The ionization time derived by fitting with STNEI model is quite low. Even
considering the largest value allowed by the statistical confidence
regions, 
appears to be not larger than 10 yr cm
in 18 out
of 21 acceptable bins. The
calculations of Itoh (1979) show that the ionization balance is very
far from equilibrium near the shock front, but such a low range of
can be regarded as unusual. We stress that the value we find is
actually a weighted average of along the line of sight.
Moreover, applying the analysis of Masai (1994) to the case of the Vela
SNR, we expect 
yr cm (eq. 24 and Figure 5 of
Masai 1994) in the thin shell responsible of most X-ray emission.
There is still a chance that this could be due to the simplifications
introduced in our modeling: in fact, Itoh (1979) shows that the
population of some ions (e.g. O VI, O VII and some N and Ne) can remain
far below the equilibrium just behind the shock front, weakening the
strong O VII (0.57 keV) and O VIII (0.66 keV) lines and resulting in a
low values of if derived from soft X-ray spectra. Nevertheless,
a recent estimate of the ionization time in the shock region of the
Cygnus loop based on a ASCA observation (Myata et al. 1994) gives 
yr cm, a value that has generally been found also in
younger SNR's (Hughes & Singh, 1994). In particular Myata et al.
(1994) found that a cosmic abundance STNEI plasma emission model is not
sufficient to describe the spectrum, but underabundances of metal
(
of the cosmic values) should be taken into account.
Moreover, Myata et al. have also found that a model based on a thermal
bremsstrahlung plus 10 lines in the band 0.5-2.2 keV gives a
description of the ASCA SIS spectrum of the Cygnus Loop as good as an
optically thin STNEI plasma model.
For a better understanding of our results, taking also into account the
abovementioned recent ASCA results, we recall that our PSPC data can be
also fitted by a bremsstrahlung model and by a two temperature
Raymond-Smith model (Paper I). The NEI spectrum of an optically thin
plasma with 
yr cm has the line contributions
largely reduced, but not completely suppressed as in the case of the
thermal bremsstrahlung. Based on the present results and those of Paper
I, we conclude that more than one temperature and density component are
likely contributing to the X-ray emission of the Vela SNR shock region
we consider. The components are possibly both in NEI condition,
but the limited spectral resolution of the
ROSAT PSPC prevents us from testing this hypothesis.
presence of two NEI temperatures.
The single temperature STNEI conditions are excluded as a general
feature of the portion of the Vela SNR we considered, but they could be
valid from a statistical point of view if applied locally in 21 spatial
bins, mostly located in regions e1-e7 and f1-f7. n these
regions, the observed value of 
is mostly in the range
0-0.5. This derived ionization time is not realistic because we expect
to see plasma with higher (§3.2), and Figure 1
shows that the NEI emissivity peaks at 
. At 
the emissivity is more than 10 times less then the maximum
value and should not contribute significantly to the total emission.
Therefore, we have further investigated the issue of the reduced line
emission contribution. In particular, we note that if the plasma has
reduced metal abundances and we fit it with models having metal
abundances fixed to cosmic values (as our STNEI model) using moderate
spectrally resolved data, the result is an underestimate of the
ionization time. In fact, our model predicts a steady evolution of the
neutral plasma toward CIE conditions over long time scales. In the
early stages of this process, (
) the contribution
of line emission in the 0.1-2.4 keV band might be reduced, either
because lines from low ionization stages have energy < 0.1 keV, or
because the population of typical ROSAT band emitting ion species
(like O VII) is intrinsically very low. The latter situation could
arise as well in low abundance plasma in more advanced stages of
ionization (e.g. 
).
Recently, Vancura et al. (1994) have pointed out that the grains
present in the ISM could survive long in the post-shock regions of
SNR's under certain conditions of temperature and density (this is
referred to as ``the dusty shock model"). Since a non negligible
fraction of the metals is locked up in the grains, they would not
contribute to X-ray emission until they are released upon grain
destruction; the plasma in the post-shock region could then have
effectively reduced metal abundances. For instance, according to
Vancura et al. (1994), the fractions of Fe, Si and O (whose lines
mostly fall in the 0.1-2.4 keV band) locked up in grains in the ISM are
respectively of 0.98, 0.98 and 0.2, and grain destruction doesn't
change significantly the fractions before 
yr for a shock
speed of 400 km/sec, similar to the shock speed we have derived.
During this time, the shock has moved across a region of 3 pc, spanning
about half of our FOV. So it is reasonable to conclude that our results
may somehow be affected by grain depletion. Detailed comparisons between
the dusty shock model and our data are not feasible at the moment
because of the clumpy nature of the plasma in this region.
The results presented in this work and in Paper I leave open two explanations, namely, the effect of grain depletion in the PSPC spectra of SNR's and the presence of at least two phases in the post-shock medium. Given the old age of the Vela SNR and therefore the lack of reverse shocks due to stellar ejecta, the most immediate interpretation of the multi temperature nature of the X-ray emission could be the expansion of the blast-wave in an inhomogeneous medium with formation of multiple secondary shocks in an optically thin environment. This will be studied with a greater detail and with the aid of numerical simulations in a following paper.