A Study of the Mechanisms Involved during Dissolution of Silica in Liquid Silicon

Abstract

By the application of well-known thermodynamics, growth kinetics, diffusion processes, and experimental results, a reasonable explanation for the behavior of silica in contact with liquid silicon is presented. Decomposition of silica to solid SiO, and diffusion of silicon and oxygen ions through solid SiO, play important parts in the discussion. The consequences for Czochralski growth of silicon crystals are outlined.Oxygen incorporation in Czochralski-grown silicon crystals has been the subject of numerous investigations, particularly over the last 15 years. Most of the investigations have been directed toward studying the influence of different growth parameters on the oxygen distribution in the crystals, e.g., see Moody (1) for a recent review; more fundamental work has been carried out by Kaiser and Breslin (2), Kubaschewski and Alcock (3), and the present authors (4, 5). In spite of these studies, there is no known investigation into the nature of the interaction that takes place at the interface between molten silicon and silica. Some experiments were done by Zulehner and Huber (6), and Chaney and Varker (7), and it was concluded that cristobalite grows on the SiO2/Si(1) interface. This opinion was opposed in Ref. (5), where it was concluded that the phase growing at elevated temperatures is solid SiO. This conclusion was based on thermodynamic calculations and measurements of the solubility of oxygen in molten silicon. However, none of the authors made any serious treatment of the reaction mechanisms involved, which is the subject of the present paper. ExperimentalSilica ampuls were filled with polycrystalline chemical vapor-deposited (CVD) silicon, 300 fl-cm, obtained from Wacker Chemitronic GmbH. The ampuls were evacuated, sealed, and heat-treated at 1420~ for 15 min, 1, 2, and 9h. The temperature was kept within _+ 2~ and measured by a Pt-Pt 10%Rh thermocouple calibrated at the eutectic point of the iron-carbon system. After the heat-treatment, the samples were subsequently quenched in cold water. The samples were mounted, ground, and polished metallographically so that the structure at the interface, Si/SiO2, could be examined. A sample from the bottom of a silica crucible, from which a high-resistivity phosphorus-doped silicon crystal had been grown, was treated in the same way as well as an ampul in which solid CVD silicon had been heat-treated for 8h at 1390~ ResultsIn Fig. 1-5, the results of the metaUographic investigation are presented. No picture from the sample heattreated for 15 min is presented, as it did not contain any information that cannot be obtained from the other micrographs. The very light phase covering the top part of the pictures is silicon, and the gray phase at the bottom is the unaffected SiO2. At the interface between these phases, the reaction products form an irregular pattern. In Fig. 1, lh heat-treatment, an isolated spot of a gray phase, herein called the A phase, has been formed and has grown into the silica. A layer of silicon between the unaffected silica and the edges...

Reaction at the interface between Si melt and a Ba-doped silica crucible

“…The mechanism of cristobalite formation with Ba doping was much different from the ordinary brownish rings generated from reaction of an undoped silica glass and Si melt. In the case of ordinary brownish rings without Ba doping, growth of the brownish rings or cristobalite depends on oxygen transportation in Si melt near the interface between silica glass and Si melt, and the driving force of oxygen transportation is considered as follows [3]. There is an oxygensaturated layer in Si melt near the interface between Si melt and inner surface of a silica crucible (amorphous SiO 2 ).…”

“…The supersaturation of oxygen in Si melt near the cristobalite (or nucleus of that) results in oxygen precipitation and finally results in growth of cristobalite in Si melt. It indicates that the ordinary brownish rings with cristobalite inside are generated in the Si side at the interface between Si melt and silica glass or silica crucible [3,7,10], which results in release of the brownish rings or cristobalite from the inner surface of a silica crucible because there is violent convection in Si melt in CZ-Si crystal growth. However, in the case of Ba doping, especially with a relative high concentration, inner surface of a silica crucible crystallizes to cristobalite uniformly from inside of silica glass or crucible.…”

“…[3][4][5][6][7]. It was reported that the density of the brownish rings depended on OH concentration in silica glass [8], and dopants with high concentration in Si melt also affected the reaction between Si melt and silica glass [6,9].…”

Thermodynamics and phase stability in the Si–O system

“…Reviewing the extensive literature in Appendix A, only very few works mention the existence of a crystalline compound ÔSiO(cr)' [108][109][110][111][112][113]. The works of [108,109] are discussed in [111] with the conclusion that the examined samples consisted of a stoichiometric mixture of crystalline Si + SiO 2 .…”

Analysis of the reaction at the interface between Si melt and Ba-doped silica glass

“…There have been several studies on the generation of the brownish rings related to their composition and generation mechanism [2][3][4][5][6], including the influence of impurities [7,8]. The brownish ring was identified to be an oxygen-deficient inhomogeneous cristobalite, SiO 2Àd [6,8].…”