The kinetics of GaAs MOCVD using in situ Laser Reflectance Monitoring

Electronic versions

Documents

  • A.J. Clayton

Abstract

The kinetics of gallium arsenide (GaAs) parasitic growth during metal-organic chemical vapour deposition (MOCVD) was investigated using in situ laser reflectance monitoring.
Experiments were carried out in a horizontal quartz reactor tube using trimethylgallium (TM Ga) and arsine (AsH3), with a reactor pressure of 200 mbar and total flow (f 101) of 1.49 standard litres per minute (slm). A quartz wafer was aligned with the laser to allow the growth rate to be measured during growth. Experiments were carried out between 400°C and 500°C where surface reactions limit the growth rate. The V /lll ratio was varied between 10, 30 and 50 by changing the partial pressure of TMGa (pTMGa), while the AsH3 partial pressure CPAsH3) was kept constant. The growth rate was observed to increase with temperature and PTMGa, which was determined to be due to the increased
supply of methyl radicals and concentration of gallium species at the surface.
Reflectance oscillations were only obtained when coalescence of the growing deposit occurred at a greater rate than vertical growth. The 2-dimensional layer growth was observed at 450°C when the V /III ratio was equal to IO and when both parameters were at high values in the range employed. Analysis on the surface morphology established that the reflectance intensity reduced when 3-dimensional growth occurred, with high intensity and the presence of oscillations when deposition was 2-dimensional.
Compositional analysis determined that the parasitic deposit was non-stoichiometric, with an average Ga:As ratio of I :( 1.3). A kinetic model was developed using the steady state approximation and rate parameters obtained from the literature for the reaction mechanisms involved in the GaAs process. The model results correlated with the experimental results such that GaAs yield increased with temperature and PTMGa· The measured growth rate from experiments was generally greater than that calculated in the model, which is expected to be due to the rough surfaces where a larger surface area increased the rate.

Details

Original languageEnglish
Awarding Institution
  • University of Wales, Bangor
Supervisors/Advisors
  • Stuart Irvine (Supervisor)
Award date2005