Dioximate- and Bis(salicylaldiminate)-Bridged Titanium and Zirconium Alkoxides: Structure Elucidation by Mass Spectrometry

The treatment of titanium alkoxides with 1,5-pentanedioxime or 2,5-hexanedioxime resulted in the formation of complexes [{TiL(OR)2}2] in which the dioximate ligands (L) bridge a dimeric Ti2(μ2-OR)2 unit. The structures of the complexes were determined by single-crystal structure analysis, ESI mass spectrometry, and 1D and 2D solution NMR spectroscopy. In contrast, the treatment of titanium alkoxides with dioximes bearing cyclic linkers, such as cyclohexyl or aryl groups, resulted in insoluble polymeric compounds. The treatment of various bis(salicylaldiminates) with titanium and zirconium alkoxides resulted in compounds with the same composition [{TiL(OR)2}2], in which, however, two monomeric Ti(OR)2 units are bridged by the ligands L. The two structural possibilities can be distinguished by low-energy collision-induced dissociation owing to their different fragmentation patterns.


Introduction
Metal alkoxides are common precursors for sol-gel processing. Modification with bidentate organic ligands, such as b-diketonates, b-ketoesterates, carboxylates, aminoalcoholates, or oximates, lowers their reaction rates and offers the possibility of introducing functional organic groups for the formation of inorganic-organic hybrid materials. [1] The bidentate ligands are retained largely during sol-gel processing.
Bifunctional ligands YÀXÀY (Y= bidentate coordinating group, X = spacer) have been used rarely for the modification of metal alkoxides. They can be chelating or bridging and result in either polymers of the type [(RO) n MÀYÀXÀY] 1 or cyclic compounds [(RO) n MÀYÀXÀY] m . Such metal alkoxide derivatives offer the possibility of obtaining structured metal oxides after sol-gel processing, similar to alkoxysilane derivatives (RO) 3 SiÀYÀSi(OR) 3 .
A few metal alkoxide derivatives with bifunctional ligands have been reported, but systematic studies were only performed in a few cases. Polymeric structures were previously only obtained by reaction of titanium and zirconium alkoxides with diamines. [2] These adducts are, however, not suitable for sol-gel processing because of the hydrolytic instability of the TiÀN bond. The tetrameric complex shown in Scheme 1 was obtained from 2,4-pentanedioxime, in which the dioximate ligands both bridge dimeric titanium alkoxide units and interconnect two of the dimeric units. [3] In the absence of an additionally coordinating substituent the salicyclaldiminate-substituted derivatives [M(OR) 2 (SA) 2 ] (M= Zr, Ti; SA = salicylaldiminate) have the same geometry as the corresponding complexes [M(OR) 2 (b-diketonate) 2 ], that is, with the OR groups in positions cis to each other. [4] This structural motif was also found in a complex in which the two SA ligands are connected through a C 6 H 4 ÀSÀSÀC 6 H 4 bridge (Scheme 1, bottom left) [5] as The treatment of titanium alkoxides with 1,5-pentanedioxime or 2,5-hexanedioxime resulted in the formation of complexes [{TiL(OR) 2 } 2 ] in which the dioximate ligands (L) bridge a dimeric Ti 2 (m 2 -OR) 2 unit. The structures of the complexes were determined by single-crystal structure analysis, ESI mass spectrometry, and 1D and 2D solution NMR spectroscopy. In contrast, the treatment of titanium alkoxides with dioximes bearing cyclic linkers, such as cyclohexyl or aryl groups, resulted in insoluble polymeric compounds. The treatment of various bis(salicylaldiminates) with titanium and zirconium alkoxides resulted in compounds with the same composition [{TiL(OR) 2 } 2 ], in which, however, two monomeric Ti(OR) 2 units are bridged by the ligands L. The two structural possibilities can be distinguished by low-energy collision-induced dissociation owing to their different fragmentation patterns.
The way in which the bifunctional ligands are coordinated and the kind of compounds that are formed depends not only on the nature of the coordinating groups Y and the length and rigidity of the spacer X, but also on the coordination geometry at the metal center. We therefore extended our studies in the current work on reactions of M(OR) n (M = Ti, Zr) with dioximes and bis(salicylaldiminates) with various spacer groups X (Scheme 2).
The TiÀO and TiÀN bond lengths and selected angles are in a comparable range as in the previous investigated compounds [{Ti(OiPr) 2 (ON=CHR') 2 } 2 ]. [3,9] The alkoxo bridges are slightly asymmetric, with TiÀ O bridging bond lengths ranging from 202.5 to 206.2 pm. The oximate groups bonded to the same titanium center are nearly coplanar.
The covered m/z range was increased to higher values to be able to monitor high molecular mass species and to eliminate the formation of oligomeric or polymeric compounds. In both cases only the molecular-ion peak was present in the mass spectrum. Increasing the Ti(OiPr) 4 /dioxime ratio resulted in the same sodiated molecular species, while the intensity of [Ti-(OiPr) 4 +Na] + at m/z 307.1 (calcd 307.1) increased. Thus, the formation of mono-substituted titanium-alkoxo-oximate derivatives can be excluded.
Solution NMR spectroscopy also proved the coordination of both oximate groups to the titanium centers. The shift of the C=N signal in the 13 C NMR spectra changed from d = 155.3 and 150.5 ppm in the starting oxime to d = 144.8/146.0 and 139.4 ppm in compounds 1-3, which is in agreement with previously investigated titanium-alkoxo-oximate derivatives. Resonances for the CH/CH 2 groups of two different alkoxo groups were observed in all 1 H NMR spectra, at d = 4.47/3.37 ppm for 1, d = 4.35/3.91 ppm for 2, and d = 4.34/3.64 ppm for 3. Similar observations were also made in the corresponding 13 C NMR spectra. NMR spectroscopy also revealed two different signals for the CH 3 group of L 1 in compounds 1 and 2.
Because of its better solubility, compound 1 was investigated by 2D NMR spectroscopy. Seven different CH signals were observed in the HSQC spectrum at room temperature ( Figure 4). This can be explained by different conformations of the nine-membered ring, formed by a dioximate and a bridging OiPr ligand (i.e., ÀTiÀO br ÀTiÀNÀCÀCÀCÀCÀNÀ), which leads to independent OiPr signals.
The exchange spectroscopy (EXSY) spectrum at room temperature ( Figure 4a) showed that the compound is highly dynamic. Exact interpretation is difficult, as the methylene signals are not split. Only two to three signals were observed, although four signals are expected for this AB system. Two ex-change signals at d = 3.7/4.5 and 4.5/4.85 ppm were observed. The resonance at d = 3.7 ppm was assigned to free 2-propanol; the first set of exchange signals is therefore caused by the exchange of residual 2-propanol and a terminal OiPr group. The second exchange signal (d = 4.5/4.85 ppm) was attributed to an exchange between different OiPr groups. One possible explanation is active exchange between terminal and bridging OiPr groups. A possible mechanism is shown in Figure 5; opening of the OiPr bridges would result in a dimer only bridged by the dioximate ligands. Rotation of the metal alkoxide moieties and recombination of the bridges would result in exchanged alkoxo groups.
The second possibility is a passive exchange in which the bridging OiPr groups follow the dynamics of the alkylene  Temperature-dependent NMR spectroscopy experiments were conducted between + 20 and À80 8C ( Figure 6). The methine signal of the OiPr group and that of the NÀCH 2 group in the 1 H NMR spectrum were broad. When lowering the temperature, the methine signal at d = 4.47 ppm broadened into two independent signals, that is, more conformers can be distinguished. The COSY spectrum at À60 8C revealed eleven to twelve independent CH signals, which proved the existence of at least three different conformers at low temperature. The previously observed exchange signals disappeared in the EXSY spectrum at À60 8C (Figure 4b). The dynamics of both processes were therefore minimized.
A similar solution structure is postulated for 2 and 3 based on NMR spectroscopy and MS measurements.

Cyclohexanedioximes
Bridging of the titanium atoms in an alkoxo-bridged dimeric Ti 2 (OR) 4 unit (type A in Scheme 3) is only possible if the spacer between the two coordinating units is flexible enough to adjust to the TiÀTi distance in this unit, which is determined by the geometry of the central Ti 2 O 2 ring. Stiffening of the spacer could therefore induce the compounds to adopt a differ-ent structure. Reaction of one molar equivalent of Ti(OEt) 4 with 1,3-or 1,4-cyclohexyldioxime resulted in the barely soluble, amorphous compounds 4 and 5. Both were only reasonably soluble in ethanol at elevated temperature. As discussed below, there is evidence, however, that the compounds degrade upon dissolution in ethanol. The solid-state 13 C NMR spectrum of 4 showed a clear shift of the oximate carbon to d = 143.0 ppm, which is proof of coordination of the oximate groups. [3,9] The spectrum also revealed the presence of OEt groups by the signal at d = 68.7 ppm assigned to the OCH 2 groups. The shift of the oximate nitrogen in the 15 N NMR spectrum from d = 278.8 ppm in L 3 H 2 to d = 274.0 ppm in 4 confirmed the coordination of the oximate groups.
ESI-and MALDI-MS measurements (the latter not described here in detail) of solutions of 4 and 5 in ethanol did not result in signals for compounds of reasonable composition related to the charged molecule, but only ions representing L 3 H 2 and titanium alkoxides were observed. A possible explanation for this is that the compounds do not dissolve in ethanol but are instead degraded in solution.
The very low solubility of products 4 and 5 indicated a polymeric structure. A glass transition was observed by differential scanning calorimetry (DSC) at approximately 80 8C for 4, which would be in line with a polymeric structure. No glass transition was observed for 5 and therefore equilibria between oligomeric and polymeric structures must also be considered. A polymeric structure could be based on either the type A or the type B motif (Scheme 3), that is, with or without bridging OR ligands.
Reaction of titanium alkoxides with aryl-bridged dioximates results in compounds with similar properties and therefore analogous polymeric structures are proposed.
Information on the composition of 6-10 was gained from ESI-MS. An intense molecular-ion peak with the composition [{ML(OR) 2 } 2 ] was observed for compounds 6-10. The titanium species in 6, 8, and 9 were detected as the sodiated molecule, whereas the zirconium compounds 7 and 10 were detected as the sodiated and chloro-adduct molecules. As an example of all the above-mentioned compounds, the positive-ion ESI mass   2 -(OiPr) 4 +Na] + ). The calculated molecular mass corresponds to a dimeric structure with two bridging bis(salicylaldiminate) ligands. MS/MS (low-energy CID) experiments were conducted to determine whether the compound is of type A or B (Scheme 3). A single peak at m/z 511.2 (calcd 511.2) in the MS/ MS spectrum was assigned to [TiL 5 (OiPr) 2 +Na] + . Similar fragmentation was previously observed for various bis(b-diketonate)-substituted titanium alkoxide derivatives. [8] This fragmentation pathway indicated a type B structure (Scheme 3). The MS/MS measurement also revealed that the signal at m/z 511.3, which appeared also in the full-scan mass spectrum (m/z 511.2), was only caused by in-source fragmentation and was not due to another compound (or contamination) in the solution. The peak appearing at m/z 775.4 (calcd 775.3) was attributed to [(TiL 5 2 )iPrOH+Na] + , an ion corresponding to the monomeric titanium complex with two coordinated bis(salicylaldiminate) ligands. As this signal did not appear at all in the MS/MS spectrum (lower spectrum in Figure 7), it was no fragment ion, but instead a byproduct of the reaction. Corresponding signals did not appear in the mass spectra of compounds 7-10. Otherwise the mass and CID spectra were the same, also for zirconium compounds 7 and 10, implying that all compounds have the same structure. A change of the M(OR) 4 /bis(salicylaldimine) ratio resulted in the same molecular ion peaks.
Analogous structures were previously reported for [{Pt(L 5 )} 2 ] and [{Cu(L 5 )} 2 ] complexes (metal ion instead of the M(OR) 2 entity), [10,11] whereas a polymeric structure was proposed for Ni II complexes of C n -bis(salicylaldiminates) (n = 6-12). [12] The solution properties of 6-10 were investigated by NMR spectroscopy experiments. A clear shift of the CH=N proton from d = 8.46 to 7.92-7.63 ppm was observed for all compounds, and the chemical-shift difference of the CH=N and C aryl ÀO signals in the 13 C NMR spectrum became smaller. Both observations are clear evidence of coordination of the salicylaldiminate groups. All spectra showed only one signal for CH/ CH 2 groups of terminal alkoxo groups, in line with the conclusions from the ESI-MS spectra.

Conclusion
Reactions of titanium or zirconium alkoxides with dioximes or bis(salicylaldimines) resulted in complexes of the type www.chempluschem.org [{TiL(OR) 2 } 2 ] (L = bridging dioximate or bis(salicylaldiminate) ligands), when the spacer between the coordinating groups was flexible enough. Two structural possibilities (Scheme 3) were observed that could be distinguished clearly by ESI-MS and especially MS/MS experiments in the low-energy CID mode. Type A complexes were formed with dioximate ligands, in which the ligands bridge a Ti 2 (m 2 -OR) 2 unit. In contrast, the bis(salicylaldiminate) ligands bridge two independent M(OR) 2 (M = Ti, Zr) moieties (type B complexes). The latter structure type was previously observed also for bis(b-diketonate)-substituted titanium isopropoxide derivatives. [8] When the spacer between the two oximate groups was stiffened, that is, when titanium alkoxides were treated with 1,3or 1,4-cyclohexyldioxime, insoluble compounds with a glass transition were formed, which appear to have a polymeric structure.

Characterization techniques
The 1 H and 13 C solution NMR spectra were recorded on a Bruker Avance 250 (250. 13 MHz { 1 H},62.86 MHz { 13 C}). Samples for solution NMR spectra were taken by dissolving the dried residue in a deuterated solvent without further purification. 2D NMR spectra were performed on a Bruker Avance 300 DPX (300. 13 MHz { 1 H},75.47 MHz { 13 C}) and measured with Bruker standard pulse programs COSY, HSQC, EXSY (t mix = 1 s), and HMBC (optimized for J = 140 Hz). Solid-state NMR spectra were recorded on a Bruker Avance 300 instrument equipped with a 4 mm broad-band magicangle spinning (MAS) probe head operating at 75.4 MHz for 13 C and 30.4 MHz for 15 N. The 13 C and 15 N NMR spectra were recorded with ramped CP/MAS at a rotor frequency of usually 6-8 kHz.
The ESI-MS measurements were performed on a Bruker Daltonics Esquire 3000 plus 3D ion-trap mass spectrometer fitted with an orthogonal ESI ion source and operated in the positive-or negativeion mode. The spray voltage was maintained at À4 kV, the drying gas temperature was set to 200 8C and all ion and transfer-line source voltages were optimized for maximum molecular-ion transmission (i.e., the sodiated or chloro-adduct molecules). For lowenergy CID MS/MS experiments, the isolation width was typically set to 10 Da to cover the entire isotopic distribution of the selected precursor ion. The fragmentation amplitude was manually set to 0.5-1 V to induce abundant product-ion formation. Solutions in pure 2-propanol or in a mixture of chloroform/2-propanol (1:3) at a concentration of 1 mg mL À1 were infused by a syringe pump into the ESI-source at a flow rate of 3 mL min À1 . No sodium chloride was added to the sample solution to enhance formation of adduct ions. All calculated m/z values of the titanium complexes were based on the naturally most abundant 48 Ti isotope, and those of the zirconium complexes based on the most abundant 90 Zr isotope.
The MALDI-MS evaluation (for 4 and 5) was performed by means of a Shimadzu Kratos Analytical Axima CFR + in the positive-ion mode by applying standard MALDI matrices.

X-ray structure analyses
Single-crystal X-ray diffraction experiments were performed at 100 K on a Bruker-AXS SMART APEX II diffractometer with a CCD area detector and a crystal-to-detector distance of 5.0 cm using graphite-monochromated Mo Ka radiation (l = 71.073 pm). Data were collected with f and w scans and 0.58 frame width. The data were corrected for polarization and Lorentz effects, and an empirical absorption correction (SADABS) was applied. [13] The cell dimensions (Table 1) were refined with all unique reflections. The structures were solved with direct methods (SHELXS97) and refinement to convergence was carried out with the full-matrix least-squares method based on F 2 (SHELXL97) with anisotropic structure parameters for all non-hydrogen atoms. [14,15] The hydrogen atoms were placed on calculated positions and refined riding on their parent atoms. CCDC 907994 (1) and 907995 (3) contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

Synthesis of ligands
The dioximes L 1 H 2 , L 2 H 2 , L 3 H 2 , and L 4 H 2 were synthesized by a modification of the method described by Bousquet. [16] In a typical procedure hydroxylamine hydrochloride (14.5 g, 209 mmol) was dissolved in deionized water (50 mL) and cooled to 0 8C. The corresponding diketone/dialdehyde (87 mmol) was added dropwise. After 30 min of stirring at room temperature a solution of potassium carbonate (14.4 g, 104 mmol) in deionized water (25 mL) was