Effect of Carbon Dioxide and Temperature on Passive Film Parameters of Superduplex Stainless Steel

Superduplex stainless steel has been frequently employed in new sites of Brazilian Pre-Salt. In these environments, chloride concentration, temperature and carbon dioxide are normally present in higher levels than those at sea water at room temperature. In these conditions, it is expected that the passive films of stainless steel also show modifications. To better understand such modifications, samples of superduplex stainless steel UNS S32750 were submitted to electrochemical impedance measurements in brine media, at two temperatures and under presence/absence of carbon dioxide. The electrochemical impedance results were initially tested using the Kramers-Kronig transform and subsequently fitted by equivalent circuit employing constant phase elements - CPE. Moreover, to quantify the effect of each factor (temperature, chloride, carbon dioxide and microstructure) on the equivalent circuit, their parameters were tested applying statistical analysis. Significant effect of carbon dioxide and temperature was found on related parameters of passive film for heat-treated samples.


Introduction
Commercial companies face challenging situations when exploiting oil and gas in offshore fields whose access is difficult and the environment may be extremely aggressive.Hydrocarbon sources existing on Brazilian Pre-Salt, for example, present temperature and salinity, besides harmful gas, in which the use of ordinary carbon steels become unfeasible since the corrosion effect is intensified by the local conditions.Another aggressive factor is the presence of carbon dioxide that promotes the pH reduction due to the H 2 CO 3 formation 1 and it well-known to be a major problem in corrosion behavior of iron based alloys [2][3][4][5][6] .An alternative is the use of corrosion resistant alloys like superduplex stainless steel that exhibit high corrosion resistance and good mechanical properties 2 .Such alloys have thin passive film, essentially constituted by Cr 2 O 3 , cohesive and self-regenerative, which is responsible for the excellent corrosion resistance 7 .The superduplex stainless steel shows, in addition to excellent pitting resistance number, small grain of austeno-ferritic microstructure and alloy elements as Cr, Mo, N and Ni, which assure superior properties 8 .Nevertheless, the heat-treatments used during the production of equipments may cause the precipitations of deleterious intermetallic components like sigma phase, making the steel more susceptible to localized corrosion 2 .
To better understand the passive film behavior of superduplex stainless steel at different conditions of environment and microstructure, measurements of electrochemical impedance spectroscopy (EIS) were performed.Such technique comprises applying a small amplitude potential excitation signal to the interface at different frequencies and registering the corresponding linear current response.The ratio between the Fourier or Laplace transforms of the applied perturbation and response signal defines the impedance 9 .Once the experimental diagrams of electrochemical impedance are obtained, it is possible to fit them with the help of equivalent electric circuits, EEC, comprising elements like resistors, capacitors and others.The overall electrical impedance of these EECs hence matches those of experimental electrochemical measurements.Moreover, each element from equivalent circuit must be related to physical aspects of the studied sample, if possible.In the present paper, the parameters of equivalent circuit were evaluated in order to investigate the effects of temperature, carbon dioxide, salinity and heattreatment on the passive film of UNS S32750 stainless steel.

Material and Methods
Sixteen experiments that compose a 2 4 orthogonal experimental design, four factors at two levels, are shown in Table 1.Numbers 1 and 2 stand for the two heat-treatments performed at 800°C for 15 minutes or 2 hours respectively.Table 2 presents the nominal chemical composition of UNS S32750.Figure 1 shows the microstructures of heat-treated stainless steels during 15 minutes (a) and 2 hours (b).Sample of Figure 1a (15 min) presents incipient intergranular sigma precipitates that were clearly observed at the alpha/gamma interface, while sample 1b (2 hours) reveals considerable amount of sigma phase.
The first heat-treatment simulates welding processes and the second mechanical hot-work process 1 .For the CO 2 factor, number 1 represents the absence of gas whilst number 2 its presence.The NaCl factor comprises the electrolytic solution of sodium chloride used in concentrations of 70 and 140 ppt (parts per thousand, simulating the high concentration of chloride that can be found in regions of the Brazilian Pre-Salt layer).Finally, the working temperatures were 30 and 60 °C thermostated thanks to a double wall electrochemical cell.

Kramers-Kronig transforms
The increase of EIS technique use in diverse systems results in the appearance of a vast variety of spectra, making it difficult to evaluate, by simple inspection, if the results are valid or contains some experimental errors.The Kramers-Kronig (KK) transformations are performed to validate EIS experimental according to four conditions 10 .The condition of causality demands that the system response is only sensible to the imposed perturbation; the linearity implies the use of sufficiently small amplitudes, about 10 mV, to assure tha tresponse is independent of perturbation amplitude but linear; the stability requires that the system return to its initial condition when the perturbationis ceased; and finally, the impedance must be a continuous function with finite values to any frequency, including frequencies close to zero or infinite.
If the four conditions above are obeyed, then the following Kramers-Kronig equations are valid 10 .
Where ω is the angular frequency (ω= 2πf) and the others are defined as usual for EIS.
Equations 1, 2 and 3 allow recalculating the components of impedance values from the EIS experiments results and therefore evaluate their validity.In the present paper the

Materials Research
Effect of Carbon Dioxide and Temperature on Passive Film Parametersof Superduplex Stainless Steel EIS300 Electrochemical Impedance Spectroscopy software was used for KK analysis.

Equivalent circuit for stainless steel
After some fitting attempts of impedance diagrams using different equivalent circuit models, the one that better represented the experiments results is shown in Figure 2 and was already performed in literature to represent stainless steels interface 11,12 .This circuit was implemented with in ZView 3.2c software.A constant phase element (CPE) circuitry was utilized, and its individual impedance formula is represented in Equation 41 ( ) Where α and Q are CPE parameters and j is the imaginary coefficient.The physical meaningof this element changes according to α values.If α = 1 CPE behaves like a capacitor and the parameter Q is then equivalent to a capacitor C. When α < 1 the system shows a behavior that has been frequently attributed to metal surface heterogeneity 13 even if several other causes for CPE behavior to appear have been proposed 14 .When α = 0.5 CPE is equivalent to Warburg element that is performed to represent diffusion processes, and if α = 0 CPE behaves like a resistor 11,12 .
The meaning of elements used in equivalent circuit is related hereafter according to literature 11,12 .R Ω represents the electrolyte resistance, Q 1 represents the capacitive behavior of passive film and R 1 the resistance of ionic flux, Q 2 represents the capacitive behavior of the metal interface and R 2 the charge transfer resistance.

Statistical analysis
In order to optimize the evaluation process of experimental results, the Statgraphics Centurion XVI software was used to perform the full factorial analysis of the 2 4 orthogonal experimental design.The microstructure and CO 2 factors were classified as qualitative and the NaCl and temperature factors were classified as quantitative, as shown in Table 1.The statistical analysis reveals the individual effects, as well as its synergisms, and the significances.The experimental runs were entitled as 1 to 16 according to their conditions, as presented in Table 3.

Results and Discussion
The results of KK analysis suggested that the EIS experiments obeyed the four necessary conditions to validate the electrochemical impedance measurements since the associated errors were acceptable.The values of goodness of fit provided by software were between 17 × 10 -6 for experiment 4 and 455 × 10 -6 for experiment 14. Figure 3 graphically depicts the KK results by Bode and Nyquist diagrams taking experiment 9 as an example since the goodness of its fit falls by the middle of the determined range (288 × 10 -6 ).The residual errors from real and imaginary impedances are dependent on frequency 15 but were small and fluctuated around zero which qualitatively corroborates the goodness of the results.
Besides the similarity between the experimental data and KK fitting observed on Bode and Nyquist diagrams, was also noted, in all experiments, that the residuals errors from real and imaginary impedances were higher in low frequencies between 10 -3 and 10 -2 Hz.This behavior is due to the passive film evolution and could be normally expected.
The equivalent circuit from Figure 2 was able to fit the EIS experimental results with good precision.The χ 2 (chi-square) deviation values presented order-of-magnitude of 10 -4 as shown in Table 3 and are in accordance with literature 11 .In Figure 4 are presented the Bode and Nyquist diagrams of equivalent circuit fitting for experiment 1.For the sake of conciseness, the other diagrams are not showed.
Table 4 summarizes the fitting values of parameters used in equivalent circuit and the respective errors for the 16 experiments.The R Ω , Q 1 and α 1 parameters presented the smallest errors and maintained such values low in all performed fitting.The other parameters showed, in average, error values slightly higher, notwithstanding in some situations such values had been considerable higher as the Q 2 parameter of experiment 1 and 2, and R 1 for experiment 9.
To a significance level of 5% the factorial analysis exhibited statistically significant variation to Q 1 , α 1 and  In this analysis, the variable effect that increase the factor significance value is described as positive, and the opposite behavior is identified as negative.In Figure 5, for example, if the CO 2 level increases, the variable Q 1 raises its value.The microstructure, NaCl, temperature and CO 2 factors were respectively represented as A, B, C and D. Figures 5 and 6 suggest that the carbon dioxide influences the passive film.Indeed, it clearly appears that it alters only the significant values of Q 1 and α 1 parameters which are related to the passive film behavior.Thus, Q 1 parameter (Figure 5) under CO 2 effect was about 2.5 times higher than the threshold value to be considered statistically significant.
The temperature/CO 2 (labeled CD) and microstructure/ temperature (AC) synergisms were also classified as significant in Figure 5, even if their significant values to these parameters were not so intense than CO 2 .In Figure 6 the temperature/CO 2 -CD synergism presented a significant value comparable to CO 2 for α 1 , but with a negative effect.This result is in good agreement with some recent findings in the literature showing that Q and α values are inversely and intrinsically coupled, that is, an increase (conversely decrease) in Q is mutually compensated by a decrease (increase) in alpha 14 as it appears from Figures 5 and 6.Finally, Figure 7 indicates that the temperature/CO 2 and microstructure/temperature synergisms are significant to α 2 .
Table 4 depicts the statistical analysis results.The Q 1 parameter values were substantially higher in CO 2 presence that, according to the literature 12 , indicates an increase of surface film defects.Indeed, capacitance is proportional to the film permittivity that increases with the presence of defects.Moreover, the increment of temperature  intensified the CO 2 effect on Q 1 .This information may be better observed in Figure 8 in which are represented the Q 1,2 /Q 1,1 ratio values as a function of temperature, where Q 1,2 represents the Q 1 parameter measured in CO 2 presence, and Q 1,1 represents the same parameter measured in the same conditions of microstructure, NaCl and temperature, but in CO 2 absence.
As all Q 1,2 /Q 1,1 ratio values are higher than unit, then for all cases Q 1,2 > Q 1,1 meaning that the presence of carbon dioxide was deleterious to passive film.Moreover, all curves present positive inclinations, so that the increment of temperature from 30 to 60 °C resulted in the increase of Q 1,2 /Q 1,1 ratio, intensifying the deleterious effect of CO 2 .Besides, the higher linear slope suggests that the raise of temperature was more aggressive to microstructure 2, which contains higher microstructural heterogeneity and presents higher sensibility.It is noteworthy that the statistical orthogonal design considers   linear the factor effect, as depicted in Figure 8.Thus, each data point represents an average of eight experiments with a given factor at a single level.
Another parameter that presented interesting behavior relative to CO 2 presence/absence was α 1 , whose values were lower in most of experiments containing the gas.Such behavior is represented in Figure 9 and it was a priori expected.Indeed, since the presence of CO 2 induces a proliferation of defects upon the film, hence increasing the heterogeneity distribution, this is effectively expected to appear in the form of smaller values of alpha, which means, larger deviations from the "ideal" capacitance predicted for "perfect" films.
While α 1 values remained within 0.8 and 0.9, α 2 presented more variable values, ranging from 0.6 to 1. Thus the latter average was lower than α 1 , as shown in Table 4.Such result suggests that at the global metallic-electrolyte interface, with passive film defects, can be influenced by diverse phenomena.Therefore, the global impedance is hardly represented by a simple equivalent circuit when a parallel diffusion across the film defect occurs, and α 2 values would be close to 0.5.Moreover, Q 2 , R Ω , R 1 and R 2 factors did notexhibited statistically significant variation to a significance level of 5%.
An additional effect of bubbling CO 2 is the reduction of pH, approximately from 7 to 4 1 .This reduction changes the valence of certain metals, as nickel, and chiefly the molybdenum.For high chloride content and moderate temperature, Wang et al 16 have remarked the strong stability of Mo(III) for large pH range at 80 °C.This solubility can worse the film properties from 30 to 60 °C as observed in Figure 8.
It is worth noticing that in the present work we were interested in identifying which variables can, isolated or couple to others, be considered as harmful parameters for the corrosion performance of heat-treated stainless steel in these media, regardless the theoretical reaction models behind it.Therefore, the comparison with theoretical reaction models is beyond the aims of the present paper.

Conclusion
The electrochemical impedance spectroscopy diagram analysis of heat-treated superduplex stainless steel UNS S32750 indicates that carbon dioxide presence is harmful to the studied passive film.The gas presence increases the defects on steel passive layer and makes it more heterogeneous.Moreover, this deleterious effect is intensified at higher temperature thus the combination of these two factors, CO 2 and temperature, further reduces the protective characteristics of films tested by electrochemical impedance and fitted by constant phase element circuit.

Figure 8 .
Figure 8.Effect of CO 2 and temperature on Q 1 parameter.Q 1,1 without and Q 1,2 with CO 2 .

Table 4 .
Fitting of equivalent circuit parameters.