Ni-Catalyzed Synthesis of Acrylic Acid Derivatives from CO₂ and Ethylene

Abstract

The story of nickelalactones finally ends well. Over three decades after their discovery, catalytic processes have been successfully established to synthesize acrylate derivatives from ethylene and abundantly available carbon dioxide. The performed research during this time in the CO₂ utilization via C–C bond formation with olefins is presented within this review. It gives detailed insights starting from the initial milestones in the 1980s up to modern strategies through cleavage auxiliaries. Different approaches are examined from an experimental and theoretical point of view as the choice of cleavage agent and the corresponding ligand is crucial for the reaction control and suppression of undesired pathways. Methylation of the lactone species led to a first successful liberation of methyl acrylate in stoichiometric amounts. These results led to a vast progress in research with auxiliaries afterward. Upon addition of Lewis acids or strong sodium bases, finally the first two different catalytic routes have been established which are discussed in detail.

1 Introduction

Acrylic acid and its derivatives, i.e., organic esters and alkali salts, are widely used as monomers for the synthesis of polymeric materials (Fig.  1 ). One of the most famous examples is poly(acrylic acid) (PAA) with its unique properties as a superabsorber [1]. Depending on the degree of neutralization to sodium poly(acrylate) (PANa), this polymer is able to absorb water up to 300 times of its own mass. Besides the application in modern diapers, a huge variety of household and daycare products in cosmetics contain PAA as thickener or film-forming substance, adhesives, coatings, and paints. Due to its ability to bind earth alkali elements like calcium and magnesium, it is utilized in sequestering agents. The applications become even more versatile by using blends or copolymers from a variety of poly(acrylate)s [2].

Fig. 1

Typical polymerization of acrylic acid and some of its important derivatives

One of the most commonly used derivatives is methyl methacrylate (MMA) in its polymerized form, poly(methyl methacrylate) (PMMA). The translucency and resistance to acids and bases make it a suitable substitute to common glass materials applied mainly in construction, the automotive sector, and medical technology. Today’s production volume of acrylic acid and its derivatives reach ~4 Mt p.a. with a further increase in the future, especially due to a rising demand in emerging countries. To accommodate this demand, efficient and keen production pathways are requested by the chemical industry [3].

An overview of the abandoned production processes is drawn in Fig. 2 [1]. The cyanohydrin method (i) was established first in industrial scale. Starting from ethylene oxide, the corresponding 3-hydroxypropionitrile is formed by addition of highly toxic HCN. Treatment with H₂SO₄ leads to acrylic acid; however, huge amounts of NH₄HSO₄ are formed as undesired side product. The Reppe synthesis (ii) was introduced in 1939 and used coal-based acetylene and CO to undergo a heterogenic nickel-catalyzed C–C bond formation. Major disadvantages of this route are the toxicity and corrosivity of the catalyst Ni(CO)₄ as well as the huge energy consumption of the overall process. Similar problems occur from the pyrolysis of acetic acid to the corresponding ketene (iii). Again, the high toxicity and demand of energy forced the industry to investigate new strategies toward acrylates. The first established synthesis with a C3 building block was the Sohio process (iv) for the synthesis of acrylonitrile via ammoxidation. Subsequent acidic hydrolysis with sulfuric acid yields the desired product. In common with the cyanohydrin route, huge amounts of NH₄HSO₄ waste are produced.

Fig. 2

Abandoned industrial synthesis pathways to acrylic acid via (i) cyanohydrin, (ii) Reppe synthesis, (iii) pyrolysis, and (iv) Sohio process [1].

Today’s acrylic acid is synthesized from propylene by direct oxidation with O₂ over molybdenum/bismuth catalysts, leading to acrolein as intermediate compound. In a subsequent second step, it is oxidized over molybdenum/vanadium catalysts to acrylic acid [2]. This process has largely replaced any other industrial method nowadays because of the increased selectivity, less undesired by-products, and hence higher profitability (Fig.  3 ) [1]. However, this pathway is limited to propylene as a substrate and thus to unsubstituted acrylic acid. Additional modifications of the carboxylic group need to be performed separately.

Fig. 3

Today’s industrial process for the synthesis of acrylic acid through oxidation of propylene

The finiteness of oil reserves accompanied by the imbalance of supply and demand for fossil fuel leads to fluctuating market prices. Thus, novel routes to acrylic acid replace crude oil-based C3 resources with cheap, renewable, and abundantly available C2 and C1 building blocks as raw materials [4]. Besides the oxycarbonylation of ethylene with CO and water, the nickel-catalyzed direct carboxylation of ethylene with CO₂ (Fig. 4) represents one promising approach for industrial application [2]. The C–C coupling of CO₂ and ethylene with Ni(0) was discovered by Hoberg in the early 1980s [5, 6]. However transformations of nickelalactones were limited to stoichiometric reactions. Postulations about the catalytic employment of nickelalactone intermediates toward acrylic acid in 2006 [7] aroused huge attention in science leading to a dramatic increase of investigations in this topic. Within this review, we want to give an insight into the beginning of this story 30 years ago up to the latest advances in research about the catalytic formation of acrylates from carbon dioxide and ethylene (Fig. 4).

Fig. 4

Ni(0)-catalyzed synthesis of acrylic acid from ethylene and CO₂

2 Formation and Reactivity of Nickelalactones

2.1 Pioneering Work

The initial milestones date back to the year 1980 when Yamamoto et al. published the first cyclic Ni(II) complexes with a nickelalactone motif [8]. The Ni(0) precursor Ni(cod)₂ was treated with acrylic acid and acrylamide in an excess of phosphine ligand. The resulting cyclic amide and ester complexes were confirmed by infrared (IR) and nuclear magnetic resonance (NMR) spectroscopy (Fig.  5 ).

Fig. 5

First published synthesis of nickelalactones and nickelalactams through addition of acrylic substrates by Yamamoto et al. [8]

They postulated a pathway involving a bond rearrangement triggered by the Michael reaction type shift of the acidic proton to the β-carbon of the π-system (Fig. 6). The influence of the phosphine’s steric demand and basicity on the reaction path was also revealed. While PCy₃ and PEt(^tBu)₂ favorably form the cyclic compound, a less basic P-ligand (PEt₃, PPh₃) promotes the reaction to the air-sensitive Ni(0) π-complexes.

Fig. 6

Postulated interconversion mechanism for the nickelalactam formation [8]

Although the cyclic structures demand more space around the metal center, a larger cone angle of the phosphine ligand affords the lactone or lactam preferably. A bulky phosphine coordinates only via one equivalent, while less sterically crowded ligands form a nickel complex with two equivalents of phosphine. This leads to an even more occupied ligand sphere and therefore promotes the π-complex formation. An interconversion from the Ni(0) π-complex to the cyclic Ni(II) species can be observed for (PCy₃)Ni by refluxing the reaction mixture in thf.

In 1982, Burkhart and Hoberg presented the first synthesis of a nickelalactone based on a C1 carbon source and an alkyne [9]. The reaction of Ni(cdt) with tmeda and CO₂ causes a disproportion to Ni(CO)₄ and (tmeda)Ni(CO₃). In the presence of 2-butyne, this pathway is suppressed and affords the five-membered oxanickelacyclopentene derivative instead. Indirect proof of the structure was given upon hydrolysis or CO addition which leads to 2-methylcrotonic acid and dimethylmaleic anhydride, respectively (Fig. 7).

Fig. 7

Reactivity of Ni(0) complexes toward CO₂ in the absence and presence of an alkyne [9]

These findings were further investigated and extended to a vast variety of alkynes and subsequently to alkenes [10–13]. Activated alkenes like norbornene easily undergo a nickel-induced C–C bond formation [14]. The cyclic intermediate was not entirely characterized, but the saturated product exo-norbornane-2-carboxylic acid formed when HCl was added. To extend the library of utilized olefins, Hoberg employed chelating σ-donor ligands (i.e., 2,2′-bipyridine (bpy), 1,2-bis(dicyclohexylphosphino)ethane, (dcpe)) which are suitable for nonactivated alkenes like ethylene [5, 6]. In this case, the lactone formation occurs if Ni(cod)₂ is replaced by the more reactive 16e⁻ precursor Ni(cdt) in the absence of additional tmeda. Tmeda promotes the alternative pathway to the reductive disproportion of carbon dioxide as depicted in Fig. 7. Subsequent acidic hydrolysis of the cyclic compounds affords propionic acid and its methyl esters, whereas CO insertion liberates succinic anhydride (Fig. 8). Already in 1983, Hoberg declared these findings as a first groundbreaking step toward nickel-catalyzed reactions of ethylene and its homologues with CO₂ to a huge variety of organic compounds [5].

Fig. 8

The first oxidative coupling with ethylene and CO₂ and subsequent conversion to organic compounds [5]

2.2 Utilization of Nickelalactones

The research on the nickel-induced C–C coupling with CO₂ expanded tremendously in the following years. A wide range of linear [6] and cyclic olefins [15, 16], 1,2-dienes [13], 1,3-dienes [17–19], alkynes [10], imines, and aldehydes have been studied. Furthermore, the addition of specific promotors (electrophiles, acids, alkyl halides, CO) affords highly selective products from one nickelalactone intermediate, which makes them a versatile tool for organic compounds [20, 16]. The C–C bond formation of alkenes and its homologues affords saturated compounds; alkynes lead to products with an olefinic moiety. Depending on the auxiliary, the deactivation of the Ni complex or hydrolytic workup inhibits a catalytic approach though. Intramolecular reactions like β-H elimination and followed reductive elimination could avoid the use of promotors, but have been elusive so far. An exception is the depicted oxidative coupling of (DBU)₂Ni(0), styrene, and CO₂ [21]. Depending on the reaction temperature, the isomers 2-phenylpropionic acid (A), 1-phenylpropionic acid (B), and additionally cinnamic acid (C) are isolated after hydrolysis. Oxidative coupling at 60°C promotes the formation of the saturated acids (A + B/C = 32/1), whereas a temperature increase to 85°C mainly yields cinnamic acid (A + B/C = 1/7). Derived from the product structure, a thermal β-H elimination is most likely initiated to afford the unsaturated alkyl species upon hydrolysis (Fig. 9). It was the first published synthesis of an α,β-unsaturated carboxylic acid from an olefin and CO₂. A catalysis has not been achieved as the regeneration of (DBU)₂Ni(0) through reductive elimination was not observed. Hoberg assumed that H-migration to the ligand occurs and deactivates the nickel compound for recycling.

Fig. 9

Synthesis of cinnamic acid from styrene via thermal β-H elimination by Hoberg et al. [21]

Besides the utilization of carbon dioxide, isoelectronic isocyanates have been explored, which react with C=N, C=O, or C=C double bonds in analogous manner affording imines, amides, carbamates, and ureas [22–29]. In contrast to the use of CO₂, catalytic conversions with isocyanates are possible.

2.3 Regioselectivity, Ring Contraction, and Expansion

The first mechanistic studies regarding the regioselectivity of monosubstituted alkenes in nickelalactones were reported for 1-hexene by Hoberg in 1984 [6]. The oxidative coupling for the two different occurring isomers was studied at 20 and 60°C (Fig. 10). Due to the higher stability of a metal–C_prim bond, the lactone formation reveals a preference of product A over B. At 20°C an isomeric ratio of A/B = 4/1 is reached. With rising temperatures, the equilibrium shifts further to the thermodynamically favored product A (A/B = 25/1).

Fig. 10

Regioselectivity and reversibility of the C–C bond formation with monosubstituted alkenes, Ni(cdt), and dcpe [6]

A subsequent treatment of a 4/1 mixture at 60°C affords a final ratio of A/B = 25/1, indicating that the C–C bond formation is a reversible step in the overall reaction. Various monosubstituted alkenes (e.g., propylene) reveal analogous regioselectivity, whereas specific ligands (i.e., (2-diethylaminoethyl)-dicyclohexylphosphine) are capable to invert the regioselectivity toward isomer B. Due to the electronic interactions of the precursors, the electrophilic CO₂ carbon favors the C–C bond formation with the nucleophilic C1 carbon of the alkene. The electronic effects play a bigger role than the thermodynamically favored formation of a M–C_prim bond and are in good accordance with the observed regioselectivity in the coupling reaction of imines and aldehydes with CO₂ [30, 31]. Yamamoto et al. discovered similar M–C_sec isomers [32] when ring contractions of six-membered nickelalactones occur (Fig. 11). The stability of six-membered nickelacycles is highly influenced by the employed auxiliary ligand.

Fig. 11

Ligand-induced ring contraction of six-membered nickelalactones [32]

Bulky ligands like 1,2-bis(diphenylphosphino)ethane (dppe) and its propyl-bridged homologue (dppp) cause a rapid isomerization to the methyl-substituted five-membered lactones, while less sterically demanding ligands like bpy in contrast do not induce a ring contraction. A similar trend can be observed for monodentate phosphines (PPh₃ > PEtPh₂ > Pet₂Ph > PMePh₂ > PEt₃ > PMe₂Ph) with regard to the Tolman’s cone angle. In conclusion, a higher bulkiness of the ligand promotes the ring contraction. The degree of contraction was determined by the ratio of the respective methyl succinic anhydride and glutaric anhydride after treatment with CO. Approaches to perform the backward ring expansion were unsuccessful indicating an irreversible process due to the high stability of five-membered lactone species.

Ring expansions are accessible via insertion of olefinic compounds into the M–C bond of saturated nickelalactones bearing two 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) ligands [33, 34]. The widely used phosphines and bpy suppress this reaction pathway; respective products have not been isolated. By contrast, DBU promotes the subsequent incorporation of a second C=C double bond (Fig. 12). The five-membered lactone precursor expands to the seven-membered lactone and can be cleaved hydrolytically to obtain elongated and functionalized acids. In certain cases, β-H elimination occurs, resulting in the cleavage of the ring, formation of a Ni–H species, and regeneration of the olefinic function. The reductive elimination to the free acid and (DBU)₂Ni(0) does not take place and therefore prevents a catalytic run. The same phenomenon was observed for the C–C coupling of styrene (Sect. 2.2) [21].

Fig. 12

Reactions of (DBU)₂NiLac with C=C-containing substrates via ring expansion [34]

With the employment of DBU, a first successful X-ray crystal structure analysis was obtained for the five-membered nickelalactone consisting of CO₂ and ethylene (Fig. 13) [34]. The 16e⁻ complex revealed a square-planar coordination of the Ni atom, which is bound to the imine nitrogen of two DBU units. Due to the planar orientation, a nonbinding interaction of the metal with the hydrogen of C _α was excluded.

Fig. 13

Crystal structure of (dbu)₂NiLac with a square-planar coordination sphere [34]

2.4 Preliminary Studies Toward Acrylic Acid

The interest in the topic of nickelalactone utilization decreased in the 1990s leaving the question if a catalytic conversion for CO₂ could ever be realized. Despite a huge variety of synthetic methods, promotors, and ligands, each approach was limited to stoichiometric conversions at some point. Either the crucial β-H elimination or the reductive elimination did not occur. For this reason, alternative metal centers (Fe [35, 36], Pd [37, 38], Pt [39, 40], Ti [41], Zr [42], Rh [43], Mo [44, 45]) have been investigated. Nevertheless, Ni seemed to be the most versatile metal center in the C–C bond formation with olefins. In 2006, Walther et al. revived the vision of the catalytic conversion to acrylic acid. They postulated a theoretical catalytic cycle of the “dream reaction” (Fig. 14) [7]. The first step involves the oxidative coupling of a Ni(0) with ethylene and CO₂ and subsequent generation of an acrylate species via β-hydride elimination of the five-membered lactone. The reductive elimination of the Ni(II) center affords the π-coordinated acid which is released by ligand exchange with ethylene and hence closes the catalytic cycle.

Fig. 14

Hypothetical catalytic cycle to the formation of acrylic acid from ethylene and CO₂ via nickelalactone intermediates by Walther et al. [7]

One of the key steps in this catalytic cycle is the β-H elimination that has been elusive so far except for styrene (Sect. 2.2) [21]. Walther investigated phenyl-substituted bis(phosphine)s with different bridge lengths. They revealed that the reaction pathway upon addition of ligand to (tmeda)NiLac is strongly dependent on the bridge length. Longer backbones either afford stable complexes (dppe), or a reductive decoupling (dppp, dppb) is induced, whereas the methylene-bridged ligand 1,1′-bis(diphenylphosphino)methane (dppm) promotes a β-H elimination and subsequent dimerization with the decoupled species (dppm)₃Ni(0) to an acrylate-bridged compound (Figs. 15 and 16). However, reductive elimination from the complex to release the acrylate does not occur. Due to its stability and the sacrificial cleavage of one dppm ligand to form a phosphido bridge (PPh2⁻), it cannot be utilized in further syntheses.

Fig. 15

Synthesis of a bridged binuclear nickel-acrylate complex through β-H elimination [7]

Fig. 16

X-ray analysis of the binuclear Ni species with the acrylate-bridged ligand motif [7]

Further investigations by Walther focused on the ligand modification [46]. They obtained model compounds of the different postulated intermediates in the catalytic cycle employing rigid and bulky 1,1′-bis(di-iso-propylphosphanyl)ferrocene (dippf) and 1,1′-bis(diphenylphosphanyl)ferrocene (dppf) (Fig. 17). Depending on the substituents at the phosphorous atom and the synthetic strategy, a stable Ni(0) π-complex with coordinated acrylic acid, the corresponding nickelalactone, or the Ni(0)(C₂H₄) π-complex is formed. Conversion of the cyclic compound to the ethylene complex was observed by CO₂ liberation; however, no isomerization of the lactone to the acrylic acid occurs.

Fig. 17

Synthesized model compounds bearing a rigid dippf ligand [46]

Buntine et al. performed detailed computational calculations on the desired catalytic C–C coupling between CO₂ and ethylene mediated by the (DBU)₂Ni(0) catalyst [47]. The quantum mechanical data was elaborated using the model ligand (mDBU) in which the seven-membered ring was substituted by two methyl groups. It decreased the computational effort. The data reveals three main barriers in the catalytic reaction: (1) the oxidative coupling (+121.8 kJ/mol), (2) β-H elimination (+147.4 kJ/mol), and (3) the reductive elimination (+104.1 kJ/mol) through a most likely “five-center three-ligand” transition state. A transition structure for the β-H elimination in the five-membered nickelacycle proved unsuccessful due to the restricted geometry in the ring. Through an elongation of the Ni–O bond, the abstraction of the hydride may occur. With regard to the values, these barriers do not hinder the catalytic process at moderate reaction conditions; however, the overall endergonic reaction (+42.7 kJ/mol) of the catalytic cycle is most likely the biggest obstacle to overcome from a thermodynamic point of view. The deactivation pathway through H-migration of Ni–H to the N=C double bond of DBU has a moderate barrier (+150.2 kJ/mol), but the instability of the product (+102.1 kJ/mol) makes it negligible, in contrast to theories of Hoberg (Sects. 2.2 and 2.3) [34]. Prospective solutions have to improve the overall thermodynamics by removal of the acrylic acid and shift the equilibrium to the product site or modify the desired acrylic acid to improve the overall energy balance.

3 Liberation of Acrylates

3.1 Methylation of Nickelalactones

Previous studies and DFT-based calculations lowered the expectations to realize a Ni(0)-catalyzed process to acrylic acid [47]. Therefore, novel approaches and strategies were required. In 2010, Rieger et al. reported the first liberation of an acrylate derivative via methylation of nickelalactones [48]. Upon addition of 2–100 equiv. of MeI to the reaction mixture of (dppp)NiLac in CH₂Cl₂, the respective methyl acrylate is obtained in yields up to 29%. Neat MeI increases the yield by 4%. The replacement of MeI with LiI yields 6% Li(acrylate) due to its low solubility and weaker interaction with the Ni–O bond. Other methylating agents like methyl triflate and trimethyloxonium tetrafluoroborate afford only traces of acrylate. Although only stoichiometric amounts of methyl acrylate were formed, a new hypothetic catalytic cycle was postulated (Fig. 18). After the oxidative coupling, the cleavage of the Ni–O moiety by MeI promotes the β-H elimination to release the desired methyl acrylate. The reductive elimination of the side product HI finally regenerates the Ni(0) catalyst.

Fig. 18

Postulated catalytic cycle for the formation of methyl acrylate from nickelalactones by Rieger et al. [48]

Analogous to reactions of platina- [40] and palladalactones [49] with MeI, the methylation induces the elongation of the Ni–O bond, so that subsequent β-H elimination is promoted. Furthermore, the unfavored reductive elimination of a stable Ni–O bond is circumvented. Closure of the catalytic cycle was not achieved due to the formation of methyl propionate as by-product and unsuccessful recovery of the active species.

Limbach et al. performed deeper experimental and theoretical studies on the methylation with the dtbpe ligand (Fig. 19) [50]. They found out that the methylation most likely occurs at the carbonyl oxygen and proceeds via S _N 2-type reaction with a low barrier. The methylated species equilibrates between a cationic four- and five-membered intermediate via β-H elimination. The final formation of methyl acrylate offers two possible pathways. Either a second nickelalactone deprotonates the cationic intermediate and affords the Ni(acrylate) π-complex, or the postulated β-H elimination occurs and liberates methyl acrylate from the complex. Furthermore, they identified unfavorable pathways, which inhibit a catalytic approach due to the formation of HI.

Fig. 19

Hypothetic catalytic cycle forming methyl acrylate (black) and examined side reactions to methyl propionate (red) by Limbach et al. [50]

Due to the unselective reaction with MeI, a range of alternative methylating agents have been examined for various nickela- and palladalactones by Kühn [51, 52]. In contrast to results of Rieger and Limbach, methyl triflate turned out to be the most suitable cleavage agent under the chosen reaction conditions, independent of the auxiliary ligand. A fast transformation of the cyclic ester to the nickel hydride and the ring opening intermediate was observed via in situ H-NMR spectroscopy. A catalytic conversion was not performed though.

3.2 Lewis Acid-Induced β-H Elimination

The spontaneous β-H elimination from a five-membered nickelalactone has not been observed so far, except for one styrene-based lactone [21]. However, the use of Lewis acids reveals a promoting effect on β-H elimination. When we take a look back, Hoberg added BeCl₂ to various substituted and unsubstituted nickelalactones [53]. An immediate ring contraction occurs, most likely via β-H elimination and subsequent 2,1-insertion. Due to the ring strain and the Ni–C_sec bond, the formed β-lactone structures exhibit a higher reactivity than their five-membered congeners and readily insert a second substrate to expand to a six-membered ring (Fig. 20). Therefore, the four-membered intermediate was not isolated. Via this route, geminal dicarboxylic acids are accessible by the addition of a second carbon dioxide. Again, the stability of the Ni–O moiety prevents a catalytic approach; hence, the acidic hydrolysis is essential to liberate the organic products.

Fig. 20

Postulated pathway to geminal dicarboxylic acids through Lewis acid-induced ring contraction by Hoberg et al. [53]

Based on these results and observations with Li⁺ [48], a similar strategy was published by Bernskoetter [54]. Treatment of nickelalactones, in this case (dppf)NiLac, with the strong neutral Lewis acid B(C₆F₅)₃ at rt promotes a rapid β-H elimination. The most likely formed nickel-acrylate hydride intermediate was not detected, but a subsequent 2,1-insertion affords the first stable four-membered β-lactone (Fig. 21). The X-ray analysis exhibits a formally zwitterionic product with the expected ring structure and the Lewis acid attached to the oxygen.

Fig. 21

Lewis acid-induced transformation of a nickelalactone via β-H elimination and subsequent 2,1-insertion to the thermodynamically favored β-lactone [54]

Further utilization of the β-lactone is possible via deprotonation with neutral organic bases. With DBU, unselective side reactions arise, and the Ni species partially decomposes, whereas BTPP selectively affords the deprotonated π-complex at rt after two days. From a mechanistic point of view, the source of the withdrawn proton is not identified. Most likely the deprotonation takes place from the nickel hydride intermediate. Alternative pathways like the proton abstraction from the C _α position in the four-membered ring or from the CH₃ group in β-position seem unlikely but cannot be fully excluded with regard to the pK _a values of the complex and the base. The subsequent ligand exchange readily occurs by addition of ethylene and expels the formed [BTPPH][(B(C₆F₅)₃)(acrylate)]. However, the establishment of the last step of the catalytic cycle remained elusive, due to the fact that dppf does not induce an oxidative coupling with CO₂ under the investigated reaction conditions (Fig. 22).

Fig. 22

Deprotonation and subsequent release of the acrylic salt from a 2,1-insertion nickelalactone [54]

Recently, Bernskoetter examined a strong influence of the chosen ligand on the lactone’s reactivity [55]. Dcpm and its ethyl-bridged congener dcpe both perform the C–C bond formation of CO₂ and ethylene to the γ-nickelalactone in average to high yields (47% for dcpm, 88% for dcpe). Upon addition of a weak Lewis acid (NaBAr^F ₄), the Na⁺ coordinates to the carboxylic moiety. Subsequent heating to 55°C in thf induces a partial γ-to-β isomerization (β/γ = 1/3.5 for dcpe, 1/8 for dcpm), verified by NMR spectroscopy and X-ray analysis. In contrast to the dcpm ligand, thf solutions of (dcpe)NiLac even isomerize in the absence of Lewis acid to an unactivated β-lactone. The unactivated species is less stable than the sodium adduct; therefore, a higher preference for the five-membered γ-lactone can be observed in equilibrium in this case (Fig. 23).

Fig. 23

Lewis acid-assisted and Lewis acid-unactivated thermal isomerization to a β-lactone [55]

Computational analysis supports the beneficial effect of Na⁺ on the γ-to-β isomerization [55]. It provides a decrease of the ring strain in the formed β-lactone and stabilizes the negative charge of the carboxylic moiety in the transition state. The addition of suitable Lewis acids as cocatalysts that can bind reversibly to the carboxylic moiety to promote β-H elimination could introduce new strategies toward catalytic reactions.

3.3 Catalytic Routes

The addition of cleavage auxiliaries has been broadly investigated recently. However, the employed electrophiles and Lewis acids do not lead to catalysis. Either one of the elementary steps remained elusive, or side reactions to inactive species occurred. Investigating a different approach, Limbach et al. reported in 2012 the first successful nickel-catalyzed synthesis of an acrylate, derived from carbon dioxide and ethylene (TON = 10) [3]. With the choice of a suitable ligand and the sacrificial use of a strong sodium alkoxide base, a selective catalytic reaction was established. Formation of Na(acrylate) as desired product shifts an overall endergonic reaction (for acrylic acid) [47] to a favorable exergonic process (−59 kJ/mol) (Fig 24) [56].

Fig. 24

Quantum mechanical energy profile for the catalyzed synthesis of Na(acrylate) by Limbach et al. [3]

The initial step of the catalytic cycle represents the oxidative coupling. It reveals a strong influence by the chosen ligand. The literature-known ligands, i.e., bpy and DBU, only offer limited solubilities and stabilities; therefore, Limbach focused on tert-butyl and phenyl-substituted bis(phosphine)s with different bridge lengths ((–CH₂−)_1–3). The phenyl-substituted bis(phosphine)s rapidly form Ni(dppe)₂ and Ni(dppp)₂ upon addition of Ni(cod)₂, even in the presence of ethylene and CO₂. The use of bulky tert-butyl groups inhibits the deactivation by two ligands attached to one Ni(0) center. Due to their strong σ-donating properties, they rapidly form the corresponding π-ethylene complexes and subsequently induce the C–C bond formation with CO₂. This is in good accordance with a mechanistic point of view. The reaction is expected to occur through the C–C bond formation of an incoming CO₂ into a Ni–C₂H₄ bond; hence, the formation of a Ni-π-(ethylene) complex is crucial. With regard to the bridge length, only (dtbpe)NiLac forms a stable isolable compound, whereas (dtbpm)NiLac is only detectable under CO₂ pressure via ³¹P NMR spectroscopy.

In the second step, strong sodium alkoxides induce a cleavage to the corresponding acrylate species. The role of the base has not been clearly confirmed yet since different mechanistic procedures seem possible according to the quantum mechanical calculations. This includes the deprotonation in the acidic α-position of the lactone with a subsequent rearrangement to the nickel π-acrylate species or the β-H elimination of an activated intermediate, as suggested for palladalactones by Aresta [38]. NaO^tBu turned out to be the best suitable base under the investigated reaction conditions. Further studies have shown that the sodium cation, as well as the base, is crucial for the transformation to the acrylate. The quaternary ammonium salt NBu₄OMe does not promote a lactone cleavage, whereas the supplementary addition of NaBAr^F ₄ accomplishes the reaction with 47% yield. The last step requires a ligand exchange of the coordinated acrylate with ethylene to regenerate the potential catalyst. Due to a weaker π-back bonding of Na(acrylate) compared to Me(acrylate) or acrylic acid [3, 50, 56], the pressurized ethylene can release Na(acrylate) from the complex and hence close the catalytic route (Fig. 25). As a disadvantage, the presented catalysis has to be performed in two repetitive steps. Alkoxide bases irreversibly form carbonic half esters under CO₂ pressure, which inhibit a deprotonation. Therefore, the overall reaction is divided into a CO₂-rich (oxidative coupling) and CO₂-poor (cleavage and liberation) regime.

Fig. 25

Ni(0)-catalyzed synthesis of Na(acrylate) from NaO^tBu, CO₂, and ethylene [3]

In further research, Limbach reported an optimized one-pot-catalysis (Fig. 26) [57]. Firstly, the sacrificial base was replaced by sodium phenoxides, due to their lower nucleophilicity. Hence, they are suitable bases under CO₂ pressure which do not form carbonate half esters. In correlation with the according pK _a values (ortho < meta < para), sodium 2-fluorophenoxide reveals the highest activity of the tested substrates. NaOPh, Na(4-fluorophenoxide), and phenoxides with +I substituents exhibit significantly lower TONs. Studies on small-bite angle ligands (dtbpm, dppm), monodentate phosphines (PPh₃, P^tBu₃), and stereogenic bis(phosphine)s revealed the highest TONs for the chiral ligands (S)-BINAPINE, (S,S,R,R)-TangPhos, (R,R,S,S)-DuanPhos, and (R,R)-BenzP*. The addition of an excess of zinc has beneficial effects on the TON; however, the specific reason remained unclear.

Fig. 26

Advanced catalytic cycle with (BenzP*)Ni(0) as active species and X-ray structure of the lactone intermediate [57]

Additional studies on various substrates have been performed with the BenzP* system at 100°C in thf. The direct carboxylation is applicable to a wide range of olefins including 1,3-dienes, styrenes, cycloalkenes, Michael acceptors, and functional group containing olefins; however, it exhibits the highest TONs for ethylene (TON = 107) and 1,3-butadiene (TON = 116) (Fig. 27). In the case of 2-vinylpyridine, styrenes, and 2,3-dimethyl butadiene, the catalytic activity dramatically decreases (TON = 2–12). Summarizing, a one-pot reaction with TONs > 100 was successfully developed, but TOF values up to 6 h⁻¹ still prevent this catalysis from applicability in a huge scale.

Fig. 27

Excerpt of the carboxylated substrates via one-pot catalysis [57]

Vogt et al. recently published an optimized version of Rieger’s initial approach with MeI [48, 58]. They realized a catalytic formation from CO₂ and ethylene employing LiI as “hard” Lewis acid (Fig. 28). As examined and calculated for various Lewis acids [56, 55], the Li⁺ cation elongates the carboxylic moiety and promotes β-H elimination. According to DFT calculations support the choice of Li⁺ over Na⁺, since the Gibbs free energy of the transition state (+104 kJ/mol without Lewis acid) is decreased by 20 kJ/mol for Na⁺ and +40 kJ/mol for Li⁺. Immediate release of the product and regeneration of the catalyst by trapping the formed HI with the base close the catalytic cycle. The role of the iodide remains unclear.

Fig. 28

Proposed catalytic cycle for the acrylate formation upon addition of LiI by Vogt et al. [58]

In order to suppress side reactions with the released HI (i.e., propionate), NEt₃ is added. DIPEA, K₂CO₃, and Cs₂CO₃ were less efficient. The addition of Zn dust prevented the formation of inactive NiI₂ species reaching a TON = 21 for dcpp after 72 h. Overall, a huge amount of sacrificial auxiliaries (base, Zn) is added to the reaction mixture, and the low TON values are far from a useful application in synthesis. But apart from Limbach’s strategy involving sacrificial bases [57], the route by Vogt provides new opportunities to the formation of acrylates with Lewis acids.

4 Conclusion

Since the discovery of the nickel(0)-induced oxidative coupling of CO₂ and ethylene over three decades ago, the formed nickelalactones have been considered as potential intermediates in catalytic approaches. Initial investigations were focused on the stoichiometric conversion to useful organic compounds, e.g., saturated carboxylic acids, esters, and anhydrides. Catalytic strategies however remained elusive. With regard to the synthesis of acrylates, the overall thermodynamics and the missing elimination reactions were pointed out as main obstacles in a catalytic formation.

After more than 25 years of research, the methylation of a nickelalactone with MeI afforded the first successful liberation of a methyl acrylate species. The methyl group elongates the Ni–O moiety and facilitates a subsequent β-H elimination to release the product. This route is accompanied by the formation of inactive Ni(II) compounds and by-products that can only be suppressed to a certain amount. Although limited to stoichiometric conversion, it was the basis for further experimental investigations and computational studies. Hence, a wide range of potential cleaving agents (i.e., electrophiles, Lewis acids) have been examined recently. Finally, two different catalytic pathways have been developed. In the presence of LiI, a catalytic conversion to Li(acrylate) with a TON = 21 was achieved. The Li⁺ ion serves as Lewis acid and promotes the Ni–O bond dissociation. However the reaction is slow and occurring side reactions were suppressed upon stoichiometric addition of base and reducing agent. Sacrificial sodium alkoxide bases offer an alternative catalytic route via deprotonation of the lactone intermediate and subsequent transformation to sodium acrylate. By ligand modification, exploration of suitable bases, and optimization of the reaction conditions, one-pot catalysis with a TON > 100 was achieved. Further examinations will focus on the acceleration and versatility of the catalysis, since this system is not limited to ethylene only. In order to compete with the current oil-based industrial processes one day, the catalyst activities have to improve dramatically. With respect to the recent breakthrough in this field, one can be excited about prospective developments in the nickelalactone utilization. Due to the rising demand for fossil resources, novel and modern strategies like the catalytic CO₂-based C–C coupling with olefins to basic chemicals are highly demanded in the future.