Exposing Elusive Cationic Magnesium-Chloro Aggregates in Aluminate Complexes Through Donor Control

The cationic magnesium moiety of magnesium organohaloaluminate complexes, relevant to rechargeable Mg battery electrolytes, typically takes the thermodynamically favourable dinuclear [Mg2Cl3] form in the solid-state. We now report that judicious choice of Lewis donor allows the deliberate synthesis and isolation of the hitherto only postulated mononuclear [MgCl] and trinuclear [Mg3Cl5] modifications, forming a comparable series with a common aluminate anion [(Dipp)(Me3Si)NAlCl3] ̄. By pre-forming the Al-N bond prior to introduction of the Mg source, a consistently reproducible protocol is reported. Usage of the green solvent 2-methyltetrahydrofuran in place of THF in the context of Mg/Al battery electrolyte type complexes is also promoted.

Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal's standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains.

Dalton Transactions
www.rsc.org/dalton Introduction A key area of study for post-lithium ion batteries can be found in neighbouring group 2, specifically magnesium, 1 due to its considerably greater natural abundance, which as a consequence makes it more economically viable long-term. 2 Furthermore, the neutral metal/cation redox couple is a two-electron process, giving magnesium a higher volumetric capacity than lithium (3833 mA h cm -3 cf. 2062 mA h cm -3 ), while its high reduction potential of -2.37 V (vs SHE) is conducive to high energy density and high voltage batteries, provided other drawbacks, such as development of suitable electrolytes and cathodes, can be adequately surmounted. One main impediment of magnesium based electrolytes is that, unlike lithium, its neutral inorganic metal salts are incapable of reversibly conducting magnesium ions in aprotic solvents sufficiently, forming passivating films on the electrode surface, while the strong reducing nature of Grignard reagents gives them low anodic stability. A possible way to circumvent these problems is to move to a bimetallic ate complex 3 such as a magnesium aluminate, 4 which typically takes the form [Mg 2 Cl 3 ] + [R x AlCl 4-x ]¯ and can be generated from reaction of Lewis basic magnesium and Lewis acidic aluminium precursors. 5 Moreover, their solvent separated ion pairing enhances their conductivity. Inorganic only haloaluminates (x = 0) have been studied but suffer from poor solubility even in THF, 6 although more recent studies on this system have confirmed its enhanced oxidative stability. 7 The solubility can be increased by grafting an organo group onto the aluminium (e.g. R = Et, 8 Ph 7a, 9 ) but their nucleophilicity can render them incompatible with the electrophilic sulfur cathodes typically employed. To prevent this problem, a bulky non-nucleophilic amido group (NR 2 ) can be utilized instead, 10 which maintains the benefit of increased solubility but without the propensity toward undesirable side reactions. Chloride ligands have been suggested as the likely culprit for corrosion of magnesium electrolytes, 11 a problem when using a non-noble metal electrode; although purity of the starting materials has also been implicated as a potential cause. 12 Despite this possibility, magnesium organohaloaluminates continue to dominate the landscape in Mg battery research, 13 including theoretical calculations on the nuclearity of the active cation. 14 Indeed, it was recently suggested that free chloride anions in the electrolyte solution adsorb at the electrode surface, enhancing magnesium electrodeposition. 15 While the dinuclear cation is the most common structurally characterized motif within these systems due to its thermodynamic stability, other aggregated cationic moieties have been implicated as playing an important role evidenced through techniques such as mass spectrometry, although they have never been isolated nor characterized crystallographically in a magnesium organoaluminate species. These include mononuclear [MgCl] + and trinuclear [Mg 3 (µ 3 -Cl) 2 (µ 2 -Cl) 3 ] + (figure 1) 16 that can all be found in a complicated equilibrium in solution which is difficult to resolve due to, for example, lack of appropriate NMR handles in the [Mg x Cl 2x-1 ·nTHF] cations. Indeed, Muldoon has contended that this equilibrium can conceivably affect Mg electrochemistry in solution and consequently it should not be assumed that the crystallographically verified dinuclear complex is solely responsible, 17 a hypothesis supported by an X-ray absorption near edge structure (XANES) study which suggests that it is a different (unidentified) magnesium species which is electrochemically active. 18 The suggestion that these different cationic oligomers were involved was particularly interesting to us given our long standing interest in the bimetallic 'ate' chemistry of the main group metals, including magnesium and aluminium. 19 We felt our synthetic expertise could be exploited to shed light on these different, important oligomeric cations in the presence of a common anion, the results of such a study we now present herein.

Synthesis and molecular structures
We commenced by formulating a simple synthetic protocol designed to consistently produce the desired crystalline material. Instead of transferring the organo anion from Mg to AlCl 3 , as many previous studies had done (equations 1-3), we decided to follow the recent work of Liu and co-workers (equation 4), 7a and Zhao-Karger and coworkers (equation 5), 8b who pre-formed the Al-C bond prior to the introduction of the magnesium source. Thus, our first step was to make the Al-N bond via a salt metathesis reaction of equimolar amounts of lithium amide and AlCl 3 20 prior to introducing the magnesium reagent. The bulky aryl/silyl amide [(Dipp)(Me 3 Si)N]¯ (Dipp = 2,6-diisopropylphenyl) was chosen as it has a complementary combination of steric and electronic properties which can stabilize low valent or low coordination main group and transition metal species 21 and so seems ideally suited for purpose. Our focus was on an aluminium mono secondary amide since a higher Cl:R ratio is understood to lead to a higher oxidative potential. 8a (3)  As seen previously in other relevant magnesium aluminates, the cation of 1 has a noncrystallographic C 3 axis of symmetry passing through the two magnesium atoms, which are in a virtually octahedral environment consisting of three bridging facchloride anions and three terminal THF molecules.
With respect to mononuclear species, the predominant form of the [MgCl] + cation in the solid state is the penta-THF solvate, although to the best of our knowledge it has never been seen as the cationic moiety in a magnesium organohaloaluminate complex, which appear to prefer to adopt the thermodynamically more stable [Mg 2 (µ 2 -Cl) 3 ] + cationic structure. 22 3 ·3TMEDA] + , 3 was trinuclear as hoped (figure 2d) although its THF solubility was poor making it difficult to adequately characterize in solution (vide infra). We then moved to 2-methyltetrahydrofuran (MeTHF), whose slightly increased bulk vis-a-vis THF might prevent tris-solvation of the metal centre, and that enforced bis-solvation would consequently promote trinuclear cation formation. MeTHF is a greener alternative to THF with some similar properties 29 and has previously found use as an electrolyte solvent for rechargeable lithium batteries 30  Following the success of our reactions with polydentate N-donors TMEDA and Me 6 TREN, we repeated our protocol using the related ligand N,N,N',N'',N''pentamethyldiethylenetriamine (PMDETA), expecting that this tridentate ligand could act as an isodentate surrogate for three molecules of THF, yielding a magnesium aluminate with a dinuclear N-solvated cation. However, the resulting product was shown by single crystal X-ray diffraction to be the neutral magnesium dichloride complex MgCl 2 ·PMDETA (5, figure 3). We note that the related tridentate O-donor bis(2-methoxyethyl)ether (diglyme) also results in a neutral magnesium complex when used in this context, specifically dimeric [MgCl 2 ·diglyme] 2 . 10b This may suggest that acyclic tridentate donors do not possess the correct spatial conformation of their donor atoms to adequately protect one end of a [Mg 2 Cl 3 ] + fragment as they are aligned for mer rather than fac coordination to an octahedral metal centre, while they do not have the requisite number of donor atoms to protect a mononuclear or trinuclear cation.   27 Al NMR spectroscopy was uninformative, with well-resolved singlets absent due to the lack of high symmetry at the aluminium centre.
Complexes 1, 2 and 4 were also sufficiently soluble in C 6 D 6 to obtain NMR spectra in the absence of bulk Lewis donor. Surprisingly, despite the anion being identical in all three cases, there were noticeable differences in their 1 H NMR spectra (Figure 4) suggesting strong ion-pairing in less polar solvent. While the amido resonances of the di-and trinuclear complexes are fairly similar, those of the mononuclear complex (2) are clearly different. Specifically, the aryl resonances are well resolved into triplet (para) and doublet (meta), while the iPr and SiMe 3 resonances are deshielded with respect to those in the di-and trinuclear complexes (1 and 4 respectively). Furthermore the iPr methyl resonance is resolved into a pair of doublets (1.64/1.53 ppm) suggesting inequivalence. There is a slight separation of these resonances in 1 and 4 but they still overlap at around 1.32 ppm in each. Temperature effects can be ruled out since all spectra were obtained at 300 K, while a variable concentration NMR experiment revealed identical spectra, showing that this change is not a consequence of concentration variation. The reason for these differences is not instantly clear, although one could perhaps speculate that the unique feature of mononuclear complex 2, namely a terminal Mg-Cl bond, is somehow able to interact with the anionic moiety in weakly-donating benzene solvent.

Mass Spectrometry
Due to the inherent difficulty thus far in characterizing the cations in solution we turned to Electrospray-ionization (ESI) mass spectrometry. This method has been effective for the characterization of complex inorganic and organometallic ions in solution. 31  were observed (figures 5a, S10 and S11). The former corresponds to the cationic component of the salt, the latter to its dinuclear homologue. It is not clear whether the dinuclear ion was already present in the original sample solution or whether it only formed during the ESI process. The ESI process produces charged nanodroplets, which permanently lose solvent molecules due to evaporation. The increased effective concentration in these nanodroplets can lead to shifts of aggregation equilibria and, thus, to formation of the observed dinuclear ions. 32 Both [MgCl(Me 6 TREN)] + and [Mg 2 Cl 3 (Me 6 TREN) 2 ] + exhibit a 1:1 stoichiometry of magnesium and ligand, which reflects the latter's polydentate nature. Likewise, the absence of any THF adducts points to the lack of empty coordination sites at the magnesium centre. In addition, ions containing the protonated ligand were detected (Figures 5a and S12). Because of its high Brønsted basicity, the ligand can easily react with traces of protic contaminants remaining in the used glassware or the ESI source. For 3, which was run at a lower concentration due to its poor solubility, vide supra, we observed ions belonging to the homologous series [Mg n Cl 2n−1 (TMEDA) n ] + , n = 1-3 (Figures 5b and S13 -S15). Like in the case of the Me 6 TREN-containing ions, these chelated ions display a 1:1, Mg:ligand stoichiometry and do not bind any THF. , were detected (figures 5c and S16 -S19). Concentration-dependent measurements (figure 5d) showed that the relative abundance of the mononuclear ion decreased as a function of concentration, as expected on the basis of the law of mass action. The fraction of the dinuclear ions slightly increased with higher concentrations, whereas that of their trinuclear counterparts decreased slightly, the reason for this decrease not being obvious.
Analysis of solutions of the MeTHF-solvated complex [(Dipp)(Me 3 Si)NAlCl 3 ]¯ [Mg 3 (µ 3 -Cl) 2 (µ 2 -Cl) 3 ·6MeTHF] + (4) in THF gave similar mass spectra ( Figure S20). This finding proves that the MeTHF molecules coordinating to Mg centres are easily exchanged by excess of less bulky THF and corroborates our NMR findings. Repeating this experiment in MeTHF resulted in the detection of the trinuclear ions [Mg 3 Cl 5 (MeTHF) n ] + , n = 4 and 5 (Figures 5e and S21). In comparison with THFsolvated 1 in THF, the nuclearity of the observed complexes was significantly shifted toward higher aggregation states. Accordingly, the behavior of these salts in solution appears to parallel their behavior in the solid state.
Next we performed the reverse control experiment and dissolved the THF-containing salt 1 in MeTHF (Figures 5f and S22). In this case, the recorded ESI mass spectrum showed mainly ions coordinated by MeTHF, but a few complexes retaining a single THF molecule as well. This incomplete exchange again indicates that THF binds to the magnesium cations more strongly than MeTHF.
Finally, further information was obtained from the gas-phase fragmentation experiments (figures S10 -S20). The Mg complexes binding THF and MeTHF exclusively dissociated by losing one or two solvent molecules ( Figures S28 -S33).
For the larger and more fully solvated ions, the loss of one THF or MeTHF molecule occurred so easily that it proceeded even without the application of any extra excitation energy, as also the poorer mass resolution of the isotope patterns for these ions indicated. 33 For the smaller and less solvated ions, the loss of one THF or MeTHF molecule occurred less easily and required the concomitant addition of one water molecule to avoid a decrease in the coordination number (the ion trap mass spectrometer inevitably contains a low partial pressure of background water). The TMEDA-containing ions exchanged a ligand for water only to a minor extent, but mainly decomposed by expulsion of a neutral [MgCl 2 (TMEDA)] fragment ( Figures  S26 and S27). The analogous loss of neutral [MgCl 2 (Me 6 TREN)] was also observed for the dinuclear complex [Mg 2 Cl 3 (Me 6 TREN) 2 ] + ( Figure S25) whereas such a fragmentation reaction was not feasible for its mononuclear counterpart. This mononuclear ion only underwent partial decomposition of the ligand ( Figure S23). This deviating behavior of the TMEDA-and Me 6 TREN-containing complexes reflects the significantly stronger binding energies of these chelating ligands in comparison with monodentate THF and MeTHF.

Conclusion
In summary, mono-and trinuclear chloromagnesium cations (charge-balanced by a common organohaloaluminate counter-anion), implicated previously as key solution species in magnesium aluminate battery electrolytes, have now been rationally and selectively prepared for the first time by controlling the magnesium solvating Lewis donor additive. Paired alongside the thermodynamically favoured dinuclear derivative, these reproducible synthetic protocols represent a significant step forward since access to such a family opens the door to a greater understanding of their solution chemistry, particularly due to their excellent solubility. The present study has already taken the first steps in this direction and demonstrated the particular suitability of ESI mass spectrometry for this purpose. We have also introduced 2-methyl-THF as a coordinating Lewis donor into this chemistry and shown that it promotes and stabilizes formation of a trinuclear, magnesium rich cationic species, both in the solid state and in bulk MeTHF solution. Given its green credentials, we further intend to pursue this solvents applicability in systems such as these and hope that other research groups, inspired by our own findings, may follow suit.