The issue concerning semi-classical methods recently developed in deriving the conditions for Hawking radiation as tunneling, is revisited and applied also to rotating black hole solutions as well as to the extremal cases. It is noticed how the tunneling method fixes the temperature of extremal black hole to be zero, unlike the Euclidean regularity method that allows an arbitrary compactification period. A comparison with other approaches is presented. *

A local Hawking temperature is derived for any future outer trapping horizon in spherical symmetry, using a Hamilton–Jacobi variant of the Parikh–Wilczek tunneling method. It is given by a dynamical surface gravity as defined geometrically. The operational meaning of the temperature is that Kodama observers just outside the horizon measure an invariantly redshifted temperature, diverging at the horizon itself. In static, asymptotically flat cases, the Hawking temperature as usually defined by the Killing vector agrees in standard cases, but generally differs by a relative redshift factor between the horizon and infinity, this being the temperature measured by static observers at infinity. Likewise, the geometrical surface gravity reduces to the Newtonian surface gravity in the Newtonian limit, while the Killing definition instead reflects measurements at infinity. This may resolve a long-standing puzzle concerning the Hawking temperature for the extremal limit of the charged stringy black hole, namely that it is the local temperature which vanishes. In general, this confirms the quasi-stationary picture of black-hole evaporation in early stages. However, the geometrical surface gravity is generally not the surface gravity of a static black hole with the same parameters.

Previous work on dynamical black hole instability is further elucidated within the Hamilton–Jacobi method for horizon tunneling and the reconstruction of the classical action by means of the null expansion method. Everything is based on two natural requirements, namely that the tunneling rate is an observable and therefore it must be based on invariantly defined quantities, and that coordinate systems which do not cover the horizon should not be admitted. These simple observations can help to clarify some ambiguities, like the doubling of the temperature occurring in the static case when using singular coordinates and the role, if any, of the temporal contribution of the action to the emission rate. The formalism is also applied to FRW cosmological models, where it is observed that it predicts the positivity of the temperature naturally, without further assumptions on the sign of energy.

The instability against emission of massless particles by the trapping horizon of an evolving black hole is analyzed with the use of the Hamilton-Jacobi method. The method automatically selects one special expression for the surface gravity of a changing horizon. Indeed, the strength of the horizon singularity turns out to be governed by the surface gravity as was defined a decade ago by Hayward using Kodama's theory of spherically symmetric gravitational fields. The theory also applies to point masses embedded in an expanding universe, were the surface gravity is still related to KodamaHayward theory. As a bonus of the tunneling method, we gain the insight that the surface gravity still defines a temperature parameter as long as the evolution is sufficiently slow that the black hole pass through a sequence of quasi-equilibrium states.