Masahiro
Shimizu
*,
Atsuhito
Nakahigashi
and
Susumu
Arai
Department of Materials Chemistry, Faculty of Engineering, Shinshu University, 4-17-1 Wakasato, Nagano, 380-8553, Japan. E-mail: shimizu@shinshu-u.ac.jp; Fax: +81-26-269-5627; Tel: +81-26-269-5627
First published on 27th July 2021
In the development of rechargeable Mg-ion batteries which are not limited by resource constraints, studies on negative electrode materials have been concentrated on efficient Mg-deposition/stripping rather than on insertion/extraction-type active materials, driven by the extremely high theoretical capacity of Mg metal (2205 mA h g−1). This work re-examined the potential of graphite, which is overlooked in electrochemical tests using a two-electrode type cell due to a large overpotential during sluggish Mg-deposition/stripping at the counter electrode caused by the passivation layer. The formation of a graphite intercalation compound (GIC) with a stage structure was demonstrated by the continual application of a constant current without considering the cut-off voltage to eliminate the detrimental impact of the counter electrode, although the intercalant was solvated Mg-ions. The GIC formed during the charging process has a blue tint just like a GIC synthesized by a vapor method. Although there is still issue with the large polarization during the deintercalation of solvated Mg ions, a reversible capacity of approximately 200 mA h g−1 could be achieved in the galvanostatic charge/discharge tests with a current density of 7.44 mA g−1. The results should facilitate future research and development of graphite as a negative electrode material.
The solvation environments of the Mg ions in electrolyte solutions depending on salt concentration was analyzed with a Raman spectroscopy system (LabRAM HR Evolution; HORIBA, Ltd) using the 532 nm line of a diode-pumped solid-state laser at room temperature. We tightly sealed the electrolyte solution in a quartz cell in an argon-filled glove box to prevent exposure to water vapor.
Natural graphite powder (average particle size: 10 μm; SEC Carbon, Ltd) and polyvinylidene difluoride (PVDF) were mixed with N-methyl-2-pyrrolidone (NMP) at a weight ratio of 90/10 wt%. The resulting slurry was uniformly cast onto a Ti current collector using a doctor blade and then dried at 120 °C under vacuum for 5 h. The loaded mass of the active material and electrode thickness were approximately 1.8 mg cm−2 and 20 μm. The electrochemical Mg2+ storage behavior of the graphite was studied using a three-electrode electrochemical cell consisting of a working electrode (∅ 10 mm), a 0.5 mm-thick Mg alloy plate (Mg: 96%, Al: 3%, Zn: 1%, AZ31) as the counter electrode, a reference electrode, an electrolyte, and a glass fiber filter (Whatman GF/A) as the separator. The Mg alloy plate (as the counter electrode) was polished with sandpaper (#120) to remove the surface oxidation layer prior to cell assembly. Since the potential of Mg is known to be unstable in some organic electrolytes due to surface passivation,3,34 we used Ag wire soaked in 0.1 M AgNO3–0.1 M tetrabutylammonium perchlorate/acetonitrile (TBAP/AN) as the reference electrode, which was separated from the main electrolyte by porous Vycor glass (Fig. S1, ESI†). The potential of Ag/Ag+ was converted to the potential of ferrocene/ferrocenium (Fc/Fc+). The electrolyte preparation and cell assembly were carried out within a purge-type glove box filled with an argon atmosphere from which oxygen and water (a dew point below −75 °C) had been removed. For electrochemical impedance spectroscopy, two identical graphite composite electrodes were charged and discharged to assemble a symmetric cell. The impedance analysis was conducted using the symmetric cells in the frequency range of 100 kHz–100 mHz with an amplitude of 5 mV.
Structural changes in the graphite and the formation of the graphite intercalation compound (GIC) under galvanostatic polarization at a current density of 7.44 mA g−1 (0.01C) were tracked using transmission electron microscopy (TEM; HD2300A, Hitachi, Ltd) and X-ray diffractometry (XRD) (SmartLab, Rigaku). In the XRD measurements, the electrochemical cell after the application of a constant current was disassembled, and then the graphite electrode covered with a Kapton film inside an argon-filled glove box. Except when measuring the electrochemical potential of graphite electrodes, two-electrode cells were used to structurally analyze the GIC.
![]() | (1) |
![]() | (2) |
![]() | (3) |
As mentioned earlier, in the case of using a two-electrode type cell without considering the potential of the Mg foil as a counter electrode, it cannot be denied that the function of graphite as a negative electrode material has been overlooked. Structural changes undergone by graphite were carefully followed by ex situ XRD at a constant applied current, until the reductive capacity exceeded 1000 mA h g−1 (which corresponds to 1.34 mol Mg2+ per C6 if the all capacity was expended only in the charge transfer reaction) without a cut-off voltage; that is, under conditions in which the stripping reaction on the Mg counter electrode proceeded sufficiently (Fig. 2a). The graphite exhibited almost no structural change at a capacity of 50 mA h g−1, although the intensity of the 002 diffraction peak of graphite was slightly lower compared to that of the original material. A new broad diffraction peak was observed at 10° at a capacity of 100 mA h g−1; the intensity of this peak became remarkable at a capacity of 150 mA h g−1, while the 002 peak of the graphite was significantly less intense and new diffraction peaks at 20.4°, 25.52°, and 30.56° were observed. A new diffraction peak was observed at 15.0° when the capacity was 500 mA h g−1 or greater. The absence of peaks that correspond to Mg deposits in the XRD patterns, or the blue tint of graphite, which is similar to that of the vapor-phase-synthesized ternary Mg/ethylenediamine–graphite intercalation compound (GIC) reported by Lerner et al.,28,29 suggest that the observed diffraction peaks are due to the formation of a ternary GIC containing DMF as solvation molecules; i.e., Mg(DMF)x–GIC. The diffraction peaks located at 25.52° and 30.56° of the Mg(DMF)x–GIC, which are prominent at 250 mA h g−1, originate from the splitting of the XRD peak caused by the periodic sequence of graphite and intercalant [Mg(DMF)x]2+ planes. When the former and later peaks are assigned as the (00n) and the (00n + 1) indices of stage n and n + 1 GIC structures, the value of n can be calculated with the following equation:42,43
![]() | (4) |
The value of n was calculated to be 4.977, which was labeled as a stage-5 GIC. Consequently, the diffraction peaks at the position of 2θ = 10.1°, 15.0°, 20.4°, 25.52°, and 30.56° are indexed to (002), (003), (004), (005), and (006), respectively, using the above equation. Using the 003- and 005-diffraction angles of the GIC and the 002- and 004-diffractions of graphite, the interlayer space with the intercalant (gallery height) in the c-direction was determined to be 10.87 Å, which is equivalent to 3.24 times the distance (3.35 Å) between graphene sheets devoid of intercalants. Given this significantly extended distance, the intercalant is most likely to be solvated ions of [Mg(DMF)x]2+, rather than Mg ions alone, as opposed to the bare Li ions in a typical electrolyte.26 In addition to this, in light of the crystal and molecule structure of Mg(ClO4)2·DMF analyzed by X-ray diffractometry,44 the solvated Mg ions as the intercalant remain in the electrolyte bulk composition (0.5 M) and are likely to be accompanied by six solvent molecules, that is, the intercalant should be [Mg(DMF)6]2+. Meanwhile, it is noteworthy that Mg neither electrodeposits on the graphite composite nor on the Ti foil in the electrolyte used in this study, even though the voltage of the electrochemical cell was below 0 V and the applied capacity exceeded 250 mA h g−1 (Fig. S5, ESI†). Thus, even considering the possibility that not all of the current capacity was consumed in the GIC formation, the obtained results lead us to conclude that [Mg(DMF)6]2+ intercalates into the interlayers to form ternary GICs with a stage structure.
Cyclic voltammetry with an Ag/Ag+ (an Ag wire in 0.1 M AgNO3–0.1 M TBAP/AN) reference electrode at a sweep rate of 1 mV s−1 was used to study the reversibility of the intercalation/deintercalation of [Mg(DMF)6]2+ into/from graphite interlayers (Fig. 2b). Note that the potential of Ag/Ag+ was converted to ferrocene/ferrocenium (Fc/Fc+) in the electrochemical measurements. The current response observed from −2.32 to −2.68 V vs. Fc/Fc+ during sweeping to the cathodic side is attributed to the intercalation of [Mg(DMF)6]2+ to form GICs. The rapid increase in the reduction current below −2.71 V should come from the electrolyte decomposition (Fig. S6, ESI†). The anodic response between −2.14 to −1.51 V and the oxidative current at the broad potential ranges between −0.2 to 0.7 V should be due to the stepwise deintercalation of [Mg(DMF)6]2+ to return to the original graphite structure. At higher potential range, the deintercalation reaction should be accompanied with the oxidative decomposition of the electrolyte on the graphite composite electrode (Fig. S6, ESI†). TEM observation reveals that the thickness of the surface layer is approximately ∼4 nm and it is believed to be composed of inorganic/organic compounds that mainly contain Mg and F (Fig. S7 and S8, ESI†). After the first cycle, the pair of peaks corresponding to the intercalation/deintercalation of the solvated Mg-ions were clearly observed. Focusing on the ternary GIC formed at a capacity of 250 mA h g−1, in which the diffraction peaks of the GIC were prominent compared with those of graphite, the XRD patterns of the graphite electrode were remarkably unchanged as capacity was further increased (Fig. 2a). We monitored the charge–discharge behavior of graphite as a negative electrode material for Mg-ion batteries at a fixed capacity level of 250 mA h g−1 and a constant current density of 7.44 mA g−1 (Fig. 2c). The potential of the graphite composite electrode was −2.6 V vs. Fc/Fc+ when the capacity reached 250 mA h g−1. Two observed potential plateaus from −2.37 to −2.65 V are probably derived from the stepwise formation of ternary Mg(DMF)x–GICs. These plateaus were also detected at the second cycle at the same potentials as in the first cycle. In the desolvation-type reactions such as the intercalation of Li+ alone into graphite interlayers, the range of operating potentials is relatively narrow, whereas the co-intercalation reaction proceeds at wide operating potentials.45 Assuming that the intercalation of Mg2+ alone into graphite interlayers is the desolvation type reaction as reported by God et al.31 and comparing with the results obtained in this work, the range of operating potential in the intercalation/deintercalation of the solvated Mg2+ seems to be wide. This also suggests that the charge/discharge reactions in this case proceed as the co-intercalation/deintercalation type. However, the graphite composite suffered from a large potential hysteresis during discharging. Possible causes for the observed large overpotential are one or more of the following: (i) the ternary Mg(DMF)x–GIC is electrochemically unstable and decomposes during charge–discharge processes, (ii) the surface layers formed by the electrolyte decomposition have a lack of good Mg-ion conductivity, and (iii) the solvated Mg ions inside the graphite layers diffuse extremely slowly. In fact, Calandra and Mauri argued using the density functional theory calculation that MgC6 is energetically unstable against phase separation in Mg and graphite.46 We therefore tracked the structural change in graphite to examine whether or not the capacity of 250 mA h g−1 leading up to the completion of the discharge was consumed for the deintercalation of [Mg(DMF)6]2+.
Fig. 3 exhibits XRD patterns and Raman spectra of the graphite composite electrodes with galvanostatic charge–discharge during the first cycle. The Raman spectrum of graphite showed an additional band at 1599 cm−1 in the charged state at a capacity of 250 mA h g−1, in addition to the E2g mode47 (G band, sp2 graphitic, 1581 cm−1), which is due to the formation of GICs; the graphite electrode had expanded at least 1.33 times by this stage (Fig. 4a–c), although the exact expansion ratio could not be determined due to the relatively sparse electrode mixture prior to charging (Fig. S9, ESI†). Considering the Mg ion radius of ∼72 pm (with coordination number of 6) that is close to the Li ion (∼76 pm) and the two-electron reaction involved in Mg2+,48 the expansion ratio of the electrode in the charged state is great. This is because of the larger size of the intercalant, [Mg(DMF)6]2+, compared to Li+ alone and Mg2+ alone. During the deintercalation process, the 005 diffraction of the GIC had almost disappeared immediately after a discharge capacity of 50 mA h g−1 had been reached, although the position of the 002 of the graphite was not completely returned to the original position as before charging (Fig. S10, ESI†). It suggests that the solid solution with solvated Mg ions intercalated between the graphite interlayers to the extent that no stage structure is formed. The diffraction position gradually shifted to a higher angle with increasing the discharge capacity, and graphite regained its original structure when the capacity reached to 200 mA h g−1. As well as this, the GIC-derived band disappeared at the same time as the G band intensified in the Raman spectrum, that is, the capacity was delivered based on the reversible intercalation/deintercalation of [Mg(DMF)6]2+. At the fully discharged state, the electrode became dense and the thickness resulted in 9.9 μm from 18.34 μm in which the graphite included the intercalants. The phenomenon also indirectly reveals that the formation of the GIC is reversible. Similar changes in the structure and the electrode thickness of the graphite were observed at high current densities, indicating that the diffusion of [Mg(DMF)x]2+ is not fatally sluggish within the host material (not shown here) and is probably not the cause of the large polarization observed during the discharge process. In the case of Li+, which has an ionic size similar to that of Mg2+, the activation energy of the intercalation of the solvated ion is lower compared with that in the intercalation of ion alone accompanied with desolvation and the rapid charge/discharge performance is superior based on the smooth charge transfer reactions though a reversible capacity is reduced.45,49 We plan to study the activation energy of the charge transfer reaction and the diffusion coefficients of solvated Mg2+ inside the graphite structure. Meanwhile, the integrated Raman intensity ratio (ID/IG),47,50 one of the parameters on the crystallinity of carbonaceous material, resulted in 0.43 from 0.70 after the charge/discharge. The TEM images of pristine graphite displayed highly ordered graphene layer stacking with a d-spacing of 3.35 Å, whereas the randomly stacked structure was recognized after the first galvanostatic charge/discharge cycle, as expected based on the Raman spectra (Fig. 4d and e). On the basis of these results, it is plausible to assume that the large polarization observed at the graphite composite electrode during the discharge process is caused by the electrochemical instability of the ternary Mg(DMF)x–GIC and/or the lack of good Mg-ion conductivity through the surface layers formed by electrolyte decomposition. We are currently investigating the details of these causes, with attention given to the binder as well.
It is very interesting to see if the electrochemical formation of GICs achieved in this work also proceeds in a Grignard reagent solution, a typical electrolyte used for efficient Mg-deposition/stripping.51 The Grignard reagent solution of 1 M CH3MgBr/THF was applied to a two-electrode type cell with a graphite composite electrode and Mg alloy as a counter electrode (Fig. 5). In the Grignard electrolyte, the voltage plateau, which corresponds to the Mg deposition, appeared at around 0 V, and there was no change in the diffraction pattern of graphite. The electrolyte of 0.5 M Mg(TFSA)2/DMF enables the formation of GICs, whereas the electrolyte of 1 M CH3MgBr/THF induced the generation of only metallic Mg without the GIC formation. These results reveal that GIC formation depends greatly on the selection of the Mg salt and the electrolyte solvent.
Finally, we performed charge/discharge cycling tests at a current density of 7.44 mA g−1 (0.01C) in the potential range from 0 to −2.65 V vs. Ag/Ag+ (corresponding to −0.06 to −2.71 V vs. Fc/Fc+). Fig. 6a shows the charge/discharge profiles of the graphite composite electrode in 0.5 M Mg(TFSA)2/DMF using a three-electrode type cell. In the first cycle, charge and discharge capacities were 574 and 178 mA h g−1, respectively. The low coulombic efficiency (31%) should be due to the electrolyte decomposition below −2.5 V and the solvated ions that cannot be completely released (Fig. S6, ESI†). We conducted the electrochemical impedance measurements using symmetric cells consisting of two identical graphite composite electrodes which were charged and discharged (Fig. 6b). At the potential of −1.03 V vs. Fc/Fc+ before the charging, one semicircle and a vertical line were detected at a high frequency region, which are assigned to the contact resistance between active materials and the dominant capacitive behavior of the electric double layer formed at the electrode/electrolyte interface, respectively, as typically observed in a Nyquist plot in electric double layer capacitors.52 On the other hand, at the potential of −2.48 V, one semicircle and a slope were confirmed at high and low frequency regions, respectively. Since the top frequency of the semicircle is one order of magnitude lower than that observed before the charging and the electrodes were charged to the potential at which solvated Mg ions were intercalated, it is indicated that the semicircle is mainly the charge transfer resistance associated with the co-intercalation (Fig. S11, ESI†). The slope should be attributed to the diffusion of solvated Mg ions inside the graphite. As clearly shown in the Nyquist plot of the symmetric cell consisting of graphite electrodes charged to −2.6 V corresponding to a capacity of ca. 200 mA h g−1, the diameter of the semicircle decreased with the charging potential. This also supports that the semicircle originates from the charge transfer reaction.49 The semicircle remained even at the upper cut-off potential of −0.06 V, suggesting that the solvated ions are not completely deintercalated from the graphite interlayers. This is also consistent with the lower reversibility in the first cycle (Fig. 6a). At the second cycle, a reversible capacity reached to 194 mA h g−1 and then remained around 180 mA h g−1. The improvement in the reversibility after the second cycle is probably due to the formation of surface layers on graphite that suppresses further electrolyte decomposition while working as the Mg-ion conductor, as observed in the Li-ion batteries system. However, after the end of the 4th cycle (corresponding to ca. 14 days), the cut-off potential in the charge/discharge test could no longer be controlled. Although the reference electrode with a double-junction structure is separated from the main electrolyte by Vycor glass, the electrolyte inside the reference electrode seeped into the main electrolyte and finally disappeared. This is thought to be the reason why the potential did not reach the upper cut-off during the 5th discharging process. The phenomenon was also observed in the reproductive experiments, and it was found that the reference electrode functions for about 14 days under the current conditions. We are investigating the conditions which enables the long-term cycling tests independent of the reference electrode.
Footnote |
† Electronic supplementary information (ESI) available: Raman spectra of Mg(TFSA)2/DMF; cyclic and liner sweep voltammograms; survey XPS spectra of graphite electrodes; STEM and TEM images of graphite after electrochemical tests, impedance spectra of graphite electrodes during galvanostatic charge/discharge tests (PDF). See DOI: 10.1039/d1cp02895b |
This journal is © the Owner Societies 2021 |