2013MNRAS.434..848R

Currently, the ability to produce black hole mass estimates using the Civ λ1549 line that are consistent with Hβ mass estimates is uncertain, due in large part to disagreement between velocity line width measurements for the two lines. This discrepancy between Hβ and Civ arises from the fact that both line profiles are treated the same way in single-epoch scaling relationships based on the assumption that the broad-line region is virialized, even though a non-virialized emission component is often significant in the Civ line and absent or weak in the Hβ line. Using quasi-simultaneous optical and ultraviolet spectra for a sample of 85 optically-bright quasars with redshifts in the range z = 0.03-1.4, we present a significant step along the path to rehabilitating the Civ line for black hole mass estimates. We show that the residuals of velocity line width between Civ and Hβ are significantly correlated with the peak flux ratio of Siiv+Oiv] λ1400 to Civ. Using this relationship, we develop a prescription for estimating black hole masses from the ultraviolet spectrum that agree better with Hβ-based masses than the traditional Civ masses. The scatter between Hβ and Civ masses is initially 0.43dex in our sample and is reduced to 0.33dex when using our prescription. The peak flux ratio of Siiv+Oiv] λ1400 to Civ is an ultraviolet indicator of the suite of spectral properties commonly known as `Eigenvector 1'; thus, the reduction in scatter between Civ and Hβ black hole masses is essentially due to removing an Eigenvector 1 bias in Civ-based masses.