Electrocatalytic CO2 Conversion to Oxalate by a Copper Complex

See allHide authors and affiliations

Science  15 Jan 2010:
Vol. 327, Issue 5963, pp. 313-315
DOI: 10.1126/science.1177981

Oxalate from Air

In light of increasing concerns about the consequences of excessive atmospheric carbon dioxide, there is demand for methods to use carbon dioxide in the preparation of more elaborate compounds. Though reactions with hydroxide salts to form carbonates tend to proceed fairly cleanly, reductive processes to form carboxylic acids, esters, and alcohols are often rather unselective. Angamuthu et al. (p. 313; see the news story by Service) discovered that a copper complex exhibited remarkable selectivity in reductively coupling carbon dioxide to form oxalate through coordinative electron transfer, even in the presence of excess oxygen, normally a much more potent electron acceptor. Precipitation of the oxalate as a lithium salt and electrochemical re-reduction of the copper produced a preliminary catalytic cycle, demonstrated through six turnovers.


Global warming concern has dramatically increased interest in using CO2 as a feedstock for preparation of value-added compounds, thereby helping to reduce its atmospheric concentration. Here, we describe a dinuclear copper(I) complex that is oxidized in air by CO2 rather than O2; the product is a tetranuclear copper(II) complex containing two bridging CO2-derived oxalate groups. Treatment of the copper(II) oxalate complex in acetonitrile with a soluble lithium salt results in quantitative precipitation of lithium oxalate. The copper(II) complex can then be nearly quantitatively electrochemically reduced at a relatively accessible potential, regenerating the initial dinuclear copper(I) compound. Preliminary results demonstrate six turnovers (producing 12 equivalents of oxalate) during 7 hours of catalysis at an applied potential of –0.03 volts versus the normal hydrogen electrode.

Research toward carbon dioxide fixation enjoys much attention at present, as a result of the alarming reports that link global warming and its potentially devastating effects with the steadily increasing concentration of CO2 in the atmosphere. Chemical activation of carbon dioxide could help to reduce its concentration in the atmosphere while at the same time exploiting it as a carbon feedstock for the production of useful organic compounds (15). Transition-metal complexes, especially of copper and zinc (6, 7), as well as simple salts such as lithium hydroxide monohydrate and soda-lime (mixture of sodium and calcium hydroxides) are well known for their assistance in the stoichiometric transformation of carbon dioxide to carbonate salts (817). Mixtures of glycol and amines (glycol-amine) as well as coordination complexes of polyamines have been reported to bind CO2 reversibly through the formation of carbamates (8, 15, 17, 18). In contrast, reductive conversion of CO2 into useful products of industrial significance such as formaldehyde, formic acid, methanol, or oxalic acid has proven more challenging to achieve selectively (19, 20).

The one-electron reduction of CO2 into the CO2•– radical anion occurs at potentials as high as –1.97 V versus NHE (normal hydrogen electrode) in N,N-dimethylformamide, and the CO2•– may further react to form CO, carbonate, formate, or oxalate (1921). Selective production of oxalate would be much preferred because dimethyl oxalate is a useful feedstock, for example, for the production of methyl glycolate. The assistance of transition-metal complexes appears mandatory to direct the reactivity of the CO2•– radical anion toward a specific product, in addition to optimizing electrochemical parameters such as current density. Moreover, the inner-sphere electron-transfer mechanisms that proceed with most transition metal systems result in less-negative reduction potentials, which may improve overall thermodynamic favorability of the reduction, assuming there is an accessible way to liberate the product after the electron-transfer reaction (20). Reductive coupling of CO2 to form the oxalate dianion has been accomplished by electrochemical methods, including outer-sphere electron transfer using mercury or lead electrodes and inner-sphere electron transfer using transition-metal complexes or anion radicals of aromatic hydrocarbons, esters, and nitriles as electrocatalysts (2022). Mechanistic understanding of the metal-catalyzed reduction of CO2 to C2 or C3 fragments is also highly relevant for an improved understanding of the natural photosynthetic transformation of atmospheric CO2 to functionalized C3 molecules (3-phosphoglycerate).

We herein report a copper complex, which spontaneously captures and reductively couples CO2 from the air selectively, yielding an oxalate-bridged copper(II) tetramer in acetonitrile solution. Moreover, we have found that this copper system can be used repeatedly as a catalyst for the reductive coupling of CO2 to oxalate upon electrochemical reduction. The reduction of the copper(II) complex occurs at a readily accessible potential that is nearly 2 V less negative than that required for outer-sphere reduction of CO2 to CO2•–.

The ligand HL [N-(2-mercaptopropyl)-N,N-bis(2-pyridylmethyl)amine] was designed for the synthesis of biomimetic models for nickel-containing superoxide dismutase. In addition to studies with nickel salts, reactions were performed with copper and zinc for comparison. Upon mixing of equimolar amounts of Cu(acac)2 (Hacac is acetylacetone), the ligand HL, and HBF4 in acetonitrile at room temperature, we obtained a yellow-colored solution, in which as expected the thiolate-containing ligand was oxidized by the copper(II) ion. The solution was analyzed with positive-ion electrospray ionization mass spectroscopy (ESI-MS); a prominent signal at mass/charge (m/z) ratio of 335.91 showed an isotopic distribution envelope matching that calculated for the dinuclear copper(I) complex [1]2+ (Fig. 1) (23). This complex [1]2+ can also be synthesized by the reaction of the preoxidized disulfide ligand with two equivalents of [Cu(CH3CN)4]BF4 in dry acetonitrile. This yellow-colored solution turned greenish-blue upon exposure to air; over the course of 3 days crystals formed, which we isolated in 72% yield and analyzed by x-ray diffraction. We observed a tetranuclear copper(II) structure [CuII2(L-L)(μ-oxalato-κ4O1,O2:O3,O4)]2(BF4)4 {[2](BF4)4, Fig. 1}, with bridging oxalate anions that must originate from CO2 in the air. A positive-ion ESI-MS spectrum acquired from the acetonitrile solution is consistent with this molecular structure, showing a prominent signal at m/z of 379.35 (figs. S9 and S10). We thus found that the initial Cu(I) complex is oxidized by CO2 rather than O2. Indeed, purging carbon dioxide into a solution of complex [1]2+ results in the formation of the tetranuclear oxalate-bridged complex [2]4+, which was fully characterized by Fourier transform infrared spectroscopy (FT-IR), ESI-MS, and elemental analysis. That carbon dioxide is the origin of the oxalate dianion was proven with the use of 13CO2; the resulting copper(II) complex showed a signal at m/z of 381.06 (fig. S11). The reaction of [1]2+ in an O2 atmosphere under strict exclusion of CO2 resulted in a deep green solution containing a copper(II) compound with a molecular ion peak at m/z = 361.16 in positive ion ESI-MS that is consistent with the expected dihydroxo complex of molecular formula [CuII(L-L)CuII(μ-OH)2(H2O)]2+(fig. S12).

Fig. 1

Schematic overview of the formation and reactivity of the complexes [1]2+, [2]4+, and [3]4+. Cu, brown; N, blue; S, yellow; O, red; Cl, green; C, black. BF4 anions, solvent molecules, and hydrogen atoms are omitted for clarity. Selected (average) bond lengths (Å) for [2]4+: Cu–Oeq, 1.963(2); Cu–Oax, 2.283(2); Cu–Npy, 1.991(2); Cu–Namine, 2.026(2); S1–S2, 2.0423(16); Cu1⋅⋅⋅S1, 2.9837(12); Cu2⋅⋅⋅S2, 2.9731(12); Cu1⋅⋅⋅Cu2, 5.3205(6); Cu1⋅⋅⋅Cu2i, 5.4295(6). Selected (average) bond lengths (Å) for [3]4+: Cu1–Cl1i 2.2479(6); Cu2–Cl2 2.2440(7); Cu1–Cl1 2.8589(6); Cu–Npy, 1.984(1); Cu–Namine, 2.052(2); S1–S2, 2.0388(11); Cu1⋅⋅⋅S1, 3.0036(9); Cu2–S2, 2.7343(7); Cu1⋅⋅⋅Cu1i, 3.5677(5); Cu1⋅⋅⋅Cu2, 6.1248(4). Symmetry operation i; 1 – x, 1 – y, 1 – z. Estimated standard deviations in the last digits are given in parentheses. Further details are provided in (23).

In the solid state, [2]4+ consists of a cyclic centrosymmetric dimer of two dinuclear moieties bridged by two oxalato dianions (fig. S13). Each dinuclear moiety consists of two crystallographically independent copper(II) ions. The two copper(II) ions within the asymmetric unit bind to the same disulfide ligand and are separated by 5.3205 ± 0.0006 [5.3205(6)] Å. The copper ions are situated in square-pyramidal environments, with the three nitrogen donors of the meridionally coordinated dipicolylamine unit occupying three corners of the basal plane. One of the oxygens from the bridging oxalato dianion is situated at the fourth corner of the basal plane, with another oxygen from the same oxalato dianion occupying the apical position. However, for both copper ions one of the disulfide sulfur atoms can be regarded as a sixth ligand at meaningful axial distances of 2.9837(12) and 2.9731(12) Å. (Further parameters are provided in table S1.)

In an attempt to crystallize the original complex [1]2+, chloroform was let to diffuse into the initial reaction mixture containing the copper(I) complex in an argon atmosphere. Interestingly, this yielded the unexpected tetranuclear compound [ClCuII(L-L)CuII(μ-Cl)]2(BF4)4 {[3](BF4)4, Fig. 1} with bridging and terminal chloride anions that can only originate from chloroform (24). The solid-state structure of [3]4+ was obtained by x-ray diffraction from a blue crystal of [3](BF4)4. The molecular structure of [3]4+ is confirmed by a positive-ion ESI-MS spectrum of the compound acquired from acetonitrile solution, which shows a prominent signal at m/z = 370.71 (figs. S14 and S15). Complex [3]4+ is a linear centrosymmetric dimer of two dinuclear moieties bridged by two chloride anions (figs. S16 and S17). The two copper(II) ions within the asymmetric unit are bound to the same disulfide ligand and are separated by 6.1248(4) Å. The copper ions in [3]4+ are situated in pentacoordinate environments resembling those in complex [2]4+; the thioether sulfur and the chloride donors replace the oxalato oxygen donors in [2]4+.

Inspired by this finding, we explored whether complex [2]4+ could be converted to this chloride complex [3]4+ by treatment with HCl, in the process liberating the CO2-derived oxalic acid.

Addition of four equivalents of hydrochloric acid to an acetonitrile solution of [2](BF4)4 indeed leads to elimination of oxalic acid with concurrent formation of [3](BF4)4 as confirmed by ESI-MS and elemental analysis. The electrochemical reduction of [3](BF4)4 occurs at the cathodic peak potential (Epc) of +0.06 V versus NHE (fig. S18), producing a copper(I) complex that selectively produces complex [2]4+ upon reaction with CO2. This result stimulated us to explore the possibility of using the copper/disulfide-ligand system as an electrocatalyst for the selective reduction of CO2.

To that end, we undertook electrochemical reduction of complex [3]4+ by using controlled potential coulometry and monitored the process by using electronic absorption spectroscopy. The copper complex [3](BF4)4 (0.9 g, 0.5 mmol) was dissolved in 100 ml of 0.1 M tetrabutylammonium hexafluoridophosphate in acetonitrile; the solution was then reduced at +0.03 V versus NHE. A current drop was observed after 195 C of charge was passed, the quantity expected for a one-electron reduction of each copper ion. The disappearance of the characteristic d-d transition band (~670 nm) of [3]4+ during electrolysis confirmed the formation of a copper(I) species (fig. S19). The resulting yellow-colored solution was shown by ESI-MS to contain the dinuclear copper(I) complex [1]2+ (fig. S20). The cyclic voltammogram of this solution showed a reversible oxidation process at the anodic peak potential (Epa) of +0.81 V versus NHE (fig. S21).

Bubbling carbon dioxide into this solution turned the color greenish-blue, indicating the formation of complex [2]4+ as confirmed by ESI-MS analysis of the solution. The cyclic voltammogram of [2]4+ produced in this reaction sequence was identical to that of the independently synthesized and isolated [2]4+ and showed an irreversible reduction process at –0.03 V versus NHE (fig. S22). The bulk electrolysis experiment was then repeated under the same conditions but with use of lithium perchlorate as the supporting electrolyte in a CO2-saturated acetonitrile solution. These conditions resulted in the precipitation of lithium oxalate as the generated copper(I) complex spontaneously reacted with the CO2 available in the solution to form oxalate (fig. S23). In order to quantify the selectivity of our electrocatalyst, we halted electrolysis after passing 195 C of charge (the charge expected for a one-electron reduction of each copper ion), purged the solution with CO2, and removed the lithium oxalate precipitate by filtration under an argon atmosphere. The 24-mg (0.24-mmol) yield of lithium oxalate [as confirmed by ESI-MS spectrometry, nuclear magnetic resonance (NMR), and FT-IR spectroscopy, figs. S24 and S25] corresponded to nearly quantitative current efficiency (96%) for formation of the desired product. The remaining blue-colored solution was shown to contain the dinuclear copper(II) complex [(CH3CN)CuII(L-L)CuII(CH3CN)]4+ [4]4+ as characterized by ESI-MS spectrometry (fig. S26). We proceeded to saturate this solution with argon to remove the remaining CO2 and then subjected it to a second electrolysis run; 185 C of charge was consumed before the current dropped, indicating regeneration of nearly 95% of the copper(I) complex.

Both complexes [2]4+ and [3]4+ upon mixing with LiClO4 in acetonitrile yield [4]4+ as confirmed by ESI-MS spectrometry (Fig. 2 and figs. S27 and S28). Therefore, in another attempt to use the complex [2]4+ as an electrocatalyst in this reaction, the electrochemical cell containing an acetonitrile solution of complex [2]4+ and lithium perchlorate (as supporting electrolyte) was stirred to precipitate all the available oxalate. Then the solution was electrolyzed at –0.03 V versus NHE with continuous purging of CO2. The consumption of current continued linearly for more than 3.3 hours, consuming three equivalents of charge (12 electrons) per four copper ions, with concurrent crystallization of lithium oxalate. Thereafter, the rate of the reaction gradually decreased as the crystallized lithium oxalate started to cover the electrode surface, thereby hampering electron transfer (fig. S29). In total, the electrocatalysis could be extended for more than 7 hours, with consumption of 6 equivalents of charge (24 electrons) and generating 12 equivalents of oxalate per molecule of [2]4+.

Fig. 2

Formation of [4]4+ from [2]4+ or [3]4+.

We have thus devised an electrocatalytic system based on a copper coordination compound that is able to activate and convert CO2 selectively into oxalate at readily accessible potentials, in the simple but very effective catalytic cycle shown in Fig. 3. The finding that a copper(I) system is oxidized by CO2 rather than O2 implies that the selective binding of CO2 to the copper(I) ions offers a low-energy pathway for the formation of the CO2•– radical anion. The copper(II) oxalate complex [2]4+ is thermodynamically favored; the binding of CO2 to the Cu(I) centers in [1]2+ and the formation of oxalate appears to be highly selective and relatively rapid. Because of the low solubility of lithium oxalate in acetonitrile, the release of the oxalate dianion from [2]4+ in the presence of lithium perchlorate is instantaneous, generating the complex [4]4+. Therefore, for the current system the electrocatalytic reduction of the copper(II) ion to copper(I) appears to be rate-limiting. The precipitation of the lithium oxalate formed during the reaction onto the electrode surface hampers efficient electron transfer. Tuning the redox potential of the copper complex by altering the ligand structure with a variety of substituents, immobilization of the complex onto the electrode surface, and improved methods for the removal of oxalate may result in improved efficiency of the catalytic system. We believe that our studies will instigate further development of coordination complexes for catalytic CO2 sequestration, its selective conversion and use as fuels such as methanol or as feedstock in the synthesis of useful organic compounds.

Fig. 3

Proposed electrocatalytic cycle for oxalate formation.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S30

Tables S1 and S2


References and Notes

  1. Synthesis and characterization details are provided as supporting material on Science Online.
  2. A similar reactivity of coordination complexes with chloroform has been observed and reported before; see, for example, (25).
  3. This work was supported by the Leiden Institute of Chemistry. X-ray crystallographic work was supported (M.L. and A.L.S.) by the Council for the Chemical Sciences of The Netherlands Organization for Scientific Research (CW-NWO). J. Reedijk and M. T. M. Koper are gratefully acknowledged for stimulating discussions. P.B. (Ithaca College, New York) was involved in the project through a summer exchange program. Crystallographic data for [2](BF4)4 and [3](BF4)4 have been deposited with the Cambridge Crystallographic Data Center under reference numbers 717726 and 717727.

Stay Connected to Science

Navigate This Article