Surface Roughness and Critical Exponent Analyses of Boron-Doped Diamond Films Using Atomic Force Microscopy Imaging: Application of Autocorrelation and Power Spectral Density Functions

Abstract

The evolution of the surface roughness of growing metal or semiconductor thin films provides much needed information about their growth kinetics and corresponding mechanism. While some systems show stages of nucleation, coalescence, and growth, others exhibit varying microstructures for different process conditions. In view of these classifications, we report herein detailed analyses based on atomic force microscopy (AFM) characterization to extract the surface roughness and growth kinetics exponents of relatively low boron-doped diamond (BDD) films by utilizing the analytical power spectral density (PSD) and autocorrelation function (ACF) as mathematical tools. The machining industry has applied PSD for a number of years for tool design and analysis of wear and machined surface quality. Herein, we present similar analyses at the mesoscale to study the surface morphology as well as quality of BDD films grown using the microwave plasma-assisted chemical vapor deposition technique. PSD spectra as a function of boron concentration (in gaseous phase) are compared with those for samples grown without boron. We find that relatively higher boron concentration yields higher amplitudes of the longer-wavelength power spectral lines, with amplitudes decreasing in an exponential or power-law fashion towards shorter wavelengths, determining the roughness exponent (α ≈ 0.16 ± 0.03) and growth exponent (β ≈ 0.54), albeit indirectly. A unique application of the ACF, which is widely used in signal processing, was also applied to one-dimensional or line analyses (i.e., along the x- and y-axes) of AFM images, revealing surface topology datasets with varying boron concentration. Here, the ACF was used to cancel random surface “noise” and identify any spatial periodicity via repetitive ACF peaks or spatially correlated noise. Periodicity at shorter spatial wavelengths was observed for no doping and low doping levels, while smaller correlations were observed for relatively higher boron concentration. These semiquantitative spatial analyses may prove useful in comparing synthesis techniques and varying compositional makeups of diamond films and other technologically important electronic materials. These findings in terms of critical exponents are also correlated with traditional Raman spectroscopy and x-ray diffraction structural properties, thus helping to provide insight into the growth kinetics, albeit in reverse manner.