Article pubs.acs.org/accounts
Renewable Formate from C−H Bond Formation with CO2: Using Iron Carbonyl Clusters as Electrocatalysts Natalia D. Loewen, Taruna V. Neelakantan, and Louise A. Berben* Department of Chemistry, University of California, Davis, California 95616, United States S Supporting Information *
CONSPECTUS: As a society, we are heavily dependent on nonrenewable petroleum-derived fuels and chemical feedstocks. Rapid depletion of these resources and the increasingly evident negative effects of excess atmospheric CO2 drive our efforts to discover ways of converting excess CO2 into energy dense chemical fuels through selective C−H bond formation and using renewable energy sources to supply electrons. In this way, a carbon-neutral fuel economy might be realized. To develop a molecular or heterogeneous catalyst for C−H bond formation with CO2 requires a fundamental understanding of how to generate metal hydrides that selectively donate H− to CO2, rather than recombining with H+ to liberate H2. Our work with a unique series of water-soluble and -stable, lowvalent iron electrocatalysts offers mechanistic and thermochemical insights into formate production from CO2. Of particular interest are the nitride- and carbidecontaining clusters: [Fe4N(CO)12]− and its derivatives and [Fe4C(CO)12]2−. In both aqueous and mixed solvent conditions, [Fe4N(CO)12]− forms a reduced hydride intermediate, [H−Fe4N(CO)12]−, through stepwise electron and proton transfers. This hydride selectively reacts with CO2 and generates formate with >95% efficiency. The mechanism for this transformation is supported by crystallographic, cyclic voltammetry, and spectroelectrochemical (SEC) evidence. Furthermore, installation of a proton shuttle onto [Fe4N(CO)12]− facilitates proton transfer to the active site, successfully intercepting the hydride intermediate before it reacts with CO2; only H2 is observed in this case. In contrast, isoelectronic [Fe4C(CO)12]2− features a concerted proton−electron transfer mechanism to form [H−Fe4C(CO)12]2−, which is selective for H2 production even in the presence of CO2, in both aqueous and mixed solvent systems. Higher nuclearity clusters were also studied, and all are proton reduction electrocatalysts, but none promote C−H bond formation. Thermochemical insights into the disparate reactivities of these clusters were achieved through hydricity measurements using SEC. We found that only [H−Fe4N(CO)12]− and its derivative [H−Fe4N(CO)11(PPh3)]− have hydricities modest enough to avoid H2 production but strong enough to make formate. [H−Fe4C(CO)12]2− is a stronger hydride donor, theoretically capable of making formate, but due to an overwhelming thermodynamic driving force and the increased electrostatic attraction between the more negative cluster and H+, only H2 is observed experimentally. This illustrates the fundamental importance of controlling thermochemistry when designing new catalysts selective for C−H bond formation and establishes a hydricity range of 15.5−24.1 or 44−49 kcal mol−1 where C−H bond formation may be favored in water or MeCN, respectively.
■
INTRODUCTION The development of alternative fuels and commodity chemicals derived from carbon dioxide (CO2) is becoming increasingly important as the negative effects of CO2 in the atmosphere intensify and as continued environmental catastrophes result from our unsustainable reliance on the products of the fossil fuel industry.1 If CO2 could be used as a feedstock for fuel and chemical production powered by electricity derived from renewable resources, then a carbon neutral, or even a carbon negative, economy could be established, and further increases in atmospheric CO2 concentration would be prevented. Chemical fuels provide a very high energy density form of storage, and their production and use is feasible on a very large scale from inexpensive substrates and renewable energy. For example, formic acid (FA) is an excellent candidate as a hydrogen © 2017 American Chemical Society
storage material in H2 fuel cell vehicles, and because it is nontoxic with low flammability, the existing gasoline infrastructure could be potentially employed for FA distribution.2 FA is currently priced at 400−650 USD/kg and has low annual production (global capacity 8 × 105 t/a). To compete with gasoline, FA prices need to drop to 300 USD/t. If production increased by a factor of 104, prices would drop dramatically and the supply would be adequate to use FA as a fuel. Current routes to FA use methanol as a feedstock, so development of cost-effective and renewable CO2-derived FA is needed. We summarize some of the best performing catalysts for formic acid or formate production before describing our own Received: June 14, 2017 Published: August 24, 2017 2362
DOI: 10.1021/acs.accounts.7b00302 Acc. Chem. Res. 2017, 50, 2362−2370
Accounts of Chemical Research
■
ELECTROCATALYSIS Electrocatalysts store energy in the form of chemical bonds (fuels). The energy is derived from renewable resources and used in the form of electricity. The proton and electron transfer reactions involved in electrochemical C−H bond formation with CO2 are a challenging combination of chemical transformations that must be appropriately directed by a catalyst that enables the desired selectivity in product formation. The simplest product of C−H bond formation with CO2 is formate. Formally, the “H” should be delivered as H− (hydride), and this hydride must be generated by an appropriately designed catalyst through a series of proton and electron transfers. There are many challenges in the design of this catalyst, and these include (1) the formation of H− from protons (or H2O) with low energy input and (2) generation of hydride reagents with the correct thermodynamic driving force or hydride donor ability for selective C−H bond formation with CO2 (Scheme 1).
work in this area. Discussion here is limited to catalysts that produce formate selectively, although the majority of heterogeneous catalysts produce a wide variety of C1 and C2+ products at the expense of selectivity.3 Those products would also be valuable if accessed selectively, and much work is ongoing in that area. For formate production, we compare operating potential, stability, current density, and selectivity of heterogeneous formate-producing catalysts in water to give an overview of their performance (Table 1). Table 1. Summary of Heterogeneous Formate Producing Electrocatalystsa ref
catalyst
E (V)
FE formate (%)
j (mA cm−2)
Stability (h)b
4 5 6 7 8 9 11 12
Pb Pd on Pt 10% Pd/C SnO2 NCc Cu32H20L12d Pd70Pt30np/C [Fe4N(CO)12]− Ir(POCOP)e
−1.4 −0.83 −0.77 −1.8 −1.18 −1.05 −1.2 −1.65
98 100 99 93 89 88 96 93
0.3 0.3 4 5.5 ∼1 ∼5 4 0.6
95%, the applied overpotential for the reaction is only 230−440 mV (pH 7−13), and the best catalytic conditions are water at neutral pH.11 The catalyst performs similarly in MeCN solution although the selectivity for formate vs H2 production is much lower.17 In mixtures of MeCN and water (95:5), the selectivity of the catalyst for formate production falls between that observed in pure water and that observed in pure MeCN. Using the abundant element iron, these metrics and particularly the stability of the catalyst, which operates with no change for >24 h in water (Figure 6, Table 3), represent a significant
Figure 7. Structure of [H−Fe4N(CO)12]−, (H-1)−. Green, blue, gray, red, and white ellipsoids are Fe, N, C, O, and H atoms. Reproduced from ref 11. Copyright 2015 American Chemical Society.
the carbonyl clusters led to months of crystallization time and ultimately both (H-1)− and 12− cocrystallized at −25 °C. We believe that 12− reacted with trace water present in the THF to produce (H-1)−. Bond lengths and angles of the cluster core in (H-1)− and 12− were not well-resolved. Further mechanistic insights were obtained from CV measurements. We determined that the mechanism for formate production is first order in both 1− and CO2, and we proposed a mechanism for CO2 reduction where 1− is first reduced to 12−, and this reacts with protons to afford the hydride intermediate (H-1)− (Scheme 4). Reaction with CO2 afforded
Figure 6. Production of formate by 1−, pH 7, 24 h (−1.2 V vs SCE, GC electrode, black). Faradaic efficiency of formate production vs time (blue). Reproduced from ref 11. Copyright 2015 American Chemical Society.
Scheme 4. Proposed Mechanism for CO2 Reduction
Table 3. Production of Formate under 1 atm CO2a
1− 1− 2− 3−
solvent
E (V)
q (C)
pH 7 aq a/w a/w a/w
−1.2 −1.2 −1.4 −1.4
45 4 16
HCOO− (%)
H2 (%)
TOF (HCOO−)
95 94 61 b
4 5 36 96
106 15 7 b
a aq = aqueous. a/w = 0.1 M Bu4NPF6 MeCN/H2O (95:5) solution. TOF reported in unit h−1; CPEs run over 50−90 min. Control experiments can be found in the primary research articles.11,18 bNot applicable.
advance over known homogeneous and heterogeneous electrocatalysts for formate production. In separate experiments, we showed that clusters 4−−82− do not react with CO2 to give C−H bond containing products. This reactivity is considered again in the context of hydricity measurements presented below.
■
CHARACTERIZATION OF THE HYDRIDE INTERMEDIATE [H−Fe4N(CO)12]− Electrocatalysts that produce C−H bonds selectively from CO2 are rare, among both heterogeneous and homogeneous attempts to develop this chemistry. In efforts to understand the origins of the selectivity we observed with 1− in C−H bond formation, we were interested in the structure and the formation and reactivity properties of the key intermediate, which we proposed is (H-1)−. The structure of (H-1)− was determined from single crystals that contained a 50:50 mixture of (H-1)− and 12− (Figure 7). These crystals were obtained by reduction of 1− with Cp*2Co, in a reaction that targeted 12− in THF. The high solubility of
Reproduced from reference 11. Copyright 2015 American Chemical Society.
formate and 1, which is very quickly reduced back to 1− under the reaction conditions of −1.2 V. Work is still needed to determine the relative reaction rates for each of these elementary steps, to identify the rate-determining step, and to probe more details of the elementary step where hydride is transferred to CO2. Our conclusion that the reduced hydride intermediate, (H-1)−, is important in C−H bond formation with CO2 suggests that the reaction pathway could be altered by installation of a proton relay that enhances the rate of protonation of 2366
DOI: 10.1021/acs.accounts.7b00302 Acc. Chem. Res. 2017, 50, 2362−2370
Article
Accounts of Chemical Research (H-1)−: if successful, this experiment would provide more evidence for the existence of the intermediate (H-1)− and for our proposed reaction mechanism. We synthesized two derivatives of 1−, each containing phosphine ligands: 2− and [Fe4N(CO)11(PPh2(CH2)2OH)]− (3−) (Chart 1).18 Only 3− contains a proton relay, and 2− was used as a control compound. The reduction potentials of 2− and 3− are −1.45 and −1.47 V vs SCE, respectively (Figure 8).
Scheme 5. Proposed H2 Evolution by 3−
Reproduced from ref 18. Copyright 2016 Royal Society of Chemistry.
by Miller26 to address comparison between water and nonaqueous solvents for metal hydride complexes. We measured the pKa values and hydricity values for the catalytic intermediate, (H-1)−, in both water and MeCN/H2O (95:5) to provide insight as to why formate formation by 1− is more selective in water than it is in MeCN (Scheme 6).11 Scheme 6. Equations 1−6 Used To Determine Hydricity
Figure 8. CVs of 0.1 mM 3− in 0.1 M Bu4NPF6 MeCN, 1 atm N2 (black), in 0.1 M Bu4NPF6 MeCN/H2O (95:5), 1 atm N2 (red), or in 0.1 M Bu4NPF6 MeCN/H2O (95:5), 1 atm CO2 (blue). Scan rate 0.1 V s−1. Inset: Solid state structure of 3−. Green, blue, gray, red, pink, and white ellipsoids are Fe, N, C, O, P, and H atoms.
CPE experiments revealed that 2− produced formate selectively from a CO2-saturated solution of MeCN/H2O (95:5), while 3− produced only H2. Both clusters produce H2 when CPE experiments are performed under N2. The observation that a proton relay enables H2 evolution and prevents C−H bond formation with CO2 supports our hypothesis that a reduced hydride, such as (H-2)− or (H-3)−, is involved in the catalytic cycle. We proposed that the mechanism for H2 evolution by 3− proceeds similarly to that we described for 1− (Scheme 4). Following formation of (H-3)−, transfer of the proton from the relay to (H-3)− afforded H2 and 3, and 3 is reduced back to 3− (Scheme 5).
Hydricity (ΔG°H−) is defined as the free energy for loss of H− (eq 1 in Scheme 6) and can be used to predict the overall free energy of C−H bond formation with CO2 when used in combination with the known hydricity values for formate. Our estimates of thermochemical properties, pKa and ΔG°H−, were made by quantification of protonation equilibria and use of a thermochemical cycle that provided a value for ΔG°H− (eqs 2−6, Scheme 6). In MeCN solution, infrared spectroelectrochemistry (IR-SEC) was used to identify and quantify (using Beer’s Law) the catalytic intermediate (H-1)−, after it was generated by application of a reducing potential to 1− in the presence of either a weak acid or water. In aqueous solution, measurements of protonation equilibria were made using IR spectra of aliquots taken from CPE experiments. Summarizing these thermochemical results, we measured ΔG°H− for (H-1)− as 15.5 kcal mol−1 in water, and 49 kcal mol−1 in MeCN (Table 4). The increased hydricity of (H-1)− in water likely arises from stabilization of the hydride anion by water.24,25 These results indicate that the driving force for C−H bond formation by (H-1)− is favorable by 8.6 (in H2O) and unfavorable by 5 kcal mol−1 (in MeCN), and this corroborated our experimental observations where we saw greater selectivity
■
THE ROLE OF THERMOCHEMISTRY IN THE SELECTIVITY OF [H−Fe4N(CO)12]− No other molecular first row transition element electrocatalyst produces formate selectively from CO2 in water. In general, the formation of hydride intermediates leads to generation of H2 as a competing reaction, which inhibits C−H bond formation (Scheme 1). To better understand the properties of 1− that lead to such high selectivity in the transfer of hydride to CO2, we have investigated the kinetic and thermodynamic factors that play a role in determining the observed reaction pathways. These data also represent the first set of data for a functional H2 or formate producing electrocatalyst that were acquired in both MeCN and water. Previous reports of thermochemical measurements for functional catalysts have appeared only in nonaqueous solvents. Work on noncatalytic systems has been published by Creutz24 and more recently by Yang25 and 2367
DOI: 10.1021/acs.accounts.7b00302 Acc. Chem. Res. 2017, 50, 2362−2370
Article
Accounts of Chemical Research Table 4. Thermochemical Properties of (H-1)−, (H-2)−, (H-6)2−, H2,27,28 and formate28,29 in MeCN and H2O ΔG°H− (kcal mol−1)
To further understand hydricity and its role in reaction selectivity for hydride transfer to CO2, we compared the hydricity values (ΔG°H−) for the carbide-containing (H-6)2−, with those for the nitride-containing (H-1)−. In aqueous solution, hydricity values for (H-1)− and (H-6)2− are 15.5 and
