The diffusion coefficient of Si in crystalline Ge between 550 and 900C (figure 2) was determined. A molecular beam epitaxially grown buried Si layer in an epitaxial Ge layer on a crystalline Ge substrate was used as the diffusion source. For samples annealed at above 700C, a 50nm-thick SiO2 cap layer was deposited in order to prevent decomposition of the Ge surface. The temperature dependence of the diffusion coefficient could be described by:

D (cm2/s) = 4.2 x 101 exp[-3.32(eV)/kT]

over the entire temperature range. These data extended previous measurements by 2 orders of magnitude at low temperatures. The diffusion of isovalent Si was slower than Ge self-diffusion over the full temperature range and the activation enthalpy was higher than that for self-diffusion. This suggested a reduced interaction potential between the Si atom and the native defect mediating the diffusion process. For Si, which was smaller in size than the Ge self-atom, a reduced interaction was expected for a Si–vacancy (Si–VGe) pair. It was therefore concluded that Si diffused in Ge via the vacancy mechanism.

Diffusion of Silicon in Crystalline Germanium. H.H.Silvestri, H.Bracht, J.Lundsgaard Hansen, A.Nylandsted Larsen, E.E.Haller: Semiconductor Science and Technology, 2006, 21, 758-62

The diffusion coefficient of Si in crystalline Ge between 550 and 900C (figure 2) was determined. A molecular beam epitaxially grown buried Si layer in an epitaxial Ge layer on a crystalline Ge substrate was used as the diffusion source. For samples annealed at above 700C, a 50nm-thick SiO2 cap layer was deposited in order to prevent decomposition of the Ge surface. The temperature dependence of the diffusion coefficient could be described by:

D (cm2/s) = 4.2 x 101 exp[-3.32(eV)/kT]

over the entire temperature range. These data extended previous measurements by 2 orders of magnitude at low temperatures. The diffusion of isovalent Si was slower than Ge self-diffusion over the full temperature range and the activation enthalpy was higher than that for self-diffusion. This suggested a reduced interaction potential between the Si atom and the native defect mediating the diffusion process. For Si, which was smaller in size than the Ge self-atom, a reduced interaction was expected for a Si–vacancy (Si–VGe) pair. It was therefore concluded that Si diffused in Ge via the vacancy mechanism.

Diffusion of Silicon in Crystalline Germanium. H.H.Silvestri, H.Bracht, J.Lundsgaard Hansen, A.Nylandsted Larsen, E.E.Haller: Semiconductor Science and Technology, 2006, 21, 758-62

The diffusion coefficient of Si in crystalline Ge between 550 and 900C (figure 2) was determined. A molecular beam epitaxially grown buried Si layer in an epitaxial Ge layer on a crystalline Ge substrate was used as the diffusion source. For samples annealed at above 700C, a 50nm-thick SiO2 cap layer was deposited in order to prevent decomposition of the Ge surface. The temperature dependence of the diffusion coefficient could be described by:

D (cm2/s) = 4.2 x 101 exp[-3.32(eV)/kT]

over the entire temperature range. These data extended previous measurements by 2 orders of magnitude at low temperatures. The diffusion of isovalent Si was slower than Ge self-diffusion over the full temperature range and the activation enthalpy was higher than that for self-diffusion. This suggested a reduced interaction potential between the Si atom and the native defect mediating the diffusion process. For Si, which was smaller in size than the Ge self-atom, a reduced interaction was expected for a Si–vacancy (Si–VGe) pair. It was therefore concluded that Si diffused in Ge via the vacancy mechanism.

Diffusion of Silicon in Crystalline Germanium. H.H.Silvestri, H.Bracht, J.Lundsgaard Hansen, A.Nylandsted Larsen, E.E.Haller: Semiconductor Science and Technology, 2006, 21, 758-62

Figure 2

Diffusivity of Si in Ge