewed from the hot component is a crucial parameter in these models. Winds with cross-sections of 2-3 red giant radii reproduce the observed fluxes. My models favor winds with acceleration regions that either lie far from the red giant photosphere or extend for 2-3 red giant radii.