Manufacturing and characterization of magnesium alloy foils for use as anode materials in rechargeable magnesium ion batteries

Journal of Power Sources 367 (2017) 138e144

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Manufacturing and characterization of magnesium alloy foils for use as anode materials in rechargeable magnesium ion batteries Daniel Schloffer a, Salar Bozorgi b, Pavel Sherstnev b, 1, Christian Lenardt c, Bernhard Gollas a, * a b c

Graz University of Technology, Institute for Chemistry and Technology of Materials, 8010 Graz, Austria AIT Austrian Institute of Technology, LKR Leichtmetallkompetenzzentrum Ranshofen GmbH, 5282 Ranshofen, Austria VARTA Micro Innovation GmbH, 8010 Graz, Austria

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


Thin foils of magnesium alloyed with zinc and gadolinium were produced.
The solid solutions are have a much better workability than pure magnesium.
Their initial degree of surface passivation depends on the alloying elements.
Zinc and gadolinium hardly affect the electrochemical behavior in APC electrolyte.
The alloy foils can be used as anodes in rechargeable magnesium ion batteries.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 8 June 2017 Received in revised form 23 August 2017 Accepted 17 September 2017

The fabrication of thin foils of magnesium for use as anode material in rechargeable magnesium ion batteries is described. In order to improve its workability, the magnesium was alloyed by melting metallurgy with zinc and/or gadolinium, producing saturated solid solutions. The material was extruded to thin foils and rolled to a thickness of approximately 100 mm. The electrochemical behavior of Mg1.63 wt% Zn, Mg-1.55 wt% Gd and Mg-1.02 wt% Zn-1.01 wt% Gd was studied in (PhMgCl)2-AlCl3/THF electrolyte by cyclic voltammetry and galvanostatic cycling in symmetrical cells. Analysis of the currentpotential curves in the Tafel region and the linear region close to the equilibrium potential show almost no effect of the alloying elements on the exchange current densities (5e45 mA/cm2) and the transfer coefficients. Chemical analyses of the alloy surfaces and the electrolyte demonstrate that the alloying elements not only dissolve with the magnesium during the anodic half-cycles, but also re-deposit during the cathodic half-cycles together with the magnesium and aluminum from the electrolyte. Given the negligible corrosion rate in aprotic electrolytes under such conditions, no adverse effects of alloying elements are expected for the performance of magnesium anodes in secondary batteries. © 2017 Elsevier B.V. All rights reserved.

Keywords: Magnesium alloy anode Rechargeable magnesium battery Electrodeposition Metal dissolution Exchange current density Magnesium corrosion

1. Introduction * Corresponding author. E-mail address: [email protected] (B. Gollas). 1 Present address: Linhardt Altai, Bijsk, Podgornaya Str. 76, 659333, Russia. https://doi.org/10.1016/j.jpowsour.2017.09.062 0378-7753/© 2017 Elsevier B.V. All rights reserved.

Magnesium alloys have been used as active materials in primary batteries for military applications like sonobuoys or electric

D. Schloffer et al. / Journal of Power Sources 367 (2017) 138e144

torpedoes [1]. Recently, the pioneering work of Gregory et al. and Aurbach et al. [2,3] has raised the interest for magnesium as active material also in secondary batteries. Magnesium has a rather negative standard reduction potential (2.362 V vs. NHE) and the following properties make it superior to Li metal. It has a high energy density (3833 mAh cm3), a high natural abundance (5th most in the earth crust) [4], it reacts only very slowly with water and air at room temperature [5], and it does not form dendrites during electrodeposition [6,7]. The first prototype of a working magnesium ion battery with a magnesium metal anode and a cathode based on the chevrel phase Mo6S8 in the electrolyte Mg(AlCl2BuEt)2/THF was described by Aurbach et al. [8] However, before the rechargeable magnesium ion battery can compete with lithium systems, there are many hurdles to overcome. Except for the chevrel phase, there currently is no other insertion cathode material for magnesium ions known to have a useful cyclability. Unfortunately, the potential for the Mg ion intercalation/deintercalation reaction of the chevrel phase is rather low (formal potential of 1,14 V versus Mg). There are very few materials, which show an intercalation reaction with Mg ions at all, for example V2O5 and a-MnO2 [9,10]. Very recently, transition metal compounds with polyanions received some attention, however those based on iron, manganese and cobalt silicate showed no reversible intercalation reaction [11,12]. A further class of cathode materials that operate at low voltage, but have high theoretical capacities, are conversion type materials, for example elemental sulfur [13,14]. A comprehensive overview of cathode materials for the magnesium ion battery has recently been published [15]. For high voltage rechargeable magnesium ion batteries, electrolytes with a sufficiently large potential window vs. magnesium are required. Unlike in lithium ion batteries, simple salts like Mg(ClO4)2 or Mg(BF4)2 decompose on the surface of the magnesium metal anode and form a blocking layer that is impermeable for the Mg2þ ions [16]. A couple of electrolytes, which possess an anodic stability of about 3 V vs. magnesium on platinum have been reported to be suitable for high voltage batteries. Examples are the (PhMgCl)2-AlCl3/THF electrolyte (so-called APC electrolyte) [17], the 2:1 MgCl2-AlCl3/ DME electrolyte (so-called MACC electrolyte) [18], the 2:1 MgCl2Mg (TFSI)2/DME electrolyte [19] and the 1:2 Mg(HMDS)2-AlCl3/ diglyme electrolyte [13]. All of these electrolytes contain a relatively high concentration of chloride ions. On the one hand, it seems that a high chloride ion concentration improves the electrochemical reversibility of the magnesium electrodeposition and dissolution [20,21]. On the other hand, such electrolytes are not compatible with conventional current collector materials like copper or nickel due to pitting corrosion [22]. There are only few halogen-free electrolytes, for example the magnesium borohydride salt (Mg(BH4)2 dissolved in DME or THF or the magnesium carborane salt (Mg(CB11H12)2) dissolved in tetragylme [23,24]. The former has an anodic stability of only 1.7 V vs. magnesium, while the latter shows a superior anodic stability of up to 3.8 V vs. magnesium, but suffers from high precursor costs and a labor-intensive synthesis. The last major component of a magnesium ion battery is the negative electrode. There are several reports about metals that form alloys with magnesium. Bismuth and antimony as well as the alloys Bi1-xSbx were utilized and, except Sb, all materials show good cyclability at high C-rates [25]. Nanostructuring of the bismuth anodes further increased the capacity retention [26]. Other materials, which were shown to work, are tin [27], indium [28], indiumbismuth [29], and lead [30]. Although a reversible magnesiation of these materials was possible, they all suffer from a capacity drop at higher C-rates. In order to fully exploit its potential in terms of energy density, however, magnesium should be used as anode in its elemental form. Since pure magnesium has a hexagonal structure, its mechanical

139

workability at temperatures below 200  C is rather bad [31]. In order to satisfy the von Mises criterion for general deformation, at least five independent slip systems are required [32]. At room temperature only the basal slip system is available. At higher temperatures additional prismatic and pyramidal slip systems can be activated (Figs. S1 and S2) [33]. One way to overcome this problem is alloying of the magnesium to change its mechanical properties. Blake et al., who have investigated the alloys of magnesium with zinc in the concentration range from 0 to 6.9 wt%, found a maximum in ductility at a concentration of about 1.5 wt% of zinc due to a softening effect of the prismatic slip system [34,35]. One important factor affecting the ductility of magnesium is the texture. Gadolinium is known as texture modifier and therefore, different concentrations ranging from 0 to 4.65 wt% and their influence on the ductility were investigated by Stanford et al.. They found that at concentrations of about 1e1.5 wt% Gd, the texture strength is at a minimum and this correlates to an improvement in ductility. Yan et al. took advantage of these positive effects and investigated the ternary alloys Mg-1.20 wt% Zn-0.79 wt% Gd and Mg-2.26 wt% Zn-0.74 wt% Gd. Both alloys possess a good ductility and showed an excellent formability at room temperature [36]. Moreover, magnesium is an active metal, which means that it is generally the anode in any galvanic couple that plays a role in macrogalvanic (in contact with another metal) and microgalvanic (second phases) corrosion [37]. There are only very few studies, which report on the electrochemical behavior of magnesium alloys with respect to the alloying elements or the manufacturing process €che et al. investigated the corrosion effect of re[38,39]. Ho deposited iron on magnesium [40,41]. However, all of these studies have been carried out in aqueous electrolytes. Here, we report for the first time on the electrochemical behavior of the alloys Mg-1.63 wt% Zn (MgZn1.6), Mg-1.55 wt% Gd (MgGd1.6) and Mg-1.02 wt% Zn-1.01 wt% Gd (MgZn1Gd1) in APC electrolyte for potential use as negative electrodes in rechargeable magnesium ion batteries. 2. Experimental 2.1. Manufacturing of the alloy foils Pure magnesium (99.95%) and the respective amounts of the alloying elements zinc (99.93%) and gadolinium (99.98%) were melted at 700  C under protective argon atmosphere in a graphite crucible. The melt was then cast into a steel mold to produce cylindrical bolts with a diameter of 48.5 mm and a height of 140 mm. The bolts were homogenized at 400  C for 10 h and then extruded at a temperature of 380  C with an extrusion speed of 50 mm/s to produce thin foils with a thickness of about 1.5 mm. The foils were hot rolled followed by an annealing step at 380  C for 5 min. for several times, to achieve a final thickness of approximately 100 mm. Another batch was extruded at a temperature of 380  C with an extrusion speed of 50 mm/s to produce rods with a diameter of 6 mm. 2.2. Alloy characterization The composition of the alloys was determined by arc/spark spectrometry (SPECTRO Analytical Instruments Inc., SPECTROMAXx6). Inductively coupled plasma optical emission spectrometry (ICP-OES, SPECTRO Analytical Instruments Inc., GENESIS FEE) was used for determining gadolinium, because this element cannot be analyzed with the former technique. Depth profiles of the oxygen and carbon content of the alloys were obtained by glow discharge optical emission spectroscopy (GDOES, Horiba Jobin Yvon, GD Profiler 2).

140

D. Schloffer et al. / Journal of Power Sources 367 (2017) 138e144

in negative direction. Five cycles in the potential range between ± 1.0 V were recorded. Subsequently, the anodic limit was extended to 1.5 V to partially dissolve the electrode. Afterwards, the chemical composition of the electrolyte was investigated via ICPOES (SPECTRO Analytical Instruments Inc., Ciros Vision) to determine the concentration of the alloying elements in the solution. Galvanostatic cycling (GC) was also performed with a battery cycler at current densities of 0.15 mA/cm2 and 1 mA/cm2. The first cycle started in positive direction for 8.65 h and 1.13 h, respectively to ensure dissolution of about 3 mm of the alloy on each side of the electrodes. 5 cycles were recorded at a current density of 0.15 mA/ cm2 and 10 cycles were recorded at a current density of 1 mA/cm2. The deposits of the experiments as well as the pristine alloys were studied with a scanning electron microscope (SEM, ESEM Tescan 500 PA) equipped with an energy dispersive X-ray (EDX) detector (INCA x-act Oxford instruments). For a semi-quantitative analysis of the deposit, the EDX results were corrected from any nonmetallic species steming from the degradation of the electrolyte and compared to the pristine alloys. The interaction volume in SEM is about 1 mm3, consequently, the whole information in the EDX spectrum originates from the deposit [46].

X-ray diffraction (XRD) of the alloys was performed with a Bruker D8 Advance X-ray powder diffractometer and Cu-Ka radiation in the range of 20e130 2-theta at 295 K (0.02 /step). Rietveld refinement was carried out with HighscorePlus Software from Panalytical. The extruded rods were grinded (P#1200, P#4000, SiC, Struers GmbH) and then polished (MD MOL/CHEM with 0.04 mm OP-S suspension, Struers GmbH). The microstructure was revealed by etching with acetic picral etchant. The etchant consisted of 5 mL of deionized H2O, 3 g of picric acid, 2.5 mL of acetic acid in 50 mL of ethanol (96%) [42].The samples were immersed into the etchant with gentle agitation until the face turned brown. Then they were washed with ethanol and investigated under an optical microscope. 2.3. Electrolyte synthesis 0.25 mol L1 (PhMgCl)2-AlCl3 in tetrahydrofuran (THF) was used as electrolyte (all phenyl complex, APC). It was prepared through the reaction between PhMgCl and AlCl3 in a 2:1 M ratio in anhydrous THF in an argon-filled glove box (MBraun 150-B-G-II, <1 ppm H2O/O2) [17]. Before use, THF (Sigma Aldrich, anhydrous, inhibitor free, þ99.9%) was dried over night with molecular sieve (0.3 nm) to a water content of less than 10 ppm, which was confirmed by Karl Fischer titration. PhMgCl (Sigma Aldrich, 2.1 mol L1 in THF) was titrated according to Blumberg and Martin to determine the exact concentration of the Grignard reagent [43]. For the preparation of 20 mL of the electrolyte, 0.667 g (5 mmol) of AlCl3 (Strem chemicals, anhydrous, 99.99þ%) were added slowly to 15.3 mL of THF during rigorous stirring. Then, 4.76 mL (10 mmol) of PhMgCl were added slowly to the solution and the as prepared electrolyte was stirred for 18 h.

3. Results and discussion 3.1. Alloy characterization The chemical composition of the alloys determined by ICP-OES and arc/spark spectrometry can be found in Table 1. The concentration of trace metals is below 0.1 wt% except for thorium. The manufacturing process of the magnesium foils is illustrated in Fig. S5. The hot rolling step had to be repeated 10 times in order to achieve a thickness of 100 mm, which is considered a standard size for battery electrodes. The microstructure of the rods is represented in Fig. 1. The addition of gadolinium leads to grain refinement (b) compared to the pure magnesium metal (a). The addition of zinc results in dendritic solidification (c). It can be deduced from thermodynamic data, that the intermetallic phases MgZn and Mg5Gd are formed in the respective alloy (Fig. S6eS7). The alloying elements and the intermetallic phases might also account for the high number of corrosion pits after etching. A slightly finer grain can be observed in the alloy with both alloying elements, but also a higher number of corrosion pits. In order to understand how the alloying elements are incorporated in the magnesium lattice, XRD was performed with a focus on the lattice parameters a and c. Fig. S8 shows the diffraction patterns of pure magnesium and the three alloys. Magnesium crystallizes in the hexagonal crystal system with space group number 194 (P63/ mmc). From Rietveld refinement we obtained the lattice parameters a and c for the hexagonal unit cell (Table 2). In the case of gadolinium, the lattice constants increase, whereas for zinc the lattice constants decrease. If we consider that the alloying elements are incorporated into the magnesium lattice, this observation would be consistent with the atomic radii in the metals, which are 1.60 Å for magnesium, 1.34 Å for zinc, and 1.79 Å for gadolinium [47]. In the alloy MgZn1Gd1 a slight increase can be

2.4. Electrochemistry Electrochemical characterization was conducted at ambient temperature (20  C) in four-necked glass cells assembled in the glove box. The respective alloy was used as working electrode and as the two auxiliary electrodes. Pure magnesium ribbon (250 mm, 99.9%, Goodfellow GmbH) was used as reference electrode. Furthermore, cyclic voltammograms of ferrocene were measured in the APC electrolyte in order to provide an additional reference potential as recommended by IUPAC for non-aqueous systems (Figs. S3 and S4, Table S1) [44]. The formal potential of the ferrocenium/ferrocene redox couple is 2.51 V versus magnesium and can be used to calculate the potential of magnesium in different solvents as suggested by Gritzner [45]. Alloy foils with a thickness of about 600 mm and a circular geometry (d ¼ 12 mm, A ¼ 2.3 cm2) were used, because of their higher mechanical stability and easier handling. The surface of the magnesium and its alloys was grinded, polished (P#320, P#1200, P#4000, SiC, Struers GmbH) and washed with ethanol prior to use. Before the cells were assembled, the alloys were once again polished with P#4000 to remove any surface layer. Cyclic voltammetry (CV) was performed with a battery cycler (BioLogic, MPG-2) at a scan rate of 50 mV/s starting from the open circuit potential

Table 1 Chemical composition of the alloys MgZn1.6, MgGd1.6 and MgZn1Gd1 (wt%, balance Mg) after manufacturing as determined by arc/spark spectrometry. Material

Zn

Gd

Th

Al

Si

Mn

Ca

Fe

Ni

Cu

Y

Ce

Pb

Zr

Pr

1.55 1.01

0.121 0.177 0.158

0.0623 0.0166 0.0256

0.0056 0.0130 0.0196

0.0030 0.0110 0.0119

0.0070 0.0062 0.0046

0.0039 0.0060 0.0056

0.0012 0.0088 0.0063

0,0010 0,00066 0,00066

0.0089 0.0076 0.0078

<0.004 0.0042 <0.004

<0.004 0.0087 0.0076

<0.001 0.0023 0.0016

<0.0015 0.0592 0.0327

wt% MgZn MgGd MgGdZn

1.63 0.0086 1.02

D. Schloffer et al. / Journal of Power Sources 367 (2017) 138e144

141

Fig. 1. Microstructure of the extruded rods of pure Mg (a) and the alloys MgGd1.6 (b), MgZn1.6 (c), and MgZn1Gd1 (d) after etching with acetic picral etchant at 200-fold magnification.

Table 2 Lattice parameters a and c for pure magnesium and the magnesium alloys MgGd1.6, MgZn1.6 and MgZn1Gd1. Material

Lattice constant a/Å

Lattice constant c/Å

Mg [42] MgGd1.6 MgZn1.6 MgZn1Gd1

3.2092 3.21177 3.20755 3.21036

5.2105 5.21257 5.20568 5.21068

observed due to the stronger influence of gadolinium over zinc on the lattice constants. This impact on the lattice parameters can also be found in the literature, and is obvious when different concentrations of the alloying elements were examined. Correa investigated the range from 0 to 1.86 wt% zinc and observed a decrease of the lattice constant a from 3.2088 Å to 3.2054 Å and for c from 5.2111 Å to 5.2049 Å [48]. The same decrease can be observed for higher concentrations of zinc [49]. Unfortunately, for gadolinium no such data can be found in the literature. Therefore, we will instead focus our discussion on a similar rare earth element. Dysprosium has a good solubility in magnesium and forms a solid solution with magnesium like gadolinium does (Fig. S9). The atomic radius of Dy is 1.75 Å [47] and thus very similar to that of gadolinium. Miura et al. investigated the change of the lattice parameters in magnesium upon addition of up to 20 wt% dysprosium and observed an increase of the lattice parameter a from 3.210 Å to 3.229 Å and of c from 5.210 Å to 5.220 Å [50]. Consequently, we assume that both zinc and gadolinium are incorporated into the lattice and form a solid solution with magnesium. Magnesium is a reactive metal and hence its surface state is expected to have a significant influence on its electrochemical behavior. A depth profile of the carbon and oxygen content of the alloys was recorded by GDOES. The oxygen content of the three alloys is highest at the surface (Fig. 2). It is well known, that MgO and Mg(OH)2 are formed on the metal surface, when magnesium is exposed to humid air and even in the presence of the low-level

Fig. 2. Depth profiles of the oxygen content of MgZn1.6, MgGd1.6, and MgZn1Gd1 determined by GDOES.

traces of water and oxygen found in the glove box [5]. The concentration of oxygen decreases for all alloys below a value of 0.5 wt % after a depth of about 2 mm. The highest oxygen content can be found at the surface of the MgGd alloy, which also decreases over the largest depth of about 50 mm. Also the carbon content is highest at the surface and falls below 0.05 wt% after a depth of about 2 mm into the material (Fig. 3). The carbon can be traced back to the diffusion of CO2 from the air through the porous surface layer, resulting in the formation of MgCO3 [51,52].

3.2. Electrochemical characterization The first and the fifth voltammetric cycle of each magnesium alloy in APC-electrolyte are presented in Figs. 4 and 5. In the first cycle, the three alloys show different overpotentials for Mg deposition. MgZn has the highest overpotential of about 0.45 V versus Mg followed by MgGd with 0.3 V and MgGdZn with 0.2 V. After 5 voltammetric cycles of deposition and dissolution in the potential range between /þ 1.0 V, these overpotentials for Mg deposition

142

D. Schloffer et al. / Journal of Power Sources 367 (2017) 138e144

Fig. 3. Depth profiles of the carbon content of MgZn1.6, MgGd1.6, and MgZn1Gd1 determined by GDOES.

Fig. 4. Cyclic voltammograms of the alloys MgZn1.6, MgGd1.6 and MgZn1Gd1 in 0.25 M APC/THF electrolyte starting cathodically from the open circuit potentials 27 mV, 26 mV, and 4 mV, respectively, with v ¼ 50 mV/s, magnesium reference electrode and the respective alloy as auxiliary electrode. Electrode surface was 2.26 cm2. Shown is the 1st cycle of each alloy.

concentration of the alloying elements in Mg-Mn, Mg-Al and MgAl-Zn and demonstrated that low concentrations of alloying elements have almost no impact on the corrosion potential in an aqueous solution [53]. After extension of the anodic potential limit to 1.5 V vs. magnesium and further five cycles, the chemical composition analysis of the electrolytes revealed, that the alloying elements zinc and gadolinium are present in the solution (Table 3). The Tafel plots of the 5th cycle of the cyclic voltammograms of each alloy are displayed in Fig. 6. For the linear regression in the Tafel region, two limitations must be kept in mind. Within the range of ± 10 mV around the corrosion potential, the current depends linearly on the potential and the Tafel equation is not valid. The second limitation is, that the Tafel analysis is only applicable for charge transfer limited currents. At values larger than 5% of the diffusion limited current, mass transport also becomes significant and could lead to misinterpretation [54]. Within these limits, the Pearson correlation coefficient r of the linear regression in the Tafel plots of Fig. 6 is between 0.98 and 0.99. We also analyzed the linear region within ± 10 mV of the corrosion potential and found correlation coefficients close to 1 (the Tafel analysis and calculation of the current densities and transfer coefficients can be found in the supplementary information). The calculated exchange current densities j0 and transfer coefficients a of the backward scan of the fifth cycle can be found in Table 4. The exchange current densities calculated from the Tafel region and those calculated from the linear region are in good agreement and in the range between 5 and 45 mA/cm2 for the individual cells. Table 3 Chemical composition of the electrolyte after cyclic voltammetry determined by ICP-OES (mg/g). Material

Zn

Gd

mg/g MgZn MgGd MgGdZn

e 0.163 0.047

0.061 e 0.039

Fig. 5. Cyclic voltammograms of the alloys MgZn1.6, MgGd1.6 and MgZn1Gd1 in 0.25 M APC electrolyte starting cathodically from the open circuit potentials 27 mV, 26 mV, and 4 mV, respectively, with v ¼ 50 mV/s in negative direction, magnesium reference electrode and the respective alloy as auxiliary electrode. Electrode surface was 2.26 cm2. Shown is the 5th cycle of each alloy.

have largely diminished, except for the alloy MgZn1.6. The different overpotentials in the first cycle are attributed to the degree of passivation of the alloy surfaces. The Tafel plots of the 1st cycle are presented in Fig. S12. From the open circuit potential in cathodic direction a plateau can be observed for each alloy. Such plateaus are best known from Tafel plots in aqueous corrosion electrochemistry and indicate the passivation of the electrode surface. As seen in the cyclic voltammograms, the passivation is strongest for the MgZn1.6 alloy (smallest current and longest plateau) followed by MgGd1.6 and MgZn1Gd1. The corrosion potentials of the three alloys are almost the same at 5 mV vs. magnesium. Koo et al. varied the

Fig. 6. Tafel plot of the 5th voltammetric cycle of the alloys MgZn1.6, MgGd1.6 and MgZn1Gd1 in 0.25 M APC/THF electrolyte.

Table 4 Exchange current densities and transfer coefficients of MgZn1.6, MgGd1.6 and MgZn1Gd1 calculated from the Tafel region and the linear region of the 5th cyclic voltammogram. Material j0, j0,

2 Tafel region (mA/cm ) 2 linear region (mA/cm )

acathodic aanodic

MgZn

MgGd

MgGdZn

14.9 14.3 0.55 0.45

14.1 19.4 0.60 0.41

22.9 33.1 0.60 0.46

D. Schloffer et al. / Journal of Power Sources 367 (2017) 138e144

The mean value of two different cells with each alloy can be found in the first and second row of Table 3. Experimental data from the literature is quite rare, but in the system EMImAlI4/(d-MgI2)x (0  x  0.023) for example, Di Noto et al. found exchange current densities of the same order of magnitude [55]. The transfer coefficients were calculated independently from the slope of the anodic and cathodic branches assuming n ¼ 2. For the cathodic reaction acathodic lies above 0.5 and for the anodic branch, aanodic lies below 0.5, which means that the energy barrier of the transition state is not symmetrical. Therefore the change of the potential will affect the rate of the cathodic reaction more than that of the anodic reaction. Figs. 7 and 8 show the potential versus time during galvanostatic cycling experiments with current densities of 0.15 and 1 mA/ cm2, respectively. Immediately after the start of the experiments, the potential is highest, because of the passivation layer on the surface. Once this layer is broken up, the alloys show similar overpotentials to supply the set current. The overpotential needed for the MgZn alloy is slightly higher compared to the other alloys, which is consistent with the higher overpotential observed already in cyclic voltammetry. At both current densities, the overpotential for the deposition increases with the cycle number. The corresponding SEM images of the pristine state of the electrodes and after cycling can be found in the supplementary information (Figs. S13eS21). After cycling, a higher amount of oxygen, chlorine, and aluminum can be detected on the surface. In the brighter parts of the images, no aluminum can be detected by EDX, but the amount of oxygen is twice and the amount of chlorine is five times higher than before cycling. This can be attributed to electrolyte decomposition or dried up electrolyte residues [56].

143

Aurbach et al. have already shown that the impedance of magnesium electrodes increases during longer periods at open circuit potential in ethereal solutions and they attributed this to some kind of surface adsorption phenomena [57]. Wetzel et al. identified products of the corrosion reaction between the magnesium electrode and the electrolyte by X-ray photoelectron spectroscopy [58]. Therefore, we assume the impedance of the electrode increases due a corrosion reaction resulting in some degree of passivation of the surface. In Table 5 the chemical composition of the deposits is listed. After having detected the alloying elements zinc and gadolinium in the electrolyte, we were also interested in finding out, whether they are re-deposited. From the analysis of the pristine alloys, we conclude that it is possible to the detect the alloying elements by means of EDX within an error margin that is acceptable compared to the results from arc/ spark spectrometry. If we assume that only magnesium is deposited, no other metals should be detected. Not surprisingly, aluminum was found in every deposit. This could originate on the one hand from the electrodeposition of aluminum, which was already shown by Yoo et al. [8] and on the other hand, it may additionally stem from trapped and dried up electrolyte. More interesting is the fact, that we find the alloying elements zinc and gadolinium in the deposit. Therefore, the alloying elements obviously not only dissolve in the electrolyte, but are also readily be redeposited. In aqueous systems, the influence of such impurities is well known, because of their contribution to macrogalvanic and microgalvanic corrosion of the magnesium [40,59]. In the aprotic electrolyte used here, magnesium does not corrode at a significant rate. Consequently, the presence of small concentrations of alloying elements does not adversely affect the electrochemical behavior of magnesium in electrolytes for rechargeable batteries. 4. Conclusion

Fig. 7. Galvanostatic cycling of the alloys MgZn1.6, MgGd1.6 and MgZn1Gd1 in 0.25 M APC/THF electrolyte at a current density of 0.15 mA cm2. The alloys were used as working and auxiliary electrodes, magnesium as reference electrode. Electrode surface was 2.26 cm2.

The alloys Mg-1.63 wt% Zn, Mg-1.55 wt% Gd and Mg-1.02 wt% Zn-1.01 wt% Gd have been manufactured. Rietveld refinement of XRD data let us conclude that both zinc and gadolinium mainly form solid solutions with magnesium. The addition of these elements improved the workability of magnesium such that thin foils with a thickness of 100 mm could be produced for potential use as anodes in rechargeable magnesium ion batteries. The added elements have a negligible impact on the corrosion potential of the alloys in all-phenyl-complex/THF electrolyte. The exchange current densities and transfer coefficients show no significant change either. Although the alloys with gadolinium have the highest oxygen concentration at the surface under atmospheric conditions, the alloy MgZn1.6 shows the highest overpotential for the electrodeposition of magnesium. MgZn1.6 also showed the highest

Table 5 Chemical composition of the alloys MgZn1.6, MgGd1.6 and MgZn1Gd1 before and after galvanostatic cycling determined by EDX (wt%). Material

Zn

Gd

Al

wt%

Fig. 8. Galvanostatic cycling of the alloys MgZn1.6, MgGd1.6 and MgZn1Gd1 in 0.25 M APC/THF electrolyte at a current density of 1 mA cm2. The alloys were used as working and auxiliary electrodes, magnesium as reference electrode. Electrode surface was 2.26 cm2.

MgZn (before) 0.15 mA cm2 1 mA cm2

1.52 2.36 e

e e e

e 4.29 3.84

MgGd (before) 0.15 mA cm2 1 mA cm2

e e e

0.96 1.18 e

e 4.21 3.59

MgGdZn (before) 0.15 mA cm2 1 mA cm2

0.89 1.41 1.41

0.89 1.61 1.51

e 6.23 2.81

144

D. Schloffer et al. / Journal of Power Sources 367 (2017) 138e144

overpotential during galvanostatic cycling. It is concluded that MgZn1.6 is more strongly passivated in the APC/THF electrolyte than the other alloys. At cathodic potentials not only magnesium is deposited from the APC electrolyte, but also aluminum together with the alloying elements zinc and gadolinium that are dissolved during the anodic half-cycles. While the co-deposition of elements plays an important role in aqueous Mg corrosion and hydrogen evolution, its impact on the anode performance of a non-aqueous secondary Mg battery looks negligible. Acknowledgments The authors acknowledge funding by the Austrian Research and Promotion Agency and the Ministry for Transport, Innovation and Technology through the program 'Mobilit€ at der Zukunft' through grant no. 840457 (Project MagIC). We thank B. Bitschnau and F. Mautner for XRD analyses, and H. Wiltsche and M. Winkler for ICPOES measurements. We gratefully acknowledge the support of C. God, S. Koller, and M. Schmuck from VARTA Micro Innovation. Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.jpowsour.2017.09.062. References [1] R.F. Koontz, R.D. Lucero, Magnesium water-activated batteries, in: D. Linden, T.B. Reddy (Eds.), Handbook of Batteries, third ed., McGraw-Hill, New York, 2002, pp. 468e494. [2] T.D. Gregory, R.J. Hoffman, R.C. Winterton, J. Electrochem. Soc. 137 (1990) 775e780. [3] D. Aurbach, Z. Lu, A. Schechter, Y. Gofer, H. Gizbar, R. Turgeman, Y. Cohen, M. Moshkovich, E. Levi, Nature 407 (2000) 724e727. [4] J. Muldoon, C.B. Bucur, T. Gregory, Chem. Rev. 114 (2014) 11683e11720. [5] C. Chen, S. Splinter, T. Do, N. McIntyre, Surf. Sci. 382 (1997) L652eL657. [6] D. Aurbach, Y. Gofer, A. Schechter, O. Chusid, H. Gizbar, Y. Cohen, M. Moshkovich, R. Turgeman, J. Power Sources 97 (2001) 269e273. [7] C. Ling, D. Banerjee, M. Matsui, Electrochim. Acta 76 (2012) 270e274. [8] H.D. Yoo, I. Shterenberg, Y. Gofer, G. Gershinsky, N. Pour, D. Aurbach, Energy Environ. Sci. 6 (2013) 2265e2279. [9] P. Nov ak, R. Imhof, O. Haas, Electrochim. Acta 45 (1999) 351e367. [10] R. Zhang, X. Yu, K. Nam, C. Ling, T.S. Arthur, W. Song, A.M. Knapp, S.N. Ehrlich, X. Yang, M. Matsui, Electrochem. Commun. 23 (2012) 110e113. [11] C. Ling, D. Banerjee, W. Song, M. Zhang, M. Matsui, J. Mat. Chem. 22 (2012) 13517e13523. [12] X. Chen, F.L. Bleken, O.M. Løvvik, F. Vullum-Bruer, J. Power Sources 321 (2016) 76e86. [13] Z. Zhao-Karger, X. Zhao, O. Fuhr, M. Fichtner, RSC Adv. 3 (2013) 16330e16335. [14] Z. Zhao-Karger, X. Zhao, D. Wang, T. Diemant, R.J. Behm, M. Fichtner, Adv. Energy Mater 5 (2014) 1401155. [15] P. Saha, M.K. Datta, O.I. Velikokhatnyi, A. Manivannan, D. Alman, P.N. Kumta, Prog. Mat. Sci. 66 (2014) 1e86. [16] Z. Lu, A. Schechter, M. Moshkovich, D. Aurbach, J. Electroanal. Chem. 466 (1999) 203e217. [17] O. Mizrahi, N. Amir, E. Pollak, O. Chusid, V. Marks, H. Gottlieb, L. Larush, E. Zinigrad, D. Aurbach, J. Electrochem. Soc. 155 (2008) A103eA109. [18] C.J. Barile, E.C. Barile, K.R. Zavadil, R.G. Nuzzo, A.A. Gewirth, J. Phys. Chem. C 118 (2014) 27623e27630. [19] I. Shterenberg, M. Salama, H.D. Yoo, Y. Gofer, J. Park, Y. Sun, D. Aurbach, J. Electrochem. Soc. 162 (2015) A7118eA7128. [20] K.A. See, K.W. Chapman, L. Zhu, K.M. Wiaderek, O.J. Borkiewicz, C.J. Barile, P.J. Chupas, A.A. Gewirth, J. Am. Chem. Soc. 138 (2015) 328e337.

[21] R. Mohtadi, F. Mizuno, Beilstein J. Nanotechnol. 5 (2014) 1291e1311. [22] S. Yagi, A. Tanaka, Y. Ichikawa, T. Ichitsubo, E. Matsubara, J. Electrochem. Soc. 160 (2013) C83eC88. [23] R. Mohtadi, M. Matsui, T.S. Arthur, S. Hwang, Angew. Chem. Int. Ed. 51 (2012) 9780e9783. [24] O. Tutusaus, R. Mohtadi, T.S. Arthur, F. Mizuno, E.G. Nelson, Y.V. Sevryugina, Angew. Chem. Int. Ed. 54 (2015) 7900e7904. [25] A. Benmayza, M. Ramanathan, N. Singh, F. Mizuno, J. Prakash, J. Electrochem. Soc. 162 (2015) A1630eA1635. [26] Y. Shao, M. Gu, X. Li, Z. Nie, P. Zuo, G. Li, T. Liu, J. Xiao, Y. Cheng, C. Wang, Nano Lett. 14 (2013) 255e260. [27] N. Singh, T.S. Arthur, C. Ling, M. Matsui, F. Mizuno, Chem. Commun. 49 (2013) 149e151. [28] F. Murgia, E.T. Weldekidan, L. Stievano, L. Monconduit, R. Berthelot, Electrochem. Commun. 60 (2015) 56e59. [29] F. Murgia, L. Monconduit, L. Stievano, R. Berthelot, Electrochim. Acta 209 (2016) 730e736. [30] K. Periyapperuma, T.T. Tran, M. Purcell, M. Obrovac, Electrochim. Acta 165 (2015) 162e165. [31] M. Huppmann, Characterization of Deformation Mechanisms of the Hot Extruded Magnesium Alloys AZ31 and ME21 under Monotonic and Cyclic €t Berlin, Berlin, GerLoading Conditions, Ph.D. Thesis, Technische Universita many, 2011. [32] J.P. Hirth, J. Lothe, Theory of Dislocations, second ed., John Wiley & Sons, New York, 1982, p. 857. [33] E. Meza García, Influence of Alloying Elements on the Microstructure and Mechanical Properties of Extruded Mg-Zn Based Alloys, Ph.D. Thesis, Technische Universit€ at Berlin, Berlin, Germany, 2010. ceres, Mat. Sci. Eng. A 483 (2008) 161e163. [34] A. Blake, C. Ca ceres, A. Blake, Phys. Stat. Sol. 194 (2002) 147e158. [35] C.H. Ca [36] H. Yan, R. Chen, E. Han, Mat. Sci. Eng. A 527 (2010) 3317e3322. [37] G. Song, A. Atrens, Adv. Eng. Mater 5 (2003) 837e858. [38] N. Wang, R. Wang, C. Peng, Y. Feng, B. Chen, Corros. Sci. 64 (2012) 17e27. [39] Z. Hongyang, B. Pei, J. Dongying, J. Environ. Sci. 21 (2009) S88eS91. € che, C. Blawert, S.V. Lamaka, N. Scharnagl, C. Mendis, M.L. Zheludkevich, [40] D. Ho Phys. Chem. Chem. Phys. 18 (2016) 1279e1291. €che, R. Petrauskas, C. Blawert, M. Zheludkevich, Electrochem. [41] S. Lamaka, D. Ho Commun. 62 (2016) 5e8. [42] M.M. Avedesian, H. Baker, ASM Specialty Handbook: Magnesium and Magnesium Alloys, ASM International, Materials Park, Ohio, 1999, p. 350. [43] S. Blumberg, S.F. Martin, Tetrahedron Lett. 56 (2015) 3674e3678. [44] G. Gritzner, J. K uta, Electrochim. Acta 29 (1984) 869e873. [45] G. Gritzner, J. Mol. Liq. 156 (2010) 103e108. [46] D.B. Williams, C.B. Carter, Transmission Electron Microscopy, second ed., Springer US, New York, 2009, p. 775. [47] N. Wiberg, E. Wiberg, A. Holleman, Lehrbuch der Anorganischen Chemie, 102nd ed., Walter de Gruyter, Berlin, 2007, p. 2149. [48] A.M. Becerra Correa, Effect of Solute Elements on the Lattice Parameters of Magnesium, ME, McGill University, Montreal, Canada, 2006. [49] S. Miura, S. Imagawa, T. Toyoda, K. Ohkubo, T. Mohri, Mat. Trans. 49 (2008) 952e956. [50] S. Miura, S. Yamamoto, K. Ohkubo, T. Mohri, Mat. Sci. Forum 350 (2000) 183e190. [51] S. Feliu, A. Pardo, M. Merino, A. Coy, F. Viejo, R. Arrabal, Appl. Surf. Sci. 255 (2009) 4102e4108. [52] H. Oettel, H. Schumann, Metallografie: mit einer Einführung in die Keramografie, fifteenth ed., WILEY-VCH, Weinheim, 2011, p. 950. [53] J. Kim, S. Koo, Corrosion 56 (2000) 380e388. [54] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals and Applications, second ed., John Wiley & Sons, New York, 2001, p. 864. [55] F. Bertasi, F. Sepehr, G. Pagot, S.J. Paddison, V. Di Noto, Adv. Funct. Mater 26 (2016) 4860e4865. [56] C.J. Barile, R. Spatney, K.R. Zavadil, A.A. Gewirth, J. Phys. Chem. C 118 (2014) 10694e10699. [57] D. Aurbach, A. Schechter, M. Moshkovich, Y. Cohen, J. Electrochem. Soc. 148 (2001) A1004eA1014. [58] D.J. Wetzel, M.A. Malone, R.T. Haasch, Y. Meng, H. Vieker, N.T. Hahn, A. Golzhauser, J. Zuo, K.R. Zavadil, A.A. Gewirth, Appl. Mat. Interfaces 7 (2015) 18406e18414. [59] A. Südholz, N. Kirkland, R. Buchheit, N. Birbilis, ECS Solid State Let. 14 (2011) C5eC7.