Investigating the Real Time Dissolution of Mg Using Online Analysis by ICP-MS (2024)

The Mg dissolution measured during a series of galvanostatic polarization steps, primarily in the cathodic direction, is shown in Figure 1. The applied current density (the current flowing from the potentiostat), i_app, can be compared to the current density associated with the measured concentration of dissolved Mg²⁺ ions, i_diss(Mg²⁺), during polarization. Equation 1 was used to convert dissolved Mg concentration, c_Mg (M), to a dissolution current density:

where z = 2 is the charge transfer number (assuming that magnesium dissolves to form divalent Mg (Mg²⁺)), F is the Faraday constant, V_f is the volume flow rate, and A is the working electrode area. The sample was initially held at the OCP for 15 min followed by a preconditioning anodic step of 237 μAcm⁻² for 10 min. This preconditioning allowed for the realization of a c_Mg greater than the open circuit dissolution rate. Following the anodic step, the Mg dissolution rate at OCP was more stable. The second OCP measurement was followed by a sequence of applied cathodic current steps each of 10 min, with the cathodic current density increased by increments of 119 μAcm⁻².

From Figure 1 it is evident that cathodic polarization led to decreasing rates of Mg dissolution (i_diss(Mg²⁺)). The application of cathodic polarization did not completely suppress i_diss(Mg²⁺) until appreciable cathodic currents were applied; in this case total suppression of i_diss(Mg²⁺) was not observed until the applied current density was ∼ −470 μA/cm². The data presented in Figure 1 gives the raw data for one test run, however the test has been successfully reproduced several times.

Anodic dissolution results are reported in Figure 2. As was done during the cathodic dissolution experiment, the sample was initially held at the open circuit for 20 min followed by a preconditioning anodic step of 237 μAcm⁻² for 10 min. The second open circuit measurement was followed by a sequence of applied anodic current steps, the duration of each step being 10 min and the increment between each step being 119 μA/cm².

Figure 2 has several salient features, each of which merit description. First, there was some 'noise' in the collected signal, the origin of which was attributed to hydrogen gas bubbles by Światowska.¹¹ Second, as i_app increased, the magnitude of i_diss(Mg²⁺) approached the magnitude of i_app. This similarity of i_app and i_diss(Mg²⁺) at higher anodic currents is also consistent with the data presented by Światowska.¹¹ This is reinforced by presentation of a snapshot of raw data collected during an independent duplicate test on various locations upon an Mg sample, as seen in Figure 3. The third important feature is that i_diss(Mg²⁺) was higher than the applied current density at small galvanostatic signals, indicating that the cathodic reaction was not reduced to zero during the (anodic) galvanostatic pulse, as is discussed further below.

In regards to further elaboration, we emphasize the setup used in this present study did not permit simultaneous hydrogen collection, and therefore the discussion is based not solely upon the results collected, but augmented by aspects from the literature and a hypothesis driven discussion that remains the focus of ongoing work.

The working electrode potential was measured during the galvanostatic polarization, which allows for the representation of the data in the form of log i – E curves. Figure 4 depicts the applied current density (i_app) and the dissolution current density (i_diss(Mg²⁺)) as a function of potential. We note that the potential values were measured, i.e. the potential was not the independent controlled variable, as was the case in a previous study performed under galvanostatic conditions.¹⁰ As expected, the log of the Mg dissolution rate increased linearly with increasing potential, showing the form of typical anodic metal dissolution behavior. The nominal Tafel slope is ∼300 mV/dec, however the plotted E values in Figure 4 were not corrected for ohmic potential (IR) drop and so the actual (anodic) Tafel slope is less than this value. Uncompensated IR drop is expected to be present during polarization due to the resistance associated with hydrogen bubble evolution at the electrode surface. This aspect is addressed further below. Significant IR drops were also evident in other works.¹⁰

Besides using a different Mg ion measurement method, one key difference of this work compared to the study by Światowska¹¹ is that the dissolution was studied under current (galvanostatic) control. Galvanostatic control allows for the determination of more data points closer to the open circuit potential, which are difficult to access by potentiostatic polarization owing to the non-polarisable nature of Mg dissolution.

The kinetics of hydrogen evolution (HE) can be evaluated based on the applied current (i_app) and measured Mg²⁺ dissolution current (i_diss(Mg²⁺)). Assuming that magnesium dissolution to form Mg²⁺ and hydrogen evolution (from the reduction of water) were the only reactions taking place on the working electrode surface, and that all of the Mg²⁺ ions were transported to the ICP-MS system (100% detection efficiency), the HE current density, i_HE, can be determined from the difference of the applied current density, i_app, and i_diss(Mg²⁺).

Included in Figure 4 is the rate of HE hydrogen evolution calculated from Eqn. 2. Because i_diss(Mg²⁺) decreased with increasing applied cathodic current, the rate of hydrogen evolution increased, which is to be expected because it is the primary cathodic reaction. The apparent Tafel slope is about 329 mV/dec. The persistence of Mg dissolution at cathodic potentials was also previously observed,¹¹ however the data presented herein were collected under galvanostatic conditions allowing for direct comparisons on a current basis, as discussed further below.

The curve for i_HE indicates that the rate of HE decreased continuously in the region of anodic polarization. The kinetics shown in Figure 4 are of the form expected for a typical mixed potential corrosion system with anodic reaction rate increasing exponentially and cathodic reaction rate decreasing exponentially as the potential increases. However, this is not a typical corroding system. Mg dissolution exhibits the negative difference effect, and previous investigations have shown that the rate of HE increases with increasing anodic potential,^2,10 in apparent contradiction to the calculated rate of HE shown in Figure 4. In fact, it is now appreciated that the assumption used to produce Figure 4, i.e. that no additional hydrogen gas is formed, is not accurate based on the more definitive works in,^24,25 with Williams clearly observing persistent hydrogen evolved as a cathodic process. None the less, we present Figure 4 as we have, in order to build the subsequent discussion from simple principles. Despite the trend shown by the i_HE line in Figure 4, significant hydrogen bubbles were observed (visually, again noting that hydrogen could not be quantified by the setup here) to evolve on the sample surface at the highest applied currents, indicating an increase of the HE rate for high anodic polarization. Data for the highest anodic current tested are shown in Figure 5, which reveals the signal collected. The average i_diss(Mg²⁺) measured during this period was 2390 μAcm⁻², whilst the i_app was 2256 μAcm⁻². However, large and somewhat periodic fluctuations in i_diss(Mg²⁺) are evident with instantaneous values both higher and lower than i_app. Such fluctuations could likely be caused by the effects of hydrogen bubbles that evolved from the surface affecting the transport of dissolved Mg ions to the ICP system, or indicative of instability and the jumping between two surface conditions.

In general, the measured potential during polarization was more stable than i_diss(Mg²⁺) at every applied current density, but the fluctuations in both the potential and i_diss(Mg²⁺) were greatest at the highest applied current densities. The fluctuations in potential were also affected by varying ohmic resistance associated with the evolution of the gas bubbles. However, there was no strong correlation between the potential and current noise. It should be noted that i_diss(Mg²⁺) was calculated from the ICP data, and was not directly responsible for any ohmic potential drop. Nonetheless, hydrogen bubbles would be expected to influence both i_diss(Mg²⁺) and measured potential.

As described above, the calculated (ideal) curve for i_HE in Figure 4, which is based on Equation 2, is not an accurate representation of the real rate of HE, which continued and even increased with high anodic polarization. We recall the setup used in this study did not permit simultaneous collection of hydrogen gas. Data from a previous study in which the rate of hydrogen evolution was measured using a conventional electrochemical cell, however which did not allow for measurement of i_diss(Mg²⁺),¹⁰ is therefore used as a proxy herein. The use of these data (from)¹⁰ requires the assumption that the relationship between HE (*i_H2) and i_app is the same in both systems despite differing cell geometry. Nonetheless, the previous data can be fitted to extract the following empirical relationship:

where *i_H2 is the current density calculated from the rate of HE as a function of i_app for pure Mg in the anodic region. This expression is shown in Figure 6 along with the values of i_Mgtot and i_app, although here on a linear scale. The rate of HE increased with potential even though the difference between i_diss(Mg²⁺) and i_app decreased.

The total rate of Mg oxidation, i_Mgtot, must therefore be represented by the total cathodic current generated, which is the sum of the current flowing through the potentiostat to the distant counter electrode, i_app, and the current consumed by the local hydrogen evolution on the anode, i_H2:

A line for i_Mgtot is shown in Figure 6 assuming that i_H2 = *i_H2. This total rate of Mg oxidation was not reflected by the current measured by ICP, i_diss(Mg²⁺), because some of the oxidization current is hypothesized to have been used in the formation of insoluble product, i.e. either a surface film or some other insoluble oxidized species.²⁴ Assuming that the insoluble product is all surface film, the current going to film formation would be the difference between i_Mgtot and i_diss(Mg²⁺). Therefore,

Figure 6 also shows a line for i_Mg,film. It is important to reiterate that i_H2 in Equations 4 and 5 was assumed to be *i_H2, which is the fit from previous data, and not the value given by Eqn. 2, which was described to be inaccurate. It is possible to integrate i_Mgfilm to determine a nominal film thickness as shown in Figure 7. The thickness will depend on the film composition and is given in Figure 7 for the cases of MgO or MgOH₂. Very recent work from Taheri²⁹ indicated that a surface film of ∼400 nm was formed following anodic polarization of pure Mg at −1.0 V_SCE (∼ −0.75V_SHE). Similarly, the film thickness calculated from the work of Lebouil²⁴ is approximately 2000 nm. Such values are consistent with those given in Figure 7, but this is obviously an important avenue of future work, which is being pursued.

An interesting revelation from the present work that has not been uniquely appreciated previously, is the implication of the data revealing that i_diss(Mg²⁺) ≈ i_app at higher values of i_app. This is in itself an important outcome, which has been determined independently herein. An implication of this, is that according to Equation 5, it is seen that i_Mgfilm ≈ i_H2, which is observed in Figure 6. It would not necessarily be expected that the current associated with film formation is equal to the rate of the local cathodic reaction as it is reasonable to assume that the electrons used for HE might come from oxidation anywhere on the electrode surface. This observation that essentially all of the current associated with local hydrogen evolution was used for film formation with high efficiency suggests that the electrolyte near the dissolving Mg electrode reached saturation quickly. Every hydroxide ion released by water reduction combined with Mg²⁺ ions to form insoluble product. This is more remarkable given that the electrolyte was flowed past the working electrode surface. In view of the relationship between i_Mgfilm and i_H2, along with anodic (i_a) and cathodic (i_c) currents on anodically polarized Mg (noting that a true cathodic current, i_c, exists and is measurable on Mg),²⁵ the observation described above is in agreement with independent SVET work of Williams²⁵ who revealed a corresponding and related increase in i_c with i_a for a fixed applied current (i_app).

In addition, Mg²⁺ generally hydrolyses very weakly, causing the pH to increase during Mg dissolution due to the effects of the cathodic reaction. This was shown visually by Kirkland, who used an indicator in the electrolyte to reveal pH increasing at the dissolving anode.⁹ Mg will hydrolyse and Mg(OH)₂ will precipitate if the pH tends toward alkaline, and therefore the OH⁻ generated by local hydrogen evolution (aspects of which are elaborated in)³⁰ is critical in forming film.

Future work will contribute to a holistic understanding of the dissolution by studies in buffer solutions that may prohibit the formation (and hence, breakdown) of any protective layers. In particular, when coupled with the findings from recent complementary data,²⁴ SVET,^12,20,25 and detailed surface sensitive electron microscopy,^29,31,32 the physical origins of the NDE are being elucidated. Enhanced catalytic behavior of Mg is in operation during anodic dissolution of Mg,¹¹ whilst it can be affirmed that there is no requirement for assuming that Mg⁺ is required, nor does Mg⁺ play a role based on the data described herein.