Correlation and Spectral-Density Roughness Analysis of Surfaces Processed With Gas-Cluster Ion Beams
- PDF / 417,771 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 68 Downloads / 222 Views
R9.32.1
Correlation and Spectral-Density Roughness Analysis of Surfaces Processed With Gas-Cluster Ion Beams D. B. Fenner Epion Corporation, 37 Manning Road, Billerica, MA 01821 and Physics Department, University of Connecticut, Storrs, CT 06269 ABSTRACT Autocorrelation, height-difference correlation, and power spectral density are used to characterize various surfaces exposed to gas-cluster ion beams (GCIB), as well as for detailed comparison with simulated surfaces which are based on models with stochastic impact and surface mobility mechanisms. This work demonstrates a close correspondance between detailed statistical analysis of AFM images from actual surfaces and the results of the same types of analysis of simulated surfaces. Surface roughness prior to and after GCIB treatment is always found to be essentially fractal in nature. Phenomenological models with both continuum surface mobility and Monte Carlo impact accumulations are presented. Accurate simulation of smoothing requires a combination of these models. INTRODUCTION Roughening of surfaces by ion beams has been a nearly ubiquitous and undesired phenomenon in practical surface science. Smoothing, under particular circumstances, and the more general roughening behavior have been reported widely and ascribed to radiation-enhanced viscous flow of the surface and stochastic ion impacts, respectively [1-3]. Gas clusters consist of cryogenic, condensed droplets with diameters of just a few nanometers. Accelerated and ionized gas clusters (the GCIB) explosively evaporate upon impact with kinetic energy ~10 eV per constituent. Since this is far more than the van der Waals binding energy, ~0.1 eV, the collisions are highly inelastic. Smoothing of a large number of materials has been demonstrated, [4,5] finding an (acceleration-dependant) exponential decline in roughness for increasing fluence [6]. Figures 1 (a) and (c) are examples of GCIB smoothing [5,6]. Statistical analysis of atomic-force microscope (AFM) images revealed that the smoothing occurs from ~10 nm to ~10 µm [7]. Individual cluster impacts have been imaged by AFM [8]. These images show the craters to be ~5 nm in depth and 10-50 nm in diameter surrounded by a small rim of ejecta. Figure 1 (b) illustrates the nanoscale craters in an oxide film on silicon [5,8]. Monte Carlo (MC) methods utilizing simplified impact shapes have been reported for simulating cluster impacts [9]. Continuum surface-diffusion models have also been reported by us and others in attempts to model GCIB smoothing [10,11]. Comparisons of these simulations with only the most general features of AFM images of actual surfaces were reported. Here a detailed statistical analysis of images is reported along with new models that are able to reproduce many aspects of both GCIB roughening and smoothing [12]. EXPERIMENTAL METHODS and RESULTS The GCIB method of smoothing has been widely described elsewhere, and commercial apparatus was used here [5-8]. AFM images were obtained with a Digital Instruments Nanoscope III in tapping mode, and dig
Data Loading...
