Uptake of one and two molecules of CO₂ by the molybdate dianion: a soluble, molecular oxide model system for carbon dioxide fixation

Abstract

Tetrahedral [MoO₄]²⁻ readily binds CO₂ at room temperature to produce a robust monocarbonate complex, [MoO₃(κ²-CO₃)]²⁻, that does not release CO₂ even at modestly elevated temperatures (up to 56 °C in solution and 70 °C in the solid state). In the presence of excess carbon dioxide, a second molecule of CO₂ binds to afford a pseudo-octahedral dioxo dicarbonate complex, [MoO₂(κ²-CO₃)₂]²⁻, the first structurally characterized transition-metal dicarbonate complex derived from CO₂. The monocarbonate [MoO₃(κ²-CO₃)]²⁻ reacts with triethylsilane in acetonitrile under an atmosphere of CO₂ to produce formate (69% isolated yield) together with silylated molybdate (quantitative conversion to [MoO₃(OSiEt₃)]⁻, 50% isolated yield) after 22 hours at 85 °C. This system thus illustrates both the reversible binding of CO₂ by a simple transition-metal oxoanion and the ability of the latter molecular metal oxide to facilitate chemical CO₂ reduction.

---

Metal oxide catalysts for CO₂ transformations are advantageous based on considerations of cost, ease of re-use, and stability,¹ but these advantages come at the expense of our ability to readily characterize such systems at a molecular level of detail. Intrigued by the paucity of soluble transition-metal oxide systems known to react with CO₂ in a well-defined manner (e.g., eqn (1)),^2–4 we decided to investigate salts of the molybdate dianion in this respect, in order to determine the behavior and mode of reaction (if any) of a simple oxoanion with carbon dioxide as either the potential basis for a new homogeneous catalytic system or as a soluble model for known heterogeneous oxide catalysts. Accordingly, herein we report the finding that molybdate absorbs not just one but two equivalents of CO₂ (the second, reversibly) together with complete characterization including single-crystal X-ray diffraction studies of the resulting mono- and dicarbonate complexes.

L_nM=O + CO₂ ⇌ L_nM(CO₃) (1)

As our studies were in progress, it was reported that the related tungstate dianion indeed serves as a homogeneous catalyst for CO₂ fixation,⁵ but so far the reaction intermediates in that system have not been isolated. The structural, spectroscopic, and computational details we disclose herein form an excellent point of reference both for understanding tungstate-catalyzed CO₂ fixation processes and for developing analogous systems based upon molybdate. Toward the latter goal, we show herein the ability of molybdate to mediate the triethylsilane reduction of CO₂ to formate. The present work follows and improves upon our earlier report of titanium trisanilide oxoanion CO₂ binding² in that the present system utilizes essentially non-interacting organic cations (in the earlier system oxophilic alkali-metal cations such as lithium were a necessary ingredient for CO₂ uptake) and in that the molybdate dianion is an entirely inorganic species well suited as a discrete, molecular analog of a solid-state metal oxide. Also serving as a precursor to the present work is our report of a cycle for CO₂ reduction to CO at a niobium nitride binding site; that system represented our initial foray into ligand-based strategies for CO₂ conversion.⁶

We began our studies with the commercially available sodium molybdate, but soon determined that this organic-media insoluble salt does not react with CO₂ under aqueous conditions (as assessed by ⁹⁵Mo NMR spectroscopy).⁷ In order to endow the molybdate dianion with solubility in organic media, we prepared [PPN]₂[MoO₄] (PPN⁺ = (Ph₃P)₂N⁺) in one step from Ag₂MoO₄ and [PPN]Cl using a modified literature procedure.⁸ Upon addition of CO₂ to a 0.04 M acetonitrile solution of [PPN]₂[MoO₄] at room temperature, a new species quickly formed. The ⁹⁵Mo NMR spectrum of the isolated product exhibits a new resonance at +46.7 ppm, no unreacted starting material (+13.2 ppm), but also a small amount of [Mo₂O₇]²⁻ by-product identified by a signal at −3.8 ppm.⁹ A new characteristic carbonyl stretch at 1638 cm⁻¹ could also be observed by IR spectroscopy.¹⁰

A preliminary X-ray crystal structure of the CO₂-addition product revealed a κ²-bound carbonate moiety and enabled us to formulate the major product obtained as [PPN]₂[MoO₃(κ²-CO₃)]. In the interest of obtaining high quality crystallographic data, [NEt₄]₂[MoO₄] (previously reported in the literature and used to prepare examples of well-behaved and crystallographically characterized molybdenum complexes)¹¹ was used to obtain the [MoO₃(κ²-CO₃)]²⁻ dianion as its tetraethylammonium salt. Colorless crystals were grown by vapor diffusion of Et₂O into a CH₃CN solution of [NEt₄]₂[MoO₃(κ²-CO₃)], and the structure obtained in the ensuing crystallographic investigation is shown in Fig. 1. The C–O distances are elongated from 1.162 Å in free CO₂¹² to 1.2258(13) Å, 1.3048(13) Å, and 1.3357(14) Å in the carbonate unit. The average Mo–O distance is 1.739 Å for the three molybdenum oxo bonds, shorter than the average Mo–O distance of 1.776 Å in tetrahedral [MoO₄]²⁻.¹³ The carbonate ligand is associated with longer Mo–O interatomic distances at 2.0674(9) and 2.2191(9) Å. The slight asymmetry of the carbonate binding mode is apparently induced by the trans influence of one of the molybdenum oxo ligands (O3–Mo1–O6 148.75(3)° and C1–O3–Mo1–O6 2.09(11)°) as reflected in the Mo–O bond lengths that differ by approximately 0.15 Å, but also in the different Mo–O–C angles of 97.75(6) and 91.74(7)°. This κ² binding mode is not surprising given the lack of steric bulk around the molybdenum center, in contrast to the arrangement in [(κ¹-CO₃)TiX₃]⁻ (X = N[^tBu](3,5-Me₂C₆H₃)) for which a combination of ancillary ligand steric bulk and external carbonate complexation by an alkali-metal counterion promotes κ¹-binding of [CO₃]²⁻ to the titanium center.²

Fig. 1 Left: ⁹⁵Mo NMR resonance of [MoO₃(κ²-CO₃)]²⁻. Right: solid-state structure of [NEt₄]₂[MoO₃(κ²-CO₃)] (ellipsoids at the 50% probability level, cations omitted for clarity). Representative interatomic distances [Å] and angles [°]: C1–O1 1.2258(13), C1–O2 1.3357(14), C1–O3 1.3048(13), Mo1–O2 2.0674(9), Mo1–O3 2.2191(9), Mo1–O4 1.7351(8), Mo1–O5 1.7390(8), Mo1–O6 1.7436(8); C1–O2–Mo1 97.75(6), C1–O3–Mo1 91.74(7), O6–Mo1–O3 148.75(3), O4–Mo1–O2 120.28(4), O5–Mo1–O2 121.93(4), O4–Mo1–O5 110.10(4), O4–Mo1–O6 105.15(4), O5–Mo1–O6 105.34(4); C1–O3–Mo1–O6 2.09(11), C1–O2–Mo1–O4 75.02(8), C1–O2–Mo1–O5 −71.35(8).

Solid [PPN]₂[MoO₃(κ²-CO₃)] is moderately stable in air, and does not lose CO₂ even after being heated at 70 °C under vacuum for 1 h. In solution, [PPN]₂[MoO₃(κ^2-13CO₃)] was heated to 56 °C without any observable loss of ¹³CO₂ as monitored by ¹³C NMR spectroscopy. However, this compound is moisture sensitive, undergoing conversion to [PPN]₂[MoO₄] when even a few equivalents of water are added to a solution of [PPN]₂[MoO₃(κ²-CO₃)]. On the other hand, solid [NEt₄]₂[MoO₃(κ²-CO₃)] is hygroscopic and converts to molybdate and dimolybdate in ca. 15 minutes by absorbing moisture from the ambient atmosphere, as monitored by IR spectroscopy.

Prepared and isolated using ¹³CO₂, [PPN]₂[MoO₃(κ^2-13CO₃)] displays a sharp ¹³C NMR resonance at 165.7 ppm, this being in a region of the spectrum that is characteristic for carbonates.^2,3,14 In its IR spectrum, an isotope-shifted carbonyl stretch is present at 1599 cm⁻¹, in good agreement with the theoretical 1602 cm⁻¹ predicted using the harmonic oscillator approximation. A small peak due to a minor impurity at 158.9 ppm could also be observed by ¹³C NMR spectroscopy, correlated with the trace dimolybdate by-product detected by ⁹⁵Mo NMR spectroscopy. Adding [PPN][HCO₃]¹⁵ to a mixture of [PPN]₂[MoO₃(κ^2-13CO₃)] and this unknown species yielded an increase in the intensity of the 158.9 ppm signal and no additional resonances, allowing us to conclusively assign the minor impurity as bicarbonate anion. The minor [HCO₃]⁻ impurity may originate from the reaction of [MoO₄]²⁻ with [MoO₃(κ²-CO₃)]²⁻ to yield [Mo₂O₇]²⁻ and free [CO₃]²⁻, the latter converting to bicarbonate upon protonation, presumably from adventitious water.

Under 1 atm of ¹³CO₂, the room temperature ¹³C NMR spectrum of a [PPN]₂[MoO₄] solution features two broad signals: one for the free ¹³CO₂ at 125.8 ppm, and one for the molybdenum carbonate at 165.3 ppm, the broad nature of the resonances suggesting that a chemical exchange is occurring on the NMR time scale. A new major resonance appeared at 162.8 ppm when acquiring the spectrum at −19 °C, but disappeared after degassing the sample. The ratio of the unknown species at 162.8 ppm to [MoO₃(κ^2-13CO₃)]²⁻ increases at higher pressure of ¹³CO₂ (3 atm), and at lower temperature (−31 °C). These data are indicative of additional reversible binding of ¹³CO₂ to the [MoO₃(κ^2-13CO₃)]²⁻ dianion.

The existence of a dicarbonate species was confirmed by X-ray crystallography, as colorless diffraction quality crystals were grown by slowly cooling a CH₃CN solution of [PPN]₂[MoO₄] under an atmosphere of CO₂. In the solid state, both carbonate ligands are bound κ² (Fig. 2), with Mo–O distances of 2.198(2) and 2.024(2), 2.175(2) and 2.0145(19) Å, respectively. The molybdenum oxo distances are 1.695(2) and 1.705(3) Å, shorter still than in [MoO₃(κ²-CO₃)]²⁻ as the Mo–O π character is shared over fewer centers. The carbonate ligand binding mode is characterized by the same type of asymmetry as seen in [MoO₃(κ²-CO₃)]²⁻ due to the trans influence of the oxo ligands (O3–Mo1–O2 152.29(10)°, O6–Mo1–O1 152.10(8)°, C1–O3–Mo1–O2 18.3(2)°, C2–O6–Mo1–O1 27.3(2)°). While we were able to find several examples of κ²-bound molybdenum carbonates reported in the Cambridge Structural Database,¹⁶ this is the first example of a molybdenum complex with two κ²-carbonates. To the best of our knowledge, [PPN]₂[MoO₂(κ²-CO₃)₂] is also the first transition-metal dicarbonate complex obtained directly from CO₂.¹⁷

Fig. 2 Right: ¹³C NMR spectrum showing the distribution of [MoO₃(κ²-CO₃)]²⁻ (165.4 ppm) and [MoO₂(κ²-CO₃)₂]²⁻ (162.8 ppm) at −19 °C under 1 atm of ¹³CO₂. Left: solid-state structure of [PPN]₂[MoO₂(κ²-CO₃)₂] (ellipsoids at the 50% probability level, cations and solvent molecules omitted for clarity). Representative interatomic distances [Å] and angles [°]: C1–O3 1.302(4), C1–O4 1.352(4), C1–O5 1.228(4), C2–O6 1.303(3), C2–O7 1.358(3), C2–O8 1.223(3), Mo1–O1 1.695(2), Mo1–O2 1.705(3), Mo1–O3 2.198(2), Mo1–O4 2.024(2), Mo1–O6 2.175(2), Mo1–O7 2.0145(19); O3–C1–O4 110.4(2), O6–C2–O7 109.9(2), O4–Mo1–O7 149.44(6), O3–Mo1–O2 152.29(10), O6–Mo1–O1 152.10(8), O1–Mo1–O2 104.86(18); C1–O3–Mo1–O2 18.3(2), C1–O4–Mo1–O6 −79.04(11), C1–O4–Mo1–O1 80.37(14), C2–O6–Mo1–O1 27.3(2), C2–O7–Mo1–O2 86.70(15), C2–O7–Mo1–O3 −76.00(12).

In order to gain further insight into the energetics and potential energy landscape of this system, we turned to computational methods (Fig. 3). Binding of the first CO₂ molecule is exothermic and exergonic with a ΔH°(298 K) = −14.2 kcal mol⁻¹ and ΔG°(298 K) = −5.2 kcal mol⁻¹. The stability of the [MoO₃(κ²-CO₃)]²⁻ species is explained by the considerable activation energy of ΔG^‡(298 K) = 17.2 kcal mol⁻¹ for regenerating the molybdate dianion with loss of CO₂. This is consistent with our inability to remove CO₂ under vacuum at room temperature from this material. As expected, binding of the second CO₂ is slightly endergonic (ΔG°(298 K) = 3.2 kcal mol⁻¹), being favored at higher CO₂ pressures and lower temperatures as observed in our ¹³C-labeling experiments. The possibility of binding a third CO₂ molecule was also investigated. However, producing such a species is endergonic with a ΔG°(298 K) = 14.6 kcal mol⁻¹, as well as ΔH° > 0 and ΔS° < 0. In contrast to the findings of Mizuno et al. who reported a calculated κ¹ structure for the related tungstate–CO₂ adduct,⁵ we were unable to locate minima corresponding to κ¹ structures for any of the molybdenum carbonate complexes studied herein.

Fig. 3 Combined calculated potential energy diagram for the first and second CO₂ binding events. Electronic structure calculations were carried out using the M06¹⁸ density functional with the Def2-QZVPP¹⁹ basis set for molybdenum, incorporating the SDD²⁰ effective core potential, and 6-311+G(3df) for all other atoms as implemented in the Gaussian 09 suite of programs.²¹ The CPCM model²² for CH₃CN was used to describe solvation effects, and the final single-point energies were calculated with QCISD(T)²³ at the optimized M06 geometries.

Curious to see whether the new molybdenum carbonate complexes can serve as a source of activated CO₂, we subjected [PPN]₂[MoO₃(κ²-CO₃)] to the mild hydride donor triethylsilane, which exhibits no background reactivity with CO₂ according to a control experiment.²⁴ A test reaction revealed a new resonance at δ = 8.73 ppm (¹H NMR spectroscopy), this being located in a region characteristic for formyl protons, as well as a new molybdenum species having a ⁹⁵Mo NMR signal at −23.7 ppm, a shift essentially identical to that reported for the [MoO₃(OSiMe₃)]⁻ anion.⁹ The production of formate improves from 16% to 71% if the reaction is run under an atmosphere of CO₂, raising the question of whether the active species facilitating CO₂ reduction is the monocarbonate complex [MoO₃(κ²-CO₃)]²⁻ or the dicarbonate complex [MoO₂(κ²-CO₃)₂]²⁻. After optimization, clean conversion to [PPN][OCHO] and [PPN][MoO₃(OSiEt₃)] (eqn (2)) as the sole products can be obtained after 22 h at 85 °C, as evidenced by the ¹H NMR spectrum of the crude reaction mixture. We were able to isolate [PPN][OCHO] in 69% yield, along with [PPN][MoO₃(OSiEt₃)] in 50% yield by taking advantage of their differential solubilities in THF.

In summary, we have synthesized and characterized two molybdenum oxo carbonate species obtained from the uptake of CO₂ by the molybdate dianion and begun exploring their reactivity in the context of CO₂ reduction to formate.²⁵¹³C-labeling experiments and computational analysis suggest that the first binding event to form [MoO₃(κ²-CO₃)]²⁻ is irreversible, while the second CO₂ molecule binds in a reversible process. We are currently investigating potential uses of [MoO₄]²⁻ as a nucleophilic catalyst for CO₂ fixation²⁶ (encouraged by the methods already developed using the analogous tungstate dianion),⁵ while also making efforts to develop an improved mechanistic understanding of this rich, yet simple system. This work illustrates the finding that a discrete, soluble molecular metal oxide system unencumbered by organic spectator ligands and unobstructed by hydrogen bonding with water is sufficient for CO₂ activation and conversion to formate in conjunction with a mild hydride source.

Acknowledgements

Dr Peter Müller is acknowledged for his invaluable help with the crystal structure refinement of [NEt₄]₂[MoO₃(κ²-CO₃)] and [PPN]₂[MoO₂(κ²-CO₃)₂]. SABIC (Saudi Basic Industries Corporation) is acknowledged for partly funding the work of IK. TO would like to thank the Spanish Ministry of Education, Culture and Sport (MECD) for financial support. MT acknowledges the Spanish Ministry of Economy and Competitiveness under CTQ2012-36966 for financial support. DT was funded by the National Science Foundation under CHE-1111357. X-ray diffraction data was collected on an instrument purchased with the aid of the National Science Foundation (NSF) under CHE-0946721.

References

1. (a) S. Wang, K. Murata, T. Hayakawa, S. Hamakawa and K. Suzuki, Appl. Catal., A, 2000, 196, 1–8; (b) B. M. Bhanage, S.-I. Fujita, Y. Ikushima and M. Arai, Appl. Catal., A, 2001, 219, 259–266; (c) M. Matsuoka and M. Anpo, J. Photochem. Photobiol., C, 2003, 3, 225–252.
2. J. S. Silvia and C. C. Cummins, Chem. Sci., 2011, 2, 1474–1479.
3. J. P. Krogman, M. W. Bezpalko, B. M. Foxman and C. M. Thomas, Inorg. Chem., 2013, 52, 3022–3031.
4. N. P. Tsvetkov, J. G. Andino, H. Fan, A. Y. Verat and K. G. Caulton, Dalton Trans., 2013, 42, 6745–6755.
5. (a) T. Kimura, K. Kamata and N. Mizuno, Angew. Chem., Int. Ed., 2012, 51, 6700–6703; (b) T. Kimura, H. Sunaba, K. Kamata and N. Mizuno, Inorg. Chem., 2012, 51, 13001–13008.
6. J. S. Silvia and C. C. Cummins, J. Am. Chem. Soc., 2010, 132, 2169–2171.
7. Sodium molybdate dihydrate (50 mg, 0.2 mmol, 1 equiv.) was dissolved in ca. 0.8 mL D₂O. Carbon dioxide (20 mL, 0.8 mmol, 4 equiv.) was bubbled through this solution, after which it was stirred for 10 min. The ⁹⁵Mo NMR spectrum of the resulting solution showed only the molybdate resonance at δ = 0 ppm and no other species.
8. J. R. Briggs, A. M. Harrison and J. H. Robson, Polyhedron, 1986, 5, 281–287.
9. Y. Do, E. D. Simhon and R. H. Holm, Inorg. Chem., 1985, 24, 1831–1838.
10. G. Busca and V. Lorenzelli, Mater. Chem., 1982, 7, 89–126.
11. (a) D. V. Partyka and R. H. Holm, Inorg. Chem., 2004, 43, 8609–8616; (b) S. Groysman, J.-J. Wang, R. Tagore, S. C. Lee and R. H. Holm, J. Am. Chem. Soc., 2008, 130, 12794–12807.
12. G. Herzberg, Electronic spectra and electronic structure of polyatomic molecules, Van Nostrand, New York, 1966.
13. P. Román, A. San José, A. Luque, J. M. Gutiérrez-Zorrilla and M. Martínez-Ripoll, Acta Crystallogr., Sect. C: Cryst. Struct. Commun., 1994, 50, 1189–1191.
14. D. J. Darensbourg, K. M. Sanchez and A. L. Rheingoldib, J. Am. Chem. Soc., 1987, 109, 290–292.
15. M. L. Meckfessel Jones, Ph.D. thesis, Texas A&M University, 1994.
16. (a) J. Chatt, M. Kubota, G. J. Leigh, F. C. March, R. Mason and D. J. Yarrow, J. Chem. Soc., Chem. Commun., 1974, 1033–1034; (b) E. Carmona, F. Gonzalez, M. L. Poveda, J. M. Marin, J. L. Atwood and R. D. Rogers, J. Am. Chem. Soc., 1983, 105, 3365–3366; (c) D. M. Curtis and K. R. Han, Inorg. Chem., 1985, 24, 378–382; (d) K. A. Belmore, R. A. Vanderpool, J.-C. Tsai, M. A. Khan and K. M. Nicholas, J. Am. Chem. Soc., 1988, 110, 2004–2005; (e) L. Contreras, M. Paneque, M. Sellin, E. Carmona, P. J. Perez, E. Gutierrez-Puebla, A. Monge and C. Ruiz, New J. Chem., 2005, 29, 109–115.
17. S. V. Krivovichev and P. C. Burns, Radiochemistry, 2004, 46, 12–15.
18. Y. Zhao and D. G. Truhlar, Theor. Chem. Acc., 2008, 120, 215–241.
19. F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005, 7, 3297–3305.
20. D. Andrae, U. Haeussermann, M. Dolg, H. Stoll and H. Preuss, Theor. Chim. Acta, 1990, 77, 123–141.
21. M. J. Frisch, et al., Gaussian 09, Revision C.01, 2010.
22. (a) V. Barone and M. Cossi, J. Phys. Chem. A, 1998, 102, 1995–2001; (b) A. Klamt and G. Schuurmann, J. Chem. Soc., Perkin Trans. 2, 1993, 799–805.
23. (a) J. A. Pople, M. Head-Gordon and K. Raghavachari, J. Chem. Phys., 1987, 87, 5968–5975; (b) P. J. Knowles and H.-J. Werner, Chem. Phys. Lett., 1985, 115, 259–267.
24. Triethylsilane (23 mg, 0.2 mmol) was dissolved in ca. 1 mL CD₃CN and transferred to a Schlenk tube which was brought outside the glovebox. The solution was frozen using liquid nitrogen, the headspace was evacuated and the tube refilled with 1 atm of CO₂. The tube was sealed and heated at 85 °C overnight. The solution was analyzed by ¹H NMR spectroscopy, showing that no reaction had occured.
25. (a) S. Itagaki, K. Yamaguchi and N. Mizuno, J. Mol. Catal. A: Chem., 2013, 366, 347–352; (b) W. Sattler and G. Parkin, J. Am. Chem. Soc., 2012, 134, 17462–17465; (c) K. Motokura, D. Kashiwame, A. Miyaji and T. Baba, Org. Lett., 2012, 14, 2642–2645; (d) R. Lalrempuia, M. Iglesias, V. Polo, P. J. Sanz Miguel, F. J. Fernández-Alvarez, J. J. Pérez-Torrente and L. A. Oro, Angew. Chem., Int. Ed., 2012, 51, 12824–12827; (e) A. Jansen and S. Pitter, J. Mol. Catal. A: Chem., 2004, 217, 41–45.
26. (a) Y.-B. Wang, Y.-M. Wang, W.-Z. Zhang and X.-B. Lu, J. Am. Chem. Soc., 2013, 135, 11996–12003; (b) J. Seayad, A. M. Seayad, J. K. P. Ng and C. L. L. Chai, ChemCatChem, 2012, 4, 774–777; (c) Z.-Z. Yang, L.-N. He, S.-Y. Peng and A.-H. Liu, Green Chem., 2010, 12, 1850–1854; (d) S. N. Riduan, Y. Zhang and J. Y. Ying, Angew. Chem., Int. Ed., 2009, 48, 3322–3325; (e) H. Zhou, W.-Z. Zhang, C.-H. Liu, J.-P. Qu and X.-B. Lu, J. Org. Chem., 2008, 73, 8039–8044; (f) W. N. Sit, S. M. Ng, K. Y. Kwong and C. P. Lau, J. Org. Chem., 2005, 70, 8583–8586.

---

Footnote

† Electronic supplementary information (ESI) available: Full experimental, crystallographic, spectroscopic, and computational data. CCDC 978136, 978137 and 981116. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c4sc00132j

---