Nelson reported epitaxiel layers that were typically 60 to 80 µm thick. Surfaces were
rough due to rapid growth in the low temperature regime near the completion of the
process and large (in the order of several µm) droplets of Ga were also noted at the
interface. A.R. Goodwin et al (1968) recognised this problem of rapid growth. Their
solution was to introduce a temperature gradient, approximately 10°C cm-1 so that the
seed was always colder than the melt by a fixed amount. As before, the boat was tipped
at 850°C and the furnace temperature lowered at a rate of 10°C/min to 600°C. Thus
layers of 150 to 200 µm thickness were grown over 3 hours. The surfaces of the epilayer
were good and occasionally mirror bright. Such experimental work led Goodwin et al to
reconsider the travelling solvent technique (TST) reported earlier by Miavsky and
Weinstein (1963). The Ga solvent is saturated with As by adding GaAs crystals, some are
dissolved while others remain floating on the surface throughout the growth cycle.