The next generation of long pulse fusion devices will impose severe requirements on the properties of plasma-facing materials. In devices such as ITER, a divertor design is being considered, using a divertor plate which would be either tungsten or a low-Z material such as graphite or beryllium. Graphite and beryllium have a relatively high light ion erosion rate. Tungsten has a much lower sputtering rate for light ion impact, but it is subject to runaway self-sputtering. Because of its limited thermal conductivity, it must be used as a relatively thin plate which might be subject to damage during a disruption. Strongly segregating lithium alloys have been proposed as a means of producing a self-sustaining low-Z overlayer which lowers plasma Zeff and resists self-sputtering. Aluminum-lithium alloys are among the better-characterized lithium-bearing alloys, and it has been demonstrated that lithium segregates strongly in aluminum. However, aluminum has a relatively low melting point, and for low lithium concentrations, the lithium diffusion rate is too slow to replenish lithium at the rate at which it is eroded by the incoming plasma. It has been suggested previously that the β phase Al-Li alloy (48–54 at.% Li) should have high enough diffusivity to be able to replenish surface lithium, and that incorporation of the β-phase AlLi in a composite with tungsten would provide improved high temperature strength and melt layer stability, along with significantly better thermal conductivity than pure tungsten. Such a composite has been fabricated, as well as a variation containing titanium as a means of controlling oxidation at grain boundaries. The Li overlayer formation, erosion, and replenishment are characterized for the β-phase LiAl alloy, and W-AlLi and W-Ti-AlLi composites. It is found that Li diffusion is extremely rapid, and the composites form an oxygen-free Li overlayer which is stable under continuous ion beam sputtering.