Isolation of chloride- and hydride-bridged tri-iron and -zinc clusters in a tris(β-oxo-δ-diimine) cyclophane ligand

Abstract

A cyclophane ligand (H₆L) bearing three β-oxo-δ-diimine arms and the corresponding tri-iron and -zinc complexes bridged by either chlorides, viz. Fe₃Cl₃(H₃L) (1) and Zn₃Cl₃(H₃L) (2), or hydrides, viz. Fe₃H₃(H₃L) (3), Zn₃H₃(H₃L) (4), were synthesized and characterized. 1 adopts a chair-shaped C_3v-symmetric [Fe₃(μ-Cl)₃]³⁺ cluster wherein only one hemisphere of the ligand is metallated and the other three ketoimine sites remain protonated as evidenced by single crystal X-ray diffraction and vibrational and NMR spectroscopic analyses. 3 and 4 were synthesized by substitution of the bridging chlorides in 1 and 2 using KBEt₃H and are accessed with retention of the three protonated ketoimine sites.

Graphical Abstract

Introduction

Active sites of the enzymes that catalyse multielectron redox reactions utilize metal cluster cofactors in conjunction with secondary coordination sphere interactions (e.g., hydrogen bonding interactions) to facilitate bond activation.^1–7 As an initial approach to understanding and replicating such reactivity, several groups have employed ligands to template the spatial and electronic environment of multiple metal centres, thereby accessing polynuclear metal complexes with a priori control.^8–14 As part of those efforts, our group has reported a number of trimetallic complexes supported by a tris(β-diketimine) cyclophane (H₃L) for which the metal ion type and the identity of the bridging ligands can be varied.^14–21 Members of this trimetallic cyclophanate family are capable of supporting low-valent high spin metal centres, can readily undergo exchange of bridging ligands, and can reductively activated small molecule. One challenge, however, is that the internal space within the cavity limits substrate binding and product egress and these compounds are limited to trimetallic species. Expansion of the internal void space of the cyclophane ligand by utilizing linking arms that are longer than acetylacetone as well as arms capable of coordinating more than one metal centre provides a simple and complementary route to address these issues.

Gerlach and Holm first reported the use of deprotonated β-ketoiminates as ligands in which the ligand sterics preclude polymer formation and allow access to four-coordinate metal species,²² and reports by others have since been communicated.^23,24 As an extension, β-oxo-δ-diimine can be accessed by Schiff base condensation of a triketone with amines. Once deprotonated, the corresponding β-oxo-δ-diiminate can be used to access dimetallic complexes wherein each metal is held within a monoanionic N,O-chelate.^25,26 In addition to the dimetallic species, others have reported selective and exclusive monometalation of these β-oxo-δ-diimine; the resultant complexes retain the N–H proton on one half of the ligand.^25,27,28 We reasoned then that employing β-oxo-δ-diimine as arms in a cyclophane could simultaneously allow access to hexanuclear metal complexes under one metalation condition or trimetallic complexes by minor changes to the synthetic approach. For the latter, the internal cavity of one hemisphere of the complex would be decorated with N–H protons, providing both a large internal cavity for substrate access, but also template secondary coordination sphere hydrogen bond donors. To our knowledge, there are only a few examples of dimetallic complexes in which the secondary coordination sphere are incorporated, and no examples for higher nuclearity compounds. Herein, we report the synthesis and characterization of a tris(β-oxo-δ-diimine) bis(1,3,5-triethylbenzene) cyclophane and its tri-iron and tri-zinc complexes with bridging chlorides or hydrides (Scheme 1). These metal complexes access the trimetallic species with potential secondary coordination sphere hydrogen bond donors within the internal cavity of the cyclophane.

Scheme 1.

Synthesis of tri- or hexa-nucleating ligands and coordination complexes of H₆L.

Experimental section

Synthetic procedures

General Considerations

All manipulations except ligand synthesis were performed inside an N₂-filled Innovative Technologies glovebox. Tetrahydrofuran (THF), benzene, toluene, and n-hexane were purchased from Sigma-Aldrich, dried under Ar using a solvent purification system (Innovative Technologies), transferred to the glovebox, and stored over activated 3A molecular sieves. Benzene-d₆, toluene-d₈, and THF-d₈ (Cambridge Isotope Laboratories) were distilled over CaH₂ or with Na/benzophenone and then degassed and stored over 3 A molecular sieves. ¹H Nuclear Magnetic Resonance (¹H NMR) spectra were recorded on a 500 MHz Varian Inova spectrometer or a 300 MHz Mercury spectrometer equipped with a three-channel 5 mm indirect detection probe with z-axis gradients. Chemical shifts were reported in δ (ppm) and were referenced to solvent resonances: δH = 7.26 ppm (CDCl₃), 7.16 ppm (C₆D₆), 2.09 ppm (C₇D₈), and 3.58 ppm (C₄D₈O). Fourier Transform Infrared (FT-IR) spectra were recorded as solids on a Thermo Fisher iS5 instrument equipped with an ATR diamond crystal stage using OMNIC software package. 2,4,6-heptatrione was synthesized following the literature procedure.²⁹

H₆L

A methanol (60 mL) solution of 2,4,6-heptatrion (3.40 g, 23.9 mmol) and 1,3,5-triaminomethyl-2,4,6-triethylbenzene (5.15 g, 20.6 mmol) was heated at 50 °C for 6 h. Resulting beige solid was collected by filtration and dried under reduced pressure (yield = 5.00 g, 77%). ¹H NMR (300 MHz, CDCl₃, 298 K): δ 9.56 (s, br, 6H), 4.72 (s, 6H), 4.29 (s, 12H), 2.73 (d, J = 7.1 Hz, 12H), 2.02 (s, 18H), 1.17 (t, J = 6.9 Hz, 18H). ¹³C NMR (75 MHz, CDCl₃, 298 K): δ 189.6, 157.3, 144.1, 131.6, 95.7, 41.2, 23.0, 20.0, 16.3. IR (cm⁻¹): 3060, 2931, 1620, 1560, 1427, 1296, 1080, 790, 710. ESI-MS calcd for C₅₁H₇₃N₆O₃ [M+H]⁺ 817.5739, found 817.5725.

Fe₃Cl₃(H₃L) (1)

A THF solution (12 mL) of H₆L (1.16 g) and LiNⁱPr₂ (0.470 g, 3.0 equiv.) was stirred at r.t. for 5–10 min. After the formation of a green solution, the solvent was removed under vacuum at 50 °C. The residue was suspended in chlorobenzene (24 mL) and FeCl₂ (0.555 g, 3.1 equiv.) added. The reaction mixture was stirred at 90 °C overnight, then filtered, and filtrate dried in vacuo at 50 °C to yield analytically pure orange solid (1.43 g, 92%). Crystals suitable for diffraction were grown by cooling chlorobenzene solution from 90 °C to r.t. C₅₁H₆₉Cl₃Fe₃N₆O₃·0.5C₆H₅Cl (calcd): C, 56.48 (56.68); H, 6.33 (6.30); N, 7.04 (7.34). ¹H NMR (500 MHz, CDCl₃, 298 K): δ 141.31, 77.17, −10.05, −11.81, −13.38, −13.58, −18.52, −32.72, −33.59. IR (cm⁻¹): 3280, 2963, 1599, 1494, 1324, 1066, 781. μ_eff (CDCl₃, 298 K) = 5.6 μ_B. UV–Vis (THF, 25 °C, M⁻¹cm⁻¹): ϵ₅₃₄=644

Zn₃Cl₃(H₃L) (2)

A THF solution (4 mL) of H₆L (323 mg) and LDA (134 mg, 3.1 equiv.) was stirred at r.t. for 5–10 min. After the formation of a green solution, the solvent was removed under vacuum at 50 °C. Chlorobenzene suspension (8 mL) of the deprotonated ligand and ZnCl₂ (166 mg, 3.1 equiv.) were stirred at 90 °C overnight. After the removal of insoluble salt by filtration, the solution was evaporated under vacuum at 50 °C to afford 2 as a yellow solid (362 mg, 82%). C₅₁H₆₉Cl-₃N₆O₃Zn₃ (calcd): C, 54.33 (54.86); H, 6.11 (6.23); N, 7.16 (7.53). ¹H NMR (500 MHz, CDCl₃, 298 K): δ 9.53 (s, 3H), 4.55 (d, 6H), 4.33 (d, 12H), 2.68 (dd, 12H), 2.03 (d, 18H), 1.18 (dt, 18H). IR (cm⁻¹): 3244, 2981, 1599, 1503, 1397, 1333, 1074, 782.

Fe₃H₃(H₃L) (3)

To a toluene solution (1 mL) of 1 (33.5 mg) was added a toluene solution (1 mL) of KBEt₃H (12.5 mg, 2.94 equiv.) at −35 °C and the mixture stirred overnight. Then, the reaction was filtered through toluene-rinsed Celite, and the filtrate was evaporated under vacuum to yield 3 as a red-brown solid (28.8 mg, 95%). C₅₁H₇₂Fe₃N₆O₃·0.5C₄H₁₀O (calcd): C, 61.76 (62.30); H, 7.67 (7.60); N, 7.83 (8.23). ¹H NMR (500 MHz, C₆D₆, 298 K): δ 58.34, 43.37, 15.71, 4.41, −0.93, −3.33, −5.24, −6.36, −7.85, −44.24. IR (cm⁻¹): 3295, 2963, 1612, 1501, 1398, 1337, 1067, 787. UV–Vis (THF, 25 °C, M⁻¹cm⁻¹): ϵ₅₃₃=1410, ϵ₅₈₂=857.

Zn₃H₃(H₃L) (4)

The procedure is as described above for 3 employing the trizinc complex 2 instead of the triiron congener and yields 4 as a yellow solid in good yield (78.6 mg, 78%). C₅₁H₇₂N₆O₃Zn₃·C₆H₆ (calcd): C, 62.66 (62.73); H, 7.37 (7.20); N, 7.26 (7.70). ¹H NMR (500 MHz, C₆D₆, 298 K): δ 8.38 (s, 2H), 5.01 (d, 6H), 4.23 (d, 12H), 3.38 (s, 2H), 2.56 (q, 6H), 2.40 (q, 6H), 1.87 (s, 9H), 1.78 (s, 9H), 1.09 (t, 9H), 1.02 (t, 9H). IR (cm⁻¹): 2869, 1612, 1507, 1405, 1341, 1068, 782.

Results and Discussion

The target ligand (H₆L) comprising three β-oxo-δ-diimine sites was synthesized in 77% yield from the condensation of 2,4,6-heptatrione with the previously used 1,3,5-triaminomethyl-2,4,6-triethylbenzene (Scheme 1). As anticipated from our prior work with the tris(β-diketimine) cyclophane, this ligand is a solution averaged D_3h symmetric compound at ambient temperature with only six resonances observed in the ¹H-NMR spectrum (Figure S1). The β-oxo-δ-diimine has been previously reported to adopt the keto-di(enamine) tautomer; similarly, the integral singlet α-hydrogens at 4.72 ppm and that of the downfield shifted amine protons at 9.56 ppm are indicative of the keto-diamine tautomer in H₆L. Notably, a weak absorbance is observed at 3060 cm⁻¹ in IR spectra of solid samples of H₆L is assigned to N–H stretching mode (Figure S2); these data together with the NMR resonance at 9.56 ppm are the diagnostic markers for successful deprotonation at a given site.

Deprotonation of H₆L in THF with 3 equiv. LiNⁱPr₂ (LDA) results in a reduction of the molecular symmetry and, likely, a distribution of slowly equilibrating conformers of Li₃(H₃L) species in solution. For example, the ¹H-NMR resonances corresponding to the backbone C–H protons of the β-oxo-δ-diimine arms and the aminomethyl protons broaden from well-resolved singlets to multiplets (Figure S3). Loss of one bridging proton in each arm also affords observable changes in the IR spectrum; the vibration in the C=O/N range at 1560 cm⁻¹ undergoes a hypsochromic shift to 1577 cm⁻¹ and loss of the vibration associated with the N–H protons (Figure S4). In contrast to the deprotonation with 3 equiv. LDA, ¹H-NMR resonances in spectra from reactions of H₆L with 6 equiv. LDA are broad and through bond couplings are unresolved. For both the proposed Li₃(H₃L) and Li₆L, the NMR spectrum of the purported Li₆L could reflect a distribution of slowly equilibrating isomers in solution, which is consistent with the decrease or absence of a discernible IR absorption for the N–H vibration in the partially deprotonated ligand. In addition, we observe loss in absorptions between 1560–1620 cm⁻¹ (Figure S5, Figure S6). The changes to the IR spectrum upon deprotonation are reminiscent of our observations for deprotonation of our tris(β-diketimine) cyclophane; we observe an almost complete loss in the absorption at ~1617 cm⁻¹ in the free base ligand upon complete deprotonation. Attempts to access the hexametalllic complexes have been unsuccessful with the reaction of the putative Li₆L with excess FeCl₂ resulting in intractable products that are insoluble in all solvents tested. Similar results were obtained for a survey of other metalating agents in which metal ion type and counteranion were varied. In contrast, treatment of a chlorobenzene suspension of Li₃(H₃L) with 3 equiv. FeCl₂ or ZnCl₂ readily affords the orange-red tri-iron (Fe₃(μ-Cl)₃(H₃L), 1) or yellow tri-zinc (Zn₃(μ-Cl)₃(H₃L), 2) complexes, respectively. The paramagnetic ¹H-NMR spectrum of 1 comprises ten resonances between 144.26 and -36.04 ppm (Figure S7). Similarly, we observe five doublets for the aliphatic protons and a singlet for N–H protons in ¹H-NMR spectra of 2 (Figure S9). Taken together, we conclude that 1 and 2 both have C_3v-symmetry in solution, hinting at selective metalation of only one hemisphere of the ligand and μ-chloride donors.^17,19 Notably, a new absorption is observed in IR spectra of 1 and 2 in the O/N–H region of the IR spectrum (viz. 3280 cm⁻¹ for 1 and 3244 cm⁻¹ for 2) as compared to spectra of Li₃(H₃L) (Figure S8, Figure S10). These absorptions support retention of three N–H protons in the isolated compounds, with incorporation of the d-block metal ions resulting in one preferred conformer and loss of the fluxionality observed for the Li derivatives. In addition, metalation results in a hypsochromic shift of the absorption at 1577 cm⁻¹ in the Li₃(H₃L) to 1599 cm⁻¹ in 1 and 2. Taken together, we surmise that 1 and 2 are isostructural and are C_3v symmetric.

The solid state structure of 1 determined by single crystal x-ray diffraction qualitatively corroborate our IR and NMR analysis: the average stoichiometry is Fe₃Cl₃(H₃L).^‡ In this structure solution, the three iron centres are positionally disordered over the six β-ketoiminate sites with occupancies ranging from 0.46–0.54 (Figure 1). The three bridging chlorides are on same plane of carbonyl oxygens (rms distance of 0.160 Å for 3Cl atoms from the O₃ plane); the elongated thermal ellipsoids of the O and Cl atoms are consistent with the Fe positional disorder and site averaging of the iron atom in one half of each arm. The average result together with the solution C_3v symmetry deduced from our ¹H NMR data affords a chair-shaped [Fe₃Cl₃]³⁺ cluster within the internal cavity of the cyclophane. This chair motif has substantial precedent in d-block chemistry with analogous [M₃Cl₃]ⁿ⁺-type clusters have been reported for rhodium,^30–32 platinum,^33,34 copper,^35,36 and palladium.³⁷ However, to our knowledge based on search of the Cambridge Structural Database, this complex represents the first case of crystallographic structure of chair-shaped triiron cluster bridged by chlorides. The average Fe–Fe distance in the cluster is 3.759(1) Å, which is the second largest value from chloride bridged Fe(II) compound reported so far.³⁸ The largest Fe–Fe distance is found from the parallel diiron β-diiminato complex, two iminate group of which are separated by xanthene moiety.³⁹ The Fe–Fe lengths and Fe–Cl–Fe angles (ϑ) are well correlated by the following equation: d_Fe–Fe = 2d_Fe–CIsin(θ/2), giving average bond length of bridging Fe–Cl as 2.392(5) Å (Figure S11). Based on the angle dependence on coupling proposed by Goodenough, Kanamori, and Anderson, we anticipate that the iron centres in Fe₃Cl₃(H₃L) are antiferromagnetically coupled as are our previous trimetallic complex.^14,40–43

Figure 1.

Single-crystal structure of 1 at 50% thermal ellipsoid (top-left). Bonds with disordered iron atoms are displayed as dashed line. The chair-shaped [Fe₃Cl₃]³⁺ cluster occupies the lower hemisphere of the ligand with each iron atom adopting a trigonal pyramidal geometry (top-right). Expanded view of the donor atoms for each metal centre in the cluster highlight the local coordination geometry (bottom-left). Connolly surface map drawn with no diameter of probe provide insight into the access to the internal cavity (bottom-right). H atoms and solvent molecules have been omitted for clarity. C, N, O, Fe, and Cl are depicted as grey, blue, red, orange, and green ellipsoid, respectively.

There are differences for coordinating halides between 1 and our previous β-diketiminate complex with [Fe₃Br₃]³⁺ cluster.¹⁴ For our previously reported trinucleating ligand, the ancillary ligands (e.g., chloride) lie in the M₃ plane, whereas the similarly μ-halides in 1 and 2 are out of the M₃ plane. Such a change in the structure is expected based on the preferred bonding interactions for halides in metal halide clusters. Such an assertion is supported by the prevalence of self-assembled chair-like clusters as compared to the dearth of analogous planar clusters. Thus, the larger cavity size of H₆L as compared to the more compressed tris(β-diketiminine) cyclophane allows the M₃X₃ cluster to adopt the preferred chair orientation. Arguably, the ligand fields differ slightly between the O,N-chelate afforded by H₆L as compared to the N,N-chelate in our previous work; given precedent, such changes are not expected to be the driving force for loss of planarity of the M₃X₃ unit.

Each iron centre in 1 adopts a pseudotrigonal pyramidal coordination geometry with τ₄ values between 0.72 and 0.75 (τ₄=0.85 for trigonal pyramidal).⁴⁴ Here, the trigonal plane is defined by the two chlorides and the nitrogen donor coordinated to a given Fe centre with the sum of these L–M–L angles being 359°. The plane is composed of two large N–Fe–Cl bond angles, which give rise to relatively small τ₄, and small Cl–Fe–Cl angle (103.29(4)°). Given the similarity of this chair motif to the synthetic 3Fe-4S clusters, we note that these Cl–Fe–Cl angles are comparable to the average S–Fe–S angle of 103.64(2)° in synthetic examples.⁸ Future directions aim to leverage these structural similarities to access high nuclearity compounds.

With the trichloride complexes in hand, we then sought to derivatize these complexes as previously demonstrated for the trimetallic complexes of our tris(β-diketiminate) cyclophane. Reaction of 1 or 2 with 3 equiv. of KEt₃BH at −35 °C effected the desired substitution, affording the triiron (3) or trizinc (4) complexes of [H₃L]³⁻ in excellent yield (95% or 89%, respectively). Unlike the reported result for monometallic β-diketiminatoiron(II) complexes, we do not observe transmetalation and adventitious formation of iron-alkyl species.^17,45 A total of ten resonances are observed in ¹H-NMR spectra of 3, consistent with the expected retention of C_3v molecular symmetry (Figure S12). Retention of the enamine protons within the cavity is evidenced by the IR absorption at 3295 cm⁻¹, which we assign as the N-H stretching mode based on deuteration experiments (Figure S13). Further corroboration for retention of the protonated state of the unmetalated enamines is provided by analysis of 4. NMR spectra recorded on samples of 4 comprise a sharp resonance at 3.38 ppm assigned to the Zn–H as well as ten resonances corresponding to the ligand protons and one singlet at 8.39 ppm for amine protons (Figure S14). We note for 4, however, that the integrals for the hydride and amine proton peak was smaller than expected, and the absorption corresponding to the N-H stretching frequency was not well-resolved from baseline in IR spectra (Figure S15). Both iron complexes showed two irreversible reduction events (viz. −2.25 and −2.58 V vs. Fc/Fc⁺ for 1 and −1.66 and −2.33 V vs. Fc/Fc⁺ for 3) from cyclic voltammograms measured in THF and using ⁿBu₄NPF₆ as the electrolyte (Figure S16, Figure S17). As we will mention about the stability of the complex below, data suggest that these complexes are significantly more labile as compared to our tris(β-diketiminate) cyclophane complexes, and this lack of stability likely contributes to irreversibility of these redox processes.

The stability of 1 was significantly affected by the presence of added chloride, solvent, and temperature. Despite the initial synthesis in an aromatic solvent, dissolution of isolated 1 in toluene or benzene followed by heating at 90 °C over the course of 24 h results in decomposition of the complex to the free base ligand as well as species with apparent C_2v symmetry, which are consistent with that expected for partially metalated complexes (e.g., di- and monometallic complexes). Surprisingly, demetalation was also apparent in the solid state even at −35 °C over the course of several weeks as evidenced by NMR and IR. The stability of 1 could be enhanced if THF was used as the solvent instead of toluene or benzene, if the bridging ligand was exchanged from chloride to hydride as in 3, or an exogenous halide source was added. The instability in aromatic solvents and enhanced stability in THF suggests the chloride donors in 1 may sample modes in which solvent can coordinate to the iron centres, increasing the metal coordination number and stabilizing the metal centres. On distinct difference between complexes 1–4 and our prior trimetallic species is the presence of the three reasonably acidic enamine protons in the upper hemisphere of the ligand. We posit then that one possible mechanism for demetalation may be triggered by proton transfer from the N–H to the coordinating carbonyl O atom, thereby labilizing the Fe centre. Alternatively, one can envision a pathway in which formal loss of HCl leads to a dianionic chelate for one metal centre; the liberated HCl may facilitate protonation and demetalation of remaining 1 in solution.

Consistent with hydridic reactivity expected for 3, the triiron trihydride reaction with chloroform at ambient temperature in near quantitative (based on ¹H-NMR spectra) to generate 1, and a mixture of dichloromethane and chloromethane. Contrastingly, 1 is stable indefinitely in chloroform. As might be expected based on the facile demetallation of 1, this complex is unstable to water affording the free ligand, whereas 3 reacts readily with water to afford a mixture of a new C_3v symmetric species with a strong IR absorption consistent with O–H ligands as well as the protonated ligand (Figure S18, Figure S19). This reactivity contrasts the reactivity of the trimetallic trihydride complexes reported previously by our group; those compounds demonstrated an unusually slow reactivity towards water as well as highly specific reactivity towards CO₂. Here, we surmise that the greater accessibility of the cavity as well as the hydrogen bonding donors allows for more facile approach of water, enhancing reactivity towards proton sources.

Conclusions

In summary, we report the synthesis and metalation with iron(II) and zinc(II) chloride of a tris(β-oxo-δ-diimine) cyclophane. The resultant trimetallic compounds contain chair-shaped [Fe₃Cl₃]³⁺ clusters, in which each metal centre is ligated by a monoanionic N,O chelate and two halides and with retention of one N–H proton per ligand arm. Facile derivatization of the chloride complexes to the corresponding hydride bridged compounds was readily affected using KEt₃BH, generating complexes bearing both hydridic and protic groups. In addition, complexes 1–4 represent rare examples of the complexes in which secondary coordination sphere interactions are templated in a designed multimetallic system, and the first example for trimetallic systems to our knowledge. The possible redox and H⁺/H⁻ transfer reactivities are the subject of future and ongoing work.