Leveraging Conformational Dynamics of MOFs to Overcome Conversion‐Selectivity Trade‐Off in CO₂ Hydrogenation to Ethanol

ABSTRACT

In this study, we exploit the structural dynamics of MIL‐53(Al) metal‐organic framework (MOF) to develop a highly efficient and selective aqueous‐phase heterogeneous catalyst, [MIL‐53‐Co(OH)], featuring atomically dispersed Co^II(OH) active species at the MOF's nodes for the hydrogenation of CO₂ into ethanol. Under mild conditions (130°C, 20 bar, H₂/CO₂: 3), MIL‐53‐Co(OH) achieves 92% CO₂ conversion with an ethanol productivity of 10 133 µmol g_cat ⁻¹ h⁻¹ and 93% selectivity. Comparisons with the rigid MIL‐68(Al) and breathing‐suppressed variants of MIL‐53 MOFs revealed that the lattice flexibility of MIL‐53‐Co(OH) enhances ethanol productivity by at least 3.5‐fold while suppressing the formation of CH₃OH. Experimental, structural and computational analysis suggest that the reversible narrow‐pore (np)↔large‐pore (lp) interconversion of MIL‐53‐Co(OH) periodically optimizes the Al₂[μ ₃‐O–Co(OH)] active‐site geometry at MOF's node, which synchronizes in situ generated CH₃OH activation followed by CO insertion to promote C‒C coupling. During the np→lp transition, the transient lattice expansion relaxes the Al–[μ₃ ‐O–Co(OH)]–Al hinge, which reduces the activation barrier of σ‐bond metathesis between the Co–H bond and C–O bond of CH₃OH, a key step in the catalytic cycle. This strategy of leveraging conformational dynamics of MOFs for active‐site engineering opens new avenues in designing highly active earth‐abundant metal catalysts for challenging chemical transformations.

We report the effect of structural dynamics of breathable MIL‐53(Al) MOF on modulating mononuclear Co^II(OH) active sites for boosting CO₂ hydrogenation into ethanol. Under mild conditions (130°C, 20 bar, H₂/CO₂: 3), MIL‐53‐Co(OH) catalyst achieves 92% CO₂ conversion with 93% ethanol selectivity, and a productivity of 10,133 µmol g_cat ⁻¹ h⁻¹.

1. Introduction

The hydrogenation of carbon dioxide (CO₂), a thermodynamically stable and abundant C₁ feedstock, into value‐added chemicals is a pivotal process in sustainable chemistry. This transformation facilitates carbon neutrality and renewable energy storage by converting CO₂ into energy‐rich molecules, thereby reducing greenhouse gas emissions and promoting sustainable chemical manufacturing when combined with renewable hydrogen sources. While significant progress has been made in the hydrogenation of CO₂ to C₁ products such as CO [1, 2], CH₄ [3, 4], and CH₃OH [5, 6], the selective synthesis of C₂+ products, particularly ethanol, remains a significant challenge [7]. Various catalysts have been investigated for this transformation, including noble metal catalysts such as Ir‐In₂O₃ [8], Ru‐In₂O₃ [9], Rh/Fe‐SiO₂ [10], and Pd/Fe₃O₄ [11], as well as earth‐abundant metal alternatives such as CoAlO_x‐600 [12], Co_0.52Ni_0.48AlO_x [13], 4.6K‐CMZF [14], Cu/Na‐Beta [15], Na‐Co₂C [16], and FeNaS‐0.6 [17]. Although these catalysts have shown promise for direct CO₂ hydrogenation to ethanol, their productivity remains fundamentally limited by the substantial kinetic and thermodynamic barriers associated with C–C bond formation, which requires the precise orchestration of multi‐electron/proton transfer processes. As a result, most catalysts face a trade‐off between CO₂ conversion and ethanol selectivity and often rely on high‐temperature pre‐activation and tend to favor the production of kinetically favored C₁ products or engage in uncontrolled hydrocarbon homologation.

Metal‐organic frameworks (MOFs) are versatile platforms used in various catalytic transformations [18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28] including CO₂ hydrogenation [29, 30, 31] due to their crystalline structure, modular porosity, and well‐defined metal‐ligand environments that stabilize active catalytic sites within rigid scaffolds. Recent studies have shown that MOFs enable precise control over CO₂ hydrogenation through atomically dispersed active metal sites, bifunctional node environments, and cooperative catalytic motifs that regulate key elementary steps [32, 33]. Within the confined environment of these frameworks, the stabilization of key reaction intermediates (methoxy, acyl, *CO, and *CHOH) is crucial for modulating the oxidation states of active sites and enforcing the spatial isolation required to facilitate C–C coupling toward ethanol [34, 35]. Furthermore, confinement effects and cooperative multisite interactions, including secondary‐sphere and enzyme inspired functionalities found in porphyrinic MOFs, have been shown to enhance CO₂ enrichment, stabilize *CO intermediates, and reduce the barriers to C–C coupling, as supported by experimental and theoretical studies [36, 37, 38, 39]. Despite recent advances in developing MOF‐catalysts for CO₂ to ethanol conversion through single‐site grafting [32, 33], MOF‐nanoparticle hybridization [40, 41], defect engineering [42, 43], linker functionalization [44], and alkali promotion [45], most of these catalysts still suffer from low turnover, rapid deactivation, poor stability and selectivity. Recently, a zirconium MOF catalyst bearing Zr₁₂‐oxo node‐supported Cu(I) species, stabilized by Cs⁺, produced ethanol in 99% selectivity with 96% CO₂ conversion [45]. However, this catalytic system requires anaerobic conditions and THF as solvent, which limits its practical applicability. Nevertheless, all these MOF‐catalysts rely on structurally rigid frameworks, wherein the static coordination environment precludes the dynamic modulation of the active‐site geometry. While catalysis within rigid porous frameworks is well‐studied, the influence of conformational dynamics in flexible porous materials on catalytic reactions remains relatively underexplored. Emerging examples demonstrate that structural adaptability through hemilabile coordination, photo‐switchable linkers, or host‐guest induced flexibility can actively regulate site accessibility, diffusion, and intermediate stabilization, thereby adding a dynamic dimension to catalytic control (Figure 1a–c) [46, 47, 48, 49, 50, 51]. We envision that integrating catalytically active single‐site earth‐abundant metal species into a flexible or breathable MOF scaffold would offer a synergistic effect: atomically dispersed active sites provide optimized geometry for CO₂ activation and H₂ heterolysis, while framework dynamics orchestrates reaction sequences en route to ethanol via C₁ homologation.

FIGURE 1.

(a–c) Schematic representations of the reported approaches to control catalysis via conformational adaptivity in porous materials. (d) Modulation of Co^II(OH) active sites at MIL‐53(Al) nodes through conformational dynamics of flexible 3D MOF frameworks, enabling selective CO₂ hydrogenation into ethanol.

In this study, we present a “breathing” MIL‐53(Al) MOF‐confined single‐site cobalt(II) hydroxyl catalyst [MIL‐53‐Co(OH)] for efficient and selective hydrogenation of CO₂ to ethanol in water. MIL‐53(Al), a commercially available MOF (BasoliteA100), is distinguished by its structural flexibility, undergoing reversible transformations between narrow‐pore (np) and large‐pore (lp) morphologies in response to external stimuli such as temperature, pressure, and guest molecules such as CO₂, H₂O, and H₂ [52, 53, 54, 55]. We demonstrate that tuning the coordination environment of the cobalt active sites at MOF's nodes by conformational dynamics of the MOF frameworks under catalytic reaction conditions facilitates C–C bond coupling within the micropores, while concurrently promoting the local reactant enrichment and product desorption, allowing MIL‐53‐Co(OH) to achieve 92% CO₂ conversion in water with an ethanol productivity of 10 133 µmol g_cat ⁻¹ h⁻¹ in 93% selectivity (Figure 1d).

2. Results and Discussion

2.1. Synthesis and Structural Elucidation of MIL‐53‐Co(OH) and Derivatives

MIL‐53(Al) MOF and its two functionalized derivatives, MIL‐53‐NO₂ and MIL‐53‐NH₂, were synthesized via a general synthetic procedure, involving a solvothermal reaction between Al(NO₃)₃.9H₂O and the corresponding dicarboxylate linkers in water or DMF/H₂O mixtures at 150°C–220°C [56]. Powder x‐ray diffraction (PXRD) revealed that all three MOFs crystallize in the characteristic MIL‐53 sra topology, with Al(OH) nodes linked by terephthalate or functionalized terephthalate ligands, forming 1D rhombic channels. Post‐synthetic modification at the μ ₂‐OH groups of the secondary building units (SBUs) was achieved via sequential treatment with n‐BuLi, CoCl₂, NaEt₃BH, and finally H₂O, affording uniformly dispersed Co^II(OH) moieties anchored at the MOF's nodes. PXRD of the cobalt‐functionalized MOFs showed that MIL‐53‐Co(OH) exists primarily in the lp form, while the functionalized analogues, MIL‐53‐NO₂‐Co(OH) and MIL‐53‐NH₂‐Co(OH), coexisted in lp and np phases (Figure 2a,b). The Co‐loading relative to the μ ₂‐OH sites in MIL‐53‐Co(OH), MIL‐53‐NO₂‐Co(OH), and MIL‐53‐NH₂‐Co(OH) was 38%, 36%, and 42%, respectively, as determined by inductively coupled plasma‐optical emission spectroscopy (ICP‐OES) of the digested MOFs. Brunauer‐Emmett‐Teller (BET) analysis revealed that MIL‐53‐Co(OH), MIL‐53‐NH₂‐Co(OH), and MIL‐53‐NO₂‐Co(OH) possess specific surface areas of 900, 534, and 348 m² g⁻¹, respectively (Figure 2c). Scanning electron microscopy coupled with energy‐dispersive x‐ray spectroscopy (SEM‐EDX) mapping confirmed a uniform distribution of Al and Co throughout the MIL‐53‐Co(OH) crystallites (Figures S6 and S7). X‐ray photoelectron spectroscopy (XPS) confirmed the +2‐oxidation state of Co ions in all three variants (Figure 2d). This assignment was further corroborated by x‐ray absorption near edge structure (XANES) analysis, where the Co K‐edge of MIL‐53‐Co(OH) aligned with that of CoO at 7719.7 eV, confirming the presence of Co^II species (Figure 2e).

FIGURE 2.

(a) PXRD patterns of simulated MIL‐53‐lp (black), simulated MIL‐53‐np (red), pristine MIL‐53 (blue), MIL‐53‐CoCl (green), MIL‐53‐Co(OH) (magenta), and MIL‐53‐Co(OH) after catalysis (brown). (b) PXRD patterns of simulated MIL‐53‐lp (black), simulated MIL‐53‐np (red), pristine MIL‐53‐NO₂ (blue), MIL‐53‐NO₂‐CoCl (green), MIL‐53‐NO₂‐Co(OH) (magenta), pristine MIL‐53‐NH₂ (brown), MIL‐53‐NH₂‐CoCl (orange), and MIL‐53‐NH₂‐Co(OH) (gray). (c) N₂ sorption isotherm of MIL‐53‐Co(OH) (red), MIL‐53‐NH₂‐Co(OH) (blue), and MIL‐53‐NO₂‐Co(OH) (black) measured at 77 K. (d) Co 2p XPS spectra of MIL‐53‐Co(OH), MIL‐53‐NO₂‐Co(OH), and MIL‐53‐NH₂‐Co(OH). (e) Co K‐edge XANES spectra of MIL‐53‐Co(OH) (green), MIL‐53‐Co after catalysis (blue), CoO (red), and Co foil (black). (f) DFT‐optimized structure of Al₂[μ ₃‐O–Co(OH)] active site in MIL‐53 (lp). Hydrogen atoms are omitted for clarity. (g) EXAFS spectra (black and blue squares) and fits (black and blue lines) of MIL‐53‐Co(OH) in the R space from 1 to 3.25 Å. (h) k ²‐weighted Co‐EXAFS χ(R) spectra of MIL‐53‐Co(OH) (red), MIL‐53‐Co after catalysis (black), CoO (green), and Co foil (blue). (i) Wavelet transform for the k ²‐weighted EXAFS signal at the Co K‐edge.

The extended x‐ray absorption fine structure (EXAFS) fitting of MIL‐53‐Co(OH) with the DFT‐optimized model of the cobalt active species in MIL‐53‐Co(OH) revealed the presence of cobalt in the lp phase of MIL‐53, with a tetrahedral coordination environment comprising of one hydroxyl ligand (Co‒OH: 1.84 Å), one μ‐oxo bridge (μ₃ ‐O–Co: 1.98 Å), and two carboxylate oxygen atoms (Co‒O_carboxylate: 2.32, 2.35 Å) (Figure 2g). The EXAFS fitting of MIL‐53‐NO₂‐Co(OH) and MIL‐53‐NH₂‐Co(OH) however revealed the existence of tetrahedral Co^II species in the np phase of MIL‐53, coordinated with one hydroxyl ligand (Co‒OH: 1.82 Å), one μ‐oxo (μ₃ ‐O–Co: 1.94 Å), and two carboxylate oxygen atoms (Co‒O_carboxylate: 2.20, 2.57 Å) (Figures S14 and S15). The phase‐uncorrected k ²‐weighted EXAFS spectrum of MIL‐53‐Co(OH) displayed a primary Co•••O scattering feature at ∼1.53 Å, consistent with Co^II–O bonding observed in CoO (Figure 2h). Notably, no significant Co•••Co or Co••O••Co scattering features were detected; only weak signals appeared at ∼2.2 and ∼2.7 Å, ruling out the presence of metallic cobalt(0) or CoO aggregates in MIL‐53‐Co(OH). This conclusion was further validated by quantitative EXAFS fitting, which showed poor agreement when models included 5%–50% contributions from metallic cobalt or cobalt oxide species (Figure S16). Additionally, wavelet transform (WT) analysis exhibited a single intensity maximum at k∼4.3 Å⁻¹, characteristic of Co–O coordination, with no high‐k contour intensity at k∼6.5 Å⁻¹, thereby excluding the presence of neighboring Co•••Co interactions (Figure 2i). These results collectively confirm the existence of atomically dispersed single‐site Co^II(OH) active species at MIL‐53 nodes.

2.2. MIL‐53‐Co(OH) Catalyzed Hydrogenation of CO₂

Catalytic testing was performed in a high‐pressure stainless steel batch reactor with a total free volume of 65 mL, using 15 mL of H₂O as the solvent and 5 mg of MIL‐53‐Co(OH) catalyst (8 µmol Co). Upon completion of the reaction, liquid‐phase products were identified and quantified by gas chromatography coupled (GC) with mass spectrometry and flame ionization detector (GC‐MS/FID), while gaseous products were analyzed by a GC equipped with a thermal conductivity detector (GC‐TCD). Under a total pressure of 20 bar (P_H2/P_CO2: 3) at 130°C over 15 h, the system produced 0.58 mmol of ethanol with a selectivity of 98.3%, corresponding to a productivity of 7733 µmol g_cat ⁻¹ h⁻¹ (Entry 1, Table 1). No detectable amounts of C₂+ hydrocarbons, methanol, or other oxygenates were observed; methane was the only measurable byproduct, formed in trace amounts (0.01 mmol). Time‐resolved monitoring confirmed a steady accumulation of ethanol, increasing from 0.49 mmol at 12 h to 0.72 mmol at 21 h, although this was accompanied by a decline in selectivity from 98% to 90% due to progressive methane build‐up of 0.07 mmol together with minor methanol formation (Entries 7–8, Table S3). Elevating the total pressure to 32 bar further enhanced ethanol yield to 0.77 mmol but resulted in decreased selectivity and CO₂ conversion values of 87% and 8%, respectively (Entry 3, Table 1). Adjusting the P_H2/P_CO2 ratio from 3 to 4 improved CO₂ conversion to 12.6%, yet ethanol yield decreased to 0.31 mmol, with acetic acid emerging as the dominant product at 0.64 mmol (Entry 4, Table 1). Temperature‐dependent studies revealed a gradual increase in CO₂ conversion between 100 and 150°C, with higher temperatures promoting acetic acid formation and lowering ethanol selectivity to 76%, indicating that the reaction reaches maximum ethanol productivity at 130°C (Entries 5–6, Table S3). Importantly, the reaction in organic solvents or without water did not produce ethanol, suggesting that water is essential for catalytic turnover.

TABLE 1.

Optimization of reaction conditions for MIL‐53‐Co(OH) catalyzed CO₂ hydrogenation.

Entry	Catalyst	P (Bar)	Temperature (°C)	n (mmol) a			EtOH Selectivity (%)	CO2 Conversion (%)
				EtOH	MeOH	CH4
1	MIL‐53‐Co(OH)	20	130	0.58	0	0.01	98.3	8.9
2	No catalyst	20	130	0	0	—	—	—
3	MIL‐53‐Co(OH)	32	130	0.77	0.01	0.11	86.5	7.9
4	MIL‐53‐Co(OH)	30 b	130	0.31	0.08	0.01	29.8	12.6 c
5	MIL‐53‐Co(OH)	20	100	0.48	0.09	0.02	81.3	8.16
6	MIL‐53‐Co(OH) d	20	130	1.52	0.03	0.09	92.6	92.1

Note: Standard reaction conditions: 5 mg MIL‐53‐Co(OH) (8 µmol of Co), H₂/CO₂: 3, P: 20 bar, time: 15 h, 15 mL H₂O, 130°C in high pressure batch reactor (free volume: 65 mL).

^aThe liquid product yields were quantified by GC‐FID using ethylene glycol as an internal standard, whereas gaseous products were analyzed by GC‐TCD.

^bH₂/CO₂: 4.

^c0.64 mmol of CH₃CO₂H was obtained as a product.

^d10 mg of MIL‐53‐Co(OH) (16 µmol of Co) was used in high‐pressure batch reactor (free volume: 17 mL).

To enhance catalytic efficiency, a compact stainless‐steel batch reactor with a reduced free volume of 17 mL was employed, and the MIL‐53‐Co(OH) catalyst loading was doubled to 10 mg (16 µmol of Co). These modifications were grounded in thermodynamic and kinetic principles in multiphase catalysis, where gas‐liquid equilibrium follows Henry's law and gas solubility scales with partial pressures; thus, reducing headspace volume at constant total pressure (20 bar) increases effective dissolved gas concentration in the liquid phase, enhancing gas solubility and mass transfer, especially important given CO₂’s low Henry's constant in aqueous media [57, 58]. Concurrently, doubling the catalyst amount increases the number of Co(OH) active sites and the volumetric site density, shortening the time to full conversion. Within confined volumes, increased collisions between dissolved reactants and catalytic sites further enhance kinetics, especially under diffusion‐limited conditions [58]. The combined effects of enhanced mass transfer, higher reactant concentration, and augmented catalyst surface area led to a kinetically superior system, achieving 1.52 mmol CH₃CH₂OH yield and 92% CO₂ conversion under optimized conditions (20 bar, 130°C, 15 h; Entry 6, Table 1). CH₃CH₂OH selectivity slightly declined to 93%, accompanied by trace amounts of CH₃OH (0.03 mmol) and CH₄ (0.09 mmol). The time‐averaged turnover frequency (TOF), defined as the mmol of ethanol produced per mmol of cobalt per hour is ∼6.3 h⁻¹ under the optimized conditions (Table S7). This value reflects the average rate across the entire reaction duration, naturally encompassing the substrate depletion observed at 92% CO₂ conversion. The MIL‐53‐Co(OH) catalyst exhibited a high productivity of 10,133 µmol g_cat ⁻¹ h⁻¹, achieving yields comparable to reported systems while operating under aqueous, mild, and pre‐activation free conditions, and simultaneously overcoming the conventional trade‐off between ethanol selectivity and CO₂ conversion efficiency (Table S18).

MIL‐53‐Co(OH) achieved 92% CO₂ conversion and 93% ethanol selectivity at 130°C in aqueous phase, outperforming most previously reported base‐metal thermocatalysts, even while operating under relatively mild conditions (Figure 3a). The ¹H‐NMR spectrum (D₂​O) and the GC‐MS analysis of the crude reaction mixture confirmed the formation of ethanol (Figure 3b). No ethanol was detected in control experiments conducted without MIL‐53‐Co(OH) or when CO₂ was replaced with N₂, thereby verifying that CO₂ was indeed the carbon source (Entry 2, Table S3, Section S.3.4). Post‐catalysis analyses confirmed the stability of both the framework and the active sites. PXRD confirmed the retention of MOF crystallinity, while EXAFS suggests the existence of [(μ ₃‐O)(carboxylate‐O)₂Co^II(OH)] species in MIL‐53‐Co after catalysis (Figures 2a and 3c). The absence of any minor quantities of Co(0) and CoO in the recovered MIL‐53‐Co after catalysis was further confirmed by EXAFS combinatorial fittings incorporating 5%–50% of these species (Figure S18). Importantly, MIL‐53‐Co(OH) demonstrated excellent recyclability, maintaining high activity over at least eight cycles, showing only minimal loss in productivity while achieving > 96% ethanol selectivity (Figure 3d).

FIGURE 3.

(a) Comparison of catalytic activity for CO₂ hydrogenation over MIL‐53‐Co(OH) and other earth‐abundant catalysts. Catalysts marked with an asterisk have not been reported with selectivity or conversion values. (b) ¹H‐NMR spectrum of the crude reaction mixture after MIL‐53‐Co(OH) catalyzed CO₂ hydrogenation and GC‐MS spectrum of the ethanol product. (c) EXAFS spectra (black and blue squares) and fits (black and blue lines) of MIL‐53‐Co after catalysis in the R space from 1 to 3.25 Å. (d) Recyclability test for MIL‐53‐Co(OH) catalyzed CO₂ hydrogenation. Reaction conditions: 5 mg MIL‐53‐Co(OH) (8 µmol of Co), 20 bar (H₂/CO₂: 3), 130°C, 15 h, 15 mL H₂O in high pressure batch reactor (free volume: 65 mL). (e) Comparison of catalytic activity of MIL‐53‐Co(OH) for CO₂ hydrogenation with other materials. (f) Control experiments for elucidation of reaction intermediates/pathways in MIL‐53‐Co(OH) catalyzed CO₂ hydrogenation to ethanol. (g) GC‐MS spectrum of ethanol obtained from MIL‐53‐Co(OH) catalyzed ¹³CH₃OH homologation using CO and H₂. (h) Comparison of MIL‐variants for methanol homologation to ethanol using CO and H₂.

2.3. Control Experiments and Mechanistic Investigation

Having established the stability and activity of MIL‐53‐Co(OH) catalyst, we focused on understanding the mechanistic role of the atomically dispersed Co^II(OH) moieties. Control experiments using the pristine, non‐metallated MIL‐53 framework and Co(0) showed only trace amounts of methanol and no detectable ethanol. This finding highlights the necessity of site‐isolated Co^II(OH) species at the MOF nodes for efficient catalysis (Figure 3e). Notably, Co(0)@MIL‐53 produced only trace ethanol, with methanol as the dominant product. Comparative tests with bulk Co(OH)₂, CoCl₂, and Co supported on Al₂O₃ also yielded no catalytic activity, emphasizing the unique role of atomically dispersed Co(OH) moieties within the MIL‐53 framework. To delineate the operative pathway for ethanol formation, we conducted a series of critical mechanistic control experiments. Under standard catalytic conditions, a CO₂ + H₂ feed yielded the highest ethanol production, establishing direct CO₂ hydrogenation as the dominant and most efficient pathway (Figure 3f). In contrast, when we fed 0.6 mmol of CH₃OH under the same conditions, we observed zero ethanol yield, which demonstrates that free methanol in the bulk phase is not a competent intermediate for C–C coupling and may even passivate the active sites (Entry 8, Table S4). A pivotal insight emerged from the CH₃OH + CO + H₂ experiment, which yielded only 0.035 mmol of ethanol, an order of magnitude lower than the CO₂ + H₂ benchmark, indicating that while methanol can enter the chain‐growth pathway when co‐activated with CO under hydrogenation conditions, its dominant effect remains site passivation, thereby underscoring the critical role of confinement in enabling selective C–C coupling (Figure 3f). Critically, while a CH₃OH + CO + H₂ feed produced ethanol, a CH₃OH + CO₂ + H₂ mixture was entirely inactive. This difference unequivocally identifies CO, not CO₂, as the essential C₁ coupling partner for the surface‐bound methyl group derived from methanol. Complementing the observation, a control experiment using isotopically labeled ¹³CH₃OH + CO + H₂ produced ¹³CH₃CH₂OH, as confirmed by GC‐MS, directly demonstrating that surface‐bound methanol‐derived methyl groups can engage in C–C bond formation at the Co(OH) sites (Figure 3g). Finally, the catalytic inertness of other potential C₁ feeds, including CH₄ + CO₂ + H₂, CH₄ + CO, and HCOOH + CO + H₂, ruled out methane homologation and formic acid‐derived pathways as viable routes for ethanol production from CO₂ hydrogenation (Figure 3f) [59, 60, 61].

2.4. Proposed Reaction Pathway for CO₂ Hydrogenation to Ethanol

Based on the above experimental findings, we propose a plausible mechanistic sequence for ethanol formation over MIL‐53‐Co(OH). The overall transformation is conceptualized as three interlinked sub‐cycles: (i) the generation of CO via reverse water gas shift (RWGS) reaction (Cycle‐1), (ii) the synthesis of methanol (Cycle‐2), and (iii) the subsequent C–C coupling to ethanol (Cycle‐3); all operating catalytically at the well‐defined Co^II(OH) nodes of MIL framework. Catalysis is initiated by heterolytic activation of H₂ at the Co–OH sites, forming a cobalt(II) hydride species (INT‐1), along with the release of a water molecule. The Co–H intermediate subsequently undergoes hydride transfer to CO₂ to generate a cobalt formate species (Co–OCHO, INT‐2). From this stage, the catalytic network bifurcates into two interconnected reduction channels. In one pathway, partial decarboxylation of the formate releases CO, thus forming a CO reservoir (Figure 4a) [62]. In the parallel competing channel, INT‐2 participates in a sequence of successive reduction steps that progressively reduce the carbon center (INT‐2 → INT‐3 → INT‐4), to a Co–O–CH₃ species (Figure 4b) [63, 64]. This cobalt methoxy species ultimately undergoes hydrolysis to release methanol while regenerating the Co–OH active sites. The formation of ethanol likely arises from the convergence of these two pathways (Figure 4c). Herein, methanol reacts with Co–H species via σ‐bond metathesis to generate a Co–CH₃ species (INT‐5). The coordination of in situ generated CO to the Co^II ion, followed by migratory insertion forms the Co‐(COCH₃) species (INT‐6) [65, 66], which subsequently undergoes H₂ reduction to give INT‐7. Finally, hydrolysis of INT‐7 liberates the ethanol and regenerates MIL‐53‐Co(OH). Notably, the overall stoichiometry of the transformation (2 CO₂ + 6H₂ → C₂H₅OH + 3 H₂O) inherently requires a 3:1 H₂/CO₂ molar ratio. The applied pressure ratio of 15 bar H₂ to 5 bar CO₂ thus not only matches the stoichiometric demand but also provides the kinetic driving force to accelerate hydrogenation steps, suppress CO over‐accumulation, and sustain the pool of CO₂‐derived intermediates essential for C–C coupling. This synergy between reaction stoichiometry and optimized pressure conditions underscores the role of Co–OH motifs as bifunctional sites for CO₂ activation, H₂ dissociation, and selective C–C bond formation to ethanol. Ethanol was not produced in the reaction without water, which suggests that water is presumably involved in the catalytic cycle, particularly in the hydrolytic steps associated with product release and regeneration of the Co–OH active sites (Figure 4b,c). Furthermore, water may act as a thermodynamic sink for continuous product removal to the liquid phase. As ethanol forms within the confined MOF pores, its high miscibility in water likely promotes rapid diffusion into the bulk aqueous phase, maintaining its low local concentration at the Co active sites and thereby continuously driving the reaction equilibrium toward the product formation [67]. Given its strong interaction with MIL‐53, water may additionally modulate framework dynamics, thus the loss of activity under dry conditions suggests coupled catalytic and structural effects, consistent with water‐assisted breathing being integral to active‐site function.

FIGURE 4.

(a–c) Proposed reaction pathways for MIL‐53‐Co(OH) catalyzed CO₂ hydrogenation to ethanol.

2.5. Effect of Framework Dynamics on Active Site Structure and C–C Coupling

To elucidate the influence of framework dynamics on catalytic performance, we systematically compared the Co(OH)‐functionalized MIL‐53 derivatives. The incorporation of –NH₂ and –NO₂ substituents on the BDC linker suppresses the intrinsic flexibility of MIL‐53 to varying degrees, with the extent of breathing generally following the trend: MIL‐53 > MIL‐53‐NO₂ > MIL‐53‐NH₂, based on the previous studies (Figure 5a) [56, 68]. Despite identical cobalt‐loadings and analogous local coordination environments as confirmed by EXAFS, ethanol productivity and selectivity exhibited a pronounced dependence on framework flexibility (Figure 5b,e). Catalytic evaluation under standard conditions (5 bar CO₂, 15 bar H₂, 15 mL H₂O, 130°C, 15 h) revealed a linear correlation of ethanol productivity with framework flexibility: MIL‐53‐NO₂‐Co(OH) and MIL‐53‐NH₂‐Co(OH) exhibited 83% and 67% CH₃CH₂OH selectivity, respectively, in contrast to 98% in MIL‐53‐Co(OH) (Figure 5e). Concurrently, CH₃OH selectivity increased with decreasing breathing capacity (0% →14% →30%), implying that restricted lattice dynamics hinder C–C coupling required for C₁ homologation to ethanol. Building on insights from MIL‐53 analogues, MIL‐68‐Co(OH): a rigid, non‐breathable structural polymorph of MIL‐53(Al); was synthesized via post‐synthetic modification of the pristine MIL‐68 MOF (Figure 5c, Section S.2.7) [69]. The Co^II ion in MIL‐68‐Co(OH) also adopts a tetrahedral coordination environment analogous to that in MIL‐53‐Co(OH), forming [(μ ₃‐O)(carboxylate‐O)₂Co^II(OH)] species at MOF's nodes, as confirmed by EXAFS analysis (Figure 5d). Under identical cobalt‐loading and standard reaction conditions (5 bar CO₂, 15 bar H₂, 15 mL H₂O, 130°C), MIL‐68‐Co(OH) was ∼3.5 times less active than MIL‐53‐Co(OH), with a significantly lower CH₃CH₂OH selectivity of 55% compared to 98% in the flexible MIL‐53 framework (Figure 5f). Simultaneously, CH₃OH selectivity increased from 0% in MIL‐53‐Co(OH) to 40%, which suggests a framework‐induced shift in product distribution. These experiments likely suggest that the flexible framework of MIL‐53‐Co(OH) is key to its high ethanol yield, as it dynamically reshapes the cobalt active sites during CO₂ hydrogenation.

FIGURE 5.

(a) Effect of functional group modification on MIL‐53(Al) MOF‘s breathing capacity, with the highly breathable MIL‐53‐Co(OH) compared to low breathable MIL‐53‐NO₂‐Co(OH) and MIL‐53‐NH₂‐Co(OH). (b) EXAFS spectra (black and green squares) and fits (black and green lines) of MIL‐53‐NO₂‐Co(OH) in the R space from 1 to 3 Å. (c) Structural comparison of breathable MIL‐53‐Co(OH) and rigid non‐breathing MIL‐68‐Co(OH) frameworks, illustrating their distinct pore flexibility yet similar Co(OH) coordination environments. (d) EXAFS spectra (black and green squares) and fits (black and green lines) of MIL‐68‐Co(OH) in the R space from 1 to 3.25 Å. (e) Effect of linker functionalization on catalytic performance of MIL‐53‐functionalized Co(OH) catalysts. (f) Effect of framework flexibility on catalytic performance of MIL‐53‐Co(OH) and MIL‐68‐Co(OH) catalysts. Reaction conditions: MIL‐Co(OH) (8 µmol of Co), 20 bar (H₂/CO₂: 3), 130°C, 15 h, 15 mL H₂O.

To elucidate the mechanistic origin of correlation of ethanol selectivity and framework breathing, we performed the CH₃OH + CO + H₂ control experiment with analogues of MIL‐53. A linear correlation of methanol conversion and ethanol production was observed with the extent of framework flexibility. MIL‐53‐Co(OH), which dynamically accesses the lp state, produced highest amount of ethanol, while MIL‐53‐NO₂‐Co(OH) and MIL‐53‐NH₂‐Co(OH), where breathing is progressively damped, show attenuated C–C coupling; and rigid MIL‐68‐Co(OH) was completely inactive (Figure 3h). Density functional theory (DFT) calculations were performed using the B3LYP functional with GD3BJ dispersion corrections and the def2‐SVP basis set, incorporating implicit solvation effects via the PCM model for water at 403 K, to compute free‐energy profiles for the hydrogenation of CO₂ to CH₃CH₂OH along Cycles 1–3 in both the narrow‐pore (np) and large‐pore (lp) conformations of MIL‐53‐Co (Figure S10) [32, 70, 71]. The DFT‐derived free‐energy profiles reveal that, for Cycles 1 and 2, the activation barriers remain largely unchanged between the np and lp phases, indicating that framework breathing has minimal influence on the energetics of the early hydrogenation steps. In the methanol homologation sequence, the initial formation of the Co–H species via TS‐1 proceeds with comparable barriers in both states (35.0 vs. 34.0 kcal mol⁻¹), indicating that hydride generation is not conformationally discriminating (Figure 6). In sharp contrast, the subsequent σ‐bond metathesis between Co–H and CH₃OH, which furnishes the surface Co–CH₃ intermediate, exhibits a marked dependence on pore geometry. The activation barrier for TS‐3 is substantially lower in the lp state (45.4 kcal mol⁻¹) than in the np state (63.8 kcal mol⁻¹), because the expanded lp lattice permits hinge‐like relaxation of the Al–[μ₃ ‐O–Co(OH)]–Al linkage, enabling the Co–H σ‐bond to align with the σ⃰ (C–O) antibonding orbital of methanol. This alignment maximizes donor‐acceptor overlap and minimizes coordination‐sphere distortion. By contrast, the compressed np conformation restricts angular reorganization at the Al–[μ₃ ‐O–Co(OH)]–Al junction, resulting in incomplete orbital alignment, increased local strain, and a significantly higher enthalpic cost for accessing TS‐3. This reduction in transition‐state distortion is quantitatively supported by the optimized dihedral alignments (∡10.95° vs. ∡−7.01°), the structurally relaxed Al–O–Co junction angles (∡105.5°/ ∡106.4° vs. ∡110.4°/ ∡100.6°) and a significantly reduced Co…H bond elongation (3.44 Å vs. 3.09 Å) for the np and lp phases, respectively (Table S9). Subsequent CO coordination and migratory insertion (TS‐4) exhibit only a modest conformation dependence (29.7 vs. 27.8 kcal mol⁻¹), and the final hydrogenation/hydrolysis sequence proceeds with similarly limited sensitivity to framework conformation (44.7 vs. 42.4 kcal mol⁻¹). The relative conformation‐insensitivity of these steps suggests they are primarily governed by localized, inner‐sphere coordination chemistry. H₂ heterolysis (TS‐1) relies on intrinsic bond polarity with an insignificant conformational dependence, while CO migratory insertion (TS‐4) lacks strict angular constraints and, despite the modest steric demand of the coordinated CO and CH₃ ligands, incurs only a minimal energetic advantage in the lp phase. In stark contrast, the sterically demanding σ‐bond metathesis (TS‐3) dictates a precise three‐dimensional trajectory, achieving the requisite orbital overlap strictly necessitates the global angular reorganization of the Al–[μ₃‐O–Co(OH)]–Al hinge uniquely afforded by the dynamic np ↔ lp transition. Thus, TS‐3 emerges as the distinctive transition state whose accessibility is controlled by the breathing amplitude of the host lattice. Accordingly, lattice breathing is not merely a geometric phenomenon but a transition‐state‐level catalytic parameter, in which relaxation of the Al–[μ₃ ‐O–Co(OH)]–Al hinge lowers the barrier to Co–CH₃ formation, thereby defining the kinetic threshold for C₁→C₂ homologation.

FIGURE 6.

DFT‐calculated free energy diagram at 403 K for MIL‐53‐Co(OH)‐catalyzed transformation of CH₃OH to CH₃CH₂OH using in situ generated CO and H₂; a comparison of the effect of narrow‐pore (np) and large‐pore (lp) conformations of MIL‐53 MOF on the transition state (TS‐3) in the turnover limiting step.

3. Conclusion

This work establishes atomically dispersed mononuclear Co^II(OH) active species grafted onto the nodes of breathable MIL‐53(Al) MOF, coupling single‐site precision with framework dynamics for efficient and selective CO₂ hydrogenation to ethanol. The catalyst delivers 92% CO₂ conversion with 93% ethanol selectivity and a productivity of 10,133 µmol g_cat ⁻¹ h⁻¹ in aqueous medium at 130°C. Structural, mechanistic, and DFT analysis together reveal that the reversible np↔lp transition actively modulates the Al–[μ₃ ‐O–Co(OH)]–Al hinge, selectively lowering the barrier for σ‐bond metathesis that generates the pivotal Co–CH₃ intermediate in breathing MIL‐53 and promotes selective homologation to ethanol. In contrast, rigid MIL‐68(Al) and functionalized MIL‐53 analogues with suppressed lattice dynamics exhibit markedly reduced ethanol yields and increased methanol selectivity, thereby establishing a direct correlation between framework flexibility and C–C bond formation efficiency. Post‐catalysis characterization of MIL‐53‐Co(OH) demonstrates the retention of crystallinity, oxidation state, and local coordination environment of the Co^II(OH) active sites, and recyclability tests confirm sustained catalytic performance over at least eight cycles. The commercial availability of the pristine MIL‐53(Al) MOF (Basolite A100) and the ease of synthesis of MIL‐53‐Co(OH) via post‐synthetic metallation highlights its potential for scalable implementation. Translation from lab‐scale to industrial application would require addressing MOF engineering considerations, including shaping into mechanically robust bodies without pore occlusion and ensuring efficient heat and mass transfer while preserving framework integrity under sustained turnover to minimize defect formation, structural rearrangement, diffusion limitations, and metal leaching. Therefore, future efforts will be devoted to further optimizing the MIL‐53‐Co(OH) catalyzed CO₂ hydrogenation under continuous‐flow operation in industrially relevant conditions. Finally, our study provides a mechanistic and conceptual basis for exploiting framework dynamics for active‐site engineering, offering a generalizable strategy to design efficient earth‐abundant metal catalysts for CO₂ valorization and other challenging chemical transformations.

Author Contributions

Manav Chauhan: conceptualization, investigation, writing ‐ original draft, formal analysis, methodology, data curation. Rahul Kalita: investigation, methodology, data curation, validation, formal analysis. Aditya Kumar: writing – review and editing, software, data curation, formal analysis. Bharti Rana: investigation, writing – review and editing, methodology, formal analysis, software, validation. Swapnaneel Sarmah: writing – review and editing, methodology, formal analysis, validation. Kuntal Manna: conceptualization, formal analysis, writing – original draft, supervision, resources, project administration, validation.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

The authors have cited additional references within the Supporting Information.

Supporting File: anie72589‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.