William J.
Ramsay
,
Jonathan A.
Foster‡
,
Katharine L.
Moore
,
Tanya K.
Ronson
,
Raphaël J.
Mirgalet
,
David A.
Jefferson
and
Jonathan R.
Nitschke
*
Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, UK. E-mail: jrn34@cam.ac.uk
First published on 24th September 2015
The incorporation of an N-heterocyclic carbene (NHC) moiety into a self-assembled MII4L6 cage framework required the NHC first to be metallated with gold(I). Bimetallic cages could then be constructed using zinc(II) and cadmium(II) templates, showing weak luminescence. The cages were destroyed by the addition of further gold(I) in the form of AuI(2,4,6-trimethoxybenzonitrile)2SbF6, which caused the reversibly-formed cages to disassemble and controllably release the AuI-NHC subcomponent into solution. This release in turn induced the growth of gold nanoparticles. The rate of dianiline release could be tuned by capsule design or through the addition of chemical stimuli, with different release profiles giving rise to different nanoparticle morphologies.
The shape and dimensions of discrete, self-assembled supramolecular structures are predominantly dictated by how ligands with a characteristic size and denticity come together in geometries defined by the coordination sphere of a given metal ion.15–28 The functionality29–36 of these assemblies can be shaped by incorporating ligands with active functional groups, which can also be modified in situ either pre-37–40 or post-synthetically.41–44 The response of a ligand to an external stimulus, such as light, can influence the guest binding affinity of the corresponding supramolecular structure.45 When a ligand contains a coordinated transition metal that is catalytically active, this active site can be orientated explicitly by the supramolecular framework, either protecting or exposing it;46 incorporating more than one metal center thus can allow for division of labor between structural and functional metal centers.
N-Heterocyclic carbenes (NHCs) are a useful class of electron-donating ligands that form strong metal–ligand bonds.47 Gold(I)-NHCs, in particular, have found applications in the fields of catalysis,48 pharmaceuticals,49 liquid crystals50 and optical devices.51,52 Despite the potential of incorporating metal-based NHC ligands into supramolecular constructs, opening new possibilities of control over their reactivity, few examples of self-assembled structures with this motif have been reported.53–56 We thus designed subcomponent A and its metallated carbene derivatives B–D (Scheme 1), to be capable of reversibly self-assembling with 2-formylpyridine and metal ions to form supramolecular structures, in order to probe the scope and limitations of the ability of this class of ligand to self-assemble into supramolecular structures. As AuI-NHC complexes can be reduced to form carbene-stabilized gold nanoparticles (Au NPs),57 we anticipated that incorporating this ligand motif into a reversibly formed supramolecular structure could allow for new means to be developed to control the nucleation and growth conditions of Au NPs.
Scheme 1 Syntheses of subcomponents A–D. Reaction conditions: (i) CuCl, Na2CO3, acetone, 60 °C, 16 h, 90%; (ii) Ag2O, DCM, 25 °C, 2 h, 90%; (iii) Au(tht)Cl, aq. Na2CO3, DCM, 25 °C, 0.5 h, 92%. |
These subcomponents were subjected to conditions that were anticipated to lead to the preparation of self-assembled MII4L6 structures.25 Each was treated in turn with 2-formylpyridine and the prospective metal templates ZnII, CdII, FeII and CoII. Although dianilines A–C did not yield discrete complexes with any of these templates, and D with FeII or CoII gave intractable mixtures of products, D (6 equiv.) was observed to generate metal–organic cages 1 and 2 following treatment with 2-formylpyridine (12 equiv.) and either zinc(II) or cadmium(II) di[bis(trifluoromethylsulfonyl)imide] (triflimide, NTf2−; 4 equiv.) respectively, as the uniquely observed products in solution (Fig. 1; ESI Sections 1.3 and 1.4†). Confirmation of the MII4L6 stoichiometry was provided by electrospray ionization mass spectrometry (ESI MS; Fig. S10a and S20b†), including high resolution MS (Fig. S10b and S20b†).
The observation that dianilines A–C did not lead to discrete cage formation suggested that a strong and inert metal-NHC bond was required in the ligand to construct a cage, which was provided by the gold(I) chloride moiety in D. Fluorescence spectroscopy for both 1 and 2 in dry acetonitrile (Fig. S11, S12, S21 and S22,† respectively) indicated weak luminescence, with bands that are attributed to both intra- and intermolecular AuI⋯AuI interactions, as observed in gold(I)-containing supramolecular architectures developed by Yam and co-workers.60 The removal of the gold-bound chloride from ligand D in cages 1 and 2 following the addition of chloride-binding metal cations, such as stoichiometric AgI or excess ZnII or CdII (Fig. S13, S14, S23 and S24†) caused disassembly, suggesting that coulombic repulsion between adjacent ligands destabilized the cage framework. Agents that engender the removal of chloride, therefore, provided a stimulus to trigger the response of cage destruction.
NMR spectra for the products 1 and 2 (Fig. S5 and S15,† respectively) yielded three imine signals. In the case of 2, satellite signals associated with the imines were observed, attributed to J-coupling with the two spin-1/2 isotopes of cadmium.61 The three magnetically inequivalent ligand environments observed in both the 1H and 13C NMR spectra of the cages suggested that the achiral S4 (ΛΛΔΔ) diastereomer of a M4L6 framework was present in solution, but not the homochiral T (ΔΔΔΔ/ΛΛΛΛ) or heterochrial C3 (ΔΔΔΛ/ΛΛΛΔ) diastereomers.62,63 The steric bulk of the isopropyl groups is expected to hold the phenyl rings orthogonal to the central imidazolium ring; this coplanar orientation of the terminal phenylene rings has been shown to favor a syn arrangement of the ligands,64 making the S4-symmetric framework the lowest-energy conformation, and thus the unique product observed in solution. Diffusion ordered spectroscopy (DOSY) was consistent in each case with all signals belonging to a single species (Fig. S9 and S19†). The 19F NMR signal for NTf2− was unchanged from its free value in cages 1 and 2, consistent with the absence of anion encapsulation (Fig. S6 and S16†).
In order to visualize the cage geometry, an energy-minimized model of 2 in the S4 arrangement65 was constructed (Fig. 2b) based upon the X-ray crystal structure of D (Fig. 2a). The achiral framework is composed of two metal centers of the same handedness (Λ) and two metal centers of opposite handedness (Δ). Each pair of metal centers of the same stereochemistry is connected by ligands adopting anti conformations (where the –AuCl moieties are directed to the outside of the cage), and the other four ligands are syn (where the –AuCl moieties are directed inwards); we acknowledge that other conformations may be possible in solution. Our model suggests that the cages are sufficiently flexible to allow the gold(I) centers to approach each other, accounting for the weak luminescence associated with the presence of gold–gold interactions (further modelling of the intermolecular cage interactions in 2 is provided in Fig. S22†).
Fig. 2 (a) ORTEP representation of the crystal structure of D, showing thermal ellipsoids at 40% probability level. Solvent molecules and disorder are omitted for clarity. (b) Two views of a model of the MM2 (ref. 66) energy-minimized structure of 2. The –AuCl units are shown in space-filling representation. Hydrogens and counterions are omitted for clarity (color scheme: carbon, grey; nitrogen, blue; cadmium, light yellow; gold, dark yellow; chloride, green). |
Following the addition of Au(tmbn)2SbF6 (6 equiv.) to a solution of 1 in dry acetonitrile (ESI Section 3.1†), the surface plasmon resonance (SPR) band in the UV-Vis spectrum, attributable to Au NPs,71 was observed to grow in following a sigmoidal time course with steps corresponding to nucleation, growth, and saturation (Fig. 3a); consistent growth features were observed in a second run (Fig. S26†). When cadmium-containing cage 2 was mixed with Au(tmbn)2SbF6 (6 equiv.) (ESI Section 3.2†), the SPR band in the UV-Vis spectra began to grow in intensity approximately 22 min after mixing (Fig. 3a), a delay 10 min longer than was observed in the case of 1 (Fig. S36†). In both experiments, NP growth was observed only after part of the cage had first disassembled (Fig. S27, S28, S37 and S39†). When cage 1 decomposed to provide the reducing agent, TEM (Fig. S29†) and AFM (Fig. S30†) images displayed aggregates (250 nm in diameter) of approximately 5 nm Au NPs, together with Au NPs approximately 50 nm in diameter (Fig. 3b), following 90 min of NP growth. In contrast, when cage 2 disassembled {111}-faceted triangular prisms approximately 300 nm in edge length were observed (Fig. 3b), in addition to aggregates of 5 nm Au NPs (Fig. S40 and S41†). Particle size distribution analysis of AFM measurements confirmed the presence of the larger triangular species alongside numerous small NP (Fig. S41d–h†); the triangular prisms constituted approximately 10% of the species counted, but approximately 52% of the total gold.
Fig. 3 (a) The evolution of the SPR UV-Vis band was monitored during Au NP formation when the reducing agent employed was provided by cages 1 and 2 and with D and either Zn(NTf2)2 or Cd(NTf2)2; the curves were fitted using the Avrami equation72,73 (ESI Section 3.6†). (b) Differences in Au NP morphology observed in the TEM images for growth conditions shown in (a). |
We also observed that when reagent grade acetonitrile (0.01% H2O) was used in place of the anhydrous solvent (<0.001% H2O) during the synthesis of Au NPs with cage 2,74 the SPR band in the UV-Vis spectra appeared more rapidly (Fig. S42†). We infer that the higher water content of the solvent increased the rate of imine hydrolysis in 2, resulting in a shortening of the nucleation stage. Instead of the triangular prisms observed with anhydrous solvent, TEM images revealed hexagonal plates (Fig. S42†) and aggregates of 5 nm particles. This difference in Au NP morphology may be attributed to the variation in the rate of cage breakdown.
Notably, control experiments involving equimolar quantities of D, Zn(NTf2)2 or Cd(NTf2)2 and Au(tmbn)2SbF6 (ESI Sections 3.1.1 and 3.2.1†) resulted in a SPR band in the UV-Vis spectra immediately following mixing (Fig. 3a) and significantly different NP morphologies. When equimolar quantities of D and Zn(NTf2)2 were mixed with Au(tmbn)2SbF6, only aggregates were observed in the TEM images (Fig. 3b); nanoparticles greater than 5 nm in diameter were not observed in contrast to the case when 1 provided reductant D through its hydrolysis. In addition, none of the anisotropic Au NP morphologies grown starting from 2 were observed in the control experiment involving equimolar quantities of D, Cd(NTf2)2 and Au(tmbn)2SbF6. Further control experiments in which only A or D were used in the reduction of AuI (ESI Sections 3.3 and 3.4,† respectively) revealed similarly rapid rates of formation and small nanoparticle morphologies in the presence of either CdII or ZnII. Energy-dispersive X-ray spectroscopy (EDS) confirmed the presence of gold (with small quantities of nitrogen) across all experiments (Fig. S29, S32, S40 and S44†), with a higher organic content observed in the aggregates of smaller Au NPs compared to the larger anisotropic species; spectral features consistent with Zn or Cd were not observed, suggesting that these metals were not incorporated into the NPs. The aggregation of the Au NPs and the observation of organic material in the EDS spectra suggested that the oxidized bis-monodentate ligand D may have acted to bridge between Au NPs, as has been observed with similar ligands.75
The kinetics of Au NP formation were assessed with the Avrami theoretical model for crystallization and growth72,73,76 (ESI Section 3.6†). In the Avrami equation, the overall kinetics are described by the apparent rate parameter kapp which depends principally upon the nucleation rate – a slower kapp is attributed to a lengthier nucleation stage.77 The kapp of Au NP growth when 1 provided the reducing agent was 1.8 × 10−4 (the units are min−n for all rate constants), with an Avrami exponent (n) of 2.59; in a control experiment using D and Zn(NTf2)2 as reductant, the rate constant was 1.70 × 10−2 with an exponent of 0.97. Similarly, Au NP growth where the reducing agent was provided by 2 displayed a rate constant of 1.5 × 10−6 with an exponent of 3.4, and the control experiment using D and Cd(NTf2)2 gave a kapp of 3.8 × 10−2 with an exponent of 0.90 (Table S1†). As the Avrami exponent n78–82 has been inferred to reflect the nucleation mechanism and directionality of growth, the constant nucleation rate (n ≈ 1) observed when D served as the reductant with different metal salts is consistent with the observation of spherical Au NPs. Similarly, the multi-dimensional growth (n > 1) observed when 1 or 2 were used as reductants is consistent with a heterogeneous nucleation mechanism being responsible for the anisotropic features.72 The rate constants and Avrami exponents observed when 1 and 2 were used as reductants suggest that the supramolecular structures impeded Au0 nucleation, in contrast to the faster nucleation observed when free D was present in solution.
Mechanistically, we infer that the addition of AuI to a solution of 1 or 2 first resulted in the abstraction of chloride, as noted above in the cases of other metal salts. This destabilizes the cage and results in its disassembly and release of free subcomponent D. We attribute the difference in rate of dianiline release between 1 and 2 to the differences in thermodynamic stabilities between these two structures. The shorter nucleation stage provided by the breakdown of 1 in comparison to the longer nucleation stage provided by 2 is consistent with the labile ZnII centers that have faster ligand exchange kinetics;83 it is this delay during the nucleation stage of Au NP growth that ultimately determines the final NP size and shape. The slow rate of D release from 1 and 2 thus brought about a longer nucleation stage and slower rate of AuI reduction, as required to produce thermodynamically-favored shapes with low-index facets, such as the {111}-faceted triangular prisms.84 Several other factors, including the presence of small amounts of different salts (such as ZnCl2 or CdCl2 from cage decomposition, or the metal triflate salts) could also influence the final particle shape and kinetics of nanoparticle growth,84 but the control experiments involving D and ZnII or CdII were not consistent with this mechanism. The different growth rates and morphologies of the observed Au NPs were thus inferred to result from the incorporation of an AuI-NHC dianiline ligand into the walls of metal–organic capsule as the diimine.
Footnotes |
† Electronic supplementary information (ESI) available: Experimental procedures, synthesis and characterization data for subcomponents and cages, and optical, microscopy, ESI-MS and NMR data. CCDC 1061936. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c5sc03065j |
‡ Current address: Department of Chemistry, University of Sheffield, Sheffield, S3 7HF, UK. |
This journal is © The Royal Society of Chemistry 2015 |