Does evaporate not only Black Holes but also neutron stars and white dwarfs?!

The really interesting part is that the closer in towards the central object you get the more of these particles are produced, but they have a relatively low chance of escape to great distance. The further you go from the object the fewer are created, but these have a greater chance of long range detection. This implies that there is a radius for which the detected particle flux ( being a product of the number that are created times their chance of escape ) is at a maximum, which is what FIG 2 illustrates. Note they are using units where the speed of light is unity.

There is another most interesting feature. If one gradually increases the central mass ( maybe by accretion of material ) it could transition from a neutron star say, very dense but no event horizon, to a black hole with an event horizon. This you know, but this paper implies a continuity of behaviour of escaped particles ie. there's no sudden production of these particles because an event horizon was formed. You just need a steep enough energy/potential gradient - in this case a substantial change in the gravitational field strength - for this to happen. So Stephen Hawking's work has been expanded.

Furthermore the event horizon* is a classical ( non-quantum ) idea. I think this paper implies the horizon at an exact radius has been replaced by a 'zone' near the classical horizon ie. of some thickness in the radial direction, where this type of radiation is produced. 

When are these particles detected ? Well, the deeper you are in the well the more time slows down compared to an unaccelerated distant observer eg. us. Put another way the closer in the pair production occurs the ( possibly very much ) later the distant detection can occur as it 'climbs out' of the gravity well. For the limiting black hole case this time tends to infinity. Of course an observer could go closer in towards the horizon/zone to see what the particles are up to, rather than await distantly for them to turn up. You could then come back up the well to report on your findings, but beware : that will be a long time in the future and no one may be about to examine your findings !

Finally this tantalising comment in the Discussion section : "In principle, the pair production process should also continue inside the event horizon, but this would not be observable outside. The process would diverge at r = 0 and the associated backreaction** could potentially tame the curvature singularity in the center. The weak-field assumptions of our work, however, would not extend to this extrapolation". So no infinite density at radius zero because of quantum effects. Outstanding.

Cheers, Mike. 

* All bodies have a Schwarzschild radius. I have one and so do you. If the body lies completely within it's own Schwarzschild radius then you have a black hole.

** ( my asterisks ) Backreaction means the change in the environment of a particle because the particle has a non-zero mass/charge/etc. This change in the environment then affects the particle differently than the case where no backreaction is considered. Here backreaction implies a large ( but still finite ) radius zero density rather than infinite density. Having infinite density is a polite way of saying your model is broken.