The Effect of Mg Morphology on the Irregular Behavior of the Electrochemical Quartz Crystal Microbalance in Mg[N(CF₃SO₂)₂]₂/glyme Solutions

Abstract

Mg plating/stripping reaction in Mg[N(CF₃SO₂)₂]₂/glyme based solution is studied by electrochemical quartz crystal microbalance method. During the cyclic voltammetry, the apparent mass decrease is observed in spite of the negative scan. The irregular response also appears in the Mg plating reaction with low constant current density apply. In the cases, Mg plating takes place locally and the size of each plating is relatively large of about 50 µm. The cross-sectional image of the plated Mg is tree-like structure, i.e., the large Mg crystal connects to the substrate with small contact area. From the results, we conclude that the specific Mg morphology causes the restoring force to the quartz substrate, resulting in the apparent mass decrease.

1. Introduction

Lithium-ion batteries (LIBs) are widely used in portable electric devices, such as smart phones, laptop PCs and so on. In addition, the recent population of the hybrid electric vehicles or electric vehicles accelerates the demands for LIBs. On the other hand, LIBs are facing some serious issues including safety problems, high cost and the limit of energy densities. In order to conquer the problems above, the research and development of post LIBs are paid much attention. Mg metal has the relatively low electrode potential of −2.37 V (vs. SHE) and large volumetric capacity of 3839 mA h cm⁻³. Besides, Mg resources, such as magnesite or dolomite are abundant and widely distributed around the world. Therefore, magnesium rechargeable batteries (MRBs) using Mg metal electrode are attractive for their essentially high energy densities and lower cost. In order to commercialize MRBs, however, a lot of problems still remain to be solved. As for Mg negative electrode, the reversible Mg plating/stripping reaction is difficult in the carbonate-based solutions and so far, Mg plating reaction has been almost limited in the ethereal based solutions. Among them, glyme based solutions containing Mg[N(CF₃SO₂)₂]₂ (Mg[TFSA]₂) show relatively high ionic conductivity (10⁻³ S cm⁻¹) and high anti-oxidation durability (ca. 3.5 V vs. Mg²⁺/Mg).^1–3 So far, we have focused on the mechanism of the Mg plating/striping or the insertion/extraction of Mg²⁺-ion into the intermetallic compounds in Mg[TFSA]₂/glyme solutions and found that the interaction between Mg²⁺-ion and solvents would be critical for the reversible Mg plating/stripping or Mg²⁺-ion insertion/extraction.^4,5 On the other hand, in Mg[TFSA]₂/glyme solutions, the coulombic efficiency is generally as poor as 10 % and large overpotential more than 1 V is needed for Mg plating/stripping reaction. The poor electrochemical performance is thought to be caused by the decomposition of TFSA⁻ anion and the following resistive film formation.^6,7 Contrary to the surface film on Li metal, which is Li⁺-ion conductive and electron insulative, the surface film on Mg metal is generally known to be Mg²⁺-ion and electron insulative. Therefore, the elucidation of the effect of the surface film formation on the Mg plating/stripping reaction process in Mg[TFSA]₂ contained solution is important.

In order to elucidate the metal plating/stripping reaction itself or the surface film formation process, the electrochemical quartz crystal microbalance (EQCM) method has widely used. For example, Serizawa et al. used EQCM for the study of Ag plating/stripping in ionic liquid and they reported that the concentration of Ag⁺-ion could be estimated near the electrode from the product of the viscosity and density, which is obtained from QCM.⁸ Naoi et al. focused on the surface film on Li metal and found that the thickness of the surface film was largely influenced by the anion species.⁹ Sonoki et al. also reported the similar result by conducting the cyclic voltammetry down to 0.1 V (vs. Li⁺/Li) with QCM, in which, the effect of Li metal plating was excluded.¹⁰ As for Mg metal plating, Aurbach et al. found that the mass change per moles of electron (m.p.e.) during Mg plating/stripping reaction in Grignard reagents was higher than the theoretical value and concluded the first stage of the plating/stripping was not the simple two electron reaction.¹¹

In the present study, we conducted EQCM measurement to elucidate the Mg plating/stripping mechanism in Mg[TFSA]₂/glyme based solutions. During the series of study, we found that the response of QCM showed the irregular behavior, i.e., the apparent mass decrease during Mg plating reaction and the mass increase during Mg stripping reaction. In order to elucidate the irregular behavior, we discussed the correlation between the specific morphology of the plated Mg and the response of QCM.

2. Experimental

The electrochemical Mg plating/stripping was studied by EQCM method using three electrode cell purchased from EC FRONTIER co., Ltd (VQ1). The working electrode was Au coated quartz (9 MHz, AT-cut). The counter electrode was Mg ribbon and the surface film was removed by abrasive paper before use. The reference electrode was Ag in 0.01 mol dm⁻³ AgNO₃ + 0.1 mol dm⁻³ Mg[TFSA]₂/triglyme, which was separated from the electrolyte solution by the porous glass. The reference electrode showed about 2.5 V vs. Mg²⁺/Mg in the Grignard reagent. Unless otherwise mentioned, all electrode potentials were referred to Mg²⁺/Mg. The electrolyte solution was 0.5 mol dm⁻³ Mg[TFSA]₂/diglyme. In order to confirm that EQCM measurement works properly, the Grignard reagent of 1.0 mol dm⁻³ EtMgBr/THF solution was also used. The Mg plating/stripping behavior was investigated by cyclic voltammetry or chronopotentiometry, and the corresponding mass change of the working electrode was monitored by QCM system (QCM922, SEIKO EG&G) coupled with the electrochemical measurement system (HZ-7000, HOKUTO DENKO). All cell fabrications and electrochemical measurements were conducted in Ar filled glove box. The morphology of the Mg plating was investigated by scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) with JCM-6000 (JEOL Ltd.) or focused ion beam SEM with MI-4000L (Hitachi High-Tech).

We estimated the mass change during the electrochemical measurements, according to the report by Sasaya et al.¹² The procedure is depicted in the followings.

The mass change of the electrode (Δm) is connected to the shift of resonance frequency (ΔF_m) by Sauerbrey equation (Eq. 1).   

\begin{equation} \Delta F_{m} = - \frac{2f_{0}^{2}}{A\sqrt{\rho_{q}\mu_{q}}}\Delta m \end{equation} (1)

(A: electrode area (0.20 cm²), ρ_q: density of quartz (2.648 g cm⁻³), μ_q: stiffness factor of quartz (2.95 × 10¹¹ g cm⁻¹ s⁻²), f₀: unperturbed frequency shift (8.94 MHz)).

However, the observed shift of resonance frequency change (ΔF_total) also contains the effect related to the viscosity (η) and the density (ρ) adjacent to the quartz (ΔF_ηρ), i.e., ΔF_total = ΔF_m + ΔF_ηρ. Therefore, we eliminated the effect of them by Kanazawa–Gordon equation (Eq. 2).   

\begin{equation} \Delta F_{\eta\rho} = -f_{0}^{\frac{3}{2}}\sqrt{\frac{\eta\rho}{\pi\rho_{q}\mu_{q}}} \end{equation} (2)

Although we can not obtain η and ρ value separately, the product of them are connected to the resonance resistance (R) by the following Eq. 3.   

\begin{equation} R = \frac{A\sqrt{2\pi f_{0}\eta\rho}}{k^{2}} \end{equation} (3)

R can be monitored from the QCM instrument and the value of the constant (k) is estimated by comparing R before and after the injection of the electrolyte solution. Then, we can calculate (ηρ)^1/2 and by substituting it for Eq. 2, ΔF_ηρ is estimated. Finally, ΔF_m (= ΔF_total − ΔF_ηρ) is substituted for Eq. 1 and we can obtain Δm.

3. Results and Discussion

As already known, Grignard reagent achieves the Mg plating/stripping reaction with excellent coulombic efficiency, so we used it as the standard system. Figure 1a shows the cyclic voltammogram (CV) of Au substrate in 1.0 mol dm⁻³ EtMgBr/THF. The redox current corresponding to the Mg plating/stripping reaction is clearly seen without any clear overvoltage and the coulombic efficiency is 87 %, which shows the good accordance with the previous reports.^11,13 Figure 1a also shows the frequency responses of the substrate. During the metal plating reaction, the metal ion is reduced and deposits on the electrode, so the concentration of the metal ion in the vicinity of the electrode surface generally decreases, leading the increase of ΔF_ηρ. However, ΔF_ηρ in Fig. 1a does not increase clearly. This will be connected with the specific rate determining step of the Mg plating process in Grignard reagent. The ionic conductivity of the Grignard reagent used is as small as the order of 10⁻⁵ S cm⁻¹, therefore, the kinetics of the Mg plating reaction is thought to be influenced by the migration process, rather than the charge transfer or the diffusion process.¹⁴ In the case, the concentration of Mg²⁺-ion will not be varied so much and as the result, the variation of η or ρ will also be small. On the other hand, during the Mg stripping process, Mg²⁺-ion is forced to be supplied from the Mg metal electrode, the increase of ηρ, i.e., the decrease of ΔF_ηρ is confirmed.

Figure 1.

(a) CV of Au substrate in 1.0 mol dm⁻³ EtMgBr/THF and the corresponding frequency response. Scan rate was set to be 5 mV sec⁻¹. (b) The total variation of the charge and the mass change obtained from QCM.

The mass change obtained from QCM and the charge (ΔQ) in CV measurement are shown in Fig. 1b. The total mass change during the reduction process was 6.99 µg and the corresponding total charge was 46.94 mC. From the result, we calculate m.p.e. to be 14.37 g mol⁻¹. The result is similar to the theoretical value (12.15 g mol⁻¹) and therefore, we can confirm that our EQCM measurement works properly.

Figure 2a shows the CV and the corresponding frequency responses in 0.5 mol dm⁻³ Mg[TFSA]₂/diglyme. Contrary to the Grignard reagent system, the mass change obtained from EQCM in Mg[TFSA]₂/diglyme is quite different. Firstly, ΔF_ηρ shows the general behavior, i.e., the positive shift during the Mg plating reaction and the negative shift during the Mg stripping reaction, as described above. The negative shift around at 0 V will be concerned with the slight Mg stripping reaction inserted in the figure. On the other hand, ΔF_m curiously shows the positive shift although the reduction current appears and then, ΔF_m decreases during the positive scan. This obviously means that QCM shows the irregular behavior in Mg[TFSA]₂/diglyme solution: the apparent mass decrease occurs during the Mg plating reaction and the mass increase during the Mg stripping reaction, as shown in Fig. 2b.

Figure 2.

(a) CV of Au substrate in 0.5 mol dm⁻³ Mg[TFSA]₂/diglyme and the corresponding frequency response. Scan rate was set to be 5 mV sec⁻¹. (b) The mass change obtained from QCM.

The morphology and purity of the plated Mg greatly depends on the electrolyte solutions. In order to reproduce the morphology in CV measurements, we conducted the linear sweep voltammetry (LSV) in both solutions and compared the plated Mg by SEM-EDS. Figure 3 shows SEM images of Mg plating obtained in 1 mol dm⁻³ EtMgBr/THF and 0.5 mol dm⁻³ Mg[TFSA]₂/diglyme. In EtMgBr/THF solution, the substrate is densely covered with the Mg fine particles with the size of about 10 µm, as seen in Figs. 3a and 3b. In addition, EDS analysis shows that the purity of the plated Mg is over 90 % by atomic ratio. The result shows good accordance with the previous reports and it is the typical for Grignard reagent system.¹¹ On the other hand, Mg plating reaction takes place locally in Mg[TFSA]₂/diglyme solution and the particle size (over 50 µm) is much larger than that of EtMgBr/THF case. Besides, the atomic ratio of Mg is as small as 48 % and impurities such as F (1.8 %) and S (0.2 %) are confirmed. As mentioned before, Mg[TFSA]₂ salt is decomposed to form the surface film on Mg, resulting in the poor reversibility and large overvoltage. In order to investigate whether the surface film formation causes the irregular response of QCM or not, we conducted the following experiment. Firstly, Mg was plated in 1.0 mol dm⁻³ EtMgBr/THF solution and then, the solution was changed to 0.5 mol dm⁻³ Mg[TFSA]₂/diglyme and Mg stripping behavior was monitored by QCM. According to the procedure, Mg fine particles are densely plated on the substrate and their surface will be covered with the TFSA⁻ derived film. Therefore, the effect of the morphology will be excluded.

Figure 3.

SEM images of plated Mg (a) and (b) in 1.0 mol dm⁻³ EtMgBr/THF, and (c) and (d) in 0.5 mol dm⁻³ Mg[TFSA]₂/diglyme.

Figure 4a shows the potential profile of the substrate with the applied current density of 0.25 mA cm⁻² and the corresponding mass change during the Mg plating reaction (solid line) in 1 mol dm⁻³ EtMgBr/THF. The total amount of charge was set to obtain 5 µg of Mg and the theoretical mass change is also shown in the figure (dashed line). As seen in the figure, the mass change increases along the theoretical line and the total mass increase is 6 µg. From the result, we confirmed that QCM measurement worked properly. Figure 4b shows the open circuit potential (OCP) profile after Mg plating reaction. In EtMgBr/THF, OCP keeps 0 V and the result means that the surface of the plated Mg remains fresh. On the other hand, once the solution is changed to Mg[TFSA]₂/diglyme, OCP jumps up to about 1.2 V, which will indicate that the plated Mg was covered with the TFSA⁻ derived surface film and from the results, we confirmed the densely plated Mg metal with TFSA⁻ derived surface film was obtained. Figure 4c shows the LSV profile for Mg stripping reaction and the corresponding mass change. Similar to Fig. 2a, Mg stripping reaction needs the large over voltage more than 2 V. On the other hand, the response of QCM is quite different from that in Fig. 2a. Δm decreases corresponding to the oxidation current and the total mass decrease shows good accordance with the amount of plated Mg. Considering the surface film formation took place, the result does not necessarily mean that all Mg was completely stripped. However, we confirm the normal QCM response and therefore, we conclude that the irregular response of QCM will not be caused by the TFSA⁻ derived surface film.

Figure 4.

(a) Potential profile for Mg plating in 1.0 mol dm⁻³ EtMgBf/THF. Applied current density was set to be 0.25 mA cm⁻² and the corresponding mass change estimated from (solid) QCM and (dash) applied charge, (b) OCP change of plated Mg in 1.0 mol dm⁻³ EtMgBr/THF and after changing to 0.5 mol dm⁻³ Mg[TFSA]₂/diglyme, and (c) linear sweep voltammogram and the corresponding mass change in 0.5 mol dm⁻³ Mg(TFSA)₂/diglyme. Scan rate was 5 mV sec⁻¹.

The positive shift during mass increase process has been reported by some groups. For example, Pomorska et al. confirmed the positive shift in the adsorption process of the mesoporous TiO₂ coated with polyelectrolyte on the substrate with the self-assembled monolayer (SAM).¹⁵ In the case, the relatively large TiO₂ aggregates (ca. 1 µm) are connected to the substrate via SAM, i.e., the point contact between TiO₂ and the substrate is formed. Therefore, TiO₂ aggregates do not move by the oscillation of the substrate and restoring force is exerted. As the result, the oscillation speeds up and leads the positive shift of ΔF_m. According to the report, the irregular response of QCM will also be caused by the specific morphology of the plated Mg. In fact, the plated Mg aggregates are enough large of about 50 µm to cause the restoring force.

Here, we discuss the effect of alloy formation between Mg and Au substrate. If the alloy formation took place, the plated Mg strongly would connect with Au substrate and oscillate with the substrate. And as the result, the restoring force will be largely influenced. However, we think the alloy formation hardly occurs in the present study. In general, the electrochemical alloy formation takes place at higher electrode potential than the metal plating. In fact, Kwak et al. reported that the nucleation overpotential for Mg plating reaction on the Au-coated copper substrate decreased about 0.5 V compared with the bare copper substrate.¹⁶ On the other hand, in Fig. 2a, the onset potential for the Mg plating reaction is about −0.5 V, which is comparable to that of Pt (no alloy formation) substrate as shown in Fig. S1. From the result, we conclude that the alloy formation will not take place. The difference of Kwak’s result will be caused by the electrolyte solution. They use all phenyl complex/tetrahydrofuran, which is know to be stable against Mg metal. In the case, the Au substrate will be kept clean and the alloy formation will not be suppressed. On the other hand, we use Mg[TFSA]₂/diglyme solution and in the solution, TFSA⁻ decomposition also occurs as well as Mg plating reaction. Therefore, TFSA⁻ derived surface film forms on the Au substrate and it will suppress the alloy formation. Therefore, we conclude that the irregular behavior will not caused by Mg-Au alloy formation.

In order to elucidate the correlation between the morphology of the plated Mg and the response of QCM, we conducted the cross-sectional SEM observation. Figure 5a shows the potential profile and the corresponding mass change in 0.5 mol dm⁻³ Mg[TFSA]₂/diglyme with applied current density of 0.25 mA cm⁻² for 810 seconds. Surprisingly, the polarization is larger than that in EtMgBr/THF case (Fig. 4a), although the ionic conductivity of Mg[TFSA]₂/diglyme (10⁻³ S cm⁻¹) is much larger than EtMgBr/THF (10⁻⁵ S cm⁻¹). This will be because the active electrode area is largely different, depending on the electrolyte solutions. As confirmed in Figs. 3a and 3c, Mg covers the substrate densely in EtMgBr/THF, while only the localized plating is confirmed in Mg[TFSA]₂/diglyme. Therefore, the net surface area is much smaller in Mg[TFSA]₂/digmlye and the internal resistance will be enlarged unexpectedly. Anyway, the typical electrochemical behavior for the metal plating reaction, i.e., the voltage spike corresponding to the nucleation overvoltage and the following potential plateau of crystal growth is confirmed. On the other hand, the response of QCM does not work properly and no clear mass change is observed. Figures 5b and 5c show the top view of the plated Mg and bird’s-eye view of the etched Mg plating, respectively. Similar to the result of CV in Fig. 3, the localized Mg plating is obtained again. As seen from the cross-sectional image, the plated Mg metal is tree-like shape and its root is only 15 µm, although the size of the plating is estimated to be over 50 µm. Although the situation is not the point contact, the contact/projected area ratio is enough small and from the result, we conclude that the small projection area will cause the restoring force as well as TiO₂ particles mentioned above.

Figure 5.

(a) Potential profile for Mg plating in 0.5 mol dm⁻³ Mg[TFSA]₂/diglyme and the corresponding mass change. Applied current density was set to be 0.25 mA cm⁻². (b) Top view and (c) Cross-sectional SEM image of the plated Mg.

In order to confirm that the specific morphology causes the irregular response of QCM, we investigated the effect of the particle size on the response of QCM. In general, the critical radius of the plated metal crystal is inversely proportional to the applied overvoltage during the nucleation process.¹⁷ Therefore, we expect that the smaller and denser Mg plating will be obtained by applying the larger current density. Figure 6a shows the potential profile and the corresponding mass change for the current density of 10 mA cm⁻² for 20.3 seconds. Contrary to the result of 0.25 mA cm⁻² case, the mass change linearly increases with the applied charge, which indicates QCM works properly. The difference from the theoretical line will mean that the side reaction, i.e., the decomposition of TFSA⁻ also takes place. Note that the total applied charge was set to be identical with the 0.25 mA cm⁻² case discussed above, so the response of QCM will be mainly reflected by the morphology. In fact, as seen in Fig. 6b, the 50 µm sized crystal no longer appears and instead of that, 10 µm sized particles are distributed more densely. In addition, the cross-sectional image shown in Fig. 6c reveals that each plating is hemisphere shape and the contact/projected area ratio is almost 1. Therefore, the restoring force does not appear and QCM works properly. The results mean that we need to pay attention to the morphology of the plated Mg in order to investigate Mg plating/stripping reaction by EQCM.

Figure 6.

(a) Potential profile for Mg plating in 0.5 mol dm⁻³ Mg[TFSA]₂/diglyme and the corresponding mass change. Applied current density was set to be 10 mA cm⁻². (b) Top view and (c) Cross-sectional SEM image of the plated Mg.

4. Conclusions

The electrochemical Mg plating/stripping reaction was studied by EQCM method. In Mg[TFSA]₂/diglyme solution, QCM showed the apparent mass decrease during Mg plating reaction and the apparent mass increase during Mg stripping reaction. QCM method also did not work properly for the Mg plating by the small constant current density apply, while the normal response was observed in case of the large current density apply. The morphology of the plated Mg was greatly different by the applied current density. In the case of small current density apply, the tree shaped Mg was observed locally. On the other hand, Mg was plated densely and each plating was small and particulate shape in the case of the large current density apply. From the results, we conclude that the irregular QCM response was caused by the specific Mg morphology: the oscillation of the quartz substrate was accelerated by the plated Mg with small contact/projected area ratio. The conclusion will tell us the importance of the effect of the morphology on EQCM for the metal plating/stripping reaction.

Acknowledgment

This study was supported by KAKENHI (18K05293) of the Japan Society for the Promotion of Science (JSPS). The authors are also grateful to Dr. Y. Yamamoto at Nagoya University for the cross-sectional SEM measurements.

Data Availability Statement

The data that support the findings of this study are openly available under the terms of the designated Creative Commons License in J-STAGE Data at https://doi.org/10.50892/data.electrochemistry.19588411.

CRediT Authorship Contribution Statement

Fumihiro Sagane: Conceptualization (Lead), Data curation (Equal), Formal analysis (Equal), Funding acquisition (Lead), Project administration (Lead), Resources (Lead), Writing – original draft (Lead), Writing – review & editing (Lead)

Akiya Muramatsu: Data curation (Equal), Investigation (Equal)

Conflict of Interest

The authors declare no conflict of interest in the manuscript.

Funding

Japan Society for the Promotion of Science: 18K05293

Footnotes

F. Sagane: ECSJ Active Member

References

• 1)  T. Fukutsuka, K. Asaka, A. Inoo, R. Yasui, K. Miyazaki, T. Abe, K. Nishio, and Y. Uchimoto, Chem. Lett., 43, 1788 (2014).
• 2)  Y. Orikasa, T. Masese, Y. Koyama, T. Mori, M. Hattori, K. Yamamoto, T. Okado, Z. D. Huang, T. Minato, C. Tassel, J. Kim, Y. Kobayashi, T. Abe, H. Kageyama, and Y. Uchimoto, Sci. Rep., 4, 5622 (2014).
• 3)  S. Y. Ha, Y. W. Lee, S. W. Woo, B. Koo, J. S. Kim, J. Cho, K. T. Lee, and N. S. Choi, ACS Appl. Mater. Interfaces, 6, 4063 (2014).
• 4)  F. Sagane, K. Ogi, A. Konno, and K. Kanamura, J. Electrochem. Soc., 168, 120528 (2021).
• 5)  F. Sagane, M. Matsui, and K. Kanamura, J. Electrochem. Soc., 169, 030517 (2022).
• 6)  N. N. Rajput, X. Qu, N. Sa, A. K. Burrell, and K. A. Persson, J. Am. Chem. Soc., 137, 3411 (2015).
• 7)  A. Baskin and D. Prendergast, J. Phys. Chem. C, 120, 3583 (2016).
• 8)  N. Serizawa, Y. Katayama, and T. Miura, J. Electrochem. Soc., 156, D503 (2009).
• 9)  K. Naoi, M. Mori, Y. Naruoka, W.-M. Lamanna, and R. Atanasoski, J. Electrochem. Soc., 146, 462 (1999).
• 10)  H. Sonoki, M. Matsui, and N. Imanishi, J. Electrochem. Soc., 166, A3593 (2019).
• 11)  D. Aurbach, A. Schechter, M. Moshkovich, and Y. Cohen, J. Electrochem. Soc., 148, A1004 (2001).
• 12)  N. Sasaya, M. Matsumiya, S. Murakami, K. Nishihata, and K. Tsunashima, Electrochim. Acta, 194, 304 (2016).
• 13)  D. Aurbach, Z. Lu, A. Schechter, Y. Gofer, H. Gizbar, R. Turgeman, Y. Cohen, M. Moshkovich, and E. Levi, Nature, 407, 724 (2000).
• 14)  M. Matsui, J. Power Sources, 196, 7048 (2011).
• 15)  A. Pomorska, D. Shchukin, R. Hammond, M.-A. Cooper, G. Grundmeier, and D. Johannsmann, Anal. Chem., 82, 2237 (2010).
• 16)  J. H. Kwak, Y. Jeoun, S. H. Oh, S. Yu, J.-H. Lim, Y.-E. Sung, S.-H. Yu, and H.-D. Lim, ACS Energy Lett., 7, 162 (2022).
• 17)  F. Sagane, K.-i. Ikeda, K. Okita, H. Sano, H. Sakaebe, and Y. Iriyama, J. Power Sources, 233, 34 (2013).