Atomic Emission Spectroelectrochemistry: Real-Time Rate Measurements of Dissolution, Corrosion, and Passivation Open Access

Aqueous metallic corrosion is an electrochemical phenomenon and much of the theory and experimental techniques for predicting and assessing corrosion are based on the application of electrochemical concepts and techniques (for example, see Kelly, et al.).¹  Electrochemical measurements provide rich and meaningful kinetic information on the underlying faradaic processes, but are often insufficient to identify the chemical mechanisms of these processes, limiting their utility as predictive tools. As a complement to conventional electrochemical information, we would like to know how do the individual elemental constituents of a material react with the environment? Can we identify and quantify the specific fate of each element? What are the stoichiometries of dissolution, film formation, and electron exchange? Do alloy components dissolve selectively, leaving behind other components in the form of oxide films or dealloyed metallic layers? Can non-faradaic corrosion processes such as oxide dissolution or the release of intermetallic particles or metallic grains be detected and quantified?

These and other questions may be addressed by a relatively new technique, atomic emission spectroelectrochemistry (AESEC). This technique involves coupling an electrochemical flow cell to a downstream inductively coupled plasma atomic emission spectrometer (ICP-AES, also referred to as optical emission spectroscopy or ICP-OES), so that the concentrations of dissolved species may be followed as a function of time yielding a direct and simultaneous measurement of the elemental dissolution rates in real time. In this way, for spontaneous reactions the changes in the electrochemical potential may be correlated with elemental reactivity and the total faradaic current of conventional electrochemistry may be decoupled into partial dissolution reactions. In many situations, the evolution of surface composition may be determined in the form of elemental enrichment as oxides or dealloyed metallic films by application of a mass and charge balance to the AESEC data.

Interest in AESEC has increased in recent years and related techniques have been developed. The objective of this article is to serve as a basic tutorial in AESEC methodology. The instrumentation is discussed and the basic quantitative relationships between the electrochemical and spectroscopic measurements are presented, illustrated with numerous recent examples. Although the focus is on the AESEC technique as developed in the author’s laboratory, many of the principles will apply to other in-line techniques such as electrochemistry—inductively coupled plasma mass spectrometry (ICP-MS).

The term spectroelectrochemistry was first coined by Kuwana²  to describe the coupling of electrochemistry with UV-visible spectroscopic chemical analysis of the electrolyte to identify and quantify electrogenerated species in an optically transparent thin layer cell (OTTLE). In this way, the transient electrochemical measurements of potential and current (E, i) could be related to concentration transients in solution, yielding a precise mechanistic interpretation. AESEC provides similar information via the coupling of an ICP-AES downstream from an electrochemical flow cell (Figure 1). The electrolyte reacts with the material under investigation in the flow cell and the dissolved species released from the material are continuously analyzed downstream by ICP-AES. In this way, the concentration transients in the electrolyte may be compared with electrochemical data (E and i) as in the original OTTLE experiment.

The coupling of electrochemical dissolution with ICP spectroscopy has been around for a long time, for example as a detector of anodic stripping voltammetry.^3-4  However, the spectroelectrochemical application, that is to say, the measurement of concentration transients correlated with electrochemistry in order to understand the electrochemical reactions, was first developed in the early 1990s^5-6  at the Institut de Recherche de la Sidérurgie (IRSID) at Saint Germain-en-Laye, France. The development of time resolved emission spectroscopy had been pioneered at IRSID by Roger Berneron, who had previously developed depth profiling by glow discharge optical emission spectroscopy (GDOES) by combining an emission spectrometer with a Grimm lamp as a plasma source,⁷  and Pierre DeGelis, who had developed a prototype for AESEC, a non-electrochemical, open dissolution cell coupled to an ICP-AES.^8-9 

The AESEC instrumentation may be divided into three modules (Figure 1): (a) the electrochemical flow cell where the reactions of interest occur, (b) the plasma and spectrometer in which the products of the reactions are analyzed, and (c) the data acquisition system which gives a real-time presentation of elemental intensities as a function of time. A diagram of a flow cell commonly used in the author’s laboratory is shown to the left. The sample or working electrode (WE) with an exposed surface area (A ≈ 1 cm²) is exposed to a small volume (V ≈ 0.2 cm³) of continuously renewed electrolyte, the flow of which is maintained by a peristaltic pump. A porous membrane separates the reaction cell from a separate compartment where the reference (RE) and counter (CE) electrodes are placed. The porous membrane allows ionic conductivity while preventing bulk mixing of the electrolytes in the two compartments.

Fresh electrolyte enters at the bottom of the cell, reacts with the specimen, and is removed from the cell at the top containing all dissolved species coming from the specimen. The temperature of the reaction is controlled by placing the electrolyte reservoir in a thermal regulated water bath and by pumping the water from the bath through a hollow copper block positioned on the back side of the specimen.

Interfacial reactivity may depend directly on the hydrodynamics of the electrolyte at the interface. Elementary processes such as corrosion product precipitation, pH changes, oxygen diffusion, etc. may be diminished or enhanced by changing electrolyte velocity. Unfortunately, the range of possible flow rates is limited by the nebulization system of the ICP-AES. The flow rate of electrolyte (f) through the cell is usually controlled at around 1 cm³/min to 5 cm³/min (0.02 cm³/s to 0.08 cm³/s). Higher rates may be used by diverting a fraction of the electrolyte with a second pump as has been done for gas measurements.¹⁰  The average diffusion layer thickness has been determined by potentiostatic reduction of K₃Fe(CN)₆.¹¹  The limiting diffusion current corresponded to a diffusion layer thickness of approximately 200 μm to 130 μm at flow rates of 2.0 cm³/min to 5.0 cm³/min. The diffusion layer thickness increased sharply at lower flow rates.

After reaction, the electrolyte is aspirated into the torch of an ICP-AES which is used to determine the elemental concentration from the emission spectrum of the plasma.¹²  This is similar in concept to a rotating ring-disk electrode or a double channel flow cell in which the electrochemical detection at the ring or second electrode is replaced by the ICP-AES down stream from the flow cell. Dissolved species are desolvated in the plasma due to the high temperature (8,000 K to 10,000 K), molecular bonds are broken, and usually, the constituent elements are reduced to atomic species. Electronic excitation and de-excitation occur with the emission of characterisitic atomic line spectra. The intensity of the individual lines is directly proportional to the concentration of the element in the plasma, which is proportional to the concentration in the electrolyte.

There are numerous advantages to using ICP-AES as a detector. It is sensitive to most elements in the periodic table and quantification is straightforward using commercially available standard solutions. It is extremely robust and stable over long periods of time even with complex electrolytes containing high salt or total dissolved solid concentrations, or containing organics, oils, etc. The detection limits are low, on the order of a few ppb (µg/dm³) for most metallic elements, and the dynamic range of quantification usually spans five to six orders of magnitude making it possible to compare electrical current and dissolution rates over a wide range of conditions. Interferences between elements are rare due to the narrow linewidth of the emission spectra and the high resolution of the spectrometer used, and when they exist they may be readily accounted for and corrected by standard methods. An important advantage of ICP-AES is the near absence of matrix effects. This is due to the destruction of almost all molecular bonds in the plasma by the high temperature (≈8,000 K to 10,000 K) and their decomposition into atoms. The drawback to this is that all molecular information is lost: molecular speciation and elemental oxidation states must be determined or inferred using other information.

A more detailed diagram of the spectrometer in our laboratory is given in Figure 2. The Paschen Runge polychromator system with photomultiplier tube (PMT) detectors is ideal for AESEC in that each element may be monitored in a completely independent fashion by adjusting the sensitivity via the high voltage on each photomultiplier. The photomultiplier has the large dynamic range needed for AESEC operation and the elemental concentrations vs. time date are truly simultaneous without stepping from one element to the other and with no dead time between measures. The Czerny-Turner monochromator is used to select a single element that was not preinstalled on the polychromator system, or that requires a higher spectral resolution, hence lower detection limit. The electrochemical data (four analog signals) are also routed into the same data acquisition system so that current and potential are measured on the same time base as the spectroscopic data.

The AESEC system used in our laboratory is equipped with a rapid, simultaneous data acquisition system, originally developed by HORIBA Jobin Yvon for GDOES depth profiling analysis. The signals from the phototubes, monochromator, and polychromator are monitored simultaneously by the Quantum™^† software and data acquisition electronics. Three 16-bit A/D converters operating at a frequency of 250 kHz are used to continually monitor the output of 31 photomultipliers, 30 on the polychromator and 1 on the monochromator. The data are transferred to the computer after averaging over a user-defined integration period, usually set for 1 s. This means that each measured value of intensity corresponds to the average of 250,000 16-bit data points giving an extremely large dynamic range. Much faster data acquisition rates are possible for certain applications, for example the measurement of particle release (Section 4.3) was performed at a rate of 10 ms per point.

Alternative detectors and/or optical mounts are available, namely semiconductor detectors, such as charge-coupled device (CCD) or the charge injection device (CID) and/or echelle type optics. These systems have the advantages of being less expensive, may in principle measure more elements, and are usually more compact if echelle systems are used. The major drawback is a limited dynamic range: the sensitivity for a specific element is controlled by optimizing the integration time for that element, making it difficult to measure several elements simultaneously and truly correlating the signals. The CCD detection systems are also less sensitive in the ultraviolet wavelength range. An ICP-AES system with CCD detection system and a quasi-identical electrochemical flow cell was used by Mercier, et al., to investigate the corrosion^13-15  and anodization^16-17  of Al, although acqusition was limited to the analysis of a single element, Al.

Although this article concerns atomic emission spectroscopy, it is of interest to briefly review the use of inductively coupled plasma mass spectrometry (ICP-MS) for similar experiments. A recent review of the electrochemistry coupled to plasma mass spectrometry and optical spectroscopy has been presented by Cherevko and Meyrhofer.¹⁸  Their technique was first developed by Homozava, et al.,^19-21  and further developed by Ott, et al.^22-24  Their technique involved a noncontinuous operation in which aliquots were taken and analyzed at regular intervals. Continuous in-line experiments have been conducted using ICP-MS in particular using a scanning droplet cell (SDC) as developed at the Max Planck Institute in Düsselorf by Klemm, et al.^25-28  Using either the SDC or a conventional flow cell, ICP-MS has been used to investigate the corrosion of Mg, Al, and Ni alloys,^29-34  the kinetics of the degradation of heterogeneous catalysts,^26-27,35-38  and to detect partial currents during anodization.^39-44  Recently, Lopes, et al.,⁴⁵  moved away from the flow cell technique to directly sample the electrolyte in the vicinity of a rotating disk electrode with transfer to an ICP-MS to investigate the dissolution of Pt single crystals.

ICP-MS is similar in concept to emission spectroscopy except that ions drawn out of the plasma are usually detected with a quadrapole mass spectrometer. ICP-MS has the advantage of very low detection limits (on the order of parts per trillion [ppt]) and is sensitive to most elements in a single measurement. The drawbacks include less versatility as compared to ICP-AES in terms of electrolyte compatibility, being limited to relatively dilute solutions and acids. Plasma stability for ICP-MS is also an issue and usually an internal standard is required for long operating times. ICP-MS is also very prone to matrix effects: as the ions move away from the plasma, they may regroup into molecules creating mass interferences such as ArO⁺ and ArC⁺ which interfere with ⁵⁶Fe⁺ and ⁵²Cr⁺. These species may be reduced by reaction with a secondary gas in a “dynamic reaction cell.”²⁷ 

AESEC offers a direct measurement of the instantaneous dissolution rate of a material on an element by element basis. The electrolyte carrying the dissolved reaction products is aspirated into the inductively coupled plasma and the dissolved species are atomized and excited by collisions with electrons and argon ions. The relaxation of the excited atoms gives rise to the emission of radiation at different wavelengths associated with different elements. Through calibration with elemental standards, the emission intensity at a specific wavelength (I_λ) is converted into concentration: where I_λ° and κ_λ,M are the background intensity and the sensitivity factor for element M, respectively, for a given wavelength (λ). These values are determined by conventional ICP-AES calibration methods and of course depend upon the specific conditions of the plasma and nebulization system and must be determined regularly. The concentration is in turn converted into dissolution rate (ν_M): where f is the flow rate of the electrolyte and A is the exposed geometric surface area of the electrode. The flow rate is measured to <1% uncertainty for each experiment. For convenience, the dissolution rate may be presented as an equivalent faradaic current density (j_M) by application of Faraday’s law: where n_M is the number of electrons exchanged in the dissolution reaction and F is the Faraday constant where ν_M is considered to be in mol·s⁻¹·cm⁻². Note that by convention the rates are normalized to the geometrical surface area.

The plot of the elemental dissolution rates as a function of time for a given specimen-electrolyte composition is referred to as a dissolution profile. Direct insight into the mechanisms of dissolution may be obtained from the dissolution profile. First and perhaps foremost, the stoichiometry of the dissolution is obtained directly by comparing the ratio of the dissolution rates (electrolyte concentrations) with the composition of the material. This reveals whether or not the dissolution occurs congruently for all elements or preferentially for some elements. It also offers a direct measurement of the faradaic yield of an anodic dissolution process by comparing the electrical current density measured by the electrometer of the potentiostat (j_e) with the sum of the elemental dissolution current densities (j_Σ = Σ_M j_M).

The MgZn₂ phase occurs in Al alloys containing Zn and Mg, i.e., the 7000 series with ≈3% Zn and Mg, and in the Zn-Al-Mg coatings of galvanized steel, usually with ≈ 1% to 3% of Al and Mg. It is a very reactive phase and is frequently the most anodic phase when present in a multiphase system. Figure 3 shows the dissolution profile obtained during the anodic dissolution of a MgZn₂ pure phase intermetallic in a slightly alkaline electrolyte. The figure gives the system response to an applied potential (E_ap) of −0.8 V_SCE, applied at t = 0, approximately 200 mV anodic to the spontaneous corrosion potential. As shown in Han and Ogle,⁴⁶  cathodic reactions are negligible at this potential. Note that the current density is presented as j_e and ⁠, the latter being determined by numerical convolution of the former described in detail later in the Time Resolution and Convolution section.

This example clearly illustrates the basic quantitative principles of AESEC. The distribution of the faradaic electrical current density (j_e) between the dissolution of the two alloy components, Zn and Mg was nearly congruent and faradaic. Congruent dissolution is indicated by the ratio of the elemental dissolution rates, j_Zn/j_Mg = 1.92±0.03, very close to the MgZn₂ stoichiometry, and faradaic dissolution is indicated by the ratio of elemental dissolution to the anodic current, j_Σ/ = 0.97±0.02, very close to 1. The elemental dissolution rates are consistent with the stoichiometry of the material: x-ray diffraction (XRD) and electron backscatter diffraction (EBSD) revealed an essentially pure MgZn₂ intermetallic phase and ICP-AES measurements in an independent laboratory confirmed the 2:1 Zn:Mg atomic ratio (=1.98±0.02).⁴⁷ 

The reaction of a material with an electrolyte may involve faradaic processes which do not result in the immediate release of dissolved species and these processes are therefore undetectable by ICP-AES. Cathodic reactions and the formation of oxides are among the electrochemical processes which may not be monitored by AESEC. In the absence of surface charging, these processes may be quantified indirectly by a mass-charge balance. The faradaic current due to the undetectable processes is given by the difference current density (j_Δ): where the * denotes the numerical process of convolution, described in the Time Resolution and Convolution section. For a faradaic dissolution yield of 100%, j_Δ = 0. For j_Δ ≠ 0, it is sometimes possible to attribute j_Δ to either a specific cathodic or anodic process making possible a further breakdown of the electrical current density into elementary processes. Oxide formation may be determined when j_Δ > 0 and the cathodic current density is negligible,^48-52  or the cathodic current density may be determined when j_Δ < 0 and oxide formation can be ignored.^53-54 

The quantitative determination of oxide formation by mass-charge balance is well exemplified by the anodic dissolution of Cu and Cu-Zn (Figure 4) in a synthetic tap water.⁽¹⁾ Galvanostatic dissolution is shown to the left of the figure for a wt% 75.8%Cu-21.0% Zn alloy (CW724R brass [UNS C69300⁽²⁾]) during the imposition of a fixed galvanostatic current density of 40 μA/cm². The dissolution profile shows that the faradaic current density is distributed between the dissolution of Zn and of Cu, as for MgZn₂ (Figure 3), however in this case, the faradaic yield of dissolution is only 42% assuming Zn(II) and Cu(II) are the dissolved oxidation states. The charge involved in the “hidden” faradaic process, in this case the formation of oxides, may be calculated by integrating Equation (4), indicated by the pink shading in Figure 4.

Following their formation, the oxides were selectively dissolved at open-circuit in a deaerated 0.2 M citrate buffer, pH = 4.9, shown to the right of Figure 4. The quantities of Cu and Zn dissolving during this step corresponded to the total amount of oxide formed during the preceding anodic dissolution. (The quantity of accumulated Zn oxides was very low and essentially independent of time and was therefore ignored in the following analysis.) This is demonstrated in Figure 5, which compares the integral of the total current density (Q_e = j_e Δt/F) with the sum of dissolved Cu and Zn in the tap water solution and the citrate buffer as shown. The results for similar experiments with a Cu reference material are also shown.

From Figure 5, n values of 0.89 and 1.1 were determined for the formation of the Cu-oxide films on Cu-Zn and pure Cu, respectively, consistent with a mechanism in which Cu is oxidized to form insoluble Cu(I)-oxide and soluble Cu(II) as majority species. In separate postmortem experiments, the presence of Cu₂O was confirmed by Raman spectroscopy and XRD for pure copper,⁵⁶  brass,^55,57  and bronze.⁵⁸ 

This experiment also demonstrates the high surface sensitivity of the technique. The integrated quantity of oxide dissolved in the citrate buffer yields an approximate film thickness of 60 nm assuming the standard density of Cu₂O (6.0 g/cm³) and a uniform film over the 1 cm² of geometric surface area. This would be significantly less if the true surface area was used in the calculation considering surface roughness. The estimated average oxide formation rate is 0.25 mmol·s⁻¹·cm⁻³, which under the same assumptions, corresponds to a rate of 0.06 nm/s.

Mechanistic detail is apparent in the dissolution profile. The early stages of dissolution show a significant Zn dissolution rate which rapidly passes through a maximum while Cu dissolution increases slowly. This early Zn dissolution correlates with a dip in the open-circuit potential and may be interpreted as the early stages of dealloying which do not seem to be affected by the formation of the Cu₂O film, followed by Cu dissolution as Cu(II).

Frequently, one or more components of an alloy or complex material will undergo selective dissolution, leaving behind a surface layer enriched in the other components. The surface enrichment may be either an oxide in the case of passivation or a metal film in the case of dealloying. The AESEC method can sometimes be used to indirectly determine these surface enrichments via a mass balance. The total quantity of dissolved M at any time (Q_M(t)) may be calculated by integrating the dissolution rate of M:

If the bulk composition of the material is known and assumed to be uniform in depth, it is possible to calculate the total quantity of accumulated M enriched at the surface (Θ_M) relative to another element (N) which is assumed to be completely soluble: where x = (%M/%N) in either wt% or at% depending on the units of ν_M(t) in Equation (5). The first term of Equation (6), xQ_N, gives the hypothetical dissolution rate of M assuming congruent dissolution with N. If the dissolution of M and N were perfectly congruent, Θ_M = 0.

This measurement of the accumulated M by Equation (6) does not distinguish between insoluble oxidation products and dealloyed metallic films. The distinction may be made by considering the charge balance. The accumulated quantity of insoluble oxidation products may be determined in units of charge (Θ_e), and expressed in terms of a specific element M (Θ_M(z)) as follows: where z is the presumed oxidation state of element M (mol/cm²) in the solid oxidation product.

To illustrate these relationships, Figure 6 presents a complete kinetic analysis of the anodic dissolution of wt% Cu-42Zn in synthetic tap water at 80 μA/cm². This alloy undergoes dealloying much more readily as compared to the wt% Cu-21Zn (Figure 4)⁽³⁾, which was specially engineered to resist dealloying. In this case, the enrichment of Cu(0) (Θ_Cu(0)) may be determined as a function of time taking into account the repartition of excess Cu between dealloyed Cu(0) and insoluble Cu(I) corrosion products, presumably Cu₂O:

The dissolution profile is shown in the upper curve and the variation of the interfacial composition is shown in the lower figure. The quantity of accumulated solid Cu(I) oxidation products (Θ_Cu(I)) was determined by the mass-charge balance of Equation (4).

The Cu(0) enrichment in Example 3 was very small, on the order of a few nm, and proved difficult to confirm by a postmortem analysis. This was achieved, however, for the wt% Zn-68Al system (Al_5.2Zn) which underwent a mechanism of cathodic dealloying in 0.03 M NaCl at pH = 10.1.⁵⁹  The quantitative analysis of this reaction was obtained with a double-step potentiostatic experiment, shown in Figure 7. The dealloying reaction (left) occurs at −1.60 V_SCE (Equation [9]) where only Al dissolution was observed.

The anodic dissolution of the excess Zn(0) occurred at −1.00 V (right, Equation [10]). The quantity of Zn dissolved at this potential was 78% of that determined by mass balance. Also shown is the effective Zn dissolution current density when the alloy was stepped directly to −1.00 V without dealloying at −1.6 V; no Zn dissolution was observed, indicating that the Zn of the pure phase Al_5.2Zn has a more positive dissolution potential than that of the dealloyed, metallic Zn(0) formed by Equation (9).

The original idea of spectroelectrochemistry was to determine a direct relationship between the electrochemical current density and the concentration of electrogenerated species so as to interpret the instantaneous current density in terms of chemical reaction rates. For AESEC, the instantaneous dissolution rates are directly related to the instantaneous downstream concentrations by mass balance (Equation [2]). However, transport in the capillary system leads to a time offset between the instantaneous concentrations and electrochemical measurement (Δt), while mixing and diffusion within the flow cell leads to a broadening of the concentration transients.

The time offset is easily accounted for by simply shifting the spectroscopic data with respect to the electrochemical data by Δt, an easily measurable value. The broadening of the concentration transients, however, presents a more complex problem. When the dissolution rate is changing rapidly, a rigorous comparison of electrical current density and dissolution rate requires “smoothing” the electrochemical data (j_e) so that it has the same time resolution as the dissolution rate data, ⁠. For example, the use of Equation (4) for the determination of oxide during electrochemical transients requires that the electrical current density (j_e) be convoluted with the residence time distribution function (h(t)) so that they have identical time resolution. This is illustrated in Figures 3, 4, and 6 where the j_e and differ significantly during the first seconds of the transient and only become equal around 120 s. Failure to use the convoluted current density in Equation (4) would lead to a significant overestimation of the quantity of oxide formed during this time period. If the time evolution of current density and dissolution rate are slow on the time scale of the experiment, convolution is not necessary.

The principle of convolution is straightforward. For a given element, the true interfacial dissolution rate (ν_M°(t)) is related to the measured elemental concentration transient (ν_M(t)) via a convolution integral:^6,60   where h(t) is the residence time distribution (RTD)^61-62  of the electrochemical flow cell. τ is simply a variable of integration. The idea is entirely equivalent to the transfer function between the electrochemical and spectroscopic measurements proposed by Gabreilli, et al.⁶³ 

The ideal situation would be to solve Equation (11) directly for ν_M°(t) and deconvolution routines for the dissolution rate data have been proposed.^15,53  However, due to the magnification of error implicit in such calculations, they have not been used routinely. In practice, it is necessary to perform a numerical convolution of the experimental current data set j_e(t). As the electrochemical current measurement may be considered instantaneous on the time scale of these experiments, the convolution integral is as follows:

The RTD will be specific for each flow cell design and must be determined experimentally. This may be obtained by measuring the concentration transient (C(t)) following the application of a delta function of dissolution at the working electrode/electrolyte interface:

A typical potentiostatic pulse experiment for Cu/HCl is shown in Figure 8. The 1.5 s anodic pulse was considered sufficient to approximate a delta function. Numerical simulations have demonstrated a good agreement with the experimental data by only considering the electrolyte flow patterns in the electrochemical cell. Empirically, it has been demonstrated that h(t) closely follows a log-normal function for the flow cell design in Figure 1: where value of τ and β are empirically determined parameters, which may be construed as measures of the time resolution of the system.

Much effort has been made to identify the different oxide species that form during anodic dissolution of Zn in alkaline electrolytes.⁶⁴  To this end, AESEC was used to distinguish two forms of Zn oxide in the AESEC polarization curves (Figure 9) of nominally pure Zn (electrogalvanized steel) in a 0.1 M NaOH solution. Figure 9(a) shows the overlay of j_e, ⁠, and j_Zn at different sweep rates as indicated. The difference between j_e and j_Zn throughout the active domain increased markedly with increasing sweep rate, which at a glance might be interpreted as oxide formation. The convoluted current density (⁠⁠), however, overlays j_Zn throughout the active domain, demonstrating that the effect was mostly due to the residence time distribution of the flow cell. Figure 9(b) gives j_Δ, which may be interpreted as the formation of an insoluble Zn(II) oxide/hydroxide film. It has been previously proposed that two types of oxide formed during linear scan voltammetry experiments and this hypothesis is confirmed by the AESEC-LSV data. For 0.5 mV/s and above, an oxide formation peak (Type I oxide) was observed at −1,210 mV and this peak became increasingly resolved from background noise as the sweep rate increased consistent with a surface reaction. This was followed by a second peak at −1,100 mV (Type II oxide), perturbed by the nonstationary peak immediately following the maximum region. In this way, the AESEC technique distinguished two types of oxide based on the kinetics of their formation. The existence of the two forms of oxide was further confirmed by potentiostatic experiments.

A fundamental problem in the investigation of corrosion processes is that they occur spontaneously without passage of electrical current through an external circuit. Therefore, the use of dynamic electrochemical methods requires perturbing the system and extrapolating the electrical current to the zero-current potential (E_j = 0). It is possible, however, to measure the potential of a metal or an alloy during free corrosion, and on occasion, the value of the potential may be used to diagnose the state of the system by comparison with thermodynamic calculations⁶⁵  or with polarization curves. This is of particular interest for monitoring the progress of industrial surface treatment processes such as pickling, etching, or conversion coating where the reactions may occur very rapidly without any real steady-state occurring. Surface treatments are often performed by simple immersion in which the material reacts spontaneously with the electrolyte and the only electrochemical measurement available is the open-circuit potential, which of course gives no direct information on the interfacial reactions. As a classic example, Ghali and Potvin (1972)⁶⁶  may be cited on the mechanism of phosphating of steel, and a more recent example of Schoukens, et al., concerning Zr-based conversion coatings on aluminum alloys.⁶⁷ 

AESEC provides a window into these processes by directly measuring the dissolution rates of the different alloying elements and correlating them with potential changes. The pretreatment of Al alloys prior to conversion coating or anodization is a case in point. The pretreatment often involves two steps: (1) alkaline or acid etching to remove the mechanically altered layer, followed by (2) a nitric acid rinse to remove residual elements, especially Cu. Figure 10 gives an example of a typical open-circuit AESEC measurement during a surface treatment.⁶⁸  In this case, a 6000 series aluminum alloy was treated with a commercial acidic etching solution. The open-circuit potential vs. time and the average mass loss (upper curves, Figure 10) are the standard in-house measurements for monitoring the progress of the reaction. In this case, the standard mass loss was determined to be 40 μg·cm⁻²·s⁻¹, corresponding to an etching rate of approximately 0.85 μm/h and the E_oc profile shows considerable fine structure: an early anodic dip, followed by a slow rise through a maximum, and then a steady rise toward the end of the experiment. Alone, these results would be almost meaningless as they cannot be directly related to either the corrosion rate as predicted by a hypothetical Evans diagram, or the equilibrium model of the Pourbaix diagram. AESEC offers a new dimension to our understanding: by following the elemental dissolution rates, one can see that the potential variations correlate with the dissolution rate variations of specific elements. Al dissolution was initially slow while Mg dissolution passed through an early maximum. This maximum correlates with the anodic dip (“a” in Figure 10). Note that Mg is multiplied by a factor of five to be on the same scale as the Al dissolution rate: its dissolution would make a negligible contribution to the total mass loss in a typical gravimetric measurement of the etching rate.

These results also demonstrate the importance of time resolution. Following the dissolution of Mg, a peak of Mn dissolution was observed, here multiplied by 100 to be on the same scale as Al. This clearly correlates with the maximum (“b” in Figure 10) in E_oc. Beyond this, a relatively stable dissolution rate was obtained with a slow, steady increase in E_oc, perhaps due to the enrichment of metallic Cu on the surface. Obviously, a detailed interpretation of these results would require additional knowledge of the metallurgy of the system: for example, how the different elements are distributed between phases at the surface and how the elements are distributed in depth in the material. Nevertheless, the elemental dissolution rates reveal a fingerprint of the mechanistic processes occurring during the surface treatment that underlies the fine structure of the E_oc profile.

The robustness of AESEC is also illustrated by the results of Figure 10. The commercial etching solution used was an extremely complex, nearly opaque electrolyte containing a high total dissolved solid content including transition metal cations, surfactants, etc. Despite the complex matrix, excellent detection limits for the dissolving elements were readily obtained.

A simulation of a complete pretreatment sequence is shown in Figure 11, in this case involving NaOH/water rinse/HNO₃ under the conditions indicated in the figure.⁶⁹  The results demonstrate the intense Al dissolution in alkaline solution, followed by the dissolution of residual Cu and the passivation of the Al in the nitric acid step. The quantitative aspects of AESEC are well illustrated: Al dissolution was rapid in the alkaline solution corresponding to an etching rate of approximately 160 μg/h. The Cu dissolution rate, on the other hand, was three orders of magnitude lower. Under these conditions, Cu dissolution would have made a negligible contribution to the overall mass change and could not have been detected by conventional gravimetric methods.

The situation changes when the surface is exposed to nitric acid: the Cu dissolution rate was approximately 3× larger than the Al dissolution rate. It was also observed that Al and Cu dissolution peaks did not have the same shape. Al dissolution occurred in two peaks, the first which overlays with the time constant distribution of the flow cell here shown as a dashed red curve.

The selective dissolution of Al during the alkaline treatment is apparent in Figure 11. This leaves behind a surface layer enriched in metallic Cu, the growth and dissolution of which may be determined indirectly via a mass balance. This is shown in Figure 12 which gives the Al mass loss and the buildup of Cu gain as a function of time. These results also demonstrate that under the conditions of these experiments, Cu dissolution was not complete during the nitric acid rinse.

Note that the measurements of Figures 10, 11, and 12 would be difficult with CCD detection because the Cu and Al are on very different orders of magnitude, or with ICP-MS because of the high Al dissolution rate for Figure 11 and the complex electrolyte containing high dissolved solids and surfactants for Figure 10.

The corrosion of Al alloys may involve the undermining and release of constituent particles, which are typically intermetallic compounds.⁷⁰  As a non-faradaic process, this phenomenon cannot be detected by electrochemistry and makes a negligible contribution to the mass loss. The high time resolution of the AESEC method and the ability to correlate the elemental signals makes it possible to detect and identify the released particles.

Constituent particle release was observed during the alkaline etch of Figure 11 as indicated by the sharp spikes of Cu dissolution. The detection of individual particles is more apparent in Figure 13, which was obtained from a similar alkaline etch with enhanced time resolution of 10 points per second. Particles released into the electrolyte were carried to the plasma giving rise to very sharp peaks of a single point. As each point represents the average emission intensity over the data collection period, the intensity of the point will be directly proportional to the data collection frequency. Previous work demonstrated that the duration of the individual particle transients was <10 ms.⁷¹  Of particular interest is that the composition of the particles may be obtained from this data. In the figure, it is clear that Cu, Fe, and Mn peaks correlate with each other as do Cu and Mg. A complete statistical analysis of the particle distribution is also given for this experiment.⁷² 

The selective dissolution of constituent particles during the anodization of Al alloys leads to the formation of defects in the final oxide film. To evaluate the fate of the constituent particles, AESEC was used to monitor the dissolution of the alloying elements during the anodization of AA7050-T74.⁷³  Based on a mass balance it was possible to identify the dissolution rates of the different particles. It was found that MgZn₂ completely dissolved, whereas Al₂Cu, Al₂CuMg, and coarse Al₇Cu₂Fe intermetallic compounds underwent a more complex dissolution process leading to the accumulation of copper on the surface and the formation of holes in the anodized layer. The kinetics of dealloying of the latter pure phases to form nanoporous Cu films was investigated as a function of potential.⁵⁰ 

The polarization curve may be used to predict the electrochemical behavior and corrosion rates over a wide range of potentials⁷⁴  and the rate laws for the elementary anodic and cathodic reactions, the input for most numerical simulations of corrosion, and are usually derived from polarization curves.⁷⁵  For multicomponent and multiphase materials, however, it is necessary to know specifically which elements are dissolving and which elements remain behind, becoming enriched on the surface, if we wish to understand what is really happening. Prediction also requires knowledge of the “steady-state” polarization curve, which in general is not available and arguably does not exist, as the interfacial reactions that occur during corrosion induce changes in both the material and the environment and preclude the existence of a true steady-state. For example, dissolved species formed by the anodic and cathodic reactions alter the nature of the environment with time and may generate films of corrosion products. Selective dissolution of one element may lead to the formation of accumulated metallic films on the surface such that even the nature of the alloy changes with time.

To illustrate the importance of the elemental dissolution rates, first consider the polarization curve of a stainless steel in sulfuric acid. This system may be considered an archetypal example of a standardized electrochemical measurement and is commonly used in the stainless steel industry.⁷⁶  A typical example of a Type 304 austenitic stainless steel (UNS S30400) in 2 M H₂SO₄ is shown in Figure 14, redrawn after Ogle, et al.⁷⁷  Note that the j_e vs. E characteristic curve shows two E_j = 0 points, defining two cathodic and two anodic domains. The second cathodic domain is referred to as a “cathodic loop,”^78-81  the origin of which has been debated.

The elemental polarization curves, shown below the j_e–E characteristic, yield a precise interpretation of the origin of this loop. First, all of the alloy components, with the exception of Cu, dissolved congruently during the first anodic peak.⁽⁴⁾ The dissolution in the active peak was faradaic, as indicated by the close correspondence between j_e and j_Σ. However, the absence of Cu dissolution in this potential range suggests that metallic Cu built up on the surface due to the selective dissolution of the other elements. Further, the end of the cathodic loop was observed to coincide precisely with the dissolution of this excess Cu.⁷⁹ 

Taken together, these results strongly support the idea that the enhanced catalytic reaction is due to the buildup of excess Cu during the active peak as suggested by Hermas, et al.⁸¹  This example demonstrates the effect that alloying elements may have on the polarization curve such that the history of the material is critical. Cu buildup may have important consequences in some applications, for example, resulting in an anodic polarization of the material due to galvanic coupling between steel and Cu. In some cases, this may lead to an enhancement of the passive film.⁷⁹ 

The dissolution of a multicomponent material is often selective for certain elements leaving behind either the more noble metals as a metallic film as illustrated for Cu/stainless steel above, or as insoluble oxides. In both cases, the residual film may have a profound effect on the reactivity of the underlying substrate. The Cr-rich passive film on the surface of Fe-Cr and Ni-Cr alloys is a case in point. Only a few nanometers thick, its measurement normally requires ex situ surface analysis, for example, by XPS. As discussed in the Mass Balance section it is often possible to indirectly determine the quantity of residual elements remaining on the surface of the dissolving material through a mass balance via Equation (6).^82-84 

A recent example is the cyclic activation and passivation of Type 304 stainless steel in 2 M H₂SO₄ in Figure 15. Shown are the Fe and Cr dissolution rates, ν_Fe and ⁠, respectively, where the apostrophe indicates normalization to the bulk composition as follows:

In this way, the dissolution profile obtained during the cyclic experiments reveals at a glance much about the mechanisms of dissolution. The open-circuit (active) period is characterized by congruent dissolution of Cr and Fe as indicated by the equality of the normalized Cr and Fe dissolution rates, ⁠. In this period, the rates of dissolution were proportional to the bulk composition and no detectable film formation occurred.

The passivation step involved the formation of a Cr-rich passive film, inferred from the dissolution of excess Fe (⁠⁠):

The total quantity of the Cr(III) enrichment was estimated from a mass balance and is indicated by the pink shaded area between the two curves. The formation of the Cr enrichment caused the system to transition into the passive state, clearly indicated by the decreased dissolution rate and negative open-circuit potential, following the release of the potential (+412 mV_SCE).

The activation step (A in Figure 15) involves the cathodic dissolution of the passive film. The dissolution reaction in this case was complex but most likely due to the reduction of the Cr(III) enriched passive film to give a soluble Cr(II) species:

The removal of the passive film caused the system to transition into the active state indicated by the markedly enhanced dissolution rate and a more negative open-circuit potential (322 mV_SCE). The dissolution of the Cr enrichment occurred during this step, indicated in Figure 15 by the excess Cr going into solution (⁠⁠). The quantity of dissolved Cr in the passive film was determined by the difference between $νCr′$ and ν_Fe, indicated by the blue shaded area.

The kinetics of passive film formation and dissolution may be obtained from the mass balance equations (Equation [6]) which give the accumulated excess Cr (Θ_Cr) as a function of time. The simultaneous measurement of the corrosion rate and film quantity allows the examination of the effect of the film on the corrosion rate in situ. Typical results are shown in Figure 16 for a sequence of four active/passive cycles. These results demonstrate the inverse relationship between the soluble corrosion rate, j_diss (= j_Σ), measured as the sum of the elemental dissolution rates, and Θ_Cr. Passive film formation under these conditions involves the formation of an oxide film containing approximately 0.4 μg Cr/cm², corresponding to approximately 0.7 nm assuming a uniform Cr₂O₃ surface film of ordinary density. This may seem very small; however, it should be borne in mind that this is also a very short time passivation corresponding to only the first steps of passive film growth.

Information on the kinetics of passivation are also available from results like Figure 16: activation (passive film dissolution) occurs very rapidly with the dissolution peak corresponding approximately to the residence time distribution in the cell. Dissolution may be even faster than the experimental RTD due to the formation of hydrogen gas which facilitates mixing in the cell. Passivation (passive film formation) occurs more slowly and continues even after the applied potential is released.

It is of interest to compare the elemental polarization curve of stainless steel with that of a Ni-Cr-Mo alloy, Figure 17.⁸⁵  The polarization curve in the upper part of the figure shows a similar form with clearly defined active peaks, although this alloy does not show any cathodic loop. The elemental dissolution, however, is quite different: the active peak of stainless steel shows nearly congruent dissolution with a near 100% faradaic efficiency, while that of Ni-Cr-Mo reveals elemental dissolution prior to the active peak, but minimal dissolution during the active peak. This suggests that the majority of the oxidation, indicated by the total anodic current density, was due to the formation of insoluble species. The absence of Mo dissolution in this domain suggests that Mo builds up on the surface during the active to passive transition and may play a role in the passivation mechanism. A mass-charge balance during the active peak (a2 in Figure 17) yields approximately 2.5 mC/cm² of undetected oxidation (blue shaded area), which is a factor of 100 less than was obtained for pure Ni. This would correspond to the formation of a nanometric film on the order of 2 nm to 3 nm assuming a standard density for Cr₂O₃.

The cyclic activation–passivation of a Ni-Cr-Mo alloy is given for both Cr and Mo in Figure 18. A unique feature of this alloy as compared to stainless steel (Figure 16) is the spontaneous passivation which occurred during the open-circuit time period labeled SP in the figure. Following the cathodic removal of the passive film (A), Θ_Cr began to increase as soon as the potential was released. This spontaneous formation of the Cr enriched passive film continued until the passive potential (P) was applied. From the Θ_Cr versus time profile, the kinetics of repassivation may be directly obtained.

The Θ_Mo vs. time profile indicates that Mo enrichment only occurred during the spontaneous passivation step, consistent with the elemental polarization curves of Figure 17 in which no Mo dissolution was observed below 0 V. Partial Mo dissolution was observed during the potentiostatic step to the passive potential, and complete dissolution occurred during activation. This complex behavior of Mo may offer a partial explanation as to why it has been difficult to elucidate the mechanism by which Mo improves the corrosion resistance of Ni-Cr alloys: whether or not Mo is enriched in the film depends on the potential at which passivation occurs.

Mo enrichment during transpassive–passive cycles was investigated for a series of Ni-Cr-Mo alloys in a neutral saline solution (1.0 M NaCl, pH = 7.4, T = 75°C) using a similar cyclic potentiostatic method.⁸⁶  The idea was to simulate the pitting/repassivation behavior that might be observed during crevice corrosion. The transpassive–passive experiments were conducted by stepping the potential from the passive domain to the transpassive domain and then returning to the passive domain.

It was found that Mo enrichment occurred during transpassive dissolution and redissolved when the potential returned to the passive domain as indicated by the normalized dissolution profiles of Figure 19. The enrichment of Mo during the transpassive step is indicated by the low dissolution rate (normalized to Ni) during this period. The quantity of enriched Mo (Θ_Mo) may be determined from the difference between Ni and Mo dissolution rates, indicated by the yellow shaded area. When the potential was potentiostatically returned to the passive domain, the excess Mo dissolved. The quantity of Mo dissolved (Q_Mo) may be determined from the shaded blue area. The inset to Figure 19 gives a plot of Θ_Mo vs. Q_Mo for a series of experiments performed at different hold times as indicted in the transpassive domain.

From these results, it was proposed that Mo precipitation and redissolution are driven by local pH changes. Even though a flow cell was used, a diffusion zone of about 130 μm to 200 μm was present at the material/electrolyte interface (Instrumentation section). During transpassive polarization the acidification was indicated by a drop in pH from 7.4 at the flow cell entrance, to approximately 3 to 4 at the exit. Obviously a more intense acidification would be expected at the interface. These results suggest a mechanism of Mo enrichment and release that could play a significant role in the repassivation of a Ni-Cr-Mo alloy in neutral chloride electrolytes, for example, during crevice corrosion.

Electrochemical impedance spectroscopy (EIS) is an important technique in the arsenal of the corrosionist. For steady-state corroding systems, the EIS spectrum may in some cases be used to estimate the corrosion rate without significant electrochemical perturbation, and the EIS spectrum itself is often considered a fingerprint for specific mechanisms revealing different kinetic processes over a wide range of time constants. It is straightforward to perform EIS simultaneously during an AESEC experiment and Cwalina, et al.,⁸⁷  used high-frequency AC electrochemical measurements to follow the growth of oxide films for the Ni-Cr-Mo system during conventional DC experiments coupled with ICP-MS measurement of elemental dissolution rates.

It is also possible to monitor the oscillating elemental dissolution rates and to decompose the global EIS data into elemental components at least for low frequencies. In this way the EIS measurement may be obtained on an element by element basis and the AC response of oxide formation or the cathodic reaction obtained by consideration of j_Δ as in the previous sections. Of particular importance is that dynamic systems may be treated directly, by analyzing data in the time domain. The latter idea was demonstrated for the Mg/1 M NaCl system in which EIS-AESEC was used to pin down the origin of the low-frequency inductive loop commonly observed and the origin of which has been debated.⁸⁸  This work revealed that the inductive loop was most likely due to a catalytic enhancement of the dissolution rate at low frequency.

An example of the use of EIS-AESEC is to distinguish situations in which the oxidation of the metal leads directly to the formation of dissolved ions or passes through a slightly soluble or insoluble intermediate.⁵³  In the former case, it was possible to measure fundamental electrokinetic parameters in a single experiment. For example, Figure 20(a) shows the dissolution profile for Zn dissolution 0.1 M NH₄Cl solution. The profile is divided into four time periods: I – measurement of the background signal in the electrolyte alone; II – open-circuit dissolution; III – measurement of potentiostatic EIS data from high to low frequency; and IV – return to open-circuit dissolution. The excellent correlation between the AC components of the electrical current density and the zinc dissolution rate in the lower frequency range demonstrates that anodic dissolution occurred directly without the formation of an insoluble intermediate on the time scale of the experiment. Analysis of the dissolution rate and total current density transients as a function of potential revealed the anodic and cathodic Tafel slopes for this system. In contrast, for Zn/0.5 M NaCl (Figure 20[b]), electrochemical oxidation leads to the formation of an intermediate corrosion product film with subsequent dissolution. As a consequence, no variation of the elemental dissolution rate was observed despite the alternating electrochemical current. This type of behavior has been previously seen for silicate-based conversion coatings using cyclic large amplitude DC potentiostatic cycles and is an interesting approach to quantifying the stability of oxide films used as conversion coatings.⁵² 

Different elements of an alloy react differently with the environment and this may sometimes be observed in the elemental impedance behavior. A fascinating example is the Zn-Al-Mg system in a pH = 10.5 ammonium buffer, shown in Figure 21.⁸⁹  For this system, Zn and Mg dissolve in phase with the AC current while Al dissolution is shifted out of phase by 180°. This clearly indicates a different mechanism for Al. The origin of this shift is the subject of current research.

The stoichiometric relationship between current and dissolution obtained by AESEC may give considerable information concerning the mechanisms of interfacial reactions. However, in many situations this is not sufficient to completely describe the reaction and additional real-time techniques are of interest.

In situ gravimetry may yield additional insight into interfacial processes which involve simultaneous film formation and substrate dissolution, such as corrosion product formation or conversion coating chemistry. The electrochemical quartz crystal microbalance (EQCM) is a technique that monitors the mass of the working electrode as a function of time: the test material is deposited on a thin quartz crystal wafer, the resonance frequency of which, under optimum conditions, is inversely proportional to the mass of the electrode. The coupling of EQCM–AESEC therefore provides information on the mass changes and dissolution simultaneously. This coupling was achieved via a special flow cell replacing the working electrode of AESEC with a quartz electrode. It has been used to investigate the anodic dissolution of Cu and the formation of phosphate⁹⁰  and chromate⁹¹  conversion coatings on zinc.

Figure 22 gives EQCM–AESEC profiles for the spontaneous, open-circuit reaction of a Zn electrode with a 0.2 M phosphoric acid solution (a) alone and (b) containing 1.37 g/L KNO₃. The upper curves shows the AESEC dissolution profile, and the lower curve gives the calibrated QCM frequency transient expressed in “mass” superimposed with the integral of the ICP transient. Also shown is the “precipitation profile,” being the difference between the total mass change from the QCM (ΔM_QCM) and the dissolution mass change from the ICP (ΔM_ICP). This difference (ΔM_film) corresponds to the growth of the Zn–phosphate film. The results of Figure 22 demonstrate that film formation was enhanced by the addition of nitrate, justifying its frequent use as an accelerator in conversion coating formulations. For the nitrate-containing solution, the ΔM_film traced a sigmoidal curve with a saturation level of approximately 250 μg/cm², roughly 10× that observed in the phosphoric acid solution without nitrate. Film formation was dominant only during the first ≈100 s, while Zn dissolution occurred throughout the exposure. At longer times, ΔM_ICP closely followed ΔM_QCM, demonstrating that, under the conditions of these experiments, the frequency of the QCM is a reliable measure of mass loss.

Gas evolution often occurs simultaneously with anodic dissolution, for example during the negative difference effect (NDE) of Mg,^92-95  the cathodic dissolution of Al,^59,96-97  and anodization of Al alloys.⁹⁸  A method of independently measuring gas evolution and thereby the stoichiometry of gas formation permits a kinetic decoupling of gas evolution, anodic dissolution, film formation, and charge transfer. The first attempt to combine a volumetric method with AESEC was obtained by Lebouil, et al., who, obtained the full stoichiometry of Mg dissolution, during the NDE.^92-93  The technique she developed used an in-line high-speed video camera to measure H₂ bubble volume by image analysis after breaking the bubbles up in a microfluidic capillary system. More recently, Han and Ogle¹⁰  have combined AESEC with a downstream gravimetric measurement of hydrogen gas based on Archimedes’ principle as proposed by Curioni.⁹⁴ 

An example of the latter work is given in Figure 23 showing the cathodic corrosion of an AA2024 (UNS A92024) alloy in a neutral electrolyte was investigated at an imposed potential of −1.8 V_SCE.¹⁰  The expected overall stoichiometry was:

The upper curve gives the dissolution profile for and j_Al. The Al dissolution rate mirrors the slow onset of the cathodic current density with a ratio of ⁠, consistent with the previously cited work. This ratio was always >1 due to the formation of excess hydroxide which diffuses away from the interface. The stoichiometry of the hydrogen evolution reaction (HER) is displayed in the lower curve showing the integrals of the dissolved Al, Q_Al, the inversed net charge, –Q_e, and the evolved H₂, n_H2. The latter is compared with the hypothetical quantity of evolved H₂, calculated from the mass balance of Reaction (4), n_H2(mb) = (3Q_Al – Q_e)/2. The ratio of the measured to the hypothetical quantity of evolved hydrogen was found to be n_H2/n_H2(mb) = 0.94 at the end of the potential pulse, showing a fairly good agreement with the stoichiometry of Equation (18).

The coupling of other in-line methods of electrolyte analysis with AESEC is quite straightforward although it has not been extensively exploited. Only a small portion (≈5%) of the electrolyte is actually aspirated into the plasma, the remainder being pumped away to a waste container. Downstream pH measurements have been used frequently. For example the variation of pH during linear scan polarization of Al alloys was presented in Serdechnova, et al.⁵¹  Other techniques are possible such as downstream electrochemical or UV-visible detection. For example, electrochemical detection was used to quantify the dissolution rates of Cu(I) and Cu(II) during the anodic dissolution of Cu-Zn alloys in a channel flow double electrode by Hoshi, et al.,^99-100  and UV detection was used to monitor xanthate adsorption and desorption during a simulated electrochemical flotation of chalcopyrite with a flow spectroelectrochemical system by Walker, et al.¹⁰¹  There is much to be developed along these lines.

• The direct, time resolved measurement of elemental dissolution rates yields insight into the reactivity of complex, multi-element materials, unavailable by conventional electrochemical, gravimetric, or ex situ spectroscopic measurements. This direct approach to interfacial kinetics removes some of the limitations imposed by electrochemical methods, such as the necessity of a steady-state reaction, the uncertainty of the specific faradaic reactions occurring at the interface, or the guess work associated with interpreting open-circuit potential measurements. This methodology gives direct insight into the questions of how specific elements interact with each other in an alloy, or how specific additives in a surface treatment solution will affect different alloying elements or different phases. The AESEC technique will no doubt prove useful for the development of more stable alloy materials and improved surface treatment solutions. Although this work has been focused on metallic materials, AESEC may be used to study the reactivity of other materials such as polymer and paint films, controlled inhibitor release, or the selective dissolution of minerals during processing.

• In this work, the basic quantitative relationships were developed and illustrated by numerous examples including the spontaneous reactivity of Al alloys with commercial surface treatment solutions and the dissolution and passivation of Fe-Cr and Ni-Cr-Mo alloys. In the case of Al surface treatment, the extremely fast etching of Al may be simultaneously measured by the extremely slow dissolution process of Cu and other alloying elements. The compatibility with complex, concentrated electrolytes was also demonstrated. AESEC detects non-faradaic processes as well such as intermetallic particle release. This type of experiment may prove useful for the development of new surface treatment technologies.

• By mass balance, the surface accumulation of small quantities of various elements may be readily detected at the sub-nanometer level. These may be metallic elements in the case of a dealloying phenomenon or oxide films in the case of passivation. The relationship between the corrosion rate and the accumulation of surface Cr yields new information on the relationship between surface composition/structure and reactivity. Such experiments may prove useful for the development of new alloy materials.

• The objective of this work has been to demonstrate that AESEC has great potential to validate many hypotheses regarding corrosion and electrochemistry. A few selected applications of AESEC would be to understand the role of different alloying elements on passive film formation, optimizing the formulation of additives during surface treatment, quantifying the impact of corrosion inhibitors on different elements and phases of an alloy, quantifying the kinetics of cathodic dissolution such as is observed for Mg, Al, or passive film dissolution during cathodic cycles of bipolar plate operation. A number of methodological ideas remain undeveloped or underdeveloped. As previously mentioned, other techniques may be coupled with the flow system such as electrochemical detection to determine oxidation states of dissolved ions or UV-visible spectroscopy to measure organic species. Numerical modeling of the flow cell would allow one to better understand the chemical and hydrodynamic conditions of the surface and perhaps optimize the geometry of the flow cell. The methodology could be extended to include gas analysis for the investigation of catalytic reactions and real-time video imaging of the surface. These and other possibilities are under consideration.

I would like to express my gratitude to my former coworkers Paul Lodi and Sophie Weber at the Institut de Recherche de la Siderurgie, who assisted in the early development of this technique. I am immensely grateful to all my former students and collaborators whose works are cited herein, and to my colleague Professor Polina Volovitch. I thank Patrick Chapon and Alice Stankova of Horiba France for technical assistance and advice concerning the instrumentation section of ICP-AES and ICP-MS.

AESEC

Atomic emission spectroelectrochemistry

ICP-AES

Inductively coupled plasma atomic emission spectroscopy

ICP-MS

Inductively coupled plasma mass spectrometry

OTTLE

Optically transparent thin layer cell

GDOES

Glow discharge optical emission spectrometry

SDC

Scanning droplet cell

RTD

Residence time distribution, also h(t)

NDE

Negative difference effect

QCM

Quartz crystal microbalance

C_M

Concentration of element M downstream from flow cell

I_λ

Emission intensity at wavelength (λ)

I_λ°

Background intensity at wavelength (λ)

κ_λ,M

Sensitivity factor relating emission intensity at wavelength (λ) to concentration of M

ν_M

Measured dissolution rate of element M normalized to the surface area

ν°

True dissolution rate of element M

Measured dissolution rate of element M, normalized for the bulk concentration of the material

E_ap

Applied potential during potentiostatic experiments

E_oc

Open-circuit or spontaneous corrosion potential

E_j = 0

Zero current potential from a polarization curve

n_M

Assumed number of electrons transferred during the dissolution of M

j_M

Dissolution rate of element M expressed as an equivalent current density

j_Σ

Sum of elemental current densities

j_diss

Soluble corrosion rate expressed as current density, normally = j_Σ if all elements are measured

j_e

Electrical current measured by the electrometer of the potentiostat

After convolution with residence time distribution

j_Δ

Faradaic current for phenomena not detected by ICP-AES; j_Δ = j*_e – j_Σ

n_H2

moles of hydrogen measured by in-line volumetry

n_H2(mb)

moles of hydrogen determined by stoichiometric mass balance of AESEC data

Q_M

The total quantity of element M dissolved in a defined time period

Q_e

Total charge, integral of j_e/F

Θ_M

Surface enrichment of element M determined by mass balance

Θ_e

Quantity of accumulated charge, integral of j_Δ/F, used in the determination of oxide formation

Θ_M(z)

Surface enrichment of M, oxidation state z, usually in the form of oxide

ΔM_QCM

Mass change of the working electrode as determined by QCM

ΔM_ICP

Mass change of the working electrode as determined by ICP

ΔM_film

Mass change due to film formation on working electrode, = ΔM_QCM – ΔM_ICP

h(t)

Functional form of the residence time distribution (RTD)

τ, β

Empirical factors of log-normal residence time distribution used in convolution

Δt

Time shift between electrochemical measurements and ICP-OES measurements due to electrolyte transport between the flow cell and plasma

f

Flow rate

A

Geometrical area of specimen exposed to the electrolyte in the flow cell