We have studied buried oxide formation as a function of implantation and annealing conditions. The layers appear to form via a nucleation and growth process, so the quality of the oxide and the perfection of the overlying crystalline Si layer depend more strongly on the substrate temperature during implantation than on the annealing temperature. Since it is easier to observe the layer formation process in 'a thin (<1000A) layer, we concentrated on sub-stoichiometric doses and chose substrate temperatures below 400 0 C to stay in a homogeneous nucleation regime. Then we varied the annealing temperature from 11500C to 1407'C. Modeling the coalescence of the oxide layer as a thermally-activated process yields activation energies of approximately 6 eV, suggesting that crystalline damage removal may be the bottleneck for this substrate temperature regime.
INTRODUCTIONMuch of the work to date on buried oxide layers formed by ion implantation involves doses of enough oxygen (>l.8xl0+lS/cm 2 ) to form a stoichiometric oxide layer during implantation [1]. This layer is then a minimum of 3000A thick after annealing. In contrast, we have been concentrating on doses which are almost an order of magnoitude lower [2]. These gppear as roughly gaussian profiles within the Si (FWHM of -1800A, at a depth of -4000A). A subsequent high temperature anneal then causes the oxygen to travel up the concentration gradient and coalesce to a narrow (-600X) stoichiometric SiO 2 layer. The advantage to studying thinner layers is that the growth process of the layers can be examined in detail. Lower doses also mean shorter implant times and lower defect densities in the Si overlayer. Moreover, Silicon-on-Insulator (SOI) structures with thin oxides have a higher capacitance and therefore require lower charging energies. The disadvantages of this technique can be summarized by noting that it is more difficult to fat ricate a continuous narrow oxide layer* the substrate temperature during the implantation is more crucial and higher temperature anneals are necessary. Our goal is to understand the mechanisms of formation of these layers in order to be able to optimize them.Our results on the role of substrate temperature in the formation of these layers were presented at the Fall 1986 MRS Meeting (3]. Basically, we showed that the layer formation involves a competition between nucleation and growth of SiO 2 precipitates. This means that the substrate temperature during implantation is particularly important. At low substrate temperatures (_310'C), the incoming ions create a substantial amount of damage which provides a myriad of nucleation sites. As a result, nucleation of SiO 2 precipitates occurs uniformly over the region of the implant and a uniform layer forms easily as the temperature of the sample is raised for the anneal. Unfortunately, regrowth of the heavily oxygenated Si does not proceed smoothly and twins are the inevitable consequence. At higher substrate temperatures (>400 0 C), dynamic annealing during the implant causes real time dam...