The article discusses using this engine for satellite station keeping, so it only needs a small force if it is on a significant time. At 100,000 alphas per pulse and 75MHz, I get 3.5 watts out and 0.6 micronewton (total momentum all alphas all directions times alpha generation pulse rate). This is likely not enough force for anything useful but I expect there is a lot of optimization to be done. It talks about being possibly only 50% efficient; perhaps some of the alphas going in the wrong direction are expected to be lost. Would depositing a few watts of alpha particles into the boron and metal film layers destroy them? How long would that take?
Note this is not a very efficient rocket in terms of force / power, because the high velocity of the alphas cause them to require large kinetic energy. However, the point is to optimize fuel use. I calculate boron used at 0.05 ng/sec, or 1.6 mg/year if the device is always active.