2.1 Synthesis and characterization

Novel molecular or supramolecular assemblies containing multiple identical redox centers have attracted much recent interest due to their fundamental and practical significance [11]. In this work, our own efforts in utilizing the redox-active [Re₆(μ₃-Se)₈]²⁺ clusters as building blocks to construct multicluster arrays and the studies of their electronic properties are described.

The synthesis of R₄ took advantage of the substitutional lability of the coordinated nitrile molecule of [Re₆(μ₃-Se)₈(PEt₃)₅(CH₃CN)](SbF₆)₂ (R) [8] when reacted with pyridyl-based ligands (Scheme 1) [12]. Thus, reacting stoichiometric amounts of the cluster solvate R and 5,10,15,20-tetra(4-pyridyl)-21H,23H-porphine in chlorobenzene under reflux led to the displacement of the solvent ligand and the concomitant formation of the tetracluster R₄. This cluster-studded porphyrin features four units of [Re₆(μ₃-Se)₈(PEt₃)₅]²⁺ stationed onto the central pyridyl-based tetratopic ligand via N–Re coordination. It is soluble in dichloromethane, acetonitrile, and other common polar organic solvents, yielding brown-red solutions. Despite numerous attempts, X-ray-quality single crystals have so far evaded us.

Several lines of spectroscopic evidence point to the successful production of R₄. Upon formation of R₄, the ¹H signal of the coordinated nitrile of [Re₆(μ₃-Se)₈(PEt₃)₅(CH₃CN)](SbF₆)₂ (R) disappears, indicating the displacement of the previously attached solvent molecule. This is accompanied by a new signal corresponding to the α-H of the coordinated pyridyl moieties at 9.72 ppm, which is downfield shifted by 0.67 ppm from that of the free ligand. There is also a small but noticeable upfield shift of the β-H of the pyridyl moiety from 8.13 to 8.18 ppm. The pyrrole protons are slightly shifted from 8.86 to 9.04 ppm. There is a peak at –3.00 ppm that is diagnostic of the unmetallated porphyrin due to the two protons associated with the interior nitrogens. The simplicity of the ¹H NMR spectrum is consistent with the anticipated symmetric structure. Additional evidence supporting the formation of R₄ comes from the unsophisticated ³¹P NMR spectrum of R₄: Two resonances in a 4:1 relative ratio consistent with the stereochemistry of a pentaphosphine-substituted cluster core are shown at –23.6 and –27.6 ppm. R₄ was further characterized by MALDI-TOF mass spectrometry: A series of peaks corresponding to the loss of 4, 5, 6, and 7 counter-ions are observed at 2729.4, 2137.9, 1742.5, and 1459.2 a.m.u, respectively. It is clear that the cationic tetracluster maintains its structural integrity under the experimental conditions of the mass spectrometry. Finally, the composition of R₄ was confirmed by satisfactory elemental analyses (CHN).

The cluster-studded metalloporphyrins (R₄–M; M = Co²⁺, Ni²⁺, Cu²⁺, Zn²⁺) were prepared via metallation of the prefabricated R₄ (Scheme 1). In a representative synthesis, R₄ and an excess of metal acetate was dissolved in dichloromethane and stirred under reflux. The resulting mixture contained the metallated product as well as unreacted metal acetate. The solvent was removed, and the product was extracted into dichloromethane. The excess metal salt was removed by repeated washing with water.

The desired metal complexes could be formally synthesized by coupling the metallated tetrpyridylporphyrins with the cluster solvate R. However, this alternative route led only to mixtures containing both the metallated and unmetallated tetracluster porphyrins. Purification by recrystallization or extraction was unsuccessful.

The complexes are soluble in dichloromethane, acetonitrile, and other common polar organic solvents to yield purple-red solutions. ¹H and ³¹P NMR and electronic absorption spectroscopy are in agreement with the proposed structure. The effect of metallation upon the spectra is small. For example, in the ¹H NMR of R₄–Ni, the α-H of the pyridyl moiety shifts the least from 9.72 to 9.82 ppm. The pyrrole protons and the β-H of the pyridyl have more noticeable shifts of 0.19 and 0.27 ppm, respectively. The most convincing evidence of successful metallation is the disappearance of the resonance at –3.00 ppm, which corresponds to the protons bound to the interior nitrogens in R₄. The ³¹P NMR has two resonances at –24.3 and –27.6 ppm in a 4:1 ratio, indicating that the pentasubstituted cluster motif is present and remains intact. UV–vis spectroscopic studies of porphyrins are diagnostic of both the porphyrin type and the nature of the metal ions, and can be used reliably as evidence for metallation. Details of such studies will be discussed in a separate section below.

Insertion of Cu²⁺ was achieved in an analogous manner to that of R₄–Ni. Upon metallation of the tetracluster porphyrin, dramatic changes to the ¹H NMR spectrum occurred. There are three important observations: (1) The peak at –3.00 ppm is no longer present, suggesting that the copper ion has replaced the two protons; (2) the upfield resonances are noticeably broader and less well-defined; and (3) the original resonances in the aromatic region are not visible and only a broad resonance at 9.30 ppm is present. The last two observations are probably due to the influence of the paramagnetic species Cu²⁺. Little change to the ³¹P NMR spectra was observed upon metallation, however. There are two slightly broadened resonances at –24.9 and –28.8 ppm in a 4:1 ratio, mandated by the stereochemistry of a pentaphosphine-substituted cluster.

The tetracluster cobalt porphyrin complex R₄–Co can be synthesized if cobalt acetate and R₄ are refluxed in dichloromethane. As with the copper complex, the ¹H NMR of R₄–Co is significantly different from that of its unmetallated parent. Again, this can be ascribed to the influence of the paramagnetic Co²⁺ ion. Disappearance of the original upfield proton signals corroborates the insertion of the metal and the complete displacement of the interior protons. The signals due to the triethylphosphine ligands are broadened, but most interesting are again the signals in the aromatic region. The well-defined resonances associated with the pyridyl and pyrrole protons are replaced by three new resonances, but only one of them is clearly visible at 10.32 ppm. The other two are very broad, at approximately 9.0 and 12.5 ppm, respectively. The ³¹P NMR spectrum shows two characteristic signals at –25.4 and –28.7 ppm in a 4:1 ratio.

The Zn²⁺ ion was incorporated into the tetracluster array in a similar fashion. However, when dissolved in dichloromethane, the resulting solution is red-green in color, distinctly different from the purple-red color displayed by the solutions of its Cu²⁺, Co²⁺, or Ni²⁺ cognate in the same solvent. The ¹H NMR spectrum clearly indicates the insertion of Zn²⁺ because the resonance of the protons bound to the interior nitrogen at –3.00 ppm disappears. The ³¹P NMR has two resonances at –19.3 and –22.6 ppm in a 4:1 ratio, in agreement with the pentasubstituted stereochemistry.

2.2 Electronic spectroscopic studies

UV–vis spectra of 5,10,15,20-tetra(4-pyridyl)-21H,23H-porphine and R₄ are shown in Fig. 2. The electronic absorption data for R₄ at 295 K are collected in Table 1. The absorptions of the porphyrin moiety are usual: An intense Soret or B band at 426 nm and weak Q bands at 517, 553, 589, and 644 nm were observed. The B and Q bands both arise from π–π* transitions and can be explained by considering the Gouterman four-orbital model [13]. The absorption spectrum of the tetracluster R₄ shows, in addition to the porphyrin Soret and Q bands, poorly defined broad absorptions between 230 and 300 nm that are due to ligand-to-cluster charge transfer [7]. Upon coordination of the clusters, the porphyrin-based absorption peaks are red shifted. This clearly shows that significant interactions exist between these building blocks, but the fundamental properties of both components remain intact upon formation of the array.

Metallation of the porphyrin core causes changes to the symmetry of the molecule, and therefore, the π–π* transition. The effects can be seen in the alteration of the Q-band absorbance pattern, and therefore, can be used for identification of specific metallated porphyrin species. As discussed above, the addition of four metal clusters only shifts the electronic absorptions of the porphyrin moiety, but not the configurational interactions. It is therefore reasonable to expect similar electronic spectra of the cluster-modified metalloporphyrins and their unmodified parent.

The electronic spectra of R₄–Ni, R₄–Cu, and R₄–Co are shown in Fig. 3, and the absorption data are summarized in Table 2. Upon metallation the Q-band changes significantly: The unmetallated porphyrin shows four Q-band peaks. In contrast, only one Q-band is visible for its Ni²⁺, Cu²⁺, or Co²⁺ complexes. Complete metallation of the porphyrin core is thus inferred.

The electronic spectrum of the non-cluster modified zinc porphyrin (Fig. 4) has only one Q-band at 559 nm. However, in the spectrum of R₄–Zn, not only is there a slightly red-shifted Q-band at 569 nm, but a new absorption at 617 nm is also observed. The appearance of this new absorption greatly affects the color observed. Furthermore, upon dissolution in solvents of different polarity, the color of the resulting solution varies significantly. To explore the solvatochromic properties of this system, the electronic absorptions of R₄–Zn in organic solvents of varying polarity were studied and the data are collected in Table 3. Similar to other zinc metalloporphyrins, the band position of R₄–Zn exhibits a pronounced solvent dependence. The spectral shift of the intense Soret absorption is the easiest to follow. This band can shift up to 14 nm, depending on the solvent employed.

Many attempts have been made to correlate various experimental parameters with the Soret shifts. But there is really no general correlation between the red shift and dielectric constant [14], Drago's parameters [15], or Gutmann's donor numbers [16]. The behavior of the present porphyrin system is similar to literature observations [14–16]; when a limited number of solvents were used, a correlation with the parameters previously mentioned could be obtained. However, when a larger number of solvents are incorporated in the study, there is no general correlation observed. It was postulated that the limited data set reflects a minor trend of a much more complex phenomenon. Other analyses based upon combinations of parameters fit better the observed shifts [17,18]. However, if the parameters include contributions from many effects, the significance of such a correlation is diminished. Clearly, more data are required before any specific contribution to the spectral shifts can be understood.

2.3 Electrochemical studies

Cluster oxidation potentials for the compounds under the present study are presented in Table 4. A dicluster (R₂) bridged by 4,4′-dipyridyl [12] and a tricluster (R₃) bridged by 2,4,6-tri(4-pyridyl)-1,3,5-triazine [19] are included in the study in order to correlate the number of electrons involved in the redox events to the number of clusters in the molecular arrays. The potentials represent the half-wave potential from cyclic voltammetry (CV) experiments in acetonitrile at a Pt disc electrode (100 mV s^–1), referenced to the ferrocenium/ferrocene (Fc⁺/Fc) redox couple (E° = 0.46 V vs. SCE; 0.70 V vs. NHE). Oxidation potentials are not significantly changed (ca. 0.70 V) in going from the free cluster solvate [Re₆(μ₃-Se)₈(PEt₃)₅(CH₃CN)](SbF₆)₂ (R) to assemblies of up to 4 clusters per molecules (i.e. the R₄ series), and all materials give highly reversible CVs on this time scale with ΔE_p values of ca. 70 mV, which is comparable to those observed for ferrocene in this cell, under the same experimental conditions.

In order to obtain high-resolution and high-signal-to-background voltammograms and to extract absolute coulometric information for these materials, a thin-layer cell electrode (TLE) was developed (see Section 4 for details) [20] and utilized to determine whether these materials exhibit coupled or uncoupled oxidations.

Thin-layer cyclic voltammetry (TLCV) and coulometry measurements were made in the same solution cell as the standard voltammetric measurements, exhaustively oxidizing the analytes in the confined volume defined by the TLE. At sufficiently slow scan rates (v, developed below), the current measured in a voltammetric experiment has no dependence on mass transfer in the cell, and it is therefore not necessary to know the diffusion coefficient (D₀) of the redox species in order to determine the number of electrons transferred per redox event (n) [21]. While the peak current in such a voltammogram is directly proportional to v, the total amount of charge passed (Q) in exhaustively oxidizing/reducing the contents of such a thin-layer is independent of v and can be related to n by the following formalism(Eq. 1)

TLCVs (1 mV s^–1) of selected subject compounds (R, R₂, R₃, and R₄–Zn), containing between one and four clusters per molecule, are presented in Fig. 5. The current axis in this plot has been normalized to the number of molecules in the confined volume, and on this scale, the peak areas are proportional to the amount of charge passed (number of electrons) per molecule. Q/C values were generally calculated by integrating the anodic peak in the original current vs. voltage plots and dividing by scan rate and concentration. These values are listed in Table 4 for all materials, along with the calibrated cell volume for that measurement. The number of electrons transferred per molecule for the cluster oxidation event, n, was then calculated using Eq. 1 and is also reported in Table 4. It was determined that for every compound studied, all clusters on a given molecule oxidize at the same potential, without influencing each other (i.e. these are ‘non-interacting’ redox couples), yielding a measured value of n that is approximately equivalent to the number of clusters per molecule. Based on this determination of n, diffusion coefficients were calculated to be approximately 5 × 10^–7 (cm² s^–1) from standard voltammograms for these materials, placing this determination (at v = 1 mV s^–1) within the valid working range of the TLE.

For the R₄ series, R₄ and R₄–Co give nearly identical CVs to that shown for R₄–Zn in Fig. 5. R₄–Ni and R₄–Cu also exhibit a 4-electron cluster oxidation, but additional oxidations are observed in this potential window. These extra peaks are attributable to core metal oxidations, which are not reversible on this very slow (1 mV s^–1) time scale. The return sweep (cathodic or top peak) of the cluster oxidation was used for cluster coulometry measurements on these materials (4 e^–), as there was no interference from the irreversible core metal oxidation. ΔE_p values (<100 mV) for all materials are comparable to those of ferrocene and are the result of uncompensated cell resistance within the TLE capillary.

At a faster time scale (10 mV s^–1), where the voltammetric response resides between thin-layer and diffusion-limited behavior for molecules of this size, the core metal oxidations observed for R₄–Cu and R₄–Ni are reversible, as shown in Fig. 6. At this sweep rate, absolute coulometric measurements are invalid, yet superior resolution and Faradaic current response is achieved over a typical Pt disc electrode, where core metal oxidations for R₄–Cu were not detectable. Relative peak areas for separate redox processes of the same molecule, however, are still expected to be proportional to the number of electrons passed per redox event. Using peak fitting, area ratios were determined to be ca. 1:1:4 (core/core/cluster) for R₄–Cu and 1:2 (core/cluster) for R₄–Ni. Based on the previous determination of a 4-electron oxidation for the cluster peak of both species at slow scan rates, the stoichiometry for core metal oxidation can be inferred: two 1-electron oxidations are obtained for R₄–Cu (Cu^II → Cu^III → Cu^IV) and one 2-electron oxidation is obtained for R₄–Ni (Ni^II → Ni^IV), which is in agreement with other coulometric studies on Cu and Ni-cored metalloporphyrins [22]. It appears that the multiple cluster oxidations on these molecules are not influenced by each other or by metal core oxidations, and remain uncoupled. Additionally, these R₄-series compounds were observed to have varying reduction energies, attributable to the porphyrin core or metal, which have no influence on the cluster oxidation.

4.1 Materials and methods

The cluster solvate [Re₆(μ₃-Se)₈(PEt₃)₅(CH₃CN)](SbF₆)₂ was prepared according to published procedures [8]. 5,10,15,20-Tetra(4-pyridyl)-21H, 23H-porphine, (n-Bu₄N)PF₆, and other reagents were of commercial origin and were used as received. NMR (¹H and ³¹P) spectra were recorded on a Varian Unity 300 spectrometer in CD₂Cl₂ solutions. Chemical shifts of ³¹P NMR spectra were referenced to 85% H₃PO₄ (0.0 ppm, with negative values meaning upfield). NMR resonance multiplicities are designated: s (singlet), d (doublet), t (triplet), q (quartet), and m (multiplet). Mass spectrometry data were collected on a Bruker Reflex III MALDI-TOF instrument. A conventional N₂ laser (337 nm) was used in the experiments with variable attenuation (typical range: 40–60%). Dithranol was used as a matrix, which was dissolved in CH₂Cl₂ (0.1 M solution). This matrix solution of dithranol was mixed with a 0.1 M CH₂Cl₂ solution of the sample in a 5:1 ratio, and 1 μl of this mixed solution was deposited on the MALDI plate. Electronic absorption spectra in dichloromethane solutions were recorded on a Perkin Elmer Lambda 10 spectrophotometer. Elemental analyses (CHN) were performed by Desert Analytics Laboratory, Tucson, AZ.

4.2 Synthesis of R₄

A mixture of [Re₆(μ₃-Se)₈(PEt₃)₅(CH₃CN)](SbF₆)₂ (R, 274 mg, 0.0962 mmol) and 5,10,15,20-tetra(4-pyridyl)-21H, 23H-porphine (10.0 mg, 0.0323 mmol) in 30 ml of chlorobenzene was stirred under reflux for 72 h. An orange-red residue was obtained upon removal of solvent. The residue was dissolved in 5 ml of dichloromethane, and the solution was triturated with ether to give an orange-colored powder. The product was obtained as dark red crystals after recrystallization from ether and dichloromethane. Yield: 157 mg (82.0%). ¹H NMR: –3.00(s), 1.05–1.30(m), 2.10–2.40(m), 8.18(d), 9.04(s), 9.72(d). ³¹P NMR: –23.6(s), –27.6(s). MS: m/z 2729.4 [M-4SbF₆^–]⁴⁺, 2137.9 [M-5SbF₆^–]⁵⁺, 1742.5 [M-6SbF₆^–]⁶⁺, 1459.2 [M-7SbF₆^–]⁷⁺. Anal. Calc. for Re₂₄Se₃₂P₂₀N₈C₁₆₀H₃₂₄Sb₈F₄₈: C, 16.19; H, 2.73; N, 0.94. Found: C, 16.50; H, 2.75; N, 0.98.

4.3 Synthesis of R₄–Ni

A mixture of R₄ (50.0 mg, 0.004 mmol) and Ni(C₂H₃O₂)₂·4 H₂O (60.0 mg, 0.241 mmol) in 30 ml of chloroform was stirred under reflux for 48 h. The red residue obtained upon removal of the solvent was extracted using dichloromethane and water. The organic phase was dried over MgSO₄. A purple–red residue was obtained upon removal of the solvent. Recrystallization from ether and dichloromethane afforded purple-red crystals as the product. Yield: 45 .0 mg (90.3%). ¹H NMR: 1.00–1.30(m), 2.00–2.40(m), 7.91(d), 8.85(s), 9.62(d). ³¹P NMR: –19.5(s), –22.8(s). Anal. Calc. for Re₂₄Se₃₂P₂₀N₈C₁₆₀H₃₂₄NiSb₈F₄₈: C, 16.13; H, 2.72; N, 0.94. Found: C, 16.48; H, 2.89; N, 1.10.

4.4 Synthesis of R₄–Cu

R₄–Cu was obtained as a purple powder in a similar manner to R₄–Ni except that Cu(C₂H₃O₂)₂·H₂O was used in place of Ni(C₂H₃O₂)₂·4 H₂O. Yield: 0.046 g (92%). ¹H NMR: 0.80–1.10(m), 1.85–2.25(m), 9.30(bs). ³¹P NMR: –24.9(s), –28.3(s). Anal. Calc. for Re₂₄Se₃₂P₂₀N₈C₁₆₀H₃₂₄CuSb₈F₄₈: C, 16.12; H, 2.72; N, 0.94. Found: C, 16.42; H, 2.90; N, 1.07.

4.5 Synthesis of R₄–Co

R₄–Co was obtained as a purple powder in a similar manner to R₄–Ni except that Co(C₂H₃O₂)₂∙4H₂O was used in place of Ni(C₂H₃O₂)₂·4 H₂O. Yield: 0.046 g (92%). ¹H NMR: 0.55–1.33(m, 180), 1.85–2.50(m, 120), 9.03(bs), 10.32(bs), 12.49(bs). ³¹P NMR: –25.4(s), –28.7(s). Anal. Calc. for Re₂₄Se₃₂P₂₀N₈C₁₆₀H₃₂₄CoSb₈F₄₈: C, 16.13; H, 2.72; N, 0.94. Found: C, 16.18; H, 2.91; N, 1.07.

4.6 Synthesis of R₄–Zn

R₄–Zn was obtained as a green powder in a similar manner to R₄–Ni except that Zn(C₂H₃O₂)₂∙2H₂O was used in place of Ni(C₂H₃O₂)₂·4 H₂O. Yield: 0.047 g (94%). ¹H NMR: 1.00–1.30(m), 2.04–2.40(m), 8.09(d), 8.95(s), 9.67(d). ³¹P NMR: –19.3(s), –22.6(s). Anal. Calc. for Re₂₄Se₃₂P₂₀N₈C₁₆₀H₃₂₄ZnSb₈F₄₈: C, 16.12; H, 2.72; N, 0.94. Found: C, 16.47; H, 2.69; N, 1.11.

4.7 Cyclic voltammetry

CVs were acquired with an EG&G Instruments 283 potentiostat, using a 500 μm diameter Pt working electrode, a freshly prepared Ag/AgCl pseudoreference electrode, and a Pt counter electrode, in N₂-degassed solutions containing (n-Bu₄N)PF₆ as supporting electrolyte. All solvents were purified by distillation and stored over activated alumina to remove trace water. Prior to preparing analyte solutions, the distilled solvent was run through a fresh column of activated alumina. Unless otherwise noted, an analyte concentration of ca. 1 mM, a scan rate (v) of 100 mV s^–1, and an electrolyte concentration of 0.1 M was used. A custom glass solution cell with fritted openings was used to stabilize the electrodes and minimize permeation of oxygen. The working electrode was cleaned between each experiment by polishing with 0.3 μm alumina paste for 1 min, followed by copious solvent rinses. The reference electrode was prepared by plating a thin-layer of AgCl onto a Ag wire from a 1 M KCl solution. After each voltammetric experiment, ferrocene was added (ca. 1 mM), an additional voltammogram was recorded, and the potential axis was calibrated against the formal potential for the Fc⁺/Fc redox couple. Reported redox potentials represent the half-wave potential in these voltammograms.

4.8 Thin-layer cyclic voltammetry/coulometry

A thin-layer cell electrode (TLE) was designed and fabricated in order to perform exhaustive electrolysis on analyte solutions, yielding coulometric information for the relevant redox event. A schematic drawing of this electrode is presented in Fig. 7 [20]. The electrode was designed such that these measurements could be made using the above setup for standard cyclic voltammetry, where the working electrode is simply replaced with the TLE. The entire electrode assembly was placed in the solution cell, and the distal end was submerged 2–5 mm into the analyte solution, causing the solution to reproducibly fill the entire cavity created between the Pt wire and the small capillary due to capillary action. Relatively large electrolyte concentrations (0.3 M) were necessary to minimize the potential drop in the TLE cavity. TLCVs were collected at 1 mV s (where there was insignificant diffusion dependence on the current), and the cell volume was first calibrated using a 1 mM standard solution of ferrocene in acetonitrile. Following the confirmation of a one-electron oxidation for the free cluster monomer R, standard solutions of R were used for volume calibration prior to analyzing multicluster molecules. Solvent background (minimal) was subtracted from all charge measurements.