In many studies of oxidation kinetics, it has been observed that SiO2 growth in dry oxygen in the thin regime (<500Å) is faster than the classic description of growth in thicker layers by a linear‐parabolic relationship. Growth‐rate enhancement in the thin regime was studied in the 800°–1000°C range under a variety of substrate doping densities and O2 partial pressures using in situ ellipsometry. The enhancement in oxidation rate is found to decay exponentially with thickness, and its thickness extent is approximately independent of substrate orientation, doping density, and oxygen partial pressure; its oxygen pressure and substrate doping dependence suggest that it is caused by physical mechanisms associated with the substrate. Such mechanisms are discussed in part II of this paper (11).

Based upon the linear‐parabolic growth model of silicon oxidation, accurate kinetic rate constants are determined for (100), (111), and (110) silicon oxidized in dry oxygen in the 800°–1000°C range. The oxide growth was monitored by high temperature automated in situ ellipsometry. It is shown that fitting the maximum number of oxidation data points to a linear‐parabolic relationship yields accurate oxidation rate constants that are unique to the oxidation process as described in the Deal‐Grove model, and not just good empirical fitting parameters. This approach is denoted the “optimum Xnormali technique.” Both linear and parabolic rate constants exhibit a break in their activation energies at 950°C. This behavior is discussed and interpreted in terms of the viscoelastic properties of SiO2 .

No abstract

Higher density SiO2 results from the thermal oxidation of Si in dry 02 at lower oxidation temperatures. More than 3% higher density is observed for SiO2 grown at 600~ as compared with 1150~ A consistent model for the formation of this material is deduced based on the following: the temperature dependence of the density, the annealing behavior of the higher density SIO2, and on the literature and new measurements of the intrinsic stress in SiO~ films. The model considers viscous flow of a Maxwell solid and hinges on the attainment of the necessary free volume for oxidation at lower oxidation temperatures.

The significance of the addition of 25 ppm of H20 to O2 has been studied for the thermal oxidation of (100), (110), and (lll) oriented silicon. With 25 ppm water addition to O2 a sharp increase in the over-all rate of oxidation was observed for each orientation and experimental temperature (800 ~ 927 ~ and 996~From data analysis in terms of a linear-parabolic oxidation law, orientation effects were found for both the linear and parabolic rate constants and for both dry 02 and H20 added oxidations. The major effect of trace H20 appears to be on the parabolic rate constants. The experimentally derived kinetic constants are interpreted in terms of surface geometry for the linear rate constants and in terms of the SiO2 structure for the parabolic rate constants. The large kinetic effects observed with trace amounts of H20 might be a cause for the disparity of literature results for dry thermal oxidation of silicon.The thermal oxidation of silicon produces a silicon dioxide (SiO2) passivation film with excellent electrical properties. However, the fabrication of advanced thin film electronic devices requires a knowledge of the parameters affecting thickness control and impurity content of the thin films. To describe SiO2 film growth via thermal oxidation of silicon by oxygen, a linearparabolic model is assumed. Numerous publications confirm the adherence of the SiO2 growth kinetics to the linear-parabolic model [see for example Ref.(1-3) ]. The qualitative effects of water and sodium on the oxidation kinetics are also known (4). However, considerable differences exist for the reported oxidation rate constants both for studies considering impurities and for so-called clean oxidation studies.The purpose of this study is to consider the kinetic role of trace quantities of H20 (20-30 ppm) in O2 on the thermal oxidation of (111), (110), and (100) oriented single-crystal silicon. Experimental ProceduresSample preparation.--Chem-mechanically polished silicon wafers measuring 3.2 cm in diameter and 0.025 cm thick with (111), (110), and (100) orientations supplied from three different vendors were utilized for this study. Both n-and p-type Si wafers with resistivities ranging from 0.5 to 10 ohm-cm were used. For a given orientation and under identical oxidation conditions no systematic differences in oxidation rate were found for different type or resistivity silicon in the above resistivity range.Prior to oxidation each experimental Si wafer was cleaned according to the following schedule: (i) deionized H,_,O rinse until H20 resistivity is 18 mohm-cm;(ii) basic peroxide rinse at 65~ with ultrasonic agitation (NH~OH:H,_,O2:H20----1:1:5); (iii) repeat step 1; (iv) acidic peroxide rinse at 65~ with ultrasonic

In many studies of oxidation kinetics, it has been observed that silicon‐dioxide growth in dry oxygen in the thin film regime (<500Å) is faster than predicted by the linear‐parabolic description of the growth of thicker layers. Oxidation‐rate enhancement in the thin film regime was studied in the 800°–1000°C range for a variety of substrate orientations, doping densities, and oxygen partial pressures using in situ ellipsometry. The results were reported in part I of this paper. In this part, the physical mechanisms previously proposed to explain the rate enhancement are discussed. No single model was found to apply under all experimental conditions. A new understanding of the growth‐rate enhancement in the early stages of silicon oxidation in dry oxygen is introduced.

No abstract

A parallel beam reflection technique has been developed in our laboratory for measuring intrinsic residual stress for Si wafers thermally oxidized at temperatures of 600–1150 °C. A detailed description of this technique is provided, and stress values calculated for thermally grown SiO2 films on Si are consistent with those reported in the literature. Low temperature thermal oxidations resulted in compressive intrinsic SiO2 stresses greater than 4×109 dyn/cm2.