INTRODUCTION
In this study, an analysis has been made of Electrochemical Noise (EN) signals of potential and current obtained on immersing samples of the alloy AA5083 (UNS A95083) in solutions of NaC1. These signals have been processed using the calculation of the Hurst exponent, tool derived from the theory of chaos that allows one to estimate the influence that events occurring in the past have on the appearance of new events at subsequent times. The results obtained have enabled the values of the Hurst exponent to be correlated with the predominant mechanism of corrosion.
The measurement of electrochemical noise is a technique that has been successfully applied for the study of the behavior of diverse metallic materials subjected to different types of corrosion. As is known, success in the application of this technique is largely determined by the validity of the mathematical method employed in the analysis of the data. Several different methods of analysis have been described in the bibliography, such as the employment of statistical parameters, analysis in the frequency domain, tools derived from the transformation of wavelets, or methods based on chaos theory. In accordance with different theories of chaos, the calculation of the so-called Hurst exponent (H) permits the quantification of the degree of auto correlation existing between the elements that form a data series, and the determination of the influence that events occurring in the past have on the emergence of future events. This H exponent was utilized for the first time by Hurst for the study of hydrological data, with the objective of designing a dam on the river Nile 1-3. Later, thanks to the work of 45 Mandelbrot ' , this mathematical tool has been successfully applied in many scientific fields.
As examples of the great versatility presented by this methodology, one can cite its employment for the analysis of solar activity 6'7 , for monitoring the evolution of the price of electricity 8 , for studying stock market fluctuations 9, in the field of particle physics ~° and in the studies of mechanical sliding in solids 11.
In the bibliography, one of the ways in which the H exponent can be useful is for detecting changes in the persistence of a data series. From a theoretical point of view, values of H=0.5 would indicate that the series analyzed is random; in other words, that each event included in the series is independent one from the other. On the other hand, if 0.5<H<1, the series analyzed is said to be persistent, in other words, the events that appear in the series are related. Thus, in this type of series, the emergence of one event would promote the initiation of the following one. On the other hand, if 0<H< 0.5, it can be said that the series is anti-persistent. It is a characteristic of this latter f~/pe of series that the emergence of one event delays or inhibits the initiation of a new event in the series .
As a consequence of its great versatility, the Hurst exponent has also been employed for the study of corrosion processes. Thus, the utilization of H as a parameter for evaluating the morphology of the pits generated in localized corrosion processes has been proposed 13. Similarly it has been employed in the analysis of data obtained by means of electrochemical noise measurement. In References 14 and 15, the use of H is proposed for evaluating the protective properties of organic coatings. More specifically, Moon and Skerry 14 postulate the existence of a linear relationship between the values of H, calculated from EN records, and the resistance to corrosion of the paints studied. In another study, Greisiger and Schauer is utilized the
