Coordination Chemistry and Applications of Medium/High Oxidation State Metal and Non-Metal Fluoride and Oxide-Fluoride Complexes with Neutral Donor Ligands

Most high and medium oxidation state (O.S. ≥ 3) metal and non-metal fluorides and oxide fluorides have Lewis acidic properties although detailed exploration of their chemistry with neutral ligands, which differs significantly from that with chloride, bromide or iodide co-ligands, has only been undertaken in recent years. The previous review (Benjamin et. al. Chem. Soc. Rev. 42 (2013) 1460) covered work published up to ~ 2011, and the present article covers new work up to early 2019, a period which has seen many new contributions to the field. This article describes work on the coordination chemistry of d, f and p-block fluorides and oxide fluorides with neutral ligands containing donor atoms drawn from Groups 15 (N, P, As or Sb) and 16 (O, S, Se or Te) and including N-heterocyclic carbenes. The incorporation of the radionuclide 18F into neutral metal complexes and their use in medical diagnosis via positron emission tomography (PET) is described, along with briefer coverage of other potential applications.


Introduction and Scope
The coordination chemistry of metal fluorides and oxide fluorides has traditionally developed separately from that of the heavier halides, and, until the last twenty-five years, the characterisation of the complexes was often poor and the properties were not investigated in any detail [1,2]. Recent years have seen much renewed interest in this area of chemistry, which results both from the recognition that the chemistry of metal fluoride complexes is fundamentally different to that of other halides [3,4,5,6,7] and from applications such as the use of 18 F in radiochemical diagnostics (PET imaging) [8,9,10]. Metal fluorides are also popular building blocks for molecular magnets, although only in a small minority of these cases are neutral ligands involved. This work is reviewed elsewhere [11,12] and is therefore not discussed in detail in the present article. Metal fluoride complexes in catalysis have been receiving increased attention [13,14]. In a review published in 2013 [7], we discussed the chemistry of metal and non-metal fluoride and oxide fluoride complexes containing neutral donor ligands with the metal/non-metal Lewis acid in oxidation states ≥ 3. include N-heterocyclic carbenes (NHCs). The aim is to showcase the new work published in this period, rather than provide fully comprehensive coverage, although we aim to include all new compounds and significant new studies of existing complexes. Fluoride complexes containing the softest donor ligands are very rare, but the recent characterisation of tertiary arsine complexes of WF6 [15] and a telluroether complex of TaF5 [16], demonstrate that some complexes of this type are obtainable. Other recent reviews covering aspects of fluorine chemistry are: Crystal Chemistry and Selected Physical Properties of Inorganic Fluorides and Oxide-Fluorides [17]; Molecular Hexafluorides [18]; A survey of titanium fluoride complexes, their preparation, reactivity and applications [14]; and Highly reactive carbenes as ligands for main group element fluorides [19]. Reviews covering relevant aspects of the chemistry of single elements are listed in the appropriate sections below.

Synthesis
All the elements in the periodic table, except He, Ne and Ar, form fluorides and many also form oxide fluorides, so a wide variety of synthesis routes for their complexes are possible, although often only one or two will be suitable for a specific complex. The Lewis acidity of these metal/non-metal fluorides is most easily demonstrated by their formation of the corresponding fluoroanions, which form for almost all examples [17].
Typically complexes may be prepared in water, sometimes using hydrothermal methods.
In the previous review [7] we suggested a pragmatic "classification" of the metal fluoride/oxide fluorides from the viewpoint of synthesising their metal complexes, and an updated version of this is shown below • No Lewis acidity -examples are NF3 or CF4; • Probably incompatible with neutral organic ligands due to their extremely strong oxidising/fluorinating power, e.g. IrF6, RuF6, BiF5, CrOF4, CrF5, PbF4; • Molecular fluorides or oxide fluorides that may react directly with neutral ligands, sometimes dispersed in an inert solvent or even in the absence of solvent, e.g. SiF4, GeF4, AsF5, BF3, PF5, WF6; • Weakly polymerised fluorides or oxide fluorides, which may react directly in a suitable non-polar or weakly polar solvent, e.g. TaF5, NbF5, VOF3, SbF5; • Strongly polymerised fluorides that may be converted into a suitable (molecular and soluble) • Sometimes the anhydrous binary metal fluoride is inert, but a hydrated form may provide an entry into the coordination chemistry, e.g. MF3•3H2O (M = Al, Ga), often using hydrothermal routes.
Complexes of these metal fluorides may be accessible via Cl/F exchange from the corresponding chloro-complex or, for oxide fluorides, via O/F exchange from the binary fluoro-complex; • Inert and intractable, e.g. BiF3, PbF2, and the 4f-block trifluorides, for which there are no reports of successful complex formation with neutral ligands. The solution may be reacted directly with the new ligand, or after isolation of the solvent adduct. The latter can be reacted subsequently with other neutral ligands in a non-coordinating solvent, which avoids competition between the donor nitrile or ether solvent present in large excess and the incoming ligand.
As indicated above, some metal fluoride hydrates, including MF3•3H2O (M= Al, Ga, In), can be complexed with N-donor ligands using hydrothermal vessels (180°C, 15h) [20]. The same approach fails with MF3•xH2O (M = Sc, Y, La), when the reagents are recovered unchanged [21]. The differences may result from the different constitution of the trifluoride "hydrates". Whilst those of Group 13 contain water coordinated to the metal centre, the Group 3 compounds were shown by PXRD to have the extensively polymerised structure of the anhydrous trifluorides with the water on the surface or interstitial [21]. The hydrated tetrafluorides of Zr  [27].
Some very aggressive systems require fluoroplastic or metal equipment (e.g. the higher fluorides of the platinum metals) or systems using (or generating) HF, but in many cases borosilicate glass vessels fitted with PFTE taps that are common in modern organometallic laboratories are adequate, along with glass vacuum lines and high quality glove box facilities. For the most reactive systems the key requirement is to use rigorously anhydrous conditions, thoroughly dried glassware, dried solvents and purified/dried ligands.
Water not only competes successfully with many ligands for the metal fluoride, but also may initiate a downward spiral of attack on the glass, formation of HF, SiF4, and in some cases, ready protonation of the ligand (especially with N-donor functions). It can also introduce HF2 − or fluorosilicate anions into the products.
One benefit of the hydrothermal reactions (above) is that the fluoroplastic vessels used mean any HF that is generated is not a problem.

Structural and Spectroscopic Characterisation
The species described in this article are mostly molecular coordination complexes, and their characterisation uses the usual suite of structural and spectroscopic techniques. We have discussed the key features of their application to metal fluoride complexes elsewhere [7]. Some new points to consider are mentioned here.
Single crystal X-ray diffraction: The continuing improvements in diffractometers and computing power means that the main challenge in the area is now the growth and handling of often very reactive and moisture sensitive crystals. One should also bear in mind that particularly with very reactive systems the species found in the single crystal X-ray structure may not be characteristic of the bulk. Often spectroscopic data on the bulk sample may provide the required assurance, while PXRD can be used to confirm whether the bulk sample is (or is not) the same as that found in the single crystal structure [23].  [15,28].
NMR and EPR Spectroscopy: 19 F NMR spectroscopy (I = ½, 100%) provides a very sensitive specific probe in diamagnetic complexes; the other halogens have only quadrupolar nuclei that are not observable in the low symmetry environments of coordinated halide groups. 19 F NMR spectroscopy is also very useful in identifying decomposition products or impurities formed by fluorination of the neutral ligands. For example, the reaction of Me2Te and TeF4 produced Me2TeF2 and elemental tellurium rather than a complex of TeF4 [29].
Triphenylarsine oxide forms complexes with some metal fluorides, but in other systems it is converted to Ph3AsF2, which is most easily identified by the characteristic 19  species were in fact due to phosphine oxide impurities with the major constituent being the V(III) [VCl4(PR3)2] − , which are EPR silent [31].

f-block complexes
There appear to be no known simple adducts of lanthanide(III) fluoride complexes containing neutral ligands.
The "hydrate" LaF3•xH2O does not react with neutral ligands, including Me3-tacn or terpy, even under hydrothermal conditions (180°C). PXRD showed the "hydrate" had the same polymeric structure as anhydrous LaF3 indicating that the water is interstitial or surface and not coordinated to the lanthanum, which accounts for its lack of reactivity [21]. Treatment of [LaCl3(Me3-tacn)(OH2)] with [NMe4]F in MeCN resulted in displacement of the Me3-tacn and precipitation of LaF3 [21].
Cerium(IV) fluoride dissolves sparingly in liquid ammonia to form [CeF4(NH3)4]⋅NH3, which has a distorted square antiprismatic structure [32] and is isomorphous with [MF4(NH3)4]⋅NH3 (M = U, Zr, Hf) [33]. CeF4•xH2O does not dissolve in or react with MeCN or dmf even on prolonged reflux, but dissolves slowly in refluxing dmso to give a yellow solution from which yellow crystals [CeF4(dmso)2] were isolated [34]. The structure ( Fig. 1) reveals a distorted square antiprismatic cerium environment with two cis disposed O-coordinated dmso molecules, linked into a zig-zag chain by fluorine bridges. Attempts to use this complex as a synthon for other complexes were unsuccessful, the dmso was not displaced by phosphine oxides or diimines in dmf solution. Fig. 1 The structure of [CeF4(dmso)2], showing part of the fluoro-bridged chain structure (including the next Ce neighbours), redrawn from reference 34. The dmso groups were disordered and only the major component is shown.
[Hydrogen atoms are omitted for clarity from this and subsequent structures unless needed to identify ligands] Actinide fluoride complexes with neutral donor ligands are more numerous, but mostly limited to uranium.
Attempts to produce thorium complexes from ThF4•xH2O were unsuccessful, the hydrate proved to be insoluble even in refluxing dmso [34]. terminal and two bridging fluorides and two N-coordinated hydrogen cyanide molecules [39].
The reaction of uranium metal, iodine and AgF2 in pyridine solution gives green [UF2I2(py)4]•2py (Fig. 3) which is soluble in pyridine and thf and is potentially a useful synthon to prepare other U(IV) fluoride complexes [36]. The corresponding [UF2Cl2(py)4] was obtained as a by-product from reaction of a uranium pincer carbene with PhCOF [40]. The hydrothermal reaction of UO2(NO3)2, 1,10-phen and HF gave yellow crystals of exploring the predicted structures and electronic properties were also reported [41].

Group 3
The "hydrates", ScF3•xH2O and YF3•xH2O do not react with neutral N-or O-donor ligands even under hydrothermal conditions, and like the LaF3•xH2O discussed above, PXRD shows them to have the same structures as the anhydrous trifluorides, explaining their inert nature [21].  The terpy complex decomposes in solution, but although 19

Group 4
The coordination chemistry of TiF4 is relatively well developed [7,14] and little new work has appeared. The the results with experimental data where available. In general, a linear L-Ti-F arrangement is the most stable in each case, which maximises the Ti-F pπ-dπ bonding [14].
The colourless Ti(IV) carbene [TiF4(NHCD i PP2)2] was prepared in good yield from TiF4 and the carbene in thf solution [44]. The structure (Fig. 7) shows an octahedral titanium centre with the bulky carbenes in a trans arrangement. The Ti-C bonds are shorter (2.26 Å) than those in the corresponding [TiCl4(NHCD i PP2)2] (2.32 Å). The coordination chemistry of ZrF4 and HfF4 is less extensive than that of TiF4, but the structures of known compounds have been reviewed in detail [49]. The majority of the structural work described concerns fluorozirconate(IV) and -hafnate(IV) anions, but the limited structural data on complexes with neutral ligands were also included. The structure of the monoclinic form of ZrF4•3H2O has been shown to be a polymer with ZrF6(H2O)2 groups [50], which contrasts with the known structure of the triclinic form, which is a discrete dimer. The structure of monoclinic HfF4•3H2O has a similar polymer unit to the zirconium [50].

Group 5 Vanadium
High valent vanadium fluoride chemistry is unusual in that whilst complexes of the oxide fluorides VOF3 and VO2F are well established [7], and that of VOF2 has recently been explored (see below), neutral ligand complexes of VF5 and VF4 are restricted to work in a few very old reports, although some data on VF3 adducts is available [7].
The first carbene complex of VOF3 has been obtained very recently. The [VOF3(NHCD i PP2)] was made by combining VOF3 and NHCD i PP2 in thf and the X-ray structure showed a square pyramidal geometry with the oxido group occupying the apical position [53].  [54]. The structure is shown in Fig. 8. A popular synthesis of extended lattice structures or frameworks containing vanadium oxide-fluoride anions is to combine polydentate amines or nitrogen heterocycles, a vanadium source, and aqueous HF, in hydrothermal or solvothermal vessels. The majority of products contain protonated organonitrogen cations, but in some cases the base is also incorporated into the framework. An example is the pyrazine complex [V2(N2C4H4)O2F4], made hydrothermally from pyrazine, VOF3 and aqueous HF [55]. The structure is a 2D polymer with zig-zag chains of V IV 2O2F4 linked to adjacent dimers by fluorine bridges and with coordinated pyrazine linking the chains to form a sheet structure. Hydrothermal syntheses were also used to prepare a series of Ag-V V MOFs with [Ag2(VO2F2)2(triazole)4] building blocks [56]. Several mixed valence (V V V IV ) polyoxofluorovanadate clusters containing coordinated pyridine or imidazole have been prepared and their structures determined [57].

Niobium and Tantalum
There has been significant new work on the pentafluorides of both elements, including the first NHC complexes, and the first characterisation of phosphine and arsine complexes, which add a second group of soft donor ligands to the thioether and selenoether complexes reported a few years ago [7]. The first complexes of NbOF3 have also been prepared, but those of TaOF3 have proved elusive. Complexes of fluorides in lower oxidation states, or of other oxide fluorides, also remain unknown. Table 1 lists the new reports of Nb(V) and Ta(V) fluoride complexes.
As can be seen from  in anhydrous MeCN [64,70]. The NMR spectra show that unlike the PR3 complexes, the diphosphine complexes do not dissociate the ligand in solution in MeCN or CH2Cl2 and both the 19 F{ 1 H} and 31 P{ 1 H} NMR spectra show binomial quintet couplings in the cations, demonstrating the equivalency of the P and F atoms ( Fig. 15). This was confirmed by X-ray crystal structures determined for five of the complexes, all of which reveal dodecahedral cations [64,70] (Fig. 16). The corresponding reactions of the MF5 with o-C6H4(AsMe2)2 [70] (Fig. 17), which are partially decomposed in MeCN solution, but stable in CH2Cl2. The formation of these diarsine complexes with Nb and Ta contrasts with the case of TiF4 [75], where diphosphine complexes form, but no diarsine analogue could be obtained.   The reaction of TaF5 with TeMe2 in CH2Cl2 at 0°C gave an unstable yellow solid identified spectroscopically as [TaF5(TeMe2)]; this is the first telluroether complex of a high valent d-block fluoride [16]. It decomposes in a few hours at room temperature, forming Me2TeF2 among other products. The corresponding reaction with NbF5 gives only a black sticky decomposition product.
NHC complexes of type [MF5(NHCD i PP2)] have been prepared from toluene solution [73]. They contain an unusually high metal oxidation state for carbene complexes. Tantalum amido carbene complexes  However, attempts to make complexes with thf, MeOCH2CH2OMe, PMe3 or MeSCH2CH2SMe failed, with NbOF3 precipitating, indicating that these ligands do not bind sufficiently strongly to the oxide fluoride to prevent dissociation and polymerisation [27,69]. In contrast, the diphosphine complexes

Group 6 Chromium
There are no complexes of chromium fluorides or oxide fluorides in oxidation states >3; it seems possible that N-bases could coordinate to CrF4 or CrOF3 if an appropriate synthetic entry can be found. Chromium(III) fluoride complexes, especially [CrF3(py)3], are useful building blocks for the synthesis of heterobimetallic magnets, particularly those containing lanthanide ions. The work has been reviewed [11] and some exemplar reports [79][80][81] [87].

Molybdenum
In   [92]. The structure of the Mo(III) complex [MoF2(Ph2PCH2CH2PPh2)2]BF4 has also been reported [93].   with identical W-N bonds for the first two and a slightly shorter bond for the last [94].    pyNO, OPPh3) also show the presence of an octahedral geometry (Fig 27), again with L trans to W=O. grow crystals for an X-ray study. Use of other phosphines, including P n Pr3, PPhMe2, initially produced white solids, which are probably the analogous complexes, but these rapidly turned yellow and then brown and appear to be too unstable to isolate in a pure state [96].

Group 7
There appear to be no new reports of neutral ligand complexes of Tc or Re fluorides in oxidation states >3.
In part this probably reflects the difficulties in accessing the metal fluorides.

Manganese
When Li2MnF5 was suspended in neat pyridine in a glass vessel for several days, orange brown crystals of nitrogen atom (Fig. 29); detailed magnetic and EPR measurements were reported.  [101]. Detailed theoretical studies of the superexchange mechanism in this dimer have been reported [102].

Group 8
There appears to be no new work on complexes of ruthenium or osmium fluorides, which probably reflects the difficulty of accessing the fluorides and the need for metal or fluoroplastic vessels to handle these aggressive species.

Iron
The  Mössbauer data confirm the mixed valence description [110]. Magnetic materials containing FeF3 in combination with Gd [81] or a range of 3d metal fluorides [111] have also been described.

Rhodium and Iridium
There are no reports of the direct reaction of rhodium or iridium fluorides with neutral ligands, and the relatively few examples of complexes with Rh III F or Ir III F bonds are organometallic species often derived from oxidation of Rh I or Ir I compounds [112,113,114,115] which fall outside the scope of the present article.

Nickel, Palladium and Platinum
High valent nickel species have been suggested as intermediates in some nickel catalysed organic transformations, but only very rarely has the species been experimentally identified or isolated. One example is the surprisingly stable yellow (formally) Ni(IV) complex [NiF2(py)2(CF3)2] obtained by oxidation of [Ni(py)2(CF3)2] with XeF2 [116]. The crystal structure shows an octahedral nickel environment with trans fluorines, and cis pyridines (Fig. 35). Two Ni(III) complexes were isolated under slightly different conditions -[NiF(py)3(CF3)2] and [(CF3)2(py)Ni(µ-F)2Ni(py)2(CF3)2], the first octahedral, the dimer with unusual five-and six-coordinate nickel centres (Fig. 36).  Similarly, Pd(II) catalysed conversions of aryl-H bonds to aryl-F have been suggested to involve Pd(IV) intermediates, but only in a few cases have the latter been isolated and/or characterised [117,118]. The strong ligand field produced by Ni-C σ or Pd-C σ bonds and the presence of bulky ligands help to stabilise the high oxidation states, but it seems quite unlikely that simple coordination complexes of the metal fluorides with neutral ligands will be isolable. The possibility of obtaining complexes of N-donor ligands with PtF4 does not seem to have been examined.

Group 11
Copper, Silver and Gold.
Complexes of copper and silver in oxidation states ≥3 are not expected to form, but gold(III) complexes with N-donor ligands have been described. The use of gold catalysts in the synthesis of fluoro-organics has been reviewed very recently [119] and this article should be consulted for details. which was sufficiently stable for crystallographic analysis (Fig. 38).

Group 13
Boron trifluoride is a reactive gas, but the anhydrous trifluorides of aluminium, gallium and indium are inert polymers containing octahedrally coordinated metal ions. No complexes of TlF3 have been described. Boron, aluminium and gallium fluoro-complexes have been developed as potential carriers for the radioisotope 18 F in medical diagnostics (PET imaging), and this is discussed in Section 7.  [126]. Many other similar adducts are known [7]; all are distorted tetrahedral monomers. Although the Lewis base involved is also important, in most cases the Lewis acidity of the boron halides falls BI3 > BBr3 > BCl3 > BF3, which was for many years attributed to π-backbonding from the halogen lone pairs into the empty boron p-orbital. Considerable effort using DFT methods has led to the view that the order arises from σ-bonding effects and sometimes a steric component [126,127], and the π-backbonding explanation has been discarded. Detailed electrochemical and spectroscopic studies have been carried out on a range of BF3 adducts with nitrogen heterocycles, with the aim of understanding how the heterocycle structures affect the electrochemical properties [128,129]. Adducts of this type have been examined as electrolyte additives to increase the lifetime of Li-ion batteries [130,131,132].

Boron
Phosphine complexes of BF3 are less stable than analogues with the other three boron halides. The complex [BF3(PMe3)] was prepared some years ago and characterised spectroscopically [7], and its structure has recently been obtained. It shows the expected distorted tetrahedral arrangement about boron with an eclipsed conformation [133]. The bond lengths are identical with those found in [BF3(PEt3)], the only other reported structure of a tertiary phosphine-BF3 adduct [134].   . The white complex is insoluble in CH2Cl2 and decomposed by dmso, or MeCN. Multinuclear NMR spectra ( 1 H, 19 F{ 1 H} and 11 B) were obtained from freshly prepared MeNO2 solutions, but slow decomposition prevented growth of crystals for an X-ray study [133].
The constitution follows from the NMR data (Fig. 40) and by comparison with the structures of more stable  The ambidentate HN(P i Pr2)2 reacts with BF3 in Et2O to form [BF3{N(P i Pr2)(P i Pr2H)}] in which the BF3 is coordinated to the nitrogen and the proton has been switched onto one phosphorus (Fig. 41) [136], which contrasts with the phosphine donor coordination by this ligand found with most heavier Group 13 centres. ; an X-ray crystal structure of the latter showed it to contain a chelating diphosphine dioxide and a rare crystallographically authenticated example of the [B2F7]anion [133]. The structure of the secondary phosphine oxide complex, [BF3(OPHMe t Bu)], has also been determined [137].
The weaker Lewis acidity of BF3 compared to the heavier boron trihalides is also evident in the complexes with thio-, seleno-and telluro-ethers [138,139] in water and can even be recrystallised quickly from hot water without significant decomposition [142,143]. also been synthesised and structurally characterised [144]. DFT calculations on B, Al and Ga complexes with both normal and abnormal NHC's have shown that the carbene-Lewis acid bond is strongly influenced by the substituents on the acceptor atom and that the covalent character increases Al < Ga < B [145]. Boron trifluoride-carbene adducts have also been investigated as electrolyte additives for Li-ion batteries [146].

Aluminium
As indicated above, AlF3 is an inert polymer that neither dissolves in nor reacts with neutral ligands, and entry into its complexes is either via Cl/F exchange from corresponding chloro-complexes or from the hydrate  [20,147].
There is clearly great similarity between the metal trifluoride N-donor diimine and aza-macrocycle complexes of Al, Ga and In: all the hydrated complexes are stable to water and show extensive hydrogen bonding and π-stacking (of the diimines) interaction in the crystals. Variable temperature 1 H and 19 F{ 1 H} NMR spectroscopy show that in CD3OD solution reversible dissociation occurs in some of the gallium and indium complexes, suggesting the stability is Al > Ga > In as would be expected based upon the relative Lewis acidity of the metal centre [147]. InF3⋅3H2O dissolved slowly in hot dmso to give a poor yield of [InF3(dmso)(H2O)2] [149].

Group 14
The coordination chemistry of SiF4, GeF4 and SnF4 is well developed [7,155] and only a modest amount of new work has appeared.

Silicon
NHC complexes have attracted considerable interest and the area has been recently reviewed [140]. The new carbene complex trans-[SiF4(NHCIP2)2] was isolated during attempts to make SiF2 adducts. The structures of this complex and its toluene solvate were determined [156]. In the former the carbenes are arranged parallel whilst in the solvate the carbenes are perpendicular to each other; this is presumably a packing effect and makes no difference to the bond lengths. The carbene NHCAAC reacted with SiF4 in thf to form the five coordinate [SiF4(NHCAAC)] (Fig 48 C) [157]. Reduction of [SiF4(NHCAAC)] with two equivalents of KC8 in thf in the presence of NHCAAC produced the dark purple Si(II) complex [SiF2(NHCAAC)2] (Fig. 48 B) , whilst using one equivalent of KC8 generated the yellow radical [SiF3(NHCAAC)] (Fig. 48 A )[157] (Fig. 48).  [158], attempts to prepare SiF4 adducts have failed [159]. The failure to isolate phosphine complexes contrasts with the stability of the NHC complexes. Low temperature IR spectroscopy and DFT calculations were used to explore the stability of nitrile and substituted pyridine complexes of SiF4 [160]. The calculations and the IR data showed no significant adduct formation between SiF4 and nitriles.

Germanium
Carbene complexes of GeF4 and SnF4 are included in recent reviews [19,140]. Oxidative addition of the appropriate bis(dialkylamino)difluoromethylenes to [GeCl2(dioxane)] gave the carbene complexes  The acyclic carbene complex has a significantly longer Ge-C bond, but there is no difference in the Ge-F bond lengths between the two complexes. If excess of the cyclic difluoromethylene is used, the product is a sixcoordinate anion [GeF5(SNHCMe2)] -. The structure of [GeF4(FCH2CN)2] has been determined [162].  structure with fluorine bridges is present (Fig. 51) [166].  [167]. 19 F{ 1 H} NMR studies show the complexes exist as a mixture of cis and trans isomers in solution. Similar complexes of (Me2N)3PO, (R2N)2FPO and (R2N)F2PO have been described [168].

Group 15
Carbene adducts of phosphorus(V) fluorides have received considerable attention and the area has been reviewed recently [19,140,169]. In this section we have focussed on the synthesis and structures; the applications in organic synthesis are covered in these reviews. Some carbene complexes have been developed as possible 18 F carriers for PET imaging (Section 7). New work on antimony fluorides has also been reported, but no complexes of highly inert bismuth trifluoride are known.
The complexes are air stable and can be recrystallised from boiling water, but anhydrous HF in CH2Cl2 cleaves the P-carbene bond to form imidazolium hexafluorophosphates. Similar oxidative addition to PF3, PMeCl2 and PPhCl2 has been reported [178]. (Difluoroorganyl)dimethylamines RCF2NMe2 (R = H, Ph, t Bu) also oxidatively add to PF3 to give sterically non-demanding asymmetric carbene complexes [178]. Phosphorus(V) fluoro-complexes containing carbenes with liquid crystal chains have been obtained [179].
Phosphorus(V) carbene complexes have been examined as additives to improve the function of Li-ion battery electrolytes [146].

Arsenic
Neutral ligand complexes of AsF3 or AsF5 are few [7] and merit further investigation. A series of adducts of HCN and various nitriles with AsF5, [AsF5(L)] (L = HCN, PrCN, c-C3H5CN, t BuCN, PhCN, NCCH2CN) and [(AsF5)2(NCCH2CN)] has been prepared from AsF5 and the nitrile in liquid SO2 at low temperatures [181]. The adducts are colourless solids at low temperatures, but decompose slowly turning brown when warmed to room temperature. They were characterised by IR, Raman, 19 [189]. The X-ray crystal structure shows the cation to have a disphenoidal ("saw-horse") geometry with axial fluorines, and most unexpectedly, one carbene has rearranged and coordinated in the mesoionic form (Fig. 60). The structure of the 2:1 SbF3-glycine complex has been determined [190].

Group 17
The first (and so far only) adduct of BrF3 with a neutral ligand, [BrF3(py)] has been isolated as a white solid by combination of the reagents at low temperature [196]. The X-ray structure revealed a distorted square planar bromine centre (Fig 62).

Group 18
Neutral ligand adducts of xenon fluorides are very rare, but include [XeOF3(MeCN)] [197,198 [200]. The structures of both adducts were determined and detailed Raman spectroscopic characterisation reported. The Xe-N distances are long but significantly shorter than the sum of the appropriate Van der Waals radii; the bonding has been discussed in detail [201].

Complexes for 18 F Radiopharmaceutical Applications.
The Although many of these new systems have anionic ligands coordinated to the metal/non-metal fluoride, and hence formally fall outside the scope of this review, they will be presented, but not discussed in detail as they are already discussed in depth in several recent reviews [8,9,10,202,203,204,205,206,207,208]. These systems have attracted much interest due to their comparable or higher bond dissociation energies with fluorine compared to the C-F bond (Table 2)  The most prominent alternatives to C-18 F radiopharmaceuticals comprising anionic ligands are: • Silicon-fluoride-acceptors (SiFA) systems.
The SiFA chemistry has been developed by the Schirrmacher Group with the first example reported in 2006 [210]. In these molecules, the silicon atom has a tetrahedral coordination in which a single fluoride atom is present, along with a phenyl and two tert-butyl groups. The tert-butyl groups protect the silicon fluoride from hydrolysis and the aromatic group can be functionalised for biomolecule conjugation (Fig. 63).  [210,211,212,213]. However, the need to have bulky groups around the hydrolytically unstable silicon atom have the major drawback of increasing the lipophilicity of the molecule, leading to very high liver uptake observed through in vivo studies [210]. The problem was overcome by introducing lipophilicity-reducing groups between the SiFA moiety and the biomolecule, such as an acetylated amino-sugar group attached to the amino acid asparagine, and making the system cationic overall [214,215].
The formation of the radiolabelled [ 18 F]RBF3 − is achieved either by converting a boronic ester moiety into the trifluoroborate species [216,217,218] or by an isotopic exchange reaction in acidic solution or aprotic solvents [218]. Isotopic exchange reactions sometimes require the use of a Lewis acid promoter (e.g. SnCl4) [8]. Two main organotrifluoroborate species have been developed and successfully 18 F-fluorinated: aryltrifluoroborate and the zwitterionic onium-trifluoroborate [8] (Fig. 64). Since RBF3 − anions also suffer from hydrolysis, forming the corresponding boronic acid, modifications to the aryl groups were explored. It was concluded that electron-withdrawing groups in the phenyl ring, such as F, confer the required stability to the trifluoroborate moiety. The development of the zwitterionic species is also related to the effort of improving the stability in water of the trifluoroborate unit; in this case the presence of a cationic group (e.g. ammonium, phosphonium) enhances the fluoride ion affinity of the borates, acting as an electrostatic anchor for the oppositely charged anion [219,220] [224,225,226,227].
Thermodynamic and kinetic stability is ensured by the presence of the anionic macrocycle, NOTA (Fig. 65).
Despite the ligand being anionic, this chemistry is quite different from the other two examples presented above; this was the first metal-based chelate used for PET application, and the coordination environment does not involve C atoms, making the system and its chemistry quite different from the organo-B/Si chemistry discussed above. In this methodology, [ 18 F]AlF 2+ is first produced through Cl/ 18 F halide exchange reactions, from AlCl3 and [ 18 F]KF, and then reacted with the NOTA ligand (one carboxylic pendant arm is conjugated to a biomolecule), in water at pH 4 (CH3COONa buffer) and T >100°C. The resulting product has the aluminium metal centre octahedrally coordinated with a N3O2F environment around the metal. Various modifications of the linker group and/or in the macrocycle backbone have been reported [228,229].   reported good stability to at least pH 6 but defluorinates in phosphate buffered solution (PBS) and human serum albumin (HAS) at pH 7.4 [153].
[GaF(1-Bn-4,7-(CH3COO)2-tacn)] could also be obtained by reaction of Ga(NO3)3•9H2O, [1-Bn-4,7-(CH3COO)2] 2− and non-radioactive KF in water. The crystal structure of the fluoride complex is shown in Fig. 67. The X-ray structural characterisation shows a distorted octahedral coordination at gallium, through the pentadentate ligand and one terminal F − ligand. two-step radiosynthesis of an NHC-BF3 adduct in which the biomolecule was attached following 18 F-labelling, was reported [143]. The same approach was employed for the 18 F-labelling of the analogue NHC-PF5 (Scheme 9) [175]. The RCY were rather poor (6.5%), a problem attributed to the stability of the P-F bond in the system which hinders the 18 F/ 19 F isotopic exchange. However, the purified radioproduct showed very good stability in vitro and in vivo [175].
The use of neutral triazacyclononane macrocycles with the Group 13 metal trihalides (M = Al, Ga) was also While the Ga system was radiolabelled in aqueous MeCN solution at room temperature, the AlCl3-chelate required a buffered pH 4 solution and 80°C to give a good RCY (30% and 24%, respectively) [20,148]. The RCYs were measured from the radio-HPLC chromatograms of the crude reactions and the product identified by comparison with the retention time (Rt) in the UV-chromatogram of the equivalent non-radioactive product used as reference standard (Fig. 68). The products from both reactions showed good stability in vitro (EtOH/PBS solution pH 7.4) with a radiochemical purity (RCP) >90% after 2/3 hours (Fig. 69).  Both chloride complexes were radiolabelled using 1 mg per mL (2.63 μmol), however attempts to reduce the amount of the metal precursor complex employed were unsuccessful. This was attributed to hydrolysis, which competes with [ 18 F]F − ions at lower concentration. Likely factors contributing to the differences observed are the higher Al-F bond dissociation energy compared to Ga-F (Table 2) and the higher Lewis acidity and oxophilicity of the Al system. In consideration of the high stability of the [MF3(RMe2-tacn)] (R = Bn, Me) complexes (they can be obtained in hydrothermal conditions, 180°C 15h) [20], 18 F/ 19 F isotopic exchange reactions were investigated. Indeed, [GaF3(BnMe2-tacn)] was radiolabelled starting with 0.1 mg per mL (268 nmol) and also 0.01 mg per mL (27 nmol) in aqueous MeCN or EtOH solution at 80°C for 10 minutes, with good RCY (66±4% and 37±5% respectively) (Scheme 11and Fig. 70) [152].  The same method was successfully employed for the radiolabelling of [FeF3(BnMe2-tacn)] which could be radiolabelled using 1 mg, 0.1 mg per mL (2.31 μmol, 230 nmol) in ~ 40% RCY and 0.01 mg per mL (23 nmol) in ~15% RCY [87].
The investigation of other transition metals in the 3+ oxidation state (Sc, Y, La, Cr, Mn, Co) with RMe2-tacn (R = Me, Bz) was also reported [21,87]. The stability of the [M 19 F3(RMe2-tacn)] was firstly investigated in the presence of common anions present in physiological conditions (phosphate, chloride, acetate, carbonate), fluoride anions, pH and temperature range by means of 19 F{ 1 H} NMR, or UV spectroscopy in the case of the paramagnetic systems. The Y and La fluoride complexes were not obtained [21] and the Mn and Co systems did not show sufficient stability; the Cl/F exchange was too slow to be useful in the Cr system [87].
[ScF3(BnM2-tacn)] was stable in water over a range of temperatures and up until slightly basic pH (<8), as well as towards chloride and acetate anions. [ScF3(RM2-tacn)] could be formed from [ScCl3(RM2-tacn)] with [Me4N]F in anhydrous MeCN, representing a promising system worth further examination [21]. The complexes of MF3 (M = Al, Ga, In, Sc, Cr, Mn, Fe, Co) with the tridentate acyclic ligand terpy were also reported, however these systems are generally less stable and are not considered to be suitable complexes for future PET applications [20,21,87].
It seems obvious that the possibility of using metal coordination complexes as 18 F PET radiotracers will be strongly dependant on the properties of the metal centre, which ultimately influence the reaction conditions to form the fluoride complexes and their stability. The main metal properties to consider are its size (dictating the coordination number), its redox chemistry and oxophilicity (the fluoride complex must be stable in water or in the presence of anions such as phosphates), its Lewis acidity and lability (allowing rapid Cl/F or 18 F/ 19 F substitution). Considering that stability tests and Cl/F halide exchange reactions on the non-radioactive complexes are usually performed using between 15-100 mg of material, whereas 18 F radiolabelling reactions require very small quantities of precursor and the amount of [ 18 F]F − is also extremely low, stability and reactivity of the complexes in the two regimes may be very different. At very low concentration, trace impurities, the vast excess of water and 19 F − in the mixture can compete with [ 18 F]F − or disrupt the coordination around the metal, resulting in low RCY or a decrease in RCP over time. The use of neutral ligand systems is of recent origin and many other possibilities for suitable metal/ligand combinations for 18 F incorporation are conceivable, and will no doubt attract further studies.

Applications
It is evident from the work described in Sections 4-6, that the synthesis and study of metal and non-metal fluoride complexes containing neutral donor ligands is an area of significant current investigation. In addition to their inherent chemical interest, a number of potential applications of the complexes are beginning to emerge. A good example is the use of metal complexes as binding sites for 18 F in the drive towards new generations of PET imaging agents (Section 7). Examples now include complexes involving direct bonds from 18 F to P, Al, Ga, and Fe, as well as other inorganic anions based upon B and S. The next stages of this work are likely to focus on bioconjugation of peptides and PET imaging studies. A completely different application is the possible use of main group boron and phosphorus fluoride-carbene complexes as electrolyte additives to increase the stability and lifetime of Li-ion batteries [131,132,140,146,169].
The longest established use of non-metal fluorides is in organic and organometallic synthesis. For example, boron trifluoride or its more easily handled adducts, [BF3(L)] (L = Et2O, Me2S, MeOH, etc.) are very widely used as Lewis acids for many organic transformations [126]. More recently, carbene complexes of boron and phosphorus fluorides have attracted much interest as active Lewis acids [140]. Many other fluoride systems have been investigated as possible Lewis acids or catalysts for organic transformations, including olefin polymerisation and C-F bond formation. The different properties conferred on the central element by fluoride ligands usually results in clearly different behaviour compared to corresponding complexes with other anions. Species based upon Ti(IV) fluoride have numerous applications for stoichiometric or catalytic transformations, as described in a recent review [14]. These are typically metallocene based or contain anionic pincer co-ligands, with the catalyst or pre-catalyst prepared in situ; usually the active catalyst is not identified. Thus far neutral ligand complexes feature only as precursors to introduce the metal fluoride in a soluble form, e.g. as [TiF4(thf)2], but further work can be expected to identify complexes that are themselves active. The specific effects of fluoride ligands are illustrated by olefin polymerisation using the pincer ligand in (Fig. 71) [241]. This complex, prepared from [TiF4(thf)2] and the pincer by HF elimination, and activated with MAO (methylaluminoxane), produces ultra-high molecular weight polyethylene. The corresponding chloro-complex is also highly active, but gives a different molecular weight distribution. Other metal systems, including for example, simple complexes of NbF5, achieve ethylene polymerisation, but the activity is low [66].  Fig. 71 The (anionic) pincer ligand pre-catalyst for olefin polymerisation [241].
Late d-block fluoride complexes, including those of nickel, palladium and gold, have also been investigated in connection with C-C coupling, C-H activation or C-F bond formation, and described in recent reviews [119,242,243]. The active species are often proposed to be M-F compounds in higher oxidation states, but it is rare for these to be identified and they have been isolated in only a few cases -see for example

Conclusions and Outlook.
Over the approximately eight years covered by this article, there have been many advances in the coordination chemistry of metal fluorides, including the first N-base complexes of AuF3, phosphine and arsine complexes of the highest oxidation states of the heavy group 5 and 6 elements, NbF5, TaF5 and WF6, and an impressive number of carbene (NHC) complexes with many element fluorides across the Periodic Table. In appear worth investigation. The tri-, tetra-and possibly penta-fluorides of the platinum metals may complex with nitrile or pyridyl ligands, although this will require elemental fluorine to produce the binary fluoride precursors, as well as fluoroplastic or metal equipment, which limits such studies to a small number of laboratories. Even the complexes of the 3d metal trifluorides were little studied until late, but this has changed as a result of their potential applications in molecular magnets and as 18 F carriers for PET applications, and it seems that a wider range of ligand types could be incorporated.
Within the p-block, areas worthy of investigation include cationic fluoro complexes of the Group 13 metals which will have enhanced Lewis acidity compared to the neutral analogues, and detailed studies of the NHC complexes of AsF5 and SbF5 to compare with the large amount of work on PF5 systems. Sulfur tetrafluoride forms only a few thermally unstable adducts, but complexes of TeF4 have been little studied, and SeF4 does not seem to have been examined. Nitrile and pyridyl complexes of Group 17 are also possible -while this is hazardous chemistry, the recent synthesis of [BrF3(py)] may lead to further examples (some complexes of iodine fluorides were reported over 40 years ago, but lack structural authentication).
Oxide fluoride chemistry is more limited, but recent work has included systematic studies of vanadium, molybdenum, niobium and tungsten oxide fluoride complexes with N-and O-donor ligands, and even for tungsten, some phosphine examples. A significant number of oxide fluorides exist for Re and Os and isolated examples of nitrile complexes have been characterised, but most have not been explored; again the need for fluorine and specialised equipment is a significant barrier. Molecular oxide fluorides in the p-block are mostly with the lighter non-metallic elements, and these appear to have little or no Lewis acidity.
It seems likely that the next few years will see some at least of this new chemistry attempted and a number of exciting new developments are to be expected.