Despite the expansion of GeSn and SiGeSn epitaxial growth research there remains only a small volume of work on fundamental CVD chemistry and kinetics for this material. There is a comparably large volume of work on the fundamental surface chemistry related to Si, Ge, and SiGe epitaxy, in both experimental and theoretical studies. There remain several open questions as to how exactly the chemical vapor deposition of GeSn occurs. Because of the low solubility of Sn in Ge, epitaxial growth typically must be done at temperatures < 400°C - and higher Sn contents can even require temperatures below 300°C. It has been a point of debate as to how the precursors GeH4, Ge2H6, and SnCl4 are able to react and sustain deposition at such low temperatures. In fact, the addition of SnCl4 acts to increase the growth rate beyond what can be achieved with GeH4 alone, when it is intuitively expected that the addition of Cl to the chemistry should decrease the growth rate. In this work a theoretical/computational chemistry exploration of the gas-phase and surface chemistry of GeH4 and SnCl4 on Ge and GeSn surfaces is presented. Molecular orbital DFT was used, in-which the semiconductor surfaces are approximated as clusters to reduce the computational cost relative to plane-wave DFT methods. The Gibbs free energy changes at 325°C, ΔGr, and kinetic activation barriers, Ea, for relevant gas-phase and surface reactions were calculated.An analysis of possible gas-phase reactions reveals that GeH4 and SnCl4 have dissociation barriers of Ea = 237 kJ/mol and 305 kJ/mol respectively. However, they readily undergo H/Cl exchange reactions of the form GeH4 + SnCl4 → SnHCl3+ GeH3Cl. This reaction is found to be thermodynamically favorable with ΔGr = -18 kJ/mol with a relatively low activation barrier of Ea = 139 kJ/mol. The products SnHCl3 and GeH3Cl are much more reactive than the starting precursors and readily dissociate into SnCl2 and GeHCl with Ea = 168 kJ/mol and 190 kJ/mol. At typical growth temperatures it is expected that most surface sites will be H or Cl terminated such that the removal of adsorbed H and Cl to generate open surface sites is the kinetically limiting factor. On the Ge-Ge surface we find an interdimer H2 desorption Ea = 215 kJ/mol, and on a Ge-Sn surface dimer we find the H2 interdimer desorption Ea = 184 kJ/mol. This result may at least partially explain the growth rate enhancement of GeSn relative to Ge, however large surface-Sn concentrations would be required to produce a growth rate enhancement. Sub-surface Sn incorporation was found to reduce the interdimer H2 desorption barrier from two surface Ge atoms to a value of Ea = 207 kJ/mol. This result suggests that Sn can increase the H desorption rate non-locally via strain and or electronic effects, however the change is minimal and cannot explain observed growth rate enhancements. A more likely explanation is that the generation of GeHCl allows for insertion reactions into the Ge-H surface bonds. Indeed, it is found that ΔGr = -87 kJ/mol and Ea = 45 kJ/mol for this insertion reaction. This theoretical activation barrier closely matches Arrhenius activation energy values reported from CVD growth studies. Therefore, the gas-phase reactions between GeH4 to SnCl4 to form more reactive intermediates appears to be the enabling factor in explaining the enhanced growth rates of GeSn relative to Ge. Figure 1
Read full abstract