A magnetocaloric heat pump works by changing the magnetic field applied to a magnetocaloric material while pumping fluid to move heat. Slaughter explained that this is typically done with permanent magnets. The core of the device involves spinning permanent magnets relative to the magnetocaloric material and using magnetic steel to keep the magnetic field contained.

The Ames model AMR consists of nine “beds” arranged into a ring. The AMR is porous — not unlike a foam mattress, actually. These hold a bunch of tightly-packed gadolinium particles, each only about 200 microns across…so about the size of a very fine coffee ground. At the start of the regenerative cycle, these particles are demagnetized…or not-yet-magnetized, more precisely. They’re kind of … well … asleep in these beds.

As they pass by each stationary AMR bed, the gadolinium magnetizes and gets hot. Then, as the magnetic field moves over to the next bed, that array is no longer exposed to it, and it loses its magnetization. This causes the the gadolinium temperature to drop and get cold, similar to what happens when you depressurize of a gas refrigerant in a traditional compressor.

From there, the working fluid cools, then moves to the hotter end of the heat pump’s internal core. The setup demagnetizes, and the cycle starts again…for the individual bed, that is. The cool thing about this arrangement is that because the magnets are rotating, they’re constantly magnetizing and demagnetizing each bed in turn…and therefore alternatively heating and cooling them off.