Thin Solid Films 428 (2003) 144–149 0040-6090/03/$ - see front matter 2002 Elsevier Science B.V. All rights reserved. PII:S0040-6090 Ž 02 . 01244-0 Selective growth of SiGe quantum dots on hydrogen-passivated Si(100) surfaces V. Le Thanh *, Tam T.T. Ngo , Huy Bui , D. Bouchier , Tuyen T.T. Le , Khoi H. Phan a, b b a b b Institut d’Electronique Fondamentale (IEF), CNRS-UMR 8622, Universite Paris-Sud, Batiment 220, 91405 Orsay Cedex, France a ´ ˆ Institute of Materials Science, NCNST of Vietnam, HoangQuocViet St., Hanoi, Viet Nam b Abstract We report, in this paper, on a new method to produce SiGe quantum dots on Si(100) surfaces. Starting from the fact that the adsorption of hydride molecules (SiH , GeH ) requires free adsorption sites on the surface, the basic idea of our approach is to 4 4 limit the number of sites for molecular adsorption. We show that etching of Si(100) surfaces in ammonium fluoride (NH F) 4 solution initially produces a flat and dihydride-terminated Si(100) surface and that longer etching leads to the formation of microscopic (111) facets, which are regularly distributed along the surface. Hydrogen atoms are found to desorb completely from surface dihydrides at ;400 8C while those from monohydride-terminated (111) facets remain stable up to 650 8C. Thus, for growth carried out in the temperature range of 400–650 8C, the adsorption of hydride molecules occurs only on the sites that have been previously terminated by dihydrides, i.e. free of hydrogen. We show that SiGe islands with size being reduced down to y200 A can be achieved by using this new approach. ˚ 2002 Elsevier Science B.V. All rights reserved. Keywords: Quantum dots; Selective growth; Ammoniun fluoride; Facet structures 1. Introduction During the past few years, the fabrication of quantum dots (QDs) has attracted growing interest both from the experimental and theoretical point of view w1,2x. The main reason for this interest is due to expected improve- ments of optical and electrical properties of materials as compared to conventional two-dimensional heterostruc- tures. In order to obtain an efficient quantum confine- ment inside the dots, it is desirable that they have size in the quantum range (a few tens of nanometers) and a narrow size distribution. These two parameters of dots determine in fact the linewidth of photoluminescence peaks which is at the heart of the performances of QD- based devices. Different methods have been proposed to fabricate QDs, which can be in general classified in two catego- ries: the first uses one or several technological steps in the fabrication processes, including local epitaxy through *Corresponding author. Tel.: q33-6-62-90-28-15; fax: q33-4-91- 41-89-16. Present address: Centre de Recherche sur les Mecanismes ´ de la Crossance Cristalline (CRMC2), Campus de Luminy, Case 913, 13009 Marseille, France. E-mail address: lethanh@crmc2.univ-mrs.fr (V.L. Thanh). microshadow masks w3x, selective epitaxy on V-grooved w4x or on SiO -pattemed substrates w5x, and dry etching 2 after growth w6x. However, in addition to its technolog- ical complexity, success with this approach appears to be very limited since from the technological point of view the fabrication of ultrasmall structures that are defect-free remains a challenge. The second approach uses strain-induced transition from two to three-dimen- sional growth regime that occurs during growth in a highly lattice-mismatched heteroepitaxial system. It was indeed shown several years ago that the growth of a highly lattice-mismatched semiconductor onto a sub- strate could lead to the spontaneous formation of islands that are three-dimensional, epitaxial and coherently strained w7x. Using this approach, it has been shown that islands with sizes in the quantum range (y300A) could ˚ be achieved in InAs yGaAs w8x and InGaAs yGaAs w9x systems. Concerning the Ge ySi and SiGe ySi systems that have smaller mismatch (4% in the case of pure Ge on Si compared to 7% in InAs yGaAs system), islands with sizes larger than a thousand of angstroms were ¨ generally observed w10x. In this paper, we report on an alternative method to produce SiGe quantum dots on Si(100) surfaces. Start-