Step-structure dependent step-flow: models for the homoepitaxial growth at the atomic steps on Si(111)7 × 7

Abstract

Step-flow models of homoepitaxial growth at atomic steps on Si(111)7 × 7 are considered within the framework of the dimer-adatom-stacking-fault (DAS) model with the aid of a scanning tunneling microscopy (STM) image published by Köhler, Demuth and Hamers [J. Vac. Sci. Technol. A 7 (1989) 2860]. The image was taken at room temperature for a surface epitaxially grown at 330°C. From the image around the growing step, we find that the step flow is separated into two processes; the initial growth on the end zone of the lower terrace where the (7 × 7) DAS reconstruction is not formed due to its insufficient space, and the subsequent growth where the (7 × 7) DAS reconstruction at the lower step-edge has to be canceled to grow epitaxially. For the initial growth we propose step-structure dependent step-flow models, according to our classification of the atomic steps. For the subsequent growth, in addition to our previously proposed model for the cancelation of the (7 × 7) DAS structure during homoepitaxial growth on terraces (termed as the coalescent destruction model), we postulate the following four guidelines to construct models. (i) The destruction of the faulted halves and dimers at lower step-edge is the rate-determining step for step-flow, (ii) The epitaxial growth on the unfaulted halves at the lower step-edge is rapid, (iii) The faulted half at the “macrokink” is much easier to be broken than that at the straight step. (iv) The advanced step grows faster than the receded step does. By using the guidelines and the coalescent destruction model, we propose models for the subsequent growth at the U and F steps whose outward normal of the risers are the [1̄1̄2] and [112̄] directions, respectively. As the step-flows proceed with increasing Si deposition, the proposed model naturally leads to the following change of macroscopic shapes of the atomic steps: The straight F step converts to a zigzag step consisting of the U steps, while the straight U step maintains the straight shape. These models have been confirmed by images of transmission electron microscopy by Tung and Schrey and a recent STM observation during epitaxial growth by Hasegawa and coworkers. We conclude that the cancelation of the DAS reconstruction is the ratedetermining process for the step flows, and this results in the contrastive difference of the growth shapes between the U and F steps after prolonged step-flows.