On consideration of the likely unworkable idea of centrifuging cold neutrons to discriminate their mass number and perhaps other features, I now see that a centrifuge still provides a possibly convenient way to give impetus via the tangential velocity seen there. I would propose that small experiments thought to involve ultracold neutrons through whatever theoretical construct, could be conducted in the buckets or wells of a supercentrifuge or even a benchtop preparatory type centrifuge. Supercentrifuges see rpm in the range of 10 to 30 k rpm. An ultracentrifuge can reach 150k rpm, but seems impractical due the small rotor radius and the stress over one million gravities can place on experimental materials. Well-bottom radii on supers that I am familiar with are in the range of 10 cm. The accessibility of supercentrifuges, or simple benchtop refrigerated types is widespread. The standard benchtop rarely is rated for more than about 10k rpm. However the working radius can be high in those, and the accessibility and space on the rotor provide room for well balanced and properly supported experimental setups to be run up to substantial tangential velocities. So, for example, a benchtop with a 10,000 rpm maximum and a 20 cm radius gives a tangential velocity on the rotor of about 209 m/s. Not huge, but "warmer" than "ultracold". Using online calculators and the back of an envelope I see, with a modest amount of crosschecking, that the energy acquired by such a single neutron would be about 3.66 X 10 e-23 Joules, or 2.28 X 10 e-4 eV. Using an online calculator I see that to be equivalent to a 5437 micron IR photon, or to 55 gigaHz.
These are not strong signals, but if tetrahedral neutron assemblies are involved, we would see 4-fold increases in those energies. A refrigerated centrifuge and the directionality of the neutrons could provide a convenient way to mask undesired signals. Further a particular screen or target can be used that would only react to neutrons. As far as that goes, electrostatic diverters could remove stray charged particles, protons etc. from the output. One drawback might be the dispersion of the neutrons around the full circumference of the centrifuge. On the other hand, if pulsed events were being interrogated, the rotation could provide some temporal resolution on a (very slow by electronic standards) millisecond time scale, that is at 10,000 rpm = 6 msecs per revolution. Higher speed centrifuges come with quite nice safety shielding to catch failing rotor parts. Such a safety bucket would incidentally likely absorb stray radiation-- although I would double check on any risks from such "warm" neutrons, and add sufficient additional shielding if it seemed advisable. After all, if applied to LENR, we would be separating components of reactions that presumably normally self extinguish very near the initiating event(s).
At this point it would appear that one is not able to generate enough tangential velocity to even reach the standard of a "thermal" neutron, which is 0.025 eV (considered so because it is presumed to be in equilibrium with air at 20 degree standard room temperature). But the directionality and focus of origin at the LENR "experiment" on the rotor may provide enough resolution to make such a technique nonetheless useful and / or informative.