Cooperative C-H Bond Activation by a Low-Spin d6 Iron-Aluminum Complex.

The reactions of transition metal complexes underpin numerous synthetic processes and catalytic transformations. Typically, this reactivity involves the participation of empty and filled molecular orbitals centered on the transition metal. Kinetically stabilized species, such as octahedral low-spin d6 transition metal complexes, are not expected to participate directly in these reactions. However, novel approaches that exploit metal-ligand cooperativity offer an opportunity to challenge these preconceptions. Here, we show that inclusion of an aluminum-based ligand into the coordination sphere of neutral low-spin d6 iron complex leads to unexpected reactivity. Complexes featuring an unsupported Fe-Al bond are capable of the intermolecular C-H bond activation of pyridines. Mechanistic analysis suggests that C-H activation proceeds through a reductive deprotonation in which the two metal centers (Fe and Al) act like a frustrated Lewis pair. The key to this behavior is a ground state destabilization of the d6 iron complex, brought about by the inclusion of the electropositive aluminum-based ligand. These findings have immediate implications for the design of reagents and catalysts based on first-row transition metals.

Introduction

The concept of metal–ligand cooperativity has greatly enriched the chemistry and catalytic applications of transition metal complexes.[1,2] Cooperative strategies that take advantage of the Lewis acidic nature of a transition metal center in combination with a Lewis basic ligand have become established features, frequently employed in the design of novel catalytic systems. Equally attractive but less common are transition metal complexes that bear a Lewis acidic functionality in the ligand.[3−5] These new design principles offer an opportunity to overturn existing paradigms in transition metal chemistry. This is particularly important for applications of inexpensive and sustainable first-row transition metals (e.g., Fe).[6−9] Despite the exciting opportunities in this field, many of the emerging reactions that involve metal–ligand cooperativity still proceed through established mechanisms, such as oxidative addition. For example, transition metal complexes bearing Lewis acidic ligands (based on B or Al) can activate the ortho C–H bond of pyridine substrates (Figure ).[10−13] C–H activation is believed to take place by an oxidative addition mechanism at the transition metal center, leading to products in which the pyridyl group is directly bonded to this metal.[13−15] The main group ligand plays a role in substrate coordination and determining the ortho selectivity but itself does not lead to new types of reactivities.

Figure 1

Well-defined bimetallic systems for the selective ortho C–H activation of pyridines. Current systems that require ligand dissociation prior to bond activation vs a novel pathway occurring at a coordinatively saturated iron aluminylene system.

In this paper, we report the synthesis and characterization of bimetallic complexes in which an unsupported[16] aluminum-based ligand is bound to an iron(II) dihydride fragment. These complexes are neutral low-spin d6 species based on an octahedral parent geometry and as such are expected to be chemically inert.[17] Textbook examples of such compounds with more classical ligand systems possess a stable 18-electron configuration and a large highest occupied molecular orbital (HOMO) (t2g)—lowest unoccupied molecular orbital (LUMO) (eg) gap, limiting direct reactivity at the metal site.[18] Inclusion of the aluminum-based ligand results in a significant distortion of the geometry away from octahedral due to favorable, but weak, interactions in the secondary coordination sphere. This distortion leads to an unusual ground state destabilization and raises the energy of the HOMO of the iron center. This effect exposes an entirely new type of reactivity of the neutral low-spin d6 center. These complexes were found to selectively break the ortho C–H bond in pyridine. Mechanistic studies reveal that the two metal centers act as a frustrated Fe–Al Lewis pair (FLP)[19−23] leading to a deprotonation rather than oxidative addition pathway.[24]

The findings both compliment and expand upon the known nucleophilicity of anionic analogues such as [Fe(η5-C5H5)(CO)2]−. More broadly, these results suggest that the shape and electronic structure of transition metal centers can be modulated through incorporation of Lewis acidic ligands, leading to perturbation of the electronic structure and exposing new types of reactivities.

Results and Discussion

Synthesis and Characterization

Complexes 2a and 2b could be readily prepared by reacting FeBr2, PMe3, and respective β-diketiminate aluminum hydrides 1a (R = Mesityl or Mes) and 1b (R = 2,6-diisopropylphenyl or Dipp) in toluene or benzene and were isolated in 80–90% yield (Figure ). Both complexes exhibit a mutually coupled spin system comprised of one triplet and one doublet resonance in the 31P{1H} NMR spectra with an integration ratio of 1:2. These resonances are consistent with the magnetic non-equivalence of the axial and equatorial phosphine ligands. The most characteristic features in the 1H NMR spectra are broad signals at δH = −13.08 (2a) or −13.20 (2b) ppm assigned to the bridging hydrides which are coupled to the quadrupolar I = 5/2 27Al nucleus.

Figure 2

(a) Preparation of complexes 2a-b and 3a-b. (b) X-ray structure of 2b. (c) X-ray and calculated structure of 3b.

Reduction of these precursors using magnesium turnings in tetrahydrofuran affords complexes 3a and 3b in almost quantitative NMR yield. Both 3a and 3b can be isolated as dark red/orange crystalline solids. In contrast to complexes 2a-b, 3a-b both show only one singlet resonance in the 31P{1H} and a well resolved quartet hydride resonance in the 1H NMR spectra. The data are consistent with a highly symmetric structure in solution—an observation that could suggest fast ligand exchange on the NMR timescale.

Structure and Bonding

Single crystals suitable for X-ray diffraction could be obtained for all four compounds. Data for 2b and 3b are depicted in Figure . The position of the hydride ligands in 3b could not be refined, but their presence is evident from the corresponding 1H NMR spectrum. When comparing these structures, it becomes apparent that drastic changes in the Fe···Al intermetallic distances occur upon reduction. For example, a decrease in the Fe···Al distance from 2.459(1) Å in 2b to 2.194(1) Å in 3b becomes visible in the solid-state structures. Although the Fe···Al separation of 2b almost matches the sum of their covalent radii[25] [formal shortness ratio[26] (FSR) = 1.02], it is clearly below this value in 3b (FSR = 0.91). Such short Fe···Al bond lengths appear to be diagnostic for an aluminylene[27] metalloligand covalently bound to iron.[28]

Based on the experimental data, it is evident that 2a-b and 3a-b are low-spin and diamagnetic. These species can thus be assigned as 18-valence electron complexes. Although 2a-b can be confidently described as σ-alane[29] complexes [see the quantum theory of atoms in molecules (QTAIM) analysis below], three extreme bonding scenarios[16] may be considered for bimetallic species 3a-b (Figure a): (A) a four-electron η2:η2-coordination of H–Al–H to the 14-electron fragment [Fe(PMe3)3], (B) two-electron coordination of an aluminylene metalloligand to a 16-electron Fe(II) dihydride fragment, and (C) strong polarization of the TM–M bond resulting in a cationic Al metalloligand and an anionic iron dihydride fragment due to the electronegativity difference between the two metals (ΔχP = 0.22).[30,31] According to the computational results presented below, an intermediate between structures B and C in Figure a appears as the most appropriate bonding description for complexes 3a-b.

Figure 3

Analysis of the bonding in 3a-b. (a) Extreme bonding descriptions. (b) Calculated Fe···Al distances and NPA charges of 2a and 3a [ωB97X//6-31g** (H,C,N,P)/SDDAll (Fe,Al)]. (c) QTAIM analysis and WBIs of 2a and 3a.

More insights into the structure and bonding in these complexes were gained by density functional theory (DFT) calculations, and the optimized structure of 3b is depicted in Figure c. In complexes 3a-b, the iron fragment adopts a pseudo-octahedral coordination geometry in an all-cis configuration of the hydride and phosphine ligands, respectively (3b: L-Fe-L = 83.1–101.6°; L = H, PMe3). Only the Al metalloligand deviates from this geometry being bent away from its axial position by about 45°. The calculated structures also reflect the short Fe···Al distances observed in the solid-state structure of 3b (Figure b). The Al–H bonds in these species appear to be significantly longer in comparison to the parent dibromide complexes (e.g., 1.838 vs 2.045 Å), and the Wiberg bond indices (WBIs) for the Fe···Al bond drastically increase from 0.14 in 2b to 0.55 in 3b.

These findings are further underpinned by QTAIM calculations (Figure c).[36] In 2b, bond critical points (BCPs) are found between Fe and H and Al and H but not between Fe and Al. In 3b, the QTAIM analysis reveals BCPs between Fe and Al and Fe and H, whereas no BCPs are found between Al and H. These data suggest the presence of a direct metal–metal bond due to a preceding double Al–H bond activation at the transition metal center.[30,32−35]

Analysis of the NPA charges in 3b reveals significant polarization of the Fe–Al bond showing values of −0.97 for Fe and 1.25 for Al, while negative charges on the hydrides are low (−0.18/0.18). The large negative charge accumulation on Fe likely results from the strong electron-donating nature of both the two hydrides and aluminylene ligand. This assumption is supported by extended transition state-natural orbitals for chemical valence (ETS-NOCV) calculations[37] on 3b revealing that the donation from the hydrides and the aluminylene ligand to iron accounts for more than 64% (Δρ1 = −74.7 kcal/mol) of the total orbital interaction energy (ΔEorb = −117.1 kcal/mol), whereas backdonation from the Fe–H bonds to the empty Al p-orbitals was identified as the second (Δρ2 = −16.6 kcal/mol, 14%) and third (Δρ3 = −14.2 kcal/mol, 12%) largest contributions to ΔEorb (see Table S5 in the Supporting Information). These secondary interactions are likely responsible for the bent position of the aluminylene metalloligand.

Natural bond orbital (NBO) calculations were used to gain further insights into the bonding in these complexes. For the sake of clarity, we considered simplified model complex 3c (P’ = PH3, R = Me) which, however, gives rise to highly similar frontier MOs as in 3a-b (see Table S6 in the Supporting Information). The NBO analysis identifies a σ-bond between Fe (4s, 40.8%, 3d 57.8%) and Al (3s 73.6%, 3p 26.4%) as the main contributor to the HOMO, while the LUMO + 2 comprises an empty p-orbital on aluminum and possesses antibonding character with respect to the Fe–Al σ-bond. The HOMO is significantly higher (about 1.0 eV) in energy than the lower MOs which are predominantly non-bonding and possess largely 3d character (HOMO-1 to HOMO-3). This difference appears to be a consequence of the distorted coordination geometry and vanishes when the metalloligand is moved toward the axial position (Figure ).

Figure 4

(a) Energy profile for the relaxed scan of the Al–Fe–Pax angle in simplified model complex 3c [ωB97X//6–31g** (H,C,N,P)/SDDAll (Fe,Al)]. (b) Energies of the occupied frontier orbitals in 3c as function of the Al–Fe–Pax bending angle. (c) Orbital isosurfaces (isovalue = 0.05) at 0° (equilibrium structure) and 40° (metalloligand in the axial position) bending angles.

Intramolecular C–H Activation

Complex 3a was found to undergo an intramolecular C–H activation affording cyclometalated complex 4a (Figure a). Heating a toluene-d8 solution of 3a to 80 °C results in slow formation of a new species as revealed by a growing new singlet resonance at δP = 30.0 ppm in the 31P NMR spectrum. In the 1H NMR, 4a gives rise to a new hydride signal (broadened quartet) that integrates to 3H. The reaction was monitored over time, and complete consumption of the starting material was observed after 24 h. The new species was formed in 71% NMR yield. Crystals suitable for X-ray diffraction could be grown by diffusion of tetramethylsilane into a saturated solution of n-pentane, and the solid-state structure of 4a is depicted in Figure b. Heating a solution of 3b in toluene-d8 to 80 °C only results in the slow decomposition of this complex and the formation of untraceable species.

Figure 5

(a) Intramolecular C–H activation of 3a. (b) X-ray structure of 4a. (c) calculated free energy profile for the conversion of 3a to 4a [B3PW91/PCM (toluene)/D3//6–31g** (H,C,N,P)/SDDAll (Fe,Al)]. Energies are given in kcal/mol. (d) Acceptor (LUMO + 6) and donor (HOMO) orbitals in intermediate 1-B.

More insights into the conversion of 3a to 4a were gained from a combination of kinetic experiments and DFT calculations.[13] Experimentally, the reaction was found to be first-order with respect to 3a. An Eyring analysis over a temperature range of 60–100 °C gave the activation parameters ΔH‡ = 23.4 ± 0.5 kcal mol–1 and ΔS‡ = 12.5 ± 1.3 cal mol–1 K–1 which correspond to an associated ΔG298K‡ = 27.1 ± 0.8 kcal mol–1. DFT calculations were benchmarked against the experimental data. Best results were obtained by using the B3PW91 functional and a 6–31g** (H,C,N,P)/SDDAll (Fe,Al) basis set. Solvent (PCM, Benzene) and dispersion corrections (D3) were directly included in the optimization of the stationary points.

The calculated free energy profile for the intramolecular C–H activation is shown in Figure c. The reaction sequence is initiated by a slight rotation of the aluminylene ligand around the Fe···Al axis (1-A to 1-B) followed by a deprotonation of the mesityl CH3 group by the iron metal center (1-TS). Intermediate 1-C features the Fe-(μ-H3)-Al bridging motif and the deprotonated mesityl residue. Another slight conformational change initiates the Al–C bond formation (1-TS) and leads to the cyclometalated product (1-D). Formation of the product is exergonic by −13.3 kcal mol–1 with an overall barrier of 27.9 kcal mol–1 being in good agreement with the experimental value (27.1 ± 0.8 kcal mol–1).

Analysis of the frontier MOs reveals that the HOMO in the ground state of 3a (Figure d) acts as the electron donor for the reductive deprotonation of the CH3 group in 1-TS, while an empty p-orbital at Al (LUMO + 6, Figure d) acts as the electron acceptor for the deprotonated CH2 group (1-TS) to form the Al–C bond in the final product (1-D).

Intermolecular C–H Activation

Complexes 3a and 3b are also capable of promoting intermolecular C–H activation. For example, 3a readily reacts with pyridine (1 equiv, C6D6, room temperature) resulting in the selective C–H activation in the 2-position of the heterocycle (>95% NMR yield, Figure a). Like 4a, 5a exhibits a sharp singlet resonance in the 31P{1H} NMR spectrum at δP = 29.6 ppm and a broadened quartet hydride signal at δH = −15.33 ppm in the 1H NMR spectrum. The reaction of 3b with pyridine required more forcing reaction conditions (40 °C, 10 equiv of pyridine, 18 h) but led to analogue product 5b in about 80% NMR yield. Crystals suitable for X-ray diffraction could be obtained for 5b, and the solid-state structure confirms the ortho C–H activation and C–Al bond formation in the 2-position of the pyridine heterocycle (Figure b).

Figure 6

(a) Pyridine ortho C–H activation with 3a-b (NMR yields are given in parentheses), (b) X-ray structure of 5b, and (c) calculated free energy profile for the conversion of 3a to 5a [B3PW91/PCM (benzene)/D3//6–31g** (H,C,N,P)/SDDAll (Fe,Al)]. Energies are given in kcal/mol.

A small array of substrates was investigated. Employing pyridines with different substituents in the 4-postion (R = Me, Ph, and NMe2) under the same conditions did not show any change in the reactivity or selectivity of the C–H activation with 3a, in all cases giving products that arise from substitution at the 2-position (Figure a). On the other hand, 2-phenylpyridine did not react with 3a, presumably due to steric interference of the phenyl substituent with the ligand system on aluminum.

DFT calculations on the ortho C–H activation of pyridine suggest a similar reaction pathway as for the intramolecular C–H activation of 3a (Figure c). Again, a two-step mechanism has been identified that proceeds via the deprotonation of the pyridine substrate. Coordination of the pyridine nitrogen to the aluminum center controls the site selectivity of the reaction. NBO calculations reveal that the coordination of pyridine also facilitates bond breaking. In 2-A, the Fe–Al bond becomes more polarized in comparison to 1-B, and the WBI drops (see Tables S7 and S8 in the Supporting Information). The following transition state 2-TS is significantly lower in energy than for the intramolecular deprotonation of the mesityl CH3 group in 3a (14.7 vs 27.9 kcal/mol) being in line with a fast reaction at room temperature. Both the intermediates after coordination (2-A, 0.0 kcal/mol) and deprotonation (2-B, 0.8 kcal/mol) of the pyridine appear to be almost thermoneutral in comparison to the starting materials. However, a final switch from N- to C-coordination of the deprotonated pyridine is facile (2-TS, a barrier of 5.5 kcal/mol) and highly exergonic due to formation of the Al–C bond in 2-C (−20.7 kcal/mol).

A comparison of the reaction rates using an excess of pyridine or pyridine-d5 in two independent reactions gave an unusually large kinetic isotope effect (KIE) of 14.0 ± 0.2 at 297 K. The experimentally observed kH/kD value likely results from a large classical KIE including a significant contribution of quantum tunneling.[38] The rate-determining elementary step (2-A to 2-B) is nearly thermoneutral with a centered transition state along the reaction coordinate, and thus, a maximum primary KIE (up to 7–8) may be expected.[38,39] On the other hand, KIEs greater than 10 are generally considered to be caused by quantum tunneling and are diagnostic for proton transfer reactions.[40−43] Both assumptions are in agreement with the calculated mechanism in Figure c and support a non-oxidative addition pathway.

Conclusions

In summary, we described two well-defined bimetallic Fe–Al complexes that possess a distorted octahedral geometry. This distortion leads to an unusual ground state destabilization and raises the energy of the HOMO (the Fe–Al bond). Consequently, the HOMO becomes accessible for an encounter substrate and leads to an enhanced reactivity of these complexes including intramolecular sp2 C–H activation and the selective ortho C–H activation of pyridine substrates. These reactions were found to follow a novel cooperative mechanism in which a reductive deprotonation of the C–H bond was identified as a key elementary step. We could show that the Fe–Al bonds in these complexes are highly polarized and react as a bimetallic FLP in which the destabilized HOMO acts as a Lewis donor orbital, while empty p-orbitals on Al serve as Lewis acceptors. The reaction leads to the formation of a Fe–H and Al–C bonds. These results may lay the foundation for the rational design of future catalytic systems with prospects in the field of base-metal catalysis.