HUBBLE-SPACE-TELESCOPE OBSERVATIONS OF PLANETARY-NEBULAE IN THE MAGELLANIC CLOUDS .3. ULTRAVIOLET SPECTROSCOPY USING THE FAINT OBJECT SPECTROGRAPH

Ultraviolet grating spectra (lambda lambda 1150-2300) of planetary neb ulae in the Magellanic Clouds, obtained with the Faint Object Spectrog raph (FOS) on the Hubble Space Telescope (HST), are used to identify a nd measure nebular emission lines and continua, and in one case to det ect the stellar wind of the central star. Eight nebulae are observed i n the Large Magellanic Cloud (LMC) and four are observed in the Small Magellanic Cloud (SMC). The HST observations are normalized to the pho tometric scale of ground-based, visible-light spectra (lambda lambda 3 400-8000) by means of FOS PRISM observations in the lambda lambda 1350 -5100 range. There is an unexplained, systematic discrepancy between t he average PRISM and ground fluxes of similar to 30%. Observed and der eddened UV emission-line intensities with respect to H beta are tabula ted. The published reddening determinations based on the Balmer decrem ent are used to deredden the data. In general, the reddening derived f rom the He II I(1640)/I(4686) emission-line ratio does not agree with that derived from the Balmer decrement. This discrepancy suggests that the two reddening determinations are truly not equal, or that the app lied reddening law deviates from the published average LMC or SMC curv es. The uncertainty in the relative PRISM to ground calibration also c ontributes to the discrepancy but is of less importance than the uncer tainty associated with the adopted reddening. The reddening is also de termined from the lambda 2200 interstellar absorption band, assuming t he published average Galactic and Magellanic Cloud reddening curves an d is found to agree with the reddening determined from the Balmer decr ement, in most cases. Otherwise, the reddening is always determined to be less than the Balmer decrement value. The exception to this rule i s SRIC-SMP6, which is the most reddened object in the sample. An alter native calibration for the HST data is achieved by assuming that the r eddening derived from the He II I(1640)/I(4686) ratio must equal the r eddening derived from the Balmer decrement. However, this procedure ca n only be applied to seven of the 12 objects in the sample. The altern ative UV line intensities are also tabulated. Spectra of six of the ne bulae are compared with archival IUE observations in the lambda lambda 1200-2000 range. There is satisfactory agreement between the FOS and IUE data for line intensities less than similar to H beta. For stronge r lines, the FOS appears to overestimate the measured line fluxes by s imilar to 30%, which may be due to the uncertainty associated with the PRISM to ground calibration: the effect is not seen in the smaller fl uxes because of larger measurement errors. The continuum fluxes agree to within 50%. Color temperatures, derived from the FOS continuum flux ratios, are uncertain by at least a factor of 2 and are not correlate d with previously determined Zanstra temperatures. Nebular densities, derived from O IV] and Si III] UV emission-line ratios, are presented. They are larger than densities derived from visible light [O II] and [S II] line ratios, consistent with the premise that the higher ioniza tion species are formed in denser regions, supposedly closer to the ce ntral star. Nebular temperatures are determined using the N+ [I(2139) + I(2143)]/[I(6548)+ I(6583)] ratio and are found to be in agreement w ith the previously published, solely optical N+ temperatures, when the PRISM to ground calibration is used. When the N II] fluxes are scaled with respect to the helium lines, the N+ temperatures reduce to simil ar to 11,000 K in all instances. Nebular temperatures implied by the O 2+ [I(1661) + I(1666)]/I(5007) ratio are consistent with those derived from the I(4363)/I(5007) ratio if the PRISM to ground calibration is used. The high O2+ electron temperatures for SMC-SMP28 and LMC-SMP87 i mply that photoionization is not the sole excitation process in the O2 + zone. However, if the UV line fluxes are rescaled such that the redd ening from the Balmer decrement and the He II lines agree, then the op tically derived temperatures for these two objects appear to be overes timated by similar to 10,000 K. Estimates of the C/O abundance ratio a re made for most objects in the sample using the C III] lambda lambda 1907, 1909, O III] lambda lambda 1661, 1666, and [O III] lambda 5007 e mission lines. The C/O ratio is anticorrelated with the previously pub lished N/O values, suggesting that the hot bottom burning process oper ated during the asymptotic giant branch phase. There are four type I o bjects in the sample, which have the lowest C/O values. The Si/C ratio is also estimated solely from ultraviolet lines. Silicon emission is only detected in the four type I objects. P Cygni-type line profiles a re detected in the spectrum for LMC-SMP76 at N v lambda lambda 1239, 1 243, O v lambda 1371, and C IV lambda lambda 1548, 1551. The edge velo cities derived from these lines are 1000-1500, 800, and 2000 km s(-1), respectively.