Toward Combined Carbon Capture and Recycling: Addition of an Amine Alters Product Selectivity from CO to Formic Acid in Manganese Catalyzed Reduction of CO₂

Abstract

Owing to the energetic cost associated with CO₂ release in carbon capture (CC), the combination of carbon capture and recycling (CCR) is an emerging area of research. In this approach, “captured CO₂,” typically generated by addition of amines, serves as a substrate for subsequent reduction. Herein, we report that the reduction of CO₂ in the presence of morpholine (generating mixtures of the corresponding carbamate and carbamic acid) with a well-established Mn electrocatalyst changes the product selectivity from CO to H₂ and formate. The change in selectivity is attributed to in situ generation of the morpholinium carbamic acid, which is sufficiently acidic to protonate the reduced Mn species and generate an intermediate Mn hydride. Thermodynamic studies indicate that the hydride is not sufficiently hydritic to reduce CO₂ to formate, unless the apparent hydricity, which encompasses formate binding to the Mn, is considered. Increasing steric bulk around the Mn shuts down rapid homolytic H₂ evolution rendering the intermediate Mn hydride more stable; subsequent CO₂ insertion appears to be faster than heterolytic H₂ production. A comprehensive mechanistic scheme is proposed that illustrates how thermodynamic analysis can provide further insight. Relevant to a range of hydrogenations and reductions is the modulation of the hydricity with substrate binding that makes the reaction favorable. Significantly, this work illustrates a new role for amines in CO₂ reduction: changing the product selectivity; this is pertinent more broadly to advancing CCR.

Graphical Abstract

INTRODUCTION

The catalytic reduction of CO₂ to fuels and fuel precursors such as CO, formic acid (FA), and MeOH represent recycling efforts pertinent to future energy schemes.^1,2 The coupling of this to carbon capture (CC) in a carbon capture and recycling (CCR) scheme offers a promising approach for mitigation of global warming and development of carbon-neutral fuels.^3,4 Presently, CC relies on the reaction of CO₂ with amines to give equilibrium mixtures of carbamic acid, carbamate, and bicarbonate (if water is present). The captured CO₂ is then released to serve as a substrate for recycling efforts. This process is not without challenges, including the cost associated with the release of CO₂; using alkanolamines, it is estimated that to capture and release all the CO₂ from a powerplant, one would consume 25–40% of the energy output.^3,5 As such, there is much interest in developing more robust and energy-efficient CC technologies.^6–9 An alternative approach is to circumvent CO₂ release in CCR efforts,¹⁰ hence using the captured CO₂ (carbamate/carbamic acid) as a substrate for reduction.

The idea of using captured CO₂ as a substrate is gaining traction, though studies remain limited. Sanford,¹¹ we,¹² and others^10,13–16 have shown that a carbamate can be hydrogenated to MeOH, likely via release of CO₂ under the reaction conditions. With regards to CO₂ reduction, Ishitani established that in the presence of triethanolamine (TEOA),¹⁷ both photo-¹⁸ and electrocatalytic¹⁹ reduction of CO₂ to CO is accelerated at a Re catalyst. TEOA coordinates to the Re and CO₂ inserts into the alkoxide to generate a Re-carbonate species; how subsequent reduction from this species occurs is not known. Fujita also found that TEOA enhances photocatalytic reduction of CO₂ to formate with a Ru catalyst;²⁰ it is proposed that a zwitterionic alkylcarbonate lowers the kinetic barriers of several key steps. In contrast to Ishitani’s system, coordination of the carbonate is not thought to be relevant to the reduction. These examples make use of the direct interaction of amines or the alcohol functionality of alkanolamines with CO₂ and hence are pertinent to a combined CCR approach.

Amines can also facilitate the electrocatalytic reduction of CO₂ without directly interacting with CO₂. Costentin and Nocera have shown that the binding of an amine to an Fe porphyrin catalyst allows more facile protonation of a bound CO₂ to give formate.²¹ Marinescu has shown that pendent amines can hydrogen bond and stabilize a Co-COOH intermediate.²² Finally, Baik, Skrydstrup, and Daasbjerg have shown that pendent amines on a Mn electrocatalyst enhance formation of an intermediate Mn–H, allowing for reduction to give formate over CO.²³ These studies illustrate how amines, present in CC, may enhance electrocatalytic reduction of CO₂ indirectly.

Herein, we investigate the role of added amine on the electrocatalytic reduction of CO₂. This study makes use of two Mn catalysts of the type (N^N)Mn(CO)₃X whereby N^N is a bipyridine ligand (bpy = 2,2′-bipyridine; mesbpy = 6,6′-dimesityl-2,2′-bipyridine) and X is an X-type ligand.^24–26 These complexes are chosen because they show excellent Faradaic efficiency (FE) for electrocatalytic reduction of CO₂ to CO, and numerous studies provide a detailed understanding of the mechanisms for this transformation.^{24,25,27–36} We find that in the presence of an amine, the product selectivity shifts from CO to mixtures of H₂ and FA. Mechanistic and thermochemical studies indicate that the amine generates a more acidic environment, allowing for a Mn–H to be a kinetically competitive intermediate (relative to the hydroxycarbonyl, Mn-COOH, formed en route to CO production). Investigation of the two complexes that differ in the steric bulk of the ligands shows that when bimetallic H₂ production is feasible, this reactivity supersedes that with CO₂, and H₂ is the dominant product. When a bimetallic pathway for H₂ evolution is not possible, CO₂ insertion into the Mn–H bond to give a formate species, Mn-OCHO, is faster than protonation of Mn–H to give H₂ and a Mn(I) species. This work provides the first thermochemical analysis of a Mn electrocatalyst, which aids in understanding how the product selectivity can be tuned. We find that the free energy associated with binding of formate to the catalyst must be considered. Moreover, this work establishes a new role for amines in combined CCR which is that of controlling the selectivity of the reaction.

RESULTS AND DISCUSSION

A. Synthesis and Characterization of Mn Complexes.

The formally Mn(I) species (bpy)Mn(CO)₃CN,³⁷ [(bpy)Mn(CO)₃(MeCN)][OTf],³⁸ (mesbpy)Mn(CO)₃Br,²⁶ and [(mesbpy)Mn(CO)₃(MeCN)][OTf]²⁶ were all prepared following literature procedures. The 2-electron reduced congeners that are formally Mn(-I), [(bpy)Mn(CO)₃][Na], and [(mesbpy)Mn(CO)₃][Na] were prepared by slight modification of literature procedures;^26,38 we found that using Na/Hg as a reductant as opposed to KC₈ gave identical results and allowed us to use a safer reductant (see SI).

To gain insight into the electrochemical reactions, the cyclic voltammograms (CVs) of the two anionic Mn(-I) species were collected in MeCN with 0.1 M TBAPF₆ electrolyte and referenced versus Fc^+/0.

The CV of anionic [(mesbpy)Mn(CO)₃][Na] shows a single 2-electron oxidation centered at −1.57 (±0.01) V (See SI). The electrochemical response is consistent with the Mn(0) species not undergoing any chemical reaction such as dimerization at the electrode and is similar to the reported 2-electron reduction of cationic [(mesbpy)Mn(CO)₃(MeCN)][OTf] (reported at −1.55 V).³⁵

By contrast, the CV of [(bpy)Mn(CO)₃][Na] shows two irreversible oxidations at −1.47 (±0.01) V and −0.61 V and two ireversible reductions at −1.49 (±0.01) V and −1.87 V (Figure 1, purple). The oxidation at −1.47 V is assigned to the one-electron oxidation of Mn(-I) to Mn(0). The Mn(0) species is known to dimerize very rapidly,^27,39 which occurs before the Mn(0) can be fully oxidized to the Mn(I). Evidence for the dimerization is in the second oxidation at −0.61 V, which corresponds to the oxidation of {(bpy)Mn(CO)₃}₂ to [(bpy)Mn(CO)₃(MeCN)]⁺.²⁵

Figure 1.

CVs of 1 mM (top orange): (bpy)Mn(CO)₃CN; (middle purple): [(bpy)Mn(CO)₃]⁻; and (bottom red): [(bpy)Mn(CO)₃(MeCN)]⁺ in 0.1 M TBAPF₆/MeCN solvent, with a scanrate of 0.1 V/s. S designates a coordinated solvent molecule. Scan directions and starting potentials are indicated by the black arrows. The electrochemical reactions that are occurring are shown; unless labeled, the reactions that occur at the same potentials correspond to the same reaction. The Mn(0) is unstable in solution and dimerizes once it is formed.

On the return reductive scan, the reduction at −1.49 V corresponds to the one electron reduction of [(bpy)Mn(CO)₃(MeCN)]⁺ to give (bpy)Mn(CO)₃, which dimerizes in solution to give {(bpy)Mn(CO)₃}₂. This interpretation is confirmed by examining the first cathodic scan of [(bpy)Mn(CO)₃(MeCN)]⁺ which also features a reduction at −1.49 V (Figure 1, red) to give (bpy)Mn(CO)₃. The Mn(0) complex dimerizes to {(bpy)Mn(CO)₃}₂ before the second reduction can occur, and hence the reduction at −1.87 V corresponds to the reduction of the dimer to [(bpy)Mn(CO)₃]⁻. Additionally, if the switching potential in the CV of [(bpy)Mn(CO)₃][Na] is set to −1.0 V, therefore precluding oxidation of the dimer to the cation, then the only reductive peak observed is that at −1.87 V, corresponding to the reduction of the dimer. From the above analysis, the 2-electron reduction of [(bpy)Mn(CO)₃(MeCN)]⁺ is taken to be −1.48 V.

The CVs are similar to that of (bpy)Mn(CO)₃CN, in that, redox events associated with the dimer are present (Figure 1, top), and monomer redox events are shifted due to coordination of cyanide.²⁵ It is prudent to point out that the reduction of {(bpy)Mn(CO)₃}₂ to give anionic [(bpy)Mn(CO)₃]⁻ occurs at a very similar potential to the reduction of (bpy)Mn(CO)₃CN to the neutral Mn(0) (vide infra).

B. Electrocatalytic Reduction of CO₂.

Given that (bpy)Mn(CO)₃X (X = Br, CN)^24,25 and [(mesbpy)Mn(CO)₃(MeCN)][OTf]²⁶ are electrocatalysts that reduce CO₂ to CO in the presence of weak acids, we sought to determine if addition of an amine would have a positive effect on the catalysis. Under an atmosphere of CO₂, morpholine gives an equilibrium mixture of the corresponding carbamic acid and carbamate (Scheme 1); reduction at a carbon electrode occurs with minimal electrode fouling.⁴⁰ Given that the pK_a of morpholinium (morph-H⁺) is 16.6 in MeCN,⁴¹ the pK_a of morph-COOH is estimated to be ~ 17. Thus, morpholine was chosen as an additive to the electrocatalytic conditions to see if the resulting carbamic acid or carbamate may serve as a substrate (see SI for studies with TEOA and DEA).

Scheme 1.

Equilibrium between morpholine and CO₂

As reported by Kubiak, (bpy)Mn(CO)₃CN electrocatalytically reduces CO₂ to CO at −1.94 V in the presence of PhOH.²⁵ We have replicated these results, obtaining very similar current densities (Figure 2b, dashed). Subsequently, we investigated the effect of morpholine on this electrocatalyst. The CV of 1 mM (bpy)Mn(CO)₃CN under CO₂ in the presence of 0.1 M morpholine and CO₂ shows a catalytic current at −1.96 V (Figure 2b, purple). The subsequent addition of PhOH (0.5 M – 1.25 M final concentration) shows a 16-fold enhancement of the current, indicating a synergistic effect on the catalytic current (Figure 2b). The resulting morph-COOH and morph-H⁺ are more acidic than PhOH (pK_a = 29.1 in MeCN)⁴² which suggests that it may serve as the proton source.

Figure 2.

CVs (100 mV/s) of the catalysts (1 mM) in the presence of CO₂ and the additives noted in the legends. For clarity, the return oxidation current is not shown. (a) Effect of increasing morpholine on the catalysis of (bpy)Mn(CO)₃CN. (b) Effect of increasing PhOH on the catalysis of (bpy)Mn(CO)₃CN in the presence and absence of morpholine. (c) Effect of increasing PhOH on the catalysis of (mesbpy)Mn(CO)₃Br in the presence and absence of morpholine.

Indeed, the CVs of 1 mM (bpy)Mn(CO)₃CN in the presence of various equiv of morpholine under a CO₂ atmosphere show a catalytic current at the same onset potential as when PhOH is present. The addition of morpholine is limited by the poor solubility of the morpholinium carbamate species beyond 0.1 M in MeCN.⁴⁰ With the addition of 0.1 M morpholine, the catalyst shows a 3.5-fold current enhancement (Figure 2a).

In all instances, catalysis occurs at the same onset potential. This suggests that the limiting reduction is the same in the presence and absence of morpholine. Kubiak suggests that catalysis with (bpy)Mn(CO)₃CN ensues from a rate-limiting disproportionation of [(bpy)Mn(CO)₃CN]⁻ which after CN loss generates the active [(bpy)Mn(CO)₃]⁻.²⁵ However, our CV studies indicate that the reduction of the dimer occurs at a very similar potential (Figure 1) and coupled to the low concentration of cyanide in solution suggests that reduction of the dimer to the anion may be limiting, as proposed for (R-bpy)Mn(CO)₃Br.^24,34 This latter scenario may be more likely in controlled potential electrolysis experiments, whereby the solution is stirred.

The CVs in the presence of CO₂, PhOH, and morpholine were repeated with (mesbpy)Mn(CO)₃Br which does not undergo dimerization. In the presence of PhOH (and no morpholine), the resulting CVs (Figure 2c) are similar to those obtained when MeOH is used as the proton source.²⁶ Upon addition of morpholine, several changes are observed. First, the initial plateau current shifts ~100 mV more positive. This initial plateau represents slow catalysis. In the presence of CO₂ and MeOH²⁶ (or PhOH), the plateau is anodically shifted relative to the reduction of [(mesbpy)Mn(CO)₃(MeCN)]⁺ because of binding and protonation of CO₂. The shift is noticeably more pronounced in the presence of morpholine which may suggest that a different electrophilic species such as morph-COOH is binding to the reduced Mn (EC mechanism). Alternatively, the shift may be due to reduction of a species such as (mesbpy)Mn(CO)₃(OOC-morph) that would form if morph-COO⁻ first binds the oxidized Mn (CE mechanism). Next, the catalytic current density at ~ −2 V is enhanced with the addition of morpholine, and the onset of this catalytic wave is shifted ~100 mV more positive.

To establish the product distribution from CO₂ reduction, controlled potential electrolysis (CPE) was carried out under CO₂ with the conditions given in Table 1. Consistent with literature precedence, in the presence of only the weak acid PhOH, (bpy)Mn(CO)₃CN and (mesbpy)Mn(CO)₃Br produce CO exclusively from CO₂ (entries 2 and 6).^25,26,43 When both morpholine and PhOH are present, (bpy)Mn(CO)₃CN primarily makes H₂ with very little CO (entry 3), and in the presence of only morpholine (entry 1), H₂ and FA are produced. Using (mesbpy)Mn(CO)₃Br as a catalyst gives twice as much FA as H₂ (entries 5, 7) when morpholine is present.

Table 1.

Product Distribution from Controlled Potential Electrolysis under the Conditions Specified^d

Entry	Catalysta	Additivea	(H2+FA):COb	H2:FAb
1	(bpy)Mn(CO)3CN	0.1 M morph	90.5:9.5	66:34
2	(bpy)Mn(CO)3CN	0.5 M PhOH	4:96	1:0
3	(bpy)Mn(CO)3CN	0.1 M morph 0.5 M PhOH	97:3	1:0
4	(bpy)Mn(CO)3CN	0.1 M [morph-H] [BF4]c	1:0	1:0
5	(mesbpy)Mn(CO)3Br	0.1 M morph	44:56	31:69
6	(mesbpy)Mn(CO)3Br	0.5 M PhOH	0:1
7	(mesbpy)Mn(CO)3Br	0.1 M morph 0.5 M PhOH	96:4	34:66
8	(mesbpy)Mn(CO)3Br	0.1 M [morph-H] [BF4]c	1:0	1:0
9	(mesbpy)Mn(CO)3Br	0.1 M TEOA	0:1	--

^aConditions: 1 mM catalyst in 10 mL 0.1 M TBAPF₆/MeCN solution; see SI for full details.

^bObtained from Faradaic efficiencies; see SI.

^cUnder N₂.

^dAll electrolysis was conducted under 0.85 atm CO₂ and at −2.2 V vs. Fc^+/0.

While it is not known whether CO₂ or carbamic acid is the substrate, the addition of morpholine changes the product selectivity from CO to H₂ and FA, representing a new role of amines in a combined CCR scheme. Related Mn catalysts that feature pendent amines show a similar shift in the product selectivity.²³ However, the approach here, whereby added amines change the product selectivity, also has the potential to concentrate the CO₂, yielding higher current densities;⁴⁴ this is particularly attractive for developing catalysts that function in water.⁴⁵

Catalysts of the type (bpy)Mn(CO)₃X (X = Br, CN) have been shown to photochemically reduce CO₂ to mixtures of FA, CO, and H₂ with a Ru photosensitizer and TEOA.^46,47 However, the Ru photosensitizer employed itself can reduce CO₂ to formic acid in the presence of TEOA, albeit with reduced turnover numbers.⁴⁷ When TEOA is used as a CO₂-capturing agent and proton source for electrocatalytic reduction of CO₂ with (mesbpy)Mn(CO)₃Br, only CO is produced (Table 1, entry 9).

H₂ evolution from [(mesbpy)Mn(CO)₃(MeCN)]⁺ is known to occur with stronger acids.⁴⁸ Given that all experiments with morpholine and CO₂ give some H₂, it is likely that the morpholine aids in the production of a Mn–H intermediate (Scheme 2). Indeed, CPE under N₂ and morpholinium solely gives H₂ (Table 1, entries 4, 8). As morph-COOH is estimated to have a similar pK_a to morpholinium, the morph-COOH, generated in situ likely protonates [(N^N)Mn(CO)₃]⁻ to give the intermediate (N^N)Mn(CO)₃H. The morpholinium may then protonate the formate to generate free FA. It should be emphasized that either morph-H⁺ or morph-COOH may serve as the acid in these reactions.

Scheme 2.

Effect of Acid on the Product Distribution of CO₂ Reduction

In the absence of morpholine, the only acid present is PhOH, which is not sufficiently acidic to generate the hydride. Rather, the anionic Mn binds CO₂ in the presence of PhOH to give a hydroxycarbonyl intermediate en route to CO production (Scheme 2). The higher current densities observed with both morpholine and PhOH may be due to facilitation of proton-transfer to the bound CO₂ or different products being formed.^22,49

C. Characterization and Reactivity of (N^N)Mn(CO)₃H.

To corroborate that H₂ is produced from (N^N)Mn(CO)₃H, the preparation of the hydrides via protonation of the corresponding anions was undertaken.

Treatment of [(bpy)Mn(CO)₃]⁻ with 5 equiv of PhOH in CD₃CN results in rapid precipitation and a color change to dark red. After 4 min, the major Mn species in solution is the dimer, {(bpy)Mn(CO)₃}₂ (Scheme 3). A small resonance at −3.02 ppm is consistent with formation of (bpy)Mn(CO)₃H,⁵⁰ and H₂ is observed at 4.57 ppm. As described in the literature, the hydride is not stable and fully converts to the dimer⁵⁰ upon losing H₂. IR analysis of the precipitate indicates that it is comprised of {(bpy)Mn(CO)₃}₂. Attempts to intercept the hydride by addition of CO₂ to a thawing solution of in situ generated (bpy)Mn(CO)₃H to give a Mn-OCHO species were unsuccessful; the only product obtained was {(bpy)Mn(CO)₃}₂ (see SI).

Scheme 3.

Observed Reactivity of (N^N)Mn(CO)₃(H)

Taken together, this suggests that a homolytic pathway for H₂ formation is operative, whereby two (bpy)Mn(CO)₃H react to lose H₂ (Scheme 3, top). To determine if a heterolytic pathway is also present, whereby the intermediate (bpy)Mn(CO)₃H is protonated by PhOH in solution, [(bpy)Mn(CO)₃]⁻ was treated with 20 equiv of PhOH. If a heterolytic pathway occurred at a competitive rate, then a Mn(I) species should result. The NMR spectrum of this reaction instead showed only resonances of the dimer, indicating that the heterolytic pathway is not competitive.

Given that the mechanism for H₂ production from [(bpy)Mn(CO)₃]⁻ is homolytic, it was anticipated that the corresponding (mesbpy)Mn(CO)₃H may be more stable, as the steric bulk precludes dimerization. NMR analysis of the reaction of [(mesbpy)Mn(CO)₃]⁻ with 5 equiv of PhOH shows a new set of resonances consistent with (mesbpy)Mn(CO)₃H that persist over several hours. Indeed, addition of CO₂ results in a reaction that gives (mesbpy)Mn(CO)₃(OCHO). This was confirmed by independent synthesis of (mesbpy)Mn(CO)₃(OCHO) (see SI). A related (N^N)Mn(CO)₃(OCHO) complex has recently been reported in the literature.⁵¹

Over time, the hydride resonance at −2.83 ppm disappears as H₂ is produced, and a new set of resonances consistent with a Mn(I) species assigned as (mesbpy)Mn(CO)₃(OPh) emerges (Scheme 3). This assignment was corroborated by independent synthesis of the phenolate species. If NaBr is present in solution, then (mesbpy)Mn(CO)₃Br is also observed. The formation of Mn(I) species suggests that heterolytic H₂ production is occurring.

The observation that PhOH can protonate the anionic species suggests that formate and/or H₂ could be produced in experiments whereby only PhOH and CO₂ are added. Only CO is produced under those conditions which suggests that the kinetics associated with initial CO₂ binding are more favorable than those of protonation, as has been observed on an analogous Re system.⁵²

D. Thermodynamic Analysis.

As the Mn catalysts can produce both FA and H₂, it is of interest to determine the hydricity of the resulting hydrides. As shown in Scheme 4, the hydricity can be obtained from knowledge of the pK_a of (N^N)Mn(CO)₃H and the reduction potentials.⁵³

Scheme 4.

Thermochemical cycle for determination of the hydricity of (N^N)Mn(CO)₃H. For simplicity, (N^N)Mn(CO)₃ is abbreviated as Mn

The free energy change associated with hydride formation in MeCN (eq 4) is known,⁵³ and those associated with eqs 1–3 for the two complexes have been measured and are summarized in Table 2.

Table 2.

Summary of the Thermochemical Properties for the Two Catalysts

Thermochemical parameter	(bpy)Mn(CO)3H	(mesbpy)Mn(CO)3H
pKa (experimental)	28.8 (±0.8)	28.5 ± 0.9
apKa (extrapolated)	26.4 (±0.1)	28.0 (±0.2)
a$E_{\mathit{Mn}0}^o$	−1.48 (±0.01) V	−1.57 (±0.01) V
b$E_{\mathit{Mn}+}^o$	−1.48 (±0.01) V	−1.57 (±0.01) V
$\Delta G_{\mathrm{H}-}^{\mathrm{o}}(\text{experimental})$	50.7 (±1.1) kcal·mol−1	46.1 (±1.3) kcal·mol−1
d$\Delta G_{\mathrm{H}-}^{\mathrm{o}}(\text{extrapolated})$	47.5 (±0.4) kcal·mol−1	45.4 (±0.4) kcal·mol−1

^aFrom a plot of pK_a versus reduction potential; see SI and ref.⁵⁴

^bReduction potential of the corresponding Mn(0) species, (N^N)Mn(CO)₃, versus Fc^+/0.

^cReduction potential of the corresponding Mn(I) species, [(N^N)Mn(CO)₃]⁺, versus Fc^+/0.

^dUsing the extrapolated pK_a.

The pK_as of both Mn–H species were obtained from titrations of [(N^N)Mn(CO)₃]⁻ with various equivalents of PhOH (see SI). From this equilibrium and the known pK_a of PhOH (29.1 in MeCN),⁴² pK_a values of 28.8 ± 0.8 and 28.5 ± 0.9 are obtained for (bpy)Mn(CO)₃H and (mesbpy)Mn(CO)₃H, respectively. It should be noted that the measured pK_as are consistent with the observed reactivity described in the previous section of the anions toward PhOH. The titrations are done at early time points before the subsequent reactions can shift the equilibrium. Due to the instability of (bpy)Mn(CO)₃H toward dimerization and H₂ loss, the pK_a was also estimated from a plot of pK_a versus reduction potential (see SI),⁵⁴ which gives a linear fit across several transition metal hydride species. The analysis provides a pK_a value of 26.4 ± 0.1 for (bpy)Mn(CO)₃H, indicating that the instability of (bpy)Mn(CO)₃H increases the apparent pK_a. Repeating the analysis with (mesbpy)Mn(CO)₃H provides a pK_a value that is within error of the experimental value (Table 2), validating the approach to extrapolate the pK_a for this system.

A combination of the 2-electron reduction potential of [(mesbpy)Mn(CO)₃]⁺ with the measured pK_a of (mesbpy)Mn(CO)₃H gives a hydricity value of 46.1 ± 1.3 kcal·mol⁻¹, in agreement with the value obtained using the extrapolated pK_a. The hydricity of formate in MeCN is 44 kcal·mol⁻¹ (Scheme 5 and caption),⁵⁵ and hence the hydride transfer reaction is endergonic by ~2.1 kcal·mol⁻¹ (eq 7 of Scheme 5). However, the reaction of eq 7 is not the reaction that is observed, as it gives [(mesbpy)Mn(CO)₃]⁺ and OCHO⁻. Instead, the CO₂ inserts to give (mesbpy)Mn(CO)₃(OCHO), and hence the free energy change of the observed reaction (eq 9) must be considered. This analysis encompasses the free energy change associated with hydride transfer (eq 7) and binding of the formate to the cationic Mn (eq 8) to give the apparent hydricity. Because eq 9 is favorable, the free energy of binding formate to the Mn must be < −2.1 kcal·mol⁻¹ (eq 8). This value was compared favorably to the value we measured of −3.7 kcal·mol⁻¹ for formate binding to a cationic Ru species.¹² The modulation of hydricity by the binding of a ligand after hydride has been observed in both water and THF solvent.^12,56

Scheme 5. Thermochemical Equations Pertinent to Formate Production^a.

^aFor simplicity, (N^N)Mn(CO)₃ is abbreviated as Mn. Except for eq 6, free energy values are provided for N^N = mesbpy (bpy). The hydricity of formate under the conditions presented may differ slightly, given that we are not operating at 1 atm of CO₂, and that the high concentrations of morpholine and PhOH may alter the CO₂ solubility.

Given that morph-H is more acidic than FA (pK_a = 20.7),⁵⁷ the end product of reduction is FA and not formate. The FA could be directly protonated off from (mesbpy)Mn(CO)₃(OCHO) (Scheme 6), or formate release may ensue upon reduction, which is then protonated by morph-H (Scheme 7). Given the differences in pK_as of FA and morph-H (eqs 10–11) for direct protonation of the formate, the free energy associated with formate binding to the Mn must be greater than −5.6 kcal·mol⁻¹ (eq 12). As the upper-limit for this value is −2.1 kcal·mol⁻¹ (eq 8), direct protonation of bound formate in (mesbpy)Mn(CO)₃(OCHO) may be feasible.

Scheme 6. Thermochemical Equations Pertinent to Direct Protonation of the Bound Formate^A.

^AFor simplicity, (N^N)Mn(CO)₃ is abbreviated as Mn. Free energies are in units of kcal·mol⁻¹.

Scheme 7. Proposed Mechanisms for the Electrocatalytic Reduction of CO₂ to Give CO (Top) or FA (Bottom)^a.

^aH₂ production is indicated by gray arrows, and the stoichiometry shown is for the heterolytic pathway. Direct protonation of Mn(OCHO) is indicated by light purple arrows, and reduction followed by formate loss is indicated by dark purple arrows.

Following Scheme 4, the hydricity of (bpy)Mn(CO)₃H is found to be 47.5 ± 0.4 kcal·mol⁻¹. Because of the error in the experimentally measured pK_a, this analysis uses the extrapolated pK_a (Table 2). Comparison of the hydricity values indicates that (bpy)Mn(CO)₃H is ~ 2 kcal·mol⁻¹ less hydridic than (mesbpy)Mn(CO)₃H (in comparing the hydricities obtained from the extrapolated pK_a).

Formate production from (bpy)Mn(CO)₃H is unfavorable by 3.5 kcal·mol⁻¹. For the reaction that gives (bpy)Mn(CO)₃(OCHO) to be favorable, the binding of formate must be < −3.5 kcal/mol (eq 8). Though we have not been able to prepare this complex via CO₂ insertion into (bpy)Mn(CO)₃H, we attribute the lack of observed insertion to kinetics (insufficient gas-mixing), not thermodynamics, as (bpy)Mn(CO)₃(OCHO) can be prepared by alternative means (see SI). The limit for formate binding does not allow to distinguish between formate loss from (bpy)Mn(CO)₃(OCHO) occurring from direct protonation or reduction.

The product distributions of the CPE can be understood in terms of the kinetics and thermodynamics associated with the intermediate Mn–H (Scheme 7). In the absence of morpholine, the only acid capable of protonating [(N^N)Mn(CO)₃]⁻ is PhOH. From the difference in pK_as, this reactivity is unfavorable by ~3.7 and 0.8 kcal·mol⁻¹ for [(bpy)Mn(CO)₃]⁻ and [(mesbpy)Mn(CO)₃]⁻, respectively. Hence, the CO₂ coordinates, and the resulting CO₂ is protonated to give the hydroxycarbonyl species that ultimately provides CO (Scheme 7, top).

In the presence of morpholine, morph-COOH forms and can serve as an acid with a pK_a of ~ 17. Now this acid can protonate both [(N^N)Mn(CO)₃]⁻ species to (N^N)Mn(CO)₃H. The hydride can then insert CO₂ to give formate, be protonated to give H₂ and (mesbpy)Mn(CO)₃X, or dimerize to give H₂ and {(bpy)Mn(CO)₃}₂. In all instances, reduction back to the anion closes the catalytic cycle.

As shown in Figure 3, the binding of formate can modulate the apparent hydricity, allowing FA to be produced with catalysts that otherwise would not be expected to reduce CO₂ to FA. H₂ and FA are always observed due to thermodynamics. The plot of hydricity versus pK_a indicates that the minimum pK_a of added acid to exclusively produce formate (and not H₂) is ~24.⁵⁵ Given that the pK_as of the hydrides are ~ 26-28, for selective formate production, a carbamic acid with a pK_a between 24 and 26-28 would need to be employed. The analysis assumes that the apparent hydricity of the hydrides is 44 kcal·mol⁻¹; the range is likely smaller because the apparent hydricity and pK_a combination must remain above the pink line. Acids with pK_as larger than those of the hydrides are not sufficient to generate the hydrides. Hence, addition of PhOH or TEOA (pK_a = 34.9)²⁰ instead results in exclusive reduction of CO₂ to CO. Figure 3 clearly shows the boundaries for generation of CO, formate, FA, and H₂.

Figure 3.

Plot of hydricity versus pK_a in MeCN. Solid lines represent boundaries for the speciation (boxed). Location of (N^N)Mn(CO)₃H are shown, with the downward arrow indicating how the hydricity is modulated by formate binding. Formate is only obtained in the gray triangle. Product distributions are shown along the x-axis as a function of pK_a. For FA, formate, and H₂, the hydricity or apparent hydricity must be sufficient enough to obtain the products.

CONCLUSIONS

A thorough thermochemical and mechanistic study that describes the catalytic reactivity of [(NAN)Mn(CO)₃]⁻ toward CO₂ in the presence of morpholine was presented. It was found that inclusion of morpholine to the catalytic system changes the product selectivity from CO to H₂ and FA. The change is due to increasing the acidity of the solution, allowing an intermediate hydride to form. This conclusion was corroborated by taking into account the pK_as of (N^N)Mn(CO)₃H and reactivity studies of the hydrides. Despite numerous reports on catalysts of the type [(N^N)Mn(CO)₃]⁻, this is the first study to measure the thermodynamic parameters, which allows for additional insight to be gleaned. Of significance is that both (NAN)Mn(CO)₃H are not sufficiently hydridic to reduce CO₂ despite the production of FA. This finding underscores the importance of formate binding to the catalyst, which serves to increase the apparent hydricity of the catalyst.

It was found that H₂ is favored over FA with [(bpy)Mn(CO)₃]⁻, which is attributed to fast kinetics associated with homolytic H₂ production. When only heterolytic H₂ production is feasible, FA is favored over H₂, as exemplified with [(mesbpy)Mn(CO)₃]⁻. The different reactivity is despite (mesbpy)Mn(CO)₃H being more hydridic than (bpy)Mn(CO)₃H. This emphasizes how different mechanisms can have different kinetics: the importance of tuning the catalyst for modulating the product selectivity.

A mechanistic scheme that is consistent with all observations is presented. In this system, the added amine changes the product selectivity and hence highlights a new role for amines in combined CCR systems. The hydricity of the Mn–H species is not sufficient to reduce CO₂ to formate which underscores the importance of instead considering the apparent hydricity, which takes into account the binding of the product to the metal. This finding is broadly applicable to hydrogenation and reduction reactions.