2021 Frank Newman Speller Award: Corrosion Mechanisms of Steel and Copper in Engineering Applications

The stability domain of Type 304 (UNS S30400⁽¹⁾) stainless steel (SS)/5 wt% sodium chloride (NaCl) system at 25°C and 60°C was analyzed using electrochemical impedance spectroscopy (EIS) data. The electrode was polarized to a potential located in the pitting region, where impedance measurements were obtained. Although Type 304 SS performance in chloride-containing environments has been widely studied,¹  little information is available on its pitting corrosion stability using EIS analysis. To make good use of the EIS method in the field of pitting corrosion, where instability conditions are the common situation, it is interesting to analyze the transfer function using a complex nonlinear least squares method and to apply the Kramers-Kronig (KK) relations to assess the robustness of the impedance data.

In a published paper, the pitting corrosion of Type 316L SS (UNS S31603) in a NaCl solution was studied using the EIS method,²  concluding that stability conditions were not fulfilled for electrode processes, which was attributed the existence of surface relaxation phenomena. The analysis performed on the electrical equivalent circuit (EEC), with negative resistance and capacitance, was consistent with the instability of the Type 316L/NaCl system.

As can be seen from Equation (3), a nonrational transfer function (Z(s)) is approximated by the sum of first-order systems with parameters a_i and τ_i. This can provide excellent fits of the impedance data within a frequency range so long as a sufficient number of terms is used.⁴  Note that in the time domain, Equation (3) means that a non-exponential behavior is modeled by a sum of exponential functions of the form: (a_i/τ_i)exp(–t/τ_i).

An austenitic Type 304 SS was studied with a chemical composition (wt%): 19.7 Cr, 9.4 Ni, 0.036 C, 0.003 S, 1.7 Mn, 0.28 Si, 0.022 P, balance Fe. Specimens were embedded in cold epoxy resin and mechanically polished with successive SiC down to 600 grain size. A 5 wt% NaCl aqueous solution was used as electrolyte, and experiments were performed at 25°C and 60°C.

EIS measurements were obtained on specimens anodically polarized after 30 min exposure to the test solution, using a scan rate of 0.16 mV/s, from the rest potential to the pitting region, at 0.5 V_SCE (saturated calomel electrode). The EIS method was used in the frequency range from 10 kHz to 1 mHz, with a logarithmic sweeping frequency of five steps per decade. EIS involved the imposition of a 10 mV amplitude sine-wave. A platinum gauze was the counter electrode.

Figure 2 shows two typical Nyquist plots for Type 304 SS immersed for 30 min in a 5 wt% NaCl solution and polarized at 0.5 V_SCE. The Nyquist plots show a similar shape for both tested temperatures. A capacitive behavior is defined, the high frequencies show a depressed semicircle, a capacitive loop is observed at intermediate frequencies approximating an inductive-type loop over the real axis, and a third capacitive loop is observed at low frequencies forming a second semicircle or a straight line in shape. The low-frequency portion of the Nyquist plots (below 0.63 Hz) show a significant amount of scattering, with no evident relaxation process; therefore, these results have not been considered in the figure. Figure 2 also shows the regression of Equation (3) considering only five terms (n = 5) to simulate impedance data. No more terms were necessary because satisfactory agreement was obtained between the experimental and simulated data. Table 1 includes the parameters used for Equation (3).

The poles of the transfer function of Equation (3) are obtained as s_i = −1/τ_i. As can be seen from Table 1, τ_i exhibits a negative value, indicating a pole located in the right half of the s plane. This signifies instabilities and relaxation processes for the Type 304 SS/5 wt% NaCl system.

When τ_i is negative (pole located in the right half of the s plane), the term: exp(–t/τ_i) grows indefinitely with the time (t → ∞) and the impulse response does not decay to zero, i.e., instable system.² 

The comparison of experimental plots with the plots calculated by the KK relations is a validation test for robustness of impedance measurements. Figure 3 reports the results obtained with the experimental impedance plots from Figure 2(a). Figure 3, obtained using the KK relationships, is visually not close to the experimental plots and does not satisfy KK relationships, signifying instability and surface relaxation processes.

It is concluded that the shape of the Nyquist plots for the Type 304 SS/5 wt% NaCl system reveals capacitive behavior for both temperatures tested: high frequencies show a depressed capacitance loop, a second capacitive loop is observed at intermediate frequencies, and low frequencies show a third capacitive loop. A nonrational transfer function has been approximated by a rational transfer function, and its analysis indicates instability for the Type 304 SS/5 wt% NaCl system. Additionally, the impedance data does not fulfill KK relations, confirming instabilities and surface relaxation processes.

Steel concrete reinforcements form a thin, passive oxide layer under alkaline conditions (pH>12) in the concrete framework pore solution. However, the initiation of corrosion in reinforced concrete structures occurs when the pH of the concrete matrix is neutralized by carbonation or by the increased presence of chloride ions, which penetrate concrete cover through the pores.^6-10 

Different methods have been used to enhance concrete durability, including the use of different admixtures and alterations to the concrete composition. A promising approach is through the use of new raw materials suitable for alkaline activation (AA); fly ash (FA) and slag are used to create new binding materials termed alkaline cements.¹¹ 

The EIS method was used to study the alkaline activated fly ash (AAFA) mortar/carbon steel system proposing a model for measurements obtained in the high-frequency and the low-frequency range. In many cases the transmission line model (TLM) is used, where the interpretation of results by available commercial programs is not possible, due to the non-uniform distribution of alternating current on the porous nature of the oxide film generated on the steel reinforcement surface.^12-13 

Carbon steel with a chemical composition (wt%) of 0.20 C, 0.22 Si, 0.72 Mn, <0.010 P, 0.022 S, 0.13 Cr, 0.13 Ni, 0.18 Cu, and balance Fe was studied. The cementitious material used was AAFA (FA type F) using an 8.0 M NaOH solution (pH 13.9). A 2 wt% NaCl was added to the AAFA. All of the specimens were stored at room temperature in an atmosphere of high relative humidity (RH) (∼95%) for 730 d.

The i_corr values calculated using the Stern-Geary equation (i_corr = B/R_p) were 2.55 µA/cm² and 2.60 μA/cm², where R_p =R_w + R_u, R_w is the resistance of the electrode/electrolyte interface, R_u is the resistance at the pore base, and B = 26 mV, for 270 d and 730 d, respectively. These i_corr values correspond to a 29.57 µm/y and 30.13 μm/y corrosion rate for 270 d and 730 d, respectively. The stable values exhibited in the corrosion rate of the carbon steel electrode may be attributed to the access of O₂ as the controlling factor for the corrosion kinetics.

Figure 4 shows the TLM for the region of a cylindrical pore constituted by the porous layer on the steel reinforcement. Figure 5 shows Nyquist plots for steel embedded in AAFA mortar for 270 d and 730 d experimentation. A not well defined semicircle at high frequency, a depressed semicircle at intermediate frequency, and a tail at low frequency can be observed (see Figure 5). An excellent fitting was obtained.

It is concluded that the TLM can be used to describe the evolution of damage of reinforcing steel embedded in AAFA mortar. A penetration depth was estimated for carbon steel very high, 30.13 μm/y for 730 d experimentation. The long-term corrosion process was controlled by the presence of O₂ at the porous corroded electrode.

Corrosion inhibitors for steel in concrete can be used by addition to the cement paste (admixture corrosion inhibitor [ACI]), or by applying with brush to the concrete surface diffusing through the pores of the concrete (migrating corrosion inhibitor [MCI]).⁶  Examples of ACI inhibitors are amines and fatty acid esters, which act through a double mechanism: first by reducing the ingress of the chlorine ion, through the hydrophobic property of the esters; and second, by forming a protective barrier layer through the ion-dipole interaction (^δ+H−N^δ−).¹⁶ 

Organic-based corrosion inhibitors have been used as an alternative to inorganic compounds because they are nontoxic and have low environmental impact, biodegradability, good corrosion inhibition efficiency, and a low cost. The inhibition of these compounds depends on the molecular structure and the affinity and compatibility with steel. Organic MCIs diffuse to the anodic or cathodic sites and adsorb on the steel surface through covalent bonds and polar groups.¹⁷ 

Amines, aliphatic carboxylic acids, and saturated fatty acids have been used as ACIs of steel embedded in concrete.¹⁸  Benzoate and its amino derivatives and dicarboxylates,¹⁹  and carboxylic acids have been used as organic ACIs as well.²⁰  The curing process of ordinary Portland cement (OPC) can be affected by the air entrainment of organic compounds.²¹  Thus, the incorporation of organic ACSs causes a decrease in the compressive strength of the concrete (∼15%).²²  Heterocyclic compounds are widely used, and the heteroatoms of the imidazole compounds contribute the lone pair electrons to the vacant iron orbital.²³ 

The use of nano/micro capsules to store an inhibitor prevents premature leaching of the active substances and reduces the loss of effectiveness.²⁴  Based on the concept of chemical self-healing, sodium citrate (C₆H₅O₇Na₃·2H₂O)²⁵  and sodium monofluorophosphate (Na₂PO₃F) (MFP) encapsulated in ethylcellulose have been used as an ACI.²⁶  The encapsulation of a corrosion inhibitor implies that the release of the active substance only takes place under the presence of aggressive agents and in conditions in which the steel is prone to corrosion.²⁴ 

The use of biological corrosion inhibitors is based on biomolecules such as those derived from green plants that are used as ACIs and, also, are used by applying the inhibitor to concrete, acting as MCIs.²⁷  Other inhibitors are made up of yeast extracts, bacterial cells, or other organic substances.²⁸  Biosurfactants emitted by bacterial cells are potentially interesting as ecological mixtures because they contain in the same molecule parts that act as hydrophilic group (head) and parts that act as hydrophobic group (tail). Using bacterial cell biosurfactants, the hydrophobic groups can be formed by an acid, peptide, or mono-, di-, or poly-saccharides.^29-30 

Because the reactions in Equations (11) and (12) are faster than those related to chloride ion, a stable and protective layer of lepidocrocite is generated. This ability to oxidize ferrous to ferric iron produces more insoluble layers, for instance, K_sp (Fe(OH)₂) = 10^−16.6 and K_sp (Fe(OH)₃) = 10^−37.2 with a thickness of 17 Å to 50 Å.³² 

Inorganic corrosion inhibiting compounds are toxic and have low corrosion inhibitor efficiency if they are not added in the appropriate amount. In the use of nitrite-type inhibitors, the ratio [NO₂⁻]/[Cl⁻] needs to be near unity, so that the corrosion inhibition efficiency is maximized.³³  Hybrid inhibitors consisting of inorganic quaternary ammonium salts or phosphates and organic compounds such as imidazoles have also shown to impart high inhibition efficiency. Additionally, tertiary amines for repair of structures have been used.³⁴ 

It is believed that phosphate inhibitors react with the iron ions generated in the corrosion process or with other ions present in the mortar, such as calcium forming calcium phosphate (Ca₃(PO₄)₂) precipitates, filling the pores and cracks of the mortar.^35-38 

Calcium monofluorophosphate (CaPO₃F),³⁹  as well as zinc monofluorophosphate (ZnPO₃F) have been used as a corrosion inhibitors for steel in 3 wt% NaCl solution;⁴⁰  in both cases, the inhibitors are shown to be effective. Manganese monofluorophoaphate (MnPO₃F) has been found to exhibit a mixed corrosion inhibition for steel in 3 wt% NaCl solution.⁴¹  Aluminum tri-polyphosphate (AlH₂P₃O₁₀·2H₂O) has been used as an ACI inhibitor with good results.⁴² 

Disodium hydrogen phosphate (Na₂HPO₄) (DHP) in simulated concrete pore (SCP) solution and in mortar acts as an anodic corrosion inhibitor.^36-37  Trisodium phosphate monohydrate (Na₃PO₄·H₂O) (TSP) in mortar acts as a mixed corrosion inhibitor⁴³  and even as a cathodic-type corrosion inhibitor.⁴⁴  It has been observed that TSP behaves as an anodic type inhibitor, which penetrates through the pores and it is necessary to use high concentration,⁴⁵  which can cause rheological changes to the mortar. TSP increases the critical concentration threshold of the ratio ([OH⁻]/[Cl⁻]) for steel in SCP and in mortar.

As an example of phosphates acting as corrosion inhibitors, Figure 6 (top) shows corrosion current density (i_corr) vs. time for MCI specimens: steel reinforcement embedded in OPC mortar and immersed in 0.2 M of soluble phosphates (MFP, DHP, or TSP) solutions or distilled water (control test) contaminated with 3 wt% NaCl solution. Figure 6 top (MIC inhibitors) shows that the i_corr values are situated at levels of low corrosion risk or passivity. The best corrosion inhibitor behavior for MCIs is shown by the MFP and DHP compounds, with i_corr values of <0.1 μA/cm². The control specimen was actively corroding. Figure 6 (bottom) shows the i_corr vs. time for ACI specimens: steel rebar embedded in OPC mortar, prepared by blending 3 wt% of solid MFP, DHP, or TSP with OPC, water, sand, and contaminated with 3 wt% NaCl. The best corrosion inhibitor behavior is shown by the MFP compound, with i_corr values situated at levels close to low risk of corrosion or passivity. Specimens containing DHP or TSP compounds also showed i_corr values situated at a low risk of corrosion. TSP specimen after 30 d shows an acceleration of i_corr. For the control test, i_corr values were between 0.3 μA/cm² and 0.9 μA/cm², similar to those obtained using the TSP compound at the end of the experiment.

The common corrosion protection of steel using organic coatings was exemplified using lacquered tinplate, a material widely used in the food industry. Tinplate is a thin sheet of carbon steel coated with pure tin on both faces. Cathodic treatments with sodium dichromate (CDC) are used to passivate by reduce the rapid growth of tin oxides and protect the coating from corrosion. This treatment causes the formation of a complex layer, composed of Cr metal as well as Cr and Sn oxides, which confers corrosion and sulfide staining resistance to the tinplate in addition to limiting Sn oxide propagation, also affecting lacquer adhesion (the end treatment) to the substrate.⁴⁶ 

The lacquer film protection depends on its physical-chemical characteristics, its compatibility with the canned product, and the characteristics of the substrate. The composition and thickness of the oxides formed have been shown to affect the lacquer adhesion. Different models have been proposed to describe the relation between adhesion, the oxide proportions, and the overall Cr content.⁴⁷  Titanium compounds were substitutes for Cr compounds as passivation agents, given their similar chemical characteristics, but with lower toxic and environmental implications. Passivation with titanium compounds has been studied, and titanium sulfate, titanium, and potassium oxalates treatments were developed.⁴⁸ 

Sheets of tinplate without passivation treatment were passivated using K₂TiF₆ and NaNO₂ aqueous solution for 10 s. Additionally, tinplate specimens with conventional CDC passivation were used as a reference. Tinplate specimens were lacquered in an industrial plant, applying a conventional epoxyphenolic lacquer 7 g/m² to 8 g/m² and cured at 205°C for 10 min. In order to obtain information on lacquer detachment mechanism, some specimens previously subjected to the T-peel test were analyzed by using scanning electron microscopy (SEM) and energy dispersive x-ray (EDX) techniques.

Lacquer corrosion-protection capacity was evaluated by EIS method. Measurements were taken using an electrochemical double cell,^47,49  using 0.1 M citric-citrate buffer test solution (pH 3.5), at room temperature.

Figure 7 shows the micrographs and the corresponding EDX spectra of two complementary points of a Ti-passivated tinplate.⁵⁰  According to the relative proportions of the metallic elements on the surfaces, obtained from the EDX spectrum, the lacquer detachment takes place in the proper lacquer, as happens in the Cr-passivated tinplate with a normal good adherence.

Figure 8 depicts two typical Nyquist plots obtained for the two lacquered tinplate studied exposed to 0.1 M citric-citrate buffer test solution for a period of 10 d. A semicircle can be observed for Ti-passivated lacquered tinplate (see Figure 8[a]), and a semicircle at high frequency accompanied by a straight-line diffusion tail or part of a second semicircle at low frequencies can be observed for Cr-passivated lacquered tinplate (see Figure 8[b]). In this latter, the high-frequency semicircle represents the bulk properties of the lacquer and the low-frequency tail represents the lacquer/substrate interfacial properties;⁵¹  the electrolyte diffuses through the pores of the lacquer and arrives to the metal surface occasioning corrosion.⁵²  Table 2 includes the fitting data for Ti- and Cr-passivated lacquered tinplate using the EECs of the insets of Figure 8. The resistance of the electrolyte (R_s) was of the same order (1,860 Ω·cm² for Ti-passivated and 1,170, Ω·cm² for Cr-passivated). The dielectric properties of the lacquer (barrier effect) were higher for Ti-passivated than for Cr-passivated, and the C_dl parameter was 6.5 × 10⁻¹⁰ F/cm² and 1.1 × 10⁻⁹ F/cm² for Ti- and Cr-passivated lacquered tinplate, respectively. The charge transfer resistance (R_ct), inversely related with the corrosion rate, was higher for Ti-passivated lacquered tinplate (1.3 × 10⁷ Ω·cm²) than for Cr-passivated tinplate lacquered tinplate (9.5 × 10⁵ Ω·cm²). Finally, for Cr-passivated lacquer the tail at low frequency was simulated by the Warburg element (W) (see Figure 8[b]), using a CPE (constant phase element) with Y_CPE = 3.6 × 10^–7 F/cm² s^1–α and an α parameter of 0.521 (see Table 2), which may indicate a diffusion process through the porous lacquer. It is concluded that the use of Ti is an excellent alternative for a passivation treatment of tinplate.⁵³ 

In air-conditioning systems copper tubing experiences pitting corrosion after relatively short service periods, during post-installation leakage tests, and during degreasing and stamping processes. This unusual type of localized corrosion usually occurs in thin-wall copper tubes, especially when copper is deoxidized, and is known in the literature by several names, including ant-nest corrosion, cavernary corrosion, formicary corrosion, and pinhole corrosion.^54-55  Ant-nest corrosion morphology is characterized by the development of longitudinal pits that form interconnecting microscopic caverns which run in random directions that contain porous copper oxide in directional pits connected by tunnels. Typically, tunnels begin to form on the copper tubing surface and continue into the tube wall. Perforation usually occurs in weeks or months, and not years. During the corrosion process, copper oxide is deposited on the inside walls of the tunnels, causing them to turn blacks. This type of corrosion is considered to be caused by the decomposition products of chlorinated organic solvents used in degreasing, cleaning, and picking treatments of the copper tubes used in the manufacturing or joining processes as well as in certain types of synthetic lubricant oils used during the copper tubing stamping process.⁵⁶  The oxygen provided by the copper oxide deposits serve to aid the decomposition process further, allowing the chemical reactions to take place. In these processes, carboxylic acids were formed by the oxidation of the respective alkyladehyde vapors.⁵⁷ 

Copper tubes used in commercial air-conditioning units were analyzed to determine the causes for premature failure by corrosion during leakage testing after installation as well as during leakage tests after installation and during their first 2 months in service, beginning with the inner surface. The copper used in the tubes was phosphorus deoxidized, having 0.019% to 0.022% P and ≥99.90% Cu. It had been degreased using perchloroethylene, pickled in sulfuric acid, and rinsed in distilled water before undergoing a pneumatic leakage test at 30 kg/cm² pressure. Figure 9 depicts the inner surface of a corroded copper tube.⁵⁹  The micrograph shows some microscopic caverns with connecting tunnels as well as pits with typical shape of ant-nest corrosion.

Figure 10 shows deconvolution of the high-resolution XPS spectrum of C1s for the inner surface of a corroded copper tube. Four components can be observed: hydrocarbon (C–C) elements are found at ∼284.8 eV, ether (C–O–C) elements are found at ∼286.5 eV, acetone (>C=O) elements are found at ∼288.0 eV, and acid (OH>C=O) components are found at ∼289.0 eV. The presence of the acid component may be ascribed to the existence of organic acids such as formic (HCOOH), acetic (CH₃COOH), and propionic (CH₃CH₂COOH), in the corroded copper tube. Organic acids present high corrosiveness on copper.⁵⁸  Carboxylic acids may result from the hydrolysis of the self-evaporating synthetic ester-based lubricant oils. Similarly, organic acids from the detergent used in cleaning may attribute to the organic acids.

Volatile organic substances, such as carboxylic acids (formic, acetic, propionic, and butyric acids), are produced by many everyday substances, such as vinegar-containing seasonings including vegetable oil dressing and grain vinegar, cosmetics such as eau de cologne, adhesives in synthetic building materials, insecticides, woods and certain paints, plastics, rubbers, and resins, which can cause the deterioration of copper.⁶⁰  The patina, or corrosion products on copper, exposed to the atmosphere for prolonged periods contain organic acid anions that constitute about 0.1% to 1% of the total ion concentration.^61-62 

Figure 11 shows the corrosion rate of copper exposed to 0 to 300 ppm (parts per million) of carboxylic acid vapors (acetic, formic, propionic, and butyric) for 21 d.⁶⁵  The corrosion rate increases with the vapor concentration. Acetic acid displays the highest corrosion, followed by formic, butyric, and propionic. There is a direct relationship between the corrosion rate and the acidity of the acid vapors, except for formic acid, as evidenced by the dissociation constants in aqueous solution (1.77 × 10⁻⁴ for formic acid, 1.76 × 10⁻⁵ for acetic acid, 1.34 × 10⁻⁵ for propionic acid, and 1.54 × 10⁻⁵ for butyric acid). For formic acid, the corrosion products formed from copper exposure to acid vapors create a more protective layer compared with the porous nature of the corrosion products formed upon exposure to the other acids.^63-67 

Despite the lack of experimental evidence, the development of similar intermediate compounds is possible for the other acids.

The new trends and the challenges to be solved in the near future on the corrosion research lines are the following. (1) For the EIS method, it is desirable to design input signals to ensure both an optimal identification of the parameters of the EEC and a sufficiently short measurement time. Specifically, using optimal experimental design, it is possible to select the amplitudes and frequencies (multisine excitation signal) at which the experimental impedance measurements are performed. Second, the use of noninteger order calculus to describe relaxation processes governed by fractional dynamics, such as those involving anomalous diffusion or nonideal capacitance. In time-based methods, new functions (e.g., Mittag-Leffler or Kohlrausch-Williams-Watts [KWW] functions) could be used for an accurate description and interpretation of corrosion processes. (2) It is of interest to study the corrosion behavior of new generations of low-nickel content SSs as reinforcements to further the study of corrosion in concrete. Additionally, it is desirable for new eco-efficient cements using industrial by-products such as blast furnace slag (BFS) and alkali-activated (AABFS) or calcined clays (CC). The presence of sulfur in the AABFS systems leads to high dissolution of the steel reinforcement, thus results are not available in the literature on their corrosion behavior. The use of CC as an addition to the OPC is growing great relevance, due to the reduction in the amount of clinker used and the emission of CO₂, as well as the improvement in the resistance to chlorides and the alkali-silica reaction. (3) The passivation of tinplate using friendly treatments needs to be solved and the lacquer adhesion phenomenon is in need of in-depth studies. (4) Finally, regarding ant-nest corrosion: it is of interest to study the ant-nest corrosion using new generation, self-evaporating synthetic lubricant oils, and to compare with the conventional OAK 70–5MB or OAK HC–A oils. Additionally, it is of interest to study copper deoxidized with high- and low-residual phosphorous (DHP).

I would like to thank NACE International for this recognition.