ABSTRACT
Department o f Physics, Queen Mary and Westfield College, University o f London, London, E l 4NS, United Kingdom
1. In tro d u c tio n ................................................................................... 53 2. Measuring Interdiffusion................................................................ 54 3. Interdiffusion of GaAs Based M aterials........................................ 59 4. The Activation Energy for In terd iffusion .................................... 60 5. Arsenic Overpressure Annealing ................................................ 65 6. Dielectric Encapsulant A n n e a lin g ................................................ 69 7. The Fermi Level Effect ............................................................... 71 8. Strained Material S y s te m s ............................................................ 76 9. C o n c lu s io n s ................................................................................... 80
The thermal stability of semiconductor heterostructures and in particular their interdiffusion has been an issue since the first heterostructures were produced. While some early work on self-diffusion in semiconductors goes back to the 1960’s with work on silicon and germanium [1], Some of the first work on interdiffusion was by Chang and Koma [2] in 1976, on the GaAs/AlAs system and since then there has been a proliferation of papers looking at most of the material systems that have been grown. In addition to the wide range of systems studied, the effects of a wide range of perturbations on the diffusion have been studied. These perturbations have included the surface condition (i.e. arsenic overpressure [3-6] or encapsulant [3-8]), doping [9-12], and ion implantation of a large variety of elements, for example [13,14], and in order to study these various effects a variety of techniques have been used, including Auger profiling [2], SIMS [9], TEM [15] and photoluminescence [14,16]. However, despite this huge quantity of work there is still some uncertainty as to the exact mechanisms responsible for the interdiffusion. Indeed the key
parameters necessary to characterise diffusion in materials, the activation energy and prefactor are still somewhat uncertain, with activation energies of 0.32 [17] to 6.2 eV [4] being quoted together with prefactors which cover a range of some 21 orders of magnitude. Much of this variation in the measured activation energies can be attributed to the use of small data sets and the resulting experimental uncertainties in Ea cause the massive variations in measured D0 values.
