Simple phosphinate ligands access zinc clusters identified in the synthesis of zinc oxide nanoparticles

The bottom-up synthesis of ligand-stabilized functional nanoparticles from molecular precursors is widely applied but is difficult to study mechanistically. Here we use 31P NMR spectroscopy to follow the trajectory of phosphinate ligands during the synthesis of a range of ligated zinc oxo clusters, containing 4, 6 and 11 zinc atoms. Using an organometallic route, the clusters interconvert rapidly and self-assemble in solution based on thermodynamic equilibria rather than nucleation kinetics. These clusters are also identified in situ during the synthesis of phosphinate-capped zinc oxide nanoparticles. Unexpectedly, the ligand is sequestered to a stable Zn11 cluster during the majority of the synthesis and only becomes coordinated to the nanoparticle surface, in the final step. In addition to a versatile and accessible route to (optionally doped) zinc clusters, the findings provide an understanding of the role of well-defined molecular precursors during the synthesis of small (2–4 nm) nanoparticles.

. For 3A, red circles are assigned to phenyl resonances from asymmetric ("planar") phosphinate ligands, whilst blue circles are assigned to resonances from symmetric ("bridging") phosphinate ligands. At 273 K the phenyl resonances for 3A are all sharp suggesting a relatively static species, however at 298 K the two phenyl environments attributed to the "planar" DPPA units broaden suggesting that these DPPA ligands may rotate, exchanging the phenyl positions, presumably via (partial) dissociation. The hydroxide signal of 3A remains sharp across all temperatures and there is no evidence of the "bridging" DPPA units interchanging positions with the "planar" positions even at higher temperatures (up to 328 K). 1 H NMR Spectra of 3B (288 K-328 K) and 4A (298 K) also show broadening exclusively in asymmetric phosphinate environments. These observations suggest a consistent ligand coordination mode exists in these clusters, with fast rotation of the phosphinate relative to the NMR timescale at room temperature, but exchange between different phosphinate environments not observed upon the same timescale. . The data show that an equilibrium exists between 4A, 2A and 3A in the presence of water, however 4A is strongly favoured even with multiple equivalents of water present. with varying amounts of water (100% = full hydrolysis of all Zn−Et bonds). "Unknown cluster" is displayed in Figure 6b of main text whilst other minor signals are included in Figure 6c of main text. The spectra reveal the initial presence of 1C which is consumed into minor clusters and 5C upon initial hydrolysis. 5C grows to become the dominant species over the partial hydrolysis regime, before declining sharply upon complete hydrolysis. Traces of 2C are clearly observed in the final spectra but a large broad signal for ligand coordinated to ZnO nanoparticles is also observed upon careful inspection. Although little change occurs to the DOPA containing species some further consumption of free ZnEt 2 does occurthis suggests that a small amount of moisture had not fully incorporated into the toluene solution during the initial 30 minute mixing period. Some minor species are lost over this waiting period, however, interestingly they re-appear upon addition of the next aliquot of water. . It is assumed that the initial reaction (before the addition of water) absorbs 20% of the total zinc into cluster 1C leaving 80% as free ZnEt 2 . The ZnEt 2 fraction is estimated from 1 H integrals of ZnEt 2 from NMR spectra relative to the internal standard (PPh 3 phenyl groups). 1C (4x Zn), 5C (11x Zn) and 2C (4x Zn) fractions are estimated from relative 31 P integrals. *ZnO fraction is assumed to make up the zinc balance to 100% -this does not take into account other unknown minor species and so will be a slight overestimate especially at low levels of hydrolysis indicated by dashed region.

NMR spectra of 5C
Isolating 5C as an oil led to partial decomposition/reaction with trace impurities and so the most representative spectra were taken directly from the reaction solution (protonated solvent). The position of the ethyl signals varies with solvent choice hence a CDCl 3 and H 8 -toluene 1 H NMR spectrum are included here. The CH 3 signal of the minor Et component of 5C is not located in either solvent and is likely obscured by other signals.

Characterisation of ZnO nanoparticles without capping ligand
Supplementary Figure 73: Powder X-ray diffraction pattern of ZnO nanoparticles prepared by the hydrolysis of ZnEt 2 without any added ligand, analysis of the peak width indicates a particle size of ~3.5 nm using the Scherrer equation. The resulting white powder was insoluble.

Supplementary Note 1 Comparison of 5A and 5C
The species 5C appears to be analogous to the known structure 5A. Both form under partial hydrolysis conditions and show two clear ethyl environments in a 4:1 ratio when dissolved in CDCl 3 . The addition of ZnEt 2 to 2C also establishes an equilibrium with a 5:1 ratio of 1C to 5C as is also observed for 2A/1A/5A and 2B/1B/5B. Whilst 5A displays a well-defined diastereotopic relationship between geminal CH 2 protons in one ethyl group in its 1 H NMR spectrum this is not observed for 5C, the addition of bulkier ligands in this case may enforce a more flexible molecular structure.

ZnO@DOPA(5:1) nanoparticles
The synthesis of DOPA capped ZnO nanoparticles by the hydrolysis of ZnEt 2 has been previously reported. 5

Van't Hoff analysis
To test the equilibrium between 2A/B and 3A/B variable temperature NMR spectra were recorded between 288 and 328 K. After reaching the required temperature 1 H and 31 P{ 1 H} spectra were recorded, these were repeated after a waiting period of at least 15 minutes. All repeat spectra show the same ratio of species as the initial spectrum indicating that equilibrium is established quickly at each temperature. The ratio of 2A/B to 3A/B was calculated from integral analysis of the 31 P{ 1 H} spectra and from this the concentrations of both species were calculated considering the known amount of ligand. The ratio of water to 3A (for 2A/3A) and to 2B (for 2B/3B) was calculated from integral analysis of the 1 H NMR spectrum comparing the dissolved water signal (δ ~1.62, slight temperature dependence) to a clear aromatic resonance of known relative integral (for 3A δ 7.82, for 2B δ 6.58).
Plotting a graph of ln(K (eq) ) vs 1/T allows estimation of the enthalpy and entropy of reaction  This shows that the equilibrium position is at around 50% conversion from 2A to 1A and 5A.
Starting with 2B, The final ratios determined by integral analysis were: 3 2B : 4 ZnEt 2 : 5 1B: 1 5B. This shows that the equilibrium position is at around 57% conversion to 1B and 5B. N.B. in this case an 'unknown' species is also observed with an integral similar in magnitude to 5B.
This experiment shows the generality of structures 1A and 5A in the alternative phosphinate (D Meo PPA) system. Whilst 1B and 5B have not been isolated we can be confident from the similar reactivity and chemical shifts to 1A and 5A that they form in this case.
A variable temperature study was conducted on the ZnEt 2 /2A ⇄ 1A/5A system, the sample was allowed to equilibrate for 25 minutes at each temperature. The same preparation was used as described above, in this case a ratio of 0.4 2A : 7 ZnEt 2 : 5 1A: 1 5A was initially obtained at room temperature, suggesting slightly less than 4 equivalents of 2A were measured (relative to 15 ZnEt 2 ). The NMR tube was heated to 298, 308 and 318 K to determine the response to increasing temperature. The ratio adjusts to approximately 1.1 2A : 9 ZnEt 2 : 5 1A: 1 5A at 308 K and to 1 2A : 13 ZnEt 2 : 3 1A: ~0 5A at 318 K, although it must be stressed that impurities grow in upon heating which hinder accurate analysis. On cooling the sample back to room temperature, the 2A signal reduces and a trace of 5A grows back in, however, decomposition products now appear as a significant fraction.

In situ NMR spectroscopy study of the synthesis of ZnO nanoparticles by hydrolysis route
A Young's NMR tube was loaded with 14 mg (0.048 mmol) of di-octylphosphinic acid and a capillary containing a CD 2 Cl 2 solution of PPh 3 was added as an internal standard. 0.5 ml of d 8 -toluene was then added followed by 30 mg (0.243 mmol) of ZnEt 2 . The resulting mixture was analysed by NMR spectroscopy. Under a flow of N 2 0.5 µL (0.028 mmol) aliquots of water were added sequentially (with NMR spectra recorded between samples, each approximately 30 minutes after the addition of water). Upon addition of 2 µL of water (50% hydrolysis) the mixture was left to stand overnight and NMR spectra recorded again after a 15 hour pause. Further additions of water were then conducted as before until 4 µL of water had been added (a further 0.5 µL was also added to ensure full hydrolysis had occurred). The total reaction can be described as:

In situ UV spectroscopy study of the synthesis of ZnO nanoparticles by hydrolysis route
Di-octylphosphinic acid (51.7 mg, 0.18 mmol)) was dissolved in 12 ml of toluene and to this ZnEt 2 (110 mg, 0.89 mmol) was also added. The solution was split into four 3 mL fractions and 1, 2, 3 and 4 µL of water were added to each fraction respectively giving 25%, 50%, 75% and complete hydrolysis (of which those at 50% and 75% displayed a yellow colour). The solutions were stirred for one hour and then diluted and sealed into UV cuvettes (using a glovebox for the partially hydrolysed samples) at [Zn] = 4x10 -3 M. UV spectra were recorded over the 400-290 nm region. A control reaction revealed that a solution of 1C (e.g. 0% hydrolysis) did not absorb in this UV region, nor did the 25% hydrolysed solution.

Synthesis and characterisation of 'yellow ZnO'
To 25 mL of toluene, ZnEt 2 (500 mg, 4.05 mmol) was added and then under a flow of N 2 0.75 equivalents of H 2 O (54.6 µL, 3.03 mmol) was added quickly. Since water is not very miscible with toluene the reaction proceeds over several minutes (ethane gas is evolved and the flask was left open to a bubbler). The mixture was stirred for 30 minutes with periodic sonication to make sure all moisture is evenly incorporated. A yellow solution forms. The solvent was removed by vacuum to leave a glassy yellow powder. This powder does not immediately change colour under air, however over prolonged periods (>1 week) it becomes white. Once isolated as a solid it is insoluble.

X-ray Crystallography Details
X-ray crystallography data for 1A, 2B, 3A and 5A were collected on an Oxford Diffraction diffractometer using graphite monochromated Mo K radiation (λ = 0.71073 Å) and a lowtemperature device [173(2) K]; 6 X-ray crystallography data for 4A was collected on an Agilent SuperNova diffractometer using graphite monochromated Cu K radiation (λ = 1.54180 Å) and a low-temperature device [173(2) K]. 6 Data were collected using SuperNova, reduction and cell refinement was performed using CrysAlis. 7 The structure was solved by direct methods using Superflip 8 and refined full-matrix least squares on F 2 using CRYSTALS. 9 All non-hydrogen atoms were refined with anisotropic displacement parameters. All hydrogen atoms were placed in calculated positions and refined before applying the riding model. Crystallographic data have been deposited with the Cambridge Crystallographic Data Centre under CCDC 1432882-1432886. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. Full bond length and bond angle data may be found in the CIFs.
Special refinement details: 3A: Two solvent molecules of toluene were located in the Fourier map, one could be sensibly refined with the use of symmetry restraints, however, the other could not be adequately modelled due to extensive disorder and so was treated using the SQUEEZE algorithm. 10 Two of the phenyl groups showed minor disorder in the form of enlarged displacement ellipsoids, one was modelled over two positions and restrained to maintain sensible geometries, the other was best modelled by adding symmetry restraints to the original position. Hydrogen atoms upon the OH groups (H192, H203 & H212) could not be located in the Fourier map and were placed in calculated positions before allowing free refinement and then applying the riding model. We anticipate these hydroxide protons could be disordered over different sites but this model does not include this.
4A: Eight solvent molecules of THF were located in the structure. Inspection of the Fourier map allowed identification of the oxygen atom (with greater electron density) in each case. Symmetry restraints were applied to each THF molecule and all eight molecules were restrained to each other to exhibit similar connectivity. One molecule of THF was split into two components with the occupancy of each part refined.
5A: Shift limiting restraints were added to aid initial refinement but were removed once the structure was correctly located, further refinement was conducted. Symmetry restraints were added to two phenyl groups which showed mild disorder. A racemic mixture is apparently present in the structure (Flack parameter = 0.397(10)), the approximately spherical shape may allow both enantiomers to cocrystallise.