Skip to Main Content U.S. Department of Energy
Institute for Integrated Catalysis

Core Basic Energy Sciences Catalysis Program

Thrust 2: Reductive Conversion of CO2

This thrust focuses on mechanistic understanding of the catalytic reduction of CO2 to energy carriers (e.g., CH4, CH3OH, formaldehyde, and formic acid, as well as C2 and fuels with higher molecular weight). The bifunctional sites for CO2 reduction in enzymes guide the synthesis of sites in heterogeneous and homogeneous catalysts. Mechanistic aspects of the chemistries observedin different environments are compared to gain a molecular-level understanding of the fundamental reaction steps, identify the reaction intermediatesand their stability, and determine the role that reaction conditionscan have on the energetics and kinetics.

The unifying theme is the catalytic activity of single metal centers in the presence of second functionality on both heterogeneous and homogeneous catalysts aimed at comparing and contrasting thermal CO2 reduction in homogeneous and heterogeneous environments.

Designing Molecular Catalysts Using an Energy-Based Approach

Aaron Appel, R. Morris Bullock, John Linehan, Eric Wiedner

Figure 16
Figure 16. Production of fuels requires designing catalysts for the individual steps, illustrated here for converting CO2 to CH3OH.

This activity is focused on understanding and controlling the reactivity of potential catalysts for the reduction of CO2 to fuels, such as CH3OH (Figure 16). With this goal in mind, the conversion is divided into steps consisting of reduction by one equivalent of H2. For each step, we aim to control the thermodynamic driving force as well as kinetic parameters, such that we can design catalyst systems with unprecedented performance for critical transformations.

Figure 17
Figure 17. Catalytic performance can be improved by removing energy mismatches between catalytic intermediates, which are quantified using thermochemical studies.

Many different precious metal complexes have been demonstrated to be active for hydrogenation reactions; however, the number of non-precious metal complexes for the hydrogenation of CO2 is far more limited. One of our goals is to use our focus on thermodynamic and kinetic parameters to enable the use of non-precious metal complexes to perform a variety of energy storage reactions (Figure 17). The conversion of CO2 to formate is an example transformation, which is dependent on the transfer of H- from the metal complex to CO2. The favorability of hydride transfer can be predicted based on the hydricity of the metal-hydride complex relative to the hydricity of formate, and therefore can serve as a design parameter for the catalytic conversion of CO2 to formate.

Figure 18
Figure 18. A cobalt tetraphosphine catalyst was developed to hydrogenate CO2 to formate using weaker organic bases than previous cobalt hydride complexes.

We have previously demonstrated that cobalt hydride complexes are active catalysts for the hydrogenation of CO2 to formate but require the use of a very strong organic base, such as Verkade's superbase to regenerate the cobalt hydride complexes. Using DFT calculations previously reported from our laboratory, we predicted that a cobalt complex containing a tetradentate phosphine ligand would be a catalyst for CO2 hydrogenation when using a much weaker organic base (Figure 18). Experimental studies on the cobalt catalyst confirmed the computational prediction: this catalyst afforded a turnover frequency of 150 h-1 for CO2 hydrogenation at room temperature and low pressure (1.7 atm) using a guanidine base that is 7 pKa units weaker than Verkade's superbase. The unusual geometric constraints imparted by the tetradentate phosphine ligand were critical for attenuating the cobalt hydride bond strengths, thereby enabling catalysis using a weaker base.

Figure 19
Figure 19. A catalyst system based on copper for the hydrogenation of CO2 has been designed starting from copper-solvento complex (L=CH3 C(CH2 PPh2)3).

While heterogeneous catalysts based on copper and copper oxide are quite effective for the reduction of CO2, homogeneous complexes of copper have had very little use in the hydrogenation of CO2. This scarcity of copper complexes stems from the inability to regenerate the active hydride complexes using mild reagents and conditions. To understand the potential for regeneration of copper hydride complexes using H2, we have isolated and characterized a series of copper hydrides starting from a single complex, and we have determined the thermodynamic hydricity of one of the isolated hydrides (Figure 19). The resulting copper hydride complex can be generated from hydrogen and a base, and the hydride is sufficiently hydridic to reduce CO2 to formate.

Figure 20
Figure 20. By controlling the free energy for H‾ transfer, a catalyst system was designed for hydrogenation of CO2 in water.

In organic solvents, nickel hydride complexes are not catalytically active for the hydrogenation of CO2, essentially because the hydricities of these complexes are inadequate to transfer H‾ to CO2. However, the choice of condensed phase has now been shown to have a huge impact on the reactivity. Specifically, while nickel hydride complexes are thermodynamically incapable of transferring H‾ to CO2 in organic solvents, the same complexes undergo favorable H‾ transfer in water. Based on this impact, a new catalyst system was designed (Figure 20), which is catalytically active for hydrogenation of CO2 in water.

For many catalytic reactions, including hydrogenation of C=O bonds, controlling the heterolytic cleavage of the H-H bond of dihydrogen is critically important. We determined how the rate of reversible heterolytic cleavage of H2 can be controlled, spanning four orders of magnitude at 25 °C, from 2.1 x 103 s-1 to =107 s-1 (Figure 21). Bifunctional Mo complexes, [CpMo(CO)(κ3‾P2N2)]+ (P2N2 = 1,5-diaza-3,7-diphosphacyclooctane diphosphine ligand with alkyl/aryl groups on N and P), were developed for heterolytic cleavage of H2 into a proton and a hydride. The heterolytic cleavage of the H-H bond is enabled by the basic amine in the second coordination sphere, giving Mo hydride complexes bearing protonated amines, [CpMo(H)(CO)(P2N2H)]+.Variable temperature 1H, 15N, and 2D 1H-1H Rotating frame Overhause Effect NMR spectra indicated rapid exchange of the proton and hydride. The pκa values determined in acetonitrile range from 9.3 to 17.7 and show a linear correlation with the logarithm of the exchange rates. The proton-hydride exchange appears to occur by formation of a molybdenum dihydride or dihydrogen complex resulting from proton transfer from the pendant amine to the metal hydride. The exchange dynamics are controlled by the relative acidity of the [CpMo(H)(CO)(P2N2H)]+ and [CpMo(H2)(CO)(P2N2)]+ isomers, providing a design principle for controlling heterolytic cleavage of H2.

Figure 21
Figure 21. Splitting of H2 in multifunctional complexes resulted in formation an amine protonated metal hydride, in which the exchange of the proton and hydride was observed to occur at high rates that depend upon the acidity of the protonated pendant amine.

Thermodynamic Linear Scaling Relationships to Optimize Catalytic Reactions in Molecular Complexes

Tom Autrey, Bojana Ginovska, Abhi Karkamkar, Gregory Schenter

Figure 22
Figure 22. Linear scaling relationships predict that Co(dmpe)2 at the top of the volcano plot should be one of the most efficient catalysts in a series of cobalt P2 complexes in agreement with experimental rate comparisons.

We have used a combination of experimental and computational methods to determine the thermodynamic parameters of the key intermediates leading to the catalytic reduction of CO2 to formate. Our operating hypothesis is that a balance in the energy landscape minimizes energy sinks in the catalytic cycle leading to optimized rates. This section illustrates how concepts successfully utilized in heterogeneous catalysis can be used to test this hypothesis. Given the linear free energy scaling relationships established between H2 addition, proton abstraction, and hydride transfer in previous work, we should be able to predict the best homogeneous catalysts. The volcano plot shown in Figure 22 predicts that the cobalt P2 complex using dmpe as the ligand (green square) should be the best catalyst in the series given the energy balance. The free energy driving force for (1) dihydride formation (orange line), (2) proton abstraction from the dihydride to form the cobalt hydride (grey line) and (3) hydride transfer to CO2 to form formate (blue line) is shown as a plot of ∇Gpds vs. ∇Grrs3, confirming the hypothesis that balance thermodynamics play a crucial role in optimizing rates of catalytic reactions.

Modulating Catalysts with an Enzyme-Like Outer Coordination Sphere

Bojana Ginovska, John Linehan, Gregory Schenter, Wendy Shaw

This research focuses on introducing the principles in enzymes that result in superior catalytic performance into molecular catalysts. The protein scaffold surrounding the active site in an enzyme has been demonstrated to contribute significantly to the performance of the enzyme. Unlocking this contribution is challenging in enzymes because of the complex interplay between features. Ultimately, we will have all of these features in molecular complexes. To understand how each of them works, and what contribution they have, we need to understand each of them individually. We have been approaching this by introducing amino acids and peptides into molecular complexes to capture the features of enzymes into molecular catalysts and probe how each function contributes to catalysis. We are focused on three aspects of the outer coordination sphere: (1) the unique environment created by the functional groups around the active site; (2) the transfer of substrates such as protons into and out of the active site; (3) controlled structural movements that dictate activity. Each of these goals is being pursued within the context of our research program.

Figure 23
Figure 23. A series of amino acid and dipeptide-substituted complexes demonstrates the sensitivity of the performance to both the presence of the COOH group, and the side chain interactions. The data provide design principles for molecular complexes that are inspired by enzymes.

In the past year, we have made several key advancements in our understanding of the outer coordination sphere. We have achieved an electro-chemically reversible room temperature catalyst for H2 oxidation/production. This was achieved by introducing key attributes into the outer coordination sphere of a molecular complex. First, the first and secondcoordination sphere of the molecular complex, Ni(PCy2Namino acid2)2, were thermodynamically matched, based on extensive work done within the CME Energy Frontier Research Center. However, this alone was not enough to achieve reversibility. Features in the outer coordination sphere were essential to lower the overpotential, while maintaining fast rates (Figure 23). Specifically, a COOH group enabled fast proton transport to and from the active site, enabling proton-coupled electron transfer (PCET). Further, interactions between the side chains altered the active site structure to facilitate H2 addition and electron transfer. We further investigated a series of complexes with different side chains and with COOH or amide groups. We found the amides were able to transfer protons, but those complexes did not result in reversible complexes, even though they were faster than those with COOH groups. Of interest is the complexes in water are all faster and more efficient than in methanol.

We also worked with collaborators to understand the implications of binding these complexes to surfaces. This information will be essential as these complexes are transferred to more practical applications in fuel cells. In one study, we demonstrated that the complex with phenylalanine amino acids that was reversible in methanol, but not soluble in water in solution, was functional in water on the surface. In a second study, with the amino acid arginine attached to the complex, putting the complex onto carbon nano-tubes and then putting this into a fuel cell shows performance within an order of magnitude of a fully Pt fuel cell. A scientific outcome of this study was that that relative performance of the glycine-substituted complex and the arginine-substituted complex were the same as in solution. This implies that binding these complexes to solution did not alter the active site, one of the first demonstrations of non-intrusive binding.

Further, we have been developing synthetic strategies to attach full peptides to any type of ligand, including a phosphine protection strategy and click chemistry to attach the ligand to nearly any point in a peptide that is appropriately substituted. These approaches should provide significant flexibility to use these ligands with a multitude of peptides so we can evaluate the impact of the outer coordination sphere with relative ease by including systematic changes.

High Surface Area Model Catalysts: Controlling Selectivity through Mechanistic Understanding

Janos Szanyi

The catalytic reduction of CO2 requires bifunctional catalysts as both CO2 and H2 need to be activated before their reaction. We have shown that in oxide-supported metal catalysts, the metal component dissociated H2, while the oxide support activated CO2 by binding it as bicarbonate. The focus of our work has been understanding what governs product selectivity in CO2 hydrogenation over Pd/γ-Al2O3 catalysts. To this end, we have conducted SSITKA/operando FTIR measurements on Pd/γ-Al2O3 catalysts to identify the nature and the kinetics of surface species present on the catalysts under steady-state conditions.

Table 1. Real mean surface residence times for ICH (‾τ0_ICH) and ICO (‾τ0_ICH), and activation energies for CH4 (ECH4) and CO (ECO) formation and adsorbed *COs (E*COs) and HCOO* (EHCOO*) conversions in CO2 reduction at 533-573 K. Uncertainties in activation energies are ± 2 kJ mol-1.

We found (Table 1) that at low temperature (533 K), the real mean surface residence time of the intermediate leading to CO formation was shorter (107 sec) than that of the intermediate leading to CH4 production (134 sec). However, at 573 K, the real mean surface residence time of the intermediate leading to CO formation was longer (55 sec) than that leading to CH4 formation (41 sec). Interestingly, the activation energies of CO and CH4 formation estimated from the real mean surface residence times were identical to those determined from the IR intensities of alumina-bound bidentate formates and Pd-adsorbed CO, respectively. These results let us conclude that the key intermediate in CO formation is formate, while in CH4 formation it is strongly adsorbed CO. The selectivities toward CO and CH4 productions were governed by the balance between the reaction rate of adsorbed formate reduction to CO, and the rate of strongly held CO hydrogenation to CH4. At low reaction temperatures, the methanation rate is low, while the formate hydrogenation rate is relatively high. Therefore, the Pd particles are populated with both strongly and weakly adsorbed CO, and most of the weakly held CO desorbed into the gas phase. This results in a high selectivity toward CO, and low selectivity toward CH4 formations. In contrast, at high temperatures where the methanation rate is higher than the formate reduction rate, most of the CO produced in the formate reduction will adsorb strongly on the Pd particles, and it will further be reduced to CH4. Under these conditions, the pool of weakly adsorbed CO is only partially full, or empty. This results in high CH4 formation, but low CO formation selectivities.

Figure 24
Figure 24. Controlling selectivities in CO2 re-duction by tailor-made catalysts through mechanistic understanding: CH4 and CO selectivities and CO2 conversion as a function of Pd loading at 573 K.

To test this hypothesis, we prepared three Pd/Al2O3 catalysts with Pd loadings of 2.5, 5, and 10 wt%, with very similar particle size distribution (estimated from TEM analysis). The changes in CO and CH4 selectivities at comparable CO2 conversion levels as a function of metal loadings are displayed in Figure 24. The catalyst with the lowest Pd loading exhibited the highest CO selectivity, while the catalyst with the highest metal content showed the highest CH4 selectivity. The pool of the key intermediate to adsorbed CO formation (formate) is full in all three cases. However, the pool of sites that can hold adsorbed CO (sites on Pd clusters) varies strongly with metal loading. At low Pd loading, the pool of adsorbed CO is small, so lot of the CO from the reduction of adsorbed formates ends up at weak adsorption sites, and desorbs into the gas phase (high CO selectivity). On the other hand, at high metal loading, due to the large pool of adsorption sites on Pd clusters, most of the CO from formate reduction is bound to strong adsorption sites and can be further reduced to CH4. As a result, this catalyst produces CH4 with high selectivity.

Institute for Integrated Catalysis

Research & Development


Seminars & Events


Get the Latest News

Transformations Newsletter



IIC Director
Johannes Lercher

Working With Us

Download Our Brochure

Catalysis Research for Energy Independence

In Memory