A Generalized Model of Beryllium Diffusion in InGaASs Epitaxial Structures Under Point Defect Nonequilibrium Conditions
- PDF / 374,041 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 27 Downloads / 216 Views
395 Mat. Res. Soc. Symp. Proc. Vol. 470 0 1997 Materials Research Society
first, an undoped ln0.53Ga 0.47As layer with thickness of 0.5 Am was grown, followed by a Bedoped ln(.5 3Gao. 47As layer of 0.2 !am in thickness with 3 x 10I19 cm-3 doping level. Finally, an undoped In0 53Ga 0.47As layer of 0.5 prm was grown on the Be-doped layer. For CBE growth technique, metalorganic sources of trimethylindium (TMIn) and triethylgallium (TEGa), gaseous source of cracked pure arsine (AsH 3 ), and solid source of beryllium have been used as precursors. Doped layers have been grown at a rate of 2.40 ptm/h and non doped ones at 1.76 ptm/h with a V/Ill flux ratio of 2 and a substrate temperature of 500'C. For GSMBE growth method, elemental gallium (Ga), indium (In), and Be beams, and gaseous source of cracked pure arsine have been employed. All layers have been grown at a rate in the range of 0.8-2 itm/h with a V/Ill flux ratio of 5 and a substrate temperature of 500'C. After post growth Rapid Thermal Annealing (RTA) have been performed in a halogen lamp furnace ADDAX XM, resulting samples were depth profiled in a Cameca IMS-4F SIMS (Secondary Ion Mass Spectrometry) apparatus. RESULTS The Be depth profiles for 700'C post-growth annealing during 1, 2 and 3 minutes are given by solid lines in Figs. I (for CBE method) and 2 (for GSMBE method). In both cases, we could observe that experimental profiles exhibit anomalous shapes, providing kinks which seem to remove with time diffusion increase. Many authors explained these double profiles by a strong dependence on the defect concentrations (self-interstitials and vacancies) of the crystal [1-7]. On the other hand, Be diffusion is found to be more pronounced for samples obtained from CBE technique. Thus, a best base confinement is obtained under our growth conditions from GSMBE technique. Although both epitaxial growth methods are similar, they contain basic different characteristics (epitaxial frame, precursors, growth parameters...). Besides, this diffusion length discrepancy could be also related to the effect of V/Ill flux ratio [8,9]. I. Kick-out Diffusion model adapted to our experimental conditions. Assuming point defect nonequilibrium conditions, the Kick-out mechanism could be used to explain p-type dopant diffusion in III-V materials [6,7]. Because of the strong influence of electric build-in field [6] with charged Be interstitial species, simulated diffusion fronts are too steep to fit our experimental profiles. Consequently, the Kick-out mechanism has been modified considering neutral Be interstitial species Be' and singly ionized gallium (Ga) and indium (In) self-interstitials I'! K
(I+
Be"'
Data Loading...
