Accurate modelling of average phosphorus diffusivities in germanium after long thermal anneals: evidence of implant damage enhanced diffusivities

Abstract

Average phosphorous diffusivities after implantation in 1 0 0 germanium have been measured for long anneals (i.e., 3-10 h) at temperatures from 600 to 800 • C. Considerable dose loss after annealing is also observed and quantified for temperatures below 800 • C. A diffusion model using an extrinsic diffusivity coefficient combined with a segregation component between the germanium and the oxide, to account for dopant loss, is found to be sufficient to completely explain the observed diffusion profiles. The best-fit diffusivity and segregation coefficients are reported for this model and the diffusivities are found to be over an order of magnitude slower than those measured after rapid thermal annealing (i.e., short anneals of only a few seconds). It is proposed that this disagreement of diffusivities between short and long anneals is due to implant damage perhaps similar to well-known transient enhanced diffusion effects observed in silicon.

Diffusion and defect reactions between donors, C, and vacancies in Ge. II. Atomistic calculations of related complexes

“…46 and 47͒ and the self-interstitial ͑most stable neutral structure is the ͗110͘ dumbbell͒ 48 formation energies from 2.3 to 4.1 eV. [49][50][51] Consequently, in Ge, self-interstitials are not as important as in Si and most diffusion processes are dominated by V. [9][10][11][12] A significant exception to the V domination of diffusion processes in Ge occurs for B. Experimental studies determined that B has a very high migration activation energy.…”

“…In Ge, boron ͑B͒ has been integrated as a successful p-type dopant; 4,5 but for n-type dopants such as phosphorus ͑P͒, there are problems with diffusion and activation control. [6][7][8] In recent experimental studies, [9][10][11][12] it has been determined that donor diffusion in Ge occurs via a vacancy ͑V͒-mediated process. This is in contrast to Si, where the diffusion of n-type dopants, such as P and arsenic ͑As͒, is mediated both by self-interstitials and V. 13,14 For the development of Ge-based devices, it is important to be able to design specific selective area doping that results in the appropriate distribution of dopants in the region of interest.…”

Modeling and experiments on diffusion and activation of phosphorus in germanium

“…14͔͒ as well as the severe dopant loss observed during annealing. 7,11,[15][16][17][18][19][20] Although recent reports are in general agreement that phosphorus diffusion depends on the square of the electron concentration, the quantitative values deduced for the intrinsic P diffusion coefficient are different. For example, Chui et al 7 reported significant higher diffusivities for P diffusion into Ge than did Brotzmann and Bracht 12 and Carroll and Koudelka.…”

“…Especially in the case of phosphorus, possible reasons for these poor-quality n + p junctions are the high P diffusivity 7,11,12 and the difference between chemical solubility limit ͑estimated approximately at 2 ϫ 10 20 cm −3 at 600°C͒ and the electrical activation ͓re-ported at 6 ϫ 10 19 cm −3 at 600°C ͑Ref. 13͒ or 7 ϫ 10 19 cm −3 at 800°C ͑Ref.…”

“…For example, Chui et al 7 reported significant higher diffusivities for P diffusion into Ge than did Brotzmann and Bracht 12 and Carroll and Koudelka. 11 As regards dopant and self-diffusion a rather broad consensus on a vacancy assisted mechanism exists, but different views are obtained about the charge state of the defect that mostly influences the dopant movement. Werner et al 21 suggested that Ge self-diffusion occurs through single negatively charged and neutral vacancies.…”