A solution to the surface arsenic stoichiometric problem at the GaAs(001) growth surface in atomic layer expitaxy

We report the previously undiscovered result that the ALE GaAs growth rate, using TMGa and AsH3 in the pulsed-jet reactor, becomes around 1.2–1.3 monolayer (ML)/cycle with a sufficient AsH3 supply, maintaining the self-limiting fashion for the TMGa injection at 500°C. The growth rate change for the AsH3 pulse duration shows a rapid increase to 1 ML/cycle followed by a gradual further increase up to a saturation value of around 1.2–1.3 ML/cycle. The results suggest the existence of over 1 ML of adsorbed As layer on the GaAs(001) growth surface, and these excess As atoms contribute to growth beyond 1 ML/cycle. The ideal self-limiting growth at 1 ML/cycle for the TMGa injection is achieved by optimizing the reaction kinetics of the producing surface As layer, so as to complete at a monolayer coverage. We also study the dependence of the growth rate on the H2 purge time after both TMGa and AsH3 from 500 to 580°C. We show that the growth rate decreases only by increasing the post-AsH3 H2 purge time. The observed decrease in the growth rate for the increase of the H2 purge time after AsH3 supply attributes to the thermal desorption of the surface As atoms. Assuming the existence of surface reconstruction and the surface As stoichiometry at GaAs(001) just as reported so far in molecular beam and metalorganic vapor-phase epitaxy and the phase transition among them through As desorption, we can reasonably explain the experimental growth rate variations in ALE. We indicate that the change of As-rich surface reconstruction and related stoichiometry through As adsorption and desorption at GaAs(001) plays an important role in determining the ALE self-limiting growth rates.