Dust-induced systematic errors in ultraviolet-derived star formation rates

Rest-frame far-ultraviolet (FUV) luminosities form the backbone of our understanding of star formation ( SF) at all cosmic epochs. These luminosities are typically corrected for dust by assuming that the tight relationship between the UV spectral slopes (beta) and the FUV attenuations (A(FUV)) of starburst galaxies applies to all star-forming galaxies. Data from seven independent UV experiments demonstrate that quiescent, normal star-forming galaxies deviate substantially from the starburst galaxy beta-A(FUV) correlation in the sense that normal galaxies are redder than starbursts. Spatially resolved data for the Large Magellanic Cloud suggest that dust geometry and properties, coupled with a small contribution from older stellar populations, cause deviations from the starburst galaxy beta-A(FUV) correlation. Folding in data for starbursts and ultraluminous infrared galaxies, it is clear that neither rest-frame UV/optical colors nor UV/Halpha colors help significantly in constraining the UV attenuation. These results argue that the estimation of SF rates from rest-frame UV and optical data alone is subject to large (factors of at least a few) systematic uncertainties because of dust, which cannot be reliably corrected for using only UV/optical diagnostics.