On the stability of perchlorate ions against reductive attacks in electrochemical systems and in the environment

The problems related to the electrochemical/electrocatalytic stability of perchlorate ions are reviewed in the light of recent experimental results. The electrocatalytic, catalytic, and electrochemical reduction processes are presented and the links between them are outlined. Some possible mechanisms of the complicated reduction processes are discussed. Various methods for the detection of reduction process are presented, e.g. voltammetry, impedance spectroscopy, and radiotracer methods. Environmental aspects and some methods for perchlorate removal and wastewater treatment are briefly summarized.


Introduction
Based on the thermodynamic data presented in Table 1, 4 ClO − should be instable against reductive attacks in acidic medium in a wide potential range. In contrast, in the electrochemical literature 4 ClO − are often considered to be very stable anions. Therefore, perchlorate containing solutions are widely used as supporting electrolytes in electrochemical studies with various electrodes. Among these investigations reports concerning dissolution, deposition, and corrosion of metals can also be found.
In such studies, it is tacitly assumed that the reduction of the 4 ClO − is not taking place under the actual experimental conditions. However, evidence for the occurrence of the reduction process has been reported for Rh [1][2][3][4][5], Pt [6][7][8][9], WC [10], Al [11], Ti [12], Ir [13], Ru [14], Re [15], Tc [16], and Sn [17] electrodes. A survey of the literature can be found in recent reviews [18][19][20][21]. These observations prompt us to keep in mind the possibility of the occurrence of the reduction process at any electrode used in an aqueous acid medium containing perchlorate ions. Besides the significance of perchlorate reduction for the fundamental electrochemistry today, a very practical reason, the so-called perchlorate contamination challenge, came into foreground directing the attention towards the reductive elimination of perchlorate ions.
Perchlorates are used as an oxidizer component and primary ingredient in solid propellants for rockets, missiles, and fireworks. Therefore, dissolved ammonium, potassium, magnesium, or sodium salts are present as contaminants in groundwater and surface waters originating from improper disposal of the solid propellants and from the wastewaters of the manufacturing plants.
As the sorption or natural chemical reduction of perchlorate in the environment is not significant, perchlorates are exceedingly mobile in aqueous systems and can persist for many decades under typical ground and surface water conditions. According to different reports [22][23][24][25][26][27], a large number of water sources in the United States have been contaminated with perchlorate. However, the perchlorate problem is not located to the US only. This type of contamination constitutes a major problem in China [28][29][30] and many other countries in the world [31][32][33]. For example, in Israel, an ammonium-perchlorate manufacturing plant that had been disposing untreated wastewater in four unlined ponds for 25 years caused extensive perchlorate contamination in the underlying aquifer. The perchlorate migration in the deep vadose zone has been found very low under natural recharge conditions in the semiarid region [34].
One of the main health hazards is connected with the very fact that perchlorate interferes with iodide uptake in the thyroid gland [35,36]. In large doses, it has been linked to anemia and fetal brain damage [30,35].
Several technologies have been studied for the treatment of perchlorate contaminated water, such as chemical precipitation and reduction, bioreduction, anion exchanges, electroreduction, membrane filtration, electrodialysis, and photocatalytic reduction. Even though some of these have been adopted for drinking water treatment, the optimization of cost, selectivity, and effective treatment is still a serious problem [37][38][39]. The electrochemical treatment method is one of the methods [20] considered applicable for solving the perchlorate contamination [24][25][26]. The elaboration of a treatment technology with dissolved complexes is significantly hindered by its high cost and significant difficulties with the regeneration or treatment of their end products. Some recent results in waste water treatment are presented below.
The electrochemical and electrophotocatalytic reduction of the perchlorate ion to the chloride ion were assessed at laboratory scale [40]. The study was carried out in two-chambered batch reactor systems in which the cathodic and anodic compartments were separated by an ion exchange membrane. Electrodes consisted of titanium coated with a thin film of small TiO 2 particles. Test water systems were buffered to an acid pH and background electrolyte was added. For photolytic systems, concentrations ranged from 5 × 10 -2 mol dm -3 to 5 × 10 -7 mol dm -3 . Reduction initially proceeded rapidly but slowed with time. The percentage of perchlorate electrochemically reduced after 2 h was found to range from less than 1% at the highest concentration to 30-35 % at lower concentrations. The extent of photocatalytic reduction at high perchlorate concentrations was approximately five times greater than the extent of electrochemical reduction at the same concentrations. An additional fourfold improvement in reduction percentage was observed when the electrode was doped with vanadium. A mathematical model suggested that the limiting factor in perchlorate reduction was the competition among anions for active sites on the electrode surface.
An electrically controlled anion-exchange process has been proposed to remove perchlorate ions from the industrial wastewaters [41]. A nanostructured conducting copolymer poly(anilineco-o-aminophenol) (PANOA) was electrochemically synthesized on a glassy carbon (GC) electrode and its redox and anion exchange properties for perchlorate removal were investigated by cyclic voltammetry, FTIR, and X-ray photoelectron spectroscopy (XPS). FTIR spectra demonstrated that 4 ClO − were immobilized in the copolymer film after its oxidation in a solution containing NaCl and NaClO 4 . The result of XPS spectra showed that the copolymer films oxidized in a solution containing 0.10 mol dm -3 NaCl and 0.010 mol dm -3 NaClO 4 at pH 5.7 and 9.0 had 4 ClO − / Cl − molar ratio of 0.53 and 0.31, respectively, which were significantly higher than the 0.10 ratio in the tested aqueous solution.
New, potentially green, and efficient synthetic routes for the remediation and/or re-use of perchlorate-based energetic materials have been suggested in two other studies [42,43]. Four simple organic imidazolium-and phosphonium-based perchlorate salts/ionic liquids have been synthesized by simple, inexpensive, and nonhazardous methods, using ammonium perchlorate as the perchlorate source. By appropriate choice of the cation, perchlorate can be incorporated into an ionic liquid, which serves as its own electrolyte for the electrochemical reduction of the perchlorate anion. The electrochemical degradation of the hazardous perchlorate ion and its conversion to harmless chloride during electrolysis was studied using IR and Cl-35 NMR spectroscopies.
Bioelectrochemical systems (BESs) including microbial fuel cells (MFCs) and microbial electrolysis cells (MECs) integrate three important wastewater treatment options, namely, biological treatment, electrolytic dissolution, and electrochemical oxidation/reduction and are regarded as new sustainable and effective strategies for wastewater processing. An up-to-date review is provided on recent research and development in BESs-based recalcitrant wastes treatment [44]. Recently, the reduction of perchlorate in the cathode compartment of a BES has been described [45]. The authors were able to isolate one of the organisms (Dechlorospirillum strain VDY) responsible for the perchlorate reduction. At a potential of -500 mV vs. Ag/AgCl reference electrode, the organism was capable of reducing perchlorate without the addition of a redox mediator. The results of these studies demonstrate that biological perchlorate remediation can be facilitated through the use of an electrode as the primary electron donor, and that continuous treatment in such a system approaches current industry standards. An MFC with a denitrifying biocathode for perchlorate reduction has been investigated to identify putative biocathode-utilizing perchlorate-reducing bacteria (PCRB), which can utilize a cathode as an electron donor, and this process can be harnessed to treat perchlorate while producing usable electrical power [46]. Perchlorate reduction has been observed with certain transition metal complexes and metal chelates, but the reactions are generally sluggish. A swift-oxo transfer reaction of perchlorate by rhenium oxazalin and thiazoline complexes has been reported quite recently [47,48]. The characteristic features of the new family of these molecular oxotransferases are their rapid kinetics and their ability in catalysing the very difficult reduction of perchlorate by atom transfer.
New (Re-Pd) bimetallic heterogeneous catalysts [49,50] for the reduction of perchlorate in water with dihydrogen have been prepared in two ways: (1) by impregnating 5 wt % Pd on activated carbon powder with one of the complexes trans-[ReO 2 (py-X)(4)](+) (py-X = 4-substituted pyridine; X = H, Me, OMe, NMe 2 ) or (2) by adsorbing perrhenate onto the Pd/C powder in the presence of the pyridine ligand under a hydrogen atmosphere. Both sets of catalysts are highly active at pH 2.7-3.0 (HCl), with observed rates increasing with varying X in the order H < Me < OMe < NMe 2 , which supports a rate-determining oxygen atom transfer reaction involving the Re(V) centers [51].
Granular activated carbon (GAC) coated with iron compounds (ICs) [52] or cetyltrimethyl ammonium bromide (CTAB) [53] were synthesized to remove perchlorate from water via adsorption. Laboratory-scale batch experiments were performed to study the factors affecting the perchlorate adsorption. In case of ICs/GAC, 97 % of the perchlorate ions were removed within 10 h at 90 °C. The experimental results also showed that FeOHSO 4 and Fe 2 O 3 nanoparticles have the function of perchlorate adsorption and play important roles in 4 ClO − removal. In case of CTAB/GAC, the optimal adsorption occurred at pH 2-3, and the mechanisms were associated with surface complexation, electrostatic interaction, and ion exchange.
Studies in heterogeneous reaction of dissolved perchlorate ions with metals open a new possible pathway for new approaches in the technology. In recent years, the possible application of iron has been investigated [54], including the rate and extent of perchlorate reduction on several types of iron metal in batch and column reactors. Mass balances performed on the batch experiments indicate that perchlorate is initially sorbed to the iron surface, which is followed by a reduction to chloride. Perchlorate removal was proportional to the iron dosage in the batch reactors, with up to 66 % removal in the period of 336 h in the highest dosage system (1.25 g ml -1 ). The most significant perchlorate removal occurred in solutions with slightly acidic or near-neutral initial pH values. Elevated soluble chloride concentrations significantly inhibited perchlorate reduction, and lower removal rates were observed for iron samples with higher amounts of background chloride contamination. The reactive iron phase is neither pure zero-valent iron nor the mixed oxide alone. A mixed valence iron hydr(oxide) coating or a sorbed Fe 2+ surface complex represent the most likely sites for the reaction. The use of zero-valent iron for treating wastewaters containing RDX (cyclotrimethylene trinitramine) and perchlorate from an army ammunition plant (AAP) in the USA at elevated temperatures and moderately elevated temperature with chemical addition was evaluated in batch and column experiments [55].
The interaction of a metal with dissolved perchlorate ions should be considered as a corrosion process, i.e. the overall transformation is composed of at least two or three reactions: This last step should be a very complex one composed of several elementary steps. The equations given above involve the coupling of the anodic dissolution of the metal with two cathodic processes: discharge of protons and reduction of 4 ClO − . In principle, the reduction of 4 ClO − could occur through a "catalytic mechanism" via the interaction with hydrogen atoms adsorbed on the metal surface.
It follows from these considerations that the problem of perchlorate reduction constitutes an intersection of several branches of sciences: electrochemistry (anodic dissolution of metals, discharge of protons); corrosion science (coupling of cathodic and anodic processes); catalysis (electrocatalysis); and heterogeneous chemical processes. In this paper, we present the historical background of the relevant topics in the light of recent developments and some new results obtained from the studies of direct interaction of perchlorate ions with various metals.

Heterogeneous catalytic/electrocatalytic aspects of perchlorate reduction
In 1971, Butula and Butula [56] found that HClO 4 could be reduced if Pd/BaSO 4 and Rh/C catalyst was used in acetic acid at 60 °C. In the same decade, the hydrogenation of perchlorate ions was studied at different temperatures and different perchlorate and hydrogen ion concentrations [20].
It was demonstrated that 4 ClO − dissolved in acidic aqueous solutions could be reduced to Cl − by molecular hydrogen in the presence of powdered platinum black catalyst [9]. Considering the link between the liquid (aqueous) phase heterogeneous catalytic hydrogenation and electrocatalytic reduction, it was easy to draw the conclusion that electrocatalytic/electrochemical reduction of 4 ClO − could be performed at electrodes prepared from materials identical to the material of catalyst powders used in the catalytic hydrogenation [57].
Electrodes with "catalytic" properties, such as Pt, Rh, Pd, Ir, and Ru, have been in the focus of interest. There was, and perhaps still is, a general belief in the literature that perfect voltamograms reflecting H adsorption on these electrodes can be obtained in HClO 4 solution. Although in addition to Pt and Rh electrodes 4 ClO − reduction on Ir [13] and Ru [14] conveyed the warning to be cautious with noble metal electrodes when using HClO 4 supporting electrolyte, most electrochemists in their various studies [58][59][60] carried out in the presence of HClO 4 supporting electrolyte ignored the possible complications that should be ascribed to the occurrence of a reduction process involving 4 ClO − . In 1990, it became evident that the contradictory results obtained during previous decades could be explained if it was assumed that the reduction of the 4 ClO − was characteristic not only for the rhodized surfaces, but also for smooth polycrystalline surfaces and well-defined crystal faces [3][4][5]. By this approach, almost all difficulties encountered in the interpretation of various phenomena could be eliminated during the last few years.
The distortion of voltammetric curves in the presence of 4 ClO − is the firm evidence of the occurrence of the reduction process. Figure 1 shows the voltammetric curves obtained in 1 mol dm -3 H 2 SO 4 , HF and HClO 4 supporting electrolytes using rhodized electrodes [61]. In the case of H 2 SO 4 and HF, despite some differences between them, the voltammetric curves behave "normally" as expected. In contrast, the voltammetric curve obtained in the presence of perchloric acid is distorted. The negative peak in the course of the positive sweep clearly indicates the occurrence of a cathodic process, namely, the reduction of perchlorate ions. Similar voltammetric curves were obtained with single-and polycrystalline smooth Rh [3,4] and platinized Pt [8] surfaces. Electrochemical behavior of Ir(100) in perchloric acid solutions has been characterized elsewhere [66]. The asymmetry of the cyclic voltammograms recorded in the double-layer region could be attributed to a slow reduction process, e.g. the reduction of perchlorate ions.
The characteristic features common to these systems are as follows: (a) Nowadays, there is a general agreement that the rate of reduction of 4 ClO − at noble metal and catalytic electrodes should be very low at both ends of the potential scale; consequently the I vs. E curve should go through a minimum. (b) The reduction process occurs almost exclusively during the positive sweep of the cyclic voltammogram. (c) Generally, no reaction can be observed on the negative sweeps, but the shape, height, and position of the corresponding hydrogen peak differ significantly from those expected. (d) The distortion of the anodic peak gradually disappears during subsequent cycles, but the anodic and cathodic peaks remain asymmetric. (e) The only reduction product found in the solution was Cl − . (f) The very rapid decrease in the electrocatalytic activity with respect to 4 ClO − -reduction can be explained by self-inhibition, namely, by the inhibiting action of Cl − formed during reduction and adsorbed on the electrode surface. The higher the Cl − concentration, the more pronounced is its inhibitive effect. (g) The desorption of Cl − from the metal surface is a very slow process. For instance, no measurable desorption of Cl − were detected until several minutes after switching the electrolyte solution from 1 mol dm -3 HF + 10 -4 mol dm -3 HCl to 1 mol dm -3 HF [62,67]. According to the results presented in the literature, the rate of electrocatalytic reduction of perchlorates on Rh is relatively high. The reduction process leading to the formation of Cl − should be a very complicated reaction as eight electrons are involved in the overall process. It is difficult to believe that it can be performed without the formation of any stable intermediates or products of side-reactions. Analogous to the reduction of perchlorate in homogeneous solution, the reduction of perchlorate at an electrode (or any surface) must involve oxygen atom transfer. Therefore, the controversy regarding the mechanism of the reduction of 4 ClO − on rhodium is not surprising. According to the "classical" view, the rate-determining step in the electrocatalytic process can be described as a reaction of adsorbed 4 ClO − species with adsorbed H atoms [1,13,19,61,62,67]. In contrast, the reduction process can be considered as a slow decomposition of 4 ClO − species on the surface followed by fast reduction steps [63]. The latter approach is based on the view that the reduction rate attains a measurable level at potentials where the surface concentration of H is very low; consequently, it is unlikely that it can play any role. Thus, it should be assumed that the rate-determining step is the decomposition of adsorbed 4 ClO − with the participation of a free adsorption site in its neighborhood.
In accordance with the above considerations, three fundamental types of possible mechanisms are distinguished for the electrocatalytic reduction of 4 ClO − [13,20,21,[61][62][63]65,67], namely the mechanism (A) involving protons and/or adsorbed hydrogen, the serial mechanism (B) and the mechanism (C) involving free-metal sites.
where "(ads)" denotes that the species is adsorbed, while "(sol)" refers to components in the solution phase. According to this scheme, the whole process starting from perchlorate ion occurs on the surface and the role of desorption of intermediates can be neglected, i.e. their desorption rate is very low compared with the rate of the chemical transformation.

Mechanism (C):
According to this mechanism, it should be assumed that the rate-determining step is the decomposition of adsorbed perchlorate ion if a free "active" site ( ) ⊗ is in its neighborhood. This assumption is the same as that made in [64] in connection with the reduction of N 2 O.
The decomposition reaction can be written as: On the other hand, it can be assumed that the surface concentration of active sites is determined by the following reactions: In this case, the potential dependence of the surface concentration of free active sites plays an important role.
In the reaction schemes proposed in the literature, the adsorption of perchlorate ion is an elementary step in the overall reduction. The decomposition of the adsorbed perchlorate ion is assumed to be slow; however, there are only speculations about the rates of the subsequent steps. For instance, in the mechanism (B) all decomposition steps are assumed to be slow, but no rigorous investigation of this problem has yet been performed.
The electrochemical reduction of perchlorate ions on rhodium and ruthenium was investigated in a series of experiments (voltammetry, chronoamperometry, impedance spectroscopy) performed at different electrode potentials (E) and different temperatures (T). Figure 2 shows cyclic voltammograms obtained for a Rh rotating disc electrode in contact with a 3 mol dm 3 HClO 4 solution [67]. There is a distinct difference between the curves recorded at stagnant electrodes and the curves obtained with the RDE. In case of stationary electrodes, the distortion of the anodic peak gradually disappears during subsequent potential cycles, i.e. the negative current during the positive sweep is continuously decreasing over the consecutive scans. In contrast, in case of rotating disc electrodes, the shape of the voltammogram does not change after the second or third cycle, and a negative current can be always observed during the positive sweep. It means that the desorption rate of -Cl generated during the reduction process is significantly influenced by the hydrodynamic conditions, probably through desorption/diffusion coupling. This conception is supported by the results of chronoamperometric measurements carried out at different temperatures. The curves obtained at T = 45 °C are presented in Figure 3 [67].   [67].
During these experiments the current was recorded as a function of time at constant electrode potentials and at different electrode rotation rates. Well-defined stationary currents were observed in all cases. The values of the stationary currents depend on the rotating rate of the RDE, the electrode potential, and the temperature.
It should be pointed out that the above measurements were carried out at potentials where the surface concentration of adsorbed H is very low, if existent. Consequently, it is unlikely that it can play any role. Thus, it should be assumed that the rate-determining step is the decomposition of adsorbed 4 ClO − with the participation of a free adsorption site. Figure 4 shows impedance spectra of Ru (deposited on gold) in 1 mol dm 3 HClO 4 solution. The impedance of the electrode was measured at different rotating rates of the RDE at the electrode potential of E=-0.15 V vs. SSCE (T = 25 °C). The impedance measurements were started as soon as stationary conditions were reached (e.g. about 20 min after changing the rotation rate of the RDE). The first observation is that the plots are very similar to what is expected qualitatively for the case of a single charge transfer step coupled with diffusion [68].
For all the spectra obtained in this study, only one time constant can be identified in the high frequency range, which suggests that the measured impedance response is dominated by a single (rate determining) charge transfer step. In the case of the reduction of perchlorate ions, the assumption that the -- 4 3 ClO ClO → transformation is the rate determining step seems to be a relevant and acceptable interpretation of the phenomena observed. Experiments with -3 ClO strongly support this hypothesis. Figure 5 shows cyclic voltammograms obtained for Rh in contact with a solution containing 0.49 mol dm -3 NaClO 4 + 0.01 mol dm -3 HClO 4 , and for the same electrode in contact with a solution containing 0.49 mol dm -3 NaClO 3 + 0.01 mol dm -3 HClO 4 . It can be clearly seen from the figure that in this potential range, the negative current is at least 2 orders of magnitude higher in the presence of -3 ClO than the current measured in the other solution. For comparison, the voltammetric curve obtained in 1 mol dm -3 HClO 4 solution is also given in the figure. Electrocatalytic activity in perchlorate reduction was found in the case of some non-noble metals, such as Tc [16], Re [15], Sn [17] and Ti [12,69]. Considering the very fact that Tc is an artificial element, emitting β-radiation, the coupled electrochemical and radiochemical study of electrodeposition of Tc species from HClO 4 supporting electrolyte gave an interesting insight in the reduction of 4 ClO − . It was found that under potentiostatic conditions during the deposition of Tc species, the increase in the radiation intensity, i.e. the increase of the amount of deposited material, is accompanied by a continuous increase of a cathodic current as shown in Figures 6 and 7 (the amount of the electrodeposited Tc species was determined directly by the measurement of the intensity of the radiation coming from the electrodeposited layer).
So far, only the occurrence of 4 ClO − reduction has been suggested as acceptable explanation [16]. The voltammetric study of the system furnished unambiguous evidence proving the validity of this assumption ( Figure 8).
It is well known [70] that the electrochemical behavior and properties of technetium and rhenium and those of their various ions are very similar.
Considering the similarity of Tc and Re, it was no wonder that the behavior found in the case of Tc can also be observed for Re and vice versa.
In contrast to the behavior of platinized platinum electrodes, the catalytic activity of "rhenized" electrodes can be observed under potentiostatic or galvanostatic conditions for relatively long periods of time, although no real steady state can be attained [15], as is shown in Figure 9 in the case of a potentiostatic experiment.  Adapted from [20].  (1) and in 3 mol dm -3 HClO 4 (2). Sweep rate, 2.5 mV s -1 (geometric surface area, 13 cm 2 ). Adapted from [20]. . Adapted from [15].
The continuous decrease in the reduction current can be explained by two important effects: i) the self-inhibition of the process by the adsorption of -Cl formed in the reduction (the current efficiency with respect to the formation of -Cl is above 90 %). The effect of -Cl was demonstrated in separate experiments by adding HCl in low concentration to the solution phase in the course of the reduction [15], and ii) deactivation of the electrode surface owing to the chemical transformation of the surface layer participating in the reaction.
The electrochemical behavior of tin in deaerated sodium perchlorate was studied using potentiodynamic and potentiostatic techniques [17]. The behavior of tin in sodium perchlorate was unexpectedly complicated by the reduction of the perchlorate anion. It was shown that the reduction process takes place within a potential region comprising the negative side of the double layer region and the positive side of the hydrogen region (-0.7 ≤ E ≤ -1.3 V). It was stated that the presence of oxide on the electrode surface favors the reduction reaction, which may occur in two steps: the formation of basic tin(II) chloride followed by its reduction, producing chloride.
Similar mechanistic conclusions were drawn in the case of titanium electrode. It was reported that perchlorate ion is electrochemically reduced to chloride ion at an active titanium electrode in aqueous 1.0 M HClO 4 [12]. It is assumed that the reduction occurs by direct reaction at the surface rather than a pathway involving catalysis by soluble titanium corrosion products. The reaction occurs by oxygen atom transfer to the titanium surface, and the implications of this mechanism for the surface composition of active titanium electrodes are discussed. Chlorate ion is also reduced at titanium, and the rate coefficient for chlorate reduction is at least 10 5 times greater than that for the perchlorate reduction.
In experiments with Ti in contact with perchlorate and nitrate solutions, Ti(III) or Ti(II) were generated by applying anodic current [69]. These multivalent Ti species are strong reducing agents and reduce perchlorate and nitrate. Kinetic experiments were carried out in synthetic perchlorate, nitrate solutions, and tap water media at different perchlorate and nitrate concentrations.

Reaction mechanisms and electrode impedance
In another study [67], the experimental results obtained for the electrochemical reduction of perchlorate ions on rhodium have been explained by the following reaction mechanism. After a potential step from a potential at which the adsorption of 4 ClO − does not occur, the reduction process starts with the adsorption of perchlorate ions at free active sites on the metal surface occupied, e.g. by OH − . After the production of a certain amount of -Cl , practically all active sites are occupied by -Cl or 4 ClO − , and a well defined stationary state with constant current is established. For the stationary state, the following reaction scheme can be suggested: ClO ads +Cl 0 (3.S1) where n is the number of electrons transferred from the metal phase to A − in forming one B − species, "(ads)" denotes here that the species is adsorbed on an active site, while "(0)" refers to components in the solution phase in the immediate vicinity of the electrode surface [71]. The rate of the first reaction (ν 1 ) depends on the electrode potential (E), the concentration of A − and B − near to the electrode surface (c A (0) and c B (0)), the surface concentration of adsorbed A − and B − ions (Γ A and Γ B ), respectively, while the rate of the second reaction (ν 2 ) depends on the electrode potential E, Γ A and the concentration of H + (c H (0)) adjacent to the electrode surface: According to the above considerations where Γ T is the total surface concentration of the active sites. It can be assumed, that Γ T is constant at a given electrode potential. Under steady-state conditions 1 2 v v = . The transport equations for the (linear) diffusion of the species B − can be written in the form: where x is the distance from the electrode surface.
If the system is perturbed with a sinusoidal signal of low amplitude, the variables can be written in terms of a stationary and a fluctuating component: where ( ) ξ x represents the steady-state quantity, is the amplitude of the fluctuation, ω is the angular frequency of the perturbing signal, t is the time, and "i" is the imaginary unit. On making an appropriate Taylor series expansion of the reaction rates -with the plausible assumptions we obtain : Each of the "rate constants" (k 0 -k 4 ) appearing in equations (3.9) and (3.10) represents a partial derivative. The "faradaic" current density related to the charge-transfer reaction is given by i F = nFν 2 . Thus the linearized form of i F becomes The linearized form of eq.(7) is : The above set of linearized equations (3.9) to (3.12) and the differential equation, (3.13) can be solved analytically [68,[72][73][74] by using the following boundary conditions: where δ represents the diffusion layer thickness, and J B is the flux of species B − . The resulting expression for the "faradaic" admittance is quite complicated: The first term of the right hand side of equation ( Thus, the faradaic impedance of the electrode (Z F ) is given as In a general case, the impedance function completed by elements representing the ohmic resistance (R u ) of the electrolyte solution and the double layer capacitance (C dl ) can be written as: The impedance function of the double layer in such type of electrochemical systems can be well approximated by Z dl (ω) = (iωA dl ) -β , where β ≤ 1 (but it is close to 1) [75]. Thus, In principle, it is possible to estimate the parameters of equation (3.21) from the measured impedance data by complex nonlinear fitting. However, the complete expression of the impedance (equation (3.20)) contains 10 free parameters, and the fitting in the case of such a high number of parameters is always a very difficult task. In addition, due to the mathematical structure of equation (3.21), the parameters are expected to be strongly correlated [76]. In order to overcome the difficulties concerning CNLS fitting, simplifications in the mathematical expressions should be introduced. It may be assumed that k 0 is very small, therefore Z 2 (ω) can be neglected or approximated by an ohmic resistance: On the other hand, by substituting The equivalent circuit analog [77] corresponding to the analytical expression given by equation (3.21) is presented in Figure 10. Detailed numeric results of the CNLS fittings are presented in the study by Ujvári et al [78]. Surprisingly, despite of the great number of adjustable parameters, many of them could be determined with a good statistics (for instance Ru, P 0 , R A , C A , P l , A dl , β, ), and reasonable estimated mean values have been obtained for the others. simulated curves calculated by using the optimized parameters. Adapted from [67].

Figures 11b and 11c
show the Bode diagrams (log|Z| versus log f and φ versus log f plots) at 4-4 different rotation rates of the RDE, respectively (E=0.01 V vs. SSCE, T = 25 °C). The Argand diagrams (complex impedance plane plots) are presented in Figure 11a. In Figures 11a-11c, beside the measured values (discrete points), the simulated curves calculated by using the optimized parameters are also displayed (continuous lines). The "visual" inspection of the transformed curves (Bode plots) supplies more valuable information concerning the goodness of a fit than that of the complex plane plot, since complex plane plots usually show a rather "good" fit, while the transformed curves reveal the problems. The physical reality of the parameters is perhaps the most important criterion in checking the correctness of a model. The variation of the appropriate fitting parameters with rotating rates of the RDE could be very informative. Particularly the changes observed in C A and P 0 are of special interest. C A increases monotonically with the rotation rate, as expected for decreasing concentration of chloride ions in the immediate vicinity of the electrode surface. The P 0 vs. ω r -1/2 plot is linear.

Reduction of perchlorate ions in the course of anodic dissolution and corrosion of metals
Most studies concerning the basics of the anodic dissolution of metals focused on relatively simple systems [79]. "Simple systems" ideally consist of single-crystalline or polycrystalline, highly pure metals in acid aqueous solutions free of dissolved oxygen, surface-active substances, and complexing agents. The dissolution behavior in alkaline or neutral media, even in deaerated or aerated aqueous solutions free of surface-active and complexing substances, is difficult to study because of the easy formation and slow dissolution of protective layers. Also, the influence of surface-active substances on the dissolution kinetics and mechanisms of metals is better known in acid than in neutral or alkaline environments.
It is an almost general conclusion drawn from anodic dissolution studies that the nature of anions present in the electrolyte solutions affects the dissolution process to a more or less pronounced extent.
In most mechanistic considerations, this effect is explained by the role of the specific adsorption of anions. Accepting this assumption, and taking into consideration that the adsorbability of the anions on metals is very different depending on the nature of the anion, it would be important to study the dissolution process in the presence of anions with very low adsorbability in order to minimize the role of anion adsorption in the dissolution process, consequently, to obtain experimental data without the interference of anion specific adsorption. According to some very recent views, 4 ClO − are assumed to fulfill this requirement [80,81]. On the basis of the assumption of stability of 4 ClO − sophisticated electrochemical methods, such as EQCM and electrochemical impedance spectroscopy (EIS), were used for the study of anodic dissolution and corrosion of various metals in perchloric acid solutions, for instance, in the case of nickel [82] and iron [80,81].
Detailed mechanistic conclusions are drawn from such studies without taking into account the possible reduction of 4 ClO − under the conditions of the experiments. In a series of communications [83][84][85], the authors pointed out that the reduction of 4 ClO − takes place during the corrosion of Cu, Al, Zn, Ni, and Fe in deoxygenated HClO 4 solutions and emphasized that, for the interpretation of the results of corrosion or anodic dissolution studies in the HClO 4 /metal systems, this fact cannot be left out of consideration. Presence of chlorate and perchlorate ions leads to passivity breakdown and pitting corrosion of iron in sulfuric acid solutions [86]. The origin of the pitting has been attributed to Cl − produced via the reduction of 3 ClO − and 4 ClO − by Fe 2+ . Similar results were obtained for Al in [87]. The XPS experiments detected perchlorate and chloride ions on the surface. Both of these ions may play role in the passivity breakdown.
Oscillatory electrodissolution of Al in perchloric acid solutions has been described [88]. According to the authors, current oscillations occurred over a large range of HClO 4 concentration and over a wide scale of applied potential. During the anodic dissolution of the Al electrode in HClO 4 solutions, perchlorate ions were reduced to chloride ions, which induced pitting corrosion.
The electrodeposition of Cu on Ru(0001) in perchlorate media has also been investigated [89]. XPS was used to determine surface composition. The spectra show that 4 ClO − dissociates into adsorbed Cl − and x ClO − species. The composition of the passive layers formed on Zn electrode in naturally aerated and deaerated 0.1 M KClO 4 solution were studied using chronopotentiometry, electrochemical impedance, and X-ray photoelectron spectroscopic measurements (XPS). Cl − from the perchlorate reduction reaction were detected in the solutions during electrode polarization [90,91].

Interaction of 4 ClOwith iron group metals in aqueous media
The dissolution kinetics and mechanism of iron-group metals has been investigated in a great number of studies [79][80][81][82]92].
The kinetics of the anodic dissolution of iron, cobalt, and nickel was found to be quite similar and a common mechanistic picture can be used for the interpretation of phenomena occurring with these metals.
In the 1960s, Göhr and Kruger investigated the behavior of cobalt in perchloric acid solution using measurements carried out under controlled potential and current [93,94]. The results were interpreted on the basis of thermodynamic data. It was concluded that at open circuit and at small polarizations cobalt corrodes exclusively with hydrogen evolution. The possibility of the reduction of 4 ClO − under these conditions was not considered as a physical reality. Similar conclusions were drawn in case of other iron-group metals as presented in the survey article by Lorenz and Heusler devoted to the problems of the dissolution of iron-group metals [79]. On the other hand, it is an almost generally accepted view that the nature of the anions present in the solution phase in contact with the dissolving metal plays an important role in the kinetics and mechanism of the dissolution process.
It is assumed that the main factor determining the effect of anions in these reactions is the adsorbability of the anion, i.e. the direct interaction of anions with the metal surface. Generally, it is assumed that this interaction does not lead to the chemical transformation of the anion. It is also assumed that the adsorbability of 4 ClO − on iron-group metals is very low; thus, the adsorption of other anions can be studied in the presence of perchlorate supporting electrolyte without significant role of adsorption competition.
However, the validity of this assumption could be questioned in the light of some recent studies [20] and revisiting some results reported in the literature more than fifty years ago [95,96].
As mentioned above, the problem of perchlorate contamination of ground waters is nowadays an important environmental issue [24,[37][38][39], and several methods were suggested and elaborated for reductive elimination of 4 ClO − . One of these methods, reported quite recently, is the reduction with metallic iron. The results obtained from these practical studies can be doi: 10.5599/jese.2011.0003 21 considered convincing evidence of the reduction interaction of Fe with 4 ClO − . Results with Co prove that the rate and extent of the reduction process could also be very characteristic [97]. Under suitable chosen experimental conditions, even this process could be the predominant reaction in the dissolution of metallic cobalt in the presence of 4 ClO − . In contrast to Fe and Co, a relatively slow reduction was found in the presence of Ni [83]. Nevertheless, it can be stated unambiguously that 4 ClO − dissolved in acidic aqueous solutions can be reduced with Ni [83], Fe [98], and Co [97] metals. According to [99], significant reduction of perchlorate to chloride ions occurred on the nickel cathode in a cell with a nickel working electrode and a platinum counter electrode in concentrated solutions of HClO 4 . It has been shown that the mechanism of this process involves platinum deposited in small amounts on the cathode as a result of oxidation of the platinum anode in the perchloric acid solution. The reduction of perchlorate was accompanied by oxidation of the nickel cathode, which can be attributed to chlorate ions formed in the initial step of perchlorate reduction.
To estimate the relative role of 4 ClO − reduction in the overall dissolution process of Ni, Fe, and Co, we summarized some data obtained in previous studies of metal-perchlorate (perchloric acid) interaction (Table 2). It may be seen that the rate of the formation of Cl − from 4 ClO − is relatively low in the case of Ni at ambient temperature, while for Fe and Co, depending on the experimental conditions, the reduction of 4 ClO − could play a substantial role in the overall process. Considering this situation, it could be expected that the results of the usual corrosion measurements carried out in the presence of acidic perchlorate solution should reflect this role in the case of Co and Fe. The data presented in the Table 2 were obtained using metal powders. Thus, they should be considered as average values as during the dissolution process the parameters of the systems are changing continuously. Some of these factors are as follows: i) The H + concentration is changing according to the following equation ( ) ii) The surface area of the metal powder is changing in the course of the dissolution process [100]. The change in the H + concentration could be very important at low initial HClO 4 concentrations. For instance, taking into consideration the data reported in Table 2 for the dissolution of Co at 0.1 mol dm -3 HClO 4 concentration at 25 °C using 25 ml solution, the final H + concentration is ( ) This means that at the end of the dissolution experiment, the hydrogen ion concentration is more than one order of magnitude lower than that in the initial state.
Only in the case of high H + concentration (4 mol dm -3 ) can be expected a practically constant acidity during the whole dissolution experiment. The problems caused by the H + consumption are well reflected in the temperature dependence of the conversion of 4 ClO − to Cl − in the presence of Co powder.
At low HClO 4 concentrations (0.1 and 1 mol dm -3 ), the scattering of the data is remarkable, while in the case of a solution of 4 mol dm -3 HClO 4 , a monotonous increase can be observed.
The change in the pH influences not only the rate of the perchlorate reduction -metal dissolution, but also the adsorption of the produced Cl − . The adsorption phenomena were studied on powdered metals [100]. The principle of the method used was the measurement of radiation intensity originating from labeled adsorbed species on a powdered metal layer sprinkled on a thin plastic foil that serves simultaneously as the window for radiation measurement. The measurements were carried out at ambient temperature in an Ar atmosphere. Radiotracer experiments at different H + -ion concentrations can be seen in Figure 12. The adsorption of Cl − decreases with the pH. It should be noted, however, that attempting the study of the adsorption of labeled Cl − on powdered Co in 0.5 M NaClO 4 solution at low pH values, for instance, pH ≈ 2 (by addition of HClO 4 ) the count-rate vs. time curves (curves 1, 2 and 3 in Figure 13) going through maximum were obtained.
The decrease in the count-rate means that no equilibrium or steady state is attained with respect to the surface concentration of the isotope emitting the radiation. This phenomenon can be explained by the production of inactive Cl − (i.e. by the decrease of specific activity of Cl − species). The displacement of the active species by non-active ones occurs immediately at the surface and this could be the very reason for the rapid decrease of the count-rate in the case of curve (1).
In the course of the corrosion, the pH also shifts from the initial pH ≈ 2 value to higher ones. In order to demonstrate that the decrease in the count-rate is mainly determined by the production of inactive Cl − , HClO 4 was added again to the solution phase shifting the pH to the original value. The results of this procedure presented by curves 2 and 3 in Figure 13 show that the maximum in the count-rate is significantly lower than that for curve 1 and that the maxima are less pronounced in the two former cases. This observation is in agreement with the assumption of the production of Cl − . It is evident that with decreasing specific activity the "isotopic dilution" caused by the production of further Cl − becomes less and less pronounced and this is reflected by the position and shape of the maxima on curves 1, 2, and 3.
Curves 4 and 5 in Figure 13 were obtained following the addition of inactive NaCl to the solution phase. In this case, the "isotopic dilution" was carried by direct addition of inactive − Cl . The effect observed is in accordance with the expectation. In the course of the corrosion of Co powder, Cl − are formed from the 4 ClO − . Thus, the pH dependence presented in Figure 12 was determined starting from pH = 6 and measuring the radiation intensity only a short period following a shift of pH to a preselected value. For more details about anion adsorption see Ref. [101].

Conclusion
The experimental evidence collected during the previous decades unambiguously indicates that the electrochemical stability of perchlorate ions holds no promise. Starting from this statement, experimental data and theoretical considerations based on the concept of stability of 4 ClO − should be rigorously revised, given the large number of such publications in the electrochemical literature. On the other hand, the electrochemical treatment method is one of the ways considered as applicable for solving the perchlorate contamination problem.

Introduction
A continual increase of the number of applications of electrochemistry in modern medical technologies is primarily due to the broader use made of theoretical electrochemical concepts in the functioning of various systems of the human organism. Of note here are the works of Sawyer et al. [1][2] on electrochemically induced intravascular thrombosis (blood clotting) by introducing anodically polarized metal conductors into damaged blood vessels in order to attract negatively charged blood components to the electrode and stop the hemorrhage. These investigations were continued by Guglielmi, et al. and applied for the treatment of aneurisms [3][4][5][6].
At the present time, one of the leading applications of electrochemical concepts in medicine is the ideas of Nordenström and his model of homeostasis as a biologically closed electric circuit [7], in which blood flowing through the blood vessels is treated as a conductor of electrical signals. Vascular walls and other tissues possess electrochemical potentials whose values depend on the state of the tissue (i.e., normal, inflamed, damaged, etc.). Potentials measured with platinum electrodes implanted into the tissue can serve as a source of information about the presence or absence of certain pathological states in the vicinity of the implanted electrode. Based on the above model, Nordenström proposed and realized an electrochemical method of treatment for malignant carcinomas by passing DC current through the affected areas of tissue in order to impose certain potentials upon it [8,9].
Another novel medical treatment also has electrochemical basis, viz., detoxification by means of electrochemically controlled hemosorption [10,11]. In this electrochemical model, the hemosorbent was considered to be a porous electrode immersed in aqueous 0.15 M sodium chloride solution. Based on such an approach, an important conclusion was reached as to the mechanism of interaction between the negatively charged formed elements of blood and the activated carbon hemosorbent. The sign and magnitude of the charge of the hemosorbent surface is determined by its potential. The latter, in turn, depends on the nature of the activated carbon material, since the composition of the electrolyte (blood) remains virtually unchanged.
The electrochemical model of hemosorption has made it possible to find a new approach to identifying new promising materials suitable as potential hemosorbents. This is important, since the requirements towards "good" hemosorbents are quite rigorous and somewhat selfcontradictory: on the one hand, such a material must be indifferent towards blood, while on the other it should maintain high adsorption activity towards the toxicants being removed from blood. By treating the hemosorbent/blood interface as an electrochemical system, authors [10,11] were able to assert the possibility of controlling the removal of toxicants from blood via polarization of the hemosorbent (which is usually a porous carbon material). An important aspect of this, however, is the mechanism of interaction between the carbon material and the electrolyte (blood), since adsorption on activated carbon may include the occurrence of Faradaic processes. Such processes may occur in cases where the potential of the hemosorbent in blood is comparable to the reduction or oxidation potential of the toxicant.
It is well known that Faradaic processes (especially electrooxidation) can lead to highly undesirable effects in blood. For example, anodic oxidation of proteins leads to their denaturation [12]. At the same time, Frumkin showed [13] that the potential of an electrode depends on the adsorption of organic and inorganic substances on it.
Unfortunately, electrochemical data are rarely given or utilized in carbon adsorption research. For instance, authors of [14] considered the properties of the carbon/electrolyte interface when certain cations are adsorbed; however, while the authors [14] recounted Frumkin's conception of activated carbon as an oxygen electrode, they considered chromene-like groups that generate resonance-stabilized carbonium ions to be the initiators of adsorption activity of carbons [15].
Thus, consideration of carbon potential may help elucidate the mechanism of adsorption on carbons and reveal the presence or absence of Faradaic processes during it. For this purpose the mathematical expressions given in [16] can be used, which are based on Frumkin's conception of activated carbon as a perfectly polarizable electrode, i.e. an electrode whose state is fully dependent on the amount of electrical charge consumed [13]. The total charge q of a perfectly polarizable electrode is a function of both electrode potential E and the amount Г of adsorbed electrochemically indifferent surface-active substance. Since the measurements are being made in an open circuit, an assumption can be made that the total charge of the carbon q is constant and therefore its total differential is equal to zero: In line with [16,17], (∂q/∂E) Г = С ∞ , i.e. the capacitance of the electrode at a sufficiently high frequency; and -(∂q/∂Г) E = nF. Therefore: which can be rearranged as Isolating n, one would obtain: Finally, if the assumption is made that the ratio dE/dГ ≈ ∆E/∆Г, i.e., replacing the first derivative of potential with respect to adsorption by the finite ratio of changes in E and Г, the following expression is obtained, from which the effective number of electrons can be estimated: where n -effective number of electrons needed for the elementary act of adsorption and/or Faradaic process, С ∞ (F) -differential capacitance of the adsorbent, ∆E / V -OCP shift corresponding to a relatively short period of time t, F (= 96500 C mol -1 ) -Faraday's constant, ∆Г / mol -change in the adsorption (Gibbs' surface excess) of the adsorbate during time t. Equation (3) can be used for computations of the effective number of electrons. Thus, the empirical data needed to calculate the value of n are the open-circuit potentials of activated carbon and corresponding values of adsorption. The value of differential capacitance С ∞ = 100 F was used for 1 g of activated carbon according to the data in [18]. An approximation was made that the capacitance С ∞ of the adsorbent is weakly dependent on the electrode potential and Gibbs' adsorption of the adsorbate.
With the above considerations, the effective number of electrons transferred in the elementary act of adsorption process or Faradaic process was chosen as the criterion of suitability of carbon materials for use as hemosorbents: viz., the virtual absence of Faradaic processes points to the suitability of such hemosorbents. The adsorption of certain undissociated organic toxicants and cupric ions was chosen for a model-based investigation of the applicability of the concept of effective charge transfer for hemosorption.

Experimental
An IPC-Pro L potentiostat (Volta Co. Ltd, Russia) was used for measurements of polarization curves; it was also utilized for OCP measurements. A saturated silver/silver chloride electrode was used as the reference electrode in all measurements. Adsorption of acetone and 2-methyl-2propanol (tert-butanol, TB) was investigated using gas chromatography. An SRI 310C chromatograph (SRI Instruments, USA) with a 1 m HayeSep-D packed column and a catalytic combustion detector, as well as a Shimadzu GC17 chromatograph with a CarboWAX 20M phase, were used. 0.150 M NaCl or 0.200 M Na 2 SO 4 solutions were used as a supporting electrolyte in adsorption and potential measurements of the organic compounds and in measurements of the initial carbon potentials.
Adsorption of organic compounds and cupric ions was investigated on AG-3 activated carbon; the OCP value of the initial sample measured in 0.200 M Na 2 SO 4 was 50 mV (vs. Ag/AgCl). Modification of the AG-3 carbon was carried out according to [10,11] in the cell shown in Fig. 1. The granules of initial activated carbon were placed in the anode (1) and cathode (2) chambers separated by perforated discs made from a dielectric material (8), with the granules subsequently compressed by the current-carrying discs (6) by turning the bolts (7). A peristaltic pump was used to pass an aqueous sodium chloride solution through the assembled cell at a flow rate of 50 mL/min. Carbon was modified galvanostatically with a current of 0.75 A for 10 to 90 min, with subsequent washing of the cathodically and anodically modified carbons to a pH = 7. As a result of the modification, carbons with OCP values in the range between +475 mV and -775 mV (vs. Ag/AgCl) were obtained and used for further experiments.  Carbon OCP was measured in a non-flow-through apparatus described in [19], shown in Fig. 2. This apparatus was immersed in solution with constant stirring at 100 rpm using a magnetic stirrer. Carbon granules were placed in the cylindrical chamber (3) of the apparatus and compressed against the stainless steel meshes (6,7) by the cover (1). The current-carrying meshes and the reference electrode were connected to the potentiostat, and the time dependence of OCP values was recorded. Since the surface area of the carbon is 1000 m 2 /g, while the surface area of the steel mesh is on the order of 10 -4 -10 -5 m 2 , the potential measured is that of the carbon and not of the current-carrying steel.
Cyclic voltammetry was used to obtain a potentiodynamic charging curve for computation of differential capacitance of AG-3 activated carbon. A single granule of AG-3 carbon was placed in a three-electrode cell. Thermally expanded graphite was used as the current collector and the auxiliary electrode. The cyclic voltammogram was collected at a scan rate of 0.5 mV/s in the range between -800 mV and +600 mV.

Results and Discussion
As noted above, carbon modification in the present work was performed according to the method previously developed in [10,11]. It was shown in [11] that anodic and cathodic treatment of carbon led to significant changes in the composition of its surface functional groups ( Table 1).
As seen from the data shown in Table 1, cathodic treatment of activated carbon drastically reduced the amount of acidic functional groups on carbon surface, while anodic treatment increases their amount. This trend is observed regardless of the acidity of the electrolyte used for modification. Thus, cathodic treatment results in the reduction of oxygenated surface functional groups, while anodic treatment leads to the oxidation of carbon atoms.
OCP data for unmodified and electrochemically modified AG-3 carbon samples during contact with solutions of acetone (in 0.150 M NaCl) and TB (in 0.20 М Na 2 SO 4 ) were obtained as a time dependence, with corresponding concentration changes of the organic compounds in solution. A typical graph of OCP and adsorption, calculated from chromatographic data, is shown in Fig. 3 (for TB on unmodified AG-3); similar dependences were obtained for all other samples. Zero adsorption corresponds to the initial potential of AG-3 in the experiment in the investigated solution.

Differential Capacitance
The differential capacitance of the AG-3 activated carbon was computed based on the potentiodynamic data. An 8.0 mg carbon sample was used, and potentials were scanned in the range between -700 mV and +500 mV at a scan rate of 0.5 mV/s. The resulting dependence of potential on differential capacitance is shown in Fig. 4.
A minimum at C min = 78.7 F per gram and a maximum at C max = 121 F per gram were observed. The average value of differential capacitance from the calculated results was C avg = 95.4 F per gram of carbon, which closely matches the literature data [18]. Thus, the value of capacitance С ∞ = 100 F per gram of carbon was used in all further computations.

Adsorption of Organic Molecules
Based on the final OCP and adsorption values for unmodified and electrochemically modified AG-3 carbon samples, effective numbers of electrons were calculated for acetone and TB adsorption. The results are summarized in Tables 2-3. It should be noted that these calculations reflect the integral character of effective charge transfer in the range from E initial to E final . Calculated values for the effective number of electrons in the course of adsorption of organic molecules on AG-3 carbon were quite small (0.0084 to 0.046). This may indicate that Faradaic processes do not contribute significantly to the adsorption process of these substances. Importantly, adsorption of all of these compounds on AG-3 carbon is accompanied by a positive shift in carbon OCP, which points to an electrochemical mechanism of adsorption (since changes in an electrochemical parameter are observed).

Adsorption of Cupric Ions
Similar adsorption experiments were performed on AG-3 carbon samples in aqueous CuSO 4 . The most important results of these experiments are summarized in Table 4. A typical time dependence of the OCP and concentration (for unmodified AG-3) is shown in Fig. 5. Fig. 6 shows the dependence of the "local" effective number of electrons (calculated between data points) on potential in the course of the experiment described in Fig. 6. Discrete values of (E, n) correspond to the elapsed time, as shown next to each point in the figure. It can be seen from Fig. 6 that the process of adsorption reaches equilibrium after about 30 min, as the "local" effective number of electrons transferred reaches zero (within the precision of the experimental data).   For initial carbon potentials in the range between -775 and -470 mV (vs. Ag/AgCl), the data obtained (Table 4 and Fig. 7) shows effective numbers of electrons n ≈ 2. This corresponds to the two-electron Faradaic open-circuit process of copper electroreduction on the carbon: This is also corroborated by the appearance of a characteristic reddish-colored copper deposit on the carbon granules after cathodically modified AG-3 is immersed in aqueous CuSO 4

Probability of One-Electron Copper Reduction
Importantly, one should consider the probability of the one-electron reduction process in the course of adsorption (as a competing process, with possible further transformation of Cu + ): Cu + + е -→ Cu 0 The determination of Cu + in solution is rather difficult, since it quickly disproportionates in solution: In electrolytic refining of copper [21], bulk [Cu + ] concentration for aqueous CuSO 4 in equilibrium with metallic copper was ca. three orders of magnitude below [Cu 2+ ], i.e. virtually negligible. Other investigations [22][23][24][25][26], however, suggest a stepwise mechanism for the deposition of cupric ions/ionization of metallic copper, with excess Cu + in the interphase region as an intermediate between Cu 0 and Cu 2+ .
While a polarographic method of Cu + determination in solution has been proposed [27], it involves the addition of organic ligands that might interact with activated carbon, making this method impractical for the present work. In addition, if Cu + only exists in excess near the interface, as suggested by [22][23][24][25][26], it would be difficult to detect it at an additional electrode removed from the vicinity of the carbon surface. Thus, it is not possible to elucidate the involvement of Cu + in the investigated processes. However, regardless of mechanistic details, Faradaic reduction of Cu 2+ to metallic Cu 0 undoubtedly occurs in the above experiments; this is also corroborated by the calculated values of n.

Conclusions
Thus, experimental data on the effective number of electrons unequivocally confirm an electrochemical mechanism of hemosorption. All effective numbers of electrons were calculated assuming that the electric double layer capacitance remained constant despite significant potential changes of the carbon. For the model processes of acetone and TB adsorption it was shown that the adsorption of an organic toxicant on AG-3 activated carbon took place virtually without any electron transfer, i.e. Faradaic processes did not occur, while the carbon potential did change in the course of adsorption. The interaction of copper cations with the surface of activated carbon depended on carbon potential: for potentials more negative than -470 mV (vs. Ag/AgCl), the two-electron Faradaic process of copper deposition Cu 2+ +2e -→ Cu 0 occurred, as reflected by calculated values of the effective number of electrons.
Therefore, it is possible to determine the mechanism of adsorbate interaction with an activated carbon hemosorbent by performing open-circuit potential and adsorption measurements. The effective number of electrons can be calculated and utilized as a criterion of hemosorbent suitability, since it allows the detection of Faradaic processes between the hemosorbent and adsorbate. It is important that the phenomenon of partial transfer of charge allows working hemosorbens to be reclassified according to their electric effectiveness.

Introduction
Electrochemical machining (ECM) is often applied to deburring and shaping components made of e.g. iron and steels in aqueous inorganic salt electrolytes [1,2]. Shape and surface of work pieces are modified by high rate anodic dissolution at high current densities (10-150 A cm -2 ) with a special designed cathode as tool. The resulting gap width is usually below one millimetre. High electrolyte flow velocities (5-50 m s -1 ) are required to remove reaction products and Joule heat. The precision of pulsed ECM is higher compared to direct current ECM because reaction products can be removed during pulse off time [3]. Sodium chloride and sodium nitrate are often used as electrolytes in industrial ECM because of low costs and labour safety aspects. Anodic dissolution of iron and many steels in pure sodium chloride and in pure sodium nitrate electrolytes has been investigated in great detail [4][5][6][7][8][9]. The dissolution mechanisms in both electrolytes differ strongly and result in very different features of the machined work piece. In chloride media an active anodic dissolution takes place whereas nitrate anions cause a passivation of the substrate surface and transpassive dissolution can be observed. Methods of characterizing the dissolution mechanism are anodic polarization curves, analysis of the resulting surface quality and weight-loss measurements in order to determine the current efficiency. The anodic dissolution of pure iron in chloride electrolyte is divalent over a wide range of current density and the current efficiency is closely to 100 % [5]. Hydrogen is developed in the counter reaction at the cathode [10]. In nitrate media the evolution of oxygen takes place as a second import anodic reaction and the anodic dissolution of iron leads simultaneously to Fe 2+ ions and Fe 3+ ions [8,11,12]. At low current densities the iron dissolution efficiency is close to zero because the anodic current originates from oxygen evolution. With increasing current density the current efficiency of the anodic iron dissolution increases up to 90 % but oxygen evolution is still present [11]. The reduction of nitrate anions takes place at the cathode [1].
The anodic dissolution of iron and many steels is mass transport-controlled and reaction products enrich at the anode interface [5,6]. Electrochemically inert metal carbides cause an apparent higher current efficiency for carbide-containing steels because the measured weight-loss includes the mass of carbide particles that were detached but not dissolved during the ECM treatment [13,14]. Qualitative anodic dissolution models explain the observed dissolution behaviour and the resulting surfaces. A two-layer model explains the dissolution in nitrate electrolyte [8,15,16]. Because of the current density-dependent divalent/trivalent dissolution of iron, different iron oxides (Fe 2 O 3 and Fe 3 O 4 ) are formed and cover the steel surface as a thin film. An adherent liquid polishing film consisting of supersaturated iron nitrate hydrates is built at the oxide/electrolyte interface and causes a decreasing oxygen evolution rate [11,16]. In chloride media a weakly attached, easily removable black, carbide-rich solid surface film and a thin polishing film consisting of FeCl 2 , Fe(OH) 2 and FeO (Fe x O y Cl z ) are formed at the active-steelanode/electrolyte interface [14]. In NaCl electrolyte the dissolution rate is higher than in NaNO 3 electrolyte due to active dissolution and high current efficiency while in NaNO 3 electrolyte a much better dimensional control can be reached.
The anodic dissolution of mild steel in chloride/nitrate mixed electrolytes has been described by Hoare, LaBoda and Mao [17,18]. From their steady-state anodic polarization curves, film stripping and current efficiency measurements, they concluded that chloride anions cause only a localized attack of passive films formed in the presence of nitrate anions on the steel surface. Depending on the chloride/nitrate ratio, the chloride anions lower the oxidation power of the nitrate anions, prevent the formation of a compact, protective film and only a non-protective salt film is present and the resulting surfaces are severe cratered. At lower potentials an active dissolution caused by the chloride anions takes place while passivation and transpassive dissolution dominates the behaviour at more positive potentials. With increasing chloride/nitrate ratio, the work piece surface is passivated at higher potentials [18]. Chloride/nitrate mixed electrolyte have been used in pulse and pulse reverse through-mask ECM of channels for flow field in stainless steel bipolar plates for PEM fuel cells [19], where the process improved the surface roughness of the channels at the expense of the process time and the rib/channel width uniformity compared to the process in pure chloride electrolyte.
According to the reported results from Mao et al [18] we describe in this paper more detailed insights and process conditions under which the dissolution mechanism of soft annealed steel in chloride/nitrate mixed electrolytes can be externally controlled, i.e. switched between two mechanisms. The influence of varying flow conditions at a rotating disc electrode and in a flow channel as well as the influence of applied potentiodynamic, galvanodynamic and galvanostatic pulse series show clearly the possibility to switch in one and the same electrolyte exclusively by external control (current or voltage, hydrodynamics) between both dissolution mechanisms allowing fast machining rates using the chloride mechanism and high precision by the nitrate mechanism in order to obtain crater-free surfaces.

Specimens
Spheroidization-annealed 100Cr6 cylinders (annealed to globular cementite, 5 mm diameter) were used as specimens from which the cross sectional surfaces were investigated. The 100Cr6 steel is mainly composed of approximately 96 wt. % iron, 1.5 wt. % chromium and 1 wt. % carbon. Carbon is contained as C in the steel matrix (20 % of total C content), as globular Fe 3 C (40 % of total C content) with diameters of approximately 1-3 μm and as 100 nm-slices Fe 2.5 C (40 % of total C content). The metal carbides were considered to be electrochemically inert under the applied conditions. Directly before and after the experiments the specimens were washed with distilled water and isopropanol.

Electrolytes
Aqueous solutions of sodium chloride and sodium nitrate served as electrolytes. The total salt concentration c tot was set to 3 mol L -1 with variation of chloride/nitrate mole ratio. The chloride/nitrate concentration ratio is given as the relative chloride concentration θ (Eq. 1).

Rotating disc electrode (RDE)
The measurements of potentiodynamic and galvanodynamic scans or cyclic voltammograms (CVs) were performed with a rotating disc electrode and potentiostat VoltaLab 80 PGZ402 & VoltaMaster4 Software (Radiometer Analytical SAS). The front surfaces of the 100Cr6 specimens were ground mechanically with SiC papers and the cylinder barrels were masked with an insulating galvanic polyester tape before each measurement. These samples were used as working electrodes (WE) in a three electrode cell completed by a platinized titanium net as counter electrode (CE) and an Ag/AgCl/sat. KCl reference electrode (RE). WE and CE had an electrode distance of approximately 1 cm. The scan rate for potentiodynamic experiments was set to 50 mV s -1 . In this paper all potentials E are referred to the standard hydrogen electrode (SHE). Under the typical rotation rate of 3000 rpm laminar flow conditions are expected. Incipient turbulences must, however, be expected because of grinding marks at the work piece surface, roughness due to anodic dissolution and recess of the surface below the galvanic polyester tape used to mask the side walls of the sample cylinders.

Flow channel
A flow channel in form of a parallel plate reactor was used. The flow velocity was set to 7 m s -1 assuming turbulent flow conditions [20]. Mechanical grinding was done for preparation of the specimens. A steel cathode with 5 mm diameter was positioned directly opposite to the 100Cr6 specimen with a gap of 0.75 mm before start of anodic dissolution. The electrodes were fixed and, therefore, the gap width increased during experiments. With this two-electrode configuration galvanostatic pulses up to 40 A cm -2 were applied with a custom made galvanostat (plating electronic GmbH, Germany). No current was applied during pulse off time. The pulse times were in the range of milliseconds. The duty cycle γ is on on off t on is the pulse on time and t off is the pulse off time. An averaged current i d can be calculated with γ and pulse current i p (Eq. 3 [21]).
Multiplying i d with measuring time t tot gives flowing charge Q (Eq. 4): The theoretical weight loss Δm theo was calculated by the Faraday law (Eq. 5): where M is the molar mass of iron and F is the Faraday constant. The dissolution valence z was taken as 2 in all calculations knowing that the valence is between 2 and 3 in NaNO 3 electrolyte depending on current density [11]. Current efficiency η is then given by equation 6: where Δm tot is the total experimental mass loss of the specimens which was obtained as mass difference before and after the dissolution. Duty cylces γ of 0.625, 0.5 and 0.167 were applied with charges Q between 50 and 300 C.

Surface characterisation
Observation with optical microscopes was used for rough classification of the dissolution mechanisms. The topography of the surface was measured with confocal microscope nanofocus μsurf (NanoFocus AG), for which regions in the middle of the surface were chosen at which minor streaming artefacts occur. The 3D average roughness S a was then calculated with software nanoExplorer XT. The topography was filtered in waviness and roughness by a robust Gauss-filter with a cut-off wavelength of 0.25 mm. S a is the arithmetic mean of the absolute distances of the surface points from the mean plane and is given by equation 7: N and L are the indexes of measured points in xand y-direction. This method facilitates a relative comparison of the resulting surface roughnesses. Additionally, the resulting surfaces were examined with scanning electron microscope (SEM) EVO 50 VPX (Carl Zeiss AG). An Everhart-Thornley-detector and an acceleration voltage of 20 kV were used for 1000× enlargement with a work distance of 10 mm.

Sharp transition from active to transpassive dissolution at resting electrode
Linear sweep voltammograms (LSV) were recorded between -0.5 V and 7.2 V. Figure 1 shows the LSVs in pure NaCl electrolyte (θ = 1), in pure NaNO 3 electrolyte (θ = 0) and in mixed electrolytes with θ = 0.67, 0.5 and 0.33 at a resting electrode. The curves of the pure electrolytes show the well known behaviour of active (NaCl) and transpassive dissolution (NaNO 3 ) [1]. In NaCl electrolyte a limiting current is reached because the concentration of reaction products increases to saturation. Thus a limiting current plateau is built until the overlimiting current is reached and the current increases again. In NaNO 3 electrolyte fluctuating current densities are caused by the evolution of oxygen at the WE which sets in at 1.8 V. The curves that were measured in the three mixed electrolytes overlap with each other. They agree qualitatively with earlier reports about anodic dissolution in mixed electrolytes [18]. Surprisingly, there is no strong influence of the relative chloride ion concentration on the anodic dissolution at the resting WE in Figure 1. At low potentials active dissolution takes place, indicated by an increasing current density from 0 V on. The surface is suddenly passivated at approximately 1.6 V and the current density decreases to nearly 0 A cm -2 , oxygen evolution starts at 1.8 V and transpassive dissolution sets in at higher current densities. With respect to the models of metal/electrolyte interface [11,14,16] the following mechanistic model can be proposed. At lower current densities which occur at potentials below 1.6 V at the resting electrode, the dissolution mechanism must be dominated by the chloride anions and the LSV is very similar to the curve in pure chloride solution. In this paper this section of LSVs will be called "chloride curve". Due to active dissolution, the chloride polishing film is formed. The nitrate anions are larger than chloride anions. Most likely their incorporation in the polishing film changes the structure of the polishing film so that no diffusion controlled current regime (plateau in the pure NaCl electrolyte) is observed in mixed electrolytes. The higher the current density, the higher is the active dissolution rate of the steel. Chloride anions have to diffuse from the bulk of the electrolyte through the diffusion layer to the steel surface. When the current density becomes higher than approximately 1.4 A cm -2 not enough chloride anions reach the steel surface. Therefore, the nitrate polishing film composed of iron nitrate hydrates is formed. This film could withdraw the water from the chloride polishing layer. Hence, iron chloride could be precipitated and could be displaced by the nitrate polishing film and the anodic dissolution is dominated by the nitrate anions at potentials higher than 1.8 V. Hence, the transpassive dissolution sets in at higher current densities at E > 1.8 V and this section of LSVs is called "nitrate curve" below. The transition from active to transpassive dissolution can also be observed visually (see Figure SM-11 in Supporting Information). During active dissolution the assumed reaction products creep down as a green viscose film. After passivation, the evolution of oxygen breaks off the visible flakes of the black carbide-containing layer from the anode surface. It is suggested that this surface layer is equal to the one in pure NaCl electrolyte. Haisch et al. [7] investigated the surface layer obtained in pure NaCl electrolyte by energy dispersive X-ray microanalysis. They reported that the particles of this layer mainly consist of Fe, Cr and C and suggested that they originate from the carbides in the steel matrix [7]. Transpassive dissolution can be observed by the formation of oxygen bubbles and yellow-brown colouration of the solution.

Influence of anion transport on the transition from active to transpassive dissolution at RDE
For further experiments the WE was operated as RDE in order to improve the removal of oxidation products and simulate the electrolyte flow under technical ECM conditions. This leads to an increasing of limiting current in NaCl electrolyte [22] and to a less fluctuating current density in NaNO 3 electrolyte in corresponding LSVs at 3000 rpm ( Figure 2).
In pure 3 M NaCl electrolyte no limiting current plateau is observed within the resulting current density range (curve 1). The electrolyte changes its colour from colourless to green and green precipitation occurs. In pure NaNO 3 electrolyte the current density rises linearly at 1.8 V (curve 14). Because transpassive dissolution sets in at even higher current densities, the colourless electrolyte changes to yellow-brown and yellow-brown precipitations are observed. With decreasing θ, the activation potential for the chloride dominating activation potential is shifted to more anodic potentials. The nitrate anions disturb the activation [23]. With respect to the proposed mechanistic model, increasing θ leads to higher current densities at which the change of dominating mechanism occurs due to the transport of chloride anions from the bulk of the electrolyte through the diffusion layer to the steel surface. With increasing θ formation of the chloride polishing film can be maintained to higher current densities until replacement by the nitrate polishing film and commencement of transpassive dissolution. In this paper the term passivation is not used for this phenomenon because the current density does not fall to nearly zero for θ ≥ 0.07. Transpassive dissolution sets in directly. The LSVs in mixed electrolytes with θ > 0.83 indicate no sharp decrease in current density. Both mechanisms compete and the relatively small concentration of nitrate anions cannot create an abrupt change of the dominating mechanism. Curve 13 shows that for θ = 0.03 the flux of chloride ions is too small to induce and maintain active anodic dissolution in comparison with θ = 0.05 (curve 12). The insert in Figure 2 shows separate forward and backward LSV in mixed electrolyte with θ = 0.5 arranged like a CV. The backward LSV corresponds completely to the backward LSVs which are obtained in pure NaNO 3 electrolyte due to stable transpassive dissolution.  (4); 0.77 (5); 0.67 (6); 0.5 (7); 0.33 (8); 0.13 (9); 0.1 (10); 0.07 (11); 0.05 (12); 0.03 (13); 0 (14). Insert: forward and backward LSV in mixed electrolyte with θ = 0.5.
The transport of chloride anions and, therefore, the transition from chloride curve to nitrate curve can also be controlled by rotation rate. Increasing the rotation rate for a given θ = 0.5 increases the current density at which the dissolution mechanism changes (Figure 3) due to a higher flux of chloride ions to the surface. This is in agreement with the proposed mechanistic model. The mass flux of chloride anions J Clto the disc surface can be estimated by using equation 8 [24]:  (2); 500 rpm (3); 1000 rpm (4); 2000 rpm (5); 3000 rpm (6), 4000 rpm (7).
where D Cl -= 2.032·10 -5 cm 2 s -1 [25] is the diffusion coefficient of Clanions at infinite dilution in water. The cinematic viscosity ν = 1·10 -2 cm 2 s -1 [24] is assumed to be the one of water. ω is the angular velocity of the RDE and c Cl -the bulk concentration of chloride anions. With equation 8 only the diffusion is considered whereas migration is neglected. In the mixed electrolyte with θ = 0.5 the concentration of chloride is 1.5 mol L -1 . With rotation rate 3000 rpm and ω = 2·π·3000 rpm = 314.2 s -1 , J Cl -is 2.64·10 -5 mol cm -2 s -1 calculated from equation 8. This calculation does not consider the larger existing nitrate anions in the solution. The nitrate anions could act as obstacles for the chloride anions flux but migration could lead to an increase of J Cl -. Therefore, the calculated mass flux of chloride anions is only an approximation. The dissolution rate of iron J Fe2+ at the surface can be calculated by using equation 9, which is derived from Faraday law (Eq. 5): 2 Cl ,lim Fe where j Cl -,lim = 3.6 A cm -2 is the current density at which the active dissolution mechanism cannot be maintained any longer. The calculated value J Fe2+ = 1.87·10 -5 mol cm -2 s -1 is in agreement with the proposed model considering that a Fe 2+ flux of 1.87·10 -5 mol cm -2 s -1 would require a Clflux of 2·1.87·10 -5 mol cm -2 s -1 to form stoichiometrically FeCl 2 . The active dissolution in mixed electrolytes is mass transport-controlled with respect to chloride ions. From the LSV experiments it is concluded that under RDE conditions a sharp distinction is possible between active (chloride anion-dominated mechanism) and transpassive (nitrate anion-dominated mechanism) anodic dissolution mechanisms.

Switching between both mechanisms by potential control at RDE
Sequential LSVs which means in this paper that several LSVs are applied directly one after the other show that switching between the two dominating dissolution mechanisms is easily possible by potential control. The start potential can be chosen in the area of the chloride or the nitrate curve. Curve 1 of Figure 4 starts in the potential region of the nitrate curve. After a short initial rise of the current density j (Figure 4 insert), j decreases and the curve follows the nitrate curve. The initial rise of j after switching from active to transpassive dissolution is reproducibly observed (curves 3, 6 in Figure 4). The reason of this effect is still unknown. Maybe chloride anions and nitrate anions which are initially present at the anode cause a situation where both mechanisms take place simultaneously. After the chloride anions are consumed, the dissolution follows the nitrate curve. Curve 2, initiated at 0 V, was applied directly after curve 1 and follows exactly the chloride curve. The surface is activated. Stepping the potential to 4.2 V, i.e. to a section of the LSV dominated by the transpassive dissolution, causes a change of dominating mechanism indicated by the large decrease of current density within a very short transition time (curve 3). The resulting current density after that switch is so high that the consumption of chloride ions cannot be maintained by mass transport and active dissolution ceases. Curve 4 starts within the section of the chloride curve (at 0.7 V). An instantaneous slope can be seen before the current density increases linearly with increasing potential. This slope fits to the chloride curve in the complete LSV but the current density is approximately 1 A cm -2 lower. It seems that the chloride curve is shifted to more anodic potentials for presently unknown reasons. One explanation could be that an acidification caused by oxygen evolution at anode/electrolyte interface influences the processes and shifts the chloride curve. A passivation is observed at approximately 1.5 V in this curve. So the next linear sweep in section of nitrate curve is following the expected behaviour of transpassive dissolution (curve 5).

Switching between both mechanisms by current density control at RDE
Programmed current chronopotentiometry (PCC) [26], i.e. the linear variation of the current in a galvanodynamic sweep experiment gave similar results like LSVs (see Figure SM -12). At low current densities active dissolution takes place in mixed electrolytes and the potential increases linearly with increasing current density. This section of PCC fits to the PCC in pure NaCl electrolyte. When a certain current density (depending on chloride mass transport) is reached the mechanism switches to transpassive dissolution which is indicated by an abrupt potential increase. From then on, the potential increases linearly with current density like the PCC in pure NaNO 3 electrolyte. This is in agreement with the proposed mechanistic model. An example for switching between both dominating mechanisms by current density control is shown in Figure 5.  (1) to (5) directly successive measured PCCs with current density limits from 0 A cm -2 to 1 A cm -2 (1); 4.6 A cm -2 to 5.1 A cm -2 (2); 0 A cm -2 to 1 A cm -2 (3); 4.6 A cm -2 to 5.1 A cm -2 (4); 0.3 A cm -2 to 1 A cm -2 (5); complete PCC (6); θ = 0.5; rotation rate 3000 rpm; scan rate 51 mA cm -2 s -1 .
Curve 1 starts at current density of 0 A cm -2 and active dissolution sets in immediately. A current density step to 4.6 A cm -2 leads to a change of the dominating mechanism to transpassive as expected (curve 2). This is indicated by an abrupt increase of potential. Stepping back to 0 A cm -2 (curve 3) leads to a surprising effect. Passivity of the surface does not break down and transpassive dissolution sets in. After a current density ramp in the range of transpassive dissolution (curve 4) a slight cathodic current density of 0.3 A cm -2 leads to an activation of the steel surface (curve 5). Curve 6 shows a complete PCC.
From the LSV experiments it is concluded that not only a sharp transition from active to transpassive anodic dissolution due to mass transfer rate of chloride anions occurs, but switching between the dominating mechanisms by potential and current density control is possible as well.

Flow channel experiments
Flow channel experiments permit a more realistic modelling of ECM experiments under various flow conditions using galvanostatic pulses. The anodic dissolution in pure NaCl electroylte, in pure NaNO 3 electrolyte and in mixed chloride/nitrate electrolytes was investigated. It was assumed that flow conditions in flow channel are more turbulent than in RDE experiments. The more turbulent flow causes a thinner diffusion layer and better mass transfer of chloride ions. Therefore, the chloride-dominated mechanism can be maintained at higher current densities, or the chloridedominated mechanism at a given current density should be sustained by an electrolyte with a lower θ. This hypothesis was tested with relative low values of θ of 0.07, 0.17 and 0.33 by determining the current efficiencies according to equation 6. The dissolution valence z was set to 2. Because of other probable reactions, such as the formation of trivalent iron and the formation of trivalent chromium species, the experimental results are useful only for relative comparisons between similar experiments of the same material.
Experiments were performed for pulse current densities j of 5, 10, 20 and 40 A cm -2 ( Figure 6). The current efficiencies in pure NaCl electrolyte (curves 5) and in pure NaNO 3 electrolyte (curves 1) correspond to literature values [11,14,15,20]. In pure NaCl electrolyte η is about 100 %. Decreasing η at higher j could be caused by additional formation of trivalent iron. In pure NaNO 3 electrolyte it can be distinguished between passivation of the surface, transition zone and transpassive dissolution. The region of the transition zone is shifted to higher current densities with lower γ. Total experimental mass loss Δm tot includes the erosion of electrochemically inert metal carbides. Often the ratio Δm tot /Δm theo is quoted as current efficiency that might exceed 100 %. From known carbon content and its compounds Δm tot can be corrected by subtracting the mass of electrochemical inert components Δm C in order to obtain the carbide-free weight loss Δm by equation 10 (Details in SM-3): tot inert corr theo theo 100 100 For example in pure NaCl electrolyte with γ = 0.625 and 0.167 at 5 A cm -2 a η = Δm tot /Δm theo of 112 % and 115 % are measured and yields a corrected η corr of 100 % and 102 %. There is still a trend of η corr being slightly larger than 100 % indicating a more complex dissolution mechanism for 100Cr6.
With γ = 0.625, the current efficiencies in the mixed electrolytes are rather uniform (curves 2 and 4, Figure 6a). At 20 A cm -2 and 40 A cm -2 η is about 87 % and is approximately the same as in pure NaNO 3 electrolyte. This is in agreement with the results of a microscopic inspection of the resulting surfaces at 40 A cm -2 with θ = 1, 0.33 and 0 (Figure 7a-c). A very similar appearance of the surfaces is obtained in mixed electrolyte (θ = 0.33, Figure 7b) and in pure NaNO 3 electrolyte (Figure 7c). The current efficiencies in the mixed electrolyte and in pure NaNO 3 electrolyte (87 % and 84 %, respectively) and the surface roughnesses S a (0.79 μm and 0.74 μm, respectively) are approximately the same (Figure 7b-c). Moreover, the SEM micrographs appear equal and are distinctly different from those obtained in pure NaCl electrolyte (θ = 1, Figure 7a). These are all indications of a nitrate anion-dominated dissolution mechanism at high current densities which is in line with the RDE results. Embedded small globular carbides can be seen clearly in the SEM micrographs in Figure 7a-c. The influence of additional turbulent flow due to the emerging edge can be seen in the optical micrographs, especially in Figure 7c for θ = 0. The higher material removal at the edge of the samples can be explained by higher current density which can be calculated by secondary current distribution. Applying 5 A cm -2 in mixed electrolyte with θ = 0.33 leads to a very porous, black surface structure with deep holes (Figure 7d). The porosity is so high that S a cannot be measured with confocal microscopy. Instead of a dominating chloride dissolution mechanism it is proposed that both mechanisms proceed simultaneously at different regions of the surface. The current efficiency of 57 % is an indication for that. Active dissolution produces deep holes while the transition zone from passive to transpassive dissolution obtained in pure NaNO 3 electrolyte produces the black surface film in between. This black surface film could correspond to the surface films described by Haisch in direct current experiments in pure NaNO 3 [15]. Described simultaneous dissolution by both mechanisms was not seen in RDE experiments. These observations illustrate the importance of flow conditions for the electrochemical dissolution reaction and the turbulence increasing from RDE to flow channel.
With γ = 0.5 the influence of γ becomes apparent. For θ = 0.07 (curve 2, Figure 6b) and 0.17 (curve 3, Figure 6b) there are current efficiencies measured at 5 A cm -2 and 10 A cm -2 which indicates a simultaneous chloride-dominated and nitrate-dominated dissolution. The current efficiencies at 20 A cm -2 and 40 A cm -2 approach those in pure NaNO 3 electrolyte. With θ = 0.33 (curve 4, Figure 6b) a minor influence of chloride anions is obtained at 20 A cm -2 and 40 A cm -2 because of little higher current efficiencies than in pure NaNO 3 electrolyte (93 % and 88 % compared to 82 % and 83 %). In this mixed electrolyte at 5 A cm -2 and 10 A cm -2 the anodic dissolution is dominated by chloride anions indicated by current efficiencies, surface roughnesses and the optical impression (Figure 8a-d). Electrolyte flow direction in optical micrographs and in SEM micrographs is from left to right.
This result is in agreement with the proposed mechanism which is based on the idea of chloride-dominated dissolution mechanism at lower and nitrate-dominated dissolution mechanism at higher j. The more pronounced edge effect due to the additional flow effect at the emerging edge can be seen at the surface in mixed electrolyte with θ = 0.33 at 10 A cm -2 as well (Figure 8d). Here the surface has a completely different appearance. This is another hint for the importance of flow conditions. For the duty cycle γ of 0.167 the current efficiencies in the mixed electrolytes approaches η in pure NaCl electrolyte (Figure 6c). The current efficiencies with θ = 0.33 are equal to the ones in pure NaCl electrolyte except for the current density of 40 A cm -2 . At this current density, the current efficiency is η = 97 % in mixed electrolyte with θ = 0.33. This value is in between the values obtained in the pure chloride and nitrate electrolytes.
Also the surface quality obtained in the mixed electrolyte (Figure 9e and g) appears similar to those obtained in pure NaCl electrolyte (Figure 9d and f) at 10 A cm -2 and 20 A cm -2 . The S a values indicate that the chloride dissolution mechanism dominates. Lower current densities show a slightly different behaviour. Figure 9b shows the surface obtained at 5 A cm -2 in mixed electrolyte with θ = 0.33 and γ = 0.167. It is rougher than those obtained in pure NaCl with the same γ ( Figure 9a) and in mixed electrolyte with the same mixing ratio and larger duty cycle of γ = 0.5 (Figure 8b). A value S a > 1 μm indicates that under these specific conditions a higher roughness is obtained in the presence of both anions. Perhaps the chloride ion flux towards the surface is insufficient with respect to competing nitrate anions. At higher current densities of 40 A cm -2 in a mixed electrolyte with θ = 0.33 ( Figure 9i) the surface show features found on surfaces obtained in both pure NaCl and NaNO 3 electrolytes (Figures 9h and 9j). It is concluded that both mechanisms proceed simultaneously at the same regions of a surface (in contrast to the situation found in Figure 7d). Applying different duty cycles in series was tested with γ = 0.625 (Q = 200 C), 0.5 (50 C) and 0.167 (50 C) at 5 A cm -2 with θ = 0.33 in order to assess the surface quality after each step.  In Figure 10, three specimens are shown, one processed just with the first step (Figure 10a), the second processed with the first two steps (Figure 10b) and the third processed with all three steps ( Figure 10c). The conditions in Figure 10a were identical to Figure 7d except for the use of a lower charge (200 C instead of 300 C). Because both dissolution mechanisms take place simultaneously, a black porous surface is observed (Figure 10a). However, less material is removed in Figure 10a than in Figure 7b causing a less distinctive porosity. S a is still larger than the measurable range. The next step (γ = 0.5) in which the dissolution is dominated by chloride anions efficiently removes the black surface (Figure 10b). Only a few pits and an edge at transition from flat plateau to the rim of specimen remain. Applying all three steps decreases the roughness in the last step (Figure 10c). The last step in which chloride anions dominate the dissolution flattens the surface and a smooth transition to the rim of the specimen is achieved. A look at the SEM micrograph shows that the surface is equal to the one in Figure 9b with the same duty cycle and chloride anion ratio (γ = 0.167, θ = 0.33).

Conclusions
The anodic dissolution of 100Cr6 in chloride/nitrate mixed electrolytes was investigated. LSV at RDE assuming mainly laminar flow conditions showed that either active or transpassive dissolution can be achieved by controllable mass transport conditions (bulk chloride concentration and rotation rate). This is indicated by an abrupt decrease in current density with increasing anodic potential. While active dissolution due to chloride-dominated mechanism takes place at lower current densities, transpassive dissolution due to nitrate-anion dominated mechanism is observed at higher current densities. By changing the potential or current density, it is possible to switch reversibly between both dissolution mechanisms.
The following dissolution mechanism is proposed. If the flux of chloride ions relative to the current density is too small, the chloride anion mechanism cannot be maintained and transpassive dissolution due to nitrate-dominated dissolution commences. This transpassive dissolution is then stable in the observed current density ranges. The anodic dissolution is mass transport-controlled and depends on the anode/electrolyte interface layer.
Flow channel experiments applying galvanostatic pulses showed that under assumed turbulent flow conditions a separation of dominating dissolution mechanisms can be observed under specific conditions while both mechanisms may also proceed simultaneously under other conditions. There is not only an influence of anion ratio but also on current density, pulse parameters and flow conditions.

Introduction
Arsenic (As) is one of the most toxic elements that can be present in wastewaters and environment in general. Therefore, the treatment of groundwater has been investigated for decades for the possible use of existing water stocks with appreciable As contents [1], in particular of those close to the mines, as reported by Leist et al. [2]. Treatment of As-containing solutions, being of natural origin or prepared by dissolving chemicals in pure water, has been tackled for years. It seems that As compounds -particularly those in the pentavalent form -exhibit interesting sorption capacities on Fe(III) hydroxides. Adsorption also occurs on Al(III) hydroxides, although not as strongly as on Fe(III) species [1,3,4]. Coagulation and electrocoagulation, therefore, seem interesting and suitable methods for the treatment of arsenic-containing species. The two treatment techniques do not differ very much. In both cases, the treatment consists of two steps: (i) supply of the liquid to be treated with Fe(III) species, either by Fe electrode dissolution to Fe(II) and subsequent oxidation to Fe(III) upon air oxidation or by addition of Fe salts, (ii) complexation-adsorption of As species on Fe(III) species. Since the treatment is often carried out at pH above 3, trivalent Fe species are in the form of solid hydroxide Fe(OH) 3 .
The largest part of research on this subject has been focused on potabilization of water [2,3,5,6]. Although the content of As-salts in the analyzed waters was as large as 100 mg/L, the chemical oxygen demand (COD) content of the solutions to be treated was very low and the presence of other species could be disregarded. In this study, we investigated the electrocoagulation treatment of As-containing industrial wastes for abatement of the COD and As species by using dissolution of iron anodes. Electrocoagulation technique relies on the destabilization of suspended matter by reduction of the absolute value of the zeta potential by the presence of electrogenerated trivalent metals. This technique has been used in and investigated for the treatment of various wastewaters [1,3,5,7,8]. The waste generated from paper mill industry has a slight arsenic content at ppb level. The addition of arsenic salts to the original waste makes it possible to investigate the treatment of mixed wastewaters. For this purpose, the published data on the treatment of single solutions of arsenic salts were compiled and analyzed using a simple model for electrocoagulation published previously [9] and relying upon two parameters. Moreover, the treatment of the industrial wastewater with negligible concentrations of As allowed for the estimation of the two parameters related to the COD abatement. Numerous electrocoagulation tests with the waste after the addition of various amount of As species were conducted. The interpretation of the data was made taking into account the presence of the two pollutants to examine whether the presence of As could affect COD removal in the electrocoagulation process, as it has been observed in the case of Cr-containing industrial solutions [7].

Wastewater samples and chemicals
The wastewater was collected from a local paper mill industry (Clairefontaine, France) downstream of a primary settling stage and biological treatment by activated sludge. The waste already used in the previous investigation [10] had pH=7.7, COD=285 mg O 2 /L, turbidity = 35 NTU, and a very low As content near 3.8 µg/L. The COD level was mainly due to the presence of dissolved or suspended organic matters. Because of the relatively high conductivity of the wastewater, 1.22 mS cm -1 , no supporting electrolyte, such as sodium chloride, had to be added.
The concentration of arsenic species was enhanced by adding small amounts of arsenic acid disodium salt, Na 2 HAsO 3 (Purissimum, Fluka), to the wastewater. Arsenic concentration in the wastewater obtained was below 1 mg/L. The moderate COD of the industrial waste allows to have a reasonable organic pollution to As pollution ratio, so that interactions between As and the organic pollution could be investigated.

Electrochemical reactor
Treatment runs have been carried out batchwise by recirculation of 2.5 liters wastewater in the flow rig consisting of a reservoir tank, peristaltic pump, flow meter, and an electrochemical cell doi: 10.5599/jese.2011.0002 57 [9,10]. The electrochemical cell had a rectangular cross-section and was provided with two flat facing Fe electrodes, being 15x7 cm 2 . The electrode gap was maintained at 2 cm and the liquid was circulated at 300 cm 3 min -1 . Runs were carried out for periods of time ranging from 45 to 90 minutes, depending on the current density applied, which ranged between 5 and 20 mA cm -2 . The cell voltage was monitored along the run..

Pollutants analysis and quantification
Liquid fractions of 10 cm 3 were collected at regular intervals along the runs. The concentration values of the various species were corrected for the change in the liquid volume caused by the regular sampling. Conductivity and pH were monitored using a multi-parameter instrument (Consort C931). A small volume of each sample was acidified by nitric acid in order to determine the Fe and As concentrations by ICP-MS (Series X7, Thermo) after suitable dilution of the acidic solution. For arsenic species, a small liquid volume was submitted to microwave digestion using a START-D instrument prior to injection to the ICP-MS instrument.
The rest of each sample was allowed to settle for at least twelve hours before the analysis of the clear fraction was carried out. The turbidity was measured using an IR beam at 890 nm and detection of the light dispersion using a Hanna Ins. LP2000 spectrophotometer, and COD was determined by the standardized colorimetric method after high temperature oxidation with excess chromic acid and subsequent measurement of the optical density at 530 nm using a HACH 2400 spectrophotometer. The estimated accuracy in the COD measurement was 25 ppm.

Electrocoagulation of the industrial waste
The industrial waste with a low content of As species was submitted to electrocoagulation at two levels of the current density. As expected, the COD level was observed to decrease regularly with time in spite of the slight scattering of the data due to the uncertainty in the assay. Moreover, electrocoagulation did not allow the entire removal of the organic charge since the final COD level was usually at approximately 50% of the initial value, even upon longer electrocoagulation runs. The concentration of dissolved iron increased regularly during the run. The overall current yield of the dissolution at time t was calculated as follows: where V is the volume of the wastewater under treatment, n Fe the number of electrons involved in the dissolution, taken as 2 because the anode dissolution leads to Fe(II) species, which rapidly oxidizes to Fe(III) in the presence of air oxygen. In equation (1), M Fe is the molecular weight of iron, i the current density applied, and A the geometrical area (105 cm 2 ). The current yield was observed to increase slightly over time, from 0.5 in the first minutes and reaching 0.8-0.85 after 30-40 minutes. This could be caused by partial inhibition of the iron electrode immersed in the liquid in the early stage of the run, as the electrode probably exhibited only moderate corrosive properties. As often reported, the reduction in COD level is mainly governed by the amount of coagulant generated at the anode surface (Fig. 1). The experimental variation was modeled using a previous overall model [11] relying upon the instantaneous adsorption or complexation of the pollutant charge -expressed by the COD -with Fe(III) hydroxide. The presence of As species below 4 µg/L was neglected here.
Fe(III) complexes organic matter, expressed by the COD level, S, as follows: where coefficient n 1 is expressed in mg/L O 2 per mg/L Fe species, since S is expressed by the COD level. For the sake of simplicity, the complex formed in the treatment is written FeS, regardless of n 1 value.
Fitting of the experimental variation of the COD led to the estimates for K 1 and n 1 , even though a narrow-valley situation in the optimization procedure was encountered. As formerly observed in the treatment of other industrial wastes [12], moderate COD levels in relation to ratio (K 1 /n 1 ) resulted in the higher accuracy obtained in determination of the ratio. Nevertheless, the two parameters could be estimated at n 1 = 2.1 mg O 2 per mg Fe, and K 1 = 0.029 L/mg (Fig. 1).

Electrocoagulation of pure solutions of arsenic salts
Before the compilation of published experimental data on As removal by electrocoagulation from waters possessing very low COD or turbidity levels, physicochemical properties of arsenic salts are briefly discussed.

General features of As species in wastewaters
Arsenic can be encountered in the form of either organic compounds with methyl groups or inorganic species. The most stable forms in ores or waste are arsenious salts (the trivalent form) and arsenate salts (pentavalent form). As(III) form is obtained from arsenious acid, HAsO 2 , or its hydrated form, H 3 AsO 3 [13]. The two acids exhibit weak acidity, with a pKa near 9.2. As(V) is obtained from arsenic acid, H 3 AsO 4 . The acidity constants of the triacid are near 2.25, 6.77, and 11.60 [8,14]. Therefore, for electrocoagulation of As-containing wastewater with an initial pH near 7 and increasing up to 9 in the course of the treatment, As mainly has the following forms: H 3 AsO 3 with slight amounts of H 2 AsO 3 for As(III) and H 2 AsO 4 and HAsO 4 2ions for As(V). The latter two species are prone to adsorption on Fe(III) hydroxide [15,16].
Pentavalent salts might be considered as oxidants, i.e. existing at appreciable extent under oxidizing conditions. Pentavalent arsenic is in equilibrium with As(III) according to: H 3 AsO 4 + 2 H + + 2 e -↔ HAsO 2 + 2H 2 O E 0 = 0.560 V Considering a solution with a pH 7, the equilibrium potential deduced from Nernst law is 0.147 V/NHE. From a thermodynamic point of view, As(V) and As(III) can be present in most aqueous media. It is generally admitted that As(V) is a predominant form in waters provided sufficient aeration.
The iron anode dissolves to Fe(II). This reducing species can oxidize to Fe(III) according to: Comparison of the equilibrium potentials of the redox couples shows that As(V) cannot be reduced by Fe(II); the latter rapidly oxidizes to Fe(III) by air oxygen in aerated media.
Nevertheless, removal of As by adsorption or electrocoagulation is more efficient for As(V) than As(III). In addition to its lower intrinsic toxicity, As(V) is thus a less hazardous form of arsenic to be used for treatment or consumption purposes. For this reason, in most investigations, pretreatment of wastewaters includes the preliminary step of As(III) oxidation using hydrogen peroxide, Fe(III), or at the anode surface, as reported by Ratna Kumar et al. [4].

Application of the electrocoagulation model to As removal
The model was directly derived from that for COD abatement. Fe(III) complexes As species as follows: Model Fe(III) + n 2 As(V) ↔ FeAs (9) where coefficient n 2 is expressed in mg Fe(III) per mg As species.
where q As is the amount of As adsorbed on Fe hydroxides, the binding constant K L can be related to product K 2 [Fe] t , and the sorbent capacity a L to the ratio (K 2 / n 2 ). As with S, the concentration of free, untreated arsenic species is obtained by solving the system formed byequations ( As relevant studies indicate, As(V) species can be totally removed by electrocoagulation; therefore, its final concentration [As] f was neglected here. The model was applied to various data reported in the literature [1,4,15,17]. Because the corresponding values for the concentration of generated iron species were rarely reported, it was assumed that iron anodes dissolved with a current efficiency equal to 0.8, as observed in previous investigations [9,10]. The postulated value is also in fair agreement with the value observed for the treatment of the industrial waste (see above section). The various data considered were approximately fitted to the model, yielding estimates for constants K 2 and n 2 . Fig. 2 shows the good correlation between theory and experimental data for As concentrations ranging from 100 µg/L to 30 mg/L within a factor 2, with K 2 = 1.12 L/mg and n 2 = 1.82 mg Fe per mg As Figure 2. Validation of the model for adsorption of As species onto Fe hydroxide (rel. 15): data issued from refs. [1], [4], [15] and [17] are related to treatment of synthetic As solutions or groundwater. The dotted lines correspond to 50% and 200% of the diagonal expressing perfect agreement between the model and experiments.
The existing scattering of the data is probably due to the lack of accuracy in the estimation of coagulant concentration. Simulation tests actually showed that Fe(III) concentration had a noticeable impact on the predicted concentration of remaining As species. Besides, the value for n 2 could not be compared directly to the molar Fe/As ratio values reported in the literature and discussed by Hansen et al. [3], because the adsorption of As species, like other pollutants, obeys an adsorption isotherm law involving equilibrium between complexed and free species. Therefore, the parameter n 2 does not correspond to the Fe/As ratio, which is deduced by global assessment of As removal depending on the amount of Fe(III) generated.

Experimental observations
Numerous batch runs have been carried out with As concentrations up to 700 µg/L. In most cases, iron dissolution in the industrial waste was shown to be unaffected by the presence of the arsenic salt (Fig. 3). The concentration in Fe species in the liquid increased regularly with the electrical charge passed, although at a slower dissolution rate in the first minutes of the run. This corresponds to the increase in current yield mentioned in a previous section with the raw industrial waste. In addition, the concentration of arsenic salt decreased with time. As for the COD and other features of the wastewater under treatment, the abatement of the toxic As-additives is governed by the amount of Fe dissolved (Fig. 4). Nevertheless, removal of As(V) species is nearly complete with 150 mg/L Fe dissolved, whereas the removal of the organic matter requires more significant dissolution (Fig. 1).  Modeling of As-removal from the industrial waste Treatment of As-containing wastewaters was modeled assuming that the generated coagulant acts independently on the organic matter and As(V) species. S adsorbs on Fe(III) hydroxide according to parameters (K 1 , n 1 ), whereas As(V) adsorbs on the solid coagulant with parameters (K 2 , n 2 ). The model was, therefore, developed on the basis of equilibrium (2) and (8) with constants K 1 and K 2 . Mass balances were written for the pollutant charge expressed by S, as well as for As(V) and Fe(III). The two first balances are given in equations (4) and (11), respectively, whereas the conservation of Fe species is written as: The model required the values for parameters K 1, K 2 , n 1 , and n 2 . The above model, assuming no interactions between the abatement of COD and As species, was applied to the set of data. Fig. 5 establishes the acceptable validity of the model for As concentrations ranging from 0.3 µg/L to 700 µg/L. Most predicted values for the concentration lie between 50 and 200% of the experimental value. Taking into account the uncertainty in the experiments, in particular in the chemical analysis of As in quite a large range of concentration and the existing uncertainty in the values for the parameters n 2 and K 2 , the approach can be considered as successful. Therefore, for the sample of industrial waste considered, the highly toxic salt added seems to be removed from the wastewater independently of the organic matter contained in the waste. In addition, the simple equations developed here are sufficient to model the waste treatment. Comparison of the experimental and predicted COD levels was not carried out since COD abatement was only slightly affected by the presence of As species, which was added in concentrations of nearly two orders below the COD level.

Conclusions
Arsenic species can be successfully removed from industrial waste by electrocoagulation using Fe electrodes. As(V) ions adsorb on Fe hydroxide in accordance with Langmuir-like adsorption isotherm. The parameters of the adsorption equilibrium could be estimated from previously published data dealing with the treatment of waters with low COD levels. This equilibrium could be incorporated into a more general model for the abatement of both As and COD, assuming no interaction in the treatment of the two pollution sources. In the example of industrial waste issued by a paper mill industry, the model was shown to hold, allowing a fairly accurate prediction of the As species. However, the approach may not be successful for industrial wastewaters containing oxidizing or reductive agents, which could change the chemical state of the arsenic salt and then affect its adsorption onto the Fe hydroxide particles. Whatever the model, a thorough characterization of the waste to be treated is required prior to designing the electrocoagulation process. For such reactive wastes, further characterization of their chemical behavior seems to be absolutely necessary.

Introduction
Mesoporous carbon (MC) is one of the most important carbon materials with high surface area, uniform pore size, and good electric conductivity. During the past ten years, many efforts have been made to develop simple and efficient methods for the preparation of MC materials and to improve their properties for various applications. Recently, one-step soft template methods for MC preparation have gradually replaced time-consuming and tedious two-step hard mesoporous silica templates, such as SBA-15, MCM-41, and MCM-48 [1,2]. The MCs and their composites are broadly applied to electrochemical capacitors [1,3], catalyst supports [4] and adsorbents [5,6], and magnetic separation [7]. Compared with mesopourous carbon materials, conducting polymers (CPs) possess pseudocapacitance, which is almost 10-100 times higher than the capacitance of electrochemical double-layer capacitors (EDLCs) [8]. Polypyrrole (PPy), one of CPs, has drawn a lot of attention for supercapacitor applications [9,10], mainly due to its oxidation-reduction properties, high conductivity in doped state, high specific capacitance, good environmental stability, and especially facile synthesis [11]. However, the shortcomings of CPs, such as low surface area and constrained power-output properties [12], markedly restrict their application to electrochemical capacitors. Hence, the most effective way is to use MC materials to improve the property of the CPs electrode. A new class of composite materials has originated from the combination of MCs and CPs [13]. Choi et al. introduced a conducting polymer layer into the pore surface of mesoporous carbon via vapor infiltration of a monomer and, by subsequent chemical oxidative polymerization, obtained mesoporous carbon-polypyrrole composites [14]. The maximum specific capacitance of these composites is 274.5 F g −1 . Pacheco-Catalán et al. synthesized mesoporous carbon/conducting polymer composites by adsorption of different monomers (aniline, pyrrole, thiophene, and 3-methyltiophene) in the gas phase onto the carbon surface, followed by oxidative chemical polymerization [15]. The electrochemical performance of carbon/polypyrrole composites electrode showed that it had a low specific capacitance (maximum value: 83.8 F g −1 ) and stable cycle life in the potential range of 0 to 1V. However, all those materials showed low specific capacitance. Recently, we prepared mesoporous carbon by onestep method [16]. The obtained results revealed that MCs display good capacitive behavior with high reversibility and reproducibility due to their unique large mesopore size, which is favorable for fast ionic transport. However, the specific capacitance of MCs reaches only 163 F g -1 , which limits their application to electric vehicles and high power electronic devices. In this study, we used PPy growth on the surface of MC via simple chemical oxidative polymerization with calcium carbonate as the sacrificial template to improve the capacitance of carbon materials. The resulting composite material combined double layer capacitance of MC with pseudocapacitance of PPy.

Chemicals
Pyrrole monomer (98%, Aldrich Chemical Co.) was distilled under reduced pressure, transferred into a refrigerator, and stored under nitrogen until further use. The diameter of nanoscale calcium carbonate was 80 nm and the BET surface area was 25 m 3 g -1 . Sodium 4-methyl benzene sulfonate (TsONa) was purchased from Sinopharm Chemical Reagent Co. Ltd, Shanghai, China. The other starting materials in this work were of analytical grade.

Synthesis of mesoporous carbons/polypyrrole composites
MCs were synthesized from F127/silica/butanol according to the procedure described previously [16]. Typically, 2 cm 3 sulfuric acid (98 wt%) and 9.3 cm 3 butanol (BuOH) were directly added into a clear solution containing 2.5 g F127 and 120 g deionized water at 318 K. After stirring for 1.5 h, 5 cm 3 tetraethyl orthosilicate (TEOS) was added, vigorously stirred at 318 K for 24 h, and aged at 373 K for 24 h. The F127/silica/butanol composites were collected by filtration and dried at room temperature for 12 h and at 433 K for 6 h. Finally, the obtained composites were transferred into a tube furnace and carbonized under pure N 2 atmosphere at 1123 K for 2 h, followed by the treatment with diluted HF solution. The chemical oxidative polymerization of pyrrole was performed using a modified procedure reported previously [17]. First, 0.3 g TsONa, 0.5 g CaCO 3 , 25 g CaCl 2 , 20 cm 3 C 2 H 5 OH, and 0.5 g Silica/Carbon were mixed into 40 cm 3 deionized water with magnetic stirring under N 2 atmosphere for 30 min; the nanoscale CaCO 3 served as a core, and TsONa was used as anionic surfactant. Then 0.3 cm 3 pyrrole was added by a syringe. After 10 min, 40 cm 3 of 0.02 mol dm -3 FeCl 3 solution was slowly added as oxidant into the reaction vessel. The polymerization was carried out for 12 h in an ice-bath with magnetic stirring and maintained under N 2 atmosphere. The mesoporous carbon/polypyrrole composite was filtered and rinsed several times with distilled water and ethanol to remove retained pyrrole monomer and oxidant. The as-synthesized powder was transferred into HCl solution for CaCO 3 removal for 24 h. The mixture was filtered and rinsed several times again with distilled water. Then the silica/carbon/PPy composite was treated by diluted HF solution to remove the silica and dried in vacuum at 333 K for 12 h. The product was denoted as MP, containing an approximate MC:PPy weight ratio of 6:4. Under the influence of the Lewis acid FeCl 3 and inhibition by abundant CaCl 2 , CaCO 3 was dissolved, slowly releasing CO 2 , which contributed to the holes formation in the composite during the overflow process. Actually, CaCO 3 as a core was the sacrificial template.

Characterization
The morphologies of MC and as-synthesized composite were examined using a high-resolution transmission electron microscopy (TEM, JEOL JEM-2100, 200 kV). Nitrogen sorption isotherms of samples were measured by a Micromertics TriStar 3000 analyzer at 77 K. The FT-IR measurements on samples were performed using Nicolet 6700 FT-IR spectrometer. The electrochemical behavior of both MC and MP was investigated by cyclic voltammetry (CV) conducted on a CHI 660 B electrochemical workstation (ChenHua Instruments Co., Shanghai, China). The measurements were carried out in a standard three-electrode cell system. A Pt-foil modified by either MC or MP was used as the working electrode, saturated calomel electrode was used as the reference electrode, and a Pt-foil as the auxiliary electrode. The working electrodes were prepared by mixing active materials (8 mg), acetylene black as a conductive reagent, and 5%-Nafion as a binder (80:10:10 wt %) and dispersed in absolute ethanol in ultrasonic bath. The dispersion was coated onto Pt foil drop by drop and then dried at 353 K. The CV experiments were carried out in aqueous solution of 1 mol dm -3 H 2 SO 4 . Fig. 1a and 1b show the N 2 adsorption-desorption isotherms and BJH pore size distribution of MC and MP. The two samples are found to yield a type IV isotherm with a type H2 hysteresis loop near relative pressure of 0.50 in the desorption branch, which is associated with sharp capillary condensation taking place in mesopores [18]. The results show that the mesoporous structure of MC was maintained after loading of PPy. However, BET surface area of MP was considerably lower. The specific surface areas of mesoporous carbon and MP were 1041 and 207 m 2 g -1 , respectively ( Table 1). The decrease of specific surface area is mainly attributed to the mesopores of the support being partially covered or filled by PPy. Therefore, pore volume and average pore size also decrease from 1.18 cm 3 g -1 and 5.2 nm to 0.22 cm 3 g -1 and 4.9 nm, respectively. This indicates that PPy was distributed evenly over the surface of MC.

Results and Discussion
In order to gain further insights into the structure of mesporous carbons, we investigated their appearance on the synthesized materials by TEM (Fig. 2). The disordered mesoporous size of MC with around 5 nm, which is clearly visible in Fig. 2a, is in agreement with the data of the adsorption-desorption measurements (Fig. 1). Fig. 2b clearly shows the mesoporous structure of MC covered over with transparent films. In comparison with Fig. 2a, we may draw a conclusion that the transparent film is PPy and that it has been successfully loaded on the MC. The functional groups of MC, PPy, and MP are characterized by FTIR spectroscopy (Fig. 3). The characteristic peak at 3423 cm -1 is observed in the IR spectrum of MC and ascribed to -OH. The small peaks at 2920 and 2850 cm -1 originate from the stretching vibrations of C-H bond. The peaks at 1731, 1632, and 1402 cm -1 could be assigned to the stretching vibration in carboxyl groups and C=O, C=C, and C-O bonds. The bands at 1537 and 1452 cm -1 in PPy and MP spectra are due to the typical pyrrole ring vibration of pure PPy and bands of =C-H in plane vibration at 1298, 1089, and 1032 cm -1 [19]. The peaks at 3434 and 1631 cm -1 correspond to N-H and C=C stretching vibrations, respectively. The spectrum of MP is very similar to that of PPy, verifying that PPy has been successfully applied onto MC. However, the peak at 1632 cm -1 for C=O disappeared due to the combination of MC and PPy. Similarly, the peak for N-H stretching exhibits a red-shift phenomenon due to the interaction with the reactive hydroxyl functional groups [20]. Fig. 4 shows CVs of MC and MP in 1 mol dm -3 H 2 SO 4 at different scan rates. CVs of MC electrode contain redox peaks and deviate from the rectangular shape (Fig. 4a). With increasing scan rates, the redox current evidently increases, indicating that it has good rate capability. FTIR spectra confirm the existence of the abundant functional groups, for example C=O and O-H bonds, on the surface of MC. These groups are supposed to remarkably improve the hydrophilicity and wettability of the surface of MC and are beneficial for the aqueous electrochemical capacitors. The redox peaks might originate from the transformation of the O-H bond of MC into the C=O and vice versa during charge and discharge processes. In addition, the C=O bond of MC interacts with the H + to become -OH in the reduced state, which would be re-oxidized into C=O during the discharge. However, the CVs of MP present a steep increase in the current range from 0.0 to 0.1 V, which is an important behavior in supercapacitors [14], as shown in Fig. 4b. CV curves of the composites do not exhibit redox peaks as pronouncedly as in the case of MC. This is because the π-bonded surface of the MC may interact strongly with the conjugated structure of PPy, especially through the pyrrole ring [21], or N-H of PPy can interact with the reactive hydroxyl functional groups of MC [20]. The gravimetric specific capacitance (C) of electrode is calculated according to the Eq. (1) from charge-discharge data of CVs: where i, W and V are the voltammetric current, the mass of active materials, and the total potential of electrochemical window, respectively. The calculated specific capacitances are listed in Table 1. The maximum specific capacitance of MP reaches as high as 313 F g -1 at the scan rate of 2 mV s -1 , which is nearly twice the specific capacitance of MC (163 F g -1 ). Although the surface area of MC dramatically decreases due to the coverage of PPy, leading to the EDLC loss, the pseudocapacitance from the Faradic reaction on the PPy contributes much more to the total capacitance, leading to a higher specific capacitance of MP. Due to the different synthesis methods and electrolytes and measured potential windows, it is difficult to compare our results with those reported in the literatures. Therefore, for comparison purpose, we synthesized the PPy using the same method and measured the electrochemical properties in 1 mol dm -3 H 2 SO 4 . The specific capacitance of pure PPy was 328.4 F g -1 at a scan rate of 2mV s -1 . If the contribution of pure MC and pure PPy were calculated, the total was only 229 F g -1 . However, in this study, we obtained a higher result that reached 313 F g −1 , revealing the intensely synergetic effect between MC and PPy. Fig. 4c shows a cycle life of MC and MP in 1 mol dm -3 H 2 SO 4 electrolyte at scan rate of 20 mV s -1 . The specific capacitance of MP drops from 238 F g -1 down to 181 F g -1 after 1000 cycles, corresponding to a 23.9% loss. Conclusions MC/PPy composites were successfully synthesized using a chemical method of oxidative polymerization. The resulting modified electrode showed some properties similar to the combination of the double layer capacitance of MC and pseudocapacitance of PPy. The results obtained by cyclic voltammetry demonstrated that the specific capacitance of the MP electrode was as high as 313 F g −1 compared to 163 F g −1 of MC electrode. In conclusion, it seems that the MP is a promising electrode material in supercapacitor field. Figure SM-1 shows photographs of the transition from active to passive dissolution. During active dissolution the assumed reaction products creep down as a green viscose film ( Figure SM-1a). After passivation, the evolution of oxygen breaks off the visible flakes of the black carbide-containing surface layer from the anode surface ( Figure SI-1b) and transpassive dissolution can be observed by the formation of oxygen bubbles and yellow-brown colouration of the solution ( Figure SM-1c).

SM-2 Programmed Current Chronopotentiometry
Active dissolution in pure NaCl electrolyte and transpassive dissolution in pure NaNO 3 electrolyte are indicated by curves 1 and 7. The PCCs in mixed electrolytes show active dissolution at low current densities. The transition from active to transpassive is indicated by the increase in potential. The current density at transition depends on the relative chloride anion concentration.