To me it seams that it is necessary that the scale factor becomes infinite in finite time, at least for a flat universe. I am not sure what happens in an universe which is not spatially flat.
It follows from this argumentation:
First, let's show that the phantom energy implies a growing Hubble parameter in a flat universe. The time derivative of the Hubble parameter is:
[tex]\dot H = \frac{d}{dt} \left( \frac{\dot a}{a} \right) = \frac{\ddot a}{a} - \left( \frac{\dot a}{a} \right)^2[/tex]
Let's take the
Friedmann equations for zero spatial curvature [itex]k = 0[/itex] and insert in the relation above:
[tex]\dot H = - 4 \pi G \left( \rho + \frac{p}{c^2} \right)[/tex]
If we have a phantom energy:
[tex]\frac{p}{c^2} < - \rho[/tex]
then:
[tex]\dot H > 0[/tex]
Second, let's go back to the first relation for [itex]\dot H[/itex]. Take:
[tex]a = t^q[/tex]
Inserting for [itex]a[/itex], [itex]\dot a[/itex] and [itex]\ddot a[/itex] and imposing the condition [itex]\dot H > 0[/itex], we get:
[tex]q (q-1) - q^2 > 0[/tex]
This can only hold if [itex]q < 0[/itex], which in turn implies some asymptotic behaviour of [itex]a[/itex].
By the way, the relation:
[tex]a = t^q[/tex]
with negative [itex]q[/itex] does not make sense at all, because [itex]a[/itex] must be zero at [itex]t = 0[/itex], but one could imagine a similar relation like e.g.
[tex]a = - a_0 + (t - t_{R})^q[/tex]
which needs of a negative [itex]q[/itex] to fulfil the phantom energy condition and has an asymptotic behaviour.
From such a relation you can see also that there exists another branch for the scale factor after t
R with a contraction. It is not clear, however, how to interpret this, because the universe goes through a singularity in the energy density (from the first Friedmann equation you can see how the energy density behaves).