Toward ideal carbon dioxide functionalization

From carbon fixation, Grignard reaction, metal-catalyzed reactions and asymmetric CO2-incorporation, what would be the ideal CO2-functionalization?


Introduction
Carbon is an essential element for all living organisms, and is present in carbohydrates, amino acids, proteins, and lipids. These biomolecules are synthesized with specic selectivities controlled by the natural molecular foundryenzymesto sustain forms of life. The sustainability of bio-and chemical networks in living organisms is powered by the seemingly unlimited solar energy. Owing to the evolution of cyanobacteria and their photosynthesis, 1 our planet became a unique biosphere where water was split into oxygen and hydrogen, while consuming (or xating) CO 2 to generate reduced organic matter.
Photosynthesis and CO 2 xation operate under ambient conditions; articial photosynthesis is yet to be realized, 2 and can ensure sustainable growth of the human civilization. The challenge lies in overcoming the thermodynamic stability and kinetic inertness of CO 2 , which possesses the highest oxidation state of carbon. Therefore, it is inevitable to employ reducing reagents (reactive metals, H 2 , electricity, and highly reducing chemicals) to overcome the intrinsic reaction barrier of CO 2activation, particularly to enable the reactions to be operative under mild reaction conditions.
Recently, the global society has raised concerns related to excessive energy consumption and uncontrollable anthropogenic CO 2 emission. 3 Although CO 2 functionalization can provide ideal solutions, chemical reactions with CO 2 currently suffer from low efficiency, making it impossible to mitigate the overwhelmingly large quantity of accumulated CO 2 in the atmosphere at low concentrations. 4 Yet, chemical recycling of carbon dioxide has been recognized as a promising supplement to the natural carbon cycle, 5 while producing valueadded ne chemicals. 6 In this context, CO 2 can serve as an inexpensive and non-toxic renewable C1-building block. 4,7 For example, light hydrocarbons and C 1 -or C 2 -units (i.e. carbon monoxide, formic acid, formaldehyde, methanol, and oxalic acid) are accessible from CO 2 , mostly catalyzed by heterogeneous materials (semiconductors, 8 zeolites, 9 COFs, 10 MOFs, 11 and g-C 3 N 4 (ref. 12)). On the other hand, homogeneous catalysis has shown remarkable potential in C-C bond formation reactions, via formal insertion of CO 2 at C-H bonds. The utility of carboxylic acids and their derivatives is certainly applicable with broad interest in organic synthesis 13 and pharmaceutical chemistry. 14 As categorized in Table 1, catalytic CO 2 -functionalization reactions have been reviewed, particularly transition-metal catalyzed C-C bond formation reactions, 15 carboxylation reactions catalyzed by palladium, 16 silver, 17 copper 18 or copper-NHC (N-heterocyclic carbene) complexes, 19 and nickel/iron 20 catalysts, asymmetric CO 2 -functionalization reactions 21 and photocatalytic CO 2 -functionalization. 22 Other types of reactions are Recycling of carbon dioxide to methanol and derived products-closing the loop also tabulated to guide the readers for further reading in specic topics of interest. For example, carbonate formation reactions with epoxides and ring-strain mediated reactions, 23 catalytic alkylation with CO 2 , 24 etc., will not be discussed in this Perspective.
The purpose of this Perspective is the following: providing a general concept of catalytic CO 2 -functionalization by exemplifying recent progress (up to 2018). Section 2 will discuss transition-metal catalysis with a hint of sustainability. Sections 3 and 4 will explore recently reported photochemical redox catalysis by utilizing synthetic dyes with the aid of preestablished transition metal catalysis, and single-electron reduction of CO 2 via a redox-neutral mechanism. Section 5 will focus on a handful but remarkable examples of asymmetric C-C bond formation reactions by the action of metalchiral ligand complexes. The future perspective on ideal CO 2functionalization will also be discussed in the context of umpolung carboxylation, redox-neutral photochemistry and asymmetric CO 2 -activation to reduce the prevailing energy input or highly reactive species. This discussion will lead to an alternative platform for sustainable CO 2 recycling, to mimick the natural carbon cycle by utilizing the combined knowledge in organic, inorganic, photo-and materials chemistry, and enzymatic engineering for improved carbon xation as well.
The Martin group employed a Pd(II)-Pd(0) cycle in the catalytic carboxylation reaction of aryl bromides using ZnEt 2 as a terminal reducing reagent. 27 This methodology was further expanded to abundant Ni(II) catalysis by the Tsuji group, 28 realizing carboxylation of aryl chloride with Mn powder as a reducing reagent. New reductive carboxylation reactions were developed later by the Martin group with a broad range of substrate scope, including organic halides, 29 sulfonates, 29b esters, 30 benzylic ammonium salts 31 (Scheme 2a), allyl acetates, 32 allyl alcohols 33 (Scheme 2b), and unsaturated hydrocarbons (Scheme 2c and d). 34 The facile insertion of CO 2 into R-Ni was tested with olen substrates, enabling olen activation without an apparent hydride donor (Scheme 2). These protocols provided a broad substrate scope and high functional group tolerance. However, it is necessary to use (over) stoichiometric amounts of reducing reagents (i.e. Mn, Zn, ZnR 2 , and etc.) to complete the catalytic cycle.
In 2017, a breakthrough CO 2 -functionalization was reported by the Martin group proposing a 'chain-walking' mechanism with catalytic Ni-H species (Scheme 3). 35 Although the b-hydride elimination is undesired in transition metal-catalyzed coupling reactions, 36 in the proposed reaction mechanism, a chainwalking process was key to generate thermodynamically more stable species, thus contributing to the high regio-and chemoselectivity of the targeted insertion reactions. 37 For carboxylation reactions with CO 2 , the Martin group showed temperature-controlled site-selectivity affording linear and branched carboxylated products (l : b ratios). The authors suggested a Curtin-Hammett scenario, where the reaction proceeded through common intermediates or transition states under fast equilibrium (Scheme 3a). More strikingly, the chainwalking mechanism was translated to a useful method starting from a mixture of alkyl bromidesexpanding the utility of the protocol signicantly. Regardless of regioisomers, linear alkanes were smoothly converted to carboxylated products under a bromination/carboxylation reaction sequence (1 atm of CO 2 ). The iterative reversible b-hydride elimination/insertion reactions occurred, converging regioisomers of alkyl bromides into a single carboxylated product (Scheme 3b).
The proposed chain-walking process with high siteselectivity represents a signicant potential toward fatty-acid syntheses from bulk petroleum raw materials. In this context, the same group extended the methodology with olen substrates, enabling carboxylation reactions in the presence of water as a proton source. 34b In the case of alkenes, water served as a way to access metal-hydride species, 38 namely Ni-H species, which in turn can participate in the above-mentioned chain-walking mechanism. Indeed, a linear carboxylic acid was the main product with high selectivity (b : l ¼ 1 : 99) even from an unre-ned mixture of olen isomers (Scheme 4a). As for alkynes, however, only a branched carboxylation product was obtained (Scheme 4b). The authors proposed that the Ni-L2 complex favored the formation of a thermodynamically more stable a,bunsaturated nickelalactone (Ni-1) with internal alkynes in a CO 2 environment. Therefore, a branched carboxylic acid was obtained with high selectivity (b : l ¼ 99 : 1) aer reduction with H 2 and Pd/C. The 'uni-directional' chain-walking mechanism highlights the potential application of this process in producing added value chemicals from CO 2 and crude industrial feedstock.
It is noteworthy that the variation of the ligand is critical in Ni-catalyzed reactions. The substituent adjacent to the nitrogen atoms in bidentate ligands (L1 and L2), such as bipyridine and phenanthroline, differentiates the site-selectivity of the carboxylation reaction. High site-selectivity is a pre-requisite for many organic transformations, for example in allylic substitution reactions. Catalytic metal-ligand complexes govern chemo-, regio-and even enantioselectivity. 39 Allyl alcohol is a substrate class with high accessibility yet low chemical utility for allylation reactions due to the apparently low leaving group ability of the hydroxide. It has been proved that in situ activation of allylic alcohol with 'activating reagents' can mediate various types of transformation, 40 shortening the synthetic steps avoiding the preparation of activated substrates 41 (like amines, 41a ammonium salts, 41b carbamates, 41c carbonates, 41d esters, 41e ethers, 41f nitro compounds, 41g phosphates, 41h and sulfones 41i ). For example, CO 2 was involved in the asymmetric Pd-catalyzed direct a-allylation of ketones. 40a The use of CO 2 as a catalyst is noticeable although only a 'catalytic'-amount of it would be necessary for the process.
The Martin group employed CO 2 as an activating reagent as well as a C1 source for the carboxylation of allylic alcohols to afford b,g-unsaturated carboxylic acids (Scheme 5a). 33 Once again, ligand-controlled selectivity was observed starting from linear or branched allylic alcohols affording high yields of linear and branched carboxylation products (Scheme 5b). The former resulted from CO 2 insertion between the a-carbon and Ni(I) center. Alternatively, a-branched acids were obtained when the tridentate ligand L3 was employed. The critical role of the ligands was rationalized by stoichiometric studies of active NiL2 or NiL3 species in the absence of Zn metal (yields: linear, 0%; branched 73%). The transition-state, Ni-2, was proposed for the nucleophilic attack from the g-carbon of an h 1 -allyl Ni(II) intermediate to CO 2 . Also, a six-membered cyclic conformation can be suggested, similar to the reported nucleophilic addition of Pd-(p) allyl intermediates to CO 2 or carbonyl substrates. 42,43 The utility of the reaction was further veried by producing useful intermediates for the synthesis of g-lactone-based bioactive compounds. 44 Dienes, abundant and accessible chemical feedstocks, have the same oxidation states as allylic alcohols. However, activation of dienes and conjugated olens poses a great challenge. Recently, Ni-based catalysts were evaluated for a catalytic carboxylation reaction of dienes toward carboxylated or dicarboxylated products in stoichiometric amounts of a Ni(0) complex. 45 Although limited only to activated substrates, alkynes 46 and silylallenes 47 were transformed to the desired dicarboxylated products. The Martin group successfully implemented a catalytic dicarboxylation reaction for 1,3-dienes with high site-selectivity (up to 90%), to furnish diesters (Scheme 6a). 34c Various functional groups were tolerated including heterocycles, organotin, nitrile, and esters. Structurally simple dienes such as butadiene, isoprene, and piperylenemajor byproducts of steam cracking in ethylene production plantswere converted to the corresponding terminal diacids with excellent site-selectivity in moderate yields (up to 65% yield, 99 : 1 selectivity, Scheme 6b). Single crystal structure analysis determined the formation of monocarboxylated h 3 -Ni nickelalactone (Ni-3). The corresponding dicarboxylation product could be obtained when Ni-3 was treated with CO 2 under optimized reaction conditions (also see Mori group's work 45c ), shedding some light on the reaction mechanism (Scheme 6c).
The transition metal-catalyzed carboxylation reactions of the above-mentioned recent examples showed unprecedented catalytic performances with a variety of substrates, yet they require stoichiometric reducing reagents to sustain the catalytic cycle.
Certain improvements have been attempted by utilizing insoluble reducing reagents (Mn, and Zn powder) replacing highly reactive RMgX, Et 2 Zn, or AlMe 3 . In an ideal CO 2 functionalization process, a redox-neutral mechanism would be more desirable, 48 where no additional oxidants or reductants are required. In this context, the next two section will describe reactions utilizing photocatalysts, demonstrating sustainable light-induced chemical reduction reactions, mimicking photosynthesis.

Photocatalytic carboxylation with CO 2
Photosynthesis is the master process in the realm of CO 2functionalization, as it is called CO 2 -xation. This ideal process operates via multi-step electron transfer and chemical transformation reactions, 49 resulting in somewhat limited CO 2 -xation efficiency, constraining the capacity of nature's carbon cycle ( Fig. 1 highlighted in green). 50 Recent efforts in enzymatic engineering in chemical biology for in vitro CO 2 xation 51 would potentially lead to enhanced photosynthesis. 52 Solar energy obviously represents one of the most promising and limitless energy sources, which can be harnessed using a photosensitizer. In the late 1970s, seminal studies were reported regarding photocatalytic CO 2 reduction by the Tazuke, Fujishima, Honda, and Lehn groups, 53 which formed the basis of the modern photoredox activation of CO 2 . Further developments in photocatalysts played a signicant role in CO 2 reduction reactions mainly targeting industrial feedstock

View Article Online
Scheme 6 Ni-catalyzed dicarboxylation of 1,3-dienes and a mechanistic study. to maintain mild reaction conditions to conserve complex molecular structures of products, while providing appropriate reduction potential for the reductive CO 2 -functionalization. The Iwasawa group demonstrated a dual catalytic system with a Pd/Ir-photocouple for carboxylation reactions of aryl halides in the absence of metallic reducing reagents (Scheme 7). 55 Hünig's base (3 eq.) served as a sacricial electron donor in photoredox cycles, generating Pd(0)-complexes in the proposed catalytic carboxylation cycles (Scheme 10b). Although Ar-Pd(II)-Br(XPhos) possesses a high reduction potential (À2.28 V, vs. Fc/ Fc + ), a new peak at À1.4 V was observed from cyclic voltammetry (CV) measurements. The coordination of CO 2 on Pd might inuence the redox chemistry of the metal complex, therefore reducing the required reduction potential. In addition to the common insertion of CO 2 into the active Pd(II)-C bond, the authors suggested the formation of two intermediates, a Pd(I)or Pd(II)-CO 2 complex (Scheme 7b, path b). Aer methylation with TMSCHN 2 , various carboxylic acid esters were obtained including a sterically hindered acid (i.e. 2,4,5-triisopropyl carboxylic acid methyl ester).
Starting from simple feedstocks, Ar-Br and alkyl-Br, the König group reported visible light-induced carboxylation mediated by nickel catalysts (Scheme 8). 56 The plausible reaction mechanism could be divided into two distinct catalytic cycles. The rst one involved a one electron delivery to a Ni(II) or Ni(I) complex from the anion radical (4CzIPNc À ). Hantzsch ester (HEH, 2 equiv. required) was used as a terminal reducing agent in the presence of a reducing excited sensitizer (4Czlpn*) and light (le circle, Scheme 8b). Second, the oxidative addition to a Ni(0) complex was suggested, which undergoes reduction and then an insertion reaction with CO 2 (right circle, Scheme 8b). The catalytically active Ni(0) species can be regenerated from Ni(I) with electron sources produced from the le circle.
The same group expanded the dual catalysis strategy to the carboxylation of styrenes, 57 affording Markovnikov (branched) or anti-Markovnikov (linear) products selectively controlled by the choice of the ligand (Scheme 9). The suggested reaction mechanism explained that the observed chemoselectivity (branched/linear) was controlled by the different ligands (L6, neocuproine and L7, dppb). According to DFT calculations, Ni(0) species with the more sterically demanding ligand dppb (L7) tend to coordinate with CO 2 , forming a 5-membered nickelalactone (Ni-4) with styrene. With the less hindered neocuproine ligand (L6), the reaction proceeds via hydrometalation of styrene to afford Ni(II), which is subsequently reduced by the catalytic action of the photosensitizer (4CzIPN, Scheme 9b). The electrons generated from the photocatalytic cycle are used to reduce Ni(II) or Ni(I) to Ni(0), which can diverge to the hydrometalation step (le) and CO 2 activation step (right) to generate branched and linear products respectively while completing the catalytic cycles.
The Jamison group reported styrene functionalization reactions in a CO 2 atmosphere to generate b-aryl carboxylic acids (Scheme 10). 58 In this case PMP (1,2,2,6,6-pentamethylpiperidine) was employed as a sacricial organic electron donor, while utilizing water as an additive under modied reaction conditions. Although it is unclear, the addition of water induced high selectivity toward the mono-carboxylated product compared to other tested hydride or proton donors. The suggested reaction mechanism shows that the carboxylation with CO 2 c À results in the formation of a stabilized benzylic radical (E 0 ¼ from À1.82 to À0.71 V vs. SCE). Therefore, further reduction is feasible leading to the generation of carboxylated benzylic anion species, which could be protonated upon addition of water.
Photoactivation of organic substrates has been a successful transformation with high chemoselectivity to produce Markovnikov (branched) or anti-Markovnikov (linear) carboxylic acids. Also, the Murakami group reported a carboxylation reaction with a-alkyl ketones and CO 2 via a Norrish type II activation mechanism. 59 The carboxylation reactions at toluenyl carbon were also conducted in natural sunlight at ambient temperature with good isolated yields of the desired products. The authors suggested an energetically feasible [4 + 2]cycloaddition reaction by DFT calculations, which was determined by the thermal reaction of benzocyclobutenol to generate an o-quinodimethane intermediate.
Recently, Hou et al. reported carboxylation reactions of internal and terminal alkynes promoted by Co/Ir dual catalysis (Scheme 11). 60 The authors proposed that the reaction proceeded via functionalization of alkynes to generate an (E)-Co-CO 2 complex which is an intermediate for various productscarboxylic acids, pyrones, a,b-unsaturated g-lactones, coumarins, and 2-quinolones, by sharing a common intermediate, (E)-int (Scheme 11a). Pyrones were formed through a formal [2 + 2+ 2] cycloaddition with terminal alkynes (R 1 ]H). In the case of internal alkynes, pharmaceutically vital heterocycles such as coumarins and 2-quinolones were obtained with high selectivity. The suggested mechanism proceeded via intramolecular cyclization of acrylic acid intermediates. The E/Z isomerization of acrylic acid was conrmed by control experiments with (or without) the Ir-photoredox catalyst under irradiation (or in the dark). Also, this newly developed carboxylation/acyl-migration cascade reaction is feasible for alkyne difunctionalization, highlighting its utility in the eld of light-driven CO 2 -xation.
The Yu group reported the photocatalytic hydrocarboxylation of enamides and imines to afford a-amino acids with excellent chemo-and regio-selectivity (Scheme 12). 61 The pre-equilibrium of enamides and imines was combined with photocatalytic reduction. Despite the inherent nucleophilicity of enamides, kinetic studies indicated that the imines underwent the desired hydrocarboxylation faster than the competitive b-carboxylation reaction. The authors proposed an umpolung reaction of the a-amino carbanion under metalfree conditions. The carboxylated products were obtained with a broad substrate scope regardless of the electronic and steric properties of substituents. In addition, the enamide and imine starting materials were equally effective, conrming the fast pre-equilibrium before the reduction/ carboxylation steps.
Very recently, the Walsh group presented photocatalytic carboxylation of benzophenone-derived ketimines by employing an Ir-complex (Ir-I) under mild conditions (Scheme 13). 62 The radical anion was generated by single electron transfer (SET) from [Ir-I]* to ketimines, which was facilitated by the coordination between the imine and Cy 2 MeNc + . 63 Spin density calculation was carried out to evaluate the radical anions (A, B) suggesting that the carbon atom was more negatively charged than the nitrogen atom (spin density, radical probability on C: 0.05-0.18 and N: 0.37). 64 Subsequently, the more reactive Ncentered radical species abstracts a hydrogen atom from Cy 2 -MeNc e to form an a-amino carbanion and an iminium cation Scheme 10 Photocatalytic direct b-selective hydrocarboxylation of styrenes.
[Cy 2 N]CH 2 ] + via an umpolung reactivity (Scheme 13b). The carbanion then undergoes nucleophilic addition to CO 2 affording the desired carboxylation product. The obtained a,adisubstituted a-amino acid shows potential application of the protocol in asymmetric synthesis to generate quaternary stereogenic centers, which are oen difficult to control. 65 Direct carboxylation of imines and amines with CO 2 represents a very promising pathway to afford a-amino acids, especially those promoted by photoredox catalysts as shown above (Schemes 12, 13 and 17). Compared to tertiary amines, however, a-functionalization (i.e. a-carboxylation) of primary amines still remains a great challenge due to the lower reactivity of the a-C-H bond. Besides carboxylation reactions, CO 2 has been used as an activating group, 33,40 a directing group 66 and a protecting group 67 in organic synthesis. Ye et al. recently reported the photocatalytic a-alkylation of primary amines to yield g-lactams with CO 2 as a temporary activator and as a protecting group (Scheme 14). 68 Various a,b-unsaturated esters were tolerated in the presence of an Ir-II photosensitizer. Quinuclidine was employed as a sacricial electron donor. According to the suggested reaction mechanism, CO 2 was regenerated aer releasing lactam products via an intramolecular cyclization reaction. The in situ carbamate formation reaction suppressed the reactivity of primary amines while increasing the reactivity of a-C-H bonds according to the computational studies. The generation of the a-radical of the substrate is highly intriguing due to the potential applications toward various electrophiles and radical-radical coupling reactions. Furthermore, the use of tertiary amines as a base will enable a potential asymmetric catalysis to afford enantioenriched products.
This section summarizes recent progress in photo-CO 2functionalization without strong metallic reducing agents. Instead, an organic sacricial electron source or a reducing reagent was employed (i.e. triethylamine, piperidine, Hünig's base and Hantzsch esters) in the presence of photocatalysts with an appropriate reduction potential to complete the catalytic cycles. Various types of substrates underwent C-CO 2 bond formation reactions to provide unique molecular structures under ambient photosynthetic conditions (low CO 2 pressure, and accessible light sources). However, there is still plenty of room to develop more elegant methodologies in terms of sustainability. The next section will discuss redox-neutral carboxylation without external reductants.

Recent developments in redoxneutral CO 2 -functionalization
It is thought that catalytic carboxylation of non-activated organic substrates would be an ideal approach to CO 2 -utilization, avoiding reactive organometallic reagents (RMgX, RLi, R 2 Zn, R 4 Sn, etc.). For example, solar energy provides chemical reduction potential to enable CO 2 conversion in the Calvin cycle, where actual CO 2 -xation and C-CO 2 bond formation reaction occur under mild conditions via an a-ketol Scheme 14 Catalytic application of CO 2 for photocatalytic a-alkylation of primary amines. rearrangement (Fig. 2). 50,69 This "enantioselective" CO 2 -xation process generates a new C-C bond while creating additional stereogenic center(s) via a redox-neutral pathway. Accordingly, recent progress in photo-redox catalysis offers a promising platform to develop sustainable CO 2 utilization reactions under mild conditions in the absence of additional reducing reagents. 54,70 When combined with practicability and scalability, redox-neutral CO 2 -functionalization strategies will provide a tangible scenario of sustainable articial carbon xation.
The following examples in this section represent their redoxneutral reaction prole in terms of the proposed reaction mechanismsno terminal reducing or oxidizing reagents. Despite the fact that these reactions require activated substrates or radical initiators or a strong base, the generation of C-CO 2 bonds with CO 2 is a remarkable step toward truly ideal CO 2functionalization. Keeping in mind that solar energy might be the only and truly sustainable energy source, a few examples of redox-neutral photocatalytic CO 2 -functionalization reactions are also highlighted in this section.
The Sato group recently reported a direct carboxylation reaction at the allylic C(sp 3 )-H bond (Scheme 15). 71 The use of the AlMe 3non-nucleophilic basewas ascribed to the initial generation of catalytically active Co(I) species, therefore the catalytic cycle is free from an external reducing reagent. The carboxylation reaction of allylarenes and 1,4-dienes was proven to be effective with a nucleophilic h 1 -allyl-Co(I) catalyst aer intensive screening of transition metal catalysts such as Cr(II), Mn(I), Fe(III), Rh(I), Ir(I), Ni(II) and Cu(I). The role of the ligand was critical; Xantphos (L9) showed high selectivity without the formation of isomerization or methylation byproducts by the use of AlMe 3 . Various terminal alkenes were smoothly converted to b,g-unsaturated acids with excellent functional group tolerance, including amides, esters, and ketones. The authors suggested that the presence of the low-valent Co(I)-complex was the key to the successful carboxylation reaction with high selectivity. This protocol expands upon the scope of carboxylation to C(sp 3 )-H bonds, which represents atom-and stepeconomic approaches to construct molecular complexity by incorporating CO 2 .
Very recently, the Yu group reported photocatalytic carboxylation of tetraalkyl ammonium salts via C-N bond cleavage (Scheme 16). 72 Trimethylamine was generated in situ by singleelectron transfer (SET) from the excited Ir-I to the substrates. In turn, the resulting active Ir-I species could be reduced by the tertiary amine. Aerwards, carbanions undergo a carboxylation reaction aer another SET step between the excited photoredox catalyst and the alkyl radical. The authors suggested that the oxidized trimethylamine was transformed to amine species, like a-radical [Me 2 NCH 2 c], or dimethylamine aer hydrolysis. As electron donors, trimethylamine and dimethylamine accounted for 2 equivalents of reducing reagents required to complete the catalytic cycle. This built-in reductant was generated and demonstrated carboxylation reactions without additional reducing reagents, compared to Ni-catalyzed reductive carboxylation of benzylic C-N bonds. 31 In the above-mentioned cases, organic amines act as sacri-cial electron donors, where the resulting radical cation trialkyl amines have dramatically reduced pK a at the a-protons. 73 In the presence of a base, a deprotonation reaction would generate an amine with an a-radical, which can couple with other reactive species. The single-electron reduction of CO 2 to CO 2 c À is in general a rate-determining step due to the high reduction potential (À2.21 V vs. SCE (saturated calomel electrode) in DMF (N,N-dimethylformamide)). 74 A viable C-C bond formation Scheme 15 Cobalt-catalyzed direct carboxylation of allylic C(sp 3 )-H bonds.
reaction with CO 2 c À and amine based a-radicals would afford aamino acids as the product. This was realized by the Jamison group demonstrating a metal-free photoredox conversion of CO 2 (Scheme 17). 54 An organic sensitizer, p-terphenyl, mediated single electron transfer reactions (reduction potential: À2.63 V vs. SCE in DMF) to perform the suggested one-electron reduction of CO 2 , providing a-amino acids in the absence of additional reducing reagents. Various aryl-substituted a-amino acids were prepared in good to excellent yields. The convenience of continuous ow chemistry 75 was an added benet of the photocatalysis to provide essential synthetic building blocks from carbon dioxide. The generation of CO 2 -radical anion is highly attractive, considering its vast application potential in organic synthesis for carboxylation reactions. This photocatalysis mediated by p-terphenyl showed promise toward metal-free CO 2 -functionalization via a single-electron reduction mechanism in terms of atom-economy (redox-neutral), feasibility (continuous ow setups), and utility of the nal products (a-amino acids) containing stereogenic centers.
Owing to the recent developments in organic photosynthesis and photosensitizers, 22,76 unprecedented reactivity patterns were achieved with CO 2 as a C1 source. For example, the Martin group showed photocatalytic dicarbofunctionalization of styrene derivatives initiated by radicals under mild reaction conditions, where stabilized benzyl carbanions react with CO 2 (Scheme 18). 70a Various radical initiators, such as triuoro-and diuorosulfonates, and triuoroborate salts, were proven to be effective under photochemical reaction conditions. The photocatalytic redox cycle was mediated by an Ir-complex (Ir-II). This protocol provides two new C-C bonds with a stereogenic center in the absence of additional stoichiometric reducing reagents. Trisubstituted alkenes were also employed to afford carboxylic acids with a quaternary stereogenic center. The convenient introduction of the (di)triuoromethyl group highlights potential applications of radical carboxylation reactions in drug discovery and pharmaceutical industry. 77 The Yu group developed the rst thiocarboxylation of styrenes by using an Fe/S complex as the photosensitizer (Scheme 19). 70b Various b-thioacids were synthesized selectively with different regioselectivities from the previous protocol (Scheme 18). Mechanistic studies revealed that single-electron reduction of CO 2 can be initiated by the excited Fe/S complex, yielding the CO 2 radical anion (CO 2 c À ). This radical intermediate was trapped subsequently by an alkene substrate to generate a stabilized alkyl radical, which led to anti-Markovnikov regioselectivity. Thiolation of alkyl radicals was mediated by the [Fe/S] radical cation, highlighting the application potential of the methodology in the synthesis of b-thioacids an intermediate for the antidepressant drug thiazensim. 78 Also, considering the Fe-and S-rich environment in the prebiotic era, the presented reaction could help us to rethink the CO 2 chemistry in the primordial soup, potentially affording complicated photoredox reactions with CO 2 to furnish chiral molecules.
The progress in redox-neutral CO 2 functionalization showed elegant reaction mechanisms operating under mild conditions, for example, via CO 2 insertion into metal-carbon bonds or CO 2 c À captured by activated substrates. This represents a promising and ideal mode of action, whereby no additional sacricial redox agents were applied to construct multiple C-C, and C-X bonds. Thus, high atom economy and step-efficiency are expected in constructing molecules with CO 2 as a nontoxic C1 source, boosting research in CO 2 -utilization from recently developed dicarbofunctionalization. 79 Meanwhile, the structural diversity of recent CO 2 -functionalization reactions shows the signicant potential of CO 2 in asymmetric synthesis and catalysis. Further investigations on the asymmetric activation of CO 2 and its utilization in CO 2 -functionalization will allow us to achieve higher values of products while recycling CO 2 .

Asymmetric catalytic carboxylation with CO 2
Besides the asymmetric synthesis of cyclic carbonates or polycarbonates with epoxides or diols, 21b,80 the construction of enantioselective C-CO 2 bonds using CO 2 has been a formidable challenge under the inuence of chiral catalysts or chiral environments. This is due to the high stability of CO 2 , limiting the scope of reaction partners; highly reactive organometallic species and/or harsh reaction conditions are necessary thus low stereoselectivity is in general expected. 21a In 2004, the Mori group reported the carboxylative cyclization reaction of bis-1,3dienes catalyzed by a Ni catalyst (Scheme 20). 81 The authors performed facile 5-membered ring formation reactions in the presence of excess amounts of dialkyl zinc (4.5 equiv.). The obtained products possess three consecutive stereogenic centers with absolute diastereoselectivity with good yield and excellent enantioselectivity.
In 2017, the Marek group developed an enantioselective Cucatalyzed carbomagnesiation reaction of cyclopropenes, which could be selectively carboxylated with CO 2 as an electrophile (Scheme 21). 82 High diastereoselectivity was observed which is not fully understood yet based on the control experiment without the copper catalyst (racemic but moderate diastereoselectivity, 9 : 1 dr). Other electrophiles such as iodine, bromine and allylbromide were smoothly incorporated to furnish the desired products. Although Grignard reagents are reactive nucleophiles, the sequential addition of the alkene and CO 2 prevented direct attack of these nucleophiles on CO 2 at low reaction temperature (0 C) in the presence of a copper catalyst. The observed stereoselectivity was attributed to the stability of the stereogenic center at the carbon-Cu moiety, explaining the cis geometry between the nucleophile and electrophilic CO 2 .
The Yu group 83 recently reported a highly regio-and enantioselective copper-catalyzed CO 2 -functionalization reaction of olens owing to enantioselective Cu-H catalysis 84 (Scheme 22). Inspired by the CO 2 reduction reaction to methanol 85 and other higher alcohols, 86 the authors developed the sequential enantioselective Cu-H addition, carboxylation and reduction reactions to achieve hydroxymethylation of olens. A preliminary mechanistic study revealed that the L12 Cu-R (C) species showed no reactivity toward reduced CO 2 (R 3 Si-OCOH) (dashed arrow), indicating the direct carboxylation of C in the chiral environment to ensure the obtained high enantioselectivity. Furthermore, the developed methodology was applied to 1,3dienes, affording (Z)-selective homoallylic alcohols with good enantioselectivities. Further derivatization of the hydroxymethylation products afforded elegant syntheses of enantioenriched (R)-(À)-curcumene 87 and (S)-(+)-ibuprofen, starting from CO 2 as a C1 building block.
Although asymmetric catalytic C-C bond formation has achieved relatively considerable progress, 88 only a few methodologies have been reported with CO 2 as a sustainable C1 source while creating stereogenic center(s) with high stereoselectivity.
This journal is © The Royal Society of Chemistry 2019 Considering that the carbon xation process produces carbohydrates and biomass with absolute enantioselectivity, it is a logical extension to implement asymmetric carboxylation reactions in articial CO 2 xation. Chemical synthesis offers various synthetic pathways and tools that can be easily tested, potentially providing a playground for facile screening and method development. For example, photochemical reactions with chiral catalysts including a chiral iridium catalyst 89 or Lewis-acid assisted photocatalysis 90 for CO 2 -functionalization are seemingly feasible methods to be developed. Considering the mode of action of RubisCo enzyme, redox-inactive metals and ligands (e.g. Mg-biotin complex) would be critical to improve the availability of CO 2 in organic reactions. 91 On the other hand, it could be inferred that chiral CO 2 -complexes may play a signicant role in CO 2 -activation via bifunctional asymmetric catalysts. 92 It would be exciting to see the development of CO 2 -functionalization, with foreseeable sustainability and increased utility of the nal products in organic synthesis.

Conclusion and outlook: umpolung reactivities towards CO 2
It is a formidable challenge to dene an "ideal" carbon dioxide functionalization considering that many factorsenvironmental impact, atom-economy, sustainability, utility of products, and reaction conditionsare involved in designing reaction processes. Harnessing the full capacity of CO 2 -functionalization can be envisaged with sustainable and accessible chemical feed stocks, catalysts, and reaction conditions. Victor Grignard, in 1912, stated this in his Nobel Lecture -"Willstatter in fact recognized that .organic magnesium compounds must form and that the absorption of CO 2 gas by chlorophyll would in every way be comparable to the Grignard reaction". The mode of action of magnesium compounds in chlorophyll differs from what Grignard speculated, however, one of the earliest umpolung reactions with CO 2 and Grignard reagents paved a way for modern CO 2 functionalization to date. Considering the formation of Grignard reagents, an umpolung process utilized polarized bonds, C d+ -X dÀ , by inverting the electronic nature of the carbon to nucleophilic by forming C dÀ -Mg-X.
In this context, recent developments in umpolung carboxylation reactions have shown unprecedented reaction patterns (Scheme 23): 61,62,93 for example, umpolung reactivity has been implemented to functionalize CO 2 for an aldehyde carboxylation reaction through a redox-neutral mechanism (Scheme 23a). 93b The obtained product, a-keto acid, was smoothly converted to the corresponding a-amino acid under reductive amination reaction conditions mimicking the biosynthesis of various amino acids. The use of nitrogen-containing nucleophiles offers direct synthesis of a-amino acid derivatives (Scheme 23c-e). By employing cyanohydrin, hydrazone, photocatalysts, and a base, in situ generated umpolung species were transformed to the desired carboxylated products under mild reaction conditions, with or without reducing agents. This is particularly interesting to hypothesize the evolution of a-amino acids from the CO 2 -rich prebiotic environment. It has been postulated that cyanide is an abundant source of a carbon nucleophile in the synthesis of biologically active molecules in the primordial soup. 94 The use of aldehydes, cyanide, and CO 2 in synthesizing biologically active molecules is a promising step toward answering the important question: what is the origin of life? Was there an involvement of photochemical CO 2 -activation? Was it promoted by an optically pure component to induce homochirality? The forthcoming ideal CO 2 -functionalization may answer these conundrums.
In summary, this Perspective collects the recent literature in CO 2 functionalization and groups it into four categories: (1) metal catalyzed direct carboxylation, (2) photocatalytic carboxylation reactions, (3) redox-neutral carboxylation, and (4) asymmetric introduction of catalytic CO 2 for C-C bond formation reactions. Even a broad scope of substrates and remote site functionalization were achieved; transition metal-catalyzed reductive carboxylation is mostly limited to CO 2 insertion reaction with (over)stoichiometric amounts of reducing reagents. However, photoredox catalysts present promising access to more diversied CO 2 reactions, like dicarbofunctionalization, single-electron reduction and radical coupling via a redox-neutral mechanism. Thanks to these developments of methodologies, as discussed at the end of Section 5, more examples in challenging enantioselective C-CO 2 bond formation will be realized in the near future. Although enzymatic CO 2 functionalization reactions are not covered in this Perspective, they have shown their very promising application in articial carbon recycling processes. 51,95 Synergetic and interdisciplinary CO 2 xation with biological and chemical catalysts will be particularly interesting in (asymmetric) photocatalytic conversion of CO 2 , truly mimicking photosynthesis to provide ideal CO 2 functionalization reactions.

Conflicts of interest
There are no conicts to declare.