portunity to make investigations with the STM, and to Dipl.-Phys. S. Br~uer for his help in recording the STM images and for helpful discussions about the results. We also thank the Stiftung Volkswagenwerk for financial support.

TEOS and O~ atmospheric pressure chemical vapor deposition has excellent conformality, film quality, and low particle generation; however, deposition rate obtained on thermal oxide is lower than that on silicon and the films have rough surface and higher etching rates if deposited with high ozone concentration. When deposited on silicon with a higher ozone concentration, films of sufficiently high quality are obtained. This drawback of the low film quality has been removed by doubleJayer deposition with two different ozone concentrations on thermal oxide, depositing the thin films at first with a low ozone concentration on thermal oxide and next with a high ozone concentration to the desired thickness. The film quality, namely, etch rate, depends on the thickness and ozone concentration for the first layer. The thicker the first film and the higher the ozone concentration, the better the film quality. By adjusting the first layer's thickness and ozone concentration, films can be obtained with quality that is sufficiently high for VLSI device applications.

Atmospheric pressure tetraethyloxysilicane (TEOS)/O3 chemically vapor deposited provides excellent step coverage for submicron device structures; however, the properties of the deposited films depend on the surface characteristics of the base materials being used. To illustrate this dependence, the deposition rate of nondoped silicon dioxide obtained on a thermal oxide surface is significantly lower than the deposition rate obtained on a bare silicon surface. A new method to eliminate this base material dependence involving plasma treatment has been investigated. The optimum treatment consists of exposing the base materials to a nitrogen plasma for 1 min while maintaining the base materials at 250~ Films deposited on thermal oxide base materials which have first been treated by this new method were found to have the same deposition rate, aqueous HF etch rate, and surface morphology as those films deposited on untreated bare silicon. In addition to a nitrogen plasma, oxygen and argon plasmas were studied and found to produce similar results when the base material temperature was raised to 350~ The elimination of base material dependence through the use of this new plasma treatment technique has resulted in higher integrity TEOS/O3 oxides and has also expanded the range of applications for this unique planarizing technology for very large scale integrated device fabrication. ) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 131.230.73.226 Downloaded on 2015-06-03 to IP ) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 131.230.73.226 Downloaded on 2015-06-03 to IP

SiO 2 films deposited on Si and on thermal SiO2 by tetraethylorthosilicate [TEOS, Si(OC2H5)4]/O3 atmospheric-pressure chemical vapor deposition (APCVD) were analyzed by thermal desorption spectra (TDS). The TDS results show that more silanols were incorporated during deposition and more water was absorbed during and after deposition in films deposited on Si substrates than on thermal oxide substrates. The latter result indicates that the elimination of water by-products is not the limiting step in TEOS/O3 APCVD. Based on the former result, a silanol model was proposed for the surface processes. On surfaces with a uniform and high density of silanol sites, or on which silanols readily form under TEOS/O3 APCVD conditions, the active silanol groups in the gas phase contribute to film formation and replenish silanol sites, resulting in continuous, high growth rates. On surfaces with few silanol sites, it is difficult to form silanol sites and the nonsilanol-containing polysiloxanes contribute to film formation, resulting in continuous, low growth rate. This model explains well both the surface dependence and the memory effect of TEOS/O3 APCVD.

Borophosphosilicate glass (BPSG) films were deposited using tetraethoxysilane (TEOS), triethylborate (TEB), trimethylphosphate (TMOP), and ozone at atmospheric pressure. A uniform distribution of borate was obtained throughout the thickness of the film using TEB instead of trimethylborate (TMB). The deposition rates were characterized as functions of dopant gas flow rates, deposition temperature, and ozone concentration. Both dopants enhanced the deposition rates. The deposition rates did not show ozone concentration dependence. The typical deposition rate was as high as 0.2 μm/min. The borate and phosphate contents were characterized as functions of dopant gas flow rates, and deposition temperature. They ranged from 8 to 15 mole percent (m/o) for borate and from 3 to 8 m/o for phosphate. The film stress was as low as 1×109 normaldyn/cm2 (tensile) when deposited at 400°C with 4% ozone and relaxed to −0.7×109 normaldynsol;cm2 a week after deposition. The film shrinkage was minimum (5%) at 400°C, and decreased with increasing ozone concentration. The step‐coverage of an as‐deposited film was superior, hence, the film reflowed smoothly even after the low temperature (850°C) annealing.