Artificial neural network algorithm for analysis of Rutherford backscattering data

Rutherford backscattering (RBS) is a nondestructive, fully quantitative tec hnique for accurately determining the compositional depth profile of thin f ilms. The inverse RES problem: which is to determine from the data the corr esponding sample structure, is, however, in general ill posed. Skilled anal ysts use their knowledge and experience to recognize recurring features in the data and;elate them to features in the sample: structure. This is then followed by a detailed quantitative analysis. We have developed an artifici al neural network (ANN) for the same purpose, applied to the specific case of Ge-implanted Si. The ANN was trained with thousands of constructed spect ra of samples for which the structure is known. It thus learns how to inter pret the spectrum of a given sample, without any knowledge of the physics i nvolved. The ANN was then applied to experimental data from samples of unkn own structure. The quantitative results obtained were compared with those g iven by traditional analysis methods and are excellent. The major advantage of ANNs over those other methods is that, after the time-consuming trainin g phase; the analysis is instantaneous, which opens the door to automated o n-line data analysis. Furthermore, the ANN was able to distinguish two diff erent classes of data which are experimentally difficult to analyze, This o pens the door to automated on-line optimization of the experimental conditi ons.