&Homogeneous Catalysis | Reviews Showcase |

Making a Splash in Homogeneous CO2 Hydrogenation:Elucidating the Impact of Solvent on Catalytic Mechanisms

Eric S. Wiedner* and John C. Linehan[a]

Chem. Eur. J. 2018, 24, 16964 – 16971 T 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16964

MinireviewDOI: 10.1002/chem.201801759

Abstract: Molecular catalysts for hydrogenation of CO2 are

widely studied as a means of chemical hydrogen storage.Catalysts are traditionally designed from the perspective of

controlling the ligands bound to the metal. In recent years,studies have shown that the solvent can also play a key rolein the mechanism of CO2 hydrogenation. A prominent exam-ple is the impact of the solvent on the thermodynamic hy-

dride donor ability, or hydricity, of metal hydride complexes

relative to the hydride acceptor ability of CO2. In some cases,simply changing from an organic solvent to water can re-

verse the direction of hydride transfer between a metal hy-

dride and CO2. Additionally, the solvent can impact catalysisby converting CO2 into carbonate species, as well as activate

intermediate products for hydrogenation to more reducedproducts. By understanding the substrate and product speci-ation, as well as the reactivity of the catalyst towards thesubstrate, the solvent can be used as a central design com-

ponent for the rational development of new catalytic sys-tems.

1. Introduction

In recent years, CO2 has been intensely studied as a potentialenergy vector for storage of “green” hydrogen that is obtained

from renewable energy sources, such as wind and solar.Formic acid, one of the simplest products of CO2 reduction,

has greater energy density than pressurized H2 storage tanks.[1]

As is typical for inert molecules, reaction of CO2 with H2 re-

quires a catalyst in order to occur at an appreciable rate. In in-

dustrial settings, CO2 can be hydrogenated to MeOH using het-erogeneous catalysts. However, the industrial processes for

CO2 conversion require large centralized facilities that operatewith high temperatures and pressures, making them unamena-

ble to small-scale decentralized conversion of CO2 using geo-graphically dispersed renewable energy sources.[2]

Molecular transition metal catalysts offer several advantages

over heterogeneous catalysts. First, molecular catalysts canoften hydrogenate CO2 at mild temperatures and pressures

that are compatible with using CO2 as an energy vector forchemical storage of renewable energy. Second, the per-

formance of molecular catalysts can be rationally controlledthrough ligand modification.[3] A prominent catalyst design

strategy is to promote metal-ligand cooperativity, where the

ligand is actively involved in making and breaking bondsduring catalysis. For example, functional groups in the second-coordination sphere of the ligand can assist in deprotonationof metal dihydrogen complexes[4] and promote hydride trans-fer to CO2.[5]

In addition to the catalyst structure, the solvent can also

play a critical role in determining catalytic reactivity. Solvent-dependent catalysis was observed in one of the earliest reportson CO2 hydrogenation by molecular catalysts, where small

amounts of water as a co-solvent were found to accelerate the

rate of formic acid production.[6] Since this early study, CO2 hy-

drogenation has been performed in a diverse array of solventsand conditions. Often the selection of one solvent over anoth-

er results from either empirical optimization or specific techno-logical considerations, such as: (1) molecular catalysts are typi-

cally more soluble in organic solvents than in water, (2) use ofsupercritical CO2 as the solvent greatly facilitates product sepa-

ration and increases the solubility of H2, and (3) water is com-

monly viewed as the ultimate green solvent.The solvent also plays a fundamental role in determining

the free energy profile for reaction of the catalyst with CO2.Recent years have seen an increase in the mechanistic under-

standing of how the solvent can influence the mechanism andenergetics of CO2 reduction. This Minireview focuses specifical-

ly on recent advances in addressing fundamental mechanistic

questions regarding the role of solvent on CO2 reduction.Many of the details surrounding the various interactions be-

tween catalyst and solvent have yet to be resolved, and it islikely that additional roles will be discovered in the future. Ac-

cordingly, the goal of this Minireview is not to provide conclu-sive answers to specific mechanistic issues. Rather, by identify-

ing the current state of knowledge, we seek to provide a basis

for future mechanistic studies and promote the use of solventas an integral, and quantifiable, component of rational catalyst

design.Hydride donor ability, or hydricity, is a thermodynamic pa-

rameter that is central to many of the studies described in thisMinireview. Hydricity is defined as a heterolytic bond cleavage

resulting in the formation of free H@ , with a correspondingfree energy value represented by DG8H- [Equation (1)] . The re-verse reaction, attack of free H@ on an acceptor molecule, cor-

responds to the hydride acceptor ability of a molecule [Equa-tion (2)] . Hydride acceptor ability is often notated as -DG8H-,

that is, the negative of the hydricity. However, in some sectionsof this manuscript the hydride acceptor ability will be referred

to as DG8H-,acc for clarity of discussion. Free H@ is a very high

energy species and is not a likely intermediate during hydridetransfer from one compound to another. However, similar to

H+ and HC transfers, knowledge of the hydricity of two com-pounds allows one to predict the direction and magnitude of

the hydride transfer equilibrium. Miller and Appel reported acomprehensive review on the thermodynamic hydricity of tran-

[a] Dr. E. S. Wiedner, Dr. J. C. LinehanCatalysis Science GroupPacific Northwest National LaboratoryP.O. Box 999, Richland, WA 99352 (USA)E-mail : eric.wiedner@pnnl.gov

The ORCID identification number(s) for the author(s) of this article can befound under :https ://doi.org/10.1002/chem.201801759.

Selected by the Editorial Office for our Showcase of outstanding Review-type articles (www.chemeurj.org/showcase).

Chem. Eur. J. 2018, 24, 16964 – 16971 www.chemeurj.org T 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16965

Minireview

sition metal hydrides,[7] and Kubiak recently reported an over-view of the role of hydricity in CO2 reduction.[8]

R-HÐ Rþ þ H@ DG2H@ ð1Þ

Rþ þ H@ Ð R-H DG2H@,acc ð2Þ

2. Hydride Transfer to CO2

One of the most significant developments in CO2 hydrogena-

tion in recent years has been the recognition that the thermo-dynamics for hydride transfer from a metal hydride complex to

CO2 are remarkably more favorable upon moving from organicsolvent into water. This solvent effect is readily quantified by

considering the hydricity of the metal hydride and formate.[7]

As shown in Scheme 1, hydride transfer to CO2 will be thermo-dynamically favorable if the metal hydride is more hydridic

(lower DG8H@) than formate. Creutz and co-workers were thefirst to demonstrate this effect for [HRu(C6Me6)(bpy)]+ : hydride

transfer from this complex to CO2 is uphill by 10 kcal mol@1 inacetonitrile,[9] but downhill by @3.9 kcal mol@1 in water.[10] Since

the pioneering work of Creutz, other metal hydride systemshave been discovered where the simple act of moving from an

organic solvent to water causes hydride transfer to CO2 to

become thermodynamically possible. To date, this solventeffect has been demonstrated for [HFe4(N)(CO)]@ ,[11] [HNi(di-

phosphine)2]+ ,[12] and [(H)2Co(dmpe)2]+ ,[13] as shown inFigure 1.

Several notable reports underscore the impact of solvent-de-pendent hydricity on the design of catalysts for CO2 reduction.

Berben reported solvent-dependent product selectivity forelectrocatalytic reduction of CO2 using the iron cluster

[Fe4(N)(CO)12]@ : high selectivity for formate (96 %) was observedin water,[11] while H2 was the major product in acetonitrile.[14]

Mechanistic studies identified a parallel between the product

selectivity and hydricity of [HFe4(N)(CO)12]@ , a key catalytic in-termediate.[11, 15] Hydride transfer from [HFe4(N)(CO)12]@ to CO2

is exergonic (@9 kcal mol@1) in water, but endergonic (+ 5 kcalmol@1) in acetonitrile (Figure 1). Wiedner and co-workers re-

cently reported that [(H)2Co(dmpe)2]+ is an active catalyst foraqueous hydrogenation of CO2 to formate, and operates by a

different mechanism in water than in organic solvent.[13] In

Eric Wiedner received his B.S. in chemistryfrom the Missouri University of Science andTechnology (2004), and his Ph.D. from theUniversity of Michigan (2009). He next movedto Pacific Northwest National Laboratory(PNNL) for a postdoctoral fellowship studyingunder Dan DuBois (2009–2012), then transi-tioned to a full-time research scientist in theCataysis Science group at PNNL. Eric’s scientif-ic interest is in the rational design of molecu-lar catalysts, guided by mechanistic and ther-modynamic information of the limiting cata-lytic intermediates.

John Linehan received his B.S. in chemistryfrom the San Francisco State University(1981), and his Ph.D. in chemistry from theUniversity of California—Davis (1986). Heheld a postdoctoral fellowship at Argone Na-tional Laboratory prior to moving to PacificNorthwest National Laboratory in 1987.John’s scientific interest is in understandinghow catalysts work, utilizing operando NMR,IR, and XAFS spectroscopies.

Scheme 1. Thermochemical cycle for determining the free energy for hy-dride transfer from a metal hydride complex to CO2.

Figure 1. Solvent dependent hydricity (DG8H@) for select metal hydride complexes and formate.

Chem. Eur. J. 2018, 24, 16964 – 16971 www.chemeurj.org T 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16966

Minireview

water, the hydricity of [(H)2Co(dmpe)2]+ is sufficient to reactwith CO2. In organic solvents, however, hydride transfer from

[(H)2Co(dmpe)2]+ to CO2 is endergonic, and this intermediatemust be deprotonated with a very strong organic base to gen-

erate HCo(dmpe)2, a strong hydride donor.[16] For both[HFe4(N)(CO)12]@ and [(H)2Co(dmpe)2]+ , the favorable impact ofwater on hydride transfer to CO2 was critical for catalytic for-mate production.

Given the key role of solvent-dependent hydricity for some

catalysts, it is worthwhile to consider the origin of thissolvent effect. Acetonitrile and water differ in both dielectric

constant (e) and in hydrogen bonding characteristics, both ofwhich could play an important role in promoting conversionof CO2 to formate. Insight into this question was provided byYang, who measured the hydricity of [HNi(DHMPE)2]+ and

[HNi(TMEPE)2]+ in CH3CN (e= 38), DMSO (e= 47), and H2O (e=

80), as shown in Figure 2.[12c, 17] Surprisingly, the relevant sol-

vent parameter appears to vary with the complex. In the case

of [HNi(TMEPE)2]+ , the free energy for hydride transfer is clearlycorrelated with the solvent dielectric. In contrast, hydride trans-fer from [HNi(DHMPE)]+ to CO2 was equally unfavorable inCH3CN and DMSO (+ 13.5 kcal mol@1), but less unfavorable inH2O (+ 5.9 kcal mol@1). Thus, either the solvent dielectric or hy-drogen bonding effects can dominate the thermodynamics of

hydride transfer to CO2, depending on the metal hydride com-plex.

Further insight into the origin of the solvent-dependence on

the thermodynamics of hydride transfer can be gainedthrough consideration of kinetic effects, which are much easier

to investigate in a range of solvents with varying properties.Meyer investigated solvent-dependent kinetics for reaction of

Re(bpy)(CO)3H and CO2 to give Re(bpy)(CO)3(HCO2), a formate

complex.[18] In this study, the second-order rate constantshowed a linear correlation with the solvent dielectric constant

(Figure 3 a). Ishitani performed a similar study using [Ru(ter-py)(bpy)H]+ , but found only a poor correlation between the

solvent dielectric constant and the second-order rate constantfor hydride transfer to CO2.[19] For [Ru(terpy)(bpy)H]+ , the rate

constant was found to correlate with the solvent acceptornumber (AN), which is a measure of the Lewis acidity of the

solvent (in other words, its ability to donate a hydrogen

bond).[20] This correlation with AN can be extended to includethe rate constant for reaction of [Ru(terpy)(bpy)H]+ and CO2 in

water, as measured by Creutz (Figure 3 b).[10a]

The discrepancy between the findings for Re(bpy)(CO)3H

and [Ru(terpy)(bpy)H]+ is likely related to the range of solventsemployed. Only solvents with low AN were used in the studyof Re(bpy)(CO)3H,[18] while solvents of higher AN (e.g. H2O,

MeOH) were used in the studies of [Ru(terpy)(bpy)H]+ .[10a, 19] Di-electric and AN are strongly correlated for solvents without OHgroups (or other heteroatom-hydrogen bonds). For weak hy-drogen-bond donor solvents, it is difficult to determine wheth-

er dielectric or AN has a greater impact on the kinetics of hy-dride transfer to CO2. However, hydrogen bonding appears to

override the dielectric in high AN solvents such as MeOH.A growing need in CO2 hydrogenation is to better under-

stand the correlation between thermodynamics and kinetics of

hydride transfer to CO2. Recent computational[21] and experi-mental[22] studies have demonstrated the presence of a linear

free energy relationship (LFER) between the driving force andactivation energy for hydride transfer to CO2. The focus of

these studies was on tuning the hydricity of the metal hydride

through variation of the electronic and steric properties of thesupporting ligands. Similar LFERs likely exist for individual

metal hydrides as the solvent is varied. However, such a studyhas not been reported to date, likely due to in part to the chal-

lenge of measuring both hydricity values and kinetics for hy-dride transfer to CO2 in multiple solvents. To the best of our

Figure 2. Free energy for hydride transfer to CO2 versus the solvent dielectricconstant for [HNi(DHMPE)2]+ and [HNi(TMEPE)2]+ . Dashed lines indicate theleast-squares fit, while the solid line for [HNi(DHMPE)2]+ is a guide to theeye to illustrate deviations from the least-squares fit.

Figure 3. Plots of the second order rate constant for H@ transfer to CO2 (inlog form) versus (a) the solvent dielectric constant and (b) the solvent ac-ceptor number. Dashed lines indicate the least-squares fit.

Chem. Eur. J. 2018, 24, 16964 – 16971 www.chemeurj.org T 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16967

Minireview

knowledge, the studies on the thermodynamics and kineticsfor reaction of [Ru(terpy)(bpy)H]+ with CO2 in acetonitrile[9]

and water[10a] come closest to establishing a solvent-basedLFER. Additional studies of this system in other solvents are

needed to clearly establish a solvent-based LFER for this com-plex.

In addition to hydride transfer, deprotonation of a metal di-hydride (or dihydrogen) complex is also a key step for CO2 hy-drogenation and can be substantially influenced by the sol-

vent. This is particularly true for hydrogenation of bicarbonate(HCO3

@), which serves as both the substrate and the base. In acomputational study of bicarbonate hydrogenation in alcoholsolvents, Ahlquist predicted that the activation energy for de-

protonation of the catalyst was inversely correlated to the freeenergy for solvation of HCO3

@ .[23] Thus, catalysis is impeded by

stabilizing the base through hydrogen bonding interactions, in

direct contrast to the trends observed for generating formatevia hydride transfer.

3. Role of Carbonates

Moving from organic solvent to water affects the speciation of

CO2 in solution due to the formation of inorganic carbonates(H2CO3, HCO3

@ , CO32@). When catalytic CO2 hydrogenation is

performed in aqueous basic conditions, the concentration of

HCO3@ is much greater than the concentration of CO2. This has

led some authors to propose that HCO3@ is the real substrate

under catalytic conditions.[24] In support of this proposal, sever-al studies have demonstrated that HCO3

@ can be hydrogenated

to formate without any added CO2.[25] Identification of the truesubstrate, CO2 or HCO3

@ , is complicated by the interconversion

of these two species in solution [Equation (3)] . The first order

rate constant for conversion of HCO3@ to CO2 and HO@ is

0.7 h@1 at room temperature,[26] corresponding to DG* =

22.5 kcal mol@1. For comparison, one the fastest reported cata-lysts for hydrogenation of NaHCO3, a ruthenium pincer com-plex, has TOF = 1700 h@1 at 80 8C and 40 bar H2.[25f] The barrierfor this catalytic TOF corresponds to DG*&21 kcal mol@1,

which is similar to the barrier for HCO3@ decomposition into

CO2. Thus, it is feasible that the generation of CO2 from HCO3@

may be a major contributor to the rate-determining process of

NaHCO3 hydrogenation.

CO2 þ OH@ Ð HCO3@ ð3Þ

One way to probe the identity of the true substrate is to

measure the kinetics of catalysis with and without added CO2.However, few studies report catalysis under conditions that

differ by only the CO2 pressure. In a batch reactor study ofHCO3

@ hydrogenation with the [RuCl2(C6H6)2]2/ dppm system,

more catalytic turnovers were obtained in the presence of

35 bar CO2 (2500 turnovers) than with no added CO2 (2100turnovers).[25c] In a recent study using a water-soluble Ni(di-

phosphine)22 + complex, catalysis was 5 V faster with 34 atm of

1:1 H2 :CO2 than with only H2 and HCO3@ .[12a] While limited in

scope, these studies suggest that hydrogenation of CO2 ismore facile than direct hydrogenation of HCO3

@ .

From a thermodynamic standpoint, HCO3@ is a poorer hy-

dride acceptor than CO2. The initial product of H@ transfer to

HCO3@ would be H2CO3

2@, a doubly-deprotonated form of or-thoformic acid (Scheme 2). This unstable intermediate would

rapidly dissociate OH@ to give the formate anion. There is in-sufficient thermochemical data to determine the true hydride

acceptor ability of HCO3@ due to the instability of H2CO3

2@.However, the effective hydride acceptor ability (DGeff

H-,acc) of

HCO3@ , corresponding to C@H bond formation and C@O bond

cleavage, can be determined using the free energy for conver-sion of HCO3

@ to CO2 (8.4 kcal mol@1 at pH 14 and 1 atm CO2)[26]

and the aqueous hydride acceptor ability of CO2 (@24 kcalmol@1),[10b] as illustrated in Scheme 2. This effective hydride ac-

ceptor ability is pH-dependent due to the presence of OH@ inthe chemical equilibrium, resulting in DGeff

H-,acc =@34.8 +

(1.364 V pH) kcal mol@1. At PCO2 = 1 atm, HCO3@ is the better hy-

dride acceptor when pH<7.8, while CO2 is the better hydrideacceptor when pH>7.8.

While this section has so far focused on inorganic carbonate(HCO3

@), similar mechanistic questions are relevant for hydro-

genation of CO2 that has been captured as an organic carbon-ate (ROCO2

@) or carbamate (R2NCO2@).[27] In these studies, CO2

capture is performed at ambient temperature, while the hydro-

genation reaction often requires temperatures>100 8C to drivecatalysis at an appreciable rate. Mechanistic studies areneeded to determine whether the substrate is CO2 that is re-leased from the organic carbonate (or carbamate) at the high

temperatures required for hydrogenation.

4. Products of CO2 Hydrogenation

Direct hydrogenation of CO2 to formic acid (HCO2H) in the ab-

sence of a strong base is challenging due to the unfavorablethermodynamics of the reaction. In water, formic acid is

+ 5.0 kcal mol@1 uphill from CO2 and H2 at standard state (25 8Cand 1 atm). Reaction conditions must be selected to drive the

unfavorable equilibrium to generate significant amounts of

formic acid, such as performing catalysis at elevated pressures.Laurenczy and co-workers reported an intriguing solvent-de-

pendence on formic acid production using RuCl2(PTA)4 as thecatalyst.[28] With this catalyst, mixtures of DMSO and water

gave much higher concentrations of formic acid than othersolvents (Figure 4). Several lines of evidence suggested that

Scheme 2. Thermochemical cycle for determining the effective hydride ac-ceptor ability of HCO3

@ .

Chem. Eur. J. 2018, 24, 16964 – 16971 www.chemeurj.org T 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16968

Minireview

the change in formic acid yield resulted from a change in theequilibrium concentration of formic acid in DMSO. First, the

RuCl2(PTA)4 catalyst could be recycled multiple times without aloss in activity, indicating the yields of formic acid were not in-

fluenced by catalyst decomposition. Second, RuCl2(PTA)4 alsocatalyzed the reverse reaction, dehydrogenation of formic acid.

To further investigate the solvent-dependent catalysis by

RuCl2(PTA)4, Laurenczy used calorimetry to measure the heat ofmixing of formic acid in a wide range of solvents [Equa-

tion (4)] .[29] Importantly, formic acid dissolution became moreexothermic as the solvent basicity was increased (Figure 5).

This finding has a direct parallel with solvent donor number,

which corresponds to its Lewis basicity.[30] This study supportsthe conclusion that the increased yield of formic acid obtained

in DMSO results from improvement in the thermodynamics forthe net reaction. The entropy for hydrogenation of CO2 to

formic acid is also likely to vary in different solvents, andfuture measurements of this thermodynamic parameter would

be valuable.

HCO2HðliqÞ ! HCO2HðsolvÞ DH2FA,liq!solv ð4Þ

Sans and Dupont recently demonstrated that increased

yields of formic acid could be obtained by hydrogenating CO2

in an imidazolium ionic liquid using Ru3(CO)12 as a precata-

lyst.[31] The highest yields of formic acid (>1 M) were achievedusing either acetate or formate as the anion for the ionic

liquid. Carboxylic acids, such as formic acid and acetic acid, areknown to homoassociate with their conjugate base in many

organic solvents [Equation (5)] .[32] The improved yields offormic acid are likely driven by hydrogen bonding with the

ionic liquid anion.

RCO2H þ R0CO2@ ! ðRCO2-H-O2CR0Þ@ ð5Þ

Hydrogenation of CO2 to methanol is an exergonic reaction(DG8=@5.0 kcal mol@1 in H2O), yet is very difficult to achieve

using molecular catalysts. Indirect methods for obtainingMeOH from CO2 include tandem catalysis[33] and combined

capture and conversion.[27a–f] Direct hydrogenation of CO2 to

MeOH has been achieved with Ru(triphos)[34] and Co(triphos)[35]

at high temperatures and pressures. Both of these approaches,

indirect and direct conversion, require high temperatures(>130 8C) to obtain MeOH at a reasonable rate. One exception

to the requirement of high temperature comes from Himedaand Laurenczy, who showed that direct hydrogenation of CO2

to MeOH could be achieved at ambient temperature using

[Cp*Ir(dhbp)(H2O)]2 + (dhbp = 4,4’-dihydroxy-2,2’-bipyridine) asthe catalyst in 2 M aqueous H2SO4 (pH&0).[36] The acidic condi-

tions were critical for MeOH production, as the same Ir catalystonly produces formic acid at pH 3.[37]

In a follow-up mechanistic study, Himeda and Laurenzy pro-posed that protonation of formic acid at low pH activates it toaccept a hydride from the catalyst.[38] The thermodynamic fea-

sibility of this hypothesis can be checked through determina-tion of the hydride acceptor abilities of both formic acid,

HCO2H, and protonated formic acid, HC(OH)2+ (Figure 6).

These values have not been previously reported, but can bereadily calculated from literature data. The first step is to deter-

Figure 4. Equilibrium concentration of formic acid obtained in different sol-vents by hydrogenation of CO2 with RuCl2(PTA)4. Conditions: t = 50 8C, P(to-tal) = 100 bar, 1:1 H2 :CO2. BMIM = 1-butyl-3-methylimidazolium tetrafluorobo-rate (or p-toluenesulfonate).

Figure 5. Heat of mixing of formic acid (DH8FA,liq!solv) in different solvents.

Figure 6. (a) Thermochemical cycle for determining the free energy for hy-drogenation of formic acid to methanediol in water. (b) ThermochemicalScheme for conversion of formic acid to methanediol by sequential transferof H@ and H+ .

Chem. Eur. J. 2018, 24, 16964 – 16971 www.chemeurj.org T 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16969

Minireview

mine the free energy for hydrogenation of HCO2H to methane-diol, H2C(OH)2, from a thermodynamic cycle that includes the

standard potential for reduction of HCO2H to formaldehyde,H2CO,[39] the equilibrium constant for hydration of H2CO to

give H2C(OH)2,[40] and the standard potential of the H+/H2

couple, that is, the SHE electrode (Figure 6 a). This cycle affords

DG8=@4.2 kcal mol@1 for hydrogenation of HCO2H toH2C(OH)2. This value can be combined with literature pKa

values, along with the constant for heterolytic cleavage ofH2,[10b] to determine the desired hydride acceptor abilities (Fig-ure 6 b). For example, the pKa of H2C(OH)2 in water is reportedto be 13.3,[41] which leads to a hydride acceptor ability of@20 kcal mol@1 for formic acid. Similarly, protonated acetic

acid, CH3C(OH)2+ , has been reported to have a pKa of @6 in

water.[42] Using this pKa as a proxy for that of HC(OH)2+ affords

a hydride acceptor ability of @47 kcal mol@1 for HC(OH)2+ .

Based on this thermodynamic analysis, hydride transfer from[Cp*Ir(dhbp)(H2O)]2 + (DG8H@&31 kcal mol@1)[43] to HCO2H is en-

dergonic (+ 11 kcal mol@1), while hydride transfer to HC(OH)2+

is exergonic (@16 kcal mol@1). These data suggest HC(OH)2+ is

the actual hydride acceptor in acidic conditions, in support ofthe mechanistic proposal from Himeda and Laurenzy. Yang

reached the opposite mechanistic conclusion in a recent com-

putational analysis, in which the barrier for hydride transfer toHCO2H was calculated to be 9 kcal mol@1 lower than for hy-

dride transfer to HC(OH)2+ .[44] Clearly further experimental and

computational studies are needed to resolve the mechanistic

consequences of performing catalysis in highly acidic condi-tions.

5. Summary and Outlook

This article describes both recent and emerging mechanisticstudies regarding the impact of solvent on the mechanisms for

hydrogenation of CO2 by molecular catalysts. Changing solventconditions can affect the thermodynamics and kinetics for hy-dride transfer from the catalyst to CO2. Carbonate species,both inorganic and organic, affects the speciation of CO2 and

allows the possibility of multiple hydride transfer pathways.

Additionally, the solvent can be used lower the thermodynam-ic barrier for hydrogenation of CO2 to formic acid, as well as

enable the formation of methanol by activating formic acid toaccept more hydride equivalents from the catalyst. These sol-

vent effects largely center around controlling hydride transferfrom the catalyst. However, hydrogen addition to the catalyst

and deprotonation of the intermediate metal hydrides are also

critical mechanistic steps in the hydrogenation of CO2, and thesolvent can potentially play a direct role in these steps as well.

Further mechanistic understanding is needed to be able to usethe solvent as a key parameter for rational control over catalyst

performance.

Acknowledgements

This research was supported by the U.S. Department of Energy,Office of Science, Office of Basic Energy Sciences, Division of

Chemical Sciences, Geosciences & Biosciences. The authorsthank Dr. Aaron Appel for helpful discussions, and Rose Perryfor creating the frontispiece artwork. Pacific Northwest Nation-al Laboratory is operated by Battelle for the U.S. Departmentof Energy.

Conflict of interest

The authors declare no conflict of interest.

Keywords: carbon dixode fixation · homogeneous catalysis ·hydrides · hydrogenation · solvent effects

[1] J. Eppinger, K.-W. Huang, ACS Energy Lett. 2017, 2, 188 – 195.[2] T. R. Cook, D. K. Dogutan, S. Y. Reece, Y. Surendranath, T. S. Teets, D. G.

Nocera, Chem. Rev. 2010, 110, 6474 – 6502.[3] a) W.-H. Wang, Y. Himeda, J. T. Muckerman, G. F. Manbeck, E. Fujita,

Chem. Rev. 2015, 115, 12936 – 12973; b) K. Sordakis, C. Tang, L. K. Vogt,H. Junge, P. J. Dyson, M. Beller, G. Laurenczy, Chem. Rev. 2018, 118, 372 –433.

[4] a) R. Tanaka, M. Yamashita, L. W. Chung, K. Morokuma, K. Nozaki, Orga-nometallics 2011, 30, 6742 – 6750; b) J. F. Hull, Y. Himeda, W. H. Wang, B.Hashiguchi, R. Periana, D. J. Szalda, J. T. Muckerman, E. Fujita, Nat.Chem. 2012, 4, 383 – 388; c) N. Onishi, S. Xu, Y. Manaka, Y. Suna, W.-H.Wang, J. T. Muckerman, E. Fujita, Y. Himeda, Inorg. Chem. 2015, 54,5114 – 5123; d) W.-H. Wang, J. T. Muckerman, E. Fujita, Y. Himeda, ACSCatal. 2013, 3, 856 – 860; e) G. A. Filonenko, D. Smykowski, B. M. Szyja,G. Li, J. Szczygieł, E. J. M. Hensen, E. A. Pidko, ACS Catal. 2015, 5, 1145 –1154; f) H. Ge, Y. Jing, X. Yang, Inorg. Chem. 2016, 55, 12179 – 12184.

[5] a) S. Roy, B. Sharma, J. P8caut, P. Simon, M. Fontecave, P. D. Tran, E.Derat, V. Artero, J. Am. Chem. Soc. 2017, 139, 3685 – 3696; b) W. H. Bern-skoetter, N. Hazari, Eur. J. Inorg. Chem. 2013, 4032 – 4041; c) T. J. Schmei-er, G. E. Dobereiner, R. H. Crabtree, N. Hazari, J. Am. Chem. Soc. 2011,133, 9274 – 9277; d) S. Ramakrishnan, K. M. Waldie, I. Warnke, A. G. DeCrisci, V. S. Batista, R. M. Waymouth, C. E. D. Chidsey, Inorg. Chem. 2016,55, 1623 – 1632.

[6] Y. Inoue, H. Izumida, Y. Sasaki, H. Hashimoto, Chem. Lett. 1976, 5, 863 –864.

[7] E. S. Wiedner, M. B. Chambers, C. L. Pitman, R. M. Bullock, A. J. M. Miller,A. M. Appel, Chem. Rev. 2016, 116, 8655 – 8692.

[8] K. M. Waldie, A. L. Ostericher, M. H. Reineke, A. F. Sasayama, C. P. Kubiak,ACS Catal. 2018, 8, 1313 – 1324.

[9] Y. Matsubara, E. Fujita, M. D. Doherty, J. T. Muckerman, C. Creutz, J. Am.Chem. Soc. 2012, 134, 15743 – 15757.

[10] a) C. Creutz, M. H. Chou, J. Am. Chem. Soc. 2009, 131, 2794 – 2795;b) S. J. Connelly, E. S. Wiedner, A. M. Appel, Dalton Trans. 2015, 44,5933 – 5938.

[11] A. Taheri, E. J. Thompson, J. C. Fettinger, L. A. Berben, ACS Catal. 2015,5, 7140 – 7151.

[12] a) S. A. Burgess, A. J. Kendall, D. R. Tyler, J. C. Linehan, A. M. Appel, ACSCatal. 2017, 7, 3089 – 3096; b) S. J. Connelly Robinson, C. M. Zall, D. L.Miller, J. C. Linehan, A. M. Appel, Dalton Trans. 2016, 45, 10017 – 10023;c) B. M. Ceballos, C. Tsay, J. Y. Yang, Chem. Commun. 2017, 53, 7405 –7408.

[13] S. A. Burgess, A. M. Appel, J. C. Linehan, E. S. Wiedner, Angew. Chem. Int.Ed. 2017, 56, 15002 – 15005; Angew. Chem. 2017, 129, 15198 – 15201.

[14] M. D. Rail, L. A. Berben, J. Am. Chem. Soc. 2011, 133, 18577 – 18579.[15] A. Taheri, N. D. Loewen, D. B. Cluff, L. A. Berben, Organometallics 2018,

37, 1087 – 1091.[16] a) M. S. Jeletic, M. L. Helm, E. B. Hulley, M. T. Mock, A. M. Appel, J. C.

Linehan, ACS Catal. 2014, 4, 3755 – 3762; b) M. S. Jeletic, M. T. Mock,A. M. Appel, J. C. Linehan, J. Am. Chem. Soc. 2013, 135, 11533 – 11536.

[17] C. Tsay, B. N. Livesay, S. Ruelas, J. Y. Yang, J. Am. Chem. Soc. 2015, 137,14114 – 14121.

[18] B. P. Sullivan, T. J. Meyer, Organometallics 1986, 5, 1500 – 1502.[19] H. Konno, A. Kobayashi, K. Sakamoto, F. Fagalde, N. E. Katz, H. Saitoh, O.

Ishitani, Inorg. Chim. Acta 2000, 299, 155 – 163.

Chem. Eur. J. 2018, 24, 16964 – 16971 www.chemeurj.org T 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16970

Minireview

[20] U. Mayer, V. Gutmann, W. Gerger, Monatsh. Chem. 1975, 106, 1235 –1257.

[21] a) B. Mondal, F. Neese, S. Ye, Inorg. Chem. 2016, 55, 5438 – 5444; b) T.Ono, S. Qu, C. Gimbert-SuriÇach, M. A. Johnson, D. J. Marell, J. Benet-Buchholz, C. J. Cramer, A. Llobet, ACS Catal. 2017, 7, 5932 – 5940.

[22] M. S. Jeletic, E. Hulley, M. L. Helm, M. T. Mock, A. M. Appel, E. S. Wiedner,J. C. Linehan, ACS Catal. 2017, 7, 6008 – 6017.

[23] R. Marcos, L. Xue, R. S#nchez-de-Armas, M. S. G. Ahlquist, ACS Catal.2016, 6, 2923 – 2929.

[24] a) T. J. Geldbach, G. Laurenczy, R. Scopelliti, P. J. Dyson, Organometallics2006, 25, 733 – 742; b) F. Joj, L. Nadasdi, J. Elek, G. Laurenczy, L. Nadas-di, Chem. Commun. 1999, 971 – 972.

[25] a) J. Elek, L. N#dasdi, G. Papp, G. Laurenczy, F. Joj, Appl. Catal. A 2003,255, 59 – 67; b) G. Laurenczy, F. Joj, L. N#dasdi, Inorg. Chem. 2000, 39,5083 – 5088; c) C. Federsel, R. Jackstell, A. Boddien, G. Laurenczy, M.Beller, ChemSusChem 2010, 3, 1048 – 1050; d) C. Federsel, A. Boddien, R.Jackstell, R. Jennerjahn, P. J. Dyson, R. Scopelliti, G. Laurenczy, M. Beller,Angew. Chem. Int. Ed. 2010, 49, 9777 – 9780; Angew. Chem. 2010, 122,9971 – 9974; e) C. Federsel, C. Ziebart, R. Jackstell, W. Baumann, M.Beller, Chem. Eur. J. 2012, 18, 72 – 75; f) J. Kothandaraman, M. Czaun, A.Goeppert, R. Haiges, J.-P. Jones, R. B. May, G. K. S. Prakash, G. A. Olah,ChemSusChem 2015, 8, 1442 – 1451; g) Z. Dai, Q. Luo, H. Cong, J. Zhang,T. Peng, New J. Chem. 2017, 41, 3055 – 3060.

[26] F. A. Cotton, G. Wilkinson, P. L. Gaus, Basic Inorganic Chemistry, 3rd ed. ,Wiley, Inc. , New York, 1995.

[27] a) E. Balaraman, C. Gunanathan, J. Zhang, L. J. W. Shimon, D. Milstein,Nat. Chem. 2011, 3, 609 – 614; b) J. R. Khusnutdinova, J. A. Garg, D. Mil-stein, ACS Catal. 2015, 5, 2416 – 2422; c) N. M. Rezayee, C. A. Huff, M. S.Sanford, J. Am. Chem. Soc. 2015, 137, 1028 – 1031; d) J. Kothandaraman,A. Goeppert, M. Czaun, G. A. Olah, G. K. Surya Prakash, Green Chem.2016, 18, 5831 – 5838; e) A. P. C. Ribeiro, L. M. D. R. S. Martins, A. J. L.Pombeiro, Green Chem. 2017, 19, 4811 – 4815; f) S. Kar, R. Sen, A. Goep-pert, G. K. S. Prakash, J. Am. Chem. Soc. 2018, 140, 1580 – 1583; g) M.Yadav, J. C. Linehan, A. J. Karkamkar, E. van der Eide, D. J. Heldebrant,Inorg. Chem. 2014, 53, 9849 – 9854; h) D. B. Lao, B. R. Galan, J. C. Line-han, D. J. Heldebrant, Green Chem. 2016, 18, 4871 – 4874.

[28] S. Moret, P. J. Dyson, G. Laurenczy, Nat. Commun. 2014, 5, 4017.[29] C. Fink, S. Katsyuba, G. Laurenczy, Phys. Chem. Chem. Phys. 2016, 18,

10764 – 10773.[30] V. Gutmann, Coord. Chem. Rev. 1976, 18, 225 – 255.[31] A. Weilhard, M. I. Qadir, V. Sans, J. Dupont, ACS Catal. 2018, 8, 1628 –

1634.[32] K. Izutsu, Acid-Base Disociation Constants in Dipolar Aprotic Solvents,

Blackwell Scientific Publications, Oxford, 1990.[33] C. A. Huff, M. S. Sanford, J. Am. Chem. Soc. 2011, 133, 18122 – 18125.[34] a) S. Wesselbaum, et al. , Chem. Sci. 2015, 6, 693 – 704; b) S. Wesselbaum,

T. vom Stein, J. Klankermayer, W. Leitner, Angew. Chem. Int. Ed. 2012, 51,7499 – 7502; Angew. Chem. 2012, 124, 7617 – 7620.

[35] J. Schneidewind, R. Adam, W. Baumann, R. Jackstell, M. Beller, Angew.Chem. Int. Ed. 2017, 56, 1890 – 1893; Angew. Chem. 2017, 129, 1916 –1919.

[36] K. Sordakis, A. Tsurusaki, M. Iguchi, H. Kawanami, Y. Himeda, G. Laurenc-zy, Chem. Eur. J. 2016, 22, 15605 – 15608.

[37] S. Ogo, R. Kabe, H. Hayashi, R. Harada, S. Fukuzumi, Dalton Trans. 2006,4657 – 4663.

[38] A. Tsurusaki, K. Murata, N. Onishi, K. Sordakis, G. Laurenczy, Y. Himeda,ACS Catal. 2017, 7, 1123 – 1131.

[39] S. G. Bratsch, J. Phys. Chem. Ref. Data 1989, 18, 1 – 21.[40] R. Bieber, G. Trempler, Helv. Chim. Acta 1947, 30, 1860 – 1865.[41] R. P. Bell, P. T. McTigue, J. Chem. Soc. 1960, 2983 – 2994.[42] A. R. Goldfarb, A. Mele, N. Gutstein, J. Am. Chem. Soc. 1955, 77, 6194 –

6196.[43] C. L. Pitman, K. R. Brereton, A. J. Miller, J. Am. Chem. Soc. 2016, 138,

2252 – 2260.[44] X. Yan, X. Yang, Organometallics 2018, 37, 1519 – 1525.

Manuscript received: April 9, 2018

Accepted manuscript online: June 6, 2018

Version of record online: July 27, 2018

Chem. Eur. J. 2018, 24, 16964 – 16971 www.chemeurj.org T 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16971

Minireview