Dust grains are highly efficient radiators and hence can thermally decouple
from the surrounding hot gas. Under the conditions envisaged here, there
are two grain destruction mechanisms that must be considered. First, if
the grains themselves become too hot, they will rapidly sublime. The grain
temperature will be set by the thermal equilibrium between the AGN
radiation incident on a given grain and the thermal radiation emitted by
that grain (e.g. see Barvainis 1987). For MCG-6-30-15, the
sublimation radius (i.e. the radius from the AGN within which dust
grains become so hot that they sublime) is 
. Thus, any
dust grains within the inner warm absorber would be rapidly sublimed by the
intense radiation field. Dust in the outer warm absorber would not be
subject to significant sublimation.
The second dust destruction mechanism that we must consider is thermal
sputtering. If we make the standard assumption that the (outer) warm
absorber is photoionized, then photoionization models suggest that the gas
temperature is only 
and thermal sputtering is
negligible. If, instead, we suppose that the outer warm absorber is purely
collisionally-ionized, gas temperatures of 
are required in
order to achieve the observed ionization states (Shull & van Steenberg
1982). From the expressions of Burke & Silk (1974), the thermal
sputtering timescale for this temperature is
where n is the electron number density in the gas. Suppose that r is
the distance of the outer warm absorber from the central engine, and L is
the (ionizing) luminosity of the central engine. Furthermore, define
to be the ionization parameter of a
photoionized plasma in which oxygen is ionized to the same degree as seen
in the outer warm absorber of MCG-6-30-15. Given our (temporary)
hypothesis that the plasma is collisionally-ionized, the density must
satisfy
or else photoionization would dominate the ionization state. Evaluating the sputtering timescale for the parameters of MCG-6-30-15 gives
For comparison, the flow timescale of the outer warm absorber is
where we have adopted a typical value of 
for the
velocity of the outer warm absorber, as indicated by UV absorption line
studies of other AGN (Mathur, Elvis & Wilkes 1995). It can be seen that
the flow timescale of the warm absorber always exceeds the sputtering
timescale unless 
. If the outer warm absorber was
situated at such a large distance, then either we would have to be viewing
the AGN along a very special line of sight, or else the mass, M, and
kinetic energy, 
, associated with the outflow would both be
huge. From the expressions of Reynolds & Fabian (1995), and assuming a
global covering fraction of 
, we get 
and
. The initial acceleration of this
material would be extremely problematic to understand. We consider this
possibility to be unphysical. Thus, in the absence of a viable,
collisionally-ionized model, the observation of a dusty warm absorber
may be taken as further evidence that photoionization dominates the state
of this plasma.
Whilst dust can survive in warm photoionized gas, it is extremely difficult to form dust in such an environment: the grains could never assemble at such temperatures. Furthermore, a comparison of the column density of the warm absorber with the cold column expected to be associated with the reddening reveals that the warm-gas/dust ratio in the warm absorber must be very similar to the cold-gas/dust ratio in our Galaxy. These two facts taken together suggest that the warm material originates from dusty cold material, possibility via radiative heating, and that a substantial fraction of the dust survives the heating process. The putative dusty molecular torus of Seyfert unification schemes might be a possible progenitor of such a radiatively-driven, warm, dusty outflow.