The Role of Surface Films and Dissolution Products on the Negative Difference Effect for Magnesium: Comparison of Cl⁻ versus Cl⁻ Free Solutions

High purity, 99.9% nominal, Mg rod (80 ppmw Fe) with a diameter of 7.9 mm was used in this study as supplied by Alfa Aesar (Table I).

Samples were mounted in epoxy resin with a Ni ribbon attached to provide an electrical connection and mechanically ground to a 1200 grit SiC finish using ethanol as lubricant. Sample surfaces were cleaned with ethanol and dried with compressed air after each successive grinding step. After completion of grinding, samples were placed under vacuum in a desiccator for up to 24 hours to ensure that all water was removed from the epoxy for accurate mass loss measurements.

Electrochemical testing was performed using a three electrode vertical 'flat-cell' with a saturated calomel reference electrode (SCE) and platinum mesh counter electrode. An electrolyte filled burette with a funnel attached was centered over the sample for H₂ capture as described elsewhere.⁸ Both the funnel and burette were made out of glass to minimize hydrogen bubbles forming on the funnel and burette walls and to prevent H₂ permeation through polymeric materials.⁵¹ In this study, quiescent 0.6 M NaCl, 0.6 M NaCl saturated with Mg(OH)₂, 0.1 M MgCl₂, 0.1 M Na₂SO₄, and 0.1 M TRIS buffer solutions were used. All solutions were prepared using 18.2 MΩ deionized water at 25°C and pre-charged with hydrogen with exception to 0.1 M MgCl₂ due to oxidation of Cl⁻ to Cl₂ gas on the counter electrode during charging. The initial pH of each solution was measured to be as their natural pH with exception to 0.6 M NaCl solution which produced an initial pH in the range of 8–9 discussed in our previous work.³¹ The TRIS buffer solution was acquired as 1 M, pH = 7 solution and diluted to a concentration of 0.1 M resulting in an initial pH of ≈7.25.

The procedure for the electrochemical testing used in this investigation is outlined in Fig. 1. All samples were subjected to 24 hour anodic potentiostatic holds in the range of −1.55 V_SCE to −1.40 V_SCE immediately following exposure to the electrolyte; additional potentials were tested in certain electrolytes to validate any trends. IR correction was not carried out for the potentiostatic polarization as it is not required for charge balance calculations to be made and is not relevant to HER in the anodic region. Following potentiostatic polarization, samples were allowed to rest at their open circuit potential (OCP) for 1 min before a cathodic potentiodynamic polarization scan from OCP to −3 V_SCE at a scan rate of 2500 mV/s. Such a fast scan rate allows for an expanded range of activation controlled kinetics to be explored which can lead to more confident estimations of Tafel kinetics and also minimize changes in solution chemistry (i.e. pH and other dissolved ion concentration) on the sample surface throughout the duration of the scan. All solutions were also retained for cathodic polarization scans of film free surfaces after removing dissolution products and for analysis by ICP-OES. Dissolution products were cleaned from sample surfaces using 200 g/L CrO₃ according to ASTM G1,⁵² which removes corrosion films whilst not causing further dissolution of the Mg metal. Samples were then placed under vacuum in a desiccator for a 24-hour period before the final mass was measured. Following mass loss measurements, the cleaned samples were then subjected to an additional cathodic polarization scan at 2500 mV/s^g from OCP to −3 V_SCE after completion of 1 minute at OCP using the solution that was retained from previous cathodic polarization scans, to enable a comparison of cathodic kinetics with the surface films intact and after cleaning of any surface films from the surface (referred to as 'film' and 'film free', respectively) without alteration of the bulk solution environment. Cathodic polarization scans on freshly ground samples which were subjected to CrO₃ acid cleaning did not reveal a statistically significant difference in the measured cathodic kinetics compared to samples to freshly ground samples which were not CrO₃ cleaned thus bringing confidence in analysis of film vs film free kinetics. It should be noted that some scans were terminated before reaching −3 V_SCE due to a 10 mA limit for potentiodynamic scans on certain potentiostats used in this study.

Zoom In Zoom Out Reset image size

Gravimetric mass loss measurements were performed using a scale with ± 0.01 mg resolution and converted to a corresponding charge density via Equation 4:

where z is the equivalent electrons per mole of Mg oxidized, n is the number of moles of Mg consumed, Δm is the change in mass, F is Faraday's constant, A is the area of the electrode, and a is the molar mass of Mg. In this study, z is assumed to be 2 for HP Mg following the half-cell reaction Mg → Mg²⁺ + 2e⁻. The hydrogen collected during testing of pre-saturated solutions was converted to a corresponding cathodic charge density using the ideal gas law in combination with Faraday's Law:

where z = 2 (i.e., 2H⁺ + 2e⁻ → H₂), P is the pressure inside the burette (1 atm), n is the number of moles of H₂ gas given by the ideal gas law, V is the volume of gas in the burette, F is Faraday's constant, R is the ideal gas constant, and T is temperature. The net charge density from the measured current was determined from the integrated signal of current density vs. time transient at each potential.

The charge balance under anodic polarization is therefore described by Equation 7:

following fundamental mixed potential theory⁵³ where is the summation of all HER occurring on the surface (ie following the notation of²¹).

All solutions tested were recovered for solution analysis using a Thermo Scientific iCAP 7200 ICP-OES. Each solution was acidified with additions of 1 M HCl to dissolve all insoluble dissolution products. Solutions were not diluted in contrast to our previous study³¹ in an attempt to boost the emission signal of impurity elements in solution. Furthermore, the CrO₃ solution used to clean dissolution products off samples was collected and analyzed to account for impurity atoms which may be trapped in the dissolution product films. However, analysis of the dissolution products collected from CrO₃ cleaning proved to be unsuccessful due to the impurity concentration of Fe in the CrO₃ which is was measured to be between 100–300 ppmw and thus did not allow for accurate quantitative analysis. In this work, the emission intensity of Mg (279.553 nm, 280.270 nm), Al (226.910 nm, 308.215 nm, 396.152 nm), Fe (238.204 nm, 239.562 nm), Mn (257.610 nm, 259.373 nm), Zn (206.200 nm, 213.856 nm), and Zr (339.98 nm, 343.823 nm) were recorded and used for the calculation of charge consumed and percent of elements in solution. The approximate detection limit for each element is shown in Table II.

Raman Spectroscopy was performed on sample surfaces after completion of potentiostatic polarization with corrosion products intact using a Renishaw Leica Raman Microscope. Scans were performed from 150 to 1200 cm⁻¹ utilizing a 200 mW, 514 nm laser at 50% power under a 20x objective lens through a 1800 l/mm visible grating; the laser spot size was on the order of 5 μm. Scans were accumulated twice with a 60 second exposure under standard confocality on at least three different areas of each sample. Background subtraction of spectra was not performed but spectral artifacts from the instrument were removed from each spectrum without interference of verified real peaks. Peak identification was made using standard spectra in the RRUFF project database.⁵⁴ No quantitative analysis of the concentration of identified species was performed as only species identification was desired for this study and a careful internal standard method must be adopted for each species.

The potentiostat recorded net current density traces as a function of time are shown in Figs. 2a through Fig. 2e for potentiostatic testing in various solutions. The corresponding initial and final bulk pH of each solution at each potential are reported in Fig. 2f.

Zoom In Zoom Out Reset image size

Two distinct regions are shown in Fig. 2a for potentiostatic polarization of Mg in 0.6 M NaCl. The first region occurs within the initial 1000 seconds of polarization where an increase in the net anodic current density is observed, followed by a decrease in net current density to a local minimum. During this time period, dark corroded regions, typically associated with enhanced cathodic kinetics, initiated on the sample surface and grew in the form of rapidly expanding disks. The maximum value and subsequent decrease in net anodic current in region one may be associated with increasing cathodic area and therefore increasing cathodic kinetics; however, H₂ was not captured as a function of time so it is unclear whether or not the rate of H₂ evolution increased during this time period. Growth of the dark corroded regions continued until the minimum current density value corresponding to when the entire Mg anode surface became covered with the dark film - which may indicate the end of dissolution governed by localized processes. The second regime of net current density would therefore occur under a more uniform type of dissolution with a decrease in net anodic current density attributed to thickening of the dark film on the Mg surface; the nature of the increase in net anodic current immediately following the minimum is unclear at this time. In addition, the dark film changes to a white color as it thickens which is normally associated with Mg(OH)₂.

Similar behavior is observed with both 0.6 M NaCl saturated in Mg(OH)₂ and 0.1 MgCl₂ solutions, over a similar range of time as indicated in Figs. 2b, 2c. However, the maximum height of the first peak in current in 0.6 M NaCl saturated with Mg(OH)₂ is less than that measured in non-saturated 0.6 M NaCl, and the decrease in net current density between this maximum and corresponding minimum is also less for saturated 0.6 M NaCl solution compared to non- Mg(OH)₂ saturated solution. In addition, the minimum in current density is shifted to later times for saturated 0.6 M NaCl solution which suggests that the possible increase in cathodic kinetics was not as great as in non-saturated 0.6 M NaCl solution and that the dissolution rate is less for 0.6 M NaCl saturated with Mg(OH)₂. This is not surprising since Mg dissolution would occur in the Mg(OH)₂ stable region of the Mg E-pH diagram for the entire duration tests in 0.6 M NaCl saturated in Mg(OH)₂ whilst non-saturated 0.6 M NaCl samples would spend some period of time corroding in the Mg²⁺ stable region before increases in bulk pH cause dissolution of Mg to shift to the Mg(OH)₂ region as indicated by the final pH values reported in Fig. 2f. Note that the change in pH is reflective of reactions occurring on both the Mg working electrode and Pt counter electrode and should vary with the amount of charge for unbuffered solutions. However, local increases in pH due to rapid H₂ evolution (2H₂O + 2e⁻ →H₂ +2OH⁻) and supersaturation of Mg²⁺ ([Mg²⁺] = 1.12 × 10⁻⁴ M required for saturation of Mg(OH)₂ in H₂O⁵⁵) in solution near the Mg surface is likely to be great enough to provide conditions where Mg(OH)₂ films form behind the advancing anodic front. The change in pH for 0.1 M MgCl₂ (Fig. 2f) was not as great as that observed in 0.6 M NaCl which is not surprising given the smaller relative net anodic current densities observed at each applied potential. These lower current densities are a result of the decreased Cl⁻ content in the 0.1 M MgCl₂ solution tested here compared to 0.6 M NaCl.

The potentiostatic behavior for 0.1 M Na₂SO₄ is shown in Fig. 2d. The net anodic current densities at each potential are similar to those in 0.1 M MgCl₂; however, there is little evidence of enhanced cathodic activity which is characteristic with the Cl⁻ containing solutions test in this investigation. The net current density is relatively stable throughout the duration of the test with decreases in the net current density likely due to thickening of films on the Mg surface. Corrosion of Mg in the presence of sulfate also differed from the behavior observed with chlorides in that the anode surface became covered with white film which became progressively thicker with time immediately after breakdown of the air formed oxide. Few black filaments formed which are noted to form in Cl⁻ containing environments. As in the case of Na₂SO₄, the net current density transient for samples tested in 0.1 M TRIS buffer solution (Fig. 2e) were relatively stable with time. Moreover, TRIS displayed no evidence of enhanced cathodic activity (typically indicated by a decrease in net anodic current density) and Mg samples exposed to TRIS buffer solution did not show the appearance of a surface film upon visual inspection. However, this does not exclude the possibility for nanoscale film formation on TRIS anode surfaces but an average final pH near 7.6 (Fig. 2f) indicates that Mg(OH)₂ films would not be thermodynamically stable throughout the duration of the test.

Characterization of sample surfaces after potentiostatic polarization by Raman spectroscopy is shown in Fig. 3. The spectra shown are representative of the typical scan on sample surfaces in each electrolyte. The peaks present at 278 cm⁻¹and 443 cm⁻¹ are indicative of brucite, Mg(OH)₂ whilst the peak observed at 1075 cm⁻¹ may be indicative of the formation of a carbonate species.⁵⁴ Note that the scattering intensities of each peak are related to incident laser intensity, molecular vibration frequencies, and the change in molecular polarizability and is not indicative of film thickness or amount while peak broadening arises from anharmonicities such as chemical (elemental) and physical (crystallinity) inhomogeneity.⁵⁶ There were no magnesium carbonate related samples in the RRUFF database which possessed a peak at 1075 cm⁻¹ but carbonate peaks are known to occur in the range of wavenumbers with magnesite, MgCO₃, possessing a peak at 1095 cm⁻¹. The peak observed at 982 cm⁻¹ is indicative of an SO₄²⁻ containing species which most closely resembles the standard spectrum of epsomite (MgSO₄·7H₂O).⁵⁴ The results in Fig. 3 are interesting because they show that there is a thin Mg(OH)₂ film present on TRIS samples despite there being no appearance of a surface film optically as shown in Fig. 4 for uncleaned TRIS sample surfaces which was also observed by Bland et al.⁵⁷ The cleaned sample surfaces for HP Mg corroded in Cl⁻ and SO₄²⁻ are also shown in Fig. 4 and revealed a difference in dissolution morphology between Cl⁻ and SO₄²⁻. It is clear from this figure that the HP Mg exposed to Cl⁻ yielded a rough surface with sharp features while the SO₄²⁻ exposed surfaces were smooth and lustrous.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The charge balance for all solutions tested by potentiostatic polarization is shown in Fig. 5. Inspection of Fig. 5a shows that the conditions for charge balance described by Eq. 7 are met assuming Mg dissolves only as an Mg²⁺ ion along with cathodic hydrogen evolution persisting on the corroded Mg anodes in 0.6 M NaCl. This is observed in every solution tested in this investigation within one standard deviation^h. The values for mass loss, hydrogen evolved, and net anodic charge measured by the potentiostat are similar for both saturated and unsaturated 0.6 M NaCl which is in agreement with expectations from the net current density vs. time transients shown in Fig. 2. However, the NDE is stronger in 0.6 M NaCl saturated with Mg(OH)₂ owing in large part due to an OCP hydrogen evolution rate about half that measured in unsaturated 0.6 M NaCl. Even lower corrosion rates were measured at OCP in 0.1 M MgCl₂ and 0.1 M Na₂SO₄ with the values for mass loss and hydrogen evolution in 0.1 M Na₂SO₄ being 1.67 ± 0.34 C/cm² and 1.58 ± 0.00 C/cm² respectively (not visible in Fig. 5d). As in the case of NaCl, the volume of H₂ captured for MgCl₂ increased rapidly at small anodic overpotentials but increased linearly at larger overpotentials. In contrast, the amount of H₂ captured increased linearly at every potential tested for Na₂SO₄. Moreover, the average charge due to mass loss at many potentials was less than that of the sum of the charge for the hydrogen evolved and applied anodic charge measured by the potentiostat. However, this average is within error and also validates the choice of z = 2.

Zoom In Zoom Out Reset image size

The charge balance for TRIS is shown in Fig. 5e. In contrast to the other electrolytes tested in this study, the average volume of H₂ captured was similar and did not change with increasing anodic overpotential for polarization between −1.65 V_SCE and −1.4 V_SCE and was less than that measured at OCP which suggests that there is no NDE in TRIS. This notion was confirmed at −1.3 V_SCE which showed a large decrease in the H₂ captured compared to −1.4 V_SCE as well as at the OCP. It should be noted that the ohmic resistance of the TRIS buffered solution was about 110 Ω-cm² producing a theoretical IR drop of 275 mV at 2.5 mA/cm², which led to the relatively small changes in measured net anodic current density with increasing applied anodic potential by increments of 100mV (Fig. 2e) and thus small changes in mass loss and H₂ captured.

The difference effect measured for each solution as a function of applied potential is shown in Fig. 6a where the difference effect, Δ, is defined as:

Zoom In Zoom Out Reset image size

The NDE is readily observed when plotting Δ vs. potential for Cl⁻ containing solutions and displayed a linear increase with increasing potential. The NDE at each potential is greatest for 0.6 M NaCl saturated with Mg(OH)₂ but the change of Δ with potential is similar for all Cl⁻ containing solutions. In comparison, the change in Δ vs. applied potential was much less for Cl⁻ free Na₂SO₄ solution but NDE was still evident. TRIS, however, did not display an NDE which is what would be expected from classical mixed potential theory applied to metal corrosion.⁵² Despite the presence of a positive difference effect for the TRIS solution, the faradaic efficiency for Mg dissolution (Q_{i net anodic}/Q_{mass loss}) is low at small anodic overpotentials as shown in Fig. 6b which suggests that the local cathodic reaction is still a significant factor to reducing the net applied anodic current compared to the true anodic current density. However, the faradaic efficiency increases with increasing anodic overpotential for TRIS and also for 0.6 NaCl and 0.1 M MgCl₂. In contrast, 0.6 M NaCl saturated with Mg(OH)₂ (Fig. 6b) displayed a relatively constant efficiency of about 0.5 within error while 0.1 M Na₂SO₄ indicated the greatest faradaic efficiency of all the electrolytes studied and revealed a decrease in faradaic efficiency at applied potentials greater than −1.50 V_SCE.

Potentiodynamic polarization was performed after anodic potentiostatic polarization with the anodic surface film intact and in a film free condition after cleaning sample surfaces with CrO₃ to examine the effects of enhanced cathodic kinetics, Fig. 7. E vs. log i polarization behavior for a freshly polished sample after 1 min immersion at OCP is displayed for reference to show the cathodic kinetics of an Mg surface that has no actively corroding filaments that might exhibit anodically enhanced cathodic kinetics. In addition, the calculated current density associated with the H₂ captured at each potential during anodic polarization is shown for each solution assuming that the HER rate is constant with time. Figs. 7a–7c reveal that the cathodic kinetics are faster on samples immersed in Cl⁻ with the surface film intact compared to those with no films which is in contrast to our previous findings with a 1 mV/s scan rate.³¹ Moreover, the cathodic kinetics are greater than that of freshly polished samples which is consistent with previous reports in chloride containing environments.³⁸ However, there appears to be no clear trend of increasing cathodic kinetics with increasing prior anodic polarization potential. This is also observed in 0.1 M Na₂SO₄ (Fig. 7d) but the cathodic kinetics for both the film and film free conditions were slower than that of a freshly polished sample. The film free condition tended to show faster cathodic kinetics than the filmed condition at large cathodic overpotentials but were similar at smaller overpotentials which is in contrast to Cl⁻ environments. TRIS (Fig. 7e) also did not show any variation in cathodic kinetics with increasing prior polarization between −1.65 V_SCE and −1.4 V_SCE. However, the rate of HER decreased on samples corroded at −1.3 V_SCE (film free runs are not shown on this plot because there was no appearance of a surface film from visual inspection). The absence of any strong variation of cathodic kinetics for applied potentials between −1.65 V_SCE and −1.4 V_SCE is consistent with the similar volume of hydrogen captured as shown in Fig. 5e. Moreover, the decrease in cathodic kinetics at −1.3 V_SCE compared to the other applied anodic potentials is also consistent with the decreased volume of hydrogen measured at this potential.

Zoom In Zoom Out Reset image size

The results of ICP-OES (not shown for brevity) did not reveal the presence of Fe or other impurities in solution above the detection limit for all Cl⁻ containing environments and for Na₂SO₄ which is consistent with our prior investigation.³¹ However, Fe was detected in solution for samples polarized in TRIS consistent with the bulk concentration of Fe in the HP Mg used in this investigation.

The various electrolytes used in this study have provided a number of different conditions to study the phenomenon of the negative difference effect of Mg. The effect of each electrolyte on the NDE is presented and summarized in Table III. It is apparent from the results that the type and concentration of the anion produces a large difference on not only the corrosion and dissolution rate of Mg anodes but also in the extent of the NDE. The results presented herein suggest that the NDE is strongly dependent on both presence of absence of Cl⁻ as well as pH changes in response to Mg²⁺ dissolution and the ability of Mg to produce stable films. In this study, the NDE was observed for all solutions that enabled the formation of thick Mg(OH)₂ surface films (Fig. 6a). Of these solutions, 0.6 M NaCl saturated with Mg(OH)₂ produced the strongest NDE which is consistent with observations of greater NDE at alkaline initial pH values.⁴¹ An initial pH in the Mg(OH)₂ stable region combined with a solution that is saturated with Mg(OH)₂ will greatly promote the formation of Mg(OH)₂ films and thus this result is not surprising if the NDE is described by the stable film forming reaction mechanism proposed by Lebouil et al.⁴³ However, a recent study by Yang et al. report that no NDE was observed for anodic polarization of Mg in 1 M NaOH which promoted the formation of a white film on the surface and a passive current density during anodic potentiodynamic scans.²² However, NDE was observed in 1 M NaOH + 3.5 wt% NaCl and no passive region formed. Their suggestion was that the NDE is related to a Cl⁻-induced film rupture mechanism where a stable Mg(OH)₂ film is present as in the case of NaOH but the addition of NaCl enabled the breakdown of stable passive films. This local breakdown enabled fresh Mg surface to be exposed to the electrolyte which resulted in an amplified anodic current which is accompanied by a proportionally large increase in cathodic current in order to maintain charge balance. The result of film rupture is rapid HER behind the advancing dissolution front in the form of hydrogen streams which are well known in literature with a number of Mg alloys and processing conditions.^29,58,59 But this fresh Mg would also be repaired by the reactions described by Equations 2–3 which facilitates HER.

The effect of film stability on the NDE is apparent in this study with comparisons between Cl⁻ environments and Na₂SO₄. Examination of the surface morphology after removing the anodically formed surface films revealed a contrast in attack as shown in Fig. 4. Both NaCl and MgCl₂ environments produced rough surfaces with many sharp features and pit like voids while Na₂SO₄ surfaces were smooth and lustrous. This suggests that Cl⁻ was able to penetrate Mg(OH)₂ films perhaps resulting in stress from electrostatic repulsion leading to local breakdown while more uniform dissolution with minimal localized breakdown or film rupture occurred in Na₂SO₄ which contributed to decreased dissolution rates and NDE. Moreover, this suggests that films formed in the presence of Na₂SO₄ are more stable than those in the presence of Cl⁻ which could be related to 0.1 M Na₂SO₄ having an SO₄²⁻ activity coefficient < 1.⁶⁰ A low anion activity coefficient would lead to a less reactive environment, which decreases the likelihood of dissolution film breakdown during anodic polarization. In addition, it was more difficult to mechanically remove the Na₂SO₄ films when removing surface films for ICP-OES analysis compared to Cl⁻ solutions films. Films formed in Cl⁻ solutions readily spalled off whilst Na₂SO₄ films were adherent. In the case of TRIS, no visible films formed on the surface during anodic polarization and thus no film rupture would occur. Raman spectroscopy did reveal that Mg(OH)₂ was present on the surface of samples tested in TRIS, but this film is likely to be tens of nanometers thick and may have formed in humid lab air.

A general description of the factors that may lead to anodic hydrogen evolution from an anodically polarized Mg surface can be presented by the form of the simple expression in Eq. 9, that includes the minimum number of terms as explored in recent works, whilst acknowledging there may other hitherto unknown phenomena necessitating other terms.

In this equation, HE_anode represents a contribution from an increase in i_{o, HER} as described in Ref. 21, HE_impurities is the contribution to enhanced rates of reduction from impurity metal enrichment, and HE_Film is the contribution to enhanced rates of reduction from filmed areas. These three phenomena have been purported to enhance the rate of the cathodic reduction reaction, and hence, the NDE. It is worth noting that Tafel extrapolation from net cathodic data is inappropriate as HE_anode would not be present in net cathodic data and HE_Impurities may be reduced as well. Furthermore, such extrapolation is not relevant if the Tafel or Heyrovsky mechanism is dominant during net cathodic polarization and H₂ evolution as described by the film forming mechanism in Eq. 2 dominates during net anodic polarization as they are distinctly different reactions. It is regarded that enrichment of impurities might be relevant from three standpoints: (1) Mg impurities may result from transition metal replating, (2) impurities may occur in the form of particles trapped in oxides or hydroxides, (3) solute in solid solution may accumulate at the moving metal/solution interface. The relative proportion of the terms in Eq. 9 is important in terms of electrochemical engineering of Mg, and a focal point of the work herein.

The physical reasoning for the large (and increasing) i_o,HER in the anodic region has never been definitively determined. Furthermore, Frankel et al.²¹ state that the contributions of HE_impurities + HE_Film have a small influence on the NDE based on extrapolation of cathodic polarization scans conducted long after anodic solutions may have dissipated from the anodic region. However, such an assertion is not entirely consistent with the growing number of works that have sought to quantify contributions of HE_impurities + HE_Film. Moreover, in such studies, slow scan rates were used to carry out the extrapolation of cathodic polarization which may lead to underestimation of the hydrogen evolution rate as the surface chemistry may have changed by the time the cathodic current density is assessed, due to rapid hydrogen evolution. Fast cathodic potentiodynamic polarization scans were used in this study to minimize changes in local solution chemistry after anodic polarization and to expand the range of activation controlled kinetics for Tafel extrapolation far away from OCP. However, applying Tafel extrapolation to fast cathodic scans (β_c = 315 mV/dec, determined for fast scans of Mg microelectrodes and in agreement with Ref. 8) still underestimates the anodic hydrogen evolution rate in each solution by several orders of magnitude at the anodic potentials explored here, and furthermore would still predict a decrease in cathodic kinetics with increasing anodic overpotentials as would extrapolations of the proposed Heyrovsky path.⁶¹ Tafel extrapolation of cathodic polarization curves performed for each solution (not shown) using the framework presented by Fajardo et al.²⁷ may be performed on the measured cathodic polarization curves to illustrate underestimation of the measured anodic HER rate. For example, the HER rate measured at −1.4 V_SCE in 0.6 M NaCl is about 1 × 10⁻² A/cm² (Fig. 7a) and extrapolation of its respective polarization curve to −1.4 V_SCE yielded an HER rate of on the order of 10⁻⁴ A/cm². Such an extrapolation may indicate the contribution of HE_Impurities + HE_Film to the anodic H₂ evolution rate but, again, would not capture possible contributions from Eq. 2. It should be noted that Tafel extrapolations were performed within 100 mV of the OCP despite an expanded region of activation controlled kinetics due to the fast scan rate used, which may lead to large errors in estimation of Tafel kinetics⁹ which is a further condemnation of performing Tafel extrapolation to determine anodic H₂ evolution rates on Mg and Mg alloys. Nonetheless, Tafel extrapolation of cathodic polarization curves could only account for the NDE if the cathodic kinetics continued to increase with increasing anodic overpotentials but this was not possible given the cathodic polarization findings in this investigation. As such, a mechanism describing the large i_o region HE_anode is still required to fully describe the NDE. This has been attempted recently by Yang et al.²² who posited that cathodic activation resulting from surface films and cathodic particles contribute to the large anodic i_o region instead of being regarded as separate from the large anodic i_o region. As such, the variation of impurity enrichment on the HP anode surfaces as a function of electrolyte remains an unresolved issue which may provide further clarity to the trends in NDE presented in this study. Surface analysis of dissolved HP Mg samples in each electrolyte will be evaluated and reported in a future study as a function of enrichment efficiency in an attempt to delineate the contributions of impurities on NDE. Additionally, contributions to the NDE from surface films have been shown to be an important factor based on the work presented herein.

These results and the framework presented herein also explain what is observed in TRIS. Here, enrichment of impurities by replating is eliminated due to the ability of TRIS to chelate transition metal ions. The buffering capacity and ability of TRIS to complex Mg²⁺ also eliminates the possibility of impurity accumulation in the oxide/hydroxide. The lack of a film has also eliminated possibilities of enhanced cathodic kinetics resulting from the hydroxide film. However, it is still possible that impurity particles are accumulating at the solution metal interface. This possible enrichment and the fact that there are always impurities present on the Mg surface could still stimulate remote anodic currents leading to H₂ evolution as described by Yang et al.²² Such an effect could account for the residual high rate of HER despite a positive difference effect and would also explain the lack in efficiency (Q_{i net anodic}/Q_{mass loss}) but it is unclear how much impurity enrichment occurs in TRIS and in the other solutions tested here. Future work will characterize HP Mg samples corroded in each solution studied here to determine the possible enrichment of noble impurities utilizing proton induced X-ray emission (PIXE), which has the ability to quantify of the near surface chemistry and also has the capability of elemental mapping which can provide insight on how impurities are enriching on the surface.