Ionization sources and physical conditions in the diffuse ionized gas halos of four edge-on galaxies

Deep long-slit spectra of the diffuse ionized gas halos of the edge-on spiral galaxies NGC 4302 and UGC 10288 are presented. Emission lines are detected up to about z = 2 kpc in NGC 4302 and to nearly z = 3 kpc on the north side of UGC 10288. For both galaxies, the line ratios [N II]/H alpha and [S II]/H alpha increase with z in accordance with dilute photoionization models. Runs of [S II]/[N II] and for UGC 10288 the run of [O III]/H alpha, however, are not explained by the models. Scale height determinations of their diffuse ionized gas (DIG) halos are generally lower than those of galaxies with more prominent extraplanar DIG features. These data, along with previously presented data for NGC 5775 and NGC 891, are used to address the issue of how DIG halos are energized. Composite photoionization/shock models are generally better at explaining runs of line ratios in these galaxies than photoionization models alone. Models of line ratios in NGC 5775 require a greater contribution from shocks for filamentary regions than for nonfilamentary regions to explain the run of [O III]/H alpha. In either case, the [S II]/[N II] ratio is not well Dt by the models. Composite models for UGC 10288 are successful at reproducing the run of [S II]/[N II] for all but the the highest values of [N II]/H alpha; however, the run of [O III]/H alpha versus [N II]/H alpha does not show any discernible trend, making it difficult to determine whether or not shocks contribute to the layer's maintenance. We also examine whether the data can be explained simply by an increase in temperature with z in a pure photoionization model without a secondary source of ionization. Runs of [S II]/H alpha, [N II]/H alpha, and [S II]/[N II] in each of the four galaxies are consistent with such an increase. However, the runs of [O III]/H alpha versus z in NGC 5775 and UGC 10288 require unusually high-ionization fractions of O++ that cannot be explained without invoking a secondary ionization source or at the very least a much higher temperature for the [O III]-emitting component than for the [S II]- and [N II]-emitting component. An increase in temperature with z is generally more successful at explaining the [O III]/H alpha run in NGC 891, with the ionization fraction of O++ remaining relatively low and nearly constant with z.