This thesis focuses on growth and etching mechanisms of diamond. To validate a model that explains a mechanism, one needs experimental results, either from ‘real world’ experiments, or from computer simulations. For the ‘real world’ experiments, we use gas phase methods to grow and etch diamond. The computer simulations we use are Monte Carlo methods to simulate the growth, and continuum simulations to reproduce etch patterns.

The thesis consists of two parts. Part I describes the influence of nitrogen on the growth of CVD diamond. On {001} faces grown in the presence of nitrogen, it induces step bunching. This phenomenon is discussed in chapter 2, where we present a mesoscopic Monte Carlo model for impurity-induced step bunching. The patterns of bunched steps resulting from the simulations and those observed on the {001} diamond surfaces appear to be a remarkably identical. We have found a parameter set that is consistent with the observations from our diamond growth experiments. The paradox in predicted and observed growth rates of these faces is explained by the combination of two effects. On the one hand, nitrogen on the surface retards the step propagation by its lower reactivity. On the other hand, sub-surface nitrogen promotes the growth by activating surface reconstruction bonds via N-donor electrons.

Chapter 3 and 4 deal with polycrystalline diamond grown by flame deposition in the presence of nitrogen. The growth rate in the central area of flame deposited diamond layers can be increased by more than a factor two by addition of relatively small amounts of nitrogen in the source gases. Furthermore, upon nitrogen addition, the central area of the diamond deposits becomes less homogeneous in morphology and growth rate. Diamond crystallites with {001} top faces are obtained with nitrogen concentrations above 2.5 ppt, but the process window to obtain continuous {001} textured layers by the flame technique appears to be quite narrow. For high nitrogen concentrations the films become increasingly deteriorated and separated by voids with amorphous features. Generally, if a {001} textured surface is formed it consists of differently sized, seperately grown crystallites bounded by flat {001} top facets and rough, deteriorated {111} side facets. In chapter 4 we demonstrate that the occurrence of such a morphology is in contradiction with the van der Drift model of evolutionary selection. From the observations we deduce a new growth mechanism in which deterioration of the {111} diamond faces by enhanced secondary nucleation and twinning and the presence of a decreasing deposition rate from the higher to the lower parts of the growing layer form the basis for the development of the {001} textured layers. Although most pronounced in flame deposition, we argue that this deterioration-gradient growth mechanism is valid for diamond CVD in general.

Part II is devoted to etching of diamond. Both {001} (chapter 5) and {111} (chapter 6) diamond is etched in the temperature range of 700E C - 900E C. Different oxidative methods are used: gas phase etching using ‘dry’ oxygen and using ‘wet’ oxygen, and liquid etching in molten potassium nitrate. In contrast to theory, the etched {001} surface, which is a K-face according to the PBC theory in all cases shows steps and behaves as an F-face. Although the exact atomic arrangement of the oxygen on the diamond {001} surface is unknown up to now, this indicates that they are strongly stabilised by the adsorption of oxygen. The rate-limiting step of the etching process is the removal of carbon atoms at the kink positions of the steps by reaction with adsorbed oxygen. For {111} faces, gas phase etching in ‘dry’ oxygen gives completely different results from gas phase etching in oxygen/water vapour. ‘Dry’ oxygen etching produces a roughened surface, which is in conflict with the PBC theory. The divalent oxygen causes a reduction in stability of the surface atoms, whereby step and kink atoms become at least as stable as surface atoms, which may lead to chemical roughening of the {111} face. ‘Wet’ oxygen etching results in etching via monoatomic steps and well-defined etch pits. Most pits have a ‘positive’ orientation, whereby the step atoms are bounded by two dangling bonds. The inclination of these pits depends on the etching temperature. Surface X-ray diffraction experiments, described in chapter 7, show that the atomic structure of these surfaces fits well with a single bond cleavage, –OH terminated diamond {111} surface with ordered water layers on top. Small relaxations are found in the first four layer spacings, whereby the contraction between the first two bilayers agrees with earlier measurements. Liquid etching results in a surface morphology dominated by ‘positive’ etch pits with rounded corners. It appears to be a good method to reveal the two types of dislocations ending on {111}. By applying the classical kinematic wave theory to the morphology of the pits, we deduced that for etching in liquid potassium nitrate, surface diffusion is not the rate-limiting step.

Chapter 8 deals with diamond etched in wet oxygen at very high temperatures, 1000E C - 1200E C. In this regime we find that two etching mechanisms are competing. On the one hand the global etching mechanism at the step edges, where diamond is directly converted to CO(2), on the other hand the local etching mechanism induced by the catalytic action of oxygen, where starting from a carbon nucleus, diamond is converted to graphite dots, which can coalesce to a closed layer.

Trigons are revisited in chapter 9. Experimental step patterns of interacting trigons on a natural diamond {111} surface allow us to formulate an analytical expression for the step velocity for etching as a function of the orientation of a step on the {111} surface. Using this step velocity function a continuum simulation of the evolution of step contours, based on the classical kinematic wave theory, reproduces the experimental step patterns very well. We conclude that apparently the necessary assumptions to obtain this result are valid: surface diffusion and bulk diffusion are not important at all during the etching process and two-dimensional nucleation is not at all dominant over step flow even for large terraces in between steps.