Y. Paul Lee

We present a statistical analysis of the ultraviolet emission lines of cataclysmic variables (CVs) based on $\approx 430$ ultraviolet spectra of 20 sources extracted from the International Ultraviolet Explorer Uniform Low Dispersion Archive. These spectra are used to measure the emission line fluxes of N V, Si IV, C IV, and He II and to construct diagnostic flux ratio diagrams. We investigate the flux ratio parameter space populated by individual CVs and by various CV subclasses (e.g., AM Her stars, DQ Her stars, dwarf novae, nova-like variables). For most systems, these ratios are clustered within a range of $\sim 1$ decade for log Si IV/C IV $\approx -0.5$ and log He II/C IV $\approx -1.0$ and $\sim 1.5$ decades for log N V/C IV $\approx -0.25$. These ratios are compared to photoionization and collisional ionization models to constrain the excitation mechanism and the physical conditions of the line-emitting gas. We find that the collisional models do the poorest job of reproducing the data. The photoionization models reproduce the Si IV/C IV line ratios for some shapes of the ionizing spectrum, but the predicted N V/C IV line ratios are simultaneously too low by typically $\sim 0.5$ decades. Worse, for no parameters are any of the models able to reproduce the observed He II/C IV line ratios; this ratio is far too small in the collisional and scattering models and too large by typically $\sim 0.5$ decades in the photoionization models.Comment: LaTeX format, uses aaspp4.sty, 28 pages, 11 Postscript figures, accepted for publication in The Astrophysical Journal 10/16/9

We present a statistical analysis of the ultraviolet emission lines of cataclysmic variables (CVs) based on $\approx 430$ ultraviolet spectra of 20 sources extracted from the International Ultraviolet Explorer Uniform Low Dispersion Archive. These spectra are used to measure the emission line fluxes of N V, Si IV, C IV, and He II and to construct diagnostic flux ratio diagrams. We investigate the flux ratio parameter space populated by individual CVs and by various CV subclasses (e.g., AM Her stars, DQ Her stars, dwarf novae, nova-like variables). For most systems, these ratios are clustered within a range of $\sim 1$ decade for log Si IV/C IV $\approx -0.5$ and log He II/C IV $\approx -1.0$ and $\sim 1.5$ decades for log N V/C IV $\approx -0.25$. These ratios are compared to photoionization and collisional ionization models to constrain the excitation mechanism and the physical conditions of the line-emitting gas. We find that the collisional models do the poorest job of reproducing the data. The photoionization models reproduce the Si IV/C IV line ratios for some shapes of the ionizing spectrum, but the predicted N V/C IV line ratios are simultaneously too low by typically $\sim 0.5$ decades. Worse, for no parameters are any of the models able to reproduce the observed He II/C IV line ratios; this ratio is far too small in the collisional and scattering models and too large by typically $\sim 0.5$ decades in the photoionization models.