Novel solution-phase structures of gallium-containing pyrogallol[4]arene scaffolds

Abstract

The variations in architecture of gallium-seamed (PgC₄Ga) and gallium-zinc-seamed (PgC₄GaZn) C-butylpyrogallol[4]arene nanoassemblies in solution (SANS/NMR) versus the solid state (XRD) have been investigated. Rearrangement from the solid-state spheroidal to the solution-phase toroidal shape differentiates the gallium-containing pyrogallol[4]arene nanoassemblies from all other PgC_nM nanocapsules studied thus far. Different structural arrangements of the metals and arenes of PgC₄Ga versus PgC₄GaZn have been deduced from the different toroidal dimensions, C–H proton environments and guest encapsulation of the two toroids. PGAA of mixed-metal hexamers reveals a decrease in gallium-to-metal ratio as the second metal varies from cobalt to zinc. Overall, the combined study demonstrates the versatility of gallium in directing the self-assembly of pyrogallol[4]arenes into novel nanoarchitectures.

The synthesis of metal-seamed architectures continues to attract wide attention among supramolecular chemists. Understanding the properties of such metal-seamed systems could lead to practical applications such as biological probes,^[1] electronic nanodevices^[2] or luminescent superparamagnetic hybrids.^[3] Investigation of metal-seamed pyrogallol[4]arene organic nanocapsules (MONCs) dates back to 2005 with the formation of copper-seamed C-propan-3-ol-pyrogallol[4]arene hexamers.^[4] In this assembly, six pyrogallol[4]arenes are partially deprotonated to yield a small rhombicuboctahedron with eight triangular [of (O-Cu-O)₃ units] and eighteen square faces. Later, zinc-seamed C-propylpyrogallol[4]arene (PgC₃Zn, where 3 is the alkyl chain length) dimeric nanocapsule formation was synthesized.^[5]

Gallium is of particular interest with respect to MONCs, owing to its semiconducting properties and its ability to assemble into interesting architectures. For example, gallium cages,^[6] gallium nanoparticles,^[7] single-wall gallium nitride nanotubes^[8] and GaP nanowires, nanocables and nanobelts^[9] have been reported. Amongst the pyrogallol[4]arene frameworks, gallium-containing scaffolds are unique in several aspects, such as the diversity of architectures, the metal to pyrogallol[4]arene ratio and the metal oxidation state.

Unlike the other observed dimeric and hexameric pyrogallol[4]arene nanocapsules, which are spherical, the gallium-seamed C-butylpyrogallol[4]arene (PgC₄Ga) hexamers have a distorted `rugby-ball' shape.^[10] Compared to 24 copper atoms in a typical spherical hexamer, only 12 gallium atoms, along with four water molecules, seam this hexameric assembly. These gallium-seamed nanocapsules are particularly interesting owing to the Ga⁺³ ions occupying Ga₃(μ₃-O)₃(O)₆ units on opposite planes seaming only four of the six PgC₄ moieties of the hexameric framework.^[10] Water molecules seam the remaining two PgC₄ moieties, providing the first partially hydrogen-bonded metal-seamed hexamer. The gated aqueous channels can be utilized for ion transportation to the interior of the capsules.^[11]

Gallium continued to show distinctive behaviour when mixed-metal hexamers were synthesized.^[11a] Addition of a second metal to a pre-formed gallium hexamer caused a transition in structure from the distorted rugby ball to a sphere. Subsequently, hexameric spheres of gallium-zinc, gallium-copper, gallium-nickel and gallium-cobalt combinations were found.^[11a] It was, however, difficult to distinguish between the metal atoms using single-crystal XRD because of the similar electronic densities of the metals involved. Hence, inductively coupled plasma (ICP) analyses were conducted to confirm the presence of the second metal and to obtain an approximate ratio of the metal ions in the framework.^[11a]

Given the interesting solid-state properties of the gallium-based nanocapsules, we have investigated their geometric dimensions and shapes in solution with small-angle neutron scattering (SANS). We have also performed prompt gamma activation analysis (PGAA) of the solid-state capsules to support the earlier ICP results. To our knowledge, the only other solution-phase study of PgC₄Ga mixed-metal hexamers is a ¹H NMR study (CD₃CN) comparing the PgC₄GaZn hexamer to the PgC₄Ga hexamer.^[12] The one definitive conclusion obtained from these studies was that solvent is encapsulated only in the presence of zinc. To obtain more specific structural details, we performed SANS measurements, 1-D NMR and 2-D HMQC and diffusion-ordered NMR spectroscopy for the PgC₄Ga and PgC₄GaZn nanoassemblies. Previously, we have employed SANS to examine the self-assembly process and geometric dimensions of the spherical PgC₃Cu and PgC₃Ni hexamers and dimers in solution.^[13] These two sets of experiments will allow us to identify any differences in the behaviour of a typical transition metal (copper/nickel)-seamed PgCn sphere and a gallium-seamed PgCn distorted rugby ball.

The PgC₄Ga hexameric crystals were formed by the earlier reported method of mixing acetone solubilized PgC₄ (0.26 × 10⁻³ mol/L) with water solubilized gallium nitrate (0.28 g in 2 mL water).^[12] Likewise, the PgC₄GaZn hexamer crystals were obtained by adding a zinc-nitrate ethanolic solution (0.28 g in 2 mL) to pre-formed and dried PgC₄Ga crystals that had been dissolved in acetone (0.015 × 10⁻³ mol/L).^[12] A similar method was followed for the preparation of the Ga/Ni, Ga/Co and Ga/Cu mixed-metal nanocapsules. Slow evaporation of each solution led to crystals suitable for structure elucidation. The unit cells for these crystals matched those obtained previously.^[11a]

Cold neutron PGAA^[14] was conducted to investigate the metal ratios in mixed-metal pyrogallol[4]arene frameworks and the results are compared with the earlier ICP analyses in Table 2. Interestingly, however, the new measurements lead to a noticeable trend in the metal ratios across the periodic table. The average mmol ratios of Ga⁺³:M⁺² decrease from approximately 2.8 for Ga⁺³:Co⁺² to 1.8 for Ga⁺³:Ni⁺² to 1.4 for Ga⁺³:Cu⁺² and to 1.3 for Ga⁺³:Zn⁺². That is, as the atomic number of the transition metal increases, the transition metal fills more sites in the Ga/M hexameric framework, suggesting an enhanced affinity to stitch the hexameric framework from cobalt to zinc (Table 1).

Table 2.

Resolution-smeared hollow cylinder fits to nanoassemblies of C-butylpyrogallol[4]arene gallium-zinc and gallium.^a

SAMPLE		Unitb (Å)	SQRT (χ2/N)
PgC4GaZn	Rcore	6.31 ± 0.15	1.07
	Rshell	14.52 ± 0.11
	L	13.10 ± 0.24
PgC4Ga	Rcore	4.78 ± 0.27	1.307
	Rshell	16.8 ± 0.17
	L	6.9 ± 0.36

^aR_core = core radius; R_shell = shell radius; L = height

^bUncertainties in the fitted parameters are one standard deviation.

Table 1.

Metal ratios from Prompt Gamma Activation Analysis (PGAA) of Ga/M-seamed pyrogallol[4]arene hexamers.^a

Metal 1:Metal 2	ICP Analysisb	PGAA
Ga : Zn	14 : 10	13.5 : 10.5
Ga : Cu	13 : 11	14 : 10
Ga : Ni	15 : 9	15 : 9
Ga : Co	16 : 8	18 : 6

^aExpanded uncertainties for PGAA values are estimated at ≤ 5 % based on counting statistics, background subtraction, purity of the standards, and neutron flux uncertainties.

^bReference 10.

The SANS measurements were conducted on the NG7 30m SANS instrument at the NIST Center for Neutron Research (NIST-NCNR) in Gaithersburg, MD.^[15] Neutrons of wavelength λ = 6 Å with full width half-maximum Δλ/λ = 15% were used. Sample to detector distances of 1.3 m and 6 m were employed to cover the q range of 0.012 Å⁻¹ < q < 0.52 Å⁻¹, where q is the scattering vector. Pre-formed PgC₄Ga hexamers^[11a] and PgC₄GaZn hexamers^[11a] were dissolved separately in deuterated (d6)-acetone at a mass fraction of 2%. Deuterated solvents improve the scattering contrast and thus the coherent scattering of neutrons that contains the structural information. The collected scattering data was then reduced and analyzed on software provided by NIST.^[16] Factors such as instrumental geometry, apertures, wavelength spread, effect of gravity on neutron trajectory, etc. have been taken into account by employing resolution-smeared models for the data fitting and modelling. Smeared models provide a more accurate representation of the (smeared) experimentally measured scattering data.

The SANS data for typical PgC₃M hexamers and dimers fit to a polydisperse sphere,^[17] with a radius of ≈10 Å and ≈7 Å, respectively.^[13] These solution phase measurements correspond well with the previously reported XRD structures, which establishes the robustness of pyrogallol[4]arene-based MONCs in solution.^{[5, 13, 18]} Given this parallel connection in shape and size in solution and solid phases, we expected the solution-phase scattering data for the PgC₄Ga and PgC₄GaZn nanoassemblies to fit to an ellipsoid and sphere, respectively. However, the fits to the uniform ellipsoid^[19] and Schulz sphere^{[17a, 20]} models are of poor quality and indicate, for the first time, differences in the architectures of the nanoassemblies in the two phases (supporting information).

The PgC₄Ga and PgC₄GaZn neutron scattering data were then fitted to spherical polydisperse core-shell,^[21] prolate core-shell^{[17b, 22]} and hollow cylinder^[19] models (see supporting information). The quality of the fit for the first two models is again very poor; the data fit exceptionally well, however, to the hollow cylinder model for both PgC₄GaZn and PgC₄Ga (Figures 1 and 2; Table 2). These are the first examples of cylindrical-shaped pyrogallol[4]arene nanoassemblies in solution. For this model, the scattering length densities (SLD) for the shell were calculated as 1.63×10⁻⁶ Å⁻² and 1.8×10⁻⁶ Å⁻² for PgC₄Ga and PgC₄GaZn, respectively. The solvent (d6-acetone) SLD was calculated as 5.39×10⁻⁶ Å⁻² and fixed to the hollow core.^[19]

Figure 1.

Pictorial representations of the geometries of gallium-seamed nanoassemblies as a function of addition of a second metal and change in phase.

Figure 2.

SANS data for the gallium-seamed C-butylpyrogallol[4]arene nanocapsule.The SANS data is fitted to a resolution-smeared hollow cylinder model (see text). Error bars on the SANS data represent one standard deviation.

Comparing the values from the fits, there is a minor variation in the core and shell sizes. The inner radius (core) increases by ≈1.5 Å and the outer radius (shell) decreases by ≈2 Å for PgC₄Ga compared to PgC₄GaZn. Overall, the shell thickness is about 4 Å larger for PgC₄Ga (≈12 Å) than for PgC₄GaZn (≈8 Å). In contrast, the length of the cylinder for PgC₄GaZn (≈13 Å) is twice that for PgC₄Ga (≈7 Å). Unexpectedly, the outer radius is larger than the length of the cylinder for both the PgC₄Ga and PgC₄GaZn species in solution. In fact, for the former species the shell diameter (≈34 Å) is more than four times the length of the cylinder, whereas for the latter species the shell diameter (≈29 Å) is more than twice the length of the cylinder (Table 2). The shape of the PgC₄Ga species in solution is thus a “donut-shaped/toroidal-shaped” assembly that expands in length on addition of the zinc metal ion. Similar self-assembled spherical and toroidal frameworks of dendrimersomes have been observed by Presec et al., reflecting the adaptibility of intermolecular interactions.^[23]

It is important to note that PgCns are not cylindrical in either the solid or solution phase. Hence, both the metal and solvent are playing a role in directing the solution-phase hollow cylindrical/toroidal architectures of the PgC₄Ga and PgC₄GaZn assemblies. It remains to be seen whether this interesting result will be observed for other gallium-based, mixed-metal species.

The variations in the structures of PgC₄Ga and PgC₄GaZn in the two phases suggest a conformational change in d6-acetone. To further study their solution-phase behavior and exhange dynamics, the PgC₄Ga and PgC₄GaZn nanoassemblies were investigated through ¹H, ¹³C, ⁷¹Ga, HMQC and diffusion-ordered (DOSY) NMR spectroscopy techniques. Because the peaks in the ¹³C 1-D NMR spectra are too broad to be observed with sufficient signal-to-noise ratio, the ¹³C peak positions for the linker and aromatic groups have been assigned via one-bond carbon-proton correlation in the HMQC (supporting information). HMQC indicates that the group of broad proton peaks between 6 and 8 ppm derive from protons directly bonded to the phenyl carbons (110 ppm) and that the peaks at 4.5 ppm derive from protons directly bonded to the linker carbons (35 ppm). The proton signals at 0.94 ppm and 1.4 ppm are correlated to the carbons of the -CH₃ (14 ppm) and -CH₂ (22 and 32 ppm) moieties, respectively, of the alkyl chains (supporting information). In addition to the peak broadening, the ¹H NMR spectra of PgC₄Ga and PgC₄GaZn show multiple peaks associated with the aromatic -CH and linker -CH protons, in contrast to the single ¹H signal from each of the two groups in PgC₄. PgC₄Ga and PgC₄GaZn show at least three environments for the aromatic -CH group and two environments for the linker -CH group. Note also the switch in the relative peak heights for the two sets of protons in the two toroidal nanoassemblies (supporting information).

Two peaks in the negative region (0 and -1.7 ppm) in the ¹H NMR spectra are observed only for PgC₄GaZn, suggesting that small molecules are encapsulated in only this toroidal nanoassembly. This result is consistent with that reported earlier for PgC₄Ga and PgC₄GaZn in CD₃CN.^[12] The peak at 0 ppm produced a carbon counterpart (29 ppm) in the HMQC. This observation indicates that the encapulated molecule is either acetone or ethanol (some ethanol was present during sample preparation), but not water (supporting information).

PgC₄Ga has a diffusion coefficient in d6-acetone of 5.37×10⁻¹⁰ m²/sec, which remains essentially unchanged even after 7 days (5.33×10⁻¹⁰ m²/sec). The corresponding diffusion coefficients for PgC₄GaZn and PgC₄ are 5.07×10⁻¹⁰ m²/sec and 8.65×10⁻¹⁰ m²/sec, respectively. Furthermore, the diffusion coefficient for the entrapped small molecule is the same as that of the host PgC₄GaZn toroid. These diffusion coefficients are similar in magnitude to those found by Cohen and coworkers for hydrogen-bonded resorcin[4]arene and pyrogallol[4]arene hexamers in chloroform.^[24]

The ⁷¹Ga NMR for the PgC₄Ga or PgC₄GaZn filtrate in d6-acetone shows a sharp peak from free gallium. In contrast, the ⁷¹Ga NMR for the crystalline hexamers dissolved in d6-acetone at both 27°C and −70°C reveals no recognizable signal, which indicates the absence of free gallium in solution (Figure 3). Overnight ⁷¹Ga NMR on both PgC₄Ga and PgC₄GaZn in d6-acetone at room temperature produced no free gallium signal.

Figure 3.

⁷¹Ga NMR of (a) PgC₄Ga crystals in d6-acetone at −70°C and (b) PgC₄Ga crystals in d6-acetone at 27°C shows no signal for free gallium; (c) supernatant of PgC₄Ga at 27°C shows a free gallium peak in d6-acetone. (The broad humps in the baseline in (a) and (b) are due to the probe's background).

Overall, these experiments provide evidence for not only the presence of gallium in the self-assembled toroidal frameworks but also the stability of the frameworks. The experiments also demonstrate that the PgC₄GaZn toroids comprise solvent as well as metal and that the guest solvent remains encapsulated over time.

Combining the above SANS and NMR data and assuming the same metal content as observed from XRD and PGAA, we propose the following structural details for the solution-phase toroidal assemblies. Calulating the shell volume using equation 1 yields a Vshell of ≈5600 ų for the PgC₄Ga toroidal assembly and ≈7000 ų for the PgC₄GaZn assembly.

$${\mathrm{V}}_{\mathrm{shell}}=\pi ({\mathrm{R}}_{\mathrm{shell}}^2-{\mathrm{R}}_{\mathrm{core}}^2)\mathrm{L}$$ (1)

Both XRD and SANS measurements reveal the dimensions of ca. 14 × 7 Å for a typical PgCn (n = 1–4) bowl. The 7 Å length is the distance between the centroid of the linker carbons at the lower rim and the centroid of the oxygens at the upper rim. The 14 Å diameter is the distance between opposing oxygens at the upper rim. Thus, the difference of ≈1400 ų in V_shell corresponds to the volume of 2 bowls of PgC₄. Assuming that the toroidal assemblies consist of only PgC₄ bowls, the individual shell volumes are consistent with 7 bowls in PgC₄Ga and 9 bowls in PgC₄GaZn.

The shorter axis of the PgC₄ bowl corresponds to the height of the PgC₄Ga toroid while the longer axis of the bowl corresponds to the height of the PgC₄GaZn toroid. This observation suggests a possible difference in the orientation of the PgC₄ macrocyles in the two toroidal assemblies, with the upper and lower rims of the macrocyles aligned with the upper and lower rims of the PgC₄Ga toroid but perpendicular to the upper and lower rims of the PgC₄GaZn toroid.

The suggested structure of the PgC₄Ga toroid has the 7 arenes in a boat conformation providing possible coordination sites for the 14 gallium ions between the flat pyrogallols on adjacent arenes and between opposite pyrogallols on the same arene. The sideways orientation of the 9 cone-shaped macrocyles in the PgC₄GaZn assembly allows guest encapsulation and provides sufficient coordination sites between adjacent arenes for the 24 metals that stitch the toroidal assembly (Figure 4). Among other possible interactions, the assemblies may also be stabilized by hydrogen-bonding between the pyrogallol hydroxyl groups and solvent and by coordinate covalent bonds between the metal and solvent.

Figure 4.

Top: Top view of the suggested arrangement of pyrogallol[4]arenes in PgC₄GaZn (left) and PgC₄Ga (right) nanoassemblies in acetone. Bottom: Top view of the suggested metal coordination sites in the PgC₄Ga toroidal framework in acetone.

The proposed metal coordination sites lead to at least three different environments for the pyrogallol -CH protons and two different environments for the linker group protons, as observed in the ¹H NMR. For example, focusing on the PgC₄Ga assemblies, because gallium interacts with only two of the three hydroxyl groups of a given pyrogallol, all four of the aryl protons of a given arene are non-equivalent (Figure 4, bottom). The (at least) two different types of protons on the linker carbons arise for the same reason. If both gallium and zinc metals are present in the framework of the PgC₄GaZn assembly, as suggested by the difference in the toroidal dimensions, the more complicated multiple proton environments would be expected.

The results of the SANS, NMR and PGAA studies conducted in this work are intriguing because they demonstrate that PgC₄Ga nanoassemblies are distinctive in five ways: (1) the toroidal structure of the PgC₄Ga and PgC₄GaZn assemblies is different from the spherical structure of other PgCnM (M = Cu, n = 3–13; M = Ni, n = 3) assemblies^[25] in the solution phase; (2) the rugby ball structure of the PgC₄Ga hexamer is different from the spherical structures of the PgC₃M (M = Cu, Ni) hexamers^[13] in the solid state; (3) the structure of PgC₄Ga is different from that of PgC₄GaZn in the solid and solution phases; (4) guest encapsulation is not observed for the PgC₄Ga toroid in solution ; (5) the Ga⁺³:M⁺² ratio in the PgC₄GaM hexamers varies with M, decreasing from Co⁺² to Zn⁺². That is, the SANS measurements reveal the first structural change from the solid state to the solution phase, the NMR results reveal structural details unique to each toroid, and the PGAA measurements reveal the trend among the late 3d transition metals in directing the formation of mixed metal-seamed pyrogallol[4]arene nanoassemblies. Overall, the cohesive SANS and NMR experiments provide important insights into the solution-phase behavior of gallium-seamed C-butylpyrogallol[4]arene nanoassemblies.

Future studies will continue to address the ways in which non-covalent interactions and metal ions govern the self-assembly process of bi- and tri-metallic gallium-seamed nanoassemblies. In particular, combined solution-phase and computation studies will be performed to obtain more insight into the mechanism of formation of and metal ion positions in gallium-seamed scaffolds.