Dissolution of metal oxides in task-specific ionic liquid

Due to their typically low reactivity, the activation of metal oxides, as found in ores, earths and minerals, is in general performed by high temperature reactions, which consume much energy. Owing to the prevalence of fossil fuels, this is accompagnied by the generation of large amounts of CO2. Alternatively, ionic liquids (ILs) can be used as solvents for metal oxide dissolution and downstream chemistry at much lower temperatures. The dissolution ability of the dry ionic liquid betainium bis(trifluoromethylsulfonyl)imide, [Hbet][NTf2], was investigated for 30 metal oxides at 175 °C and compared to chloride containing IL [Hbet]2[NTf2]Cl. A general dissolution-promoting effect of chloride anions was found, regarding reaction time as well as the variety of dissolved metal oxides. Up to now, the dissolution in [Hbet]2[NTf2]Cl is limited to basic or amphoteric metal oxides and assumed to be influenced by multiple factors, such as reaction conditions and the lattice energy of the metal oxide as well as its crystal structure. Comprehensive investigations were performed for the dissolution of CuO, which led to the discovery of the water-free complex compound [Cu2(bet)4(NTf2)2][NTf2]2. Proceeding from this compound, a complete exchange of the O-coordination sphere by other ligands was demonstrated, opening up promising possibilities for downstream chemistry.


Introduction
Metals are important resources for many applications, for which reason their extraction from naturally occurring ores, earths and minerals is of great industrial interest. Such natural metal sources oen consist of metal oxides alongside concomitant matter (paragenesis). Many of the established metal production processes starting from these resources are based on reactions at high temperatures, typically around 1000 C, and, thus, are extremely energy consuming and, owing to the prevalence of fossil fuels, CO 2 producing. 1,2 Due to the challenges of climate change and the increasing shortage of non-renewable energy resources, a development towards to more sustainable economic processes is of great importance. Innovative, more efficient ways of extracting metal oxides from their natural mixtures of substances and the subsequent production of metals and diverse metal compounds have to be developed.
During the last decades, ionic liquids (ILs) arouse a great deal of interest due to their promising applications to low temperature reactions. ILs, per denition, are salts with a melting point below 100 C and favourable properties, such as negligible vapour pressure, high thermal stability or a good solubility of numerous inorganic materials. 3 Recently, it was shown that ILs might even be suitable solvents for not readily dissolvable metal oxides. In this context, the presence of coordinating functional groups, as found in socalled task-specic ILs, turned out essential. [4][5][6][7][8] In the IL betainium bis(triuoromethylsulfonyl)imide ([Hbet][NTf 2 ]), interactions to metal cations occur through the carboxyl group at the cation. In a typical dissolution of a metal oxide, this functional group protonates the oxygen atom of the metal oxide yielding water, while the remaining carboxylate group, subsequently, coordinates to the metal cation, 4 as illustrated in Fig. 1.
Numerous of such metal-betaine complexes were investigated by Nockemann et al., who also noted the crucial role of water as an additional ligand. Thus, the IL [Hbet][NTf 2 ] was merely used as a reactant in an aqueous solution. 4,6 Other ILs with a slightly varied cation structure, but the carboxyl group still being present, show a similar reaction behaviour with metal oxides. 5 The possibilities of dissolving metal oxides in as far as possible water-free [Hbet][NTf 2 ] was investigated by the group of Binnemans in the course of recycling Nd 2 Fe 14 B magnets. 175 C was identied as the optimal reaction temperature, avoiding the decompositions of the IL, but supporting reaction kinetics at the same time. A signicant increase in reaction time in the absence of water as well as a decrease in the coordination number of the metal ion was reported. This is attributed to the [Hbet] + cation being too sterically demanding to saturate all coordination sites. For this purpose, smaller, additional ligands appear to be necessary, such as water or chloride anions. As demonstrated in a rst experiment, the addition of the latter resulted in enhanced dissolution rates. 8 Based on these ndings, in the present work, the dissolution behaviour of various metal oxides in the IL [Hbet][NTf 2 ] was investigated. Different from the approach of Nockemann et al. [4][5][6] but similar to the work of Binnemans, 8 no additional water was used as this usually involves two major problems regarding downstream chemistry: rst, with O-coordinated aqua ligands, a leaching of metal ions from potentially resulting complexes or ligand exchange reactions might prove difficult. Second, the relatively small electrochemical window of water oen involves the evolution of oxygen and hydrogen gases in electrolysis. Instead, the complex formation ability of metal oxides with betaine was investigated in pure [Hbet][NTf 2 ] and compared to reaction mixtures of the IL and betaine hydrochloride ([Hbet]Cl) as additional ligand source. The long-term goal of these studies is the application to downstream chemistry of metal oxides without the detour over the reduced metal as well as the metal separation from natural sources via leaching and subsequent electrolysis. However, the present work is concerned with the initial step of these problems, the dissolution and complex formation of metal oxides in the ILs [ were purchased from Iolitec. NiO (p.a.) originates from VEB Berlin-Chemie, Co 3 O 4 (p.a.) from Chempur and CaO (p.a.) as well as WO 3 (p.a.) from Reachim. MnO (99%) and MnO 2 (85-90%) were purchased from Sigma-Aldrich. ZnO (p.a.) was obtained from Grillo. ReO 3 was synthesised from Re powder (99.5%, Alfa Aesar) by repeated O 2 addition and subsequent evacuation in a silica ampoule, resulting in Re 2 O 7 . A subsequent chemical transport reaction with I 2 from 500 C to 400 C yielded ReO 3 .
PXRD of ThO 2 revealed the presence of an unidentied impurity (Fig. S1, ESI †). Similarities with the diffraction pattern of several intermetallic oxide phases suggest (intermediate) products of the ThO 2 decay chain. Caution! ThO 2 is a weak alpha emitter. All radioactive materials were handled in radioactively controlled facilities that are equipped with personal safety equipment.
All chemicals were used without any further purication, except for CaO, which was dried at 1000 C aer purchase.

Synthesis of [Hbet][NTf 2 ]
The For several metal oxides, mixtures with varying molar ratios or reaction times were investigated. Details are listed in Table  S4, ESI. † Work-up of the CuO samples was performed by the addition of acetone, thus separating CuO. The product yield h was determined by weighting the unreacted CuO and using the following formula.
found. Nevertheless, the product was suitable to perform ligand exchange experiments.

Calculation of lattice energies and U/x values
Lattice energies U of the metal oxides M x O y were calculated according to literature 9 with a Born-Haber cycle as phrased in the following equation.
For metals whose enthalpy of sublimation was not tabulated, the sum of the enthalpy of fusion (DH m ) and the enthalpy of vaporisation (DH v ) was used. The resulting slight error is <2%. 9 Furthermore, the ionisation energy of the Co(II,III) mixed oxide Co 3 O 4 was calculated according to P U/x was calculated by dividing the lattice energy by the number of metal atoms in the chemical formula.
DH f values were obtained from Thermochemical Data of Pure Substances, 10

Characterisation
Powder X-ray diffraction (PXRD) was performed on two similarly constructed X'Pert Pro and Empyrean diffractometers (PANanalytical) at 296 K equipped with a curved Ge(111) monochromator in Bragg-Brentano geometry using Cu-Ka 1 radiation (l ¼ 154.0598 pm). Rietveld renement was performed with TOPAS 13 by using the fundamental parameter approach and Le Bail method for modelling preferred orientation.
Infrared (IR) measurements were performed on a Bruker Vertex 70 FTIR spectrometer with attenuated total reection (ATR) accessory in a radiation range from 500 cm À1 to 4000 cm À1 and 32 scans per measurement. Data analysis was performed with the programme OPUS. 14 For Raman spectroscopy, a FT-Raman spectrometer Bruker RFS 100 was used with an excitation wavelength of 1064 nm and 50 to 200 scans per measurement. Due to the strong dispersion of radiation in coloured substances and the resulting high background, additional scans were taken for these samples. Analysis of the data obtained was performed with the programme OPUS. 14 NMR samples were prepared by dissolving a spatula tip of the samples in deuterated dimethylsulfoxide (DMSO-d 6 ) as deuterium lock and lling a NMR tube to a height of approximately 3 cm. 1 H NMR spectra were recorded on a Bruker Avance WB NMR 300 MHz spectrometer with the resonance frequency 300 MHz.
The collection of diffraction intensities of single-crystals (SCXRD) for crystal structure analysis took place on a singlecrystal X-ray diffractometer SuperNova (Rigaku Oxford Diffraction) with Cu-Ka radiation (l ¼ 154.184 pm) at 100 K under owing nitrogen gas. Empirical multi-scan absorption corrections were applied to the data. Structure solution was performed with the programme SHELXT 15 in OLEX2 16 via intrinsic phasing. For structure renement, the method of full-matrix least squares on F o 2 with the programme SHELXL 15 in OLEX2 16 was used. Non-hydrogen atoms were rened with anisotropic displacement parameters, while hydrogen atoms were rened with riding coordinates and displacement parameters. The crystal structure was plotted in the programme Diamond 4.5.2. 17 The full crystallographic data is deposited at the Cambridge Crystallographic Data Centre, deposit number CCDC 1947513. † For energy dispersive X-ray spectroscopy (EDX), a drop of a sample dispersed in acetone was placed on a silicon wafer xed on a SEM sample holder and stored in a vacuum chamber for three days. The studies were performed on a Hitachi SU8020 scanning electron microscope with an Oxford Silicon Dri Detector (SDD) X-Max N . An accelerating voltage of 25 kV was used.  3 . The powder X-ray diffractograms as well as the Rietveld renement are shown in Fig. S2 and S3, ESI. † The fact of roughly one sixth of the cations being unprotonated betaine might affect the maximum amount of a metal oxide dissolvable in the IL. However, as the experiments performed in this work are of rather qualitative than quantitative nature, the resulting error is considered insignicant. In the following, the synthesized IL is referred to as  Fig. S9, ESI † reveals an intact IL. The colour of the liquid might result from very ne, dispersed SnO particles. An overview about the products obtained is given in Table S1, ESI. † In agreement with this, all reections in the PXRD patterns of the latter samples could be attributed to the starting materials (see Fig. S4 and S5, ESI †). In contrast to this, the eight soluble metal oxides BaO, CaO, MgO, MnO, PbO, PbO 2 , SrO and ZnO gave homogenous solutions or pastes, whereby the high viscosity of the latter products could be lowered by using a higher amount of [Hbet][NTf 2 ]. Apart from the relatively low-viscosity liquids of the CaO and MgO samples, crystallisation occurred during cooling or upon application on the PXRD sample holder, giving completely unidentied reection patterns. Reections at low 2q angles suggest a large unit cell that might result from metal-betaine complex compounds.

Results and discussion
Besides visual examination and PXRD, evidence for complex formation is given by IR spectroscopy, where two bands attributed to the [Hbet] + cation are a good indicator: the asymmetric stretching vibration of the carboxyl OH (n as OH) at 3301 cm À1 and the asymmetric stretching vibration of COO (n as COO) at 1770 cm À1 in the spectrum of the pure IL, which are marked with dotted lines in Fig. 2. Upon coordination, n as OH disappears due to the deprotonation of the carboxyl group. Meanwhile, a shi of n as COO to lower wavenumbers can be observed as the excitation of this vibration needs less energy when the carboxylate group binds to a heavier metal atom instead of a proton. In agreement with visual observations and PXRD patterns, no shi of n as COO is found for the samples of The IR spectra of several samples suggest the presence of metal-betaine complexes, however, precipitations occur. For CuO, no complete reaction was achieved, but small amounts of the starting material were present in a blue solution from which blue crystals precipitated. The same blue compound was obtained from Cu 2 O, as identied by PXRD, indicating the oxidation of copper(I). Similarly, the V 2 O 3 sample contained a greenish blue powder in a colourless liquid. In order to clarify the nature of the solid, it was washed with acetone, hence, removing the IL. IR spectroscopy suggests the presence of V 3+betaine complexes due to the bands typical for betaine and [NTf 2 ] À but n as COO shied to 1595 cm À1 (highlighted in green in Fig. 3). Furthermore, from Bi 2 O 3 , a white powder in a colourless liquid was obtained. Aer washing with acetone, it was identied as (BiO) 2 CO 3 by PXRD (Fig. S5, ESI †). Bi 2 O 3 is known to readily react with the acid anhydride CO 2 to (BiO) 2 CO 3 , 18 however, [Hbet][NTf 2 ], apparently, signicantly catalyses this reaction. In agreement with this, the IR spectrum in Fig. 3 is typical for this compound according to literature, 18 indicating the absence of solid bismuth(III)-betaine complex compounds. The repetition of the reaction under argon ow reveals a signicantly lower reactivity of the reactants. Meeting general expectations, no reaction with CO 2 takes place, but a suspension of a yellow powder in a colourless liquid forms. By PXRD, only reections corresponding to the reagent Bi 2 O 3 are found, which already contained some (BiO) 2 CO 3 (Fig. S6, ESI †). However, the IR spectrum, displayed in Fig. 3, shows two n as COO bands at 1755 cm À1 and 1619 cm À1 . This suggests a majority of unreacted IL alongside with a small amount of betaine coordinated to bismuth. Therefore, Bi 2 O 3 is assumed to have a low solubility in [Hbet][NTf 2 ]. The fast reaction on air might be ascribed to the precipitation of (BiO) 2 (CO 3 ) shiing the reaction to the product side. Some kind of special behaviour was also observed for Co 3 O 4 and V 2 O 5 . Very small amounts of both oxides dissolve in the IL, yielding a violet, respectively green or sometimes brown liquid. As both metal oxides are strong oxidising agents, 2 their partly dissolution is attributed to the oxidation of the IL. This is supported by the NMR spectra displayed in Fig S9, † where the occurrence of additional signals compared to pure [Hbet][NTf 2 ] suggests the partial decomposition of the IL. In agreement with this, the PXRD pattern of the V 2 O 5 sample product, shows several unidentied reections besides the unreacted metal oxide. These might result from crystallised decomposition fragments of the IL. IR bands of low intensity at 1630 cm À1 and  1598 cm À1 are also attributed to the decomposition of the IL or small amounts of vanadium(III)-betaine complexes. However, it also has to be taken into account, that the synthesis of [Hbet] [NTf 2 ] does not guarantee the complete absence of chloride. Therefore, also the oxidation of chloride impurities to chlorine by cobalt(III) and vanadium(V) might play a role in the dissolution of the metal oxides.
Several low intensity IR bands in the range highlighted in green in Fig. 2 can also be found for the sample of MoO 3 , despite the unreacted appearance of the product as well as PXRD only identifying the reagents. IR spectroscopy of the remaining white powder washed with acetone gives evidence for the presence of molybdenum-betaine complexes, as shown in Fig. 3. The low intensity bands in the range 1300 cm À1 to 1500 cm À1 suggest the presence of betaine, while the nonappearance of the bands at lower wavenumbers might indicate the absence of [NTf 2 ] À anions. Furthermore, the occurrence of n as COO at 1626 cm À1 , but no band at 1770 cm À1 gives evidence for the presence of betaine only in a coordinated state. Accordingly, MoO 3 is suspected to show a low solubility in [Hbet][NTf 2 ]. As a reason for this behaviour deviating from other MO 3 type metal oxides, such as WO 3 and ReO 3 , the crystal structure might serve. In contrast to the latter salts, MoO 3 is built from layers of MoO 6 octahedrons, which are expected to be affected by dissolution signicantly easier than 3D cross-linked structures.
Similarly, the sample of ThO 2 appears like a suspension of the reagents, but the IR spectrum shows a band at 1640 cm À1 . However, PXRD of the starting material reveals some minor reections besides the main pattern of ThO 2 , which could not be assigned to a phase and might result from the decay chain of ThO 2 (Fig. S2, ESI †). Therefore, it is assumed that ThO 2 does not dissolve in [Hbet][NTf 2 ], but merely the impurities.
In order to substantiate these experimental observations, the dissolved, respectively undissolved metal oxides were investigated for common properties. A method to estimate the solubility of metal oxides in water-saturated [Hbet][NTf 2 ] was recently suggested by Fan et al. who proposed a correlation between the lattice energy of a metal oxide and its solubility in the IL. Therefore, they introduced the U/x value of a metal oxide M x O y where U is the lattice energy. For U/x < 7000 kJ mol À1 , a good solubility of metal oxides was found while U/x > 10 000 kJ mol À1 indicated metal oxides insoluble in watersaturated [Hbet][NTf 2 ]. Corresponding to their method, 9 we calculated the lattice energies for all metal oxides examined experimentally by using the Born-Haber cycle. The resulting U/x values are listed in Table 1, detailed information about the data used for calculation can be found in Table S2 From these observations it is concluded that the lattice energies of metal oxides, respectively their U/x values, merely allow a rough assessment of their solubility in [Hbet][NTf 2 ]. The assumed reason is the lattice energy not being the only factor inuencing solubility. Hence, the reaction conditions, such as temperature might have a signicant inuence, as assumed by Jayachandran et al. 19 As illustrated by the example of MoO 3 (layered structure) in contrast to WO 3 and ReO 3 (3D frameworks), the crystal structure of the starting materials might also have an effect on their disintegration. Furthermore, the product side of the reaction should not be neglected as complex formation constants as well as a possible change in the coordination number of the metal ion could affect the dissolution behaviour. A signicant inuence on the dissolution, in our opinion, also originates from the acidity or basicity of the metal oxide. Apparently, in the here performed experiments only relatively basic or amphoteric oxides readily react with the IL, meeting general expectations as [Hbet] + is an acidic cation. However, these suggestions are based on qualitative observations that should be supported by measuring data. Yet, to our knowledge, no comprehensive table about the acid-base properties of metal oxides exists to provide such evidence.
Many further examples of copper paddle-wheel structures showing the same characteristics, such as the Cu-Cu distance, the distance ratio of coordination bonds and a distorted octahedral geometry, are known in literature. 23,24 A comparison of the experimental PXRD pattern of the CuO-[Hbet][NTf 2 ] reaction product to the one simulated from singlecrystal data of [Cu 2 (bet) 4 (NTf 2 ) 2 ][NTf 2 ] 2 gives evidence for the phase impurity of the synthesis product, as shown in Fig. S8, ESI. † All theoretical reections turn up in the experimental pattern, whereby the shi to lower 2q values meets the expectations due to the temperature difference of PXRD (RT) and SCXRD (100 K) measurements. By Rietveld renement, a room temperature unit cell was obtained with 0.9%, 1.5% and 1.6% enlarged a, b and c lattice parameters, respectively (a ¼ 1459.0  pm, b ¼ 1470.2 pm, c ¼ 1522.9 pm, b ¼ 95.3 ).
Besides, apparently, another unidentied phase is present, as numerous reections cannot be ascribed to any known phase. Attempts to determine a possible unit cell were not successful, while a microscopic separation by crystal morphology also is not possible due to irregular crystal shapes. However, due to the low 2q angles of several reections, indicating a similarly large unit cell, another unknown copperbetaine compound phase should be considered.
Ligand exchange experiments. In order to investigate whether the weakly coordinating [NTf 2 ] À anions can be replaced by other ligands, the reaction of [Cu 2 (bet) 4 25,26 The consequent conclusion of [Hbet] + cations not being dissolved in [P 66614 ]Cl is supported by IR spectroscopy, as  shown in Fig. 6. Clearly observable are the main peaks of [P 66614 ] Cl, but also the most intensive signals of [Hbet][NTf 2 ] are visible in the spectrum of the sample (highlighted in green). According to literature [27][28][29][30] all of them are attributed to vibrational modes of the [NTf 2 ] À anion (see Table S3, ESI †), while n as COO at 1770 cm À1 (respectively the shied signal of coordinated [Hbet] + at 1655 cm À1 ) is not observed despite its medium intensity.
Further evidence of copper(II) being leached from the [Cu 2 (bet) 4 (NTf 2 ) 2 ] 2+ complex is given by the Raman spectrum displayed in Fig. 7. A band at 297 cm À1 suggests a Cu-Cl stretching vibration (n s CuCl, highlighted in green), 31,32 while no signal is found around 385 cm À1 , where Cu-O vibrations would be expected. 31 Furthermore, Suffren et al. reported a correlation of the band position of n s CuCl and the trans angles in [CuCl 4 ] 2À complex anions, whereby they found a shi to higher wavenumbers with increasing tetrahedral distortion over a square planar geometry. 33 Therefore, the experimental Raman band at a relatively high wavenumber is in good agreement with the yellow colour of the solution suggesting a distorted tetrahedral geometry for [CuCl 4 ] 2À . A distorted tetrahedron is the hightemperature phase of the thermochromic [CuCl 4 ] 2À . Despite being the electrostatically more favoured geometry, it is assumed that the square-planar RT phase is stabilized by hydrogen bonds at lower temperatures. 34 As in an environment of [P 66614 ] + cations, hydrogen bonds to [CuCl 4 ] 2À are not likely, the distorted tetrahedral geometry of this complex anion seems reasonable.
Altogether, these investigations suggest that copper(II) ions are leached from the Cu-betaine complex and form [CuCl 4 ] 2À complex anions in a distorted tetrahedral geometry. Thus, a complete exchange of the O-coordination sphere around copper(II) by chloride could be shown, opening promising possibilities for downstream chemistry starting from metal oxides.  Fig. S9, ESI, † the peaks at 3.1 ppm and 4.1 ppm are attributed to the protons of the CH 3 and CH 2 groups of betaine, respectively. Another signal of the carboxyl proton is not observed which is attributed to its low intensity and the expected broadness. 35 Due to the complexity of the spectrum obtained from heated [Hbet] 2 [NTf 2 ]Cl, no attribution to specic decomposition fragments was realised, yet. Concomitant in the IR spectrum of the resulting brown liquid (Fig. 8), the sharp, medium-intensity n as COO band at 1770 cm À1 transforms into two low intensity bands at 1743 cm À1 and 1666 cm À1 . The rst band is attributed to n as COO of undecomposed [Hbet] + , while the second band is assumed to result from a not yet identied decomposition fragment. In Fig. S9, ESI, † 1 H NMR spectra of several samples give evidence for the decomposition of the IL taking place in mixtures with metal oxides as well. Further studies of several systems indicate, that this can be reduced by increasing the amount of [Hbet][NTf 2 ] as well as by decreasing the reaction time.
The inuence of chloride anions on the dissolution of a metal oxide was investigated in detail on the example of CuO. As a result of this, a general dissolution promoting effect of chloride can be stated, however, the specic impact strongly depends on the amount of [Hbet]Cl added.
On one side, the addition of chloride in small amounts (0.3 mol% based on the total number of counter ions of [Hbet] + ) reveals a strong catalytic effect. Thus, in a reaction for 24 h, the  yield of [Cu 2 (bet) 4 (NTf 2 ) 2 ][NTf 2 ] 2 could be increased from 13% to 66% by the addition of [Hbet]Cl. Such a promoting effect of chloride anions is not uncommon, e.g. was a similar outcome observed for the synthesis of Cu 3Àx P from copper and red phosphorous, whereby the product yield could signicantly be increased by the addition of a few mol% of [P 66614 ]Cl to [P 66614 ] [NTf 2 ]. 36 While the ideal reaction time for the sample with 0.3 mol% chloride amounts to 24 h, by the addition of equimolar amounts of [Hbet]Cl, i.e.
[Hbet] 2 [NTf 2 ]Cl, it reduces to only 4 h. However, besides an effect on the reaction time, also an inuence on the product phase is observed. As shown in Fig. 9, the PXRD pattern does not correspond to the pattern of [Cu 2 (bet) 4 (NTf 2 ) 2 ][NTf 2 ] 2 . Instead, the low 2q angles of the numerous unidentied reections suggest a large unit cell that might result from another copper-betaine complex compound. Evidence for this is found in the IR spectrum displayed in Fig. 8. A low intensity band at 1753 cm À1 might result from a small amount of uncoordinated [Hbet] + , while another band at 1669 cm À1 might be conform with the respective band in the [Cu 2 (bet) 4 (NTf 2 ) 2 ][NTf 2 ] 2 spectrum resulting from bridging coordination of betaine. However, two additional bands at 1655 cm À1 and 1618 cm À1 are solely found in the spectrum of the chloride-rich sample and might indicate another coordination of betaine to copper(II), i.e. monodentate or bidentate. Thus, the positive effect of chloride anions on the reaction rate, as suggested by the group of Binnemans, 8 can be conrmed. Hence, this property is not unique to water, but H 2 O in principle appears to be substitutable by other small molecules or ions.
In order to investigate whether chloride ions are only suitable to decrease reaction time or might also support the dissolution of so far insoluble metal oxides, their effect in other reaction systems was examined. In agreement with the previous ndings, signicantly more metal oxides compared to the pure IL could be dissolved by the addition of chloride. Thus, in In contrast to this, for the samples of BaO and Cu 2 O, the powders present in the product suspensions were identied as the respective chloride salts by PXRD, while for Bi 2 O 3 , BiOCl was found. Apparently, the interactions of the metal ions with chloride are favoured over betaine, leading to precipitation following dissolution of the oxide. The reaction to chloride salts is, furthermore, found for the two lead oxides, however, they appear to be unstable as time-resolved PXRD measurements reveal (Fig. 10). Thus, for the sample of PbO a white precipitation occurs immediately aer the melting of the IL and, aer 5 minutes reaction time, PbCl 2 is identied as the only crystalline phase by PXRD. With increasing time, the PbCl 2 phase disappears, giving way for another, yet unidentied phase. This change of composition continues during storage of the sample at room temperature. It is conceivable that a partly ligand exchange reaction takes place, yielding Pb-Cl-betaine complexes or the decomposition products of the IL (Fig. S9 Despite the precipitation of chloride salts in several cases, the IR spectra display a n as COO band shied to lower wavenumbers towards the pure IL (highlighted in green in Fig. 8). This suggest the presence of some amount of metal ions coordinated by betaine for these samples. The same assumption is made for MnO 2 , where an unidentied solid precipitation was obtained in terms of numerous colourless crystals in a brown liquid. The crystal colour clearly deviating from the black reagent suggests dissolution.
Furthermore, for the mentioned metal oxides, assumed dissolved, the reagent oxide cannot be identied by PXRD, as shown in Fig. S10 and S11, ESI. † This is different from Fe 2 O 3 , NiO, V 2 O 3 and ThO 2 . From Fe 2 O 3 a red powder identied as Fe 2 O 3 by PXRD was obtained in a brown liquid. Additional unidentied reections are attributed to decomposition fragments of the IL in agreement with the 1 H NMR spectrum in Fig. S9, ESI. † However, two bands in the IR spectrum at 1737 cm À1 and 1641 cm À1 suggest the occurrence of uncoordinated as well as coordinated betaine. The presence of dissolved iron(III) in the separated liquid phase was conrmed by testing with aqueous KSCN and NaF solutions. Therefore, a low dissolution for Fe 2 O 3 is assumed. For the solid NiO sample product, a medium intensity IR band at 1632 cm À1 suggests the coordination of betaine. Further evidence for the partly dissolution of NiO is given by the addition of acetone, which according to our experience is a suitable solvent for some metal-betaine complexes, but not for the respective oxides. Thus, a green liquid is obtained, indicating the presence of dissolved nickel(II) ions, alongside with green NiO powder. Similarly, for V 2 O 3 , a suspension of a brown paste and a black powder is obtained, which is identied as V 2 O 3 by PXRD. Changing the reagent ratio to n V : n [Hbet][NTf 2 ] : n [Hbet]Cl ¼ 1 : 6 : 1 yields a clear solution with only a few black particles. Corresponding to this, the IR spectrum shows a band at 1662 cm À1 suggesting the coordination of betaine to vanadium(III).
In contrast to this, ThO 2 appears to react to a white, amorphous powder. Only ThO 2 is identied by PXRD, which is attributed to a small amount of black particles in the sample. Hence, no complete reaction was achieved (neither with triple the amount of IL). However, EDX gives evidence for the white powder being an organometallic thorium compound with Th, O and C in an approximate ratio of 2 : 7 : 10 ( Fig. S12 and Table  S5, ESI †). As only semiquantitative measurement conditions were applied and H atoms cannot be detected by EDX, no assumptions of the precise composition of this compound should be made. However, as no N is detected, the presence of thorium-betaine complexes appears unlikely. Instead, the powder might be formed by thorium and a decomposition product of the IL. This is in agreement with the IR spectra of the powder washed with acetone, as shown in Fig. 3. No bands are observed for the washed powder indicating the absence of betaine as well as [NTf 2 ] À ions. Further experiments in this system should be performed in order to investigate whether the precipitation of a thorium compound can be avoided by suppressing the decomposition of the IL and to clarify the nature of this white powder. Similar to the reaction in a chloride-free system, also the dissolution of very small amounts of MoO 3 is suspected due to several bands in the highlighted range in the IR spectrum. As PXRD indicates the presence of unreacted MoO 3 , the product was washed with acetone and again studied by IR spectroscopy, giving evidence for molybdenum-betaine complexes, as shown in Fig. 3. Hence, a low solubility of MoO 3 is assumed in the presence of chloride as well as in the pure IL.
For other products, such as Al 2 O 3 , Ga 2 O 3 , GeO 2 , Nb 2 O 5 , ReO 3 , TiO 2 and WO 3 , the reagent metal oxide is present in a brown liquid and no dissolution is assumed. The respective IR spectra typically correspond to uncoordinated betaine in a [Hbet][NTf 2 ]-[Hbet]Cl mixture and PXRD patterns show unidentied reections. This is attributed to decomposition fragments of the IL crystallising. For the presumably undissolved metal oxide In 2 O 3 , the two bands in the IR spectrum at 1743 cm À1 and 1666 cm À1 are present in intensity proportions deviating from metal-free [Hbet] 2 [NTf 2 ]Cl. This is attributed to varied amounts of decomposition fragments of the IL, but as the band positions are similar, no complex formation is assumed. In contrast to this, in the Cr 2 O 3 sample, the metal oxide is present in a colourless liquid. However, the 1 H NMR spectrum in Fig. S9, ESI † reveals the partly decomposition of the IL. The IR spectrum does not suggest any dissolution.
Altogether, again, a general trend is observed, that very acidic metal oxides do not react with [Hbet] + , but only basic or amphoteric ones. However, as V 2 O 5 and Co 3 O 4 form an exception from this trend, another dissolution mechanism was identied as causal. Although the black powder in the V 2 O 5 sample could not be identied by PXRD, it clearly is not the yellow reagent V 2 O 5 . Instead, a green colour in solution usually is associated with vanadium(III), 2 suggesting a redox reaction. The IR spectra shows only one broad band at 1628 cm À1 that might result from betaine coordinated to vanadium(III). As V 2 O 5 is known to oxidize chloride to chlorine, 2 the reduction of vanadium(V) to vanadium(III) by the oxidation of chloride is conceivable. However, the 1 H NMR spectrum in Fig. S9, ESI † also suggests the decomposition of the IL.
Similarly, the dissolution of Co 3 O 4 is attributed to a redox reaction. Many small, blue crystals were obtained as reaction product. As CoCl 2 could be ruled out by a different PXRD pattern and differing physical properties (less hygroscopic, different dissolution behaviour in acetone), the small crystals are assumed to consist of cobalt-betaine complexes. The blue colour, respectively the violet colour of an aqueous solution, suggest an oxidation state of cobalt(II), as cobalt(III) would be expected to form green aqua-complexes. 37 This is in agreement with the known oxidising effect of cobalt(III) in the presence of chloride by the formation of chlorine, 2 but the 1 H NMR spectrum also suggests the decomposition of the IL. Therefore, the nature of the reductant of V 2 O 5 and Co 3 O 4 is uncertain, yet.
Summarising, for the here studied metal oxides, the presence of chloride in several cases has a positive effect on the dissolution. So in a 1 :  3 and ZnO in the pure IL as well as in the presence of chloride. Altogether, the addition of chloride ions enables not only the dissolution of more metal oxides than in pure [Hbet] [NTf 2 ], but, as studied in detail for the CuO, system, also has a decreasing effect on the reaction time and might affect the product phase. This reaction-promoting effect of small coligands is in agreement with numerous investigations of other research groups. 4,7,8,38 However, not only water molecules can be such co-ligands, but the extensive investigations in this work give evidence for the applicability of chloride ions. Nevertheless, it has to be considered, that several metal ions form not readily soluble chloride salts aer dissolution, which might complicate downstream chemistry. Altogether, the dissolution of metal oxides in task-specic reaction systems seems possible. Further investigations have to be performed to determine whether the chloride ions act as permanent ligands in metal-betaine complexes or exhibit a catalytic effect. Furthermore, experiments with different small ions or molecules as potential coligands will be interesting.

Conclusions
[Hbet][NTf 2 ] is a suitable IL for the dissolution of various, basic or amphoteric metal oxides by the formation of metal-betaine complexes. The dissolution is assumed to be inuenced by a combination of various factors, whereas the basicity is considered important, but also the lattice energy of the metal oxide, the crystal structure and the reaction temperature play a role. Furthermore, a decrease of the reaction rate in the absence of water can be conrmed. The addition of chloride ions improves the solubility of metal oxides. Thereby, the amount of chloride in the reaction mixture appears to not only have an effect on the reaction time by means of a catalyst, but also to affect the product phase.

Conflicts of interest
There are no conicts to declare.