Size and kinematics of the low-ionization broad emission line region from microlensing-induced line profile distortions in gravitationally lensed quasars

Microlensing-induced distortions of broad emission line profiles observed in the spectra of gravitationally lensed quasars can be used to probe the size, geometry, and kinematics of the broad-line region (BLR). To this end, single-epoch Mg <ns0:sc>II</ns0:sc> or Hα line profile distortions observed in five gravitationally lensed quasars, J1131-1231, J1226-0006, J1355-2257, J1339+1310, and HE0435-1223, have been compared with simulated ones. The simulations are based on three BLR models, a Keplerian disk (KD), an equatorial wind (EW), and a polar wind (PW), with different sizes, inclinations, and emissivities. The models that best reproduce the observed line profile distortions were identified using a Bayesian probabilistic approach. We find that the wide variety of observed line profile distortions can be reproduced with microlensing-induced distortions of line profiles generated by our BLR models. For J1131, J1226, and HE0435, the most likely model for the Mg <ns0:sc>II</ns0:sc> and Hα BLRs is either KD or EW, depending on the orientation of the magnification map with respect to the BLR axis. This shows that the line profile distortions depend on the position and orientation of the isovelocity parts of the BLR with respect to the caustic network, and not only on their different effective sizes. For the Mg <ns0:sc>II</ns0:sc> BLRs in J1355 and J1339, the EW model is preferred. For all objects, the PW model has a lower probability. As for the high-ionization C <ns0:sc>IV</ns0:sc> BLR, we conclude that disk geometries with kinematics dominated by either Keplerian rotation or equatorial outflow best reproduce the microlensing effects on the low-ionization Mg <ns0:sc>II</ns0:sc> and Hα emission line profiles. The half-light radii of the Mg <ns0:sc>II</ns0:sc> and Hα BLRs are measured in the range of 3 to 25 light-days. We also confirm that the size of the region emitting the low-ionization lines is larger than the region emitting the high-ionization lines, with a factor of four measured between the sizes of the Mg <ns0:sc>II</ns0:sc> and C <ns0:sc>IV</ns0:sc> emitting regions in J1339. Unexpectedly, the microlensing BLR radii of the Mg <ns0:sc>II</ns0:sc> and Hα BLRs are found to be systematically below the radius-luminosity (R − L) relations derived from reverberation mapping, confirming that the intrinsic dispersion of the BLR radii with respect to the R − L relations is large, but also revealing a selection bias that affects microlensing-based BLR size measurements. This bias arises from the fact that, if microlensing-induced line profile distortions are observed in a lensed quasar, the BLR radius should be comparable to the microlensing Einstein radius, which varies only weakly with typical lens and source redshifts.