It's a little risky, but you can get a faster turn-off time out of a solenoid by introducing a resistance in the freewheeling diode circuit. The time constant is L/R, and for a low resistance coil like that, with no appreciable resistance from the diode, the "off" time can be quite large.
Downside is that the peak turn-off spike voltage goes higher. You can mitigate this somewhat by replacing the diode (probably a crappy 1N4004 or something) with a zero-recovery time schottky and choosing the resistor carefully by monitoring the PIV spikes on your driver. A 5Ω resistor will reduce the time constant to a tenth of what it is now, allowing faster solenoid drop-out and maybe faster cycle times.
Of course, if you are getting around that 18A peak current, you will see close to a 90V spike, and you have to consider the breakdown voltage of the driver. I suspect it won't be close to 90V, though. Got a scope?