Ionic Liquid-Based Low-Temperature Synthesis of Crystalline Ti(OH)OF·0.66H2O: Elucidating the Molecular Reaction Steps by NMR Spectroscopy and Theoretical Studies.

We present an in-depth mechanistic study of the first steps of the solution-based synthesis of the peculiar hexagonal tungsten bronze-type Ti(OH)OF·0.66H2O solid, using NMR analyses (1H, 13C, 19F, and 11B) as well as modeling based on density functional theory (DFT) and ab initio molecular dynamics (AIMD) simulation. The reaction uses an imidazolium-based ionic liquid (IL, e.g., C x mim BF4) as a solvent and reaction partner. It is puzzling, as the fluorine-rich crystalline solid is obtained in a "beaker chemistry" procedure, starting from simple compounds forming a stable solution (BF4 --containing IL, TiCl4, H2O) at room temperature, and a remarkably low reaction temperature (95 °C) is sufficient. Building on NMR experiments and modeling, we are able to provide a consistent explanation of the peculiar features of the synthesis: evidently, the hydrolysis of the IL anion BF4 - is a crucial step since the latter provides fluoride anions, which are incorporated into the crystal structure. Contrary to expectations, BF4 - does not hydrolyze in water at room temperature but interacts with TiCl4, possibly forming a TiCl4 complex with one or two coordinated BF4 - units. This interaction also prevents the heavy hydrolysis reaction of TiCl4 with H2O but-on the other side-spurs the hydrolysis of BF4 - already at room temperature, releasing fluoride and building F-containing Ti(OH) x Cl4-x F y complexes. The possible complexes formed were analyzed using DFT calculations with suitable functionals and basis sets. We show in addition that these complexes are also formed using other titanium precursors. As a further major finding, the heating step (95 °C) is only needed for the condensation of the Ti(OH) x Cl4-x F y complexes to form the desired solid product but not for the hydrolysis of BF4 -. Our study provides ample justification to state a "special IL effect", as the liquid state, together with a stable solution, the ionic nature, and the resulting deactivation of H2O are key requirements for this synthesis.

Introduction

Syntheses of metal oxides involving ionic liquids (ILs) have gained substantial interest recently since it is possible to synthesize many different metal oxides with various morphologies.[1−3] In comparison to other types of metal oxide syntheses, IL-based strategies often enable to use lower reaction temperatures. Hence, in this respect, IL-based syntheses of metal oxides are potentially consistent with the concept of sustainable chemistry.[4,5] Interestingly, it was found that ILs possibly are not only solvents in such reactions, but they also act as reactants and thereby strongly direct the compounds obtained. This property was, for example, demonstrated in the phase-pure synthesis of the uncommon, bronze-type compound “TiO2(B)” with the help of imidazolium-based ILs.[6,7] Compared to other literature-known syntheses of this compound, quite “soft” conditions, requiring a temperature of only 95 °C, are sufficient.[8] Another major advantage is the small number of reactants, as only ILs, TiCl4, H2O, and EtOH are needed. As a crucial step of this synthesis, the usage of a mixture of two different ILs, C16mim Cl and C4mim BF4, was proposed. Voepel et al. further investigated the reaction mechanism of this synthesis and found that the concentration of BF4– significantly influences the product composition.[9] It was proposed that BF4– is partly hydrolyzed during the reaction, providing fluoride anions, which can coordinate to Ti chloro complexes. The finally obtained bronze-type TiO2(B) material is not phase-pure but contains a low fraction of fluorine, which substitutes oxygen positions and thereby directs the crystallization. Depending on the amount of available fluoride anions, different amounts of blocked positions for the hydrolysis of the titanium complexes can be obtained, which for higher fractions of fluorine can even result in the formation of different titanium oxyfluoride compounds. Hence, using the synthesis of Voepel et al. beyond a certain concentration of the BF4-containing ILs, the peculiar hexagonal tungsten bronze (HTB) compound Ti(OH)OF·0.66H2O was observed,[9] which before had been accessible by a different synthetic concept.[10] Ti(OH)OF·0.66H2O possesses a quite interesting crystal structure with channels along the c-axis, endowing the compound with interesting electrochemical properties with respect to the incorporation of Li+ probably in the channels.[11]

In addition to the influence of the IL anion, the length of the alkyl chain of imidazolium-based ILs influences the obtained products as well,[12] allowing us to use just one BF4–-containing IL to synthesize Ti(OH)OF·0.66H2O instead of a mixture of ILs, contrary to previous studies.[9] Furthermore, within or after the formation of Ti(OH)OF·0.66H2O nanoparticles, the polar imidazolium head group attaches on the nanoparticles’ surface. It is probably this attachment that stabilizes the nanoparticles against conversion into the thermodynamically more stable polymorphs TiO2(B) and anatase. This stabilization effect is more pronounced when using ILs with longer alkyl chains demonstrating how essential the choice of the IL cation is.[12]

While thus already several details of the reaction mechanism of the presented IL-based synthesis leading to Ti(OH)OF·0.66H2O and TiO2(B) have been clarified, the first steps in the reaction, i.e., involving molecular species, are still unclear. However, the described empirical findings suggest that the final crystal structure is already predetermined at the level of F- and O-containing Ti complexes. Hence, in the present study, we target the initial reaction steps, involving Ti complexes and especially the BF4– anion with the help of NMR spectroscopy.

In the past, NMR spectroscopy has already been successfully used to investigate IL-based syntheses.[13] Saihara et al. investigated the hydrolysis process of the IL N,N-diethyl-N-methyl-N-(2-methoxyethyl)ammonium tetrafluoroborate (DEME BF4–) in the presence of water by means of 19F and 11B NMR measurements.[14] They found that during the hydrolysis of BF4–, HF is generated and reacts with the surrounding glass container since SiF62– was detected in the 19F NMR spectrum. Lin et al. focused their work on anion-exchange reactions in imidazolium-based ILs leading to the formation of hydroxometalates.[15] Since imidazolium-based ILs are also used in the already presented synthesis of Ti(OH)OF·0.66H2O (HTB), it can be expected that liquid-state NMR measurements can contribute to elucidate the underlying initial reaction steps. In addition, it was demonstrated by Giernoth et al. that different interactions in imidazolium-based ILs are detectable by NMR.[16]

Our previous studies have already shown that the hydrolysis of the IL anion BF4– is a crucial step, as it provides the fluoride anions that are incorporated into the crystal structure of Ti(OH)OF·0.66H2O. Theoretical calculations and Raman spectroscopy measurements performed in previous studies of our working group[9] indicated that BF4– is stepwise hydrolyzed to B(OH)4–. However, it has not been clarified yet whether the hydrolysis occurs via a direct reaction between BF4– and H2O or via an interaction/reaction between BF4– and a Ti complex, e.g., Ti(H2O)(OH)6–. Hence, in the present study, we conducted 1H, 13C, 19F, and 11B NMR measurements on solutions containing different mixtures of the used reactants with the goal to elucidate intermediate products and interactions present in the solutions, to shed further light on the hydrolysis mechanism of BF4– in the mixture with Ti complexes. Since the heating step is crucial for the synthesis of Ti(OH)OF·0.66H2O, additional NMR measurements were performed on solutions heated in situ to 95 °C (standard reaction conditions, see Scheme ).

Scheme 1

Schematic Illustration of the Synthesis Procedure

The colors inside the boxes refer to the different mixtures of reactants, which were investigated in this work (blue: IL + H2O, green: IL + TiCl4, yellow: IL + TiCl4 + H2O). The shown images of the solutions correspond to the subsequent reaction steps at room temperature.

To understand the role of the Ti compound, we also used titanium isopropoxide (TTIP) for the synthesis of Ti(OH)OF·0.66H2O. TTIP was chosen, as here Ti is bonded to alkoxy groups, thus possessing a different hydrolysis behavior as TiCl4, which was previously used.

To determine if complexes between TiCl4 and BF4– are present during the reaction, theoretical calculations were carried out. For this purpose, possible complexes that could form (based on the given reactants) were postulated. In the next step, these complexes, as well as their starting materials, were post-modeled and then geometry-optimized via density functional theory (DFT) calculations (see the Computational Details: Static Calculations section).

In summary, the conceptual methodology of this study is to elucidate the first reaction steps and species in the IL-mediated formation of the inorganic solid Ti(OH)OF·0.66H2O, to clarify how the generation of such F- and O-containing crystal is possibly predetermined at the level of molecular species, taking advantage of NMR spectroscopy and state-of-the-art modeling.

Results and Discussion

The main goal of our investigations was to get insight into the reaction mechanisms of the IL-based synthesis of Ti(OH)OF·0.66H2O, especially the role of the BF4– anion and the formation of Ti complexes, on the basis of NMR spectra. To understand the interactions of the reactants with each other, the composition of mixtures of the involved compounds was systematically varied, and these solutions were subjected to 1H, 13C, 19F, and 11B NMR measurements. A scheme showing all measured solutions and the questions we tried to answer with different NMR measurements can be found in the Supporting Information (SI) (see Scheme S1). Generally, in boron NMR, the 11B nucleus is used because its sensitivity is higher than that of 10B. Note that NMR using the Ti nucleus inherently cannot provide relevant information, as 47Ti and 49Ti possess a low sensitivity and as only a symmetric environment provides distinct signals.

Investigation of Pure C4mim BF4

The special part of our presented synthesis is the use of imidazolium-type ILs. In our previous works, we proved that the ILs act not only as a solvent but also play a crucial role in the formation of Ti(OH)OF·0.66H2O, as they provide fluorine, which is integrated into the HTB compound. Since evidently BF4– has to be hydrolyzed during the reaction, the IL anion acts as a reactant.[11,12]

Prior to interpreting the NMR spectra of mixtures, it is illustrative to discuss the NMR patterns of the pure compounds, for comparison. Figure shows the measured NMR spectra of pure C4mim BF4. The 1H NMR and 13C NMR spectra (Figure a,b) correspond to the structure of the IL cation, while the 11B NMR and 19F NMR spectra (Figure c,d) of C4mim BF4 show the typical signals of BF4–. It is noticeable that in the 19F NMR spectrum (Figure c) an isotopic chemical shift of the BF4– peak can be observed, which is caused by the two boron isotopes 10B and 11B. The calculated integrals of the two peaks indicate that these isotopes are present in a ratio of 1 (11B):0.25 (10B), which corresponds to the natural occurrence of the isotopes and therefore proves that the different chemical shifts are caused by these isotopes.[14,25] The 11B NMR spectrum (see Figure d) shows only one peak, which is quite broad due to the nuclear quadrupole moment of 11B.[26]

Figure 1

(a) 1H NMR (b) 13C NMR (c) 19F NMR, and (d) 11B NMR spectrum of pure C4mim BF4. All spectra were measured with 400 MHz at 298 K, and a solution containing 0.1 M trifluoroacetic acid (TFA) in dimethyl sulfoxide (DMSO)-d6 was used as external standard. For 11B NMR measurements, boron trifluoride etherate was used as a reference.

Interaction of the IL C4mim BF4 with Water

The first interaction we focus on is the interaction of C4mim BF4 with H2O. Therefore, we prepared a solution containing 3.85 mmol C4mim BF4 and 25 mmol H2O, which are the standard amounts in this synthesis (see the Experimental Section), and measured 1H, 13C, 19F, and 11B spectra of this solution. The solution was prepared and measured at room temperature.

The 1H and 13C spectra (see Figure S4) prove that the cation is not affected by the presence of water. Similar observations have been reported in the theoretical studies of similar systems, providing the radial distribution functions (RDFs) of different solutions.[24] The RDFs calculated for a mixture of IL and water (see Figure ) show that water interacts for the most part only with the polar components of the imidazolium cation, in this case with the most acidic hydrogen atom H4 of the imidazolium ring, which is reflected in the RDFs of O(H2O)-H4. In contrast, the terminal hydrogen atoms of the butyl side chain of the imidazolium cation (denoted as Hterm) are counted as nonpolar components and the corresponding RDFs between Hterm and water show hardly any signals. The different RDFs therefore demonstrate that water only exhibits interactions with H4 and not with Hterm. Therefore, we conclude that the butyl side chain is not affected by the addition of water, which is in agreement with the already mentioned 1H and 13C NMR results.

Figure 2

Calculated radial distribution functions (RDFs) between H(H2O)/O(H2O) and the most acidic hydrogen atom H4 and the terminal hydrogen atoms of the butyl chain Hterm. IW refers to a solution containing IL and H2O; IWT refers to a solution containing IL, H2O, and TiCl4. The data are adapted from ref (24).

The results of the 19F and 11B NMR measurements suggest that the anion is not affected either, as in the 19F spectrum still the characteristic signal of BF4– is present, with the two signals originating from isotopic chemical shift showing a ratio of 1 (11B):0.24 (10B) (see Figure a). It is notable that in both spectra the signals are shifted compared to the pure IL (see Figure ). This shift can be explained by the different concentration of C4mim BF4 in the measured solutions. Since no significant changes in the spectra were observed, it can be stated that the addition of water at room temperature does not affect the IL. No hydrolysis of BF4– seems to have taken place at the time of the measurement, although literature reports that hydrolysis of BF4– might occur at room temperature, but it needs several days (depending on the amount of water used) to produce a detectable amount of hydrolysis products (e.g., BF3(OH)−).[14,27] It is therefore understandable why no hydrolysis products were observed in our spectra since they were measured within a period of 12 h after mixing the IL with H2O.

Figure 3

(a) 19F NMR and (b) 11B NMR spectrum of a mixture containing C4mim BF4 and H2O with a molar ratio of 1:6.5 which was heated to 95 °C for 4 h. All spectra were measured with 400 MHz at 298 K, and a solution containing 0.1 M TFA in DMSO-d6 was used as external standard. For 11B NMR measurements, boron trifluoride etherate was used as reference.

After heating the solution for 4 h at 95 °C (typical reaction time and temperature, see the Experimental Section) and subsequent cooling to room temperature, it was possible to detect hydrolysis products of BF4–. The 19F NMR spectrum (see Figure a) contains, besides the BF4– signals, several signals in the range of −143.5 to −143.8 ppm. Based on the characteristic splitting of the peak, which was already reported in the literature, these signals can be assigned to BF3(OH)−, a hydrolysis product of BF4–.[14] The signal at −128.0 ppm can be attributed to SiF62–, which is generated as a result of the formation of small amounts of HF during the hydrolysis reacting with the glass of the NMR tube. The peaks corresponding to BF4– and BF3(OH)− are also visible in the measured 11B NMR spectrum (see Figure b). It can therefore be concluded that a higher reaction temperature leads to a faster hydrolysis of BF4–. At the same time, the IL cation is not affected since the 1H and 13C NMR spectra of this solution (see Figure S5) are comparable to the spectra of pure C4mim BF4.

Interaction of Different ILs with TiCl4

The next interaction we wanted to focus on was the interaction/reaction of C4mim BF4 with TiCl4. We assumed that an interaction is crucial for a successful synthesis of Ti(OH)OF·0.66H2O, in order to reduce the reactivity of TiCl4, establishing a stable solution in the first step of the synthesis (Scheme ). Otherwise, after the addition of H2O to the solution an immediate and heavy reaction of TiCl4 would occur, resulting in the formation of other titanium oxides (e.g., anatase) instead of the uncommon HTB compound.[28]

Hence, we prepared a sample containing 3.85 mmol C4mim BF4 and 1.82 mmol TiCl4 (standard ratio for a reaction leading to Ti(OH)OF·0.66H2O)[12] and measured different NMR spectra of this sample. The structure of the cation is not affected by the presence of TiCl4, since the 1H and 13C NMR spectra (see Figure S6) are comparable to the spectra of pure C4mim BF4. Interestingly the position of the peaks is shifted in both spectra. In the 1H NMR spectrum, all signals are shifted to higher values, with the shifts of the protons located on the imidazolium ring ranging from 0.16 to 0.18 ppm. The shifts of the protons of the alkyl side chain, on the other hand, are larger, ranging from 0.25 to 0.29 ppm. Voronoi analysis of a solution containing C4mim BF4 and TiCl4 performed in a previous study[24] indicated that the reference surfaces of TiCl4 are largely covered by the imidazolium ring, and possible surface coverage by the alkyl chain is prevented. The differing shift ranges observed in the 1H NMR spectrum are therefore a result of TiCl4 located mostly near the imidazolium ring. The visible shifts in the 13C NMR spectrum support this prediction since the shifts also vary depending on the position of carbon inside of the cation.

The 19F and 11B NMR spectra (see Figure ) clearly show that an interaction of the BF4– anion with TiCl4 must be present in this sample. The 19F NMR spectra (see Figure a) contain no longer the significant peak of BF4–observed for pure C4mim BF4. Instead, several other peaks, some of them being quite broad, can now be observed. The 11B NMR spectrum (see Figure b) of the mixture shows a quite broad signal, in comparison to the pattern of pure C4mim BF4, and the position of this peak is slightly shifted due to the different concentration of the IL and probably a different environment of the boron center. The broad signal visible in the spectrum can be explained by the different chemical environment of boron as well. Literature has shown that the line width in 11B NMR measurements strongly depends on the coordination and symmetry around the boron center. Moving from the highly symmetric BF4– to a less symmetric compound increases the line width therefore we can conclude that BF4– is no longer present, which is in agreement with the 19F NMR results.[29] These observations indicate some kind of interaction between the IL anion and TiCl4, while at the same time no interaction between the IL cation and TiCl4 takes place.

Figure 4

(a) 19F NMR and (b) 11B NMR spectrum of a mixture containing C4mim BF4 and TiCl4 in a ratio of 1:0.5. All spectra were measured with 400 MHz at 298 K, and a solution containing 0.1 M TFA in DMSO-d6 was used as external standard. For 11B NMR measurements, boron trifluoride etherate was used as a reference.

Comparable observations were found in recent theoretical studies using the ab initio molecular dynamic (AIMD) simulations.[24] In the solvation structure of TiCl4 in both pure and water-diluted systems, there are minor interactions between IL cations and TiCl4. In contrast, interactions between titanium and fluorine of tetrafluoroborate can be observed in the radial distribution functions Ti-F([BF4]−) in a mixture without water.[24]

The NMR spectra thus prove that a part of the fluorine atoms lies within a different chemical environment. Given the small number of compounds, presumably, the coordinative environment of Ti has changed. The change in coordination of Ti is further evidenced by the yellow color of the obtained solution, which is markedly more intense than pure TiCl4 (see Scheme ).

To elucidate the nature of such Ti complexes, we had to look closer into the measured NMR spectra. As mentioned above, there was no significant shift of the signal in the 11B NMR spectrum (see Figure b). We thus assumed that a significant portion of B–F bonds is unperturbed in this solution. It can be excluded that any compound containing B–Cl bonds was formed, as literature reports that the chemical shifts of comparable compounds are different from the observed shifts (BCl3: δ = 46.5 ppm; BClF2: δ = 20.0 ppm; BCl2F: δ = 31.2 ppm).[30] With this in mind, we concluded that there are still BF4– units present in our solution and that complexes containing TiCl4 and BF4– are built. Figure shows some of the possible complexes.

Figure 5

Possible complexes of TiCl4 and BF4–.

To clarify, which complexes are possibly formed, DFT calculations were performed. In particular, for structures (1) and (4) (Figure ), the electron, total thermal, and total enthalpic energies, as well as the Gibbs free enthalpy appear energetically favorable. Thus, according to the quantum chemical calculations, the formation of a binary TiCl4 complex via side-linking by tetrafluoroborate and the formation of a TiCl4 complex with one coordinated BF4– unit is realistic. As an important result, these simulations suggest the complexation proceeds via a direct linkage to fluorine. By contrast, the coordination of two tetrafluoroborate units to TiCl4 in cis or trans configuration (structures (2) and (3)) is unlikely. The calculated electron, total thermal, and total enthalpic energies, as well as the calculated Gibbs free enthalpies are given in Table S2 in the Supporting Information. Figure shows the sterical configuration of complexes (1) and (4) according to theoretical calculations. At first glance, it appears that the initial assumptions of complexes (1) and (4) (Figure ) agree well with the geometry-optimized structures shown in Figure . A comparison of our calculated bond lengths (see Figure ) with the bond lengths of literature-known compounds[31] (TiCl4 (bond lengths (Ti–Cl) = 2.17–2.18 Å), TiF4 (bond lengths (Ti–F) = 1.75–1.77 Å) and Ti2F8 (bond lengths (Ti–F) = 1.73–1.76 Å, bond lengths (Ti–F–Ti) = 1.89–2.13 Å)) show that our calculated bond lengths are in all cases slightly larger but they are on the same order of magnitude.

Figure 6

Sterical configuration of the two complexes (1) and (4) (numbering according to Figure ), based on theoretical calculations. Titanium atoms are depicted in gray, chlorine atoms in green, fluorine atoms in blue, and boron atoms are depicted in pink.

The assumption that linkage of two TiCl4 units via BF4– is possible can be inferred from the results in the literature[30] as well since it is shown that for two TiF4 units the linkage via fluorine to form Ti2F8 dimers is possible. This shows that a linkage via F–, where F– is connected to at least one titanium atom, is possible which is in agreement with our presented results. In contrast, the connection of two single TiCl4 units into a Ti2Cl8 dimer is unfavorable, in this case only weakly interacting van der Waals dimers are formed.

The quite broad signal in the 19F NMR spectrum (see Figure a) between −130 and −150 ppm can be attributed to a mixture of different complexes, which are, based on the theoretical calculations, complexes (1) and (4). We assume the broadness of the signal is caused by the different complexes in which different coordination environments of the fluorine are present. Interestingly, besides this broad maximum, there was an additional quite sharp signal at −131.3 ppm, which can be possibly interpreted as either isolated F– or a Ti–F bond in complex (1). To clarify which of the two possibilities applies, we performed reference measurements using C4mim F. It should be noted that the pure compound C4mim F is not stable; therefore, it was necessary to use a solution of this IL containing 5%w MeOH, the latter evidently impeding the interpretation of the NMR spectra. For comparison, NMR spectra of a solution containing C4mim F and 5%w MeOH were recorded (Figure ).

Figure 7

(a) 1H NMR, (b) 13C NMR, and (c) 19F NMR spectrum of a mixture containing C4mim F and 5%w MeOH. All spectra were measured with 400 MHz at 298 K, and a solution containing 0.1 M TFA in DMSO-d6 was used as external standard. (d) Structural formula of the IL.

The peaks in the 1H NMR spectrum were comparable to the spectrum of pure C4mim BF4 (see Figure a), except that no signal originating from H4 was visible. However, the C4 peak was still observable in the 13C NMR spectrum (see Figure b). Consequently, an exchange reaction at C4 probably occurred in this solution (C4mim F and 5%w MeOH), which is supported by the fact that two different peaks are visible in the 19F NMR spectrum (see Figure c). During the exchange reaction, hydride anions are released, which react with MeOH to form H2. Since the chemical has already been delivered as a mixture of C4mim F and MeOH, it was evidently impossible to observe the formation of H2, which would support this interpretation. With the help of literature-known compounds “AlkylFluor” (δ(C–F) = −107.51 ppm) and “PhenoFluor” (δ(C–F) = −34.15 ppm) (the structure of both compounds is shown in Figure S2), we were able to conclude that the imidazolium ring in this compound is still intact and that the peak at −108.7 ppm is related to a C–F bond at the C4 position (see Figure d).[17] The second peak at −145.3 ppm can be assigned to HF/F–, HF being formed in the reaction of F– with MeOH.[32]

After the addition of TiCl4 to the C4mim F-in-MeOH solution (in a molar ratio of approximately 0.5:1), a change in the measured NMR spectra can be observed. The 1H and 13C NMR spectra (see Figure S7) show that the overall structure of the cation is preserved. Interestingly, in this solution, the H4 atom is visible, which means that there is no C–F bond at position C4, which is also proven by the 19F NMR spectra (see Figure a) since the peak at −108.7 ppm is no longer visible. Instead, the 19F NMR spectrum exhibits three peaks with a quite low intensity. The peak at −150.3 ppm is attributable to the HF/F– peak, based on a comparison with the bare C4mim F-in-MeOH solution (see Figure c). For the other two peaks (−131.3 and −79.9 ppm), we assume titanium-fluorine compounds. Comparing them with the 19F NMR spectrum of the C4mim BF4/TiCl4 mixture (see Figure a), interestingly the peak at −131.3 ppm occurs in both solutions. Since TiCl4 and an F– containing IL anion are present in both solutions, this peak originates from species containing Ti–F bonds. In turn, for the synthesis using C4mim F a complex comparable to the complexes in Figure is present, with the difference that instead of the BF4– ligand now F– is bound to the Ti atom. Also, we conclude that in the case of the synthesis applying C4mim BF4, the BF4– unit is attached to Ti via a fluorine atom, and probably a chlorine–fluorine complex of the type TiClF is generated. This interpretation is in agreement with the already mentioned AIMD simulations, as such an interaction between fluoride and titanium was observed in the RDFs Ti-F([BF4]−) in a mixture without water.[24]

Figure 8

(a) 19F NMR spectrum of a mixture containing C4mim F (with 5%w MeOH) and TiCl4 in a ratio of approximately 1:0.5. (b) 19F NMR spectrum of a mixture containing C4mim F (with 5%w MeOH), TiCl4, and H2O in a ratio of approximately 1:0.5:6.9. For both spectra, the chemicals were mixed at room temperature and all spectra were measured with 400 MHz at 298 K. It was not possible to use a solution containing 0.1 M TFA in DMSO-d6 as external standard since the intensity of the 19F NMR signals of the solution is quite low in comparison to the peak intensity of TFA. This high intensity would mask the signals of interest.

The 19F NMR spectrum of the C4mim BF4/TiCl4 mixture (see Figure a) shows that the bridging F– atom has a different shift range (−131.3 ppm) from the other, free-moving F– attached to boron (broad signals between −130 and −150 ppm) due to its bonding with Ti. In addition, this interpretation explains the sharpness of the observed peak in comparison to the other visible peaks, since there is less movement of the bridging F– atom possible in comparison to the other fluorine atoms. Such different shift ranges of the bridging fluorine atom have also been observed for other metal complexes with BF4–.[33]

We assume that the peak at −79.9 ppm (Figure a) observed for the mixture of C4mim F (with 5%w MeOH) and TiCl4 can also be attributed to a Ti–F bond with the difference that in this case a substantial fraction of chlorine is replaced by methoxy groups (−OCH3), generated by the reaction of TiCl4 with MeOH. This is supported by the fact that gas formation (HCl) was observed after the addition of TiCl4 to the C4mim F/MeOH solution. It is therefore reasonable to assume that the peak at −79.9 ppm belongs to a Ti(OMe)Cl4–F species.

Such assumption is additionally supported by the fact that after adding water in excess to the solution containing C4mim F, MeOH, and TiCl4, the peak at −131.3 ppm (see Figure b) disappears, while the other two peaks are still present. In this solution, due to the high amount of water, no Cl–-containing titanium is left, explaining the absence of the −131.3 ppm signal. Also, in the presence of water TiCl4 swiftly reacts with H2O, forming Ti–OH bonds and gaseous HCl,[9] finally resulting in Ti(OH)Cl4–F, in which fluorine experiences a similar environment as in Ti(OMe)Cl4–F species, thereby causing the signal at −79.9 ppm.

Another evidence for the 19F NMR signal at −79.9 ppm belonging to F–- and OH–-containing titanium complexes is its nonappearance in the C4mim BF4/TiCl4 solution (Figure a), which is understandable in the light of the absence of H2O or MeOH. However, the peak pops up as soon as water is added, which will be discussed in detail in the Interactions Inside of the Reaction Solution section.

Solutions of C4mim BF4 with Titanium Isopropoxide (Ti[OCH(CH3)2]4, TTIP)

Based on the finding that different complexes containing fluorine and titanium can be detected in 19F NMR spectra, the proposed reactions and complexes were further studied using a different Ti precursor. Ti[OCH(CH3)2]4 (TTIP) was chosen, as here Ti is bonded to alkoxy groups, thus resulting in a substantially different hydrolysis behavior compared to TiCl4. Like TiCl4, TTIP is not stable against hydrolysis, and TiO2 is built as soon as TTIP gets in contact with water. It is thus necessary to stabilize the compound, for example with the help of a conc. aqueous HCl solution, as already reported in the literature.[34] Since a solution containing just C4mim BF4 and TTIP was not stable either, it was thus not possible to measure NMR spectra without the addition of conc. HClaq., therefore all NMR measurements with TTIP contain H2O. Table summarizes the amount of all precursors present in the two different investigated solutions with TTIP.

Table 1

Quantities of the Precursors in the Two Investigated Solutions Containing TTIP

solution	n(C4mim BF4) (mmol)	n(TTIP) (mmol)	n(HCl) (mmol)	n(H2O) (mmol)
IL + TTIP + HClaq	3.85	1.70	8.09	27.89
IL + TTIP + HClaq + H2O	3.85	1.70	8.09	52.89

Figure S9 shows the 1H, 13C, and 11B NMR spectra of a solution containing C4mim BF4/TTIP/HCl/H2O in a molar ratio of approximately 1:0.44:2.10:7.24. The 1H and 13C NMR spectra are comparable to the spectra measured for pure C4mim BF4, indicating that the IL cation is not affected in this solution. The 11B NMR spectrum shows a signal at approximately −1.75 ppm, which is comparable to the observed peak for pure C4mim BF4; therefore, it can be concluded that B–F bonds are still present.

In the 19F NMR spectrum (see Figure a) of the solution, it is possible to observe multiple signals in a shift range of −147 to −151 ppm. It is noticeable that all peaks with quite high intensities in this range have either a nearby, less intense peak or a shoulder. This finding indicates that these peaks can be assigned to compounds containing fluorine and boron since boron has two different isotopes, which influences the spectra (this finding was already explained in the Investigation of Pure C4mim BF4 section). The peaks at −149.46 and −149.52 ppm can, depending on their position and high intensity, be matched to BF4–. The integrals of these two peaks have a ratio of 1:0.24 being again in agreement with the natural occurrence of the two boron isotopes 11B and 10B.[14] The other observable peak can be assigned to different hydrolysis products of BF4– (BF4–OH–). In addition to the signals of the different B–F compounds, two other peaks are observed (−131.2 and −79.7 ppm). They are in agreement with the peaks observed in Figures a and 8, suggesting that comparable compounds (TiClF and Ti(OH)Cl4–F) are present. Interestingly, these results prove that not all isopropoxide units at TTIP were replaced with OH– units, although this would be possible by stoichiometry due to the amount of water present within the solution. Theoretical calculations have shown that large parts of H2O are located on the surface of the IL cation and anion in any solution containing C4mim BF4 and H2O.[24] Therefore, not all H2O molecules are available for the hydrolysis of every TTIP unit, resulting in the presence of the NMR signal indicative of TiClF (−131.2 ppm). It should be noted that the intensity of the peak at −131.2 ppm is quite low in comparison with the peak at −131.3 ppm in Figure . This finding can be explained by the fact that there is more H2O in this solution than MeOH in the C4mim F + 5%w MeOH + TiCl4 solution. As a result, more isopropoxide units are already replaced by H2O, which decreases the intensity of the peak at −131.2 ppm. The low intensity of the peak at −79.7 ppm can be explained by the lack of F– inside of the solution since the hydrolysis of BF4– is inhibited due to the low amount of H2O inside of the solution.

Figure 9

(a) 19F NMR spectrum of a mixture containing C4mim BF4, TTIP, and conc. HClaq. in a molar ratio of approximately 1:0.44:2.10 (HCl):7.24 (H2O). (b) 19F NMR spectrum of a mixture containing C4mim BF4, TTIP, conc. HClaq, and H2O in a molar ratio of approximately 1:0.44:2.10 (HCl):13.74 (H2O). All spectra were measured with 400 MHz at 298 K, and a solution containing 0.1 M TFA in DMSO-d6 was used as external standard.

The presence of similar signals proves that, regardless of the used titanium precursor, similar detectable titanium complexes are built.

After the addition of water in excess, the peak at −131.2 ppm can no longer be detected, while the signal at −79.7 ppm is still present (see Figure b). This observation was already noticed after the addition of water to a solution containing C4mim F, MeOH, and TiCl4 (see Figure b) which is another evidence that comparable complexes are built in both solutions. In the region from −145 to −151 ppm, again several peaks can be observed, but there are also a few changes detectable, compared with the spectrum measured for the solution without water (see Figure a). This can be explained by the fact that the hydrolysis of BF4– can take place to a larger extent due to the larger amount of water, and thus different signals of the hydrolysis products (BF4–OH−) can occur. The cation of the IL is not affected in this solution (for 1H NMR and 13C NMR spectra, see Figure S10a,b), the 11B NMR spectrum (see Figure S10c) is comparable to the spectrum measured for C4mim BF4 + TiCl4 + H2O (see Figure b).

Figure 10

(a) 19F NMR and (b) 11B NMR spectrum of a mixture containing C4mim BF4, TiCl4, and H2O in a ratio of approximately 1:0.5:6.5. The chemicals were mixed at room temperature, and all spectra were measured with 400 MHz at 298 K, and a solution containing 0.1 M TFA in DMSO-d6 was used as external standard. For 11B NMR measurements, boron trifluoride etherate was used as reference.

Interactions inside of the Reaction Solution

Building on the insight provided by the comparative experiments described above, we now aim at understanding the reactions and interactions in the actual reaction solution used to synthesize Ti(OH)OF·0.66H2O. Since the color of this solution, i.e., C4mim BF4 + TiCl4 + H2O, is different from a solution containing only C4mim BF4 + TiCl4 (see Scheme ), it can be assumed that the previously formed complexes of Ti with chlorine and BF4– as ligands (see Figure ) convert into other Ti complexes due to reactions with water. To peer further into the molecular structures and reactions, 1H, 13C, 19F, and 11B NMR spectra were measured. The 1H and 13C NMR spectra (see Figure S11) confirm the IL cation being unaffected since the spectra are comparable to the spectra of pure C4mim BF4 (see Figure a,b). The 19F and 11B NMR spectra, on the other hand, are different from the previously measured spectra. The 11B NMR spectrum (see Figure b) shows a broad and sharp signal in the range of approximately −1.5 to −2.0 ppm. Based on their position we attribute these signals to different B–F bonds. As discussed earlier, the chemical shift of boron bonds to chlorine is in a different range, and therefore it can be excluded that such compounds are present in this solution. The 19F NMR spectrum reveals several different signals (see Figure a). The peaks at −149.67 and −149.61 ppm are in conformity with BF4– because of the high intensity and the integral ratio (1:0.25). This finding is quite interesting, if not amazing, because one would assume the more or less complete hydrolysis of BF4– in such an acidic mixture. Note that these signals did not appear in a solution containing C4mim BF4 and TiCl4 (Figure ). The only explanation for the BF4– related 19F NMR signals coming up after the addition of water again is that complexes of Ti with BF4–as ligand (see Figure ) were formed and that BF4– is released from this kind of complexes upon addition of water. In addition, gas formation is observed after the addition of water to the reaction solution. The generated gas was HCl, which is formed in a reaction of the TiCl4 complex with water. Thus, after the addition of water, OH– is bound to Ti. It is not possible to confirm if at this point of the reaction every Cl– is replaced with OH–, therefore we refer to the resulting complex as Ti(OH)Cl4–F.

The experimental observations are in good agreement with already published theoretical works, based on domain and Voronoi analyses.[24] In these analyses, the molecules or ions, as well as possible functional groups, are divided into subsets. Thereby, the subsets are investigated with respect to their connectivity and their neighborhood behavior. In the already published study,[24] the subsets were divided into polar, nonpolar, TiCl4 and water domains. It was found that the presence of water disturbs the microheterogeneous structure of the whole system, especially the microheterogeneous structure of the nonpolar and TiCl4 domains, which is manifested by a more scattered ordering of these domains. The addition of Voronoi analysis shows the surface coverage of the respective molecular or ionic moieties. In particular, for the system with all components, it is shown that the reference surfaces of titanium tetrachloride and tetrafluoroborate are largely covered by water and possible surface coverage by the cation is prevented. This disturbs the molecular order of the ionic liquid and possible interactions between cations and anions, and therefore the complexation between TiCl4 and BF4– is prevented in the presence of water.

The 19F NMR peaks at about −147.0 ppm (see Figure a) can be assigned, based on the characteristic splitting of the peak (for reference, see Figure a), to the compound BF3(OH)−. The appearance of peaks, which can be assigned to different hydrolysis products of BF4– (BF4–OH–), clearly proves that the hydrolysis must proceed via a different mechanism in the presence of TiCl4, since in this case the hydrolysis can take place quite fast at room temperature. It was not possible to detect hydrolysis products in a solution containing C4mim BF4 and H2O after a comparable waiting time (see Figure a), therefore it can be assumed that the hydrolysis proceeds only to a small extent via a direct interaction of BF4– and H2O, instead it mainly occurs via interactions between BF4– and OH– ligands bound to titanium. This finding is in agreement with thermodynamical calculations already published in a previous study of our working group.[9]Table summarizes the calculated interactions between BF4–OH– and Ti(OH) out of the mentioned study.

Table 2

Thermodynamic Calculations of the Reaction Energy ΔE and the Free Reaction Enthalpy ΔG (in kJ/mol) of Interactions between BF4–OH– and Ti(OH), Performed at a Temperature of 370 K, Taken from the Literature[9],a

		ΔE	ΔG
(1)	[Ti(OH)4] + BF4– → [Ti(OH)3F] + BF3(OH)−	–4.1	–2.0
(2)	[Ti(OH)5]− + BF4– → [Ti(OH)4F]− + BF3(OH)−	–32.2	–29.7
(3)	[Ti(OH)6]2– + BF4– → [Ti(OH)5F]2– + BF3(OH)−	–26.7	–28.9
(4)	[Ti(OH)4] + BF3(OH)− → [Ti(OH)3F] + BF2(OH)2–	–2.5	–0.6
(5)	[Ti(OH)5]− + BF3(OH)− → [Ti(OH)4F]− + BF2(OH)2–	–30.7	–28.3
(6)	[Ti(OH)6]2– + BF3(OH)− →[Ti(OH)5F]2– + BF2(OH)2–	–25.1	–27.5
(7)	BF4– + 2 H2O → BF3(OH)− + F– + H3O+	124.4	150.9

Adapted with permission from Voepel, P.; et al., Cryst. Growth Des. 2017, 17, 5586–5601. Copyright 2017 American Chemical Society.

These calculations show that, from a thermodynamic point of view, the hydrolysis of BF4– can proceed in this way since the values for ΔE and ΔG are negative for all of the calculated reactions. It should be noted that the calculations were performed for a temperature of 370 K, compared to the here applied temperature (298 K). However, we believe that the sign and magnitude of the values in Table are not severely different to T = 298 K, as the entropic contribution ΔS is moderate. Also, the peak at −79.8 ppm observed in the 19F NMR spectrum (Figure a), which we relate to Ti–F bonds, supports the view that peaks at this position can be assigned to a titanium complex with OH– and F– as ligand. Hence, the theoretical thermodynamic parameters (see Table ) and NMR data suggest that the hydrolysis of BF4– does not occur just by the action of water, but requires the presence of Ti compounds in the solution. Interestingly the values of ΔE and ΔG for hydrolysis of BF4– with water without TiCl4 at 370 K are positive (see Table (7)). Therefore, from a thermodynamic point of view, the hydrolysis of BF4– with water is not favored. This explains the high amount of BF4–, which is still present after heating a solution with C4mim BF4 and H2O for 4 h at 95 °C (see Figure ).

As the main outcome with respect to elucidating the overall synthesis, the hydrolysis of BF4– can already take place at room temperature, mediated by the Ti compound, i.e., surprisingly the heating step is not crucial to spur the release of fluorine from BF4–. In addition, the presence of the peak at −79.8 ppm proves that Ti(OH)Cl4–F complexes are already present at room temperature, by systematically comparing the absence and presence of this signal in all measured solutions (see Table ).

Table 3

Summary of the Solutions in Which the Signal at Approximately –79.8 ppm (19F NMR) Was Detected and in Which It Was Absenta

solution	peak at approx. –79.8 ppm? (19F NMR)	does the solution contain a titanium species?	does the solution contain H2O/MeOH?
C4mim BF4	no	no	no
C4mim BF4 + H2O	no	no	yes
C4mim BF4 + TiCl4	no	yes	no
C4mim BF4 + TiCl4 + H2O	yes (−79.83 ppm)	yes	yes
C4mim F + MeOH	no	no	yes
C4mim F + MeOH + TiCl4	yes (−79.85 ppm)	yes	yes
C4mim F + MeOH + TiCl4 + H2O	yes (−79.85 ppm)	yes	yes
C4mim BF4 + TTIP + HCl	yes (−79.67 ppm)	yes	yes
C4mim BF4 + TTIP + HCl + H2O	yes (−79.67 ppm)	yes	yes

The presence/absence of this signal serves as proof for the presence of Ti(OH)Cl4–F in the reaction solution (C4mim BF4 + TiCl4 + H2O) already at room temperature.

It is noticeable that the peak at approx. −79.8 ppm can be detected in solutions containing an IL (C4mim BF4 or C4mim F), a titanium precursor (TiCl4 or TTIP), and H2O/MeOH, which is a clear proof that the peak can be assigned to Ti(OH)Cl4–F complexes (Ti(MeOH)Cl4–F if only MeOH is present).

We now focus on unraveling the heating step, which is inevitable to obtain Ti(OH)OF·0.66H2O. To tackle this question, the prepared reaction solution was heated up to 95 °C for 4 h (typical reaction time, see the Experimental Section) inside the NMR tube. After cooling down to room temperature, NMR spectra of the resulting solution were acquired. It should be noted that upon heating nanoparticles were formed, affecting the intensity of the NMR spectra. The 1H NMR and 13C NMR spectra (see Figure S12) prove that the IL cation is unaffected by the heating. Surprisingly, the 19F and 11B NMR spectra (see Figure ) were comparable to the spectra measured prior to the heating step. As already discussed above, the heating step does therefore not substantially initiate or accelerate the hydrolysis of BF4–. Instead, the treatment at 95 °C induces the formation of Ti(OH)OF·0.66H2O nanoparticles out of the Ti(OH)Cl4–F complexes, i.e., the condensation of single Ti-containing entities into the crystalline array by the release of water. As already mentioned it is not possible to confirm if there are still Cl– units present in the Ti complex prior to the heating step. However, X-ray photoelectron spectroscopy (XPS) and thermogravimetric analysis-mass spectrometry (TGA-MS) results of the finished product published in a previous study of our working group[11] proved that no Cl– is present within the product. Therefore, the remaining Cl– must be released from the complex during the heating step.

Figure 11

(a) 19F NMR and (b) 11B NMR spectrum of a mixture containing C4mim BF4, TiCl4, and H2O in a molar ratio of approximately 1:0.5:6.5. The solution was heated at 95 °C for 4 h, and after that, the solution was cooled down for the measurements. All spectra were measured with 400 MHz at 298 K, and a solution containing 0.1 M TFA in DMSO-d6 was used as external standard. For 11B NMR measurements, boron trifluoride etherate was used as reference.

Another approach to peer into the details of the synthesis is the analysis of the hydrolysis products after the reaction, and therefore, we analyzed the solutions obtained within the washing step as part of the synthesis (see the Experimental Section). Thus, we performed three washing steps after the synthesis and measured the NMR spectra of the washing solutions. Figure shows relevant parts of the 19F and 11B NMR spectra, the full 19F NMR spectra and the 1H and 13C NMR spectra can be found in the SI file (see Figures S13–S15). In the first and second washing steps, hydrolysis products of BF4– were detected, namely, BF4–(OH)–. It was possible to observe the same signals in both washing steps, although in comparison the position of each signal is shifted. The shift can be explained by a different concentration of the respective species. In the third washing step, in contrast, these species were no longer detectable. Since the spectra of the first two washing steps are comparable to the measured spectra of the reaction (see Figure ), we can conclude that the washing steps are crucial for the purification of the different products, but they do not influence the reaction itself.

Figure 12

Washing steps of the produced nanoparticles. In each washing step, 2 mL of abs. EtOH was used. All spectra were measured with 400 MHz at 298 K, and a solution containing 0.1 M TFA in DMSO-d6 was used as external standard. For 11B NMR measurements, boron trifluoride etherate was used as reference.

Summary and Conclusions

In this work, we investigated the mechanism of an IL-based synthesis of the special fluorine-containing solid Ti(OH)OF·0.66H2O possessing a peculiar hexagonal tungsten bronze (HTB)-type structure. The synthesis is puzzling with respect to the simplicity of the procedure generating such a distinct crystalline solid, involving simple “beaker chemistry”, just heating a solution of commonplace chemicals, namely, the simple ionic liquid C4mim BF4, TiCl4, and H2O, as well as applying quite moderate temperature in the final heating step (95 °C). While previous studies had already indicated that the BF4– anion plays a vital role by releasing fluorine and thus generating F-containing Ti clusters, the first reaction steps and products in this solution were still a matter of discussion. In particular, it had remained unclear which reaction steps take place in the mixture already at room temperature and which reactions are spurred by heating at 95 °C.

Here, we peered into the birth of the first Ti complexes generated upon reaction with BF4– and H2O upon mixing already at room temperature, by the help of 1H, 13C, 11B, and 19F NMR spectroscopy measurements. For this purpose, we performed various NMR measurements of solutions with different, systematically varied compositions of the reactants. Surprisingly, in the first step of the reaction a complex containing TiCl4 and BF4– is formed, already upon mixing at room temperature. Advanced quantum chemical calculations showed that, for instance, a binary TiCl4 complex, formed via side-linking by tetrafluoroborate and a TiCl4 complex with one coordinated BF4– unit are plausible and possible. 19F NMR measurements performed on systematically varied solutions support these theoretical results in that the BF4– ligand in such a complex is bound to titanium by a bridging fluorine atom: the bridging fluorine atom has a different shift range compared to the nonbridging fluorine atoms. This complexation is probably a crucial step for the synthesis, for instance, because it prevents the strong hydrolytic reaction between TiCl4 and H2O, which needs to be addressed by further theoretical studies.

After the addition of H2O to the solution containing IL and TiCl4, the signal originating from isolated BF4– units appeared again in the 19F NMR spectrum, which shows that the complex between TiCl4 and BF4– was destroyed, by forming Ti–O–bonds. In addition, a new peak at approximately −79.8 ppm (19F NMR spectrum) was detected, being attributable to a Ti(OH)Cl4–F complex. Again, the systematic comparison of 19F NMR spectra of different solutions provided ample evidence for this complex, as this peak is only detectable in solutions containing an IL, a titanium precursor, and H2O/MeOH (see Table ). It is important to note that BF4– does not undergo significant hydrolysis in H2O at room temperature. This finding is supported by positive ΔG values calculated for the reaction of BF4– with water without TiCl4 at 370 K (see Table ). Hence, it is the presence of the Ti species that initiates the decomposition of the BF4– anion already at room temperature.

The experimental proof for such single Ti(OH)Cl4–F species being formed already at room temperature represents one of the major insights and advancements of this study, especially because it is difficult to predict the position of such signals even by advanced DFT methods. A further surprising result is thus the unexpected pronounced hydrolysis of BF4– in the presence of Ti compounds, already at room temperature, while a mixture of C4mim BF4 and H2O exhibits a comparably slow formation of fluoride ions. This faster hydrolysis at room temperature is thus due to the fact that it occurs through interactions between Ti(OH) and BF4– instead of interactions between H2O and BF4–. Since the hydrolysis of BF4– therefore does not require the reaction temperature (95 °C), we conclude that the elevated temperature of 95 °C is only crucial for the condensation of the built Ti(OH)Cl4–F complexes, to overcome the activation energy and thus to spur the formation of Ti–O–Ti bonds and H2O.

The crystal structure of Ti(OH)OF·0.66H2O has been investigated in detail in a previous work of our working group.[11] With the help of XRD measurements and Rietveld refinements, it was proven that the crystal structure is built up of corner-sharing Ti(X)6 octahedra (X = O, F) and that the fluorine atoms occupy the apical positions of the built octahedra. To obtain this apical occupation in the product, it is therefore necessary that the fluorine atoms are arranged in trans position in the built Ti(OH)Cl4–F octahedra prior to the condensation step, as shown in Scheme . Based on these results, it is now possible to propose an adapted overall reaction mechanism for the investigated synthesis (see Scheme ).

Scheme 2

Schematic Illustration of the Investigated Reaction Mechanism

Hence, this study provides important insight into the first steps of the reaction, in particular, why the simple synthetic procedure is able to generate Ti(OH)OF·0.66H2O at moderate temperature, and to clarify the role of the IL: the IL provides fluorine in the form of the IL anion BF4–, spurs the hydrolysis of BF4– already at room temperature in the presence of Ti species, and, moreover, is able to provide a stable solution of the involved compounds as well as the formed Ti complexes. In the light of previous works using imidazolium-based ILs for the synthesis of metal oxides, it is therefore justified to say that there is a specific “IL effect” for this type of synthesis, which involves the hydrolysis of transition-metal compounds. Hence, our study might help to develop strategies to synthesize other transition-metal oxyfluorides under ambient conditions.

Experimental Section

Synthesis Procedure

All ILs used in this work were purchased from IoLiTec (1-butyl-3-methylimidazolium tetrafluoroborate (C4mim BF4), purity: >99%, product code: IL-0012; 1-butyl-3-methylimidazolium fluoride methanol adduct (C4mim F), purity: 95% IL in methanol, product code CS-1608M). HCl, TiCl4, and titanium isopropoxide (TTIP, purity: 97%, product code: 205273) were purchased from Merck. All chemicals were used without further purification or modification.

In a typical synthesis, 3.85 mmol of IL (870.2 mg of C4mim BF4 or 609.2 mg of C4mim F) are heated up to 95 °C (80 °C for C4mim F) and mixed with 0.2 mL of TiCl4 (1.82 mmol). After stirring the yellow and transparent solution for at least 5 min, 0.45 mL (25 mmol) of H2O was added dropwise (caution: heavy reaction of TiCl4 with water under the release of HCl and potentially also HF). The solution was heated for 4 h at 95 °C (24 h at 80 °C for reactions with C4mim F). After a few minutes, it was possible to observe an increasing opacity of the reaction solution, indicating the formation of nanoparticles inside of the solution. The built nanoparticles were washed four times with technical ethanol and dried after the reaction suspension was cooled down. To prove that the received product is composed of Ti(OH)OF·0.66H2O and TiO2(B), we performed XRD measurements and Rietveld refinements of the synthesis product in the past.[12] The results of these measurements can be found in the SI (see Figure S1 and Table S1). The single steps of this synthesis are documented by corresponding photographic images in the Supporting Information of ref (9).

In this work, we focus mainly on the interaction of different reactants. Therefore, solutions containing mixtures of different reactants, in the same ratio used in the described typical synthesis, were produced and NMR spectra were measured of these solutions (see Scheme ).

Synthesis with TTIP

In a typical synthesis, 870.2 mg of C4mim BF4 (3.85 mmol), 0.518 mL of TTIP (purity: 97%, 1.7 mmol), and 0.67 mL of conc. HClaq. (37% solution, 21.87 mmol) were mixed at 95 °C. After stirring for at least 5 min, 0.45 mL of H2O (25 mmol) was added to the solution and the solution was heated for 4 h at 95 °C. After a few minutes, we observed an increasing opacity of the reaction solution, indicating the formation of nanoparticles inside of the solution. The nanoparticles were washed four times with technical ethanol and dried after the reaction suspension was cooled down.

Preparation of NMR Samples

For the NMR measurements, a solution of 0.1 M trifluoroacetic acid (TFA) in DMSO-d6 was used as standard.[17] To avoid interaction of the standard with the analyte, which could possibly distort the results, the analyte was sealed into a small capillary, which was then placed inside of an NMR tube. Figure S16 shows a schematic illustration of the prepared NMR samples. For some samples, it was not possible to use the 0.1 M TFA-in-DMSO-d6 solution as a standard since the peak intensity of the compounds was too low (e.g., Figure ). In these cases, pure DMSO-d6 was used inside of the capillary and trichlorofluoromethane was used as a standard for the 19F NMR measurements.

Instrumental Settings

The NMR spectra were measured with a Bruker Avance III 400 MHz HD and a Bruker Avance II 400 MHz spectrometer. All spectra were measured with 400 MHz at 298 K. The chemical shifts δ are reported in parts per million (ppm) relative to the solvent signal of DMSO-d6 (1H and 13C NMR measurements) or the solvent signal of the 0.1 M TFA in DMSO-d6 solution (19F NMR measurements, the shift of this external standard solution was reported in the literature).[17] For 11B NMR measurements, boron trifluoride etherate was used as reference. The coupling constants J for 1H NMR measurements can be found in the SI.

The starting geometries of the structures to be investigated were initially built using the software package MOLDEN (version 5.4).[18] Subsequently, all quantum chemical calculations were performed using the ORCA 4.0 program.[19] The geometry-optimized structures were performed using the B3LYP functional[20−22] and the def2-TZVPP basis set.[23] Tight convergence criteria were applied for the SCF cycle and geometry optimization. To verify if the obtained structures were ground states, it was ensured that the Hessian did not have negative eigenvalues for minima. To take into account the solvation effects, approximations were performed with the conduction-like polarizable continuum model (CPCM), which is also possible with the ORCA program with the functional and basis sets mentioned above. A dielectric constant of 38.3 was used, which reflects well the mixture of water and C4mim-based ionic liquids.

The experimentally obtained results are compared not only with static DFT calculations but also with the results of theoretical investigations of so-called ab initio molecular dynamic (AIMD) simulations previously published.[24] The more detailed description, explanation, and setup, as well as the procedure of these are explained in detail in this recent publication.[24] The compositions of the simulation boxes reflect the compositions of the used reaction mixtures.