Order and Disorder in Semiconductors
- PDF / 4,513,137 Bytes
- 7 Pages / 576 x 777.6 pts Page_size
- 49 Downloads / 223 Views
MRS BULLETIN/DECEMBER 1995
examples of ongoing research to illustrate the power of defect utilization and the degree of sophistication that has been achieved. This principle of doping by replacement can only be safely practiced for well-controlled, homogeneous materials—of which we still have only a few semiconductors. Uncritical transfer of this principle to inhomogeneous materials does not guarantee success. Inhomogeneous materials derive many of their electronic properties not from the substituted point defects but from the capricious details of the inhomogeneity itself. A similar situation arises in the current quests foi novel quantum devices where geometry plays an overriding role in establishing energy levels, densities of electronic states, or other electronic properties, thus diminishing the influence of substituents. Dimension-controlled quantum device structures, however, also command highly perfected base materials. Dopant atoms, their interactions with native defects and their diffusion properties, are discussed first. Multiatomic defects, now of vivid interest, and their stabilities are touched upon next. The topic of gettering, still of irreplaceable technical significance, is mentioned later. The themes of inhomogeneity and quantum structures conclude this article. Current Research Topics Dopants: Their Interactions and Diffusion The diffusion of a simple trivalent acceptor or pentavalent donor in a silicon host still constitutes an essential element of device processing for bipolar and even for unipolar devices. An amazing degree of empirical knowledge rules these processes even today because the atomistic details of diffusion—even for these simplest substitute atoms—are in reality very complex. A multitude of dopant interactions with native defects, especially vacancies and interstitialcies, govern the
diffusion processes and thus determine the technologically vital properties such as depth of a junction, space charge widths, or channel properties. Diffusion of the donor phosphorous near a SiC^/Si interface provides a good example. The solubility of P in the oxide is lower than in silicon itself. Therefore, one expects a pileup of P atoms in the Si near an interface to the oxide, well-known from the pioneering work by Grove.3 However, no really satisfying model simulation with reliable, predictive power was available for a long time. Lau and his colleagues4 at the Siemens Laboratories then proposed a new model, which included a third phase sandwiched in between oxide and elemental silicon. This hypothetical interface phase was assumed to provide ample amounts of sites for phosphorous atoms. Such P atoms are then assumed to become electrically inactive, which is an easily acceptable proposition since the direct substitution inside a regular lattice no longer prevails. Lau et al. then proceeded to simulate diffusion conditions with a set of rate equations and found good agreement with experimental data. One single curve was presented to fit data for different P concentrations, as well as a variety
Data Loading...
