ALBERT

Description: Ongoing studies with XMM–Newton have shown that powerful accretion disc winds, as revealed through highly ionized Fe K-shell absorption at E  ≥ 6.7 keV, are present in a significant fraction of active galactic nuclei (AGNs) in the local Universe (Tombesi et al. 2010a ). In Gofford et al., we analysed a sample of 51 Suzaku -observed AGNs and independently detected Fe K absorption in ~40 per cent of the sample, and we measured the properties of the absorbing gas. In this work, we build upon these results to consider the properties of the associated wind. On average, the fast winds ( v w  〉 0.01 c ) are located 〈 r 〉 ~ 10 15–18  cm (typically ~10 2–4 r s ) from their black hole, their mass outflow rates are of the order of $\langle \skew{3}\dot{M}_{\rm w}\rangle \sim 0.01\hbox{--}1$  M  yr –1 or ${\sim }(0.01\hbox{--}1)\skew{3}\dot{M}_{\rm Edd}$ and kinetic power is constrained to 〈 L w 〉 ~ 10 43–45  erg s –1 , equivalent to ~(0.1–10 per cent) L Edd . We find a fundamental correlation between the source bolometric luminosity and the wind velocity, with $v_{\rm w} \propto L_{\rm bol}^{\alpha }$ and $\alpha =0.4^{+0.3}_{-0.2}$ (90 per cent confidence), which indicates that more luminous AGN tend to harbour faster Fe K winds. The mass outflow rate $\skew{3}\dot{M}_{\rm w}$ , kinetic power L w and momentum flux $\dot{p}_{\rm w}$ of the winds are also consequently correlated with L bol , such that more massive and more energetic winds are present in more luminous AGN. We investigate these properties in the framework of a continuum-driven wind, showing that the observed relationships are broadly consistent with a wind being accelerated by continuum-scattering. We find that, globally, a significant fraction (~85 per cent) of the sample can plausibly exceed the L w / L bol  ~ 0.5 per cent threshold thought necessary for feedback, while 45 per cent may also exceed the less conservative ~5 per cent of L bol threshold as well. This suggests that the winds may be energetically significant for AGN–host-galaxy feedback processes.