CHEMPHYSCHEM COMMUNICATIONS DOI: 10.1002/cphc.201402666
Revisiting the Electrochemical Impedance Spectroscopy of Magnesium with Online Inductively Coupled Plasma Atomic Emission Spectroscopy Viacheslav Shkirskiy,[b] Andrew D. King,[c] Oumama Gharbi,[b] Polina Volovitch,[b] John R. Scully,[c] Kevin Ogle,[b] and Nick Birbilis*[a] The electrochemical impedance of reactive metals such as magnesium is often complicated by an obvious inductive loop with decreasing frequency of the AC polarising signal. The characterisation and ensuing explanation of this phenomenon has been lacking in the literature to date, being either ignored or speculated. Herein, we couple electrochemical impedance spectroscopy (EIS) with online atomic emission spectroelectrochemistry (AESEC) to simultaneously measure Mg-ion concentration and electrochemical impedance spectra during Mg corrosion, in real time. It is revealed that Mg dissolution occurs via Mg2 + , and that corrosion is activated, as measured by AC frequencies less than approximately 1 Hz approaching DC conditions. The result of this is a higher rate of Mg2 + dissolution, as the voltage excitation becomes slow enough to enable all Mg2 + -enabling processes to adjust in real time. The manifestation of this in EIS data is an inductive loop. The rationalisation of such EIS behaviour, as it relates to Mg, is revealed for the first time by using concurrent AESEC.
The use of electrochemical impedance spectroscopy (EIS) to assess electrochemical corrosion behaviour of magnesium and its alloys is of great technical relevance. Given the growing use of Mg alloys as structural alloys, an accurate assessment of corrosion is imperative and the unambiguous determination of electrode kinetics in the case of Mg electrodes is also essential. To date, EIS has been used in numerous studies to assess the corrosion rate of Mg or Mg alloys.[1–6] The approach commonly taken is to determine the charge transfer resistance (Rt), defined as the value of impedance (Z’) when Z’’ = 0 at intermediate frequencies.[6–8] The frequency at which Z’’ = 0 is variable, and does not coincide with the low frequency impedance limit approaching DC conditions, which defines the polarisation resistance, RP (i.e. Z’ as frequency!0). A summary of the conventional impedance analysis of Mg was provided recently by King and co-workers.[1] However, to provide some critical [a] Prof. N. Birbilis Department of Materials Engineering, Monash University Wellinton Road, Clayton, VIC. 3800 (Australia) E-mail: [email protected] [b] V. Shkirskiy, O. Gharbi, Dr. P. Volovitch, Prof. K. Ogle Chimie ParisTech, Ecole Nationale Suprieure de Chimie de Paris 11 Rue Pierre et Marie Curie, Paris 75005 (France) [c] Dr. A. D. King, Prof. J. R. Scully Department of Materials Science and Engineering 395 McCormick Road, University of Virginia, Charlottesville, VA. 22904 (USA)
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
context, recent works have avoided the analysis of an inductive loop in EIS data for Mg, suggesting that the EIS plot is rather complicated and that inductance is immaterial to electrochemical corrosion,[6] or that the inductive loop seems complicated and is of no practical relevance;[2] in such works Rt was used to determine the corrosion rate, rather than Rp. An exception to the works the inductive loop has not been analysed, is a study from Baril and co-workers,[5] in which inductance is ascribed to the existence of unipositive Mg + as a reaction intermediate on corroding Mg surfaces. However, there was no evidence of Mg + in the study, and indeed the existence of Mg + in solution away from the Mg surface has since been summarily discredited;[9–15] albeit, some works continue to suggest the existence of lower valence Mg on the basis of incomplete analysis of EIS data with no physical evidence.[8] Criticism have been made of the electrochemical methods applied to Mg, relating to their ability to predict mass loss;[4] such criticism is based on the notion that chemical reactions cannot be monitored by electrochemical methods.[7] This has, however, been addressed by online inductively coupled plasma (ICP) methods, as first demonstrated by Ogle and coworkers[16–17] and since further developed [including inductively coupled plasma optical emission spectrometry (ICP–OES) and inductively coupled plasma mass spectrometry (ICP– MS)].[12, 13, 18–21] Hence, given the urgent need for the unambiguous determination of corrosion rates for Mg, and the state of the current literature regarding Mg EIS, a definitive experiment coupling EIS with real time online atomic emission spectroelectrochemistry (AESEC) analysis is timely. The open-circuit potential of the pure Mg used herein was approximately1.67 [vs. a saturated calomel electrode (SCE)] in 1.0 m NaCl (Figure 1). At open circuit, a free corrosion rate is able to be measured using AESEC, which is a major advantage of the online technique; these results indicate that the free corrosion rates are in the range of several hundreds of mA cm2. The variation in the electrode potential of Mg during the application of EIS is shown in Figure 1, along with the corresponding iMg2þ measured from AESEC. It is implied that AESEC is only capable of determining the soluble-Mg2 + concentration (i.e. Mg surface films or insoluble products are unable to be detected). During the initial application of EIS (to 1 Hz), there is no systematic variation in the iMg2þ signal measured that is substantially different. However, at frequencies lower than 1 Hz, a causal variation in the iMg2þ signal measured, which responds to the excitation voltage signal, is observed (Figure 1). In esChemPhysChem 0000, 00, 1 – 4
&1&
These are not the final page numbers! ÞÞ
CHEMPHYSCHEM COMMUNICATIONS
Figure 1. The results from online AESEC determination of Mg2 + dissolution (iMg2þ ) during the simultaneous measurement of EIS for pure Mg in 1.0 m NaCl. The potential (VSCE) and dissolution current (iMg2þ ) is seen prior to, during and after the application of an AC signal of varying frequency (100 kHz–0.005 Hz).
sence, the effect of the polarising voltage signal on Mg dissolution can be readily measured on the basis of the iMg2þ signal (which quantifies the amount of Mg2 + dissolved). Additionally, at the termination of the EIS test when the electrode returns to rest at open-circuit potential, the free corrosion rate realised ( 2 mA cm2) is significantly higher than that prior to the application of EIS. Over the time period of the EIS test, the Mg has been activated, with an attendant higher increase in the open-circuit potential. This activation per se, has been recently described as a result of DC polarisation or even free corrosion,[22] however its observation or relationship with the EIS polarising signal and the resultant iMg2þ signal has not been previously reported. Nevertheless, in the absence of AESEC, King and co-workers associated the so-called activation as a contributor to the manifestation of an inductance in the electrochemical response.[1] The inductance associated with the results in Figure 1 cannot be ascertained from Figure 1 itself, but can be electrochemically observed in Figure 2, which reveals the Bode magnitude and Bode phase plot for pure Mg in 1.0 m NaCl. The Bode phase data indicates a positive phase angle for frequencies below 1 Hz, which is commensurate with inductive behaviour. In addition, the j Z j value from the Bode magnitude plot decreases with decreasing frequency, concomitant with a decrease in the polarisation resistance (and hence indicative of increased dissolution rate). These observations are somewhat analogous to an electrode leaking Mg2 + , which would reconcile with an activation and enhancement of iMg2þ measured by using AESEC; however, such an activation being a result of the EIS itself cannot be ascertained. The activation of Mg is known to be a temporal and DC perturbation phenomenon,[10-11] and may occur independent of the superimposed AC signal. A unified manner in which to observe the electrochemical response as it relates to the iMg2þ signal from Figure 1, is to 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
Figure 2. Bode magnitude and phase plot (100 kHz to 0.005 Hz) for pure Mg in 1.0 m NaCl.
Figure 3. Nyquist plot accompanying Bode plot (Figure 2) of pure Mg in 1.0 m NaCl. Symbols represent experimental EIS data, and the line (in red) represents the fit to the equivalent circuit of King and co-workers[1] given in the inset. The diamonds (in grey) represent the Z’ values that would be required to corroborate the EIS and AESEC data at a given frequency, indicating how important the inductive loop is.
present the associated (annotated) Nyquist plot as in Figure 3. Such a form of Nyquist plot, in which data resides in the socalled third quadrant (i.e. positive Z’ and positive Z’’), is what is often called the inductive loop and is a characteristic of Mg that is observed in essentially all reported Nyquist plots for Mg and its alloys in NaCl electrolytes.[1–7] The Nyquist plot in Figure 3 includes the measured data, along with a fit employing the equivalent circuit proposed and validated by King and co-workers,[1] which includes a nested Randle’s element and inductor. This fit has been included on the basis of validating the circuit proposed by King and coworkers, and the requisite inductor; however, the fit is inconsequential in the presence of AESEC data, the latter providing a definitive iMg2þ to compare the electrochemical response and actual dissolution. ChemPhysChem 0000, 00, 1 – 4
&2&
These are not the final page numbers! ÞÞ
CHEMPHYSCHEM COMMUNICATIONS Overlaid on Figure 3 are three real impedance data points that were calculated directly from the iMg2þ measured when an AC signal is at frequencies of 1, 0.1 and 0.005 Hz. The Z’ value was determined from average DEAC DiMg2þ obtained from AESEC data corresponding to the E/i data for the cycle at the frequency of interest. The data points (shown as diamonds) reveal the Z’ value that is required to allow the EIS data to correctly determine the AESEC measured dissolution rate. There are several salient points on the plot, each of which merits comment. The EIS data corresponding to frequencies lower than approximately 1 Hz all resides in the third quadrant of the Nyquist plot. What is observed from Figure 3, is that there is good alignment between the EIS-determined Z’ value (shown as squares) and the Z’ value from the AESEC expectation of Z’ (shown as diamonds). The alignment is assessed by looking at the vertical alignment of data, as there is no imaginary component associated with AESEC-derived impedance data (the raw AESEC data can be observed in Figure 1). Put simply, by using the low frequency impedance limit and employing RP (which mandates analysis of the inductive loop for a reactive system), the determination of the appropriate iMg2þ value at the low frequency impedance limit is possible. It is noted that the DC value of iMg2þ (as frequency!0) would agree with icorr obtained from EIS data by using the Stern– Geary relationship in which current is inversely proportional to resistance/impedance when EIS data near 0.005 Hz is used[1] (and the utility of a value of B 40 mV in the Stern–Geary Equation, where icorr = B/RP). This value of B does not necessitate assumption, as it can be determined as the required value based on the AESEC data, which happens to, however, match the values quoted by King and co-workers,[1] from DC measurement of Tafel slopes. AESEC has revealed several key findings and indicates that in cases when the inductance (which is a characteristic of the system) is ignored, then the resultant analysis will underpredict iMg2þ , giving rise to incorrect assumptions of lower valence or that electrochemical methods are somehow deficient, which was shown to not be the case. The alignment between the EIS and AESEC data is sound, however if using the Rt value as the parameter for which to calculate corrosion rates by using the Stern–Geary relationship, this would result in a gross overestimation of the Z’ value required to predict the low frequency impedance value commensurate with the AESEC-determined iMg2þ value (at the completion of the EIS test). In a similar vein, it was also previously indicated that the means for which dissolving Fe in acid was best analysed was by utilisation of RP (and not Rt) when corroborating EIS with atomic absorption.[23] A more detailed analysis of the electrochemical results allows the electrochemical current (which is the response that is measured by the potentiostat, ipstat) during the application of controlled potential EIS to be plotted against the convoluted total current from the potentiostat with the AESEC-determined iMg2þ value(Figure 4), as per the procedure in Ref. [16]. This analysis allows the current measured by the potentiostat to be compared directly to that from the AESEC-determined current (from dissolved ions and assuming z = 2).
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
Figure 4. The AC current recorded by the potentiostat during the application of potentiostatic EIS was recorded (i.e. where potential was the controlled parameter and current is the resultant measured parameter) and plotted here in concert with the deconvoluted AESEC determined iMg2þ . The baseline DC current is omitted. Agreement is seen between the two currents indicating a dissolution stoichiometry of Mg2 + , whilst the notion of cathodic activation is apparent.
Such an analysis yields two important outcomes from Figure 4. These are: 1) there is high agreement between the iMg2þ and ipstat values, indicating that the dissolution of Mg is via Mg2 + , and 2) the cathodic current measured in each downcycle following anodic polarisation shows a higher cathodic current. This latter point is in direct agreement with recent work that has ascribed the so-called negative difference effect (NDE) upon Mg as a consequence of enhanced cathodic (catalytic) activity in response to anodic dissolution.[11] As such, when AC is cycled between the same potentials, a higher cathodic current can be sustained by the Mg specimen as cycling proceeds, in accordance with previous reports from DC tests.[22] However, what is highlighted here, is that we do not have the evidence to suggest that the EIS itself is causing the activation (i.e. as an AC effect, which would be presumably evident during the + half wave of the AC cycle), rather, that EIS is capable of measuring this effect when coupled with AESEC. In summary, this work indicates that the inductance should not be ignored in EIS measurements of Mg, and RP is favoured for the measurement of corrosion rate over Rct. Electrolytes that give rise to this inductive response have recently been summarised by Bland and co-workers.[24] Justification for the use of an inductor in the assessment of EIS data for Mg was provided herein, and the inductive response was correlated to the enhancement of Mg dissolution with time, manifest as test frequency!0. This has two major implications; the first, practical implication, is that an accurate assessment of Mg dissolution in the absence of AESEC requires the utility of an inductor. The second, more general implication, is that an obvious advantage of the EIS–AESEC technique is the ability to treat dynamic systems directly, by analysing data in the time domain that takes into account the system evolution. This is an important notion, as the Mg system, like many active-metal systems, is not stable on the basis of the anticipated causality. It was ChemPhysChem 0000, 00, 1 – 4
&3&
These are not the final page numbers! ÞÞ
CHEMPHYSCHEM COMMUNICATIONS shown by King and co-workers[1] that Mg EIS data does still satisfy the Kramars–Kronig transform, and thus, the dynamic nature of the system and the use of EIS is a matter for each researcher to contend with, as a complex response is common to the majority of corroding systems. Importantly however, it has been shown herein that proper assessment is capable of yielding meaningful results. Finally, there is no need to invoke any notion of lower valence Mg to describe results herein, which are wholly consistent with Mg2 + -dissolution stoichiometry. The aspects clarified herein regarding the instability of Mg, the activation of Mg, the enhanced catalytic activity of Mg and the dissolution stoichiometry of Mg, were all revealed by using online AESEC and EIS methods and all are relevant in the assessment of Mg and Mg alloys as electrode materials or in determination of corrosion rates.
Experimental Section The AESEC method used herein has previously been described in detail, including a schematic of the electrochemical flow cell and a description of the data acquisition and the calibration procedure.[9, 25] The system consists of a Horiba Jobin Yvon inductively coupled plasma optical emission spectrometer coupled with a Gamry Reference 600 potentiostat and an electrochemical flow cell. In Brief, the working electrode (Mg) is exposed to a flowing electrolyte (at 3 mL min1) in a small volume (0.2 mL) flow cell. The geometrical area of the sample exposed to the electrolyte was 0.5 cm2 as defined by the contour of an O ring. The flow channel has an entrance at the bottom and exit at the top so that electrolyte passes through the cell from bottom to top, such that any gas generated during the experiment flows out of the cell. The electrolyte from the electrochemical flow cell is continuously fed into the plasma by using a peristaltic pump. The electrolyte is atomised by the plasma and the elemental components are monitored in real time by following the emission at specific wavelengths by using an array of photomultipliers in a polychromator. In this work, the emission intensity of Mg (279.553 nm) was recorded simultaneously as a function of time, along with the electrochemical potential and current. Calibration of intensity versus concentration was performed by using standard ICP-OES methods and normalised against calibration standards purchased from SPC Science, France. A Pt counter electrode and saturated calomel reference electrode were used. In the electrochemical flow cell they are positioned in a 10 mL cylindrical secondary compartment of stagnant electrolyte. This compartment is separated from the working electrode compartment by a porous membrane , which allows ionic current to pass from one compartment to the other while preventing bulk mixing of the two electrolytes. All experiments were performed at ambient temperature, approximately 24 1 8C. Potentiostatic EIS was conducted over a frequency range of 100 kHz to 0.005 Hz with an AC amplitude of 20 mV rms. The experiments were performed using commercially pure (termed pure herein) 99.9 % Mg (Good Fellow) with a nominal composition of Mg, Al 70 ppm, Cu 20 ppm, Fe 280 ppm, Mn 170 ppm, Ni < 10 ppm, Si 50 ppm, Zn < 20 ppm. The sample was ground to
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org 4000 grit with SiC disks with ethanol. The 1.0 m NaCl solution was prepared from analytical grade materials and deionised water of 18.2 MW cm1 (prepared using a Millipore system).
Acknowledgements The support of Ecole Nationale Suprieure de Chimie de Paris, Centre National de la Recherche Scientifique (ENSCP-CNRS) is gratefully acknowledged. Keywords: corrosion · electrochemistry · ICP–AES · impedance spectroscopy · magnesium [1] A. D. King, N. Birbilis, J. R. Scully, Electrochim. Acta 2014, 121, 394 – 406. [2] S. Feliu, C. Maffiotte, A. Samaniego, J. C. Galvan, V. Barranco, Appl. Surf. Sci. 2011, 257, 8558 – 8568. [3] Z. X. Qiao, Z. M. Shi, N. Hort, N. I. Z. Abidin, A. Atrens, Corros. Sci. 2012, 61, 185 – 207. [4] F. Y. Cao, Z. M. Shi, J. Hofstetter, P. J. Uggowitzer, G. L. Song, M. Liu, A. Atrens, Corros. Sci. 2013, 75, 78 – 99. [5] G. Baril, G. Galicia, C. Deslouis, N. Pebere, B. Tribollet, V. Vivier, J. Electrochem. Soc. 2007, 154, C108 – C113. [6] A. Pardo, M. C. Merino, A. E. Coy, F. Viejo, R. Arrabal, S. Feliu, Electrochim. Acta 2008, 53, 7890 – 7902. [7] A. Pardo, S. Feliu, M. C. Merino, R. Arrabal, E. Matykina, Int. J. Corros. 2010, DOI:10.1155/2010/953850. [8] Z. Shi, F. Cao, G. L. Song, A. Atrens, Corros. Sci. 2014, DOI: 10.1016/ j.corsci.2014.07.060. [9] J. S´wiatowska, P. Volovitch, K. Ogle, Corros. Sci. 2010, 52, 2372 – 2378. [10] G. Williams, N. Birbilis, H. N. McMurray, Electrochem. Commun. 2013, 36, 1 – 5. [11] G. S. Frankel, A. Samaniego, N. Birbilis, Corros. Sci. 2013, 70, 104 – 111. [12] S. Lebouil, A. Duboin, F. Monti, P. Tabeling, P. Volovitch, K. Ogle, Electrochim. Acta 2014, 124, 176 – 182. [13] L. Rossrucker, K. J. J. Mayrhofer, G. S. Frankel, N. Birbilis, J. Electrochem. Soc. 2014, 161, C115 – C119. [14] M. Curioni, Electrochim. Acta 2014, 120, 284 – 292. [15] A. Samaniego, B. L. Hurley, G. S. Frankel, J. Electroanal. Chem. 2014, DOI: 10.1016/j.jelechem.2014.04.013. [16] K. Ogle, S. Weber, J. Electrochem. Soc. 2000, 147, 1770 – 1780. [17] K. Ogle, A. Tomandl, N. Meddahi, M. Wolpers, Corros. Sci. 2004, 46, 979 – 995. [18] S. O. Klemm, A. Karschina, A. K. Schuppert, A. A. Topalov, A. M. Mingers, I. Katsounaros, K. J. J. Mayrhofer, J. Electroanal. Chem. 2012, 677, 50 – 55. [19] P. Volovitch, M. Serdechnova, K. Ogle, Corrosion 2012, 68, 557 – 570. [20] A. Ulrich, N. Ott, A. Tournier-Fillon, N. Homazava, P. Schmutz, Spectrochim. Acta Part B 2011, 66, 536 – 545. [21] M. Mokaddem, P. Volovitch, F. Rechou, R. Oltra, K. Ogle, Electrochim. Acta 2010, 55, 3779 – 3786. [22] N. Birbilis, A. D. King, S. Thomas, G. S. Frankel, J. R. Scully, Electrochim. Acta 2014, 132, 277 – 283. [23] W. J. Lorenz, F. Mansfeld, Corros. Sci. 1981, 21, 647 – 672. [24] L. Bland, A. D. King, N. Birbilis, J. R. Scully, Corrosion 2014, DOI: 10.5006/ 1419. [25] K. Ogle, J. Baeyens, J. Swiatowska, P. Volovitch, Electrochim. Acta 2009, 54, 5163 – 5170. Received: September 25, 2014 Published online on && &&, 2014
ChemPhysChem 0000, 00, 1 – 4
&4&
These are not the final page numbers! ÞÞ
COMMUNICATIONS In the loop: Reactive metals such as magnesium reveal significant inductance during electrochemical impedance spectroscopy (EIS) measurements. Combined atomic emission spectroelectrochemistry and EIS provided a definitive assessment for the electrochemical analysis of Mg dissolution.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
V. Shkirskiy, A. D. King, O. Gharbi, P. Volovitch, J. R. Scully, K. Ogle, N. Birbilis* && – && Revisiting the Electrochemical Impedance Spectroscopy of Magnesium with Online Inductively Coupled Plasma Atomic Emission Spectroscopy
ChemPhysChem 0000, 00, 1 – 4
&5&
These are not the final page numbers! ÞÞ
Revisiting the Electrochemical Impedance Spectroscopy of Magnesium with Online Inductively Coupled Plasma Atomic Emission Spectroscopy Viacheslav Shkirskiy,[b] Andrew D. King,[c] Oumama Gharbi,[b] Polina Volovitch,[b] John R. Scully,[c] Kevin Ogle,[b] and Nick Birbilis*[a] The electrochemical impedance of reactive metals such as magnesium is often complicated by an obvious inductive loop with decreasing frequency of the AC polarising signal. The characterisation and ensuing explanation of this phenomenon has been lacking in the literature to date, being either ignored or speculated. Herein, we couple electrochemical impedance spectroscopy (EIS) with online atomic emission spectroelectrochemistry (AESEC) to simultaneously measure Mg-ion concentration and electrochemical impedance spectra during Mg corrosion, in real time. It is revealed that Mg dissolution occurs via Mg2 + , and that corrosion is activated, as measured by AC frequencies less than approximately 1 Hz approaching DC conditions. The result of this is a higher rate of Mg2 + dissolution, as the voltage excitation becomes slow enough to enable all Mg2 + -enabling processes to adjust in real time. The manifestation of this in EIS data is an inductive loop. The rationalisation of such EIS behaviour, as it relates to Mg, is revealed for the first time by using concurrent AESEC.
The use of electrochemical impedance spectroscopy (EIS) to assess electrochemical corrosion behaviour of magnesium and its alloys is of great technical relevance. Given the growing use of Mg alloys as structural alloys, an accurate assessment of corrosion is imperative and the unambiguous determination of electrode kinetics in the case of Mg electrodes is also essential. To date, EIS has been used in numerous studies to assess the corrosion rate of Mg or Mg alloys.[1–6] The approach commonly taken is to determine the charge transfer resistance (Rt), defined as the value of impedance (Z’) when Z’’ = 0 at intermediate frequencies.[6–8] The frequency at which Z’’ = 0 is variable, and does not coincide with the low frequency impedance limit approaching DC conditions, which defines the polarisation resistance, RP (i.e. Z’ as frequency!0). A summary of the conventional impedance analysis of Mg was provided recently by King and co-workers.[1] However, to provide some critical [a] Prof. N. Birbilis Department of Materials Engineering, Monash University Wellinton Road, Clayton, VIC. 3800 (Australia) E-mail: [email protected] [b] V. Shkirskiy, O. Gharbi, Dr. P. Volovitch, Prof. K. Ogle Chimie ParisTech, Ecole Nationale Suprieure de Chimie de Paris 11 Rue Pierre et Marie Curie, Paris 75005 (France) [c] Dr. A. D. King, Prof. J. R. Scully Department of Materials Science and Engineering 395 McCormick Road, University of Virginia, Charlottesville, VA. 22904 (USA)
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
context, recent works have avoided the analysis of an inductive loop in EIS data for Mg, suggesting that the EIS plot is rather complicated and that inductance is immaterial to electrochemical corrosion,[6] or that the inductive loop seems complicated and is of no practical relevance;[2] in such works Rt was used to determine the corrosion rate, rather than Rp. An exception to the works the inductive loop has not been analysed, is a study from Baril and co-workers,[5] in which inductance is ascribed to the existence of unipositive Mg + as a reaction intermediate on corroding Mg surfaces. However, there was no evidence of Mg + in the study, and indeed the existence of Mg + in solution away from the Mg surface has since been summarily discredited;[9–15] albeit, some works continue to suggest the existence of lower valence Mg on the basis of incomplete analysis of EIS data with no physical evidence.[8] Criticism have been made of the electrochemical methods applied to Mg, relating to their ability to predict mass loss;[4] such criticism is based on the notion that chemical reactions cannot be monitored by electrochemical methods.[7] This has, however, been addressed by online inductively coupled plasma (ICP) methods, as first demonstrated by Ogle and coworkers[16–17] and since further developed [including inductively coupled plasma optical emission spectrometry (ICP–OES) and inductively coupled plasma mass spectrometry (ICP– MS)].[12, 13, 18–21] Hence, given the urgent need for the unambiguous determination of corrosion rates for Mg, and the state of the current literature regarding Mg EIS, a definitive experiment coupling EIS with real time online atomic emission spectroelectrochemistry (AESEC) analysis is timely. The open-circuit potential of the pure Mg used herein was approximately1.67 [vs. a saturated calomel electrode (SCE)] in 1.0 m NaCl (Figure 1). At open circuit, a free corrosion rate is able to be measured using AESEC, which is a major advantage of the online technique; these results indicate that the free corrosion rates are in the range of several hundreds of mA cm2. The variation in the electrode potential of Mg during the application of EIS is shown in Figure 1, along with the corresponding iMg2þ measured from AESEC. It is implied that AESEC is only capable of determining the soluble-Mg2 + concentration (i.e. Mg surface films or insoluble products are unable to be detected). During the initial application of EIS (to 1 Hz), there is no systematic variation in the iMg2þ signal measured that is substantially different. However, at frequencies lower than 1 Hz, a causal variation in the iMg2þ signal measured, which responds to the excitation voltage signal, is observed (Figure 1). In esChemPhysChem 0000, 00, 1 – 4
&1&
These are not the final page numbers! ÞÞ
CHEMPHYSCHEM COMMUNICATIONS
Figure 1. The results from online AESEC determination of Mg2 + dissolution (iMg2þ ) during the simultaneous measurement of EIS for pure Mg in 1.0 m NaCl. The potential (VSCE) and dissolution current (iMg2þ ) is seen prior to, during and after the application of an AC signal of varying frequency (100 kHz–0.005 Hz).
sence, the effect of the polarising voltage signal on Mg dissolution can be readily measured on the basis of the iMg2þ signal (which quantifies the amount of Mg2 + dissolved). Additionally, at the termination of the EIS test when the electrode returns to rest at open-circuit potential, the free corrosion rate realised ( 2 mA cm2) is significantly higher than that prior to the application of EIS. Over the time period of the EIS test, the Mg has been activated, with an attendant higher increase in the open-circuit potential. This activation per se, has been recently described as a result of DC polarisation or even free corrosion,[22] however its observation or relationship with the EIS polarising signal and the resultant iMg2þ signal has not been previously reported. Nevertheless, in the absence of AESEC, King and co-workers associated the so-called activation as a contributor to the manifestation of an inductance in the electrochemical response.[1] The inductance associated with the results in Figure 1 cannot be ascertained from Figure 1 itself, but can be electrochemically observed in Figure 2, which reveals the Bode magnitude and Bode phase plot for pure Mg in 1.0 m NaCl. The Bode phase data indicates a positive phase angle for frequencies below 1 Hz, which is commensurate with inductive behaviour. In addition, the j Z j value from the Bode magnitude plot decreases with decreasing frequency, concomitant with a decrease in the polarisation resistance (and hence indicative of increased dissolution rate). These observations are somewhat analogous to an electrode leaking Mg2 + , which would reconcile with an activation and enhancement of iMg2þ measured by using AESEC; however, such an activation being a result of the EIS itself cannot be ascertained. The activation of Mg is known to be a temporal and DC perturbation phenomenon,[10-11] and may occur independent of the superimposed AC signal. A unified manner in which to observe the electrochemical response as it relates to the iMg2þ signal from Figure 1, is to 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
Figure 2. Bode magnitude and phase plot (100 kHz to 0.005 Hz) for pure Mg in 1.0 m NaCl.
Figure 3. Nyquist plot accompanying Bode plot (Figure 2) of pure Mg in 1.0 m NaCl. Symbols represent experimental EIS data, and the line (in red) represents the fit to the equivalent circuit of King and co-workers[1] given in the inset. The diamonds (in grey) represent the Z’ values that would be required to corroborate the EIS and AESEC data at a given frequency, indicating how important the inductive loop is.
present the associated (annotated) Nyquist plot as in Figure 3. Such a form of Nyquist plot, in which data resides in the socalled third quadrant (i.e. positive Z’ and positive Z’’), is what is often called the inductive loop and is a characteristic of Mg that is observed in essentially all reported Nyquist plots for Mg and its alloys in NaCl electrolytes.[1–7] The Nyquist plot in Figure 3 includes the measured data, along with a fit employing the equivalent circuit proposed and validated by King and co-workers,[1] which includes a nested Randle’s element and inductor. This fit has been included on the basis of validating the circuit proposed by King and coworkers, and the requisite inductor; however, the fit is inconsequential in the presence of AESEC data, the latter providing a definitive iMg2þ to compare the electrochemical response and actual dissolution. ChemPhysChem 0000, 00, 1 – 4
&2&
These are not the final page numbers! ÞÞ
CHEMPHYSCHEM COMMUNICATIONS Overlaid on Figure 3 are three real impedance data points that were calculated directly from the iMg2þ measured when an AC signal is at frequencies of 1, 0.1 and 0.005 Hz. The Z’ value was determined from average DEAC DiMg2þ obtained from AESEC data corresponding to the E/i data for the cycle at the frequency of interest. The data points (shown as diamonds) reveal the Z’ value that is required to allow the EIS data to correctly determine the AESEC measured dissolution rate. There are several salient points on the plot, each of which merits comment. The EIS data corresponding to frequencies lower than approximately 1 Hz all resides in the third quadrant of the Nyquist plot. What is observed from Figure 3, is that there is good alignment between the EIS-determined Z’ value (shown as squares) and the Z’ value from the AESEC expectation of Z’ (shown as diamonds). The alignment is assessed by looking at the vertical alignment of data, as there is no imaginary component associated with AESEC-derived impedance data (the raw AESEC data can be observed in Figure 1). Put simply, by using the low frequency impedance limit and employing RP (which mandates analysis of the inductive loop for a reactive system), the determination of the appropriate iMg2þ value at the low frequency impedance limit is possible. It is noted that the DC value of iMg2þ (as frequency!0) would agree with icorr obtained from EIS data by using the Stern– Geary relationship in which current is inversely proportional to resistance/impedance when EIS data near 0.005 Hz is used[1] (and the utility of a value of B 40 mV in the Stern–Geary Equation, where icorr = B/RP). This value of B does not necessitate assumption, as it can be determined as the required value based on the AESEC data, which happens to, however, match the values quoted by King and co-workers,[1] from DC measurement of Tafel slopes. AESEC has revealed several key findings and indicates that in cases when the inductance (which is a characteristic of the system) is ignored, then the resultant analysis will underpredict iMg2þ , giving rise to incorrect assumptions of lower valence or that electrochemical methods are somehow deficient, which was shown to not be the case. The alignment between the EIS and AESEC data is sound, however if using the Rt value as the parameter for which to calculate corrosion rates by using the Stern–Geary relationship, this would result in a gross overestimation of the Z’ value required to predict the low frequency impedance value commensurate with the AESEC-determined iMg2þ value (at the completion of the EIS test). In a similar vein, it was also previously indicated that the means for which dissolving Fe in acid was best analysed was by utilisation of RP (and not Rt) when corroborating EIS with atomic absorption.[23] A more detailed analysis of the electrochemical results allows the electrochemical current (which is the response that is measured by the potentiostat, ipstat) during the application of controlled potential EIS to be plotted against the convoluted total current from the potentiostat with the AESEC-determined iMg2þ value(Figure 4), as per the procedure in Ref. [16]. This analysis allows the current measured by the potentiostat to be compared directly to that from the AESEC-determined current (from dissolved ions and assuming z = 2).
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
Figure 4. The AC current recorded by the potentiostat during the application of potentiostatic EIS was recorded (i.e. where potential was the controlled parameter and current is the resultant measured parameter) and plotted here in concert with the deconvoluted AESEC determined iMg2þ . The baseline DC current is omitted. Agreement is seen between the two currents indicating a dissolution stoichiometry of Mg2 + , whilst the notion of cathodic activation is apparent.
Such an analysis yields two important outcomes from Figure 4. These are: 1) there is high agreement between the iMg2þ and ipstat values, indicating that the dissolution of Mg is via Mg2 + , and 2) the cathodic current measured in each downcycle following anodic polarisation shows a higher cathodic current. This latter point is in direct agreement with recent work that has ascribed the so-called negative difference effect (NDE) upon Mg as a consequence of enhanced cathodic (catalytic) activity in response to anodic dissolution.[11] As such, when AC is cycled between the same potentials, a higher cathodic current can be sustained by the Mg specimen as cycling proceeds, in accordance with previous reports from DC tests.[22] However, what is highlighted here, is that we do not have the evidence to suggest that the EIS itself is causing the activation (i.e. as an AC effect, which would be presumably evident during the + half wave of the AC cycle), rather, that EIS is capable of measuring this effect when coupled with AESEC. In summary, this work indicates that the inductance should not be ignored in EIS measurements of Mg, and RP is favoured for the measurement of corrosion rate over Rct. Electrolytes that give rise to this inductive response have recently been summarised by Bland and co-workers.[24] Justification for the use of an inductor in the assessment of EIS data for Mg was provided herein, and the inductive response was correlated to the enhancement of Mg dissolution with time, manifest as test frequency!0. This has two major implications; the first, practical implication, is that an accurate assessment of Mg dissolution in the absence of AESEC requires the utility of an inductor. The second, more general implication, is that an obvious advantage of the EIS–AESEC technique is the ability to treat dynamic systems directly, by analysing data in the time domain that takes into account the system evolution. This is an important notion, as the Mg system, like many active-metal systems, is not stable on the basis of the anticipated causality. It was ChemPhysChem 0000, 00, 1 – 4
&3&
These are not the final page numbers! ÞÞ
CHEMPHYSCHEM COMMUNICATIONS shown by King and co-workers[1] that Mg EIS data does still satisfy the Kramars–Kronig transform, and thus, the dynamic nature of the system and the use of EIS is a matter for each researcher to contend with, as a complex response is common to the majority of corroding systems. Importantly however, it has been shown herein that proper assessment is capable of yielding meaningful results. Finally, there is no need to invoke any notion of lower valence Mg to describe results herein, which are wholly consistent with Mg2 + -dissolution stoichiometry. The aspects clarified herein regarding the instability of Mg, the activation of Mg, the enhanced catalytic activity of Mg and the dissolution stoichiometry of Mg, were all revealed by using online AESEC and EIS methods and all are relevant in the assessment of Mg and Mg alloys as electrode materials or in determination of corrosion rates.
Experimental Section The AESEC method used herein has previously been described in detail, including a schematic of the electrochemical flow cell and a description of the data acquisition and the calibration procedure.[9, 25] The system consists of a Horiba Jobin Yvon inductively coupled plasma optical emission spectrometer coupled with a Gamry Reference 600 potentiostat and an electrochemical flow cell. In Brief, the working electrode (Mg) is exposed to a flowing electrolyte (at 3 mL min1) in a small volume (0.2 mL) flow cell. The geometrical area of the sample exposed to the electrolyte was 0.5 cm2 as defined by the contour of an O ring. The flow channel has an entrance at the bottom and exit at the top so that electrolyte passes through the cell from bottom to top, such that any gas generated during the experiment flows out of the cell. The electrolyte from the electrochemical flow cell is continuously fed into the plasma by using a peristaltic pump. The electrolyte is atomised by the plasma and the elemental components are monitored in real time by following the emission at specific wavelengths by using an array of photomultipliers in a polychromator. In this work, the emission intensity of Mg (279.553 nm) was recorded simultaneously as a function of time, along with the electrochemical potential and current. Calibration of intensity versus concentration was performed by using standard ICP-OES methods and normalised against calibration standards purchased from SPC Science, France. A Pt counter electrode and saturated calomel reference electrode were used. In the electrochemical flow cell they are positioned in a 10 mL cylindrical secondary compartment of stagnant electrolyte. This compartment is separated from the working electrode compartment by a porous membrane , which allows ionic current to pass from one compartment to the other while preventing bulk mixing of the two electrolytes. All experiments were performed at ambient temperature, approximately 24 1 8C. Potentiostatic EIS was conducted over a frequency range of 100 kHz to 0.005 Hz with an AC amplitude of 20 mV rms. The experiments were performed using commercially pure (termed pure herein) 99.9 % Mg (Good Fellow) with a nominal composition of Mg, Al 70 ppm, Cu 20 ppm, Fe 280 ppm, Mn 170 ppm, Ni < 10 ppm, Si 50 ppm, Zn < 20 ppm. The sample was ground to
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org 4000 grit with SiC disks with ethanol. The 1.0 m NaCl solution was prepared from analytical grade materials and deionised water of 18.2 MW cm1 (prepared using a Millipore system).
Acknowledgements The support of Ecole Nationale Suprieure de Chimie de Paris, Centre National de la Recherche Scientifique (ENSCP-CNRS) is gratefully acknowledged. Keywords: corrosion · electrochemistry · ICP–AES · impedance spectroscopy · magnesium [1] A. D. King, N. Birbilis, J. R. Scully, Electrochim. Acta 2014, 121, 394 – 406. [2] S. Feliu, C. Maffiotte, A. Samaniego, J. C. Galvan, V. Barranco, Appl. Surf. Sci. 2011, 257, 8558 – 8568. [3] Z. X. Qiao, Z. M. Shi, N. Hort, N. I. Z. Abidin, A. Atrens, Corros. Sci. 2012, 61, 185 – 207. [4] F. Y. Cao, Z. M. Shi, J. Hofstetter, P. J. Uggowitzer, G. L. Song, M. Liu, A. Atrens, Corros. Sci. 2013, 75, 78 – 99. [5] G. Baril, G. Galicia, C. Deslouis, N. Pebere, B. Tribollet, V. Vivier, J. Electrochem. Soc. 2007, 154, C108 – C113. [6] A. Pardo, M. C. Merino, A. E. Coy, F. Viejo, R. Arrabal, S. Feliu, Electrochim. Acta 2008, 53, 7890 – 7902. [7] A. Pardo, S. Feliu, M. C. Merino, R. Arrabal, E. Matykina, Int. J. Corros. 2010, DOI:10.1155/2010/953850. [8] Z. Shi, F. Cao, G. L. Song, A. Atrens, Corros. Sci. 2014, DOI: 10.1016/ j.corsci.2014.07.060. [9] J. S´wiatowska, P. Volovitch, K. Ogle, Corros. Sci. 2010, 52, 2372 – 2378. [10] G. Williams, N. Birbilis, H. N. McMurray, Electrochem. Commun. 2013, 36, 1 – 5. [11] G. S. Frankel, A. Samaniego, N. Birbilis, Corros. Sci. 2013, 70, 104 – 111. [12] S. Lebouil, A. Duboin, F. Monti, P. Tabeling, P. Volovitch, K. Ogle, Electrochim. Acta 2014, 124, 176 – 182. [13] L. Rossrucker, K. J. J. Mayrhofer, G. S. Frankel, N. Birbilis, J. Electrochem. Soc. 2014, 161, C115 – C119. [14] M. Curioni, Electrochim. Acta 2014, 120, 284 – 292. [15] A. Samaniego, B. L. Hurley, G. S. Frankel, J. Electroanal. Chem. 2014, DOI: 10.1016/j.jelechem.2014.04.013. [16] K. Ogle, S. Weber, J. Electrochem. Soc. 2000, 147, 1770 – 1780. [17] K. Ogle, A. Tomandl, N. Meddahi, M. Wolpers, Corros. Sci. 2004, 46, 979 – 995. [18] S. O. Klemm, A. Karschina, A. K. Schuppert, A. A. Topalov, A. M. Mingers, I. Katsounaros, K. J. J. Mayrhofer, J. Electroanal. Chem. 2012, 677, 50 – 55. [19] P. Volovitch, M. Serdechnova, K. Ogle, Corrosion 2012, 68, 557 – 570. [20] A. Ulrich, N. Ott, A. Tournier-Fillon, N. Homazava, P. Schmutz, Spectrochim. Acta Part B 2011, 66, 536 – 545. [21] M. Mokaddem, P. Volovitch, F. Rechou, R. Oltra, K. Ogle, Electrochim. Acta 2010, 55, 3779 – 3786. [22] N. Birbilis, A. D. King, S. Thomas, G. S. Frankel, J. R. Scully, Electrochim. Acta 2014, 132, 277 – 283. [23] W. J. Lorenz, F. Mansfeld, Corros. Sci. 1981, 21, 647 – 672. [24] L. Bland, A. D. King, N. Birbilis, J. R. Scully, Corrosion 2014, DOI: 10.5006/ 1419. [25] K. Ogle, J. Baeyens, J. Swiatowska, P. Volovitch, Electrochim. Acta 2009, 54, 5163 – 5170. Received: September 25, 2014 Published online on && &&, 2014
ChemPhysChem 0000, 00, 1 – 4
&4&
These are not the final page numbers! ÞÞ
COMMUNICATIONS In the loop: Reactive metals such as magnesium reveal significant inductance during electrochemical impedance spectroscopy (EIS) measurements. Combined atomic emission spectroelectrochemistry and EIS provided a definitive assessment for the electrochemical analysis of Mg dissolution.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
V. Shkirskiy, A. D. King, O. Gharbi, P. Volovitch, J. R. Scully, K. Ogle, N. Birbilis* && – && Revisiting the Electrochemical Impedance Spectroscopy of Magnesium with Online Inductively Coupled Plasma Atomic Emission Spectroscopy
ChemPhysChem 0000, 00, 1 – 4
&5&
These are not the final page numbers! ÞÞ
