Formation of a hydride containing amido-zincate using pinacolborane

Amido-zincates containing hydrides are underexplored yet potentially useful complexes. Attempts to access this type of zincate through combining amido-organo zincates and pinacolborane (HBPin) via Zn–C/H–BPin exchange led instead to preferential formation of amide–BPin and/or [amide–BPin(Y)]− (Y = Ph, amide, H), when the amide is hexamethyldisilazide or 2,2,6,6-tetramethylpiperidide and the hydrocarbyl group was phenyl or ethyl. In contrast, the use of a dipyridylamide (dpa) based arylzinc complex led to Zn–C/H–BPin metathesis being the major outcome. Independent synthesis and full characterisation of two LnLi[(dpa)ZnPh2] (L = THF, n = 3; L = PMDETA, n = 1) complexes, 1 and 3, respectively, enabled reactivity studies that demonstrated that these species display zincate type reactivity (by comparison to the lower reactivity of the neutral complex (Me-dpa)ZnPh2, 4, Me-dpa = 2,2′-dipyridyl-N-methylamine). This included 1 performing the rapid deprotonation of 4-ethynyltoluene and also phenyl transfer to α,α,α-trifluoroacetophenone in contrast to neutral complex 4. Complex 1 reacted with one equivalent of HBPin to give predominantly PhBPin (ca. 90%) and a lithium amidophenylzincate containing a hydride unit, complex 7-A, as the major zinc containing product. Complex 7-A transfers hydride to an electrophile preferentially over phenyl, indicating it reacts as a hydridozincate. Attempts to react 1 with >1 equivalent of HBPin or with catecholborane led to more complex outcomes, which included significant borane and dpaZn substituent scrambling, two examples of which were crystallographically characterised. While this work provides proof of principle for Zn–C/H–BPin exchange as a route to form an amido-zincate containing a hydride, amido-organozincates that undergo more selective Zn–C/H–BPin exchange still are required.


General considerations
Unless otherwise indicated all manipulations were conducted under inert conditions either using standard Schlenk techniques or in a MBraun UniLab glovebox (< 0.1 ppm H2O / O2). D6-benzene and d8-THF were dried over CaH2 and Na respectively, and distilled prior to storage over 3 Å molecular sieves. Toluene was dried using an Innovative Technology SPS system and stored over activated molecular sieves. LiTMP, 1 KTMP 2 and N-methyl-2,2'-dipyridylamine 3 were prepared according to literature procedures. L iHM DS was prepared by deprotonation of hexamethyldisilazane (HMDS(H)), and KHMDS was prepared by metathesis from LiHMDS and KOtBu in hexanes. Unless otherwise stated all other compounds were purchased from commercial sources and used as received. NMR spectra were recorded on Bruker Avance III 400 MHz and 500 MHz spectrometers. Chemical shifts are reported as dimensionless δ values and are frequency referenced relative to residual protio-impurities in the NMR solvents for 1 H and 13 C{ 1 H} respectively, while 11 B, 19 F and 7 Li shifts are referenced relative to external BF3·Et2O, hexafluorobenzene, and LiCl in D2O, respectively. Coupling constants J are given in Hertz (Hz) as positive values regardless of their real individual signs. The multiplicity of the signals are indicated as "s", "d", "t" "q" "pent", "sept" or "m" for singlet, doublet, triplet, quartet, pentet, septet or multiplet, respectively. 1 H-7 Li HOESY spectra were recorded using a Bruker PRO 500 spectrometer equipped with RT BB probe. 48 scans were recorded with mixing time of 150 ms.
Electrospray ionization (ESI) measurements were performed at the Scottish Instrumentation and Resource Centre for Advanced Mass Spectrometry (SIRCAMS) based in the School of Chemistry at the University of Edinburgh. High resolution mass spectra were recorded on a Bruker Daltonics 12T SolariX Fourier Transform Ion Cyclotron Resonance Mass Spectrometer (FT-ICR-MS). Elemental analyses were carried out by Elemental Analysis Ltd., measured in duplicate.

[(dpaZnPh)3] (2)
Diphenylzinc (0.11 g, 0.5 mmol, 1 eq) was loaded into a Schlenk tube and dissolved in toluene (5 mL). The solution was added to another Schlenk tube containing neat 2,2'-dipyridylamine (85 mg, 0.5 mmol, 1 eq) and stirred for one hour affording an off-white suspension. The product was isolated by filtration (240 mg, 77%). Crystals suitable for analysis by X-ray diffraction were grown from a concentrated C6D6 solution of 2 standing at ambient temperature over a period of several days. Anal. Calc.

[(PMDETA)Li(µ-dpa)ZnPh2] (3)
A solution of diphenylzinc (0.11 g, 0.5 mmol, 1 eq) in toluene (3 ml) was added to neat 2,2'-dipyridylamine (85 mg, 0.5 mmol, 1 eq) and stirred for 15 min at ambient temperature. In another Schlenk tube, PhLi (47 mg, 0.5 mmol, 1eq) was suspended in 5 mL toluene and PMDETA (0.11 mL, 0.5 mmol, 1 eq) was added affording a yellow solution which was then added to toluene solution of 2 and stirred for another 2 h at room temperature affording an off-white suspension. The product was isolated by filtration (184 mg, 65 %). Crystals suitable for analysis by X-ray diffraction were grown by slow mixing of toluene solutions of the separate organometallic reagents by diffusion without stirring at room temperature. Anal. Calc.

[(Me-dpa)·ZnPh2] (4)
Diphenylzinc (0.11 g, 0.5 mmol, 1 eq) was loaded into a Schlenk tube and dissolved in toluene (3 mL). The solution was added to another Schlenk tube containing neat 2,2'-dipyridyl-N-methylamine (0.5 mmol, 94 mg, 1 eq) and stirred for 1 hour. Complex 4 was formed as a white precipitate, observable after 15 min of stirring. The solvent was exchanged in vacuo for 1,2-difluorobenzene (5 mL) and heated to reflux affording a clear solution which upon slow cooling to room temperature deposited crystals of 1 (121 mg, 60 %). Accurate elemental analysis was not obtained despite fully homogeneous samples being very clean by multinuclear NMR spectroscopy.

1 H DOSY NMR comparison of 2 and 4
1 H DOSY NMR experiments were conducted in C6D6 using the External Calibration Curve (ECC) method at 15µM as described by Stalke. 4 Data was accumulated by linearly varying the diffusion encoding gradients over a range of 2% to 95% for 16 gradient increment values. The signal decay dimension on the pseudo-2D data was generated by Fourier transformation of the time-domain data. The diffusion profile and coefficients were established by use of the DOSY processing features of TopSpin software. Using residual proton signal of solvent as a reference standard we have been able to approximate the aggregates of complexes 2 and 4 as compact spheres (CS) and determine their mean diffusion coefficients (Table S1). H DOSY NMR spectrum of 4 revealed co-diffusion of the phenyl and dipyridylamine peaks, suggesting that the solid-state structure of 4 is retained in solution. A mean diffusion coefficient of D = 9.327 x 10 -10 m 2 /s was measured ( Figure S39) and determined molecular weight of 428 g/mol correlates well with calculated molecular weight of monomeric complex 4 of 406 g/mol with error of -5%. A mean diffusion coefficient of D = 5.992 x 10 -10 m 2 /s was measured for complex 2 ( Figure S38) and determined molecular weight of 709 g/mol. This value does not correlate well with monomeric constitution (MWcalc = 313 g/mol), nor trimeric constitution (MWcalc = 938 g/mol, error = 32 %) found in the solid state. The closest correlation would be with dimeric constitution (MWcalc = 625 g/mol, error = -12 %) or with an equilibrium state between dimeric and trimeric constitution (MWcalc = 782 g/mol, error = 10 %). Considering that the ECC method employed is valid for MW up to 600 g/mol, the exact structure of 2 in C6D6 solution is not clear but it is a strong indication that 2 exists as oligomer(s) in solution.

Conversion of 1 to 3 by addition of PMDETA in THF solution
In order to assess the propensity of PMDETA to bind to the Li-centre of [{(THF)3Li(µ-dpa)ZnPh2}] scaffold by displacing THF, 1 (30 mg, 4.89 x 10 -5 mol, 1 eq) was loaded into a J Young's NMR tube and dissolved in d8-THF (0.5 mL). To this solution, PMDETA (10.2 µL, 4.89 x 10 -5 mol, 1 eq) was added and the reaction mixture was stirred for 2 h. A change was firstly evident in observed precipitation, indicative of formation of 3 which was previously observed to exhibit lower solubility than 1 (in THF, while 3 is insoluble in C6D6). The mixture was analysed by 1 H and 7 Li NMR spectroscopies ( Figure S40-S42) which showed that the starting ate 1 has changed. Furthermore, comparison of obtained spectra with spectra of pure sample of 3 is in excellent agreement.

Reactivity studies 4.1 Deprotonation of terminal alkyne and isolation of 5
Lithium zincate 1 (30 mg, 4.89 x 10 -5 mol, 1 eq) was loaded into a J Young's NMR tube and dissolved in d8-THF (0.5 mL). 4-ethynyltoluene (12.4 µL, 9.79 x 10 -5 mol, 2 eq) was added and after 2 h stirring at room temperature, the reaction mixture was analysed by 1 H and 7 Li NMR spectroscopies ( Figure S43-S44) which showed formation of 5 as the major species and minor amounts of two more species containing alkynyl substituents. A sharp signal for benzene forming as a side product is also detected in the 1 H NMR spectrum. The product crashed out of the solution upon standing at room temperature and a crystal suitable for X-ray measurement was obtained. Poor solubility of crystalline 5 precluded subsequent recording of a meaningful 13 C NMR spectrum. In order to compare with other available zinc species, deprotonation of two equivalents of 4ethynyltoluene with lithium zincate 3 (PMDETA), neutral Ph2Zn and 4 was probed. Like 1, 3 exhibits enhanced basicity and after 2 h at room temperature the deprotonation is near complete ( Figure S45). There is only trace amount of alkyne left, but due to the poor solubility of the product, the exact amount cannot be determined by integration.
Both neutral organozinc species, Ph2Zn and 4, showed sluggish deprotonation. For instance, alkyne deprotonation with 4 was not completed even after a week at room temperature ( Figure S46). Relative ratio, based on the integration of the CH3 resonance of 4-ethynyltoluene, shows about 80% of Znalkynyl species and 20% leftover unreacted alkyne even after this time.
Similarly, with Ph2Zn only extensive heating pushed the deprotonation to completion ( Figure S47). Nevertheless, we attempted this as an alternative route to obtain clean product 5. Once diphenylzinc (22 mg, 0.1 mmol, 1 eq) completed deprotonation of 4-ethynyltoluene (25.4 µL, 0.2 mmol, 2 eq) in d8-THF, Lidpa (17.7 mg, 0.1 mmol, 1 eq) was added and immediately white solid started crashing out. The product was isolated by filtration and analysed by 1 H and 7 Li NMR spectroscopy which showed formation of 5 as major species and minor amounts of two more species containing alkynyl substituent.     4.2 Arylation of α,α,α-trifluoroacetophenone and synthesis of 6 Compound 1 (100 mg, 1.63 x 10 -4 mol, 1 eq) was placed in a Schlenk tube and dissolved in dry benzene (2 mL). To this solution, α,α,α-trifluoroacetophenone (23 µL, 1.63 x 10 -4 mol, 1 eq) was added and after 2 h stirring at room temperature, the reaction mixture was quenched with HCl (1M, 10 mL) and left stirring overnight. The colourless solution was diluted with water and extracted with diethyl ether (3 x 3 mL). Combined organic layers were dried over Na2SO4 and volatiles removed under vacuum. The obtained crude was dissolved in CDCl3 ( Figure S50-S51) to which CH2Br2 (11.4 µL, 1.63x10 -4 mol, 1 eq) was added to determine NMR yield (71% In order to gain understanding on organometallic species formed pre acid quench, the reaction was repeated in a J Young's NMR tube. Lithium zincate 1 (30 mg, 4.89 x 10 -5 mol, 1 eq) was loaded into a J Young's NMR tube and dissolved in C6D6 (0.5 mL). α,α,α-trifluoroacetophenone (7 µL, 4.89 x 10 -5 mol, 1 eq) was added and after 2 h stirring at room temperature, the reaction mixture was analysed by 1 H and 19 F NMR spectroscopy ( Figure S48-S49) which showed formation of new species assigned as Li-OC(CF3)Ph2 and formation of complex 2.
In order to compare the reactivity of 1, with other prepared zinc species, complexes 2-4 were reacted with α,α,α-trifluoroacetophenone in C6D6 at room temperature. Unlike in the deprotonation of alkyne, lithium zincate 3 does not display the same reactivity as lithium zincate 1 in this case. Lithium zincate 3 (30 mg, 5.26 x 10 -5 mol, 1 eq) was loaded into a J Young's NMR tube and suspended in C6D6 (0.5 mL). α,α,α-trifluoroacetophenone (14.8 µL, 1.0 x 10 -4 mol, 2 eq) was added and after 2 h stirring at room temperature, the reaction mixture was analysed by 1 H and 19 F NMR spectroscopies ( Figure S54-S55) which showed only starting material present as a fluorine-containing species.
The reaction between neutral zinc reagent 2 and one equivalent of α,α,α-trifluoroacetophenone afforded no reaction even after stirring for 5 days at room temperature as evidenced by 1 H and 19 F NMR spectroscopies ( Figure S56-S57). The same outcome was evident if neutral monomeric zinc reagent 4 was reacted with two equivalents of the substrate ( Figure S58-S59). It should be noted that the seemingly incorrect stoichiometry in this final reaction is caused by poor solubility of 4 in C6D6. Figure S48. 1 H NMR spectrum of reaction mixture of 1 with 1 eq of α,α,α-trifluoroacetophenone in C6D6 at room temperature.

Reactivity of 1 towards HBPin
A J Young's NMR tube was charged with 1 (30 mg, 4.89 x 10 -5 mol, 1 eq) followed by the addition of C6D6 (0.5 mL) and HBPin (7.1 μl, 4.89 x 10 -5 mol, 1 eq). The sealed tube was spun at room temperature and checked by NMR spectroscopy after 15 min ( Figure S60-S62). The same reaction was performed in d8-toluene by adding HBPin at -40 °C ( Figure S64-S65) and immediately recording the NMR spectra. Analysis of the data shows that in both instances the outcome is the same -no free HBPin is observed and it is transformed (mostly) into PhBPin, along with the formation of two Zn-H containing species (7A and 7B) in approximately 2:1 ratio. Repeating the NMR analysis after 1 h or overnight revealed no visible change. 1 H DOSY NMR analysis showed that 7A and 7B have similar diffusion coefficients ( Figure  S63) confirming that they are both molecular zinc hydrides containing dpaZn units.
The reaction also was probed in THF as solvent, this gave significantly more complex outcomes. For example -a J Young's NMR tube was charged with 1 (30 mg, 4.89 x 10 -5 mol, 1 eq) followed by the addition of THF (0.5 mL) and HBPin (7.1 μl, 4.89 x 10 -5 mol, 1 eq or 14.2 μl, 9.79 x 10 -5 mol, 2 eq). The sealed tube was spun at room temperature and checked by NMR spectroscopy after 1 h. Analysis of the data showed that no free HBpin is observed but only one resonance for Zn-H is observed. The formation of PhBPin is again dominant, but more ligand redistribution occurs ( Figure S66-S67) than in aromatic solvents (i.e. benzene or toluene) and this is further increased in case of 2 eq. HBPin ( Figure  S68-S70). Repeating the reaction in THF with 2 eq of DBPin affords spectra ( Figure S71-S74) in agreement with the ones obtained when using HBPin but with a difference in the absence of any resonances in the 4.5 -6.5 ppm region in the 1  Going back to arenes as solvent, the reaction was repeated with excess HBPin. A J Young's NMR tube was charged with 1 (30 mg, 4.89 x 10 -5 mol, 1 eq) followed by the addition of C6D6 (0.5 mL), and hexamethylbenzene (8 mg, 4.89 x 10 -5 mol, 1 eq) as internal standard. The reaction mixture was checked by 1 H NMR spectroscopy showing that clean starting material 1 and internal standard are present in a 1:1 ratio. Then, HBPin (14.2 μl, 9.78 x 10 -5 mol, 2 eq) was added and immediately (5 min) analysed by 1 H and 11 B NMR spectroscopy, followed by further analysis after 15 and 30 min ( Figure  S77) which showed that the reaction is complete after 5 min and no change is observed within next hour. Further analysis revealed major products to be PhBPin and Zn-H species 7B, both present in 70% against internal standard ( Figure S75-S76). Multinuclear NMR spectroscopy (including 2D COSY, HSQC, HMBC and 1 H DOSY experiments) imply that a dpa-ZnPh(H) species is formed ( Figure S78-S84).
If the excess of HBPin is increased further, to 3 eq, the same outcome is evident -two equivalents of HBPin are consumed and an equivalent is left unreacted, as evidenced in both 1 H and 11 B NMR spectra ( Figure S85-S86). The reaction is again fast and no change is observed on standing overnight ( Figure  S87-S88).
Attempts to observe and more accurately identify the two Zn-H species observed in arene solvents by mass spectrometry were unsuccessful. The key identifiable fragment from multiple MS experiments was: (MS-EI + ) found: 311.04208 [C16H13N3Zn = dpaZnPh], simulated 311.03954

Reactivity of 1 towards HBCat
A J Young's NMR tube was charged with 1 (30 mg, 4.89 x 10 -5 mol, 1 eq) followed by the addition of 0.5 mL of C6D6 and HBcat (5.2 μl, 4.89 x 10 -5 mol, 1 eq). The sealed tube was spun at room temperature and after an hour analysed by 1

Reactivity of 1 towards silane
A J Young's NMR tube was charged with 1 (34 mg, 5.55 x 10 -5 mol, 1 eq) followed by the addition of 0.5 mL of C6D6 and Ph3SiOH (15.3 mg, 5.55 x 10 -5 mol, 1 eq). The sealed tube was spun at room temperature and analysed by 1 H and 7 Li NMR spectroscopy after one hour revealing complete consumption of starting materials and formation of benzene ( Figure S91-S92). To this in situ formed alkoxide species, PhSiH3 (7 µL, 5.55 x 10 -5 mol, 1 eq) was added and analysed by 1 H and 7 Li NMR spectroscopy ( Figure S93-S94) after one hour and 3 days later. In both cases only traces of Zn-H species are evident and majority of silane is unreacted. With prolonged reaction times (i.e. 3 days) only increased degradation is evident, rather than formation of desired product. Figure S91. 1 H NMR spectrum of reaction of 1 with 1 eq of Ph3SiOH in d8-THF at room temperature. Figure S92. 7 Li NMR spectrum of reaction of 1 with 1 eq of Ph3SiOH in d8-THF at room temperature. Figure S93. Stacked 1 H NMR spectra of reaction of 1 with 1 eq of Ph3SiOH, followed by reaction with 1 eq of PhSiH3 in d8-THF at room temperature. Figure S94. Stacked 7 Li NMR spectra of reaction of 1 with 1 eq of Ph3SiOH, followed by reaction with 1 eq of PhSiH3 in d8-THF at room temperature.

Reactivity of 3 towards HBPin
A J Young's NMR tube was charged with 3 (30 mg, 5.26 x 10 -5 mol, 1 eq) followed by the addition of 0.5 mL of C6D6 and HBpin (15.3 μl, 1.05 x 10 -5 mol, 2 eq). The sealed tube was spun at room temperature and analysed by 1 H and 11 B NMR spectroscopy ( Figure S95-S96) after 90 min revealing that majority of HBPin had not reacted. The reaction mixture was left over night to react (18 h) and again analysed. Both 1 H and 11 B NMR spectra now show almost complete consumption of HBPin ( Figure S97-S98) and formation of the same Zn-H species 7B as when lithium zincate 1 is used, albeit it is formed at a much slower rate ( Figure S99-S100).
Control reaction between PMDETA and HBPin was finally performed by mixing the two reagents in C6D6 in a sealed J Young's NMR tube. 1 H and 11 B NMR spectroscopy ( Figure S101-S102) showed no interaction between these reagents.

Reactivity of 2 towards HBPin
A J Young's NMR tube was charged with 2 (30 mg, 9.64 x 10 -5 mol, 1 eq) followed by the addition of 0.5 mL of C6D6 and HBpin (14 μl, 9.64 x 10 -5 mol, 1 eq). The sealed tube was monitored over days followed by prolonged heating (Figure S103-S104) showing consumption of HBPin and formation of PhBPin. Repeating the reaction and heating the sealed tube at 80 °C for 5 h ( Figure S105-S106) afforded the same outcome with evident decomposition of formed Zn-H containing species.

Reactivity of hydride species towards ketones
A J Young's NMR tube was charged with 1 (30 mg, 4.89 x 10 -5 mol, 1 eq) followed by the addition of 0.6 mL of C6D6 and HBpin (7.1 μl, 4.89 x 10 -5 mol, 1 eq). As soon as the reagents were added the mixture was analysed by 1 H and 11 B NMR spectroscopy showing the predominant formation of Zn-H species assigned as complex 7A and PhBPin. To this mixture, α,α,α-trifluoroacetophenone (6.9 μl, 4.89 x 10 -5 mol, 1 eq) was added and the reaction mixture was spun at room temperature for 2 h. Multinuclear NMR analysis ( Figure S110-S114) showed consumption of Zn-H species 7A and formation of predominantly M-OC(Ph)(CF3)H, and small amounts of M-OCPh2(CF3) and hydroborated product PinBOC(Ph)(CF3)H. 6 The formed Zn-containing species has resonances identical to complex 2.
The potential of 7A to transfer hydride to several other ketones was tested -specifically 4fluorobenzophenone, benzophenone and acetophenone. Compound 1 (100 mg, 1.63 x 10 -4 mol, 1 eq) was placed in a Schlenk tube and dissolved in dry benzene (1 mL) and HBPin (23.7 µL, 1.63 x 10 -4 mol, 1 eq) was added. After 15 min, the selected ketone (1 eq) was added and after 2 h stirring at room temperature, the reaction mixture was quenched with HCl (1M, 2 mL) and left stirring for 3 h. The colourless solution then was diluted with water and extracted with ethyl acetate (3 x 2 mL). Combined organic layers were dried over Na2SO4 and the solvent was evaporated. Obtained crude was dissolved in CDCl3 to which dibromomethane (11.4 µL, 1.63x10 -4 mol, 1 eq) was added to determine NMR yield.
Control reactions showed that HBPin on its own hydroborates α,α,α-trifluoroacetophenone only in trace amounts even after 3 days ( Figure S126-S128), and the complex Lidpa·HBPin is unreactive ( Figure  S129-S131).    Figure S114. 19 F NMR spectrum of reaction mixture of 1 and HBPin (1:1) in C6D6 and α,α,α-trifluoroacetophenone at room temperature. Figure S115. 1 H NMR spectrum of hydrolysed crude reaction mixture in CDCl3 and with 1,3,5-trifluorobenzene (1 eq) as internal standard at room temperature to determine yield. Figure S116. 19 F NMR spectrum of hydrolysed crude reaction mixture in CDCl3 and with 1,3,5-trifluorobenzene (1 eq) as internal standard at room temperature to determine yield.   Figure S120. 11 B NMR spectrum of reaction mixture of 1 and HBPin (1:2) in C6D6 and 1 eq of α,α,α-trifluoroacetophenone at room temperature after 21 h. Figure S121. 19 F NMR spectrum of reaction mixture of 1 and HBPin (1:2) in C6D6 and 1 eq of α,α,α-trifluoroacetophenone at room temperature. Figure S122. 1 H NMR spectrum of the crude product 11 (58%) using CH2Br2 (1 eq) as internal standard in CDCl3 at room temperature. Figure S123. 19 F NMR spectrum of the crude product 11 (58%) in CDCl3 at room temperature. Figure S124. 1 H NMR spectrum of the crude product 12 (64%) and side-product triphenylmethanol (10%) using CH2Br2 (1 eq) as internal standard in CDCl3 at room temperature. Figure S125. 1 H NMR spectrum of the crude product 13 (54%) and left-over starting material (45%) using CH2Br2 (1 eq) as internal standard in CDCl3 at room temperature.     6. X-Ray Crystallography   For data sets 1, 2, 4, 5 and 9 the CrysAlisPro 9 software package was used for data collection, cell refinement and data reduction. The CrysAlisPro software package was used for empirical absorption corrections, which were applied using spherical harmonics, implemented in SCALE3 ABSPACK scaling algorithm. For data set 3 Bruker APEX3 software package was used for data collection, the applications SAINT 10 and SADABS 11 were used for the data reduction and absorption corrections of the data, respectively. For data set 8 Xia3/DIALS software package was used for data collection, the applications SAINT 10 and SADABS 11 were used for the data reduction and absorption corrections of the data, respectively. All further data processing was undertaken within the Olex2 software package. 12 The molecular structures of all compounds were solved with the ShelXT 13 structure solution program using Intrinsic Phasing and refined with the ShelXL 14-16 refinement package using Least Squares minimisation. Non-hydrogen atoms were refined anisotropically. Hydrogen atoms were all located in a difference map and repositioned geometrically.
Complex 2 was positionally/rotationally disordered over two positions, so was split into two parts and refined with an occupancy ratio of 0.95:0.05. Split same function in Olex2 was used to rotate the whole molecule fitting the Zn centres on the residual Q peaks. The anisotropic refinement of minor component resulted in an unstable refinement and was instead refined isotropically, with fixed U(iso) values for C and N-atoms, while Zn was freely refined.
The {BCat2}anion of complex 9 was modelled as disordered over a symmetry element, exchanging places with a neutral catechol molecule to retain charge balance. Restraints on ellipsoids and geometry were used where appropriate and are given in the embedded res file. Other ellipsoids were not split, despite elongation, as this brought no improvement in refinement and indeed the ellipsoids take care of disorder across the symmetry element.
Selected crystallographic data are presented in Table S2 and S3 and full details in cif format can be obtained free of charge from the Cambridge Crystallographic Data Centre via w ww .ccdc.cam.uk /data_reques t/c if.

Computational details
All calculations were performed using the Gaussian09 series of programs. 17 Geometries were optimized with the DFT method using M06-2X functional 18 with a LANL2DZ basis set for Zn and 6-311G(d,p) basis set for the rest of the atoms. All geometry optimizations were full, with no restrictions. All stationary points were characterized as minima by vibrational analysis. Solvent effects of tetrahydrofuran were introduced using the self-consistent field approach, by means of the integral equation formalism polarizable continuum model (IEFPCM). 19 Full Cartesian coordinates for the optimised geometries are included in the .xyz file.