Calibration and Limitations of the Mg II line-based Black Hole Masses

TL;DR

This study calibrates UV and optical single-epoch black hole mass estimators using Mg II and Hβ lines, revealing systematic biases and providing correction terms, but also highlighting intrinsic uncertainties in UV-based mass estimates.

Contribution

It offers new calibration relations for Mg II-based black hole mass estimates and quantifies their limitations and biases compared to Hβ-based masses.

Findings

Mg II-based masses have >0.2 dex intrinsic scatter compared to Hβ-based masses.

A correction term based on spectral slope improves Mg II mass estimates.

UV mass estimators show no significant bias with Eddington ratio or Fe II strength.

Abstract

We present the single-epoch black hole mass (M$_{\rm BH}$) calibrations based on the rest-frame UV and optical measurements of Mg II 2798\AA\ and H$\beta$ 4861\AA\ lines and AGN continuum, using a sample of 52 moderate-luminosity AGNs at z$\sim$0.4 and z$\sim$0.6 with high-quality Keck spectra. We combine this sample with a large number of luminous AGNs from the Sloan Digital Sky Survey to increase the dynamic range for a better comparison of UV and optical velocity and luminosity measurements. With respect to the reference M$_{\rm BH}$ based on the line dispersion of H$\beta$ and continuum luminosity at 5100\AA, we calibrate the UV and optical mass estimators, by determining the best-fit values of the coefficients in the mass equation. By investigating whether the UV estimators show systematic trend with Eddington ratio, FWHM of H$\beta$, the Fe II strength, and the UV/optical slope, we find no significant bias except for the slope. By fitting the systematic difference of Mg II-based and H$\beta$-based masses with the L$_{3000}$/L$_{5100}$ ratio, we provide a correction term as a function of the spectral index as $\Delta$C = 0.24 (1+$\alpha_{\lambda}$) + 0.17, which can be added to the Mg II-based mass estimators if the spectral slope can be well determined. The derived UV mass estimators typically show $>$$\sim$0.2 dex intrinsic scatter with respect to H$\beta$-based M$_{\rm BH}$, suggesting that the UV-based mass has an additional uncertainty of $\sim$0.2 dex, even if high quality rest-frame UV spectra are available.