Calibration of nebular emission-line diagnostics. I. Stellar effective temperatures

We present a detailed comparison of optical H II region spectra to photoion ization models based on modern stellar atmosphere models. We examine both s patially resolved and integrated emission-line spectra of the H II regions DEM L323, DEM L243, DEM L199, and DEM L301 in the Large Magellanic Cloud. T he published spectral classifications of the dominant stars range from O7 t o WN3, and morphologies range from Stromgren sphere to shell structure. Two of the objects include SNR contamination. The overall agreement with the p redictions is generally within 0.2 dex for major diagnostic line ratios. An apparent pattern in the remaining discrepancies is that the predicted elec tron temperature is similar to 1000 K hotter than observed. [Ne III] intens ities are also slightly overpredicted, which may or may not be related. We model the shock emission for the SNR-contaminated objects and find excellen t agreement with the observations for composite shock and photoionized spec tra. DEM L301's emission apparently results from both shocks and density-bo unded photoionization. The existence of contaminating shocks can be difficu lt to ascertain in the spatially integrated spectra. Our analysis of the co mplex DEM L199 allows a nebular emission-line test of unprecedented detail for WR atmospheres. Surprisingly, we find no nebular He II lambda 4686 emis sion, despite the fact that both of the dominant WN3 stars should be hot en ough to fully ionize He I in their atmospheres. The nebular diagnostics are again in excellent agreement with the data, for stellar models not produci ng He+-ionizing photons. The optical diagnostics are furthermore quite inse nsitive to the ionizing energy distribution for these early WR stars. We co nfirm that the eta' emission-line parameter is not as useful as hoped for d etermining the ionizing stellar effective temperature, T*. Both empirically and theoretically, we find that it is insensitive for T* greater than or s imilar to 40 kK and that it also varies spatially. The shock-contaminated o bjects show that eta' will also yield a spuriously high T* in the presence of shocks. It is furthermore sensitive to shell morphology. We suggest [Ne III]/H beta as an additional probe of T*. Although it is abundance dependen t, [Ne III]/H beta has higher sensitivity to T*, is independent of morpholo gy, and is insensitive to shocks in our objects. These observations should be useful data points for a first empirical calibration of nebular diagnost ics of T*, which we attempt for LMC metallicity.