AVS 55th International Symposium & Exhibition | |
Surface Science | Wednesday Sessions |
Session SS1-WeA |
Session: | Structure of Oxide Surfaces and Oxide Heterostructures |
Presenter: | M.S.J. Marshall, University of Oxford, UK |
Authors: | M.S.J. Marshall, University of Oxford, UK M.R. Castell, University of Oxford, UK |
Correspondent: | Click to Email |
It is well established that misfit strain between epitaxial islands and their substrates significantly influences the shape evolution of the islands as they increase their volume during growth. We have studied the growth of anatase islands on SrTiO3 surfaces. Extended annealing in UHV causes the surface region of single crystalline SrTiO3(001) to become enriched with TiO2. This results in the formation of epitaxial islands of anatase TiO2 (001). These islands are studied using UHV scanning tunneling microscopy (STM) and UHV scanning electron microscopy (SEM), which reveals the changes in morphology during growth induced by misfit strain. Screw dislocations observed with STM on some of the islands enable the rapid growth of micron-sized square islands. Starting from a square island, two types of shape transitions are observed. In the first, above 1000°C, the square anatase islands elongate in length and narrow in width. This growth behavior follows the established Tersoff and Tromp model of strain relief in epitaxial islands.1 In the second growth mode, below 1000°C, the islands relieve strain by the formation of trenches in the middle of each side of the square, thereby evolving into crosses. This shape arises because a lower annealing temperature imposes a kinetic constraint on the detachment of growth units necessary for island narrowing. While it might be expected that the growth of the notches would proceed to form four squares, each of optimal dimension, this does not occur. The shape transitions occur because the strain energy term in square islands increases more rapidly with volume than the surface and interface energy terms. In our system the islands are widely spaced, thus avoiding inter-island interactions, but there are still substantial differences between the theoretically predicted critical island size of a few 10s of nm edge length and our observed 1µm length. This can probably be explained because of partial strain relief due to dislocations in the anatase islands.
1J. Tersoff, and R. M. Tromp, Phys. Rev. Lett. 70, 2782 (1993).