Image Analysis of Semicontinuous Metal Films
J. Boldy, R. Hrach · Superficies y Vacío · 1999
In this paper the morphology of discontinuous metal films with irregular objects and semicontinuous films is studied by a combination of computer experimentation and image analysis of film micrographs. For the discussion of physical mechanisms which take part during thin film growth, an atomistic model based on direct Monte Carlo simulation technique was performed. For the analysis of experimental photographs, as well as simulated micrographs, the basic algorithms of mathematical morphology are used. Inadittion it is proposed a new set of features for thin film structure description, which are based on coefficients of Haar wavelet transform. In the initial stage of their growth, very thin metal films deposited on dielectric substrates consist of individual islands. In this case a Volmer-Weber mechanism of threedimensional nucleation takes place. During further growth the islands increase, coalesce, create larger irregular objects and then form the semicontinuous metal layer. The basic information about film properties can be derived from the analysis of film photographs obtained from a transmission electron microscope. In the study of thin film morphology, two kinds of problems arise. It is necessary to find convenient methods for quantitative characterisation of film micrographs, and to interpret the derived characteristics from the physical point of view. For the study of initial stages of growth the methods based on mathematical morphology [1] are commonly used. Both the size and spatial distribution of islands in very thin films can be characterised fully by standard morphological methods (e. g. [2]). However, the accuracy of these methods decreases with increasing film thickness and only a limited number of them can be used for the study of semicontinuous films. In the present paper two tasks are solved: an attempt is made to extend the limits of some standard morphological methods to higher film thickness and to suggest new methods based on Haar wavelet transform. For both these tasks a computer experiment was suggested. 2. Models of film growth