Group 3 metal trihalide complexes with neutral N-donor ligands – exploring their affinity towards fluoride†

Fluorination of [ScCl3(Me3-tacn)] (Me3-tacn = 1,4,7-trimethyl-1,4,7-triazacyclononane) and [ScCl3(BnMe2tacn)] (BnMe2-tacn = 1,4-dimethyl-7-benzyl-1,4,7-triazacyclononane) by Cl/F exchange with 3 mol. equiv. of anhydrous [NMe4]F in CH3CN solution yields the corresponding [ScF3(R3-tacn)] (R3 = Me3 or BnMe2). These are the first examples of scandium fluoride complexes containing neutral co-ligands. The fluorination occurs stepwise, and using a deficit of [NMe4]F produced [ScF2Cl(Me3-tacn)]. Attempts to fluorinate [YCl3(Me3-tacn)], [YI3(Me3-tacn)], [LaCl3(Me3-tacn)(OH2)] or [MCl3(terpy)] (M = Sc, Y or La; terpy = 2,2’:6’2’’-terpyridyl) using a similar method were unsuccessful, due to the Cl/F exchange being accompanied by loss of the neutral ligand from the metal centre. Fluorination of [ScCl3(Me3-tacn)] or [ScCl3(terpy)] with Me3SnF was also successful. The products were identified as the very unusual heterobimetallic [Sc(Me3-tacn)F2(μ-F)SnMe3Cl] and [Sc(terpy)F(μ-F)2(SnMe3Cl)2], in which the Me3SnCl formed in the reaction behaves as a weak Lewis acid towards the scandium fluoride complex, linked by Sc–F–Sn bridges. [Sc(terpy)F(μ-F)2(SnMe3Cl)2] decomposes irreversibly in solution but, whilst multinuclear NMR data show that [Sc(Me3-tacn)F2(μ-F)SnMe3Cl] is dissociated into the [ScF3(Me3-tacn)] and Me3SnCl in CH3CN solution, the bimetallic complex reforms upon evaporation of the solvent. The new scandium fluoride complexes and the chloride precursors have been characterised by microanalysis, IR and multinuclear NMR (H, F, Sc) spectroscopy as appropriate. X-ray crystal structures provide unambiguous evidence for the identities of [Sc(Me3-tacn)F2(μ-F)SnMe3Cl], [ScF2Cl(Me3-tacn)], [YI3(Me3-tacn)], [{YI2(Me3tacn)}2(μ-O)], [ScCl3(terpy)], [YCl3(terpy)(OH2)], and [{La(terpy)(OH2)Cl2}2(μ-Cl)2]. Once formed, the [ScF3(R3-tacn)] complexes are stable in water and unaffected by a ten-fold excess of Cl − or MeCO2 , although they are immediately decomposed by excess F. The potential use of [ScF3(R3-tacn)] type complexes as platforms for F PET (positron emission tomography) radiopharmaceuticals is briefly discussed. Attempts to use the Group 3 fluoride “hydrates”, MF3·xH2O, as precursors were unsuccessful; no reaction with R3-tacn or terpy occurred either on reflux in CH3CN or under hydrothermal conditions (H2O, 180° C, 15 h). PXRD data showed that these “hydrates” actually contain the anhydrous metal trifluorides with small amounts of surface or interstitial water.


Introduction
The coordination chemistries of scandium and yttrium have been explored much less than those of the other 3d and 4d metals. The presence of colourless metal ions, in a single (3+) oxidation state and with closed shell configurations (hence no magnetic or d-d spectroscopic fingerprints), coupled with their limited availability, low purity and high cost, restricted early work. 1,2 Often their chemistry was included in studies of the lanthanide elements, which tended to see similarities rather than explore differences. 2 More recent work 3 has shown that there are significant differences, especially for scandium, and the structural chemistry of scandium is surprisingly diverse. 4 A rich, but synthetically challenging, organometallic chemistry of both metals has been explored in recent years, 5 and C-H bond activation, ethene, styrene and α-olefin polymerisation, and aromatic C-F bond activation have all been Experimental All complex syntheses were carried out using standard Schlenk and vacuum line techniques. Samples were handled and stored in a glove box under a dry dinitrogen atmosphere to exclude moisture, which decomposes many of the samples.

and [Sc(H 2 O)
Metal trifluoride "hydrates" ScF 3 ·xH 2 O. Sc 2 O 3 (2.9 g, 0.021 mol), and a 6 M solution of HCl (43 mL) were heated to reflux for 3 h, during which period the mixture changed from a cloudy white suspension to a clear yellow solution. The solvent was removed in vacuo whilst heating at 65°C. ScCl 3 ·6H 2 O was obtained as a white solid. This was dissolved in water in a plastic beaker and 6 mL of 40% HF (aq) (CARE) were added causing the precipitation of a white solid. The mixture was heated to boiling and the solvent evaporated, giving a white gel-like solid. A portion of the gel was suspended in water, causing the formation of the solid, which was isolated by evaporation of the solvent. The same procedure was repeated portion by portion and the solid combined (3.94 g, 93%).
92 mmol) was dissolved in water. 5 mL of a solution of 40% HF (aq) was added and a white precipitate formed. The precipitate was left to settle overnight. The solution was filtered and the solid washed with water and dried in vacuo (1.07 g, 75%).
Method 2: Y 2 (SO 4 ) 3 ·8H 2 O (3.0 g, 4.92 mmol) was suspended in hot water (80°C) until most of the solid dissolved. The liquid was decanted off from any residue and a solution of 40% HF (aq) (3 mL) was added to the solution. A white solid precipitated immediately. The reaction was left stirring for 1.5 h and then the solid was left to settle overnight. The solution was decanted off and the solid dried overnight in a desiccator (1.24 g, 86%).
[LaCl 3 (Me 3 -tacn)(OH 2 )]. LaCl 3 ·7H 2 O (0.101 g, 0.41 mmol) was dissolved in ethanol (10 mL). Me 3 -tacn (0.06 mL, 0.41 mmol) in ethanol (5 mL) was added to form a white precipitate. After stirring for 45 min, the solvent was removed in vacuo leaving a white sticky solid which was dried in a desiccator for two hours. The solid was then washed with diethyl ether (3 mL) and dried again in vacuo, leaving a white powder. Yield: 0.09 g, 54%. Required for C 9  [Sc(terpy)F(µ-F) 2 (SnMe 3 Cl) 2 ]. [ScCl 3 (terpy)] (0.05 g, 0.13 mmol) was suspended in CH 3 CN (8 mL) and a suspension of Me 3 SnF (0.083 g, 0.45 mmol) in CH 3 CN (15 mL) was added. After one hour most of the solid had dissolved. The mixture was left stirring for 6 hours. The liquid was decanted via cannula and the solvent removed in vacuo, giving a slightly pink solid. The solid was washed with hexane and dried in vacuo (0.031 g, 33%). Required for C 21

X-ray experimental
Crystals of the complexes were grown as reported in the Experimental section. Data collections used a Rigaku AFC12 goniometer equipped with an enhanced sensitivity (HG) Saturn724+ detector mounted at the window of an FR-E+ SuperBright molybdenum (λ = 0.71073 Å) rotating anode generator with VHF Varimax optics (70 micron focus) with the crystal held at 100 K. Structure solution and refinement were performed using SHELX(S/L)97, SHELX-2013 or SHELX-2014/7. 27 H atoms bonded to C were placed in calculated positions using the default C-H distance, and refined using a riding model. Details of the crystallographic parameters are given in were collected on a Bruker D2 diffractometer using Cu K α X-rays and refined using the GSAS software. 28

Results and discussion
Three possible routes to Group 3 metal trifluoride complexes of R 3 -tacn and terpy (L) were considered: (1) Direct reaction with the "hydrated" Group 3 metal trifluorides with the neutral ligands; ( x" typically appears to be less than one. The addition of a fluoride source, either KF or aqueous HF, to aqueous solutions of yttrium or lanthanum salts, gave immediate white precipitates, MF 3 ·xH 2 O. Using scandium salts and aqueous HF also gave ScF 3 ·xH 2 O, but using alkali metal fluorides gave mixtures, and in one case pure KSc 2 F 7 , which was identified by its PXRD pattern (see ESI †). The PXRD patterns obtained from MF 3 ·xH 2 O sometimes showed rather broad reflections, but corresponded to the patterns reported for anhydrous MF 3 ( Fig. 1 and ESI †).
Thus, we conclude that the MF 3 ·xH 2 O actually comprise of the "anhydrous" MF 3 polymer, with water of crystallisation on the surface or occupying voids in the crystal lattice, rather than coordinated to the metal ion. This also explains the range of values of x in MF 3 ·xH 2 O reported in the limited literature available. 1,3 Attempts to react the MF 3 ·xH 2 O with terpy or Me 3 -tacn under hydrothermal conditions (180°C, 15 h) were unsuccessful, with the MF 3 ·xH 2 O being recovered, although with noticeably increased crystallinity (sharper PXRD patterns).
These results explain why 'hydrated' Group 3 fluorides are not a viable entry into the coordination chemistry of these fluorides with neutral ligands. This contrasts sharply with the Group 13 fluoride hydrates, 19,29,30 and the Group 4 compounds, [MF 4 (OH 2 ) 2 ] (M = Zr, Hf ), 31 whose crystal structures show the water is coordinated to the metal, from which it can be displaced by neutral ligands. These are effective synthons for wider coordination chemistry. We reported previously that the f-block tetrafluorides, [MF 4 ·xH 2 O] (M = Ce, Th) have very limited coordination chemistry (CeF 4 ·xH 2 O dissolves only very slowly in refluxing dmso to form [CeF 4 (dmso) 2 ]). They too are likely to contain only lattice/surface water. 32 The wider implications of these results in metal fluoride coordination chemistry suggest that for other metals, only those 'hydrated' fluorides that contain water within the metal coordination sphere, are likely to be viable synthons for neutral ligand complexes.

Chloride/iodide precursor complexes
Many trichloride complexes of Sc(III), Y(III) and La(III) with neutral ligands are highly moisture sensitive and must be synthesised and handled in anhydrous systems. In work with other early d-block systems, we noted that the corresponding metal iodides and iodo-complexes were often more soluble in weakly coordinating solvents, probably due to lower lattice energy, although the gain in solubility comes at the cost of even greater moisture sensitivity. 17,33,34 In the present study we synthesised complexes with Me 3 -tacn and terpy, which give examples of pseudo-octahedral complexes with fac and mer geometries for scandium, respectively, although for Y and La higher coordination numbers were often produced. The known 26 (Fig. 2), with the expected tridentate Me 3 -tacn and three mutually facial iodides completing the distorted octahedral environment, and the partial hydrolysis product, [{YI 2 (Me 3 tacn)} 2 (μ-O)] (Fig. 3)    The reaction of [ScCl 3 (thf ) 3 ] with terpy in anhydrous CH 3 CN gave mer-[ScCl 3 (terpy)] (Fig. 4).
The complex has a distorted octahedral coordination around the metal centre conferred by the rigid terpy ligand, the angles involving the ligand are significantly less than the 180/90°expected for a regular octahedron, with N1-Sc1-N3 = 142.3°. The extended crystal structure of [ScCl 3 (terpy)] shows π-stacking interactions (3.82 Å) between the aromatic ring of the terpy ligand of the adjacent molecule, connecting them into 1D zig-zag chains (see ESI Fig. S4 †).
The crystal structure (Fig. 5) shows a pentagonal-bipyramidal coordination around the metal centre with the Y-Cl bond lengths in the axial positions shorter than that in the equatorial plane. The angle between the yttrium centre and the nitrogen atoms are less than the 72°value expected for the perfect pentagonal-bipyramidal conformation, due to the rigid terpy ligand and the equatorial plane is puckered. Furthermore, the packing in the crystal structure shows both H-bonding (Cl⋯HOH) between adjacent molecules to form associated dimers, and weak π-stacking (4.04 Å) linking the dimers into zig-zag chains (Fig. S5 †). The bond lengths are generally shorter than in the eight-coordinate [YCl(terpy)(OH 2 ) 4 ]Cl 2 ·2H 2 O. 36 Using the heavier f-block ions, La(III) and Lu(III), [LaCl 3 (terpy)(OH 2 )]·4H 2 O and [LuCl 3 (terpy)(OH 2 )] were obtained via reaction of LaCl 3 ·7H 2 O or LuCl 3 ·6H 2 O, respectively, with one mol. equiv. of terpy in ethanol. X-ray crystallographic analyses show that the La(III) complex exists as a chloro-bridged dimer, [{La(terpy)(OH 2 )Cl 2 } 2 (μ-Cl) 2 ] (Fig. 6) involving eightcoordinate La(III) with adjacent molecules linked into chains via H-bonding interactions between the coordinated water molecule on one La(III) centre and the Cl ligands on adjacent molecules (Fig. S6 †). On the other hand, [LuCl 3 (terpy)(OH 2 )] (Fig. S7 †) is a seven-coordinate monomer, isostructural with the Y(III) analogue above (and hence also displaying the same H-bonding and π-stacking interactions in the solid state).

Chloride(iodide)/fluoride exchange reactions using [NMe 4 ]F
Addition of three mol. equiv. of anhydrous [NMe 4 ]F to a CH 3 CN solution of [ScCl 3 (Me 3 -tacn)] gave a colourless solution whose 19 F{ 1 H} and 45 Sc NMR spectra 37 each showed three    [ScF 3 (Me 3 -tacn)], respectively. The modest quadrupole moment of 45 Sc (I = 7/2) means that resonances are observed in many systems, but couplings to other nuclei are often lost in the line broadening, unless the scandium is in a high symmetry environment. 17,34,38,39 Since these complexes are the first examples of scandium fluoride species with neutral ligands, there are no comparable literature data, but the chemical shifts of the chloro-species are reasonable compared to data on other ScCl 3 adducts. 34,38,39 Adding further small aliquots of [NMe 4 ]F in CH 3 CN initially led to depletion of the resonances assigned to the mixed chloro/fluoro complexes, and enhancement of the broadened quartet at δ = 104 ppm, attributed to [ScF 3 (Me 3 -tacn)] (Fig. 7). The broad quartet shows coupling to three equivalent fluorides with 1 J ScF = 219 Hz (since the efg is small). However, excess fluoride caused complete loss of all the 45 Sc and 19 F{ 1 H} resonances from the tacn complexes (see below).
The 19 F{ 1 H} and 45 Sc NMR data for the fluorination reaction are also strongly indicative of the fluorination of [ScCl 3 (Me 3tacn)] occurring in a stepwise manner. Further confirmation of this follows from a single crystal X-ray structure determination on [ScF 2 Cl(Me 3 -tacn)], a few crystals of which were grown by concentrating the NMR solution in acetonitrile (Fig. 8).

Chloride/fluoride exchange reactions using Me 3 SnF
Me 3 SnF is a useful fluorinating agent, its polymeric structure makes it insoluble in most solvents, 40,41 but it dissolves as the Cl/F exchange reaction proceeds, and usually, the Me 3 SnCl (which contains tbp tin centres weakly chlorine-bridged into polymeric chains) 42 formed is easily removed from the products by washing with hexane. The reagent does not provide free fluoride ions so an excess can be used without the risk of the decomposition observed using [NMe 4 ]F. The fac-octahedral trifluoro complexes [ScF 3 (R 3 -tacn)] were readily obtained by treatment of the trichloro species with three mol. equiv. of Me 3 SnF (and adding excess Me 3 SnF has no further effect) (Fig. 9). The broad "doublets" observed in the 19 F{ 1 H} NMR spectra result from partial collapse of the couplings to 45 Sc (I = 7/2) in the low symmetry environments.
However, the product obtained had a microanalysis corresponding to [Sc(Me 3 -tacn)F 2 (µ-F)SnMe 3 Cl]; note that whilst three equivalents of Me 3 SnCl are produced in the reaction, only one is retained in the scandium complex.