Molybdenum-Containing Metalloenzymes and Synthetic Catalysts for Conversion of Small Molecules

Abstract

The energy deficiency and environmental problems have motivated researchers to develop energy conversion systems into a sustainable pathway, and the development of catalysts holds the center of the research endeavors. Natural catalysts such as metalloenzymes have maintained energy cycles on Earth, thus proving themselves the optimal catalysts. In the previous research results, the structural and functional analogs of enzymes and nano-sized electrocatalysts have shown promising activities in energy conversion reactions. Mo ion plays essential roles in natural and artificial catalysts, and the unique electrochemical properties render its versatile utilization as an electrocatalyst. In this review paper, we show the current understandings of the Mo-enzyme active sites and the recent advances in the synthesis of Mo-catalysts aiming for high-performing catalysts.

1. Introduction

Fixation of atmospheric nitrogen to ammonia has been promoted by the Haber–Bosch process, and the capacitated mass production of foods accelerated the increase of the world population [1,2]. The rapid growth of fossil fuel-based industry unavoidably increased atmospheric CO₂, and resultantly aggravated the global warming and climate crisis. To suggest solutions for those anthropogenic issues and mitigate the current environmental problems, scientific progress to convert fossil fuel-based energy systems toward sustainable ways is keenly desired along with other social and economic efforts [3]. In the line of scientific cognition, we review catalysts for energy conversion reactions including small molecules such as CO₂, N₂, H₂, and O₂ as reactants or products. Numerous catalysts have been developed in recent decades owing to the high interest in energy conversion reactions [4,5,6,7,8]. Mo ion is found to be an essential component in biological catalysis [9], as well as a very promising metal candidate among Earth-abundant metals for applications to electrocatalysts and photocatalysts [4,10,11]. Herein, we review versatile utilization of Mo ions in natural and synthetic catalysts.

Mo-containing homogeneous catalysts utilize both low- and high-valent Mo ions, whereas metalloenzymes and their model complexes have high-valent Mo centers. Except for FeMo-nitrogenase [12], Mo-enzyme active sites always use dithiolene ligation of the molybdopterin (MPT) moiety to stabilize a Mo(IV/VI) ion during the redox reaction and to assist electron transfer between metal and ligand [13]. Moreover, the chalcogenide ligand and nearby amino acid play essential roles to determine the selective reactivity of Mo active sites. Although enzyme mechanisms remain unclear despite significant interests, ligand identity and coordination structures seem to be critical factors to determine catalytic activity. We can propose synthetic routes for the design of new Mo-based catalysts from comparing structures, properties, and reactivities of metalloenzymes and artificial catalysts [14]. Although relatively less studied, Mo-clusters have shown promising electrocatalytic activities due to their intermediate behaviors between metal-coordination complexes and nanoparticle catalysts [15]. In the perspective of developing efficient electrocatalysts, Mo-nanomaterials, which have shown remarkable electrocatalytic activities, should be considered together. Preparation methods of Mo-nanomaterials determine their catalytic activities, and proximal atoms and electronic conditions around Mo ions are closely related to reactivities of surficial Mo ions [16,17].

2. CO₂ Electrocatalysts

Considering current energy systems, a drastic diminution of fossil fuel consumption seems unrealistic. However, continuous research for recycling anthropogenic CO₂ will be able to provide feasible solutions to current energy and climate issues. Among various CO₂ transformation methods, electrochemical conversion of CO₂ may be an attractive way to be conjugated with solar/electric energy conversion reactions [18]. Mo-enzymes of the formate dehydrogenases [19] (Mo-FDH) and the carbon monoxide dehydrogenases [20] (MoCu-CODH) are the most efficient CO₂-catalysts on Earth, and their synthetic models are known, albeit short of the enzyme activity [21]. Moreover, many homogeneous and heterogeneous catalysts have been developed for CO₂ conversion reactions.

2.1. Metalloenzymes with CO₂ Reactivity

2.1.1. Formate Dehydrogenases

Formate dehydrogenases (FDHs) is a group of metalloenzymes that catalyze a reversible conversion of formate to CO₂. The CO₂ reactivity of the FDH active site has obtained research interests because the enzyme functions near the potential of −420 mV vs. NHE (Normal hydrogen electrode) scarcely requiring overpotential. It is known that the reaction rate is varied by bacteria. Rhodobacter capsulatus FDH facilitates oxidation of formate to CO₂ at a rate of 36.5 s⁻¹ and the reverse direction at a rate of 1.6 s⁻¹ [22]. Desulfovibrio desulfuricans FDH oxidized formate much more quickly at 543 s⁻¹ and reduced CO₂ at 46.6 s⁻¹ [23]. The FDH reactivity is directly related to the active site structure, which has a Mo center surrounded by two pyranopterin-conjugated dithiolene ligand, a chalcogenide (hydroxide or sulfide) ligand, and SeCys amino acid as a sixth ligand (Figure 1). In addition, additional roles of adjacent amino acids such as Arg and His seem to be important to assist the FDH reaction by stabilizing reaction intermediates.

$${\mathrm{CO}}_2+2{\mathrm{e}}^{-}+{\mathrm{H}}^{+}\leftrightarrows {\mathrm{HCOO}}^{-},{\mathrm{E}}^{°\prime }=-420\mathrm{mV}$$

In 1997, Boyington and coworkers first reported the crystal structure of E. coli FDH [24]. Later, in 2006, Raaijmakers and coworkers reinterpreted the crystal structure and proposed a reaction pathway of the FDH active site (Scheme 1) [25]. They suggested that SeCys is reversibly bound to the Mo center during the redox reactions so that offers a vacant site to interact with substrates. Once formate bound to Mo⁶⁺ site, a proton migrated to His residue, and CO₂ was released along with the reduction of Mo^6+/4+ site. If two electrons were transferred to [4Fe4S] cluster, SeCys recombined at the Mo site. In the proposed reaction, the sulfide ligand did not participate in the reaction.

Another reaction pathway including the role of sulfide ligand has been proposed (Scheme 2). In 2011, Moura and coworkers suggested that the sulfide ligand assists the reversible dissociation/association of Mo-SeCys bond on the basis of theoretical studies. A probable Mo-S-SeCys intermediate could open a coordination site to interact with formate. Once a proton migrated from formate to SeCys, a thiocarbonate intermediate was formed with the dissociation of the S-SeCys bond, and adjacent Arg⁺ amino acid possibly stabilized the negatively charged intermediate. The roles of nearby amino acids seemed to be critical for the catalytic reaction at the active site [26,27].

The same group proposed that a hydride transfer occurred in the conversion of Mo(VI) = X and Mo(IV)-XH, and the formate oxidation would likely occur as an indirect pathway [23]. In a substitution experiment of Mo-bound formate for isoelectronic azide, the azide entered between Arg/His and Mo=S, not at the binding site of SeCys (Scheme 3).

Haumann and coworkers reported that the Mo=S bond length of the oxidized form is ~2.16 Å on the basis of X-ray absorption spectroscopy [28].

They suggested that Mo-SH possibly exists based on the Mo-S distance of ~2.4 Å in the X-ray crystal structure and SeCys stays apart from the Mo center during the reaction. They suggested a direct interaction of formate with Mo center based on X-ray absorption spectroscopy. When azide entered in the Mo-S_Cys site of the oxidized form, the Mo-Cys bond distance decreased. Conversely, when azide was removed, the Mo-Cys distance increased [29].

The reaction mechanism of the FDH active site is a continuous research topic because an in-depth understanding of the active site will lead to the development of efficient electrocatalysts for CO₂ conversion.

2.1.2. MoCu-Carbon Monoxide Dehydrogenases

In 2002, MoCu-carbon monoxide dehydrogenase (CODH) from Oligotopha carboxidovorans was found to catalyze the oxidation of CO to CO₂ under aerobic conditions. The enzyme exists as a dimer form of heterotrimer [30]. The hetero-trimer is composed of 88.7 kDa L (809 residues), 30.2 kDa M (288 residues), and 17.8 kDa S (166 residues). The L subunit contains molybdopterin-cytosine dinucleotide (MCD), and the M subunit has a flavin adenine dinucleotide (FAD) cofactor. Two [2Fe-2S] clusters stay in the S subunit [31]. As characterized by X-ray crystallography and extended X-ray absorption fine structure (EXAFS), the MoCu-CODH active site is composed of a hetero-bimetallic cluster of (MCD)Mo^VIO(μ²-S)-Cu^I(S-CYS), and the bent μ²-S bridge connects the Cu and Mo center (Figure 2a) [32]. The resting state of the active site was characterized as a Mo^VI(O)₂SCu^I core, and the catalytic oxidation possibly occurs with CO binding at the Cu center.

In 2002, Dobbek and coworkers reported the X-ray structure of the MoCu-CODH with N-butylisocyanide, where the N-butylisocyanide entered between µ-S–Cu bond by cleaving the bridging sulfide bond [31]. Accordingly, they suggested that CO would interact with the active site in a similar way as the isoelectronic molecule as forming a µ-thiocarbamate intermediate (Figure 2b).

However, Siegbahn and coworker suggested that CO binding to Cu is energetically favored from the computational studies. Next, the Cu-CO likely experiences a nucleophilic attack by the equatorial Mo=O to form a thiocarbamate intermediate, and CO₂ can be released with coordination by H₂O (or –OH) (Figure 2c) [33].

The formation of a thiocarbamate intermediate was suggested previously, but Hofmann and coworkers claimed that a thiocarbamate intermediate is not relevant to the MoCu-CODH catalytic cycle, because a thiocarbamate moiety is too stable to be involved in the catalytic steps (Figure 2d) [34]. Hille and coworkers also proposed that the Cu-CO-O-Mo bond is formed after CO insertion (Figure 2e). They emphasized the role of nearby Glu763 residue to deprotonate H₂O. A nucleophilic attack of –OH to the labile equatorial Mo=O could promote the release of CO₂ [30] The aerobic MoCu-CODH catalyzed CO oxidation with k_cat 93.3 s⁻¹ at pH 7.2 condition [35].

2.2. Structural Analogs of CO₂ Enzymes

2.2.1. FDH Analogs

The high efficiency of the FDH enzyme spurs researchers to synthesize structural analogs of the enzyme active site. Synthetic procedures for various dithiolene complexes enabled modeling studies of the enzyme active site having the Mo-bis(dithiolene) moiety. High-valent Mo(IV-VI)-bis(dithiolene) appeared to be highly reactive. Although the Mo enzyme active site is protected by surrounding proteins, the air-sensitivity of high-valent Mo in solution conditions makes the modeling study difficult. On the other hand, the high reactivity could be a clue to the CO₂ reactivity of the Mo site. Previous studies have revealed that the selective reactivity of Mo-bis(dithiolene) complexes is related to auxiliary ligands.

The Mo-FDH and W-FDH active sites have very similar coordination environments except for the metal center, although their reaction conditions are different because of the bacterial habitat. In 2012, Kim and Seo reported CO₂ reactivity of the in-situ generated [W^IV(OH)(S₂C₂Ph₂)₂]⁻ (2) [36] as an example of the functional model of the W-FDH active site (Scheme 4) [25]. Complex 2 generated by hydrolysis of [W^IV(OPh)(S₂C₂Ph₂)₂]⁻ (1) [37,38] showed the CO₂ reactivity at mild condition, and it was suggested that a nucleophilic attack of –OH to CO₂ formed a W-(bi)carbonate intermediate. Although the model complex did not show a catalytic activity, the plausible formation of a W-(bi)carbonate intermediate was similar to the reactivity of the FDH active site [26]. They also reported a stoichiometric reduction of CO₂ to formate by [W^IVO(S₂C₂Me₂)₂]²⁻ (3) at 90 °C [39]. A W-carbonate intermediate was suggested as an intermediate before the CO₂ reduction, but the oxidized W complex was dimerized to [W^V₂O₂(μ-S)(μ-S₂C₂Me₂)(S₂C₂Me₂)₂]²⁻ (4) at the high temperature as frustrating catalytic reaction (Scheme 4).

Fontecave and coworkers reported catalytic CO₂ reduction by Ni-bis(dithiolene) complex using a dithiolene derivative of quinoxaline-pyran-fused dithiolene (qpdt²⁻), which is similar to the molybdopterin (MPT) structure of the FDH active site (Scheme 4) [40]. In 2018, they used the qpdt²⁻ ligand to prepare Mo complexes, Mo^IVO(qpdt)₂ (5), [Mo^IVO(H-qpdt)₂]²⁻ (6), and [Mo^VO(2H-qpdt)₂]⁻ (7), as the Mo-FDH models, and examined the photocatalytic CO₂ reduction activity [41]. The reduced forms of H-qpdt and 2H-qpdt had similar oxidation states with the MPT. Using [Ru(bpy)₃]²⁺ photosensitizer, 1,3-dimethyl-2-phenyl-2,3-dihydro-1H-benzoimidazole as a sacrificial electron donor in the solution of triethanolamine and acetonitrile (1:5), and irradiation with a 300 W Xe lamp on the Mo complex solutions afforded CO₂-derivated products (CO and HCO₂H) and H₂. Complex 5 gave the CO₂-reduction products with 19% and mostly H₂ (81%). However, complex 6 increased the CO₂-reduction yield to 47%, and complex 7 provided the CO₂-reduction products in the highest yield of 58% with over 100 turnover number (TON) for 68 h.

Structural and functional analogs of the FDH active sites have been reported with rare examples; however, the catalytic activities remain much less than the enzymes. Secondary coordination ligands mimicking amino acids surrounding the enzyme active site would give new catalyst designs to achieve similar activities as the enzyme.

2.2.2. MoCu-CODH Analogs

In 2020, Mougel and coworkers synthesized [(bdt)Mo(O)S₂CuCN]²⁻ (8) (bdt = benzenedithiolate) as a MoCu-CODH model complex (Scheme 5) [42]. The Mo and Cu were bridged by μ²-sulfide ligand as the enzyme active site, and a bdt ligand was used to model the MPT. Complex 8 showed a catalytic reduction peak at E_1/2 = −2.07 V vs. Fc⁺/Fc in CO₂-saturated acetonitrile. Controlled potential electrolysis (CPE) at −2.62 V vs. Fc⁺/Fc in the presence of 0.1 M 2,2,2-trifluoroethanol (TFE) produced the reaction products of formate (Faradaic efficiency (FE) = 69%), CO (8%) and H₂ (19%). It was a rare example to describe the MoCu-CODH model with the CO₂ reduction activity, although the MoCu-CODH enzyme previously showed only the CO oxidation activity [20].

2.3. Synthetic Electrocatalysts for CO₂ Reduction

2.3.1. Homogeneous Mo Complexes for CO₂ Reduction

In 1978, Reichert and coworkers reported the CO₂ reactivity of a dinuclear Mo₂(O^tBu)₆ (9) [43]. Two CO₂ molecules were reversibly inserted into the Mo-O^tBu bonds to form Mo₂(O₂CO^tBu)₂(O^tBu)₄ (Scheme 6) (10). From the X-ray structure, the Mo-Mo triple bond, bridged by two O₂CO^tBu, was measured to be 2.241 Å. The CO₂ insertion occurred in both the solid and liquid states of the complex, and sublimation of the CO₂ adduct at 100 °C afforded back the complex 9 by releasing CO₂.

Metal ligand cooperation (MLC) is a useful method to fixate CO₂. The trans-Mo(C₂H₄)₂(PMe₃)₄ (11) reacted with CO₂ under mild conditions [44], where CO₂ was inserted into Mo-ethylene bond to form a Mo-acrylate moiety in [Mo(H₂CCHCO₂H)(C₂H₄)(PMe₃)₂]₂ (Scheme 7) (12). The acrylate frequency was measured as 1500 cm⁻¹ and the ¹³C NMR (Nuclear magnetic resonance) chemical shift at 175–180 ppm. Complex 11 reacted with a stoichiometric amount of CO₂ under 1 atm to give 12, which could be converted back to 11 by treatment of n-BuLi under H₂ with concomitant LiO₂CCH₂CH₃ product. In another MLC type reaction, a PNP-amide-assisted CO₂ fixation was shown with Mo(NO)(CO)(PNP) (13) (PNP = N(CH₂CH₂PiPr₂)₂) [45]. Complex 13 converted stoichiometric amount of CO₂ to formate in the presence of Na[N(SiMe₃)₂] and additional base of DBU, Et₃N (Scheme 7). However, high pressure of CO₂ (10 bar) and H₂ (70 bar) was also required. The MLC method effectively lowered the binding energy of CO₂ to Mo complex, but the formation of a highly stable ML-CO₂ adduct rather prevented the catalytic cycle.

Lau and coworkers utilized hetero-bimetallic systems of early/late transition metal ions for electrocatalytic reactions. They synthesized (η⁵-C₅H₅)Ru(CO)(μ-dppm)Mo(CO)₂(η⁵-C₅H₅) (14) (Scheme 8) from a reaction of (η⁵-C₅H₅)Ru(dppm)Cl and Na[(η⁵-C₅H₅)Mo(CO)₃] [46]. Complex 14 promoted catalytic CO₂ reduction to formic acid at 120 °C with 43 TON per catalyst for 45 h in benzene containing triethylamine base. Interestingly, the same complex also promoted converse conversion of formic acid to CO₂ and H₂ at 80 °C. They suggested a hydride-bridged bimetallic species is formed as a reaction intermediate on the basis of a hydride detection at −17.16 ppm by ¹H NMR spectroscopy.

Minato and coworkers synthesized a series of Mo-silyl hydrido complexes of [MoH₃([Ph₂PCH₂CH₂P(Ph)-C₆H₄-o]₂(R)Si-P,P,P,P,Si)] (R = Ph (15a), C₆F₅ (15b), 4-Me₂NC₆H₄ (15c), cyclohexyl (15d), and n-C₆H₁₃ (15e)) [47]. Silyl ligands with a strong trans effect were employed to control the reactivity of the Mo center (Scheme 9) [48,49]. The RSi-Mo could make an open coordination site at trans position to allow binding of electrophilic molecules such as O₂, CO₂, and carboxylic acid. Under the reaction conditions of 2.0 M of dimethylamine, 30 atm of CO₂, and 20 atm of H₂, complexes 15a–e catalyzed the CO₂ hydration as producing N,N-dimethylformamide (DMF) and H₂O. The catalytic conversion of CO₂ to DMF could be controlled by the Si-substituent. The complex 15a with a Si–phenyl moiety achieved 92 TON per catalyst, but complex 15b with Si–C₆F₅ showed only 1 TON. Electron donating substituent increased TON of complexes as seen with 15c (TON = 130) and 15e (110), because the stronger trans effect assisted binding of substrates to increase the catalytic reactivity.

Redox-active ligand could assist central metal ions for reductive CO₂ reactions. Trovitch and coworkers have used bis(imino)pyridine ligand for hydrosilylations of ketones [50]. They synthesized (κ⁶-P,N,N,N,C,P-^Ph2PPrPDI)-MoH (17) (PDI = pyridine diamine) by a reduction of ((^Ph2PPrPDI)MoI)I (16) with excess K/Hg (Scheme 10) [51]. Complex 17 reacted with CO₂ under 0.2 atm of CO₂ condition, where CO₂ was inserted into Mo-H bond to form η¹-formate-bound Mo(OC(O)H) species [52,53]. The Mo-formate ¹³C NMR resonance appeared at 169.7 ppm and the infrared (IR) frequency was detected at 1625 cm⁻¹. Complex 17 also promoted a catalytic conversion of HBPin to H₃COBPin and O(BPin)₂ under 1 atm of CO₂ as recording 40.4 h⁻¹ turnover frequency (TOF) and 323 TON with consuming 97% of HBPin. Distillation of H₃COBPin in water for 24 h generated MeOH, eventually producing 58% of MeOH starting from HBPin. They showed a potential pathway to convert CO₂ to methanol. Kubiak and coworkers used bipyridine (bpy) ligands to prepare Mo-carbonyl complexes, Mo(bpy)(CO)₄ (18a) and Mo(bpy-tBu)(CO)₄ (18b) [54]. Complexes 18a and 18b showed electrocatalytic CO₂ reduction activity in acetonitrile solution [55]. Complex 18a had a reversible redox couple at −1.58 V vs. SCE (saturated calomel electrode) and an irreversible wave at −2.14 V vs. SCE, and catalyzed CO₂ reduction near the second reduction potential. Complex 18b shifted the reduction potential a bit negatively to −2.20 V vs. SCE. The addition of 1.6 M of TFE into the reaction solutions generated only CO product as recording 1.0 s⁻¹ TOF with 18a and 1.9 s⁻¹ with 18b.

Bernskoetter and coworkers used tertiary amine pincer type ligands to prepare CO₂ hydrogenation catalysts [56]. Reduction of (PN^MeP)MoCl₃ with Na(Hg) under ethylene afforded (PN^MeP)Mo(C₂H₄)₂ (19), and the following oxidative C-H addition generated the Mo-H species of (PN^CH2P)MoH(C₂H₄)₂ (20) (Scheme 11) [57]. Blowing 4 equiv. of CO₂ into complex 20 gave the Mo-O₂CH moiety in complex 21. Catalytic production of formate was achieved by using LiOTF as cocatalyst and DBU base under 69 atm of H₂ and CO₂ (1:1 ratio) at 100 °C as recording 35 TON for 16 h.

The low-valent Mo complexes have shown promising activities for CO₂, but the formation of relatively stable adducts or requirement of high-pressure conditions remain issues to solve for efficient catalyst development.

2.3.2. Heterogeneous Mo-Containing CO₂ Reduction Electrocatalysts

In 1986, Frese and coworkers reported the reduction of CO₂ to methanol using Mo electrode at the potential range of −0.57 to −0.67 V vs. SCE in CO₂-saturated aqueous media (0.05 M H₂SO₄, pH 4.2) [58]. As a result, 18 μmol of methanol was obtained for 40 h with 50% FE together with 0.044 μmol of CO production.

MoS₂ surface has shown CO₂ reduction ability, but in general, the intrinsic properties such as poor electrical conductivity and few active sites cause low electrocatalytic activity. In 2014, Asadi and coworkers reported the layer-stacked MoS₂ nanomaterial [59]. The MoS₂ surface conducted selective reduction of CO₂ to CO in the presence of 4% 1-ethyl-3-methylimidazolium tetrafluoroborate (EMIM-BF₄) with a high reduction current of 65 mA/cm² at −0.76 V vs. RHE (FE ~98%). The activity at a low overpotential (0.1 V) was measured to be 25 times higher than Au nanoparticles. They reported vertically aligned (VA) MoS₂ had higher CO₂ reactivity than bulk MoS₂. The high catalytic performance was due to the increased MoS₂ edge site. In 2017, Salehi-Khojin and coworkers synthesized VA-MoS₂ with ~20 nm thickness by chemical vapor diffusion (CVD) method, and they also studied the metal-doping effect for the CO₂ reduction activity [60]. The 5% Nb-doped VA-MoS₂ began the CO₂ reduction with onset overpotential of 31 mV, and it reached 237 mA/cm² current density at −0.8 V vs. RHE. However, Ta-doped (3~18% Ta) VA-MoS₂ showed 98–68 mA/cm² at −0.8 V vs. RHE, which was rather lower than a pristine VA-MoS₂ (121 mA/cm²).

Previously, Bi-deposited glassy carbon has shown a good catalytic activity for reduction of CO₂ to CO because CO₂^•− radical species are well stabilized on Bi site [61]. Interestingly, a formation of hetero-bimetallic MoBiS_x (Mo/Bi 1:1) nanosheet showed different CO₂ reduction product of methanol (FE 71.2%) with generation of side products of methane (7.8%) CO (9.6%) H₂ (11.4%) [62]. It was observed that, if the Bi ratio was raised, the CO percentage increased. They explained that the further reduction process is probably promoted by the synergistic effect with MoS_x–catalytically active for the proton reduction.

Nano-sized Cu particles have shown CO₂-to-hydrocarbon electrocatalytic reduction activity [63]. The Cu-doping effect on the CO₂ reactivity of MoS₂ was studied by Yu and coworkers [64]. Flower-like MoS₂ was prepared by hydrothermal method, and the surface was reacted with Cu(NO₃)₂·3H₂O to give Cu/MoS₂ composite with varying the Cu/Mo ratio. Specific Cu ratio (12.76%) exhibited four times higher catalytic current density at −1.7 V vs. SCE than a bare MoS₂ as producing CO (FE = 35.19%), CH₄ (17.08%), C₂H₄ (2.93%). The high Faradaic efficiency for CO₂ reduction was because the surface-deposited Cu enhanced the electronic conductivity and CO₂ adsorption ability of MoS₂.

Few active sites have been an issue for the preparation of MoS₂ electrodes, thus, synthetic methods are pursued to enlarge the MoS₂ edge site area. Wang and coworkers used zeolitic imidazolate frameworks to prepare hollow MoS₂ nanostructure, which effectively enlarged the exposed MoS₂ edge area. The edge-exposed MoS₂ was supported by N-doped carbon to improve electron transfer, which gave two times higher current density (34.31 mA/cm² at −0.7 V vs. RHE) and 1.4 times higher FE (CO₂-to-CO 93% FE) compared to MoS₂ [65]. The N-doped carbon served as efficient conductive support, because the more electronegative N atom decreases electron density at nearby carbon sites by improving electron transfer at the electrode surface. An enhanced surface electron transfer to MoS₂ edge could lower the energy barrier for the formation of CO₂ reduction intermediates on the catalytic site [66]. Zhu and coworkers prepared N-doped MoS₂ nanosheets on N-doped carbon nanodots (N-MoS₂@NCDs) through a solvothermal method [66]. N-MoS₂@NCDs loaded on glassy carbon electrode exhibited high catalytic activity for the CO₂ reduction to CO as recording 36 mA/cm² at −0.9 V vs. RHE (90% FE).

The selection of electrolytes is also a critical factor to improve CO₂ reduction efficiency. EMIM-BF₄ electrolyte was shown to increase selectivity for CO product from CO₂ reduction [59]. Salehi-Khojin and coworkers reported that the use of a hybrid electrolyte of choline chloride/KOH was also effective to improve the CO₂ reduction selectivity of the MoS₂ electrode [67]. Using the choline chloride/KCl buffer instead of EMIM-BF₄ afforded 1.5 times higher current density at −0.25 V vs. RHE. The high catalytic activity was related to an accumulation of small size K⁺ ion on the MoS₂ surface, which allowed exposure of the MoS₂ edge site.

Xie and coworkers devised MoS/Se alloy monolayer to induce poor overlapping of frontier orbitals, which was helpful to enhance the surface activity for the CO₂ reduction [68]. The MoS/Se alloy monolayer improved CO production efficiency with 45% FE at −1.15 V vs. RHE compared to MoS₂ (17%) and MoSe₂ (31%) monolayer. On the other hand, in the report by Han and coworkers, the replacement of the chalcogenide with phosphide gave different CO₂ reduction products. MoP nanoparticle as supported by In-doped porous carbon (In-PC) converted CO₂ to formic acid as recording 43.8 mA/cm² at −2.2 V vs. Ag/AgNO₃ with 97% FE [69].

In a rare example, a Mo-cluster has shown promising photocatalytic CO₂ reduction ability under visible light illumination. Hexanuclear [Mo₆Br₁₄]²⁻ clusters [70] immobilized on graphene oxide [71] showed photocatalytic CO₂ reduction to methanol. High nuclearity transition metal cluster of [Mo₆Xⁱ₈L^a₆]²⁻ (Xⁱ = inner, L^a = apical) enabled visible light activities due to delocalized valence electrons overall metal centers [72,73]. The catalytic CO₂ reduction activities of the Mo-based electrocatalysts are compared in

Table 1. Compared CO₂ reduction catalytic efficiencies of Mo-based electrocatalysts.

Catalyst	FE	Current Density	CO2 Reduction Product (Major)	Ref.
Mo electrode	50%	-	Methanol	[58]
layer-stacked MoS2	~98%	−65 mA/cm2 at −0.76 V vs. RHE	CO	[59]
5% Nb-doped VA-MoS2	82%	−237 mA/cm2 at −0.8 V vs. RHE	CO	[60]
Ta-doped VA-MoS2	-	−98~68 mA/cm2 at −0.8 V vs. RHE	CO	[60]
MoBiSx nanosheets	71%	−12.1 mA/cm2 at −0.7 V vs. SHE	Methanol	[62]
MoS/Se monolayer	45%	−43 mA/cm2 at −1.15 V vs. RHE	CO	[68]
Cu-doped MoS2	85%	−17 mA/cm2 at −1.7 V vs. SCE	CO	[64]
NCMSH	93%	−34.31 mA/cm2 at −0.7 V vs. RHE	CO	[65]
MoS2 nanoflake (choline chloride)	93%	−315 mA/cm2 at −0.8 V vs. RHE	CO	[67]
N-MoS2@NCDs	90%	−36 mA/cm2 at −0.9 V vs. RHE	CO	[66]
MoP@In-PC	97%	−43.8 mA/cm2 at −2.2 V vs. Ag/AgNO3	Formic acid	[69]

.

2.4. FDH-Electrode Biohybrid

Formate dehydrogenase (FDH) is the most efficient electrocatalyst for interconversion between CO₂ and formate. There have been research interests in the enzyme activity under electrochemical conditions as adsorbed on an electrode surface, or as free in solution. FDHs are classified as two types of NADH-dependent and metal (Mo or W)-dependent enzymes (NADH = nicotinamide adenine dinucleotide). The bio-electrochemical set-ups of NADH-dependent Candida boidinii (CbFDH) have been reported to show selective CO₂ reduction to formate [74,75,76,77,78,79], where the regeneration efficiency of NADH was critical to improving the catalytic system.

Metal-dependent FDH has been studied as attached to an electrode surface. Hirst and coworkers showed reversible interconversion between CO₂ and formate by Escherichia coli FDH (EcFDH) as adsorbed on a graphite-epoxy electrode [80]. The reversible redox current for the CO₂/formate conversion was observed during the cyclic voltammetry scans, and the formate oxidation became favored at higher pH within 6–8 range. Controlled potential electrolysis at −0.6 V vs. SHE produced only formate product with 101.7 ± 2.0% FE.

Redox-active polymer could enhance electron transfer between FDH and electrode. Milton and coworkers used cobaltocene-functionalized polyallylamine (Cc-PAA) to promote electron transfer between EcFDH and glassy carbon electrodes [81]. The electroenzymatic CO₂ reduction was detected at −0.66 V vs. SHE close to the Cc/Cc⁺ redox couple of −0.576 V vs. SHE. Bulk electrolysis at −0.66 V vs. SHE using 50 mM NaHCO₃ as the CO₂ source provided formate production with FE 99 ± 5%, but the FE dropped to 65% after continuous electrolysis for 12 h.

Bio-hybrid system gives a synthetic vision to develop electrocatalysts with high selectivity and efficiency, although low current density and limited catalytic sites remain as limitations for scale-up. In addition, a hybrid of molecular catalysts with electrode surface would be an alternative method to obtain well-performing catalysts. Ligand design and synthesis are other difficulties, but the conjugation of coordination complexes by electrode surface can provide chances to modify surface reactivity and efficiency of electrocatalysts.

3. Nitrogen Fixation

Nitrogen is an essential element in the metabolism of living organisms and one of the major elements that make up the body. The only way to take atmospheric N₂ into organisms is via the nitrogen fixation process [82,83]. In a symbiotic relation, plants feed nitrogen-fixing bacteria with sugars as a reducing equivalent to perform the reduction of N₂ to NH₃, which is in return absorbed by plants. Although nature performs the process very efficiently, the direct N₂ reduction has been one of the challenges in the research area of synthetic catalysts [84]. The industrial Haber–Bosch process takes charge of global ammonia production, but this process occupies 1% of annual fossil fuel consumption and 3% of annual CO₂ emission because of the production conditions demanding high pressure and high temperature [85,86]. Mild reaction conditions are highly craved to decrease the consumption of the limited energy resources as well as anthropogenic CO₂ emission [87,88]. There exist many research efforts to elucidate the mechanism of the natural catalytic system, nitrogenase, by investigating the structural features and synthetic model complexes [89,90,91]. In addition, heterogeneous catalysts were developed for nitrogen reductions, and bio-hybrid catalysts were studied in a way of attaching nitrogenase to an electrode surface.

3.1. Nitrogenase

In the 1960s, X-ray structure of the FeMo-nitrogenase active site was obtained, and thereafter the structural features and the relation with the enzyme reactivity have been studied to elucidate the nitrogen reduction process [92,93,94]. The N₂ reduction is a series of electrochemical reactions requiring overall 6e⁻/6H⁺ input. The FeMo-cofactor facilitates the nitrogen reaction, and the co-existing P cluster is in charge of the electron-transfer steps. The FeMo cluster is buried under the enzyme protein to selectively promote the multi-electron reduction process (Figure 3).

N₂ + 8H⁺ + 16MgATP+ 8e⁻ → 2NH₃ + H₂ + 16MgADP + 16Pi

The FeMo-cofactor has a double-cubane type of Mo₇Fe₇S cluster, where each cubane is bonded by a carbide (C⁴⁻) center and a Mo atom is positioned at an end of the cluster [95,96,97]. Although the reaction mechanism remains obscure in the research of N₂ reduction catalysts, both the Fe and Mo sites have been considered as the active sites for the nitrogen reactions. In the current review, we focus on the Mo-based model complexes.

3.2. Structural Analogs of Nitrogenase

In 1978, Holm and coworkers synthesized [Mo₂Fe₆S₈] (22) cluster as a model of the FeMo-nitrogenase through a self-assembly method [98,99]. Later, in 1982, the same group reported a cubane-type structure of [MoFe₃S₄] (23) by removing a Mo ion from [Mo₂Fe₆S₈], [100,101] which showed very similar EXAFS data as the original FeMo-cofactor [102]. In 1993, Coucouvanis and coworkers synthesized a citrate-stabilized [MoFe₃S₄] (24), similar to the homocitrate-coordinated FeMo-nitrogenase [103]. In addition, [(Tp)MoFe₃S₄Cl₃]¹⁻ (25) cluster was synthesized using hydrotrispyrazolylborate(Tp) ligand as a model of homocitrate [104]. Along with the structural modeling of the FeMo-cofactor, Mo-based complexes were studied for the catalytic N₂ reduction [105,106,107] (Scheme 12).

3.3. Synthetic Electrocatalysts for N₂ Reduction

3.3.1. Homogeneous Mo Complexes for N₂ Reduction

In 1988, Hidai reported examples of converting N₂ to silylamines using zero-valent Mo coordinated by phosphine ligands, cis-Mo(N₂)₂(PMe₂Ph)₄ (26) and Mo(N₂)₂(dpe)₂ (27) (Scheme 13) [108]. Complex 26 produced 23.7% N(SiMe₃)₃ under optimized conditions of N₂ using Me₃SiCl and Na (or Li) as reductant, whereas complex 27 showed less activity of 9.7% conversion yield under the same condition. The monodentate phosphine made a more active Mo site than the bidentate phosphine. Even though the conversion yield was low, it gave early examples of Mo-phosphine catalysts for N₂ conversion reactions.

In 2003, Yandulov and Schrock used a high-valent Mo(III-IV) ion stabilized by a tetradentate triamidoamine ligand for the reduction of N₂ to NH₃ [109]. The coordination of a hexaisopropylterphenyltriamido amine (HIPT) ligand to Mo(IV) ion gave [HIPTN₃N]MoCl, where the six hexaisopropylterphenyl groups sterically enclosed the Mo active site [110]. The bulky ligand system prevented dimerization of the complex by conserving the single Mo center [106,111]. Unlike the FeMo-cofactor, the Mo complex was coordinated by organic N donors, but the reactivity study of [HIPTN₃N]Mo^III(N₂) (28) species provided implemental results to understand an N₂ reduction process (Scheme 13). Using a proton source of [(2,6-lutidinium)(BArF)] and decamethyl chromocene as a reductant, complex 28 generated 7.56 equiv. of NH₃ per a Mo ion and 63% yield per a reductant, which was comparable to that of FeMo-nitrogenase (75% NH₃, 25% H₂) [112]. They proposed an N₂ reduction mechanism catalyzed by a single Mo site based on spectral data and X-ray crystal structures of intermediates. X-ray crystal structures and spectral data characterized the nitrogen reaction intermediates such as Mo(N₂), Mo-N=NH, [Mo=N-NH₂]⁺, Mo≡N, [Mo=N-NH₃]⁺, [Mo=NH]⁺, [Mo-NH₃]⁺, and MoNH₃, which suggested the N₂ reduction pathway including 6e⁻/6H⁺ processes [113,114,115,116,117]. In the proposed pathway, protons bind to one nitrogen to release the first NH₃, and the next protonation occurs on the remaining nitrogen to produce the second NH₃ (Scheme 14). Their studies gave a research basis for a Mo-based N₂ reduction process [118].

In 2011, Nishibayashi and coworkers tried to improve the N₂ reduction activity of Mo complexes by utilizing redox-active ligands. They modified the Hidai complex (26) with ferrocenyl diphosphine ligand to prepare trans-Mo(N₂)₂(depf)₂ (29) (depf = diethylphosphinoferrocene) [119]. Complex 29 performed the silylamine production under 1 atm of N₂ at room temperature using Na and Me₃SiCl. Me₃SiCl formed Me₃Si∙ radical in the presence of Na reductant, and a Me₃Si∙ radical derived the N₂ reaction to produce N(SiMe₃)₃ (226 equiv. per a Mo atom for 200 h). Acid post-treatment of the produced silylamine afforded NH₃. The same group used other redox-active PNP type ligands for Mo catalysts [120]. Reduction of the [MoCl₃(PNP)] species with Na/Hg under N₂ formed an N₂-bridged di-Mo complex [Mo(N₂)₂(PNP)]₂(μ-N₂) (30a–c) (Scheme 13). Subsequent reaction with 4 equiv. of HBF₄∙OEt₂ and pyridine generated a Mo-hydrazide species, and further reaction with [LutH]OTf produced NH₃ (23.2 equiv. of NH₃ per a catalyst). The reaction mechanism was explained similarly as proposed by Schrock, where one nitrogen site receives 3e⁻/3H⁺ to release NH₃ and subsequent 3e⁻/3H⁺ reaction of the remaining Mo(N) generates (Scheme 15). In the screening of reductants of Cr-Cp (E_1/2 = −0.88 V vs. Ag/Ag⁺ in MeCN), Co-Cp (E_1/2 = −1.15 V), Cr-Cp* (E_1/2 = −1.35 V) and proton sources such as [LuTH]OTf (pKa 14.4 in MeCN), pyridinium trifluoromethanesulfonate (pKa 12.6), and HOTF (pKa 2.6), the Nishibayashi complex exhibited the N₂ reduction activity with Co-Cp and Cr-Cp*, but not with Cr-Cp [121]. The complex showed the best performance with the combinatory use of Co-Cp and [LuTH]OTf.

The electronic property of the pyridine ligand affected the catalytic reaction rate by modulating Mo–N≡N–Mo bond strength [122]. In comparison with the v(NN) frequency (1944 cm⁻¹) of unmodified pyridine (30a), the para-substitution with a phenyl group increased the frequency to 1950 cm⁻¹ (30b), whereas the methoxy group decreased to 1932 cm⁻¹ (30c). The relative bond strength was reflected in the N₂ reduction reactivity of catalysts. Complex 30a provided 21–23 equiv. of NH₃ production, but 30c with the electron-donating methoxy substituent induced the highly efficient NH₃ formation (34 equiv. per catalyst) by accelerating the protonation step. Similar to the P cluster of the nitrogenase, the substitution of a redox-active ferrocene assisted the catalytic reaction (30d–g) [123]. Ferrocene-attached Mo complexes showed a similar v(NN) stretching frequency as the complex 30a but increased the NH₃ production to 37 equiv. per catalyst. Although unmodified ferrocene did not affect the NN bond strength, modification of the ferrocene with ethyl or phenyl groups caused the slight shift of the v(NN) value indicating electronic connectivity through the chemical bonds. The v(NN) value was measured as 1939 cm⁻¹ with the ethyl-ferrocene and 1951 cm⁻¹ with the phenyl-ferrocene. Consistent with the electron-donating effect, the ethyl-ferrocene resulted in 30 equiv. of NH₃ formation per catalyst, which was higher than 10 equiv. of NH₃ of the phenyl-ferrocene.

The Nishibayashi type PNP-Mo catalysts began the N₂ reduction in the N₂-bound Mo state after a halide abstraction step. They recently reported the presence of iodide ligand rather improved the catalytic activity because of the electron-withdrawing property. PNP-MoI₃ complexes (31a–d) were synthesized with phosphine ligands varied by different substituents such as isopropyl, tert-butyl, adamantyl, and phenyl groups (Scheme 16) [124]. The P(tert-butyl)₂ group gave the highest NH₃ production. The catalytic activity was further examined by changing the para-position of pyridine with MeO, Me, Ph, Fc, and Rc (32a–e) with keeping the P(tert-butyl)₂ moiety of PNP ligand. At the time, the para-Ph substituent showed the highest activity of 90 equiv. of NH₃ production, which was much higher than the case of para-MeO substituent [125]. The trend was opposite from the case of N₂-bridged dinuclear Mo complexes. Complex 31b exhibited a maximum 415 equiv. of NH₃ production per catalyst, which was 35 times higher activity than the N₂-bridged Mo₂ complex. Furthermore, they examined the effect of PCP-type carbene ligands of 1,3-bis((di-tert-butylphosphino)methyl)benzimidazol-2-ylidene (PCP-1), 1,3-bis(2-(di-tert-butylphosphino)ethyl)imidazol-2-ylidene (PCP-2) [126] (Scheme 16). The PCP-1-Mo (33) produced 100 equiv. of NH₃ per catalyst, but the PCP-2-Mo (34) showed a low reactivity of 1.6 equiv. of NH₃ per catalyst. The identity of the reducing agent and proton source also affected the nitrogen reduction activity of Mo catalysts. Complex 33 showed the high activity when used together with SmI₂ and H₂O (or ethylene glycol) instead of CoCp* and [LutH]OTf. The combinatory use of SmI₂ and ethylene glycol resulted in turnover frequency (TOF) of 7000 h⁻¹ for the NH₃ generation, and the condition using H₂O as a proton source slightly decreased TOF to 6800 h⁻¹ [127]. Using ethylene glycol as a proton source produced 22% (relative to a reductant) of H₂ as a side product, but H₂O reduced the ratio to ~2%. The optimized condition using H₂O and SmI₂ increased the NH₃ productivity of complex 33 to 4350 equiv. per catalyst and TOF to 112.9 min⁻¹, which was close to the nitrogenase activity of 40–120 min⁻¹ [128].

3.3.2. Heterogeneous Mo-Containing N₂-Reduction Electrocatalysts

Nitrogen reduction catalysts should deliver protons selectively to a bound nitrogen on their active sites, because the thermodynamic potential of a competitive proton reduction is relatively low. For this, the intrinsic property of an active metal center is an important factor to determine the N₂ reduction reactivity of catalysts. Novel transition metals have shown nitrogen reactivities, but their limited reserves and high-cost issues are obstacles for their industrial scale-up as catalysts [129,130,131,132,133,134,135]. Earth-abundant metals are promising alternatives for nitrogen catalysts. Mo ions as the forms of MoS₂, MoO₃, MoN, Mo₂N, and Mo₂C were proven to be active for the N₂ reduction. The specific affinity of Mo ion to nitrogen has been reported in the previous experiments [136,137]. In addition, the utilization of mixed metal elements could increase N₂ reactivity and promote electron-transfer by tuning surface energy states.

The N₂ reactivity seems to be varied by Mo crystal orientation. Wang and coworkers reported that Mo(110) plane has higher reactivity with N₂ than other planes of (200) and (211) [138]. They prepared four types of Mo electrodes of Mo-foil, Mo-A-R, Mo-D-R-1h, and Mo-D-R-5h through electrodeposition, and Mo-D-R-5h showed the highest (110) orientation ratio based on XRD patterns [139,140,141].

Outermost Mo is an active site, but the reactivity and mechanism are determined by supporting elements. Sun and coworkers studied different reactivities of Mo, conjugated by O, S, and N atoms, by comparing MoO₃, MoS₂, and MoN [142,143,144]. Edge sites of Mo chalcogenides are known to be active for N₂ reaction. MoN surface has shown ~0.2 lower overpotential than Mo chalcogenides. It is possibly understood that as a surface Mo-N is reduced to NH₃, an empty Mo site facilitates subsequent N₂ adsorption and reduction process.

Conductive support was also an important component to improve the electrocatalytic performance of MoS₂, because the large conductive surface area supports more MoS₂ active area as well as increases surficial electrical conductivity. Tian and coworkers achieved significant improvement of FE (10.94% at –0.3 V vs. RHE) of MoS₂ using Ti₃C₂ MXene [145]. The catalytic efficiencies and reaction rates of the Mo-based electrocatalysts are compared in

Table 2. Faradaic efficiencies, NH₃ formation rates of different Mo-based catalysts.

Catalyst	FE	NH3 Formation Rate	Ref.
(110)-oriented Mo nanofilm	0.72%	3.09 × 10−11 mols−1cm−2 at −0.49 V vs. RHE	[138]
MoO3 nanosheets	1.9%	4.80 × 10−10 mols−1cm−2 at −0.5 V vs. RHE	[142]
MoN nanosheets	1.15%	3.01 × 10−10 mols−1cm−2 at −0.3 V vs. RHE	[143]
MoS2	1.17%	8.08 × 10−11 mols−1cm−2 at −0.5 V vs. RHE	[144]
1T-MoS2@Ti3C2	10.94%	30.33 μg h−1mg−1cat. at −0.3 V vs. RHE	[145]

.

3.4. Nitrogenase-Electrode Biohybrid

In general, synthetic modeling studies focus on enzyme active sites, but protein structure should be an important factor to control the enzyme reaction. Even a small mutation of a nearby amino acid could distort the entire enzyme reactivity. For example, the substitution of a nearby β-98^Tyr amino acid to β-98^His deactivated the N₂ reduction ability of the mutated FeMo-nitrogenase [146]. However, a reduction of hydrazine (N₂H₂) to NH₃ still occurred in the mutated FeMo-nitrogenase. Interestingly, the mutation by β-98^His improved electron-accepting ability from an un-natural reducing agent to promote reductions of azide (N₃⁻) or nitrite (NO₂⁻) to NH₃ [147]. Using electron mediators such as polyaminocarboxylate-ligated Eu [126] and polyallylamine (PAA) polymer-CoCp₂ [148] further improved reduction efficiencies. Rather in the presence of Fe-protein, without using an additional electron mediator, electron transfer between Fe-protein and FeMo-cluster became slow. Badalyan and coworkers showed experimentally that electron transfer between FeMo-cluster and Fe protein is the rate-determining step by calculating reaction rate constants using cyclic voltammetry data [149].

King and coworkers made a biohybrid system by attaching FeMo-nitrogenase to CdS nanorods for applications to a photocatalytic reduction of N₂ to NH₃ [128]. In the biological reaction, electron transfer steps from the P cluster to the FeMo-cluster require 16ATP (E_m = −0.42 V) (ATP = adenosine triphosphate). In the bio-hybrid system, irradiation of CdS nanorod with 405 nm wavelength generated excited electrons with −0.8 eV, which was transferred to the FeMo-nitrogenase to proceed with the N₂ reduction reaction. The TOF was 75 min⁻¹ and the NH₃ production rate was 315 ± 55 nmol(mg MoFe protein⁻¹)min⁻¹, which was 63% less activity compared to the enzyme activity with the presence of Fe protein and ATP. Upon using nitrogenase inhibitors such as acetylene, CO, and H₂ [150,151,152], the enzyme activity became silent excluding a possible N₂ reduction by CdS only.

Biohybrid of FeMo-nitrogenase with electrode enabled the enzyme to function in the absence of ATP, and thereby provided ways to study the enzyme reaction out of intracellular conditions. In future research, new biohybrid designs are required to decrease overpotential for the N₂ reduction as well as to develop methodologies for the transformation of N₂ to useful molecules.

4. H₂ Evolution

The proton reduction is always a competitive reaction in the electrocatalytic reductions of CO₂ and N₂, because the 2H⁺/H₂ reduction is thermodynamically favored. However, on the other hand, hydrogen molecules can be used as the simplest energy carrier for solar or electric energy conversion and storage [153]. Due to the importance, various transition metal-based electrocatalysts were developed for H₂ evolution [154,155,156].

4.1. Homogeneous Mo Complexes for H₂ Evolution

Nature does not use Mo metal for metabolic hydrogen reactions of H₂ evolution and splitting, but hydrogenase active site inspired ligand design for Mo complexes. It was shown that anionic cyclopentadienyl (Cp) ligand has a similar electronic property as the Ni ligation sphere of the [NiFe]-hydrogenase active site [157]. Felton and Donovan reported that [(η⁵-C₅H₅)Mo(CO)₃]₂ (Scheme 17) (35) complex promoted the reduction of acetic acid in acetonitrile with 0.9 V overpotential [158]. Fan and Hu utilized polyhapto ligands to prepare Mo-carbonyl complexes of (η³-C₃H₅)Mo(CO)₂(CH₃CN)₂Br (36), (η³-C₃H₅)Mo(CO)₂(dppe)Br (37), (η⁵-C₅H₅)Mo(CO)₃I (38), and (η⁵-C₅H₅)Mo(CO)₂PPh₃I (39) (Scheme 17) [159]. Complex 36–39 showed similar reactivities of the halide dissociation during the cyclic voltammetry in MeCN solution and also exhibited similar catalytic activity of 16–27 TON for 5 h for the reduction of trifluoroacetic acid as requiring ~1 V overpotentials.

Hydrodesulfurization utilizes Mo-sulfides as a catalytic surface [160]. Similar to the MoS₂ active site, Mo-sulfide moiety has been synthesized in the Mo coordination environment. DuBois and coworkers used Cp ligand to obtain a sulfide bridged dinuclear Mo complex, [(η⁵-C₅H₅)Mo(µ-S)]₂(µ-S₂CH₂) (Scheme 17) (40). Complex 40 facilitated reduction of p-cyanoanilinium tetrafluoroborate at −0.7 V vs. Fc⁺/Fc in MeCN solution containing 0.3 M of Et₄NBF₄ electrolyte [161]. Controlled-potential electrolysis at −0.96 V vs. Fc⁺/Fc generated 10.6 mol of H₂ gas per catalyst with 98 ± 5% FE. The catalytic activity was compared using different proton sources such as triflic acid (pKa = 2.6 in MeCN), p-cyanoanilinium tetrafluoroborate (7.6) and p-cyanoanilinium tetrafluoroborate (9.6). Higher pKa of the proton source shifted the proton reduction potential to the negative direction. Kinetic studies suggested that the elimination of H₂ from the catalytic site is the rate-determining step.

Chang and coworkers synthesized the Mo(S₂) moiety in a pentadentate polypyridyl Mo complex [(κ⁵-PY5Me₂)MoS₂]²⁺ (42) (PY5Me₂ = 2,6-bis(1,1-bis(2-pyridyl)ethyl)pyridine) from a reaction of [(κ⁵-PY5Me₂)Mo((κ¹-CF₃SO₃)](CF₃SO₃) (41) with S₈ (Scheme 18) [162]. Complex 42 generated H₂ in acetate buffer (pH 3) at the onset potential of −0.58 V vs. SHE. Controlled potential electrolysis at −0.83 V overpotential produced H₂ gas with ~100% FE and 280 mol/s TOF. Albeit less active than 42, Complex 41 also showed catalytic activity to produce H₂ gas in 0.6 M phosphate buffer (pH 7) requiring 0.52 V overpotential. Controlled potential electrolysis at 0.64 V overpotential produced H₂ gas as 1600 mol/h TOF per catalyst. Complex 41 also produced H₂ gas in seawater at the onset potential of −0.81 V vs. SHE with 1200 mol/h TOF at −1.40 V vs. SHE. [163].

Streb and coworkers observed that partial substitution of terminal disulfide in [Mo₃S₁₃]²⁻ by H₂O to [Mo₃S₇(H₂O)x]^(2−x)− increased the catalytic activity. Instead, the substitution of terminal disulfide with halides to [Mo₃S_7×6]²⁻ (X = Cl, Br) decreased significantly the HER activity [164]. They suggested that, if terminal disulfide is protonated, energetically stable and catalytically inactive species are formed. It looks critical the formation of a vacant coordination site on a terminal Mo to generate catalytically active Mo-H.

Eisenberg and coworkers used dithiolene ligands to obtain Mo-sulfides moiety in MoL₂(bdt)₂ (43a–g) (bdt = benzene-1,2-dithiol) (Scheme 19) [11]. Two-electron reduction of complexes 43a–e in MeCN/H₂O (9:1) solution dissociated isonitrile ligands to generate Mo(bdt)₂, which was the active species for the catalytic HER. Complexes 43a–g also showed photochemical HER in MeCN/H₂O solution under the conditions of using [Ru(bpy)₃]Cl₂ as a photosensitizer, 0.2 M ascorbic acid (pH 4) as an electron donor. Complexes 43a gave the highest TON of 520 for 24 h than other complexes (<475 TON).

Moly-oxo complexes have shown interesting proton reduction activities. Zhan and coworkers synthesized cis di-oxo Mo complex, [Mo^VIL(O)₂] (Scheme 20) (44) (L = 2-pyridylamino-N,N-bis(2-methylene-4-methoxy-6-tertbutylphenol)ion) [165]. Complex 44 promoted proton reduction at 0.25 M phosphate buffer (pH 7) recording 360 mol/h TOF (907 mV overpotential). Next, fluorine-substitution of [MoL’(O)₂] (Scheme 20) (45) (L’ = 2-pyridylamino-N,N-bis(2-methylene-4,6-difluorophenol)ion) gave higher TOF of 756 mol/h (89% FE) at −1.61 V vs. Ag/AgCl in mixed acetonitrile/water (2:3) solution containing phosphate buffer (pH 6) [166].

Kim and coworkers reported the proton reactivity of [MoO(S₂C₂Ph₂)₂]²⁻ (46) in MeCN solution. Treatment of protons (p-toluenesulfonic acid) with complex 46 in MeCN processed dehydration of the complex to give [Mo(MeCN)₂(S₂C₂Ph₂)₂] (47) (Scheme 21) [167]. Complex 47 had poor solubility in polar solvents, and the fast dehydration of 46 prevented the use as an electrocatalyst. However, modification of the dithiolene ligand could make Mo ion function as HER electrocatalyst. Fontecave and coworkers utilized quinoxaline–pyran-fused dithiolene (qpdt²⁻), analogous to the molybdopterin (MPT), to prepare (Bu₄N)₂[MoO(qpdt)₂] (48) (Scheme 22) [168]. The qpdt²⁻ ligand assisted the Mo-oxo site to catalyze HER. Electrocatalytic proton reduction of trifluoroacetic acid was detected at −0.55 V vs. Ag/AgCl, and TOF of 1030 s⁻¹ was obtained at −1.3 V vs. Ag/AgCl (FE 86% for 3 h). Complex 48 also facilitated photocatalytic HER with [Ru(bpy)₃]²⁺ in 0.1 M ascorbic acid (pH 4) as recording TON 500 for 15 h.

4.2. Heterogeneous Mo-Containing H₂-Evolution Electrocatalyst

In 2005, Norskov and coworkers calculated the free energy of atomic hydrogen bonding (∆G°_H) to MoS₂ surface and compared it with other catalysts such as enzymes (FeMo cofactor and NiFe hydrogenase) and metal surfaces of Au, Pt, Ni, and Mo [169]. Density functional theory (DFT) calculation results suggest that Ni and Mo bind strongly to atomic hydrogen; however, since the next proton/electron transfer steps are thermodynamically uphill, the H₂–releasing step becomes slow as making those metals not suitable for HER. Interestingly, the metalloenzymes of FeMo-nitrogenase and NiFe-hydrogenase using Ni and Mo elements had similar hydrogen-binding energy (∆G°_H ~ 0) as Pt. Plane MoS₂ was inactive for HER, but the edge of MoS₂ had a bit positive ∆G°_H ~ 0.1 eV in a suitable range for HER. Graphite-supported MoS₂ with enlarged edge area showed electrocatalytic HER with 0.1–0.2 V overpotential.

Chorkendorff and coworkers could modify the MoS₂ edge area by varying sintering temperature (Figure 4a) [170]. The MoS₂ prepared from 400 °C sintering had a longer edge length than another case of 550 °C sintering, and accordingly showed a higher HER current density of 3.1 × 10⁻⁷ A/cm² compared to 1.3 × 10⁻⁷ A/cm² of the latter. Albeit lower than the Pt electrode, the MoS₂ edge exhibited higher current density than other common metal elements. They also showed photocatalytic HER activity of mixed MoS_x surface as deposited on Ti|n⁺p-Si photocathode [171]. The MoS_x was electrodeposited on Ti-protected n⁺p-Si electrode by cyclic voltammetry (CV) scans at 0.137–1.247 V vs. RHE. The MoS_x|Ti|n⁺p-Si photocathode showed HER activity above 0.33 V vs. RHE under the illumination of red light.

The electrodeposition method seems to be not suitable to produce a large number of catalysts, and unstable species may not persist during cyclic voltammetry scans. Hu and coworkers reported that unsaturated MoS_x provides a more active site for HER [172]. They prepared MoS_x film by electro-polymerization method (Figure 4b), which exhibited higher catalytic activity than MoS₂ single crystal and MoS₂ nanoparticles. They explained that is because amorphous MoS_x increases sulfur defects at the surface [173]. The same group investigated the effects of growth mechanism and catalyst mass, but they reported that the catalyst mass dominantly affected the catalytic efficiency (Figure 4c) [174].

Mo sulfur clusters have shown similar active sites as MoS₂. Chorkendorff and coworkers adsorbed cubane type [Mo₃S₄]⁴⁺ cluster to highly oriented pyrolytic graphite [175]. [Mo₃S₄]⁴⁺ cluster showed similar overpotential as MoS₂ nanoparticle. Even higher TOF of 0.07 s⁻¹ was measured than 0.02 s⁻¹ of MoS₂ nanoparticle, but it was unstable during successive potential scanning.

Bensenbacher and coworkers synthesized [Mo₃S₁₃]²⁻ nanocluster by a wet chemical method of (NH₄)₆Mo₇O₂₄ and ammonium polysulfide [176]. [Mo₃S₁₃]²⁻ showed HER activity at 0.1 V onset potential and 0.18 V overpotential at 10 mA/cm² (TOF 3 s⁻¹). Li and coworkers synthesized Mo₃S₁₃ film by electrodeposition, which showed a higher catalytic rate and lower overpotential by ~40 mV than Mo₃S₁₃ film prepared by drop-casting of (NH₄)₂Mo₃S₁₃·2H₂O [177]. The increased electrical conductivity enhanced catalytic efficiency. Xu and coworkers improved HER efficiency by attaching [Mo₃S₁₃]²⁻ to highly conductive reduced graphene oxide-carbon nanotube (rGO-CNTs) aerogels, which recorded a lower overpotential by 74 mV than pure [Mo₃S₁₃]²⁻ [178]. Wu and coworkers reported that dimeric [Mo₂S₁₂]²⁻ cluster slightly improved the HER activity by recording 0.16 V overpotential at 10 mA/cm² (TOF ~3 s⁻¹) compared to [Mo₃S₁₃]²⁻ [179] They reported hydrogen adsorption free energy △G_ads(H) of [Mo₂S₁₂]²⁻ (−0.05 eV) is closer to zero than other surfaces such as [Mo₃S₁₃]²⁻ (−0.08 eV), MoS₂ (0.08 eV), and Pt(111)(−0.09 eV).

Min and coworkers reported HER activity of [Mo₃S₁₃]²⁻ using [Ru(bpy)₃]Cl₂ and ascorbic acid under visible light (≥420 nm). Initial TOF of 335 h⁻¹ was measured, but it had a problem of gradual decomposition [180]. The photocatalytic stability of [Mo₃S₁₃]²⁻ was improved by Zang and coworkers [181]. They encapsulated [Mo₃S₁₃]²⁻ in ethidium bromide covalent organic frameworks (EB-COFs). The COF-encapsulation did not decrease significantly the catalytic activity relative to a free [Mo₃S₁₃]²⁻ cluster. Mo₃S₁₃@EB-COF maintained photocatalytic HER for 8 h by recording 21,465 μmol g⁻¹ h⁻¹ rate, furthermore which showed the stable catalytic performance during the recycling test of 4 times of 5 h experiment.

Artero and coworkers studied the structure and reaction mechanism of amorphous MoS_x (α-MoS_x), which had polymer-based on [Mo₃S₁₃]²⁻ cluster sharing disulfide ligand [182]. They suggested that the reaction of terminal disulfide with proton and electron releases HS⁻ to generate unsaturated Mo^IV active site.

Octahedral molybdenum clusters are red near-IR phosphorescent emitters and showed high photocatalytic HER activity. Feliz and coworkers studied the catalytic performance of the (TBA)₂[Mo₆Brⁱ₈F^a₆] (TBA = tetra-n-butylammonium) cluster in aqueous solution in the presence of triethylamine (TEA) [183]. The catalytic activity of the {Mo₆Brⁱ₈}⁴⁺ cluster unit was enhanced by the in situ generation of the [Mo₆Brⁱ⁸F^a₅(OH)^a]²⁻ and [Mo₆Brⁱ₈F^a₃(OH)^a₃]²⁻ and [Mo₆Brⁱ₈(OH)^a₆]²⁻ species. In the same work, the cluster unit was coordinatively immobilized onto graphene oxide (GO) surfaces. The resulting material, (TBA)₂Mo₆Brⁱ₈@GO, enhanced the cluster stability in a water/MeOH mixture and under photoirradiation. Its catalytic performance was superior to that of GO, it decreased with respect to that of the molecular cluster complex. The TOF values with respect to atomic molybdenum were 5 × 10⁻⁶ s⁻¹ and 3 × 10⁻⁴ s⁻¹ for the heterogeneous and the homogeneous materials, respectively. Recently, the (TBA)₂[Mo₆Iⁱ₈(O₂CCH₃)^a₆] was also coordinatively immobilized onto GO to give (TBA)₂Mo₆Iⁱ₈@GO [184]. To assure the cluster stability of the cluster units under photocatalytic conditions, both materials were tested in vapor water photoreduction. Even after longer radiation exposure times, the catalysts remained stable and recyclability of both catalysts was demonstrated. The TOF of (TBA)₂Mo₆Iⁱ₈@GO is three times higher than that of the microcrystalline (TBA)₂[Mo₆Iⁱ₈(O₂CCH₃)^a₆], in agreement with the better accessibility of catalytic cluster sites for water molecules in the gas phase. In the framework of the study of the photocatalytic properties of the {Mo₆I₈}⁴⁺ cluster units, one of the most emissive octahedral metal clusters, [Mo₆Iⁱ₈(OCOC₂F₅)^a₆]²⁻, was immobilized onto graphene sheets through pyrene-containing organic cations as supramolecular linkers [185]. These non-covalent interactions enhanced the photocatalytic activity of the nanocomposite by 280% with respect to the molecular cluster and graphene counterparts. The improvement of the H₂ production activity is attributed to the synergetic effect between graphene and the hybrid cluster complex because graphene facilitates the charge-separation activity and enhances electron transfer of the cluster photocatalyst.

Lana-Villarreal and coworkers reported photocatalytic HER activity of Mo₃S₇ cluster as immobilized on TiO₂ surface [186]. Functionalized bipyridyl ligand of Mo₃S₇Br₄(diimino) cluster enabled adsorption of the cluster on TiO₂ surface. The immobilized Mo₃S₇ cluster decreased the HER overpotential by 0.3 V as recording 1.4 s⁻¹ TOF. Despite the moderate catalytic activity, it gave an example to immobilize molecular cluster on electrode surface.

Fabrication of hetero-metal chalcogenides was an effective method to improve catalytic activities of Mo-based electrode materials. Doping of metal promoters possibly improves the intrinsic activity of unsaturated Mo sites. The electrocatalytic effect of Cu/Mo hetero-metals was examined by Tran and coworkers [187]. Cu₂MoS₄ was synthesized by the solvothermal reaction of [Cu(CH₃CN)](BF₄) and (NH₄)₂[MoS₄], and the obtained Cu₂MoS₄ crystals were stable under air for weeks.

Both MoS₂ and CoSe₂ were active for HER, and the synergistic effect of combined MoS₂ and CoSe₂ was examined by Gao and coworkers [188]. MoS₂/CoSe₂ showed the HER activity close to commercial Pt/C. Interaction of the first-row transition metal Co with S could form S²⁻ and S₂²⁻ states, which possibly assisted MoS₂ growth.

MoSe₂ was relatively less studied compared to MoS₂. Since the atomic hydrogen adsorption energy was measured to be lower with MoSe₂ than that of MoS₂, MoSe₂ also showed an efficient HER activity [189]. Sasaki and coworkers improved corrosion stability of Ni/Mo-alloy by NiMoN_x nanosheet, where metal stabilizing effect of nitride possibly improved the stability [190]. Mo phosphide (MoP) is a known hydrodesulfurization catalyst. Jaramillo and coworkers compared catalytic HER activity of crystalline MoP and molybdenum phosphosulfide (MoP/S) [191]. Both the MoP and MoP/S were active for HER, but MoP/S showed the higher electrocatalytic activity than MoP.

Electrocatalytic HER activities of commercial MoB and Mo₂C have been reported [192]. Two electrode materials showed similar catalytic activity and were usable under both acidic and basic conditions, exhibiting similar activity and durability in continuous electrolysis for 48 h.

Leonard and coworkers reported the synthesis of various Mo-carbide materials such as α-MoC_1−x, β-Mo₂C, η-MoC, and γ-MoC with different stacking sequences, and compared their HER activities [193]. The β-Mo₂C, η-MoC, and γ-MoC had similar hexagonal crystal structure, but the stacking sequence of β-Mo₂C was ABAB, η-MoC was ABCABC, and γ-MoC was AAAA packing. α-MoC_1-x had a cubic structure as ABCABC stacking sequence. The β-Mo₂C showed the highest HER activity, and the catalytic activities of the others were obtained in the order of: γ-MoC > η-MoC > α-MoC_1−x. The γ-MoC showed stable catalytic activity with −1.95 mA/cm² at −340 mV vs. RHE in the continuous electrolysis for 18 h.

Nakanishi and coworkers reported the synthesis of Mo carbonitride (MoCN) nanomaterial [194]. The reaction of Na₂MoO₄ with diaminopyridine (DAP) in acidic condition gave DAP-bound Mo oxide, which was polymerized by NaHCO₃/(NH₄)₂S₂O₈ as emitting CO₂ gas. The CO₂ emission step formed nano-sized (PDAP)-2H⁺/MoO₄²⁻ (PDAP = polydiaminopyridine) complex causing a high density of the catalytically active site. Finally, the polymer was pyrolyzed at 800 °C to become MoCN nanomaterial (Figure 4f). MoCN nanomaterial showed the HER activity above −0.05 V vs. RHE and reached 10 mA/cm² at −0.14 V vs. RHE in sulfuric acid (pH 1) solution. MoCN nanomaterial had electrocatalytic stability during 1000 CV scans between −0.35 and 0.2 V vs. RHE.

Yu and coworkers used hybrid phosphorous-doped nanoporous carbon (PC) and RGO to support MoO₂ [195]. MoO₂@PC-RGO had a carbon skeleton, which prevented aggregation of catalytically active MoO₂ nanoparticles. Additionally, the synergistic combination of PC and RGO assisted to enhance the catalytic activity.

The use of adhesives for power-deposition on electrode generally lowers catalytic efficiency, so nickel foam (NF) has been used to prepare self-supported electrode. Huang and coworkers used NF to support Mo₂N/CeO₂ hetero-nanoparticles (Figure 4d) [196]. CeO₂ was coated on NF by hydrothermal method, and the CeO₂-fabricated NF was covered by Mo₂N with varying amounts of (NH₄)₄Mo₇O₂₄·4H₂O (0.02, 0.04, 0.05, 0.06 mmol) precursor by solvothermal method. Post-annealing at 500 °C for 5 h under NH₃ atmosphere afforded Mo₂N/CeO₂@NF. The synthetic condition using 0.05 mmol Mo source gave the best HER activity. Mo₂N/CeO₂@NF-0.05 sample showed smaller overpotential than commercial 20 wt% Pt/C. The high catalytic activity was probably because electronic interaction between Mo₂N and CeO₂ lowers the energy barrier for hydrogen intermediate formation (Figure 4e). The catalytic efficiencies of the above Mo-based HER electrocatalysts are compared in

Table 3. Compared catalytic efficiencies of Mo-based HER electrocatalysts.

Catalyst	Onset Potential	Overpotential	Tafel Slope (mV/dec)	Exchanged Current Density (A/cm2)	Ref.
MoS2 nanoparticles	-	-	55~60	1.3 × 10−7	[170]
[Mo3S4]4+ cluster	−0.2 V vs. NHE	-	120	2.2 × 10−7	[175]
MoSx|Ti|n+p-Si photocathode	0.33 V vs. RHE	-	39	-	[171]
MoS3–CV	-	200 mV at −15 mA/cm2	40	1.3 × 10−7	[172]
amorphous MoS3	-	200 mV at −4.8 mA/cm2	42	-	[173]
MoS2+x film	-	170 mV at −20 mA/cm2	-	-	[174]
MoSe2 nanosheets	−0.15 V vs. RHE	290 mV at −10 mA/cm2	101	-	[189]
MoSe2|RGO	−0.05 V vs. RHE	115 mV at −10 mA/cm2	69	-	[189]
Cu2MoS4		135 mV (onset)	95	4.0 × 10−5	[187]
MoS2/CoSe2	−11 mV vs. RHE	68 mV at −10 mA/cm2	36	7.3 × 10−5	[188]
NiMoNx nanosheet	−78 mV vs. RHE	-	35.9	2.4 × 10−4	[190]
commercial MoB		100 mV (onset)	55	1.4 × 10−6	[192]
commercial Mo2C		100 mV (onset)	56	1.3 × 10−6	[192]
MoP/S	-	90 mV at −10 mA/cm2	50	2.0 × 10−4	[191]
MoP	-	117 mV at −10 mA/cm2	50	5.0 × 10−5	[191]
γ-MoC	−0.24 V vs. RHE (at 0.18 mA/cm2)	-	121.6	3.2 × 10−6	[193]
Mo carbonitride	−0.05 V vs. RHE	-	46~51	-	[194]
MoO2@PC-RGO	0 V vs. RHE	64 mV at −10 mA/cm2	41	4.8 × 10−4	[195]
Mo2N/CeO2@NF-0.05	-	26 mV at −10 mA/cm2	37.8	-	[196]
[Mo3S13]2− nanocluster		180 mV at −10 mA/cm2	40		[176]
Mo3S13 film	−130 mV vs. RHE	200 mV at −10 mA/cm2	37		[177]
[Mo3S13]2− attached at (rGO-CNTs) aerogels	−110 mV vs. RHE	179 mV at −10 mA/cm2	60.2		[178]
dimeric [Mo2S12]2− cluster		161mV at −10 mA/cm2	39		[179]

.

5. Heterogeneous Mo-Containing O₂-Evolution Electrocatalysts

Water splitting is an ideal method for the conversion of electric and solar energy to the chemical bond energy of dihydrogen. However, oxidation of water to O₂, the counterpart reaction of H₂ evolution, is kinetically slow, which decreases the overall efficiency of the water-splitting reaction. Since oxygen evolution reaction (OER) is a multi-electron process demanding large overpotential, efficient electrocatalysts are highly desired. RuO₂ is a benchmark catalyst for the OER, but the high cost and easy decomposition to RuO₄ are drawbacks of using Ru. Research efforts have been made to develop Earth-abundant metal-based electrocatalysts with high-performance and durability, and, among those, Mo-containing electrocatalysts have shown promising OER activities.

MoS₂ edge site is also known to be active for the OER, and, thus, various synthetic methods were developed to enlarge the edge site area. Mohanty and coworkers investigated the OER activity of MoS₂ quantum dots [197]. The MoS₂ quantum dot compounds (MSQDs) of 2–5 nm size were synthesized without aggregation through single-step hydrothermal reaction using (NH₄)₂MoS₄) precursor (Figure 5a,b). Repeating linear sweep voltammetry (LSV) scans up to 50 cycles increased the OER current density, because the electrochemically active surface area of MSQDs was enlarged with the potential scans.

Direct growing methods, useful to increase contact, conductivity, and electron transfer, are generally applied to prepare OER electrocatalysts. In 2016, Yan and Lu prepared a porous MoS₂ microsphere on nickel foam (NF) [198]. The MoS₂ on NF showed a higher catalytic rate (105 mV/dec Tafel slope) than commercial RuO₂ on NF (127 mV/dec) and 20% Pt/C (150 mV dec⁻¹). Cui and coworkers synthesized mesoporous MoO₂ nanosheets on NF [199]. The mesoporosity of MoO₂ enlarged the active surface area and slightly improved the catalytic activity than that of compact MoO₂.

Boride increases a reverse electron transfer to the metal site, which accordingly increases electron density on the catalytic site. Gupta and coworkers synthesized ternary Co-Mo-B hetero-metal nanocomposite, which gave enhanced OER activity [200].

Li and coworkers used a well-defined Cu nanowire to deposit Ni/Mo-alloy catalyst [201]. Alkaline anodization and subsequent cathodic reduction process modified the surface of Cu foam. A mixture of Ni(SO₃NH₂)₂ and Na₂MoO₄ were electrodeposited on the Cu nano-scaffold to obtain Ni-Mo/Cu nanowire. The Ni-Mo/Cu nanowire exhibited comparable OER activity with commercial Pt/C(RuO₂) on Cu foam.

Yang and coworkers examined the synergistic effect of trimetallic FeCoMo nanocomposite for the OER test [202]. Amorphous FeCoMo nanocomposite was synthesized as a rod-like shape by hydrolysis of a mixture of CoCl₂∙6H₂O, FeCl₃∙6H₂O and (NH₄)₆Mo₇O₂₄∙4H₂O, hexamethylene tetramine at 90 °C. The existence of high-valent Mo⁶⁺ was characterized by Raman spectroscopy. In situ X-ray absorption near edge structure (XANES) showed that Mo⁶⁺ tends to attract electrons from 3d transition metals, which possibly promoted the OER process.

Gao and coworkers prepared a bifunctional catalyst using MoS₂ HER catalyst and Ni₃S₂ OER catalyst (Figure 5c–e) [203]. MoS₂−Ni₃S₂ heteronanorods on NF afforded higher OER activity than Ni₃S₂/NF and MoS₂/NF.

The bimetallic system enables manipulation of electronic structure and gives abundant active sites, but a synthesis of pure phase is difficult. Lan and coworkers could make pure-phase bimetallic Co/Mo carbides using mesoporous carbon substrate (Figure 5f) [204]. Transition metal carbide also gave the high catalytic activity because of the good electrical conductivity, metallic property, and chemical stability.

Ni oxyhydroxide is an active site for OER, but the low stability decreases the catalytic efficiency. Shen and coworkers increased the stability by fabricating MoFe:Ni(OH)₂/NiOOH nanosheet and studied synergistic effect [205].

Ma and coworkers synthesized one-dimensional MoO₂-Co₂Mo₃O₈@C nanorods with the enlarged catalytic surface area [206]. The ZIF-67 MOF was attached to MoO₃ nanorods to form MoO₃@ZIF-67, and subsequent hydrothermal reaction at 700 °C gave MoO₂-Co₂Mo₃O₈@C nanorods (containing Co/Mo in 9.1/10.5% ratio). The organic ligand of ZIF-67 was carbonized into carbon, which served as a reductant to reduce excess MoO₃ to MoO₂. The MoO₂-Co₂Mo₃O₈@C nanorods showed similar activity as the benchmark RuO₂.

Yang and coworkers studied the OER activity of Ni₂Mo₃N hetero-metal nitride [207]. They reported that an amorphous surface oxygen-rich activation layer (SOAL) was generated on Ni₂Mo₃N nanoparticle surface by applying high oxidative potential in alkaline conditions. The computational study suggested that SOAL could increase the OER activity, because the conjugation of Ni-nitrides active sites and the secondary Mo-electron pump promoted the catalytic reaction.

Jiang and coworkers reported the synergistic effect of Co₃Mo alloy nanoparticles as attached to nanoporous Cu skeleton [208]. They synthesized Mo-doped Co₃O₄ nanoflakes on CuO/Cu skeleton by electro-oxidation (EO) at 1.57 V vs. RHE. The EO Co₃Mo/Cu showed lower OER onset overpotential than Ir/C supported by nanoporous Cu. Since Co₃Mo/Cu also acted as a highly active HER catalyst, they composed a water electrolysis set-up by Co₃Mo/Cu cathode and EO Co₃Mo/Cu anode. The Co₃Mo/Cu-based electrolyzer showed overall water splitting delivering a current density of 100 mA/cm² at 1.62 V vs. RHE. The water-splitting system acted more efficiently than Pt/C/Cu-Ir/C/Cu set-up (Figure 5g,h). The catalytic efficiency parameters of the Mo-based OER electrocatalysts are compared in

Table 4. Onset potentials, overpotentials, Tafel slopes Mo-based OER catalysts.

Catalyst	Current Density (mA/cm2)	Overpotential	Tafel Slope (mV/dec)	Ref.
MoS2 on NF	20	310 mV	105	[198]
Mesoporous MoO2 nanosheets on NF	10	260 mV	54	[199]
Co-Mo-B	10	320 mV	56	[200]
Ni-Mo/Cu nanowire	20	280 mV	66	[201]
FeCoMo nanocomposite	10	277 mV	27.74	[202]
MoS2-Ni3S2 heteronanorods	10	249 mV	57	[203]
bimetallic Co/Mo carbides	10	260 mV	50	[204]
MoFe:Ni(OH)2/NiOOH nanosheet	100	280 mV	47	[205]
MSQDs-AC	10	370 mV	39	[197]
MoO2-Co2Mo3O8@C nanorods	10	320 mV	88	[206]
Ni2Mo3N hetero-metal nitride	10	270 mV	59	[207]
EO Co3Mo alloy nanoparticles	164	350 mV	82	[208]

.

6. Conclusions

This review summarizes Mo-containing metalloenzymes and their model complexes, homogeneous and heterogeneous catalysts under categories of reactivities with CO₂, N₂, H₂, and O₂. Compared to the Mo enzymes, catalytic activities of synthetic systems remain at a low-efficiency level. Thus, continuous research efforts in synthetic and theoretical perspectives are requested to achieve high-performing catalysts. The revealed Mo-enzyme structures enabled synthetic research on the enzyme reactivities, and concomitant theoretical studies gave an in-depth understanding of the enzyme mechanism. However, previously proposed mechanisms need further supports by synthetic experiments, and spectroscopic research on living cells still has limitations; thus, more active modeling studies are required.

Along with the modeling studies, new catalysts by ligand design should be developed to find an optimal condition to prepare active Mo ion. Ligand design for a low-valent Mo ion to function under lower gas pressure is pursued. Examples of homogeneous catalysts using high-valent Mo ions are relatively rare. Mimicking the enzyme active site partially or conceptually could provide synthetic ideas to develop both low- and high-valent Mo-based catalysts. In addition, catalytic stability of the Mo complex is a prerequisite for commercialization. Utilization of hetero-metallic surface is a method of preventing deactivation of catalytic sites as well as providing synergistic effects. Another method would be to use polymer or organic scaffold for regulating access of reagents into a catalytic site, which is similar to the enzymatic strategy to protect the active site by surrounding it with polypeptide chains.