Mechanisms and reaction paths for surface roughening and epitaxial breakdown during molecular beam epitaxy: Fundamental limits

Experiments were designed to probe roughening pathways leading to epitaxial breakdown (EB) during low-temperature (T_s = 95–190°C) growth of Ge(001) by molecular beam epitaxy (MBE). Layers grown at T_s $\gtrsim$ 170°C remain epitaxial to thicknesses h > 1.6 μm, while deposition at T_s < 170°C leads to EB at critical thicknesses h₁(T _s) which increase exponentially with T_s. Low-temperature (LT) Ge(001) surface morphology evolves via the formation of periodic arrays of growth mounds preferentially aligned along ⟨100⟩ directions. Surface widths w and in-plane coherence lengths d increase monotonically with h. As h → h ₁(T_s), deep cusps bounded by {lcub}111{rcub} facets form at the base of interisland trenches and EB occurs as the surface reaches a critical aspect ratio, w/d ≃ 0.02. I demonstrate that EB is a growth mode transition driven by kinetic surface roughening, summarize my results in a microstructural phase map, and propose a growth model.; My results demonstrate that interlayer mass transport plays a decisive role in determining surface roughening during multilayer growth. I show that the introduction of dilute Sn concentrations (C_Sn = 1 × 1018–6 × 1019 cm −3) during Ge(001) low-temperature MBE (LT-MBE) not only decreases the roughening rate, but also increases h₁( T_s). Nevertheless, the mound aspect ratio at EB remains constant, independent of T_s and C _Sn. I attribute the increases in h₁( T_s) to Sn-induced enhancements in both the Ge surface diffusivity and probability of interlayer mass transport. This, in turn, results in more efficient filling of interisland trenches thus delaying EB.; Fully-strained Ge_1−xSn_x layers were grown on Ge(001) to probe the role of incorporated Sn concentrations (C _Sn = 1 × 1018 cm−3 to 6.1 at%) on surface roughening pathways leading to EB during LT-MBE (155°C) of compressively strained films. Sn mediates surface morphological evolution through two competing pathways. With x $\lesssim$ 0.02, the dominant effect is a Sn-induced smoothening of the surface. At higher x there is a change in Ge_1−xSn _x(001) growth kinetics due to a rapid increase in compressive strain. This leads to a reduction in h₁ with increasing x as strain-induced roughening overcomes the smoothening effects and increases in the roughening rate. I show that by varying C _Sn I can controllably manipulate the roughening pathway, and hence h₁, over a very wide range.