Artificial Neural Network: A tool for Rapid Quantitative Elemental Analysis Using Neutron Activation Analysis

Abstract

An artificial neural network (ANN) has been trained with real-sample NAA (neutron activation analysis) spectra of irradiated materials. Following the training stage ANN was applied to a subset of similar samples thus obtaining the elemental concentrations by NAA method. Results obtained with the ANN method are in good agreement with results obtained from NAA techniques, showing the high potentiality of ANN in NAA quantitative analyses. KEYWORDS:Artificial Neural Network, Neutron. I.INTRODUCTIONDevelopments from artificial intelligence like artificialneural networks, ANN, can be regarded as an engineeringprocedure emulating the human brain activity. It can bedefined as a set of non-linear and non-stationary interconnectionof elemental processes able to carry out at least oneof the following functions: training, remembrance and generalization,or, abstraction of substantial properties. Thefundamentals of the ANN technique have been described elsewhere[1-2] and several commercial and free codes area accessible nowadays.Applications of ANN to atomic and nuclear physics hasincreased during the last two decades [3][4][5][6][7][8][9][10][11][12][13], mainly inproblems related to nuclear reactors. The most used paradigms in artificial intelligence applications to nuclear scienceand particle physics are the expert system, general algorithms, fuzzy system, neural networks and hybrid system.Some applications to α, β, and  spectra have beenreported in the last decade. Further application ofANN in other cases where strong non-lineal effects arepresent like in the spectral analyses generated in analyticaltechniques like PIXE and XRF (X-ray fluorescence)are scarce and in BIXE spectra (beta induced X-rayemission) remain unexplored [14,21].Application of ANN is found in several issues where similar patterns have been represented by many spectra with specific differences. A typical case is presented in the elemental analysis of samples of unknown composition by neutron activation analysis (NAA) technique. The corresponding spectra represent the peaks containing known energy with variable intensities holding necessary information for the evaluation of the primary concentrations present in each sample. NAAis a nuclear technique based on the detection of -rays emitted by radionuclides neutrons [2][3][4][5]. Activation analysis differs from the other methods such as chemical analysis; it depends on the properties of the nucleus and not on the behavior of electrons in the outer shells of the atom. In general, NAA has many advantages over many conventional methods such as simultaneous analysis of multi-component samples and high degree of sensitivity for the majority of elements, many elements can be determined at levels of parts per billion.The basic essentials required to carry out analysis of samples by NAA are; source of neutrons, suitable instrumentation for detecting gamma rays, and a detailed knowledge of the reactions that occur when neutrons interact with target nuclei.One of the factors that define the quality of t...