We cross-correlate a cosmic microwave background (CMB) lensing map with the projected space densities of quasars to measure the bias and halo masses of a quasar sample split into obscured and unobscured populations, the first application of this meth
od to distinct quasar subclasses. Several recent studies of the angular clustering of obscured quasars have shown that these objects likely reside in higher-mass halos compared to their unobscured counterparts. This has important implications for models of the structure and geometry of quasars, their role in growing supermassive black holes, and mutual quasar/host galaxy evolution. However, the magnitude and significance of this difference has varied from study to study. Using data from planck, wise, and SDSS, we follow up on these results using the independent method of CMB lensing cross-correlations. The region and sample are identical to that used for recent angular clustering measurements, allowing for a direct comparison of the CMB-lensing and angular clustering methods. At $z sim 1$, we find that the bias of obscured quasars is $b_q = 2.57 pm 0.24$, while that of unobscured quasars is $b_q = 1.89 pm 0.19$. This corresponds to halo masses of $log (M_h / M_{odot} h^{-1}) = 13.24_{-0.15}^{+0.14}$ (obscured) and $log (M_h / M_{odot} h^{-1}) = 12.71_{-0.13}^{+0.15}$ (unobscured). These results agree well with with those from angular clustering (well within $1sigma$), and confirm that obscured quasars reside in host halos $sim$3 times as massive as halos hosting unobscured quasars. This implies that quasars spend a significant portion of their lifetime in an obscured state, possibly more than one half of the entire active phase.
Recent studies of luminous infrared-selected active galactic nuclei (AGN) suggest that the reddest, most obscured objects display a higher angular clustering amplitude, and thus reside in higher-mass dark matter halos. This is a direct contradiction
to the prediction of the simplest unification-by-orientation models of AGN and quasars. However, clustering measurements depend strongly on the mask that removes low-quality data and describes the sky and selection function. We find that applying a robust, conservative mask to WISE-selected quasars yields a weaker but still significant difference in the bias between obscured and unobscured quasars. These findings are consistent with results from previous Spitzer surveys, and removes any scale dependence of the bias. For obscured quasars with $langle z rangle = 0.99$ we measure a bias of $b_q = 2.67 pm 0.16$, corresponding to a halo mass of $log (M_h / M_{odot} h^{-1}) = 13.3 pm 0.1$, while for unobscured sources with $langle z rangle = 1.04$ we find $b_q = 2.04 pm 0.17$ with a halo mass $log (M_h / M_{odot} h^{-1} )= 12.8 pm 0.1$. This improved measurement indicates that WISE-selected obscured quasars reside in halos only a few times more massive than the halos of their unobscured counterparts, a reduction in the factor of $sim$10 larger halo mass as has been previously reported using WISE-selected samples. Additionally, an abundance matching analysis yields lifetimes for both obscured and unobscured quasar phases on the order of a few 100 Myr ($sim$ 1% of the Hubble time) --- however, the obscured phase lasts roughly twice as long, in tension with many model predictions.
