Geometric modeling of homoepitaxial CVD diamond growth: I. The (100)(111)(110)(113) system F. Silva * , X. Bonnin, J. Achard, O. Brinza, A. Michau, A. Gicquel Laboratoire d’Ingénierie des Matériaux et des Hautes Pressions, CNRS, UPR 1311, 99 Av. J.B. Clément, F-93430 Villetaneuse, France Abstract : Plasma-assisted CVD homoepitaxial diamond growth is a process that must satisfy many stringent requirements to meet industrial applications, particularly in high-power electronics. Purity control and crystalline quality of the obtained samples are of paramount importance and their optimization a subject of active research. In the process of such studies, we have obtained high purity CVD diamond monocrystals with unusual morphologies, namely with apparent (113) stable faces. This phenomenon has led us to examine the process of CVD diamond growth and build up a 3D geometrical model, presented here, describing the film growth as a function of time. The model has been able to successfully describe the morphology of our obtained crystals and can be used as a predictive tool to predetermine the shape and size of a diamond crystal grown in a given process configuration. This renders accessible control of desirable properties such as largest usable diamond surface area and/or film thickness, before the cutting and polishing manufacture steps take place. The two latter steps are more sensitive to the geometry of the growth sectors, which will be addressed in a companion paper. Our model, applicable to the growth of any cubic network material, establishes a complete mapping of the final morphology state of growing diamond, as a function of the growth rates of the crystalline plane considered, namely (100), (111), (110), and (113) planes, all of which have been observed experimentally in diamond films. It is also possible to deduce transient behavior of the crystal morphology as growth time is increased. The model conclusions are presented in the form of a series of diagrams, which trace the existence (and dominance) boundaries of each face type, in presence of the others, and where each boundary crossing represent a topology change in terms of number of faces, edges and vertices. We validate the model by matching it against crystals published in the literature and illustrate its predictive value by suggesting ways to increase usable surface area of the diamond film. I. Introduction : Monocrystalline diamond CVD growth has been an active research topic in recent years. It is now possible to obtain diamond films of excellent quality with high deposition rates (10 - 60 μm/h), opening the door to industrial applications. However, significant obstacles remain, principally the limited size of the available substrates. It is therefore essential that usable surface area and film thickness be made as large as possible for manufacture of diamond electronic devices. Reaching these objectives requires good control of diamond crystal morphology during the growth. Indeed, for thick film growth, the occurrence of (111) faces on the corners of the crystal can lead to structural defects such as twinning or non-epitaxial crystallites, which can jeopardize the film quality. These defects generate large stresses inside the crystal which can even lead to film fracture during the growth or during the delicate cutting and polishing steps. Consequently, self-supported films described in the literature generally show an irregular or octahedral shape. Preventing such structural defects from occurring and controlling the stresses inside the growing crystal is therefore needed to raise the usable surface area and thickness of the monocrystalline films. In previous work, we reported that, during homoepitaxial CVD diamond growth on a <100> oriented substrate, crystalline faces (110) and (113) were observed on the edges and corners of the crystal respectively [1]. The opportune occurrence of the (113) faces on the crystal corners prevented the (111) faces, which were very often defect-prone, from being present. These unusual