Upcoming sensor technology allows for dark matter direct detection all the way down to the warm dark matter limit of ~ 10 keV. At such low masses, the usual nuclear recoil picture breaks down, as the dark matter recoils against athermal phonon modes instead. I will show how the rate for these processes can be calculated and why superfluid helium and polar materials are good targets for this type of dark matter. In the latter case the crystal axis can provide a daily modulation of the scattering rate.