This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: VOLUME 81, NUMBER 16 P H Y S I C A L R E V I E W L E T T E R S 19 O CTOBER 1998 Diffusion of Atomic Oxygen in SiO 2 D. R. Hamann Bell Laboratories, Lucent Technologies, Murray Hill, New Jersey 07974 (Received 19 May 1998) Density functional calculations using an a-quartz supercell as a model silica host identify the peroxy linkage as the lowest energy configuration of atomic O in SiO 2 , and find that its energy in this site and in interstitial molecular O 2 are nearly equal. Using ab initio molecular dynamics modified to converge to a saddle point, the barrier for concerted exchange of the peroxy linkage is found to be 1.3 eV. While O is generally believed to diffuse in molecular form in SiO 2 , measured diffusion activation energies are consistent with the peroxy exchange barrier. [S0031-9007(98)07404-3] PACS numbers: 66.30.Jt, 71.15.Pd, 81.65.Mq The growth of SiO 2 on silicon is a key process in mi- croelectronics technology. As dimensions of integrated circuit elements continue to shrink and oxide thicknesses of a few tens of angstroms are required, understanding this process on an atomic scale becomes increasingly impor- tant to achieve the required control. Growth kinetics are reasonably well explained by the Deal-Grove model , which postulates that gas-phase oxygen diffuses through the growing SiO 2 layer, and reacts at the Si-SiO 2 inter- face. The study of atomic O in SiO 2 was initially stimu- lated by the observation that it is more active in promoting oxide growth than molecular O 2 . Vitreous SiO 2 , as well as the many crystalline poly- morphs of SiO 2 formed at low pressures, consists of a continuous network of corner-sharing SiO 4 tetrahedra . While the random network topology of the vitreous form may be a key feature for certain properties, diffusion bar- riers are likely to be dominated by local bonding effects. The present choice of a supercell based on the a-quartz structure to study an “impurity” O in SiO 2 was moti- vated by this assumption and the following considera- tions: (i) An extended system avoids the surface effects inherent in cluster models. (ii) Elastic relaxation effects are expected to play an important role and cannot be mod- eled by a cluster. (iii) A supercell model of a vitreous sys- tem of computationally accessible size would of necessity contain many highly strained bonds, and be less repre- sentative of real vitreous SiO 2 than a crystalline model. (iv) Insofar as strained bonds may enhance diffusion, the crystalline model should provide an upper bound to the effective average vitreous energy barrier. Even a modestly sized supercell model of SiO 2 has many internal degrees of freedom which need to be relaxed in the presence of the added O (henceforth O*). SiO 2 mod- els based on parametrized classical force fields are com- putationally efficient, but typically include only repulsive O—O interactions , precluding covalent O—O bonding....
View Full Document
- Spring '09
- The American