Maybe you could directly convert the higher energy alphas and let the lower energy alphas bounce back into the wiffle ball and heat the plasma.
In the case a space engine, imagine you have a polywell out in the vacuum of space, fueled by gas-puff, beams etc. out of this polywell alphas are emitted in 4 pi steradians, all you have to do is set up a potential grid at several MeV to bounce the alphas coming at the spacecraft off it so that the go away from the spacecraft, you could probably approach efficiencies of 90% that way.
Heat Transfer Limitations Re: Power Plants and Rockets
Bad idea. You don't want to "heat" the plasma - it's doing just fine on the drive voltage, and the only thing a bunch of alphas cycling through the core is going to do is scatter ions out of the well before they have a chance to fuse. Also, this would dramatically increase impingement heating on the magrid surface.
Regarding your space engine, if you add a bit of extra stuff along with the ejected alphas you have what Bussard called a Diluted Fusion Product (DFP) engine. Isp in the range of a million or so. Great for cis-Oort flight but not so hot for SSTO...
Maybe there's a way to shape the fields in order to direct the alphas into some configuration of plates extended from the wall, probably in the magrid shadow, such that the energy loss is minimal and the higher-energy alphas just get closer to the outer wall before touching a plate...
Regarding your space engine, if you add a bit of extra stuff along with the ejected alphas you have what Bussard called a Diluted Fusion Product (DFP) engine. Isp in the range of a million or so. Great for cis-Oort flight but not so hot for SSTO...
Maybe there's a way to shape the fields in order to direct the alphas into some configuration of plates extended from the wall, probably in the magrid shadow, such that the energy loss is minimal and the higher-energy alphas just get closer to the outer wall before touching a plate...