Synthesis and Characterisation of Lanthanide N-trimethylsilyl and -mesityl Functionalised Bis(iminophosphorano)methanides and -methanediides

We report the extension of the series of {BIPM

Whilst a range of complexes of the formula [{Ln(BIPM TMS H)(Cl)(µ-Cl)} 2 ] have been reported previously by the reaction of [Ln(Cl) 3 ] and [{K(BIPM TMS H)} 2 ] [20], we have solely utilised iodide precursors.This is to reduce the likelihood of salt occlusion occurring as KI, RbI and CsI have a much lower propensity for salt occlusion in comparison to LiCl or KCl, and as salt occlusion can often negatively affect reactivity and promote unwanted side reactions this is a major advantage of utilising [Ln(I) 3 (THF) n ] precursors.As ligand scrambling and Schlenk-type equilibra proved problematic in alkane elimination strategies, we focused on salt metathesis strategies in our attempts to complete the range of rare earth BIPM R H (R = TMS, Mes) complexes of the general formula [Ln(BIPM R H)(I) 2 (THF) n ].These complexes would provide an insight into the structure and bonding of rare earth complexes as well as having utility in the preparation of methanediide complexes of the general formula [Ln(BIPM R )(I)(THF) n ].With these complexes in hand we would be able to investigate what, if any, effect the varying metal size would have on their structure and reactivity.

Results and Discussion
As rare earth ions are predominantly paramagnetic, any NMR spectroscopic studies performed are often difficult to interpret, which makes it problematic to both follow reaction progress and to identify the products of reactions.For this reason single crystal X-ray diffraction studies are essential to unambiguously identify the outcome of reactions involving paramagnetic rare earth metals.Unfortunately, we often observe that some complexes do not crystallise readily, despite repeated attempts, one reason for which may be that they do not possess an optimal metal/ligand size ratio for crystal growth [21].For this reason we commonly employ a range of rare earths of varying sizes, Table 1, in our studies to ensure we have the best chance at fully identifying the products of reactions.Due to the difficulties in identifying reaction products without X-ray diffraction studies, we have only included structurally authenticated complexes in this report.Table 1.Six-coordinate ionic radii and covalent radii (Å) of Ln(III) complexes [22,23].

Preparation of BIPM TMS H Rare Earth Methanides
As previously discussed, we have reported the preparation of [Ln(BIPM TMS H)(I) 2 (THF)] (Ln= La, Ce) in high yields (77% and 89%, respectively) via the reaction of [Ln(I) 3 (THF) 4 ] with the heavy group 1 ligand transfer reagent [Rb(BIPM TMS H)(THF) n ] under ambient conditions [11,13].As this strategy was successful for the two largest rare earth metals (see Table 1), we extended this methodology across the rare earth series using the cesium ligand transfer reagent to maximise our chances of success.The reactions of [Ln(I) 3 (THF) 3.5 )] (Ln= Nd, Gd, Tb) with [Cs(BIPM TMS H)] afforded, following filtration and workup, [Ln(BIPM TMS H)(I) 2 (THF)] (Ln= Nd, 1a, 37%; Gd, 1b, 50%; Tb, 1c, 34%) with concomitant elimination of CsI (Scheme 1).As Nd(III), Gd(III) and Tb(III) are highly paramagnetic, NMR spectroscopic studies were inconclusive, and due to this each complex was crystallised from either THF or toluene to ensure purity of the sample.Unfortunately this also led to lower isolated yields (36-54%) compared to the diamagnetic La (77%) or weakly paramagnetic Ce (89%) analogues which could both be identified by NMR spectroscopy.

Scheme 1. Preparation of 1a-c.
The identities of 1a-c were confirmed by single crystal X-ray diffraction studies, elemental analysis and solution magnetic studies (vide infra, Table 2).The solid-state structures of 1a C 7 H 8 and 1c OC 4 H 8 are shown in Figures 1 and 2, respectively, with selected bond lengths and angles shown in Table S1.Each complex exhibits a distorted octahedral geometry with the metal centre coordinated by tridentate {BIPM TMS H} − through the methanide carbon atom and the two imino nitrogens [N-Ln-N angles: 1a, 116.28( 14 1).However, this could also be due to solvent effects on crystallisation favouring different conformations during crystal packing as 1a and [Ln(BIPM TMS H)(I) 2 (THF)] (Ln = La, Ce) [11,13] were crystallised from toluene and adopt the trans geometry, whereas 1b-c and [Ln(BIPM TMS H)(I) 2 (THF)] (Ln = Dy, Y) [7,12] were crystallised from THF and adopt the cis conformation.There is a general shortening of the Ln-C, Ln-N, Ln-I and Ln-O bond distances between 1a-1c which is in agreement with the lanthanide contraction (see Table 1) [19], and is consistent with the bond distances reported for [Ln(BIPM TMS H)(I) 2 (THF)] (Ln = La, Ce, Dy, Y) [7,[11][12][13].
The bond distances about the metal centres in 1a-c are all within the range of previously reported bond distances [24], and are unremarkable.The mean endocyclic P-C and P-N bond distances in 1a-c and the previously reported [Ln(BIPM TMS H)(I) 2 (THF)] (Ln = La, Ce, Dy, Y) [7,[11][12][13], are statistically indistinguishable suggesting the {BIPM TMS H} − coordinates to the metal centre in a similar way despite the variation in metal centre.During our studies we have prepared many samples of [Ln(BIPM TMS H)(I) 2 (THF)] and on one occasion during preparation of [La(BIPM TMS H)(I) 2 (THF)] we isolated, from toluene solution, a small crop of colourless crystals which appeared to be of a different morphology to our previously reported [La(BIPM TMS H)(I) 2 (THF)] [13].To ensure purity of the sample a single crystal X-ray diffraction study was performed which confirmed its identity as [{La(BIPM TMS H)(I)(µ-I)} 2 ], 2 2C 7 H 8 (Scheme 2).Complex 2 is dimeric in the solid-state with each lanthanum centre adopting a heavily distorted octahedral geometry and being coordinated by a {BIPM TMS H} − in a tridentate fashion [N1-La1-N2 angle: 118.3(2)°], by one terminal iodide and by two iodides that are bridging the two lanthanide centres.The dimeric motif of 2 is likely caused by a scarcity of coordinating solvent present during crystallisation with the dimeric form allowing greater saturation of the coordination sphere of the lanthanide centres compared to a solvent free monomeric form.Once isolated, 2 has a very low solubility in arene solvents (e.g., d 6 -benzene and d 8 -toluene), which precluded solution NMR spectroscopic studies, while dissolution in more polar solvents to increase solubility (e.g., d 8 -THF) led to solvated monomeric complexes such as [La(BIPM TMS H)(I) 2 (d 8-THF)].Although NMR spectroscopy was not conclusive, the identity of 2 was confirmed by elemental analysis.The dimeric form of 2 is analogous to the chloride congeners reported by Roesky et al., namely [{Ln(BIPM TMS H)(Cl)(µ-Cl)} 2 ] (Ln = Sm ,Dy, Er, Yb, Lu, Y), which despite being prepared in THF solution adopt solvent free dimeric conformations when crystallised from toluene [20].
The solid-state structure of 2 2C 7 H 8 is depicted in Figure 3 with selected bond lengths and angles shown in Table S1.There is a variation in La1-I bond distances in 2, with the two bridging iodides being bound asymmetrically, with bond lengths of 3.2253(8) Å (La1-I2) and 3.3114( 6) Å (La1-I2a), but both are still longer than the La1-I1 distance of 3.1144

Preparation of {BIPM Mes H} Rare Earth Methanides
We have previously reported the preparation of [Ln(BIPM Mes H)(I) 2 (THF) 2 ] (Ln = La, Ce, Pr, Nd, Sm) in variable yield (40-78%) via the reaction of [Ln(I) 3 (THF) n ] with [{K(BIPM Mes H)} 2 ] [12,14].In contrast to the preparation of the {BIPM TMS H} − analogues (vide supra), no forcing conditions or use of heavier group 1 ligand transfer reagents (e.g., [Cs(BIPM Mes H)]) was required.This is likely due to the decreased steric bulk of the {BIPM Mes H} − ligand compared to {BIPM TMS H} − leading to more reactive ligand transfer reagents as the alkali metal is less sterically shielded.
As we had previously prepared [Ln(BIPM Mes H)(I) 2 (THF) 2 ] for the larger rare earths (Ln = La, Ce, Pr, Nd, Sm; See Table 1), we attempted to extend this series to investigate any changes in structure or reactivity imparted on the complex by varying the metal centre.Analogous to our previous reports, the reaction of half an equivalent of [{K(BIPM Mes H)} 2 ] with [Ln(I) 3 (THF) 3.5 ] (Ln = Gd, Yb) afforded [Gd(BIPM Mes H)(I) 2 (THF) 2 ], 3, and [Yb(BIPM Mes H)(I) 2 (THF)], 4, respectively (Scheme 3).Each complex was identified by elemental analyses, solution magnetic studies (vide infra) and by single crystal X-ray diffraction studies.However, while the X-ray diffraction data for 3 3C 7 H 8 is of good quality (R int = 0.034, R = 0.0419, R w = 0.109) and confirmed its identity as [Gd(BIPM Mes H)(I) 2 (THF) 2 ] 3C 7 H 8 , the data-set obtained for 4 3C 7 H 8 is poor (R int = 0.0744, R = 0.1097, R w = 0.2865), and despite exhaustive attempts more satisfactory data could not be obtained.While the data-set collected on crystals of 4 3C 7 H 8 is of poor quality and precludes any assessment of the metrical parameters of the complex, the connectivity is clear-cut, and together with elemental analysis and comparison to lighter rare earth analogues, we are confident in our assignment of 4 .The difficulty in obtaining crystals of 4 of suitable quality for single crystal X-ray diffraction studies is perhaps surprising as [Ln(BIPM Mes H)(I) 2 (THF) 2 ] (Ln = La, Ce, Pr, Nd, Sm, Gd) all crystallise readily from toluene solutions to afford large crystals suitable for single crystal X-ray diffraction studies.The change in crystallinity is possibly due to 4 possessing only one THF molecule coordinated to ytterbium, compared to the two THF molecules coordinated in [Ln(BIPM Mes H)(I) 2 (THF) 2 ] (Ln = La, Ce, Pr, Nd, Sm, Gd) [12,14].This variation in coordination number and ligand environment may lead to less efficient crystal packing, however, the lack in crystallinity may also be due to 4 not containing the optimal metal/ligand size ratio for crystal growth [21].
The solid-state structure of 3 3C 7 H 8 is shown in Figure 4, with selected bond lengths and angles shown in Table S1.As 3 3C 7 H 8 crystallises with the same cell settings as [Ln(BIPM Mes H)(I) 2 (THF) 2 ] (Ln = La, Ce, Pr, Nd, Sm) [12,14], they adopt the same distorted pentagonal bipyramidal structure, with the gadolinium centre being coordinated by tridentate {BIPM Mes H} − [N1-Gd1-N2 angle: 113.23 (10)°], two iodides and two THF molecules.The two iodides can be considered as occupying the axial sites, with the I1-Gd1-I2 angle of 155.765(9)° revealing the degree of distortion away from idealised pentagonal bipyramidal geometry due to the coordination of the bulky {BIPM Mes H} − ligand.As expected, due to the lanthanide contraction, the bond distances about the gadolinium centre in 3 are generally shorter than those observed in [Ln(BIPM Mes H)(I) 2 (THF) 2 ] (Ln = La, Ce, Pr, Nd, Sm), due to the smaller radius of Gd(III) (Table 1), and are each within the range of previously reported distances [24], but are otherwise unremarkable.The {BIPM Mes H} − ligand in 3 appears to be bound to the metal centre in an analogous manner to [Ln(BIPM Mes H)(I) 2 (THF) 2 ] (Ln = La, Ce, Pr, Nd, Sm) [12,14], as evidenced by the mean endocyclic P-C, P-N distances and P1-C1-P2 angle in 3 Whilst we were successful in the preparation of [Ln(BIPM Mes H)(I) 2 (THF) n ] for the larger rare earths (Ln = La, Ce, Pr, Nd, Sm, Gd) and smaller rare earth ytterbium, we had difficulty in the preparation of analogues for intermediate sized rare earths, namely dysprosium and erbium (Table 1).It is difficult to follow reaction progress between [Ln(I) 3 (THF) 3.5 ] (Ln = Dy, Er) and half an equivalent of [{K(BIPM Mes H)} 2 ], as the highly paramagnetic nature of Dy(III) (4f 7 ) and Er(III) (4f 9 ) lead to inconclusive NMR spectra, and also because [Ln(I) 3 (THF) 3.5 ] has low solubility in THF so the elimination of KI cannot be monitored easily as a suspension is observed throughout the reaction.As we followed the same preparative method that yielded [Ln(BIPM Mes H)(I) 2 (THF) n ] (vide supra), we worked up the reaction mixture in a similar manner, namely filtration and recrystallisation from hot toluene solutions.In the cases of dysprosium and erbium this did not afford [Ln(BIPM Mes H)(I) 2 (THF) n ] (Ln = Dy, Er) as expected, but instead led to the isolation of the separated ion pair species [Ln(BIPM Mes H) 2 ][BIPM Mes H] (Ln = Dy, 5a; Er, 5b) in moderate yields (5a: 39%; 5b: 31%, based on BIPM) (Scheme 3).Our hypothesis for the isolation of 5a-b is that the reaction of [Ln(I) 3 (THF) 3.5 ] (Ln = Dy, Er) with half an equivalent of [{K(BIPM Mes H)} 2 ] proceeds sluggishly under ambient conditions, primarily due to the inherent insolubility of [Ln(I) 3 (THF) 3.5 ] (Ln = Dy, Er) in THF [25].During workup, filtration separates insoluble [Ln(I) 3 (THF) 3.5 ] (Ln = Dy, Er) and yields a solution of [{K(BIPM Mes H)} 2 ] in excess to [Ln(I) 3 (THF) 3.5 ] (Ln = Dy, Er), which, when heated in toluene during attempted recrystallisation, would lead to a rapid reaction, where [Ln(I) 3 (THF) 3.5 ] (Ln = Dy, Er) would effectively react with three equivalents of [{K(BIPM Mes H)} 2 ] to afford 5a-5b, with concomitant elimination of KI.
Repeated attempts to prepare [Ln(BIPM Mes H)(I) 2 (THF) n ] (Ln = Dy, Er) by varying reaction times and solvents (THF, 1,2-dimethoxyethane, Et 2 O) were not successful, and when forcing conditions were employed the reaction proceeded rapidly rather than in a slow and controlled manner, leading to isolation of 5a-b.Complexes 5a-b were identified by solution magnetic moment measurements (vide infra), elemental analyses and by single crystal X-ray diffraction studies.As complexes 5a 4C 7 H 8 and 5b 5C 7 H 8 have very similar solid-state structures and metrical parameters, only 5a 4C 7 H 8 is discussed in detail.Selected bond lengths and angles of 5a 4C 7 H 8 and 5b 5C 7 H 8 are shown in Table S1.The cationic [Dy(BIPM Mes H) 2 ] + component of 5a is shown in Figure 5, while, for clarity, the anionic [BIPM Mes H] − component is shown separately in Figure 6.
The dysprosium centre in 5a is coordinated by two mutually orthogonal {BIPM Mes H} − ligands, each exhibiting a tridentate binding mode, leading to a distorted octahedral geometry.Each {BIPM Mes H} − is coordinated to dysprosium in a similar mode, as shown by the statistically indistinguishable Dy1-C1    12 The 31 P{ 1 H} NMR spectrum of 6a exhibits a resonance at 15.70 ppm which is slightly downfield to the corresponding resonance reported for [La(BIPM Mes H)(I) 2 (THF) 2 ] (11.46 ppm) [12].
While elemental analysis and NMR spectroscopy supported the formulation of 6a, a single crystal X-ray diffraction study was performed to confirm its structure.The solid-state structure of 6a 3C 7 H 8 is shown in Figure 7, with selected bond lengths and angles complied in Table S1.The lanthanum centre in 6a adopts a heavily distorted pentagonal bipyramidal geometry with the two iodides occupying the axial sites [I1-La1-I2 angle: 158.395( 12 Following the successful isolation of 6a, we prepared a heavier rare earth analogue, namely [Gd(BIPM Mes H)(I) 2 (TMEDA)], 6b, via the same methodology.As Gd(III) is smaller than La(III) (Table 1) this would provide us with a range of metal sizes to fully investigate if the size of the metal centre has any effect on the successful preparation of our target {BIPM Mes } 2− methanediide complexes (vide infra).Complex 6b was prepared and utilised in situ in attempted deprotonation/salt elimination reactions, but a small sample was recrystallised from toluene to afford single crystals suitable for X-ray diffraction studies.Although 6b crystallises in a different cell setting to 6a (Cmca vs. P-1), leading to the TMEDA molecule possessing positional disorder about the mirror plane, the overall structures are very similar with the gadolinium centre in 6b adopting a heavily distorted pentagonal bipyramidal geometry.As expected due to the lanthanide contraction the bond distances about the gadolinium centre in 6b are each shortened by ca.0.07-0.14Å compared to the corresponding distances about lanthanum in 6a, and are all well within the range of previously reported bond distances [24].

Attempted Preparation of BIPM Mes Rare Earth Carbene Complexes
We have previously reported that the reaction of [La(BIPM Mes H)(I) 2 (THF)] with [K(Bn)] afforded the methanediide complex [La(BIPM Mes )(I)(THF) 3 ] in 53% yield [12].Unfortunately, this reaction appears to be capricious in nature, and despite following the reported methodology the reaction often yields a mixture of products from which [La(BIPM Mes )(I)(THF) 3 ] cannot be isolated cleanly.As NMR spectroscopy revealed a mixture of products, of which none could be unambiguously identified, the reaction mixture was recrystallised from toluene which afforded [La(BIPM Mes )(BIPM Mes H)], 7a, in 8% yield.The low yield of 7a is a reflection of the reaction not proceeding smoothly and affording a mixture of products, of which only 7a was crystalline and able to be extracted cleanly.Complex 7a was identified by single crystal X-ray diffraction studies (vide infra), NMR studies and elemental analysis.The 31 P{ 1 H} NMR spectrum of 7a exhibits two singlet resonances, at 2.88 and 10.51 ppm, which are assigned to the methanediide and methanide centres, respectively, by comparison to the resonances reported for [La(BIPM Mes )(I)(THF) 3 ] (4.84 ppm) and [La(BIPM Mes H)(I) 2 (THF) 2 ] (11.46 ppm) [12].The 1 H and 13 C{ 1 H} NMR spectra of 7a reveal methanide resonances of 2.26 (no coupling observed) and 4.59 ppm (J PC = 127 Hz), which are both upfield of the values reported for [La(BIPM Mes H)(I) 2 (THF) 2 ] (3.36 ppm, 2 J PH = 14.5 Hz and 5.73 ppm, J PC = 136 Hz, respectively), while the methanediide is observed in the 13 C{ 1 H} NMR spectrum of 7a at 45.06 ppm, which is also upfield to the value reported for [La(BIPM Mes )(I)(THF) 3 ] (58.76 ppm, J PC = 148.4Hz) [12].
The isolation of 7a is reminiscent of our attempts to prepare [La(BIPM TMS )(Bn)(THF) n ] via the reaction of [La(Bn) 3 (THF) 3 ] and BIPM TMS H 2 which instead afforded [La(BIPM TMS )(BIPM TMS H)], the BIPM TMS analogue of 7a [16].In the cases of both 7a and [La(BIPM TMS )(BIPM TMS H)], the observed products are likely due to the instability of the intermediate complex [La(BIPM Mes H)(Bn)(I)(THF) n ] or [La(BIPM TMS H)(Bn) 2 (THF) n ], which would each be susceptible to ligand scrambling due to the highly labile nature of lanthanide alkyl bonds.
As  The complexes 7a-d exhibit very similar solid-state structures, with the rare earth metal coordinated by mutually orthogonal {BIPM Mes } 2− and {BIPM Mes H} − ligands, each bound in a tridentate fashion.A representative structure of 7a 3C 7 H 8 is shown in Figure 8 with selected bond lengths and angles for each complex shown in Table S1.Despite exhibiting the same overall structure there is a variation in metrical parameters between 7a, c-d and 7b, which we propose is due to a variation in solvent of crystallisation.As each of 7a-d exhibit one {BIPM Mes H} − ligand bound to the metal centre as a methanide and one {BIPM Mes } 2− ligand bound as a methanediide, varying Ln-C distances would be expected as the metal centre would have an increased electrostatic interaction to the dianionic methanediide centre over the monoanionic methanide centre.This is the case for 7b, which exhibits a longer Ce1-C1 distance of 2.819(11) Å, and a shorter Ce1-C44 distance of 2.681(11) Å, which leads to the assignment of C1 as being the methanide centre, and C44 being the methanediide centre.This is also analogous to the previously reported BIPM TMS congeners [Ln(BIPM TMS )(BIPM TMS H)] (Ln = La, Ce, Pr, Sm, Gd), which each exhibit one long and one short Ln-C interaction [16].
However, in the cases of 7a and 7c-d, each {BIPM Mes } ligand appears to be bound to the metal in an identical manner, which is exemplified by statistically indistinguishable Ln-C bond distances in each case [7a: La1-C1: 2.725(5) Å, La1-C44: 2.731(4) Å; 7c, Pr1-C1: 2.667(6) Å, Pr1-C44: 2.662(6) Å; 7d, Gd1-C1: 2.605(11) Å, Gd1-C44: 2.573 (10) Å].We propose that this variation in coordination is simply an artefact of crystallisation, with the crystal packing in 7a,c-d being random, with the resulting data-set revealing an averaged geometry about the metal centre leading to equivalent Ln-C bond distances.This is supported by the mean La1-C distances in 7a of 2.728(5) Å being intermediate to the Ce1-C distances in 7b of 2.681(11) and 2.819(11) Å, which would be expected if the two bond distances were averaged out.The other possibility is that each {BIPM Mes } ligand in 7a is bound as a methanide leading to a formal oxidation state assignment of La(II), analogous to the previously reported [Sm(BIPM Mes H) 2 ] [27], but this is easily ruled out by 7a being diamagnetic and its 31 P{ 1 H} NMR spectrum exhibiting two resonances consistent with the presence of two different ligand environments.

Solution State Magnetic Properties of BIPM TMS and BIPM Mes Rare Earth Complexes
The room temperature solution magnetic moments of 1a-c, 3, 4, 5a-b and 7b were determined utilising the Evans method [28].These are compiled in Table 2 along with their ground state terms and theoretical magnetic moments.The theoretical magnetic moments of rare earth complexes can be approximated by µ J = gJ√J(J + 1) [where gJ = 3/2 + [S(S + 1) − L(L + 1)]/2J(J + 1)] according to the Van Vleck equation for magnetic susceptibility [29], assuming the 2S+1 L J ground state is well separated from excited states and crystal field splitting is negligible.This is generally true for rare earths, with the exception of Sm(III) and Eu(III), which exhibit low lying excited states of 6 H 7/2 and 7 F 1 , respectively, which each contribute to the room temperature magnetic moments [19].As shown in Table 2 there is reasonable agreement between the theoretical and observed magnetic moments for each complex reported in this work.a µ J = gJ√J(J + 1) where gJ = 3/2 + [S(S + 1) − L(L + 1)]/2J(J + 1).

Experimental Section
All manipulations were carried out using standard Schlenk techniques, or an MBraun UniLab glovebox, under an atmosphere of dry nitrogen.Solvents were dried by passage through activated alumina towers and degassed before use.All solvents were stored over potassium mirrors (with the exception of THF which was stored over activated 4 Å molecular sieves).Deuterated solvents were distilled from potassium, degassed by three freeze-pump-thaw cycles and stored under nitrogen.

X-ray Crystallography
Crystal data for compounds 1-7d are given in Table S2.Bond lengths and angles are listed in Table S1.Crystals were examined variously on a Bruker APEX CCD area detector diffractometer using graphite-monochromated MoKα radiation (λ = 0.71073 Å), or on an Oxford Diffraction SuperNova Atlas CCD diffractometer using mirror-monochromated CuKα radiation (λ = 1.5418Å).Intensities were integrated from data recorded on 0.3 (APEX) or 1° (SuperNova) frames by ω rotation.Cell parameters were refined from the observed positions of all strong reflections in each data set.Semi-empirical absorption correction based on symmetry-equivalent and repeat reflections (APEX) or Gaussian grid face-indexed absorption correction with a beam profile correction (Supernova), were applied.The structures were solved variously by direct and heavy atom methods and were refined by full-matrix least-squares on all unique F 2 values, with anisotropic displacement parameters for all non-hydrogen atoms, and with constrained riding hydrogen geometries; U iso (H) was set at 1.2 (1.5 for methyl groups) times U eq of the parent atom.The largest features in final difference syntheses were close to heavy atoms and were of no chemical significance.The data-set obtained for 4 is of low quality, and while the connectivity is clear, no assessment could be made of the geometric parameters, and despite exhaustive attempts, a better data-set could not be obtained.Highly disordered solvent molecules of crystallisation in 4 and 7a-d could not be modelled and were treated with the Platon SQUEEZE procedure [33].Programs were Bruker AXS SMART [34] and CrysAlisPro (control) [35], Bruker AXS SAINT [34] and CrysAlisPro (integration) [35], and SHELXTL [36] and OLEX2 [37] were employed for structure solution and refinement and for molecular graphics.Crystal data have been deposited with the Cambridge Structural Database CCDC numbers 970500-970513.

Figure 1 .
Figure 1.Molecular structure of 1a C 7 H 8 with selected atom labelling.Displacement ellipsoids are drawn at 50% probability, with lattice solvent and non-methanide hydrogen atoms omitted for clarity.

Figure 2 .Scheme 2 .
Figure 2. Molecular structure of 1c OC 4 H 8 with selected atom labelling.Displacement ellipsoids are drawn at 50% probability, with disordered components, lattice solvent and non-methanide hydrogen atoms omitted for clarity.The structure of 1b is very similar.

Figure 3 .
Figure 3. Molecular structure of 2 2C 7 H 8 with selected atom labelling.Displacement ellipsoids are drawn at 50% probability, with lattice solvent and non-methanide hydrogen atoms omitted for clarity.

Figure 4 .
Figure 4. Molecular structure of 3 3C 7 H 8 with selected atom labelling.Displacement ellipsoids are drawn at 50% probability, with lattice solvent and non-methanide hydrogen atoms omitted for clarity.

Figure 7 .
Figure 7. Molecular structure of 6a 3C 7 H 8 with selected atom labelling.Displacement ellipsoids are drawn at 50% probability, with lattice solvent and non-methanide hydrogen atoms omitted for clarity.The molecular structure of 6b 2C 7 H 8 is very similar.

Figure 8 .
Figure 8.Molecular structure of 7a 3C 7 H 8 with selected atom labelling.Displacement ellipsoids are drawn at 50% probability, with lattice solvent and non-methanide hydrogen atoms omitted for clarity.The molecular structures of 7b-d are very similar.

Table 2 .
Theoretical and observed solution magnetic moments.