Aluminium porphyrins catalyse the hydrogenation of CO₂ with H₂

Abstract

Boron-based frustrated Lewis pairs (FLPs) have become well-established catalysts for the hydrogenation of a wide range of functional groups. Conversely, aluminium-based FLP hydrogenation catalysts are less common, especially for CO₂ reduction. They are mostly confined to the hydrogenation of imines, alkenes, and alkynes even though aluminium is much more abundant than boron and forms structurally related compounds. Moreover, aluminium forms penta- and hexa-coordinated complexes, which remain untested in FLP hydrogenation catalysis. Herein, we demonstrate that cationic, hexa-coordinated diaqua-meso-tetraphenylporphyrin aluminium complexes [Al(TPP)(OH₂)₂]X and [Al(^tBuTPP)(OH₂)₂]X (X = Cl⁻, OTf⁻, ClO₄⁻) form FLPs with nitrogen bases, activate H₂, and reductively couple CO₂ to amines, yielding N-formylamines and water. Our experimental results and DFT analysis indicate that H₂ activation involves the formation of an FLP, base-promoted CO₂ reduction and formate salt elimination from the FLP, as proposed for transition metal-catalyzed N-formylations. These similarities in the reaction mechanism and structure of aluminium complexes brings Al-based FLPs closer to transition metal catalysis and may enable us to apply this knowledge to ligand design to enhance main group metal-promoted hydrogenations.

---

Introduction

Frustrated Lewis pairs (FLPs) are combinations of Lewis acids (LA) and bases (LB) sterically hindered from generating classical LA-LB adducts. FLPs have been used to activate small molecules,¹ especially H_2,^2,3 and to catalyse hydrogenations.⁴ As a case in point, boron-based FLPs catalytically hydrogenate ketones, esters, imines, enamines, N-heterocyclic compounds, silyl ethers, oxime ethers, alkenes, alkynes, aromatics and to some extent even CO₂.^5–8 In contrast, functional aluminium-based FLP hydrogenation catalysts remain rare, even though aluminium is chemically related to and, by far, much more abundant than boron.

In fact, aluminium is the most abundant metal in the Earth's crust and, like boron, forms predominantly trivalent compounds, which are LAs and, as such, potentially suitable for FLP chemistry. Aluminium-based FLPs are well established^9,10 and known to facilitate hydroborations,^11–13 to react with ammonia boranes,¹⁴ C–H,^15,16 C≡C,^17,18 C=O^19–21 bonds and CO₂ (ref. 22–25) and even to activate H₂,^12,26 in some cases.

Produced in the same reaction, Al-hydrides are, however, less stable than B-hydrides (c.f. NaBH₄ and LiAlH₄),²⁷ and Al–C bonds of parent LAs are less thermally and hydrolytically stable than their boron equivalents.^28,29 For example, B(C₆F₅)₃ can be used as the LA component of FLP hydrogenation catalysts at temperatures up to 100 °C, even in wet solvents,^30,31 while Al(C₆F₅)₃ decomposes at 60 °C, facilitating only the stoichiometric reduction of alkenes.³²

Alkenes have been catalytically reduced by incorporating three coordinate aluminium catechol LAs into porous organic polymers (Scheme 1a), but this approach failed to hydrogenate other substrates, such as ketones and imines.³³ By then, though, imine hydrogenation had already been catalysed by Al-based FLPs, with di-isobutylaluminium hydride and tri-isobutylaluminium (DIBAL, TIBAL) LAs at 103 bar of H₂.³⁴ More recently, this system has also proved effective with MAlH₄ (M = Li, Mg, Ca, Sr) at 1 bar of H₂ (Scheme 1b).^35,36 Under these reaction conditions, aluminium likely binds to the imine substrate, thereby forming an FLP suitable for H₂ activation.^37,38 Imine hydrogenation has also been catalysed by [(^DippBIAN)Al-(H)₂–Li(OEt₂)₂]³⁹ at 50 bar of H₂ and 100 °C and by cationic (β-diketiminate)AlMe⁺ (ref. 40) at 1.5 bar of H₂ and 60 °C, but attempts at expanding the scope of hydrogenation activity beyond imines, alkenes and alkynes have not been successful thus far.⁴⁰ In other words, Al-catalysed hydrogenations continue to fail when applied to other substrates, such as aldehydes, ketones and CO₂, and succeed only with catechol-borane when using Al-based FLPs.^29,41,42

Scheme 1 (a) Hydrogenation of alkenes and alkynes; (b) hydrogenation of imines by Al-based catalysts; (c) N-formylation of amines with CO₂ and H₂ catalysed by Al-porphyrin.

In this study, we demonstrate that cationic, hexa-coordinate diaqua Al-porphyrin complexes form FLPs with nitrogen bases, activate H₂, and act as CO₂ hydrogenation catalysts in the presence of amines, yielding N-formylamines and water (Scheme 1c). Our experimental results and DFT analysis indicate that weaker bases, such as 2,4,6-collidine (pKa of conjugate acid 7.43 in H₂O), activate H₂ more readily than morpholine (pKa of conjugate acid 8.36 in H₂O) or 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU pKa of conjugate acid 23.9 in MeCN), which bind to CO₂ and the LAs and hinder H₂ activation. Furthermore, the N-formylation reaction proceeds as proposed for well-established transition metal catalysts,^43–59 suggesting that the hexa-coordinate aluminium complexes behave as transition metals in hydrogenation chemistry. Combined, our findings may stimulate ligand and catalyst design knowledge transfer from the field of transition metal catalysis to FLPs.

Result and discussion

Diaqua-meso-tetraphenylporphyrin aluminium complexes [Al(TPP)(OH₂)₂]X, where X = Cl⁻, OTf⁻, ClO₄⁻ and diaqua-meso-tetra(4-tert-butylphenyl)porphyrin aluminium chloride [Al(^tBuTPP)(OH₂)₂]Cl, were prepared as reported in the literature (ESI†).⁶⁰N-formylation of morpholine with CO₂ and H₂ was selected as the initial model hydrogenation reaction because CO₂ reduction is particularly challenging with FLPs^5,7,61 but well established with main group hydrides.⁶² For the reaction to proceed, these main group hydrides must be formed during FLP-catalysed hydrogenation from H₂. The preliminary reaction conditions (1 mmol of morpholine, 10 mol% of [[Al(TPP)(OH₂)₂]Cl:2,4,6-collidine], 4 bar of CO₂, 96 bar of H₂, 180 °C, 4 mL of sulfolane) were selected based on the only FLP hydrogenation catalyst [R₃SnX:2,4,6-collidine] known for this reaction.⁶³ Under these conditions, the desired product, N-formylmorpholine, was synthesized in 33% yield, a value that is below average as the yields reported for [R₃SnX:2,4,6-collidine] FLPs range from 39 to 95%, depending on the R and X groups.⁶³

Moreover, [[Al(TPP)(OH₂)₂]Cl:2,4,6-collidine] and [[Al(TPP)(OH₂)₂]Cl:morpholine] can be formed in the system (vide infra) and act as FLP in H₂ activation and CO₂ reduction, as confirmed by the results. At RT, the Lewis bases deprotonate the aqua ligands, but at high temperatures, the same ligands are reversibly displaced (ESI†). Minor adjustments were then made to the reaction conditions to suit the Al-porphyrin system (Scheme 2) before introducing structural modifications to the LA and LB used as the FLP catalyst (Table 1).

Scheme 2 Standard reaction conditions for the study of Al-based FLPs in the N-formylation reaction of morpholine.

Table 1 The structures of LA and LB affect the N-formylation of morpholine with CO₂ and H₂^a

Entry	LA	Added LB	Yield (%)	Error^b (σ)
a Reaction conditions: morpholine (1 mmol), sulfolane (4 mL), LA and LB (10 mol%), CO2 (7 bar), H2 (100 bar), 200 °C, 24 h, average yield after three runs. b The error indicated is one standard deviation rounded to a whole number. Yield was determined by ^1H NMR with an internal standard.
1	[Al(TPP)(OH2)2]Cl	2,4,6-col.	60	4
2	[Al(TPP)(OH2)2]OTf	2,4,6-col.	44	2
3	[Al(TPP)(OH2)2]ClO4	2,4,6-col.	24	2
4	[Al(^tBuTPP)(OH2)2]Cl	2,4,6-col.	53	1
5	[Al(Pht)Cl]	2,4,6-col.	19	1
6	[Al(TPP)(OH2)2]Cl	—	26	2
7	[Al(TPP)(OH2)2]Cl	DBU	30	5
8	[Al(TPP)(OH2)2]Cl	NMM	15	4
9	[Al(TPP)(OH2)2]Cl	2,6-Lut	20	6
10	TPP	2,4,6-col.	4	—
11	—	2,4,6-col.	4	—
12	[Al(TPP)(OH)]	2,4,6-col.	4	—
13	[Al(TPP)(OH)] + [morpholineH]Cl	2,4,6-col.	4	—

---

Under these modified conditions and with [Al(TPP)(OH₂)₂]Cl, N-formylmorpholine was formed in 60% yield (Table 1, entry 1). Substituting the counter anion Cl⁻ for OTf⁻ or ClO₄⁻ decreased the yield to 44 and 24%, respectively (Table 1, entries 2 and 3). By contrast, the ions OTf⁻ and ClO₄⁻ interact less in the tetra-substituted R₃SnX (X = Cl⁻, OTf⁻, ClO₄⁻), accelerating the catalytic turnovers.⁶³ During catalysis, the counter anion likely interacts with the cationic complex [Al(TPP)(OH₂)₂]⁺ and promotes the reaction, as shown by adding tetrabutylammonium chloride (10 mol%) to [Al(TPP)(OH₂)₂]OTf, which increases the reaction yield from 44 to 50% (ESI†).

In addition, the axial ligand of the porphyrin complex must be labile because the stable [Al(TPP)(OH)] complex is inactive (Table 1, entry 12). Together with morpholinium or collidinium chloride, [Al(TPP)(OH)] may be a reaction intermediate (vide infra). But adding morpholinium chloride to the N-formylation reaction with [Al(TPP)(OH)] and 2,4,6-collidine does not promote formamide formation (Table 1, entry 13). Therefore, [Al(TPP)(OH)] should not be formed during the reaction.

Substituting the meso-tetraphenylporphyrin ligand for meso-tetra(4-tert-butylphenyl)porphyrin [Al(^tBuTPP)(OH₂)₂]Cl slightly decreased the catalytic activity, leading to 53% N-formyl morpholine yield (Table 1, entry 4). The electron-donating tert-butyl group on the porphyrin ligand likely diminishes the Lewis acidity of the complex, decreasing the activity. The commercial phthalocyanine complex [Al(Pht)Cl] further decreased the yield of N-formylmorpholine to only 19% (Table 1, entry 5).

Considering that porphyrins and phthalocyanines are planar, steric hindrance should not account for the decrease in reactivity. This difference more likely derives from electronic effects associated with changes in the LA of the metal centre and, above all, from the cationic charge of [Al(TPP)(OH₂)₂]⁺, which enthalpically promotes the reaction (vide infra).

FLP catalysts can also be electronically controlled by LB, so we assessed the LB effect on the reaction. Two possible FLPs can be formed in the N-formylation of morpholine with [Al(TPP)(OH₂)₂]X and 2,4,6-collidine, namely [[Al(TPP)(OH₂)]Cl:2,4,6-collidine] and [[Al(TPP)(OH₂)]Cl:morpholine], since the reaction substrate is also a LB. Moreover, morpholine is a stronger base (pKa 8.36) than 2,4,6-collidine (pKa 7.43) and, accordingly, should more efficiently activate H₂ in combination with LA. Yet, when 2,4,6-collidine was removed from the reaction mixture, the product yield decreased from 60 to 26% (Table 1, entries 1 and 6), indicating that 2,4,6-collidine is crucial for the reaction.

Morpholine may react with CO₂, forming a carbamic acid, which removes the base from the reaction, thereby preventing FLP formation.⁶³ This hypothesis was partly supported by the lack of changes in yield when using stronger bases, such as DBU (pKa 23.9 in MeCN). DBU also reacts with CO₂, forming carbamates and the desired product in 30% yield, which was as high as that obtained when running the reaction without DBU (Table 1, entry 7). In turn, N-methylmorpholine (NMM, pKa 7.38) and 2,6-lutidine (pKa 6.6) decreased the yield to 15 and 20%, respectively (Table 1, entries 8 and 9). Although aliphatic NMM and 2,4,6-collidine have similar basicities (pKa of conjugate acid in H₂O of 7.38 and 7.43, respectively), NMM is presumably not sterically hindered enough for FLP formation. By contrast, 2,6-lutidine has the same steric profile as 2,4,6-collidine but is too weak of a LB.

When optimizing the reaction conditions, we observed that the reaction only proceeded at temperatures above 160 °C, suggesting a high energy barrier (Fig. 1A). In line with the ease of CO₂ reduction by main group hydrides,^62,63 the reaction readily proceeded at low CO₂ partial pressures between 2 and 10 bar (Fig. 1B). Conversely, transition metal-catalysed reactions usually require a high CO₂ partial pressure, typically between 25 and 30 bar.^43–59 Nevertheless, high partial pressures of H₂ are required to improve the N-formylmorpholine yield from 4% at 20 bar to 60% at 100 bar (Fig. 1C), further indicating that H₂ activation or its binding is, most likely, limiting the reaction.

Fig. 1 Effect of H₂ (C) and CO₂ (B) partial pressure and temperature (A) on the N-formylation reaction of morpholine with CO₂, H₂ and [Al(TPP)(OH₂)₂]Cl catalyst. All yields were determined by ¹H NMR with an internal standard as the averages of three runs, and with error bars representing one standard deviation.

Despite extensive efforts to optimize the conditions, the reaction yields did not surpass 60% (Fig. 1). Neither prolonging the reaction nor adding more catalyst, whether [Al(TPP)(OH₂)₂]Cl or 2,4,6-collidine, increased the yield. Thus, reaction equilibria, not insufficient reaction time or catalyst instability, accounted for the maximum reaction yield.

Knowing that the reaction is thermodynamically favourable, we studied its mechanism in detail using both experimental approaches and DFT calculations (Fig. 2). Numerous reaction pathways have been hypothesized for N-formylations catalysed by FLPs, which differ in CO₂ reduction and N-formylation steps.⁶³ The first, main group-like hydride reduction of CO₂, presumably involves CO₂ insertion into the Al–H bond and [Al(TPP)(OCHO)] and [baseH]Cl formation followed by a reaction with morpholine, yielding the desired N-formylmorpholine, [Al(TPP)(OH)] and [baseH]Cl. The catalyst is regenerated by water elimination from [Al(TPP)(OH)] through protonation with [baseH]Cl. Nevertheless, [Al(TPP)(OH)] is catalytically inactive in the presence of an equimolar quantity of [baseH]Cl (Table 1, entry 13). Even with a 10-fold excess morpholinium chloride, required for [Al(TPP)(OH)] protonation, the reaction is slower than with [Al(TPP)(OH₂)₂]Cl, so [Al(TPP)(OH)] formation is unlikely to significantly contribute to the reaction. Similarly, the second reaction pathway via direct carbamate reduction leads to inactive [Al(TPP)(OH)]. Both reaction pathways contradict the experimental results and are, therefore, unlikely to be the working catalytic cycles.

Fig. 2 DFT study of the reaction mechanism of N-formylation of morpholine with CO₂ and H₂ catalyzed by Al-based FLPs at the M06-2X/6-311++G** level of theory, using IEFPCM model for solvation corrections; the reaction pathway with [Al(TPP)Cl:morpholine] FLP is indicated in blue (left-hand side), the one with [Al(TPP)Cl:2,4,6-collidine] FLP is indicated in orange and the one with [[Al(TPP)(OH₂)]Cl:2,4,6-collidine] FLP is indicated in green (right-hand side). Relative enthalpies are indicated in kcal mol⁻¹, and E0 represents the sum of the enthalpies of the initial compounds without any interaction between them.

The third catalytic cycle includes H₂ splitting, subsequent CO₂ reduction to formate, formate release and regeneration of the initial FLP, which closes the catalytic cycle. To gain further insights into the third reaction mechanism, we performed solvation-corrected DFT calculations at the M06-2X/6-311++G** level of theory so as to describe the enthalpy of the reaction coordinate at 473 K and 100 bar (ESI†). At first, the effect of 2,4,6-collidine on the reaction was studied using [Al(TPP)Cl:morpholine] i.e., without 2,4,6-collidine (Fig. 2, blue pathway), and with [Al(TPP)Cl:2,4,6-collidine] FLP when this base was present (Fig. 2, orange pathway). The effect of the cationic complex [Al(TPP)(OH₂)₂]Cl, replacing the neutral [Al(TPP)Cl], was then assessed on the more favourable pathway in the presence of 2,4,6-collidine (Fig. 2, green pathway).

In the presence of a LB, morpholine (B_a) or 2,4,6-collidine (B_b), [Al(TPP)Cl] immediately completes its coordination sphere, forming a hexa-coordinate octahedral complex (Fig. 3, R_a and R_b), as observed in transition metal catalysts during the N-formylation reaction.⁶⁴ In the presence of morpholine, this complex is stabilized by −24.8 kcal mol⁻¹ (Fig. 2, R_a), with a bond length between morpholine and the aluminium centre of 2.17 Å and an almost straight angle of 177.3° between TPP and the aluminium centre (Fig. 3, R_a). Similar energetics are found for morpholinium carbamate binding (−25.4 kcal mol⁻¹), suggesting reversible CO₂ insertion into the Al-morpholine bond (ESI†). In contrast, the octahedral complex with 2,4,6-collidine is stabilized by only −14.3 kcal mol⁻¹ (Fig. 2, R_b). Since R_b can be converted into R_a, R_a acts as an off-cycle intermediate for the catalytic cycle with 2,4,6-collidine. In R_b, the distance between 2,4,6-collidine and aluminium is 3.35 Å, and the distortion from octahedral geometry is larger, with an angle of 161.0° between TPP and aluminium (Fig. 3, R_b). As previously proposed,^65,66 the distance between 3 and 5 Å is optimal for H₂ activation by FLP and, accordingly, favoured by R_b.

Fig. 3 Optimized geometries of [Al(TPP)Cl:morpholine] (R_a), [Al(TPP)Cl:2,4,6-collidine] (R_b) and [[Al(TPP)(OH₂)]Cl:2,4,6-collidine] (R′b) (top), and the corresponding transition states for H₂ splitting and CO₂ hydrogenation for [[Al(TPP)(OH₂)]Cl:2,4,6-collidine] catalysts (TS1′_b and TS2′_b) (bottom). C–H hydrogens are omitted for a better visualization, at the M06-2X/6-311++G** level of theory, using IEFPCM model for solvation corrections.

H₂ activation proceeds in a heterolytic fashion.^66,67 H₂ activation from R_a has a higher energy transition state (Fig. 2, TS1_a) than H₂ activation from R_b (Fig. 2, TS1_b). The local activation barrier for direct H₂ activation from R_a is 31.7 kcal mol⁻¹, and only 20.3 kcal mol⁻¹ from R_b. Moreover, considering the equilibrium between R_a and R_b, the energy barrier for H₂ activation from R_a is favoured by its transformation to R_b, followed by H₂ activation viaTS1_b, with the energy barrier (30.8 kcal mol⁻¹) mildly favouring this process. The Al-hydrides, I1_a and I1_b, are formed in endothermic reactions with relative enthalpies of 5.5 and 2.9 kcal mol⁻¹, respectively. Based on these positive energies, their formation is enthalpically unfavourable, so they are likely rapidly protonated by [baseH]⁺ or by the ever present carbamic acid, which is produced in the reaction of morpholine with CO₂.

Al-hydride protonation is a standard decomposition pathway of such compounds.²⁷ Hydrides of comparably planar calix[4]pyrrole aluminium complexes exist only in an unfavourable equilibrium and are rapidly displaced by even weakly basic ligands, such as THF.^27,68 An unfavourable pre-equilibrium between H₂ and Al-hydrides, I1_a and I1_b, is thus expected, where I1_b is slightly favoured over I1_a and should persist longer or in larger quantities in the reaction. Together with the high local activation energy for H₂ activation, the unfavourable pre-equilibrium is likely responsible for the high H₂ partial pressure(s) required for the reaction. Experimentally, dihydrogen activation is confirmed when using D₂, which results in complete deuteration of the formamide moiety of N-formylmorpholine. Thus, dihydrogen is the sole source of reducing agent of the reaction (ESI†).

The high reactivity of Al-hydrides produced in the reaction is also clearly shown by the low local activation barrier for CO₂ reduction (Fig. 2, TS2_a and TS2_b). Although most Al-hydrides formed in this reaction are likely protonated back to H₂ in the reverse reaction, this reaction can proceed thanks to the comparably low local energy barriers of CO₂ reduction (6.5 and 4.3 kcal mol⁻¹, respectively). Attempts at isolating [Al(H)(TPP)] failed, but adding CO₂ to the solution led to its immediate reduction to formate, as detected by ¹H NMR (ESI†).

Based on the enthalpic energetic span of the catalytic cycle, the transition states TS2_a and TS2_b for CO₂ reduction are also the turnover frequency (TOF)-determining transition states (TDTS), with a global energy barrier of 36.8 kcal mol⁻¹ starting from the TOF-determining intermediate (TDI) R_aviaTS1_a and I1_a. For the 2,4,6-collidine pathway (orange), the global energy barrier is only 21.5 kcal mol⁻¹ between R_b and TS2_bviaTS1_b and I1_b. Alternatively, the energy span from R_a to TS2_bviaR_b, TS1_b and I1_b is 32.0 kcal mol⁻¹ considering the equilibrium between R_b and R_a and the ability of R_a to act as an off-cycle thermodynamic well for the 2,4,6-collidine pathway. In effect, the reaction with 2,4,6-collidine is favoured in all cases, in line with the experimental results (Table 1, entries 1 and 6).

After TS2_a and TS2_b, stable, intermediate complexes between [Al(TPP)Cl], formic acid and morpholine (I2_a) or 2,4,6-collidine (I2_b) are formed. With morpholine, I2_a has a stabilization energy similar to that of R_a (−23.2 vs. −24.8 kcal mol⁻¹, respectively), potentially allowing the reverse reaction. The reverse reaction, HCOOH decomposition to CO₂ and H₂, is also well established in the literature⁶⁹ and was confirmed here by evolution of H₂ gas, characterized by ¹H NMR, from a solution of formic acid and [Al(TPP)(OH₂)₂]Cl (ESI†). Considering these results, a thermodynamic equilibrium may be established in the reaction, thereby limiting the reaction yield.

Thermodynamically, the reaction is driven by [morpholinium][formate] elimination from either I2_a or I2_b, which spontaneously condenses into N-formylmorpholine and water under these reaction conditions.^47,64 This condensation also regenerates the FLP catalyst and allows the reaction to proceed. Formate salts are eliminated in transition metal-catalysed reactions,^47,64 whereas the formate remains attached to the central element^70–72 in reactions of main group hydrides, including LiAlH₄,⁷³ so the [Al(TPP)Cl] catalyst behaves like a transition metal rather than a main group reductant. In the complex I2, the protonated base continuously interacts with the LA and the generated formate. Therefore, another base, such as morpholine, is required for the reaction to proceed (Fig. 2, TS3_{a and b}) prior to catalyst regeneration.

If the reaction is limited by an equilibrium established between R_{(a or b)} and I2_{(a or b)} and driven by morpholine, higher concentrations of morpholine should promote the reaction. Indeed, increasing the quantity of morpholine increased the turnover number (TON) in the 24 hours reaction (Fig. 4). More specifically, increasing the substrate concentration from 1 to 22.5 mmol increased the TON from 7 to 80. This increase confirmed that the 60% yield obtained with 1 mmol of morpholine (Table 1, entry 1) is limited by the quantity of the substrate, not by the efficiency of the catalyst.

Fig. 4 Morpholine effect on the TON of the catalytic system; TON calculated based on the quantity of [Al(TPP)(OH₂)₂]Cl under the following reaction conditions: morpholine (1 to 22.5 mmol), sulfolane (4 mL), [[Al(TPP)(OH₂)]Cl:2,4,6-collidine] (0.1 mmol), CO₂ (7 bar), H₂ (100 bar), 200 °C, 24 h. The reaction yield was determined by ¹H NMR with an internal standard.

In the catalytic cycle with the [Al(TPP)(OH₂)₂]Cl pre-catalyst (Fig. 5), dissociation of one aqua ligand leads to the [[Al(TPP)(OH₂)]Cl:2,4,6-collidine] FLP (R′_b), whose geometry is similar to, but more stable than, that of [Al(TPP)Cl:2,4,6-collidine] (R_b) (Fig. 3). Nevertheless, the subsequent transition states, TS1′_b and TS2′b (Fig. 3), are more stabilized (Fig. 2, TS1′b and TS2′b), lowering the overall activation barrier of the catalytic cycle. The H₂ splitting step proceeds via a late transition state TS1′b, wherein the H–H bond is 0.2 Å more stretched than in H₂. Despite having the largest local activation barrier (16.9 kcal mol⁻¹), this transition viaTS1′b is still much more energetically favourable than the transition from the neutral complex R_b to TS1_b (20.3 kcal mol⁻¹). In comparison with the neutral TS2_b, TS2′b stabilization decreases the energetic span of the catalytic cycle from 21.5 (R_b → TS2_b) to 17.5 (R′b → TS2′b) kcal mol⁻¹. When including potential off-cycle intermediates, R_a and R′a, where morpholine binds to the catalyst, the energy span decreases from 32.0 (R_a → R_b → TS2_b) to 29.4 (R′a → R′b → TS2′b) kcal mol⁻¹ (ESI†). This difference supports our hypothesis that the cationic aqua complexes of aluminium and the associated FLP with collidine [[Al(TPP)(OH₂)]Cl:2,4,6-collidine] is the most favourable catalyst in the system.

Fig. 5 View of the molecular structure of [Al(TPP)(OH₂)₂]Cl as observed in the crystal structure of [Al(TPP)(OH₂)₂]Cl·CH₂Cl₂·2 Et₂O. C–H hydrogen atoms were omitted for clarity. Thermal displacement ellipsoids are plotted at the 30% probability level. Applied colours: C – black, H – black contour, Cl – green, N – blue, O – red, Al – yellow.

While the DFT calculations fit the experimental data, the pathways involve, at most, tri-molecular transition states and the formation of an Al-hydride. [Al(TPP)(OH₂)]Cl is a hard LA, whose interactions with the relatively soft hydride donor are potentially less favourable than with the considerably harder CO₂. Alternative pathway(s) may involve H₂ activation by a tetra-molecular transition state involving CO₂ interactions with the Al-catalyst and H₂ activation by the C-atom of CO₂ and the LB, resulting in the direct CO₂ reduction to formate. Such a possibility is currently under investigation in our laboratory.

After studying the catalytic system, we assessed the substrate effect on the N-formylation of amines with CO₂ and H₂ (Scheme 3). In the N-formylation reaction, all secondary amines tested yielded the corresponding N-formylated products between 40 and 60% yield, in line with the expected reaction equilibrium. The alcohol, tertiary amine and amide functional groups were well tolerated. Diisopropyl amine was almost unreactive in the only other FLP system reported thus far, namely [R₃SnX:2,4,6-collidine].⁶³ But in our system, due to its low steric hindrance around the LA centre, diisopropyl amine provided an average yield of 54%. Primary aliphatic amines were also N-formylated for the first time with FLP catalysts, albeit in yields ≤20%. Aromatic and benzylic amines were unreactive.

Scheme 3 Substrate scope (a) N-formylation of amines with CO₂ and H₂. Reaction conditions: amine (1 mmol), sulfolane (4 mL), [[Al(TPP)(OH₂)]Cl:2,4,6-collidine] (10 mol%), CO₂ (7 bars), H₂ (100 bars), 200 °C, 24 h. All yields were determined by ¹H NMR, and the structures were confirmed by GC-MS.

Conclusion

Hexa-coordinated complexes of aluminium incorporating a labile ligand such as water can be used as LAs in FLP chemistry, H₂ activation, and hydrogenation chemistry. Catalytic hydrogenation of CO₂ in the presence of amines yields formamides with [Al(TTP)(OH₂)₂]Cl as the Lewis acid and the sterically demanding 2,4,6-collidine as the Lewis base. Energetically, the reaction is favoured when using the cationic complex [Al(TTP)(OH₂)₂]Cl over the neutral [Al(TPP)Cl], which is also unstable and hydrolyses to the cationic form. The advantages of this FLP system are the absence of labile and hydrolytically unstable Al–C bonds, which allows hydrogenation reactions that produce water. Structural and reactivity similarities of the LAs to transition metal complexes may facilitate knowledge transfer from transition metal catalysis to FLP in ligand and catalyst design. Ligand modifications may enable us to tailor Lewis acidity and steric hindrance around the metal centre to improve the reactivity of FLPs and to tune their reactivity to reaction requirements.

Data availability

All the data supporting this article have been included in the main text and the ESI.†

Author contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We thank the Czech Science foundation (GAČR 21-27431M) for funding this study, the National Scientific and Technical Research Council (Consejo Nacional de Investigaciones Científicas y Técnicas – CONICET) for Gabriela Gastelu's fellowship and Charles University Research Centre program No. UNCE/24/SCI/010 for supporting Martin Zábranský. We also thank Carlos V. Melo for editing the manuscript and Alexandros Paparakis for preparing 4-piperidinemethanol, which was used in the substrate scope. This research relied on computational resources of the High-Performance Computing Center (Centro de Computación de Alto Desempeño – CCAD) of the National University of Córdoba (Universidad Nacional de Cordoba – UNC) (https://ccad.unc.edu.ar/), part of the National System of High-Performance Computing (Sistema Nacional de Computación de Alto Desempeño – SNCAD) of the Ministry of Science, Technology and Innovation (Ministerio de Ciencia, Tecnología e Innovación – MinCyT), Argentina.

References

1. D. W. Stephan and G. Erker, Chem. Sci., 2014, 5, 2625–2641.
2. D. W. Stephan and G. Erker, Angew. Chem., Int. Ed., 2010, 49, 46–76.
3. D. W. Stephan, J. Am. Chem. Soc., 2021, 143, 20002–20014.
4. J. Lam, K. M. Szkop, E. Mosaferi and D. W. Stephan, Chem. Soc. Rev., 2019, 48, 3592–3612.
5. A. E. Ashley, A. L. Thompson and D. O'Hare, Angew. Chem., Int. Ed., 2009, 48, 9839–9843.
6. M. A. Courtemanche, A. P. Pulis, É. Rochette, M. A. Légaré, D. W. Stephan and F. G. Fontaine, Chem. Commun., 2015, 51, 9797–9800.
7. T. Wang, M. Xu, A. R. Jupp, Z. W. Qu, S. Grimme and D. W. Stephan, Angew. Chem., Int. Ed., 2021, 60, 25771–25775.
8. J. Lam, K. M. Szkop, E. Mosaferi and D. W. Stephan, Chem. Soc. Rev., 2019, 48, 3592–3612.
9. M. J. Ingleson, Annu. Rep. Prog. Chem., Sect. A. Inorg. and Phys. Chem., 2013, 109, 28–52.
10. S. A. Weicker and D. W. Stephan, Bull. Chem. Soc. Jpn., 2015, 88, 1003–1016.
11. D. M. C. Ould, J. L. Carden, R. Page and R. L. Melen, Inorg. Chem., 2020, 59, 14891–14898.
12. G. Ménard, L. Tran and D. W. Stephan, Dalton Trans., 2013, 42, 13685–13691.
13. N. Sarkar, R. Kumar Sahoo and S. Nembenna, Chem.–Eur. J., 2023, 29, e202203023.
14. C. Appelt, J. C. Slootweg, K. Lammertsma and W. Uhl, Angew. Chem., Int. Ed., 2013, 52, 4256–4259.
15. G. Ménard, J. A. Hatnean, H. J. Cowley, A. J. Lough, J. M. Rawson and D. W. Stephan, J. Am. Chem. Soc., 2013, 135, 6446–6449.
16. G. Ménard and D. W. Stephan, Angew. Chem., Int. Ed., 2012, 51, 4409–4412.
17. Y. Chen, W. Jiang, B. Li, G. Fu, S. Chen and H. Zhu, Dalton Trans., 2019, 48, 9152–9160.
18. C. Appelt, H. Westenberg, F. Bertini, A. W. Ehlers, J. C. Slootweg, K. Lammertsma and W. Uhl, Angew. Chem., Int. Ed., 2011, 50, 3925–3928.
19. M. Lange, J. C. Tendyck, P. Wegener, A. Hepp, E.-U. Würthwein and W. Uhl, Chem.–Eur. J., 2018, 24, 12856–12868.
20. D. Pleschka, M. Layh, F. Rogel and W. Uhl, Philos. Trans. R. Soc., A, 2017, 375, 20170011.
21. W. Uhl, P. Wegener, M. Layh, A. Hepp and E. U. Würthwein, Z. Naturforsch., B: Chem. Sci., 2016, 71, 1043–1050.
22. G. Ménard, T. M. Gilbert, J. A. Hatnean, A. Kraft, I. Krossing and D. W. Stephan, Organometallics, 2013, 32, 4416–4422.
23. P. Federmann, T. Bosse, S. Wolff, B. Cula, C. Herwig and C. Limberg, Chem. Commun., 2022, 58, 13451–13454.
24. P. Federmann, R. Müller, F. Beckmann, C. Lau, B. Cula, M. Kaupp and C. Limberg, Chem.–Eur. J., 2022, 28, e202200404.
25. S. Roters, C. Appelt, H. Westenberg, A. Hepp, J. C. Slootweg, K. Lammertsma and W. Uhl, Dalton Trans., 2012, 41, 9033–9045.
26. S. Styra, M. Radius, E. Moos, A. Bihlmeier and F. Breher, Chem.–Eur. J., 2016, 22, 9508–9512.
27. M. M. D. Roy, A. A. Omaña, A. S. S. Wilson, M. S. Hill, S. Aldrich and E. Rivard, Chem. Soc. Rev., 2021, 121, 12784–12965.
28. N. A. Shcherbina, A. V. Pomogaeva, A. S. Lisovenko, I. V. Kazakov, N. Yu. Gugin, O. V. Khoroshilova, Y. V. Kondrat'ev and A. Y. Timoshkin, Z. Anorg. Allg. Chem., 2020, 646, 873–881.
29. M. A. Courtemanche, J. Larouche, M. A. Légaré, W. Bi, L. Maron and F. G. Fontaine, Organometallics, 2013, 32, 6804–6811.
30. D. J. Scott, T. R. Simmons, E. J. Lawrence, G. G. Wildgoose, M. J. Fuchter and A. E. Ashley, ACS Catal., 2015, 5, 5540–5544.
31. T. Mahdi and D. W. Stephan, J. Am. Chem. Soc., 2014, 136, 15809–15812.
32. G. Ménard and D. W. Stephan, Angew. Chem., 2012, 124, 8397–8400.
33. J. Camacho-Bunquin, M. Ferrandon, U. Das, F. Dogan, C. Liu, C. Larsen, A. E. Platero-Prats, L. A. Curtiss, A. S. Hock, J. T. Miller, S. T. Nguyen, C. L. Marshall, M. Delferro and P. C. Stair, ACS Catal., 2017, 7, 689–694.
34. J. A. Hatnean, J. W. Thomson, P. A. Chase and D. W. Stephan, Chem. Commun., 2014, 50, 301–303.
35. H. Elsen, C. Färber, G. Ballmann and S. Harder, Angew. Chem., Int. Ed., 2018, 57, 7156–7160.
36. H. Elsen, J. Langer, M. Wiesinger and S. Harder, Organometallics, 2020, 39, 4238–4246.
37. H. Elsen, J. Langer, G. Ballmann, M. Wiesinger and S. Harder, Chem.–Eur. J., 2021, 27, 401–411.
38. Z. Qu, H. Zhu and S. Grimme, ChemCatChem, 2021, 13, 3401–3404.
39. J. Pölker, D. Schaarschmidt, J. Bernauer, M. Villa and A. Jacobi von Wangelin, ChemCatChem, 2022, 14, e202200144.
40. A. Friedrich, J. Eyselein, H. Elsen, J. Langer, J. Pahl, M. Wiesinger and S. Harder, Chem.–Eur. J., 2021, 27, 7756–7763.
41. F. Krämer, Angew. Chem., Int. Ed., 2024, 63, e202405207.
42. F. Krämer, J. Paradies, I. Fernández and F. Breher, Chem.–Eur. J., 2024, 30, e202303380.
43. P. Ju, J. Chen, A. Chen, L. Chen and Y. Yu, ACS Sustain. Chem. Eng., 2017, 5, 2516–2528.
44. J. Klankermayer, S. Wesselbaum, K. Beydoun and W. Leitner, Angew. Chem., Int. Ed., 2016, 55, 7296–7343.
45. C. Federsel, A. Boddien, R. Jackstell, R. Jennerjahn, P. J. Dyson, R. Scopelliti, G. Laurenczy and M. Beller, Angew. Chem., Int. Ed., 2010, 49, 9777–9780.
46. C. Ziebart, C. Federsel, P. Anbarasan, R. Jackstell, W. Baumann, A. Spannenberg and M. Beller, J. Am. Chem. Soc., 2012, 134, 20701–20704.
47. P. Daw, S. Chakraborty, G. Leitus, Y. Diskin-Posner, Y. Ben-David and D. Milstein, ACS Catal., 2017, 7, 2500–2504.
48. Z. Liu, Z. Yang, Z. Ke, X. Yu, H. Zhang, B. Yu, Y. Zhao and Z. Liu, New J. Chem., 2018, 42, 13933–13937.
49. Z. He, H. Liu, H. Liu, Q. Qian, Q. Meng, Q. Mei and B. Han, ChemCatChem, 2017, 9, 1947–1952.
50. H. Liu, Q. Mei, Q. Xu, J. Song, H. Liu and B. Han, Green Chem., 2017, 19, 196–201.
51. M. A. Affan and P. G. Jessop, Inorg. Chem., 2017, 56, 7301–7305.
52. L. Zhang, Z. Han, X. Zhao, Z. Wang and K. Ding, Angew. Chem., Int. Ed., 2015, 54, 6186–6189.
53. K. Zhang, L. Zong and X. Jia, Adv. Synth. Catal., 2021, 363, 1335–1340.
54. A. A. Dabbawala, N. Sudheesh and H. C. Bajaj, Indian J. Chem., Sect. A:Inorg., Bio-inorg., Phys., Theor. Anal. Chem., 2015, 54A, 753–756.
55. K. Kudo, H. Phala, N. Sugita and Y. Takezaki, Chem. Lett., 1977, 6, 1495–1496.
56. Y. Zhang, J. Wang, H. Zhu and T. Tu, Chem.–Asian J., 2018, 13, 3018–3021.
57. S. Schreiner, J. Y. Yu and L. Vaska, J. Chem. Soc., Chem. Commun., 1988, 602–603.
58. S. Schreiner, J. Y. Yu and L. Vaska, Inorg. Chim. Acta, 1988, 147, 139–141.
59. M. Minato, D. Y. Zhou, K. I. Sumiura, R. Hirabayashi, Y. Yamaguchi and T. Ito, Chem. Commun., 2001, 1, 2654–2655.
60. Y. Qin, H. Guo, X. Sheng, X. Wang and F. Wang, Green Chem., 2015, 17, 2853–2858.
61. M. A. Courtemanche, A. P. Pulis, É. Rochette, M. A. Légaré, D. W. Stephan and F. G. Fontaine, Chem. Commun., 2015, 51, 9797–9800.
62. M. Hulla and P. J. Dyson, Angew. Chem., Int. Ed., 2020, 59, 1002–1017.
63. A. Paparakis, R. C. Turnell-Ritson, J. S. Sapsford, A. E. Ashley and M. Hulla, Catal. Sci. Technol., 2023, 13, 637–644.
64. U. Jayarathne, N. Hazari and W. H. Bernskoetter, ACS Catal., 2018, 8, 1338–1345.
65. L. L. Zeonjuk, N. Vankova, A. Mavrandonakis, T. Heine, G. Röschenthaler and J. Eicher, Chem.–Eur. J., 2013, 19, 17413–17424.
66. L. Liu, B. Lukose, P. Jaque and B. Ensing, Green Energy Environ., 2019, 4, 20–28.
67. D. W. Stephan and G. Erker, Angew. Chem., Int. Ed., 2010, 49, 46–76.
68. F. Ebner, H. Wadepohl and L. Greb, J. Am. Chem. Soc., 2019, 141, 18009–18012.
69. K. Sordakis, C. Tang, L. K. Vogt, H. Junge, P. J. Dyson, M. Beller and G. Laurenczy, Chem. Rev., 2018, 118, 372–433.
70. S. Ishida, T. Hatakeyama, T. Nomura, M. Matsumoto, K. Yoshimura, S. Kyushin and T. Iwamoto, Chem.–Eur. J., 2020, 26, 15811–15815.
71. M. Hulla, G. Laurenczy and P. J. Dyson, ACS Catal., 2018, 8, 10619–10630.
72. G. Gastelu, D. Savary, M. Hulla, D. Ortiz, J. G. Uranga and P. J. Dyson, ACS Catal., 2023, 13, 2403–2409.
73. A. E. Finholt and E. C. Jacobson, J. Am. Chem. Soc., 1952, 74, 3943–3944.

---

Footnote

† Electronic supplementary information (ESI) available. CCDC 2357273, 2357274 and 2386191. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4sc03665d

---