_{1}

^{*}

This paper describes the corrosion behavior of aluminum, copper, and mild steel when exposed to chloride media using both electrochemical noise analysis (ENA) and electrochemical impedance spectroscopy (EIS). Analysis of electrochemical noise (EN) data demonstrated the need for removal of drifts in both potential and current fluctuations. Statistical analysis such as noise resistance, localization index, skewness and kurtosis has been evaluated. Noise resistance showed a good agreement with polarization resistance. Fast Fourier transformation (FFT) has been applied to convert EN data from the time domain to the frequency domain. Spectral noise plots showed a good agreement with impedance spectra for the different alloys determined at the same exposure time. Spectral and statistical analysis can extract useful information from EN data.

Electrochemical noise is a general term for the random fluctuations in current or potential, which occur as an electrochemical process proceeds. The measurement of electrochemical noise (EN) for corrosion studies was first described by Iverson in 1968 [_{sn}) obtained from analyzing the EN in the time and frequency domains have been discussed [

A theoretical and experimental analysis of the effects of trends in the potential and current noise fluctuations observed during measurement of EN data have been studied before [

The test solution was 3.5% NaCl. Three different alloys have been selected for this study, mild steel 1018, aluminum 2024 and pure copper 99.9%. These alloys will represent three different corrosion mechanisms when expose to 3.5% NaCl solution. The exposed surface of the electrodes were polished with SiC paper up to 1200 grit, washed with distilled water and dried in air.

EN were performed in a set-up with electrochemical cell consisted of two identical working electrodes with an exposed area of 1.0 cm^{2} for each electrode and a saturated calomel electrode (SCE) as a reference electrode within a Faraday cage. An AutoCAD DSP device (ACM Instruments) was used to collect potential and current fluctuation simultaneously. Potential and current fluctuations were obtained, with a sampling rate of 2-point s^{−1} during a time period of 1024 seconds, which fixed the frequency range (Δf) in region between 1 Hz and about 1 mHz. The instrument noise was tested and shown to be with no influence on the noise measurements. A polynomial trend removal method was performed to remove the direct current (DC) drift contained in the noise data and a Hann window was applied before computing the PSD plots. The analysis of EN data in time and frequency domain was developed using Mathcad. The time-series data of the potential and current fluctuations were trans- formed into frequency domain by using the fast Fourier transformation (FFT) method.

EIS measurements were made with AutoCAD DSP device (ACM Instruments) and by applying a sinusoidal voltage signal of 10 mV over a wide frequency range of 10^{4} - 10^{−3} Hz at open circuit potential (Ecorr) and room temperature.

The potential and current fluctuations were recorded simultaneously as described in the experimental section.

Accordingly, polynomial trend removal method was applied to remove the DC trends in this study. EN data for copper exposed to NaCl for 1 day after removing the trend (de-trended data) are shown in

Noise resistance (Zn = σV/σI) is defined as the ratio of the standard deviations of the potential fluctuations to the current fluctuations [

Eskew and Iskew are the skewness of the potential and current fluctuations, respectively, which describes the asymmetry of the probability distribution of the EN data. For a symmetric distribution the skewness will be zero, positive values indicate that the data are skewed to the right from the normal distribution, which means that the distribution is concentrated in its leftmost part with along thin tail to the right, while negative values mean that the distribution is skewed to the left. Ekurt and Ikurt are the kurtosis of the potential and current fluctuations, respectively, which is a measure of the shape of the ENA data. The kurtosis will be 3 for normal distribution for values larger than 3 the normal distribution will be sharply peaked and for values less than 3 it will be flatter. Skewness and kurtosis are calculated as [

where N is the number of points used.

It has been assumed that LI values can be used to differentiate between types of corrosion, for example LI close to zero suggests general corrosion, while Il close to 1 corresponds to localized corrosion (9). The values obtained in this study indicate that Il cannot be used to differentiate between types of corrosion, which is in agreement with our previous results (14). It was concluded that Il could not be used to determine the corrosion mechanisms

The experimental EN data were transformed from the time domain into the frequency domain by fast Fourier transform (FFT) [_{sn} (f)) can be estimated according to [

where V_{FFT} and I_{FFT} are the Fast Fourier Transformer functions of the potential and current noise, respectively, and V_{PSD} and I_{PSD} are the corresponding power spectral density (PSD) plots. SV and SI are the slopes of the PSD plots of potential and current noise, respectively. The spectral noise resistance (Z_{sn}) could be defined as the DC limit of Z_{sn} (f) plots, in the same way like Zp, which is defined as the limit of impedance when the frequency tends to zero [_{sn} plot for raw data will be higher in amplitude than the de-trended one (

agreement with our previous results, these results also show how the DC trend greatly affects noise data calculated in the frequency domain [

In this paper, the corrosion behavior of MS 1018, Al 2024 and pure copper 99.9% have been investigated using EN and EIS techniques. The EN data was affected by drift during the measurement period, which was removed using polynomial trend removal method. Zn was calculated after analysis of EN data in the time domain and showed an agreement with Zp obtained from EIS analysis. The analysis of EN data in the frequency domain are in good agreement with those obtained by EIS. Analysis of EN data demonstrated that EN was successfully able to compare between different alloys as indicated by the difference of Zn values obtained for Al, Cu and MS. Analysis of ENA data did not reveal significant changes in the distribution of data that remained normal in all cases as indicated by the skewness and kurtosis data. Increases of Zn and the spectral noise impedance were ob-served for Al 2024 and mid steel, while for copper the spectral noise impedance did not agree with the electrochemical impedance. LI obtained for the three metals exposed to NaCl were close to zero for Al 2024, copper and mild steel. It was concluded that EN data did not provide mechanistic information for the material/solutions system studied.