Article pubs.acs.org/accounts
Reversible Hydrogenation of Carbon Dioxide to Formic Acid and Methanol: Lewis Acid Enhancement of Base Metal Catalysts Wesley H Bernskoetter*,† and Nilay Hazari*,‡ †
The Department of Chemistry, University of Missouri, Columbia, Missouri 65211, United States The Department of Chemistry, Yale University, P.O. Box 208107, New Haven, Connecticut 06520, United States
‡
CONSPECTUS: New and sustainable energy vectors are required as a consequence of the environmental issues associated with the continued use of fossil fuels. H2 is a potential clean energy source, but as a result of problems associated with its storage and transport as a gas, chemical H2 storage (CHS), which involves the dehydrogenation of small molecules, is an attractive alternative. In principle, formic acid (FA, 4.4 wt % H2) and methanol (MeOH, 12.6 wt % H2) can be obtained renewably and are excellent prospective liquid CHS materials. In addition, MeOH has considerable potential both as a direct replacement for gasoline and as a fuel cell input. The current commercial syntheses of FA and MeOH, however, use nonrenewable feedstocks and will not facilitate the use of these molecules for CHS. An appealing option for the sustainable synthesis of both FA and MeOH, which could be implemented on a large scale, is the direct metal catalyzed hydrogenation of CO2. Furthermore, given that CO2 is a readily available, nontoxic and inexpensive source of carbon, it is expected that there will be economic and environmental benefits from using CO2 as a feedstock. One strategy to facilitate both the dehydrogenation of FA and MeOH and the hydrogenation of CO2 and H2 to FA and MeOH is to utilize a homogeneous transition metal catalyst. In particular, the development of catalysts based on first row transition metals, which are cheaper, and more abundant than their precious metal counterparts, is desirable. In this Account, we describe recent advances in the development of iron and cobalt systems for the hydrogenation of CO2 to FA and MeOH and the dehydrogenation of FA and MeOH and provide a brief comparison between precious metal and base metal systems. We highlight the different ligands that have been used to stabilize first row transition metal catalysts and discuss the use of additives to promote catalytic activity. In particular, the Account focuses on the crucial role that alkali metal Lewis acid cocatalysts can play in promoting increased activity and catalyst stability for first row transition metal systems. We relate these effects to the nature of the elementary steps in the catalytic cycle and describe how the Lewis acids stabilize the crucial transition states. For all four transformations, we discuss in detail the currently proposed catalytic pathways, and throughout the Account we identify mechanistic similarities among catalysts for the four processes. The limitations of current catalytic systems are detailed, and suggestions are provided on the improvements that are likely required to develop catalysts that are more stable, active, and practical.
■
INTRODUCTION Due to environmental concerns and the decline in our petroleum reserves, there is enormous interest in the development of alternative sustainable energy vectors to replace fossil fuels.1 H2 is a potential clean energy source, which can be directly combusted or electrochemically oxidized within a proton-exchange membrane fuel cell.2 However, as a result of problems with the storage and transport of gaseous H2 and its low volumetric energy density, chemical H2 storage (CHS) based on the dehydrogenation of small molecules is an attractive alternative.2,3 Both formic acid (FA, 4.4 wt % H2) © 2017 American Chemical Society
and methanol (MeOH, 12.6 wt % H2) can potentially be obtained renewably and are therefore excellent possible liquid CHS materials. MeOH also has considerable value as a direct replacement for gasoline and as a fuel cell input.4 Balanced chemical equations for the dehydrogenation of FA and MeOH along with the corresponding free energies are shown in eqs 1 and 2, respectively. Received: January 19, 2017 Published: March 17, 2017 1049
DOI: 10.1021/acs.accounts.7b00039 Acc. Chem. Res. 2017, 50, 1049−1058
Article
Accounts of Chemical Research Table 1. Selected First Row Transition Metal Catalysts for FA Dehydrogenation
catalyst
T (°C)
solvent
Aab Ba Ca D E F (2a)
80 80 60 40 80 80
MeOH propylene carbonate propylene carbonate 1,4-dioxane propylene carbonate 1,4-dioxane
base
additive terpy PP3 P4
50 mol % NEt3d 100 mol % NEt3d 0.58 M LiBF4
a
TON
ref
126 92 417c 6061 100 000 10 000 983 600
10 12 13 14 15 16
b
Catalyst is formed in situ by mixing metal precursor and appropriate ligand. Xenon lamp used to irradiate reaction. cTON from a reaction under flow conditions with constant dosage of FA. dmol % with respect to FA.
HCOOH(l) ⇌ CO2(g) + H 2(g)
■
ΔG°298 = − 7.6 kcal mol−1
DEHYDROGENATION OF FORMIC ACID Several heterogeneous and homogeneous catalysts are able to dehydrogenate FA to a mixture of H2 and CO2 (eq 1), which is directly suitable as a feed for fuel cells.3 One of the advantages of homogeneous catalysts is that they do not produce large amounts of CO as a byproduct, which can poison fuel cells. Traditionally, the most productive homogeneous systems are based on expensive precious metals such as Ir, Ru, Au, Ag, and Pd.3 Additionally, many of these catalysts require an exogenous base or have a complicated pendant base attached to the ligand framework, in order to achieve high turnover numbers (TON). In 2010, Beller, Ludwig, and co-workers reported light-driven first row transition metal based catalysts for H2 production from FA (Table 1).10 The most productive system was generated in situ from a mixture containing Fe3(CO)12, 2,2′:6′,2″-terpyridine (terpy), and PPh3 and gave a TON of 126. The active catalyst is likely [FeH(CO)3(PPh)]− (A), with the terpy playing a role in reducing the rate of catalyst deactivation. Subsequently, using the tetradentate ligand PP3 (PP3 = P(CH2CH2PPh2)3), which was first shown to stabilize complexes for FA dehydrogenation in catalytic transfer hydrogenation,11 Beller et al. generated a productive thermal catalyst by mixing Fe(BF4)2·6H2O with PP3 in situ.12 Similar catalytic activity was observed using the well-defined complex [FeH(PP3)]+ (B), suggesting that this species may be a catalytic intermediate. Using either the in situ generated system or complex B, the best catalytic activity was observed when 1 extra equivalent of PP3 was present in the reaction mixture. The maximum TON was 1942 in a batch reactor. However, in a flow reactor, which allowed for constant injection of FA, and presumably limited decomposition due to water or chloride accumulation, the maximum TON was 92,417. Gonsalvi and co-workers studied FA dehydrogenation using the related tetradentate phosphine ligand, 1,1,4,7,10,10-hexaphenyl1,4,7,10-tetraphosphadecane (P4).13 This ligand can exist as either a rac or meso isomer, and the best TON of 6061 was observed using rac-P4 mixed with Fe(BF4)2·6H2O, which initially generates C. As with Beller’s system, the best catalytic activity was observed in the presence of excess ligand. Milstein’s investigations using Fe complexes supported by tridentate pincer ligands led to complex D, which gave a maximum TON of 100 000 over 10 days in the presence of 50 mol % NEt3.14 Using a related catalyst platform E, Kirchner, Gonsalvi, and co-
(1) MeOH(l) + H 2O(l) ⇌ CO2(g) + 3H 2(g)
ΔG°298 = 2.1 kcal mol−1
(2)
Currently, FA is produced in two steps.5 In the first step, CO reacts with MeOH to form methyl formate (MF), which is then converted into FA, either through reaction with ammonia to generate formamide, followed by acidification with H2SO4, or through direct hydrolysis with water. The hydrolysis step requires high pressure and an excess of water as well as rapid reduction of the pressure and cooling to generate FA. Alternatively, FA can be produced as a byproduct of the liquid-phase oxidation of hydrocarbons to acetic acid. These routes do not use renewable carbon sources, and all are atom inefficient. MeOH is industrially synthesized from syn gas, which is in turn derived from natural gas.6 The production of syn gas is an energy intensive process, which requires temperatures between 700 and 1100 °C. Additionally, the use of a nonrenewable feedstock means that this process does not represent a sustainable long-term route. If FA and MeOH are to be used for CHS, sustainable syntheses of these molecules are required. One attractive option is the direct hydrogenation of CO2.7 In particular, the possible development of environmentally friendly and sustainable catalysts for H2 production8 could open a pathway to the large scale renewable syntheses of FA and MeOH. Moreover, CO2 is a nontoxic, readily available, and inexpensive source of carbon and it is likely that there will be economic and environmental benefits from using it as a feedstock.7 Catalysts are required to facilitate both the dehydrogenation of FA and MeOH and the hydrogenation of CO2 to FA and MeOH. In this Account, we focus on recent developments in homogeneous transition metal catalysts. We emphasize first row transition metal catalysts, especially results from our own groups on Fe based systems, which are likely to be cheaper and less toxic than systems involving precious metals.9 The crucial role of LA cocatalysts in our Fe systems is described. The account is divided into four sections, with each section focusing on catalysts for one of the four reactions of interest. 1050
DOI: 10.1021/acs.accounts.7b00039 Acc. Chem. Res. 2017, 50, 1049−1058
Article
Accounts of Chemical Research
especially relevant for other catalysts that are coordinatively saturated. Almost all coordinatively saturated systems use a base or additive such as excess ligand and we suggest that the role of these reagents is to stabilize charge in the transition state for decarboxylation. This may occur via protonation of the base or ligand by FA to create a cation which stabilizes the H-bound formate ligand. Consistent with this hypothesis, we have demonstrated that a LA can be used instead of excess ligand in FA dehydrogenation using Beller’s best catalyst, B, and that this gives improved reactivity.16 The enhancement of FA dehydrogenation by LA cocatalysts is so dramatic that future work in this area is likely to focus on developing practical applications of the catalyst technology. To maximize the impact of LA enhancement, further work will likely require immobilization of the catalytically active species and incorporation into a flow reactor, which will result in more practical systems where the catalyst can more easily be recycled.
workers obtained a TON of 10 000 over 6 h in the presence of 100 mol % NEt3.15 As with several precious metal systems, the role of the NEt3 in these systems is unclear. Building on work from Milstein, our laboratories have been exploring the chemistry of Fe complexes supported by pincer ligands of the type HN{CH2CH2(PR2)}2 (R = iPr (iPrPNHP) or Cy (CyPNHP)).16,17 We demonstrated that the five coordinate complexes (RPNP)Fe(CO)H (RPNP = N{CH2CH2(PR2)}2−; R = iPr (1a) or Cy (1b)) and the six coordinate complexes (RPNHP)Fe(CO)H(COOH) (R = iPr (2a) or Cy (2b)), formed through the 1,2-addition of FA to 1, are productive catalysts for FA dehydrogenation (eq 3). In fact, when these
■
HYDROGENATION OF CO2 TO FORMIC ACID The direct hydrogenation of CO2 to FA is endergonic (eq 1), and a base such as an amine or hydroxide is commonly used to drive the reaction through deprotonation to generate a formate salt. In some applications, FA is actually used in the form of a formate salt, and in these cases, the use of a base does not represent a problem in terms of atom efficiency.7b For example, sodium formate is used in fabric dyeing and printing processes, as a drilling fluid, and as a food additive. In applications where FA is required, several innovative engineering strategies have been developed that allow separation of FA from a FA-amine salt and subsequent recycling of the amine, meaning the reaction is essentially catalytic in base.19 The first systems for CO2 hydrogenation to formate featuring first row transition metal catalysts were reported in 1976.20 However, TONs were low, and the field was essentially dormant while research focused on precious metal systems. In recent years, a series of well-defined Co and Fe systems for CO2 hydrogenation have been described. The performance of many of these catalysts is summarized in Table 2. In 2012, Beller reported that mixing Co(BF4)2·6H2O with PP3 generated a complex capable of catalytically hydrogenating CO2 and H2.21 When these reactions were peformed using MeOH or ethanol as the solvent and a trialkylamine as the base, the products were
catalysts are used in the presence of a LA cocatalyst such as LiBF4 or NaCl, there is no need for a base or extra ancillary ligand. In the presence of 10 mol % LiBF4, our best system gave a TON of 983 600 (98% yield) and a turnover frequency (TOF) of 197 000 h−1. These TONs and TOFs are comparable with the best precious metal based systems. The proposed mechanism for FA dehydrogenation using our PNP based Fe systems is shown in Scheme 1.16 The 5coordinate complex 1 can enter the catalytic cycle through addition of FA (eq 3) to give the crystallographically characterized formate complex 2.18 Subsequent turnover limiting decarboxylation forms dihydride 3, which has been observed using NMR spectroscopy. Stoichiometric experiments suggest that the LA is crucial for facilitating decarboxylation (Scheme 2),16 as it is able to stabilize the negative charge that develops on the formate ligand in the transition state. Regeneration of 2 and liberation of H2 occurs through the reaction of 3 with FA. This reaction may involve the formation of molecular H2 complexes, but species of this type have not been observed spectroscopically. We propose that this FA dehydrogenation pathway applies to many other systems, and is
Scheme 1. Proposed Mechanism for FA Dehydrogenation Using 1 or 2
1051
DOI: 10.1021/acs.accounts.7b00039 Acc. Chem. Res. 2017, 50, 1049−1058
Article
Accounts of Chemical Research Scheme 2. Pathway for LA Assisted Decarboxylation of 2
Table 2. Selected First Row Transition Metal Catalysts for CO2 Hydrogenation to Formate
catalyst
pressure CO2 and H2 (atm)
T (°C)
solvent
base
Ga H I J Ba Ka L D E M N (1b) O (4a)
30 and 59 39 and 39 20 and 20 35 and 35 30 and 60 30 and 70 29 and 29 3.3 and 6.5 4.3 and 4.3 3.3 and 6.2 35 and 35 35 and 35
100 100 21 45 100 100 100 80 80 55 80 80
MeOH H2O THF-d8 MeCN MeOH THF 100:1 MeOH/THFd 10:1 H2O/THFd EtOH 10:1 H2O/THFd THF THF
1.3 M HNMe2 1 M NaHCO3 0.5 M Verkade’sc 4.8 M DBUf 1.3 M HNMe2 2 M HNMe2 0.72 M NEt3 2 M NaOH 0.5 M DBUf 3 M NaOH 2.4 M DBUf 2.4 M DBUf
additive
0.48 M LiOTf
0.32 M LiOTf 0.48 M LiOTf
TON
ref
1308b 59 9400 29 000 727b 5104b 200e 788 10 275 388 8910 58 990
21 22 23 24 25a 25b 26 27 28 29 30 30
a
Catalyst is formed in situ by mixing metal precursor and appropriate ligand. bTON for DMF production, originating from condensation between FA and HNMe2. cVerkade’s base = 2,8,9-triisopropyl-2,5,8,9-tetraaza-1-phosphabicyclo[3,3,3]undecane. dRatio of volumes. eTON includes MF, originating from condensation between MeOH and FA. Ratio of formate to MF was 2:1. fDBU = 1,8-Diazabicyclo[5.4.0]undec-7-ene.
methyl or ethyl formate, respectively, due to a condensation reaction between FA and the alcohol. If the reactions were performed in the presence of a HNMe2, the formamide DMF was formed as a result of condensation between FA and the amine. Control experiments indicated that the well-defined Co complex [CoH2(PP3)]+ (G) is formed during catalysis and this species was independentely synthesized and shown to be catalytically active. Subsequently, Muckerman, Himeda, Fujita and co-workers reported a Cp*Co (Cp* = η5-C5Me5) complex, H, that is capable of 59 turnovers to formate in aqueous bicarbonate.22 Linehan described a particularly active diphosphine Co catalyst, I, which gives a TON of 9400, but requires the use of a strong and expensive base (Verkade’s base).23 However, unlike most other systems, I shows high activity at room temperature. Drawing from our concurrent work with related Fe catalysts (vida infra), we reported the most
productive Co catalyst to date, J, which can achieve a TON of nearly 30 000.24 In work closely related to studies with the Co complex G, Laurenczy, Beller, and co-workers described in situ generated Fe systems for CO2 hydrogenation formed from Fe salts and tetradentate phosphine ligands.25 Using a complex generated from Fe(BF4)2·6H2O and PP3, they reported 727 turnovers for CO2 hydrogenation to DMF in the presence of HNMe2.25a The mechanism of the reaction was probed using high pressure NMR spectroscopy, and B was proposed to be a key intermediate. This needs to replace the current sentence which starts 'The second and third row transition metal systems. Subsequently, they showed that replacing the PP3 ligand with the similar tetradentate phosphine P(o-C6H4PPh2)3 resulted in a system which gave 5104 turnovers.25b It was demonstrated that reaction of Fe(BF4)2·6H2O with P(oC6H4PPh2)3 generates complex K, which gives comparable 1052
DOI: 10.1021/acs.accounts.7b00039 Acc. Chem. Res. 2017, 50, 1049−1058
Article
Accounts of Chemical Research Scheme 3. Proposed Mechanism for CO2 Hydrogenation Using 4
Table 3. Transition Metal Catalysts for Aqueous Phase MeOH Dehydrogenation to CO2 and Three Molecules of H2
a
catalyst
T (°C)
ratio MeOH/H2Oa
solvent
P Q R S T F (2a)
94 105 90 94 reflux 77
4:1 MeOH/H2O 1:5.5 MeOH/H2O 1:1.3 MeOH/H2O 4:1 MeOH/H2O 4:1 MeOH/H2O 4:1 MeOH/H2O
neat toluene THF triglyme neat ethyl acetate
additive 8 M KOH 10 M KOH
8 M KOH 10 mol % LiBF4
TONb
yield (%)
ref
350 000 ∼29 000 1260 4200 9800 51 000
27 77 84 26 6 50
33 34 35 36 37 38
Molar ratio. bEach equivalent of H2 generated is counted as a TON.
We demonstrated that Fe complexes supported by iPrPNHP and CyPNHP, such as 1 and 2, were active catalysts for CO2 hydrogenation to formate, giving TONs up to 9000 in the presence of LA cocatalysts.30 Mechanistic investigations implicated the LA in disrupting an intramolecular hydrogen bond between the PNP ligand N−H moiety and the carbonyl oxygen of a formate ligand in the catalytic resting state. When the secondary amine PNHP ligand was replaced with an analogous ligand containing a tertiary amine to give complexes such as 4a, hydrogen bonding can no longer occur.30 Using 4a as a precatalyst, TONs of nearly 60 000 were achieved; the
catalytic activity to the in situ generated system. Peters and Fong studied complexes with a related tetradentate anionic SiP3 ligand (SiP3 = Si(o-C6H4PPh2), for example L, which gave lower TON, suggesting that the charge on Fe is important.26 Another noteworthy Fe system was reported by Milstein who showed that complex D incorporating a pyridine-pincer ligand gave 788 turnovers under mild pressures (10 atm) in 2 M aqueous NaOH.27 Modification of the ligand to include a 2,6diaminopyridine (E) or pyrazine backbone (M) resulted in attenuation of the reactivity, particularly for E, which achieved a TON of 10,275 at relatively low pressure (200 °C) and often produce CO, which rapidly poisons the system.32 In contrast homogeneous catalysts offer the possibility of MeOH dehydrogenation at low temperature (
