Statistical analysis of Al III and C III] emission lines as virial black hole mass estimators in quasars

Aims. We test the usefulness of the intermediate ionisation lines Al III lambda 1860 and C III]lambda 1909 as reliable virial mass estimators for quasars. Methods. We identified a sample of 309 quasars from the Sloan Digital Sky Survey Data Release 16 (SDSS DR16) in the redshift range 1.2 <= z <= 1.4 that have [O II]lambda 3728 recorded on the same spectrum as Al III lambda 1860, Si III]lambda 1892, and C III]lambda 1909. We set the systemic quasar redshift using careful measurements of [O II]lambda 3728. We then classified the sources as Population A (Pop. A), extreme Population A (Pop. xA), and Population B (Pop. B), and analysed the 1900 A blend using multi-component models to look for systematic line shifts of Al III lambda 1860 and C III]lambda 1909 along the quasar main sequence. Results. We do not find significant shifts of the Al III lambda 1860 line peak in Pop. B or in the vast majority of Pop. A. For Pop. xA, a small median blueshift of -250 km s(-1) was observed, motivating a decomposition of the Al III line profile into a virialised component centred at rest frame and a blueshifted component for an outflow emission. For Pop. B objects, we proved the empirical necessity to fit a redshifted very broad component, clearly seen in C III], and analysed the physical implications on a Pop. B composite spectrum using CLOUDY simulations. We find consistent black hole mass estimations using Al III and C III] as virial estimators for the bulk of Pop. A. Conclusions. Al III (and even C III]) is a reliable virial black hole mass estimator for Pop. A and Pop. B objects. The Pop. xA sources deserve special attention due to the significant blueshifted excess observed in the line profile of Al III, although it is not as large as the excess observed in C IV lambda 1549.