Symmetric Assembly of a Sterically Encumbered Allyl Complex: Mechanochemical and Solution Synthesis of the Tris(allyl)beryllate, K[BeA′₃] (A′ = 1,3-(SiMe₃)₂C₃H₃)

Abstract

The ball milling of beryllium chloride with two equivalents of the potassium salt of bis(1,3-trimethylsilyl)allyl anion, K[A′] (A′ = [1,3-(SiMe₃)₂C₃H₃]), produces the tris(allyl)beryllate K[BeA’₃] (1) rather than the expected neutral BeA’₂. The same product is obtained from reaction in hexanes; in contrast, although a similar reaction conducted in Et₂O was previously shown to produce the solvated species BeA’₂(OEt₂), it can produce 1 if the reaction time is extended (16 h). The tris(allyl)beryllate is fluxional in solution, and displays the strongly downfield ⁹Be NMR shift expected for a three-coordinate Be center (δ22.8 ppm). A single crystal X-ray structure reveals that the three allyl ligands are bound to beryllium in an arrangement with approximate C₃ symmetry (Be–C (avg) = 1.805(10) Å), with the potassium cation engaging in cation–π interactions with the double bonds of the allyl ligands. Similar structures have previously been found in complexes of zinc and tin, i.e., M[M′A′₃L] (M′ = Zn, M = Li, Na, K; M′ = Sn, M = K; L = thf). Density functional theory (DFT) calculations indicate that the observed C₃-symmetric framework of the isolated anion ([BeA′₃]⁻) is 20 kJ·mol⁻¹ higher in energy than a C₁ arrangement; the K⁺ counterion evidently plays a critical role in templating the final conformation.

1. Introduction

The physical and chemical properties of first-row elements often differ appreciably from their second-row and heavier counterparts; for the group 2 metals, the outlier (”black sheep” [1]) designation belongs to beryllium. To a considerably greater extent than its heavier congeners, even magnesium, the small size of the Be²⁺ cation (0.27 Å for CN = 4; cf. 0.57 Å for Mg²⁺) [2] and its corresponding high charge/size ratio ensures its bonds will be strongly polarized and possess substantial covalent character. Not surprisingly, beryllium compounds with the same ligand sets commonly have different structures from those of the other, more electropositive alkaline earth (Ae) metals. The bis(trimethylsilyl)amides of Mg–Ba, for example, have a common dimeric bridged structure, [Ae(N(SiMe₃)(µ-N(SiMe₃)₂]₂ [3], whereas that of beryllium is a two-coordinate monomer [4]. Similarly, the bis(cyclopentadienyl) complex Cp₂Be has an η¹,η⁵-Cp structure [5] that is unlike that of the heavier metallocenes [6]. Investigation of these differences, and indeed research with all beryllium compounds, has traditionally been limited because of concerns about toxicity [7], but that has not prevented its compounds from serving as useful benchmarks of the steric and electronic consequences of crowded metal environments [8,9,10].

One of these consequences is the relative stability of η¹- vs. η³-bonded allyl ligands in compounds of highly electropositive metals. We found some time ago that the bulky allyl [A′]⁻ (A′ = [1,3-(SiMe₃)₂C₃H₃]) can be used to form the ether adduct BeA’₂∙OEt₂, which displays η¹-bonded A’ ligands in the solid state [11]. The compound is fluxional in solution, and exhibits symmetric, “π-type” bonding in its NMR spectra (e.g., only one peak is observed for the SiMe₃ groups). Density functional theory (DFT) calculations suggested that a base-free Be(C₃H₃E₂)₂ (E = H, SiH₃) complex would be more slightly more stable with delocalized, π-type allyls than with monodentate, sigma-bonded ligands (Scheme 1). If so, beryllium allyls would join those of magnesium, in which monodentate allyl ligands are uniformly found in complexes that are ether-solvated [12], but that in the absence of ethers, cation–π interactions with the metal can create “slipped-π” bonding [13].

The coordinated ether in BeA’₂∙OEt₂ proved impossible to remove without destroying the complex [11], and thus we investigated mechanochemical methods of synthesis as a means to bypass the use of ethereal solvents [14]. As detailed below, an unsolvated neutral complex was not isolated via this route, and the beryllate anion that was produced instead has structural parallels with previously described -ate complexes of Zn [15] and Sn [16]. In all of these species, the alkali metal counterion, usually K⁺ but sometimes Na⁺ and Li⁺, appears to play a critical role in the assembly of the symmetric complexes.

2. Results and Discussion

2.1. Solid-State Synthesis

The reaction of BeCl₂ and K[A’] was conducted mechanochemically with a planetary ball mill, followed by an extraction with hexanes. Initial investigations used 2:1 molar ratios of BeCl₂ and K[A’], based on the assumption that the product formed would be BeA’₂ (Equation (1)). Although the reagents are off-white (K[A’]) and white (BeCl₂), the ground reaction mixture (15 min/600 rpm) is orange. Extraction with hexanes, followed by filtration, yielded an orange filtrate and ultimately a dark orange, highly air-sensitive solid (1) on drying.

(1)

A single crystal analysis (described below) revealed that 1 is the potassium tris(allyl)beryllate, K[BeA’₃]. This forms in spite of the fact that the 2:1 ratio of reagents used is not optimum for its production. As detailed below, conducting the reaction with 1:1 and 3:1 molar ratios of K[A’] and BeCl₂ still yields 1 as the sole hexane-extractable product. It is possible that the excess halide is captured in the form of polyhalide anions such as [BeCl₄]²⁻ or [Be₂Cl₆]²⁻ [17], although these have not been definitively identified.

2.2. Synthesis in Solution

The reaction of K[A’] and BeCl₂ was also examined in solution, using diethyl ether and hexanes. These results are summarized in

Table 1. Summary of K[A’] and BeCl₂ reactions; amounts of reagents given as molar ratios.

No.	K[A’]:BeCl2	Medium a	Time	Organoberyllium Product(s)	Yield (%) b
1	1:1		15 min	K[BeA’3]	97
2	2:1		15 min	K[BeA’3]	21
3	3:1		15 min	K[BeA’3]	25
4	1:1	Et2O	1 h	2A’BeCl ⇄ BeA’2 + BeCl	n/a c,d
5	2:1	Et2O	2 h	BeA’2∙OEt2	77 c
6	2:1	Et2O	16 h	K[BeA’3]	98
7	1:1	hexanes	1 h	K[BeA’3]	24

^a = ball milling at 600 rpm. The symbol for mechanical milling has been proposed in ref. [14]; ^b Unrecrystallized; limiting reagent taken into account; ^c Ref. [11]; ^d Products were observed with ⁹Be NMR, and were not isolated.

. Previous reactions with diethyl ether involved stirring for 2 h at room temperature, which formed BeA’₂∙OEt₂ from a 2:1 reaction (#5); Schlenk equilibrium was observed in a 1:1 mixture that was allowed to react for one hour (#4). When the 2:1 reaction in Et₂O is allowed to proceed for 16 h, however, the formation of 1 is observed (#6) exclusively. Reaction in hexanes mimics the solid-state reactions, in that 1 is the exclusively detected organoberyllium product from a 1:1 reaction after 1 h (#7). Longer reactions and a higher ratio of K[A’] to BeCl₂ (e.g., 3:1) do not change this outcome.

2.3. NMR Spectroscopy

The ¹H NMR spectrum of 1 displays resonances typical of a “π-bound” A’ ligand, with a triplet representing H_(β), a broad resonance (ν_1/2 = 39 Hz, presumably an unresolved doublet) representing the equivalent H_(α) and H_(γ), and a singlet for the two equivalent trimethylsilyl groups (Figure 1). The appearance of such a symmetric spectrum even when σ-bound ligands are expected is consistent with a high degree of fluxionality, as was also observed in the σ-bound complex BeA’₂∙(Et₂O) [11]. The triplet resonance of the allyl ligands, at δ6.97, is shifted downfield from that of BeA’₂∙(Et₂O) (δ6.53); the resonance at δ3.19 is slightly upfield (cf. δ3.33 in BeA’₂∙(Et₂O)). The NMR chemical shifts for 1 are in line with those observed for other M[M’A’₃] complexes (

Table 2. ¹H NMR shifts (ppm) and bond distances in M[M’A’₃] complexes.

Complex	δ H(α)/H(γ)	δ H(β)	δ SiMe3	M–C (σ) Å	M’···C(olefin) Å	Ref.
Li[ZnA’3]	6.46	3.50	0.15	2.117(3) b	2.745(4), 2.268(3) b	[15]
Na[ZnA’3]	7.59	4.00	0.16	2.103(3)	2.857(3), 2.567(3)	[15]
K[BeA’3]	6.97	3.19	0.20	1.805(10)	3.153(7), 2.940(7)	this work
K[ZnA’3]	7.05	3.42	0.23	2.068(4)	3.205(3), 2.945(3)	[15]
K(thf)[SnA’3]	6.43	4.42	0.42, 0.23 a	2.344(7)	3.201(7), 3.164(8), 3.065(8)	[16]

^a Two resonances are observed for the SiMe₃ groups, as the A’ ligands are not fluxional; ^b Distance(s) affected by crystallographic disorder.

). In particular, the NMR shifts of the allyl ligands are sensitive both to the identity of the central divalent metal and to that of alkali metal counterion, evidence that the compounds exist as contact ion pairs in solution. Compound 1 and K[ZnA’₃] share the greatest similarities, which may reflect their having the same counterion (K⁺) and central metals of similar electronegativity (χ Be (1.57); Zn (1.65)) [18].

John and co-workers have demonstrated that ⁹Be NMR chemical shift values can be diagnostic for coordination numbers in solution [19]. Typically, organoberyllium complexes with low formal coordination numbers, such as BeMe₂∙Et₂O (coordination number 3, δ20.8 ppm in Et₂O), are observed well downfield of 0 ppm. BeA’₂∙(Et₂O) has a ⁹Be chemical shift of δ18.2 ppm, which is consistent with a three-coordinate geometry in solution [11]. It should be noted, however, that the correlation between coordination number and ⁹Be chemical shift is not exact, and can be strongly influenced by the electronic properties of the ligands. The 2-coordinate complex beryllium bis(N,N´-bis(2,6-diisopropylphenyl)-1,3,2-diazaborolyl), for example, has an extreme downfield shift of δ44 ppm [20], whereas the 2-ccordinate Be(N(SiMe₃)₂ displays a ⁹Be NMR shift at δ12.3 ppm [4]. Nevertheless, the ⁹Be of 1 is at δ22.8 ppm, which to our knowledge is the most positive shift yet reported for a three-coordinate species. DFT methods were used to predict the ⁹Be chemical shift value of 1 (B3LYP-D3/6-311+G(2d,p)//B3LYP-D3/6-31G(d)). It was calculated at δ25.9 ppm, in reasonable agreement with the observed value (referenced to [Be(OH₂)₄]²⁺ with an isotropic shielding constant of 108.98 ppm).

2.4. Solid State Structure

The structure of 1 was determined from single crystal X-ray diffraction. In the solid state, 1 exhibits approximate C₃-symmetry, with σ-bound A’ ligands and a potassium cation engaging in cation–π interactions with the three double bonds of the allyls. It is isostructural with the previously reported M[ZnA’₃] (M = Li, Na, K) and K(thf)[SnA’₃] complexes [15,16]. The beryllium center is in a nearly planar trigonal environment (sum of C–Be–C´ angles = 357.7°) (Figure 2).

The average Be–C distance of 1.805(10) Å has few direct points of comparison with other molecules, as 1 is only the second crystallographically characterized [BeR₃]⁻ complex, the other being lithium tri-tert-butylberyllate [21]. The latter’s Be center, like that in 1, is in a nearly perfectly planar trigonal environment (sum of C–Be–C angles = 359.9°). In the solid state, however, tri-tert-butylberyllate is a dimer, [Li{Be(t-C₄H₉)₃}]₂, with some corresponding distortions in the Be–C bond lengths; Be–C distances range from 1.812(4) Å to 1.864(4) Å, averaging to 1.843(6) Å. The Be–C length in 1 is indistinguishable from the Be–C_carbene length of 1.807(4) Å in the [Ph₂Be(IPr)] (IPr = 1,3-bis(2,6-di-isopropylphenyl)-imidazol-2-ylidene) complex, which also has a three-coordinate Be center [22]. The anionic methyl groups in [Ph₂Be(IPr)] are at a noticeably shorter distance, however (1.751(6) Å, ave.). A similar relationship between the Be–C_carbene and Be–CH₃ bond lengths exists in the related [Me₂Be(IPr)] [23] and [Me₂Be(IMes)] (IMes = N,N’-bis(2,4,6-trimethylphenyl)imidazol-1-ylidene) complexes [1]. A comparison of the Be–C length in 1 could also be made with the Be–C distance of 1.84 Å in lithium tetramethylberyllate, Li₂[BeMe₄], although the bond distance would be expected to be slightly longer in the latter owing to the higher coordination number of beryllium and the greater negative charge [24].

The C–C and C=C bonds in the alkyl groups in 1 are localized at 1.475(5) Å and 1.353(9) Å, respectively. The K⁺···C(olefin) contacts average 3.153(7) Å and 2.940(7) Å to the carbon atoms β (C2, C11, and C20) and γ (C3, C12, and C21) to the beryllium atom, respectively. These distances are comparable to, but slightly shorter than, the range of K⁺···C contacts found in the related zincate structure (3.205(3) Å and 2.945(3) Å, respectively), which reflects the shorter M–C_(α) bonds in 1. The distance between Be and K (3.59 Å) is long enough to rule out significant metal-metal interactions.

2.5. Computational Investigations

It has previously been suggested that the occurrence of C₃-symmetric M[M’A’₃L] (M’ = Zn, M = Li, Na, K; M’ = Sn, M = K; L = thf) complexes is the result of a templating effect of the associated alkali metal counterion [25]. The rationale for this proposal is that the neutral MA’₃ (M = As, Sb, Bi) complexes always occur in two diastereomeric forms, with R,R,R (equivalently, S,S,S) and R,R,S (or S,S,R) arrangements of the allyl ligands around the central element. The anionic [MA₃´]⁻ complexes, in contrast, are always found in the C₃-symmetric R,R,R (or S,S,S) configuration, and it is not unreasonable to assume that the counterion is responsible for the difference.

A DFT investigation was undertaken to explore the possible origins of this effect. The geometry of the free [BeA’₃]⁻ anion was optimized with calculations employing the dispersion-corrected APF-D functional [26]. Three confirmations were examined: the C₃-symmetric form (S,S,S) found in the X-ray crystal structure of 1, a related S,S,S form with one A’ ligand rotated antiparallel to the other two (C₁ symmetry), and a R,R,S form, also with one ligand antiparallel to the other two, derived from the structure of the neutral AlA’₃ complex (Figure 3) [27].

Not surprisingly, the calculated structures possess similar average Be–C bond lengths, ranging from 1.782 Å (the C₃-symmetric form (Figure 3a)) to 1.788 Å (for the rotated S,S,S form (Figure 3b)). Energetically, the R,R,S form is the most stable; the rotated S,S,S form is 10.1 kJ·mol⁻¹ higher in energy (∆G°), and the C₃-symmetric form is higher still (20.3 kJ·mol⁻¹ in ∆G°). The origin of these energy differences is not immediately obvious, but it may be related to the relative amounts of interligand congestion present. The low energy R,R,S form, for example, has no Me···Me´ contacts less than 4.0 Å, the sum of the van der Waals radii [18]. In contrast, the C₃ symmetric form has multiple contacts between methyl groups of less than 4.0 Å, including two as short as 3.76 Å. At this level of theory, the energetics of the free anions do not provide a rationale for the exclusive formation of the S,S,S form.

Not surprisingly, incorporation of the K⁺ ion into the complex alters the relative stability of the species. The optimized geometries of the C₃-symmetric K[BeA’₃] found in the X-ray crystal structure of 1 and a related S,S,R form were calculated similarly to the isolated anions, and are depicted in Figure 4.

The C₃-symmetric form is 6.1 kJ·mol⁻¹ more stable than the S,S,R form. This is not a consequence of closer K⁺···(C=C) distances, which are nearly the same (avg. 3.91 Å in the C₃ form; 2.86 Å in the S,S,R arrangement). The asymmetric arrangement of the ligands in the S,S,R form does lead to closer interligand C···C contacts in the allyl frameworks, however, as small as 3.37 Å, whereas there are no similar contacts less than 3.78 Å in the C₃ form. The somewhat greater stability of the C₃ form, possibly coupled with greater ease of crystal packing, may contribute to the exclusive appearance of that form in the crystal structure. It is likely that a similar analysis holds for the isostructural Zn and Sn complexes.

The failure to produce an unsolvated BeA’₂ in the absence of a coordinating solvent (i.e., either mechanochemically or in hexanes) was also examined computationally with the aid of the Solid-G program [28]. Both BeA’₂∙Et₂O and 1 are found to have coordination sphere coverage (G_complex) above 90% (i.e., 97.0% (Figure 5a) and 92.6% (Figure 5b), respectively). Although the coverage of the metal center in the hypothetical BeA’₂ varies somewhat with the angle between the ligands, the minimum energy position depicted in Figure 5c (C₂ symmetry) has only 78.7% coverage. It is not unreasonable to assume that a monomeric BeA’₂ may be too coordinately unsaturated to be readily isolable, and will bind an ethereal solvent molecule during synthesis, or, if that is not available, an additional A’ ligand, counterbalanced with a K⁺ ion.

3. Materials and Methods

General Considerations: All syntheses were conducted under rigorous exclusion of air and moisture using Schlenk line and glovebox techniques. (NOTE: Beryllium salts are toxic and should be handled with appropriate protective equipment.) After grinding was completed, the jars were opened according to glovebox procedures to protect the compounds and to prevent exposure to dust [10]. Proton (¹H) and carbon (¹³C) spectra were obtained on Bruker DRX-500 or DRX-400 spectrometers (Karlsruhe, Germany), and were referenced to residual resonances of C₆D₆. Beryllium (⁹Be) spectra were obtained on a Bruker DRX-500 at 70.2 MHz, and were referenced to BeSO₄(aq). Combustion analysis was performed by ALS Environmental, Tucson, AZ, USA. Beryllium chloride was purchased from Strem, stored under an N₂ atmosphere and used as received. The K[A’] (A’ = 1,3-(SiMe₃)₂C₃H₃) reagent was synthesized as previously described [29,30]. Toluene, hexanes, and diethyl ether were distilled under nitrogen from potassium benzophenone ketyl [31]. Deuterated benzene (C₆D₆) was distilled from Na/K (22/78) alloy prior to use. Stainless steel (440 grade) ball bearings (6 mm,) were thoroughly cleaned with hexanes and acetone prior to use. Planetary milling was performed with a Retsch model PM100 mill (Haan, Germany), 50 mL stainless steel grinding jar type C, and safety clamp for air-sensitive grinding.

3.1. Mechanochemical Synthesis of K[BeA’₃] (1)

Solid BeCl₂ (56.7 mg, 0.71 mmol) and K[A’] (319 mg, 1.42 mmol) were added to a 50 mL stainless steel grinding jar (type C). The jar was charged with stainless steel ball bearings (6 mm dia, 50 count) and closed tightly with the appropriate safety closer device under an N₂ atmosphere. The reagents were milled for 15 min at 600 rpm, resulting in a light orange solid. The product was extracted under an inert atmosphere with minimal hexanes (<100 mL) and filtered through a medium porosity ground glass frit, providing a dark orange filtrate. Drying under vacuum yielded a dark orange solid (61.5 mg, 21% yield of K[BeA’₃]) which was recrystallized by the slow evaporation of toluene over one month to provide dark orange-brown crystals of 1 suitable for single crystal X-ray diffraction. For a 3:1 K[A’]:BeCl₂ reaction, 812 mg (3.62 mmol) K[A’] and 95.2 mg (1.19 mmol) BeCl₂ were added to a grinding jar. After extraction, 183 mg (25% yield) of orange solid was collected. Anal. Calcd. (%) for C₂₇H₆₃BeKSi₆: C, 53.65; H, 10.51; Be, 1.49. Found: C, 52.09; H, 9.79; Be, 1.04. The values are somewhat low, possibly from the high air-sensitivity of the compound, but the C:H molar ratio is 2.34:1.00, close to the expected 2.33:1.00. ¹H NMR (500 MHz, C₆D₆, 298K): δ 0.20 (s, 54H, SiMe₃); 3.19 (br s (ν_1/2 = 39 Hz), 6H, H_(α,γ)); 6.97 (t, 3H, J₁ = 16 Hz, H_(β)). ¹³C NMR (100 MHz, C₆D₆, 298K): δ1.02 (s, SiMe₃); 70.71 (s, C_(α,γ)); 166.09 (s, C_(β)). ⁹Be NMR (70.2 MHz, C₆D₆, 298K); δ22.8 (s) (ν_1/2 = 360 Hz).

3.2. General Procedures for Synthesis of K[BeA’₃] (1) with Solvents

Reactions were performed for either 1 or 16 h, and were run under inert atmosphere at room temperature. The ratio of K[A’] and BeCl₂ was varied such that the reactions of emphasis were 1:1, 2:1, and 3:1. A general reaction involved dissolving the beryllium chloride (ca. 0.1 g) in the solvent of choice (Et₂O or hexanes); to this solution solid K[A’] was added slowly and solvent was used to quantitatively transfer all material. Upon mixing, the solution was allowed to stir for the given time. In the case of Et₂O, the solvent was removed in vacuo, and the resulting material was extracted with hexanes, filtered through a medium porosity glass frit, and then dried in vacuo. In the case of reaction in hexanes, the reaction mixture was filtered through a medium porosity fritted glass filter, and the hexane was removed in vacuo. The resulting material in all cases was then analyzed with ¹H and ⁹Be NMR.

3.3. Procedures for X-ray Crystallography

A crystal (0.20 × 0.20 × 0.08 mm³) was placed onto the tip of a thin glass optical fiber and mounted on a Bruker SMART APEX II CCD platform diffractometer (Karlsruhe, Germany) for a data collection at 100.0(5) K [32]. The structure was solved using SIR2011 [33] and refined using SHELXL-2014/7 [34]. The space group P1 was determined based on intensity statistics. A direct-methods solution was calculated that provided most non-hydrogen atoms from the E-map. Full-matrix least squares/difference Fourier cycles were performed which located the remaining non-hydrogen atoms. All non-hydrogen atoms were refined with anisotropic displacement parameters. The allylic hydrogen atoms were found from the difference Fourier map and refined freely. All other hydrogen atoms were placed in ideal positions and refined as riding atoms with relative isotropic displacement parameters.

3.4. General Procedures for Calculations

All calculations were performed with the Gaussian 09W suite of programs [35]; an ultrafine grid was used for all cases (Gaussian keyword: int = ultrafine). Each conformation of the [BeA’₃]⁻ complexes was studied with the APF-D functional, a global hybrid with 23% exact exchange [26]. The 6-31+G(d) basis set was used for C,H,Si; the 6-311+G(2d) basis was used for Be. For the neutral K[BeA’₃] conformations, the APF-D functional was used with the 6-31G(d) basis set for C,H,Si; 6-311G(2d) was used for Be and K. The nature of the stationary points was determined with analytical frequency calculations; all of these optimized geometries were found to be minima (N_imag = 0). For the Solid-G calculations, the structures were preoptimized with the APF-D/6-311G(2d) (Be,K); 6-31G(d) (C,H,Si) protocol.

4. Conclusions

The generation of products from reagents that are not in the optimum stoichiometric ratio is a known feature of some Group 2 reactions [36,37], a testament to the role that kinetic factors play in s-block chemistry. It is perhaps not surprising that when mechanochemical activation is used with alkaline earth reagents, a nonstoichiometric product such as the organoberyllate 1 is formed, as grinding and milling environments are often far from equilibrium [38,39,40,41]. However, the fact that 1 is also generated in hexanes indicates how non-ethereal synthesis can reveal features of reactions that are obscured when they are conducted in coordinating solvents. It is now apparent that the production of the previously described BeA’₂∙Et₂O, which was the expected complex from a 2:1 reaction of K[A’] and BeCl₂ in diethyl ether [11], actually depends critically on the presence of the solvent to prevent further reaction of the beryllium center with an additional A’ ligand. (In a preliminary study, the reaction of K[A′] and BeCl₂ in a 2:1 molar ratio in THF (1 h) was found not to produce K[BeA′₃]. A species with a ⁹Be NMR shift of δ16.6 ppm was present instead, tentatively identified as BeA′₂(thf). If correct, this indicates that THF, like Et₂O, can block the formation of the tris(allyl) anion with Be). Without such ethereal solvent support, whether conducted mechanochemically or in hexanes, the reaction between K[A’] and BeCl₂ rapidly forms the kinetic product 1.

Parallels of the beryllium chemistry to the related tris(allyl) -ate complexes of Zn and Sn are instructive, although they cannot be pushed too far. All the [MA₃´]⁻ species possess approximate C₃ symmetry, and it is likely that the associated alkali metal cation is intimately involved in templating their constructions. The formation of the zinc species K[ZnA’₃] is also similar to that of 1 in that it is formed from the reaction of 2 equiv. of K[A’] and ZnCl₂, i.e., in a non-stoichiometric reaction [15]. However, both it and K(thf)[SnA’₃] are synthesized in THF, so it is clear that the driving force for -ate formation compared to that for the neutral (Zn,Sn)A’₂ species is greater than that for 1. This may reflect the somewhat lesser covalency of Be–C versus Zn–C and Sn–C bonds, and the greater robustness of M²⁺⟵:OR₂ interactions with beryllium.