Dihydrogen complexes as prototypes for the coordination chemistry of saturated molecules

Abstract

The binding of a dihydrogen molecule (H₂) to a transition metal center in an organometallic complex was a major discovery because it changed the way chemists think about the reactivity of molecules with chemically “inert” strong bonds such as HH and CH. Before the seminal finding of side-on bonded H₂ in W(CO)₃(PR₃)₂(H₂), it was generally believed that H₂ could not bind to another atom in stable fashion and would split into two separate H atoms to form a metal dihydride before undergoing chemical reaction. Metal-bound saturated molecules such as H₂, silanes, and alkanes (σ-complexes) have a chemistry of their own, with surprisingly varied structures, bonding, and dynamics. H₂ complexes are of increased relevance for H₂ production and storage in the hydrogen economy of the future.

Dihydrogen (H₂) and hydrocarbons are vital in chemical processes such as hydrogenation and conversions of organic compounds. Catalytic hydrogenations are the largest-volume chemical reactions: all crude oil is treated with H₂ to remove sulfur/nitrogen, and >100 million tons of ammonia fertilizer are produced annually to support much of the world's population. The H₂ molecule is married together by a very strong two-electron HH bond but is only useful chemically when the two H atoms divorce in controlled fashion. This also applies to other strong σ-bonds such as CH in alkanes. However, the mechanism at the molecular level by which the union splits was established only relatively recently because such electronically saturated molecules were never caught in the act of chemically binding to a metal or other “third party,” usually the first step in breaking apart a strong bond. The discovery by Kubas and coworkers (1) in 1984 of coordination of a nearly intact H₂ molecule to a metal complex (L_nM; L = ligand) caught this in intimate detail and led to a new paradigm in chemistry (1–4) (see Sketch 1).

Sketch 1.

The H₂ binds side-on (η²) to M primarily through donation of its two σ electrons to a vacant metal orbital to form a stable H₂ complex. It is remarkable that the electrons already strongly bonded can donate to a metal to form a nonclassical 2-electron, 3-center bond, as in other “electron-deficient” molecules such as diborane (B₂H₆). MH₂ and other “σ-complexes” (3, 5), encompassing interaction of any σ-bond (CH, SiH, etc) with a metal center, are the major theme of this special feature.

Introduction and Historical Perspective

Certain discoveries and how they came about are fascinating sagas, e.g., that for buckminsterfullerene (C₆₀) (6). That its existence remained hidden for so long adds to the lore, and our unexpected revelation of metal–H₂ complexes has some commonality. Metal dihydrides formed by oxidative addition (OA) of the HH bond to a metal center had been known early on to be a part of catalytic cycles (7), as documented in a 1980 retrospective on catalytic hydrogenation by a pioneer in the field, Jack Halpern (8). Although some form of metal–H₂ interaction was assumed to participate in dihydride formation, it was thought to be unobservable. We were fortunate to observe it in the complex W(CO)₃(PR₃)₂(H₂), as detailed by this author (2, 3). This was the first molecular compound synthesized and isolated entirely under ambient conditions that contained the H₂ molecule (albeit “stretched”) other than elemental H₂ itself. The HH bond length in W(CO)₃(PⁱPr₃)₂(H₂) (0.89 Å) is stretched ≈20% over that in H₂ (0.74 Å), showing that H₂ is not physisorbed but rather chemisorbed, where the bond is “activated” toward rupture. Like H₂, other saturated molecules such as alkanes were thought to be inert to such binding, although their CH bonds somehow could also be broken on metals. The “somehow” is why the finding of an H₂ complex was important: it is the prototype for activation of all σ-bonds.

This discovery of W(CO)₃(PⁱPr₃)₂(H₂) ensued the serendipitous synthesis of its novel, “unsaturated” 16-e precursor, M(CO)₃(PCy₃)₂ (M = Mo, W; Cy = cyclohexyl) (9). Its unusual purple color changed instantly and reversibly to yellow on exposure to N₂ and H₂ in both solution and solid states, signifying adduct formation (Eq. 1).

Crystallography later revealed a phosphine CH bond weakly occupying the sixth binding site in W(CO)₃(PCy₃)₂ (10). This type of “agostic” interaction (11) relieves electronic unsaturation in coordinatively unsaturated complexes and is entropically favorable because it is “intramolecular.” “Intermolecular” binding of a CH bond as in an alkane σ-complex (often also termed “agostic”) is less stable. Irrefutable evidence for H₂ binding in Eq. 1 came slowly because pinpointing H positions crystallographically is difficult, even by neutron diffraction. A consultant, Russ Drago, suggested an experiment elegant in its simplicity: synthesize the HD complex and look for a large HD coupling constant in the proton NMR that would show that the HD bond was mostly intact. It worked beautifully: the ¹H NMR of W(CO)₃(PⁱPr₃)₂(HD) showed a 1:1:1 triplet (deuterium spin = 1) with J_HD = 33.5 Hz, nearly as high as in HD gas, 43.2 Hz. Observation of J_HD higher than that for a dihydride complex (>2 Hz) became the premier criterion for an H₂ complex.

One reason that H₂ complexes were so well hidden was the notion that such complexes could not be stable relative to classical dihydrides, as exemplified by the controversy over our initial findings. This paralleled the discovery of metal–dinitrogen complexes by Allen and Senoff, whose seminal paper was initially rejected (12). At the time of our finding, spectroscopic evidence for unstable MH₂ interactions was found by photolysis of Cr(CO)₆ in the presence of H₂ at low T (13–16). Cr(CO)₅(H₂) was postulated based on IR CO stretching frequencies but could not be conclusively demonstrated; only recently has its ¹H NMR spectrum been observed at low T (17, 18). Even theoretical bases for interaction of H₂ and other σ-bonds with a metal was still in its infancy at the time of our discovery. Ironically, a computational paper by Saillard and Hoffmann (19) in 1984 on the bonding of H₂ and CH₄ to metal fragments such as Cr(CO)₅ was published shortly after our publication (1) of the W–H₂ complex, without mutual knowledge. Such interplay between theory and experiment has continued as one of the most valuable synergistic relations in all of chemistry (3, 4, 20). The innate simplicity of H₂ was attractive computationally, but the structure/bonding/dynamics of H₂ complexes turned out to be unimaginably complex and led to extensive study (>300 computational publications).

Initially, H₂ binding in M(CO)₃(PR₃)₂(H₂) seemed unique because the bulky phosphines sterically inhibited formation of a 7-coordinate dihydride through OA. Kaesz and coworkers (21) viewed this as “arrested OA,” a descriptive term for the bonding in a silane complex, CpMn(CO)₂(η²-HSiPh₃). Silane complexes (22, 23) were among the first examples of σ-bond complexes but were initially unrecognized as such because the asymmetrically bound silane ligand (Eq. 2) lacked the superb clarity of the H₂ ligand, which has electrons only in the HH bond.

The hundreds of H₂ complexes synthesized after our discovery could not initially have been imagined, and it was difficult to know where to search for new ones. It would take more than a year before others were identified, notably by Morris, Crabtree, Chaudret, and Heinekey. This quartet has since performed elegant synthetic, reactivity, and NMR studies on H₂ and silane complexes (5, 25–30) and was eventually joined by >100 investigators worldwide. Remarkably, several complexes initially believed to be hydrides were revealed to be H₂ complexes by Crabtree and Hamilton in 1986 (5, 31), by using as criteria the short proton NMR relaxation times of H₂ ligands (T₁ < 100 msec). Particularly interesting was RuH₂(H₂)(PPh₃)₃ first reported in 1968 (32); it possessed unusual H₂ lability that Singleton in 1976 commented was characteristic of “H₂-like bonding” (33). However, attempts to prove H₂ binding here was problematic, even long after H₂ binding was established (34).^‡

More than 600 H₂ complexes are known (most of them stable) for nearly every transition metal and type of coligand and are the focus of 1,500 publications, dozens of reviews, and three monographs (2–5, 20, 25–30, 35–43). The view on H₂ complexes has shifted from significance in basic science to a more practical bent, e.g., H₂ fuel production and storage. Two frequent questions after their discovery were as follows. Are H₂ complexes relevant in catalysis, i.e., does direct transfer of hydrogen from an H₂ ligand to a substrate occur? And could methane bind to metal complexes? The answer to both is yes, and although a stable methane complex has not been isolated, alkane binding has been observed.

Synthesis and Diagnosis of H₂ Complexes

Most H₂ complexes contain low-valent metals with d⁶ electronic configurations that favor side-on binding of σ-bonds. Reversibility of H₂ binding is often a key feature, i.e., H₂ can be removed simply upon exposure to vacuum and readded many times at ambient temperature/pressure, as in Eq. 1. Virtually all σ-complexes are diamagnetic, with one exception (44). σ-Ligands have not been definitively shown to bridge metals. Surprisingly, the coligands on H₂ complexes can be simple classical nitrogen-donor ancillary ligands such as ammonia, as in [Os(NH₃)₅(H₂)]²⁺ (45), which has a very long HH distance (d_HH), ≈1.34 Å (46), more characteristic of a dihydride, which it was initially believed to be (47). Complexes containing only H₂O (48) or CO (17, 18) coligands are also known, but are marginally stable (Scheme 1).

Scheme 1.

Determining the presence of a H₂ ligand and its d_HH is nontrivial because even neutron diffraction has limited applicability and can give foreshortened d_HH because of rapid H₂ rotation/libration (49). ¹J_HD is the best criterion, and values determined in solution correlate well with d_HH in the solid state through Eqs. 3 and 4 (50, 51).

Data include d_HH from crystallography and also solid-state NMR measurements by Zilm and Millar (52) that gave the most accurate d_HH (direct measure of internuclear HH separation). For W(CO)₃(PⁱPr₃)₂(H₂), J_HD = 34 Hz, giving d_HH = 0.86–0.88 Å vs. 0.89 Å from solid-state NMR and 0.82(1) Å from neutron diffraction [uncorrected for H₂ libration (49)]. Short T₁ values for the H₂ ligand (27) are also diagnostic (e.g., 4 msec for the W complex), although care must be exercised in interpretation (53–55). A powerful spectroscopic tool developed by a colleague at Los Alamos, Juergen Eckert, is inelastic neutron scattering studies of rapid H₂ rotation in solid H₂ complexes that provide unequivocal evidence for molecular H₂ binding and also the presence of M–>H₂ backdonation (56).

Several synthetic routes to H₂ complexes are available; the simplest is reaction of H₂ with an unsaturated complex such as W(CO)₃(PR₃)₂ (Eq. 1). Displacement of weakly bound “solvento” ligands such as CH₂Cl₂ (57) or H₂O from [Ru(H₂O)₆]²⁺ (Scheme 1) is also effective. Protonation of a hydride complex by acids is most often used (28, 38) and is widely applicable because it does not require an unsaturated precursor that may not be available. Neutral polyhydrides L_nMH_x are convenient targets for protonation to cationic H₂ complexes, [L_nM(H₂)H_x-1]⁺, which can be more robust than complexes prepared from H₂. Only a few stable solid bis-H₂ complexes are known, e.g., [RhH₂(H₂)₂(PCy₃)₂]⁺ (58), Tp*RuH(H₂)₂ (59), and RuH₂(H₂)₂(PR₃)₂; R = Cy (30) and cyclopentyl (60), for which the neutron structure shows unstretched cis–H₂ ligands.

Structure, Bonding, and Dynamics of H₂ Complexes

The 3-center metal–H₂ interaction complements classical Werner-type coordination complexes where a ligand donates electron density through its nonbonding electron pair(s) and π-complexes such as olefin complexes in which electrons are donated from bonding π-electrons (Scheme 2). It is remarkable that the bonding electron pair in H₂ can further interact with a metal center as strongly as a nonbonding pair in some cases. The resulting side-on bonding in M-H₂ and other σ-complexes is “nonclassical,” by analogy to the 3-center, 2-electron bonding in carbocations and diborane. The M center may be considered to be isolobal with H⁺ and CH₃⁺ (61), mimicking carbocation chemistry; i.e., a σ-complex such as M⁺–CH₄ is related to CH₅⁺, which is viewed as a highly dynamic H₂ complex of CH₃⁺ (62). H₂ is thus a weak Lewis base that can bind to strong electrophiles, but transition metals are unique in stabilizing H₂ and other σ-bond complexes by “backdonation” of electrons from a filled metal d orbital to the σ* antibonding orbital of H₂ (Scheme 2), a critical interaction unavailable to main group atoms (3, 4, 20). The backdonation is analogous (4) to that in the Dewar–Chatt–Duncanson model (63, 64) for π-complexes, e.g., M–ethylene.

Scheme 2.

A large variety of σ-bonds XH interact inter- or intramolecularly with metal centers (3, 42). In principle any XY bond can coordinate to a metal center, providing substituents at X and Y do not interfere. Backdonation of electrons from M to H₂ (or to σ* of any XY bond) is crucial not only in stabilizing σ-bonding but also in activating the bond toward homolysis (3, 4, 20). If it becomes too strong, e.g., by increasing electron-donor strength of coligands on M, the σ-bond cleaves to form a dihydride because of overpopulation of the H₂ σ* orbital. There is often a fine line between H₂ and dihydride coordination, and in some cases equilibria exist in solution for W(CO)₃(PR₃)₂(H₂) (Eq. 5), showing that side-on coordination of H₂ is the first step in HH cleavage (2).

Although electronic factors for OA are well established, the role of steric factors is less clear. Bulky phosphines can inhibit H₂ splitting: for less bulky R = Me the equilibrium lies completely to the right, i.e., the complex is a “dihydride” (65). However, as shown above, H₂ complexes are also stable with only small coligands L such as NH₃ (Scheme 1), in some cases with greatly elongated d_HH, two further paradigm shifts. This led to extensive efforts to vary M, L, and other factors to study stretching of the HH bond. Within the large regime of hundreds of L_nMH₂ complexes, the reaction coordinate for the activation of H₂ on a metal (Scheme 3) shows d_HH varying enormously, from 0.82 to 1.5 Å (3, 18, 25–31, 35–38, 44–60, 65–68). This “arresting” of bond rupture along its entire reaction coordinate is unprecedented. Although the d_HH ranges shown are arbitrary, each category of complexes has distinct properties. The d_HH is relatively short (0.8–1.0 Å), and H₂ is reversibly bound, in “true” H₂ complexes best exemplified by W(CO)₃(PR₃)₂(H₂), much as in physisorbed H₂ where d_HH is <0.8 Å. Elongated H₂ complexes (d_HH = 1–1.3 Å) (29, 46, 66–69) were first clearly identified in 1991 in ReH₅(H₂)(PR₃)₂ where neutron diffraction showed a d_HH of 1.357(7) Å (67). Complexes with d_HH > 1.3 Å are now viewed as “compressed hydrides,” with NMR features differing from elongated H₂ complexes, e.g., J_HD increases with T for the former and decreases for the latter (69). These are terms because a near continuum of d_HH has been observed.

Scheme 3.

Activation of H₂ is very sensitive to M, L, and charge, e.g., changing R from phenyl to alkyl in Mo(CO)(H₂)(R₂PC₂H₄PR₂)₂ leads to splitting of H₂ (49). Strongly donating L, third-row M, and neutral charge favor elongation or splitting of HH, whereas first-row M, electron-withdrawing L, and positive charge (cationic complex) favor H₂ binding and shorten d_HH. The ligand trans to H₂ has a powerful influence: strong π-acceptors such as CO (and also strong σ-donors such as H) greatly reduce backdonation and normally keep d_HH < 0.9 Å. Thus one can favor a σ-complex by placing the potential σ-ligand trans to a strong π-acceptor. Conversely, mild σ-donors such as H₂O or π-donors such as Cl trans to H₂ elongate d_HH (0.96–1.34 Å), as dramatically demonstrated by the isomers in Scheme 4(70). The cis-dichloro complex is actually a “compressed trihydride” (d_HH ∼ 1.5 Å) in solution, but in the solid state it is an elongated H₂ complex (d_HH = 1.11 Å) due to Ir–Cl···H–Ir hydrogen bonding, illustrating the hypersensitivity of d_HH to both intra- and intermolecular effects (71). Exceptions exist: the isomers of an “electron-poor” system, Cr(CO)₄(PMe₃)(H₂), have similar J_HD (≈34 Hz, hence d_HH ≈ 0.86 Å) whether H₂ is trans to CO or PMe₃ (18).

Scheme 4.

At what point is the HH bond “broken”? Theoretical analyses suggest 1.48 Å, i.e., twice the normal length (72), but little HH bonding interaction remains for d_HH > 1.1 Å (29). In certain “elongated” H₂ complexes, e.g., [OsCl(H₂)(dppe)₂]⁺, the energy barrier for stretching the HH bond from 0.85 Å all of the way to 1.6 Å is calculated (29, 69) to be astonishingly low, ≈1 kcal/mol. The H₂ is highly delocalized: the H atoms undergo large amplitude vibrational motion along the reaction coordinate for HH breaking. Remarkably, d_HH is both temperature and isotope dependent in [CpM(diphosphine)(H₂)]ⁿ⁺ (M = Ru, Ir; n = 1, 2) (73, 74). These phenomena illustrate the highly dynamic behavior of coordinated H₂ (40), which can exhibit quantum-mechanical phenomena such as rotational tunneling (56) and exchange coupling (75). MH₂ and other σ-bond interactions are among the most dynamic, complex, and enigmatic chemical topologies known. The H₂ ligand can bind/dissociate, reversibly split to dihydride, rapidly rotate, and exchange with cis hydrides, all on the same metal. Often these dynamics cannot be frozen out on the NMR time scale even at low T.

It is clear that H₂ binding followed by OA serves as a prototype for other σ-bond activation processes, e.g., CH and SiH. Silanes (H_nSiR_4-n) bind in η²-Si-H fashion (as in Eq. 2) (21–24, 42, 76). The η²-SiH₄ structure in cis-Mo(CO)(SiH₄)(Et₂PC₂H₄PEt₂)₂, the first transition metal complex of SiH₄, exists in equilibrium with its OA tautomer, MoH(SiH₃)(CO)(Et₂PC₂H₄PEt₂)₂, analogous to that for the W(η²-H₂) system (Eq. 5), with similar structures and thermodynamic parameters (77). SiH₄ binding and SiH cleavage directly model that believed to occur for CH₄ activation. SiH distances in hydrosilane complexes vary widely, analogous to H₂ complexes (3). A valuable yardstick for measuring activation in M(η²–X–H) bonds is the value of the NMR coupling constant J_XH compared with that in the free ligand. There is typically a 50–80% reduction in J_HD in unstretched HD complexes, a 74% reduction in J(¹³CH) for low-temperature cyclopentane coordination in CpRe(CO)₂(C₅H₁₀) (78), and 65% in J(¹¹BH) in complexes of neutral borane ligands (79). J_SiH in M(η²–SiH) are normally closer to those in OA products. η²–GeH bonds undergo OA much more easily than SiH, and in general, the ease of OA of H₂ lies between that of germanes and silanes (80). Backdonation is critical: silanes bind more strongly than alkanes and cleave much like H₂ because the SiH bond is a good acceptor whereas CH is not (the much higher energy of its σ* orbital reduces interaction with M d orbitals). However, the situation is more complex than for H₂ activation because substituents at C or Si alter both electronics and sterics.

Reactivity of σ-Complexes: Acidity and Heterolysis of XH Bonds

Aside from loss of H₂, reactions of MH₂ are dominated by homolytic cleavage of H₂ (OA) and heterolytic cleavage, essentially deprotonation of bound H₂ on electrophilic metal centers (Scheme 5) (25). σ-Complexes have several advantages in catalytic and other reactions. Foremost is that the formal oxidation state of M does not change on binding of H₂, whereas formation of a dihydride formally increases the metal oxidation state by two. H₂ ligands can also have far greater thermodynamic and kinetic acidity than hydrides, which is important in the ability of acidic H₂ ligands to protonate substrates such as olefins and N₂. In heterolytic cleavage (25, 40, 81, 82), the H₂ ligand is deprotonated, and the remaining hydrogen ligates to the metal as a hydride. Both pathways have been identified in catalytic hydrogenation and also may be available for other σ-bond activations, e.g., CH cleavage. Heterolysis of XH bonds through proton transfer to a basic site on a cis ligand or to an external base is a crucial step in many industrial and biological processes involving direct reaction of H₂, silane, borane, and (possibly) alkane ligands.

Scheme 5.

H₂ complexes can undergo heterolysis in two distinct ways (Scheme 6). Intramolecular heterolysis is extremely facile for proton transfer to a cis ligand L (e.g., H or Cl) or to the counteranion of a cationic complex. The proton can also end up at a trans ligand (Eq. 6) (83).

Scheme 6.

Intermolecular heterolysis involves protonation of an external base B, e.g., an ether solvent, to give a metal hydride (H⁻ fragment) and the conjugate acid of the base, HB⁺. This is the reverse of protonation reactions used to synthesize H₂ complexes (all reactions in Scheme 6 can be reversible), and the [HB]⁺ formed can relay the proton to internal or external sites (base-assisted heterolysis). Crabtree and Lavin (84) first demonstrated heterolysis of H₂ by showing that the H₂ in [IrⁱH(H₂)(LL)(PPh₃)₂]⁺ is deprotonated by LiR in preference to the hydride ligand. A milder base, NEt₃, was shown by Heinekey and Chinn (85) to more rapidly deprotonate the η²-H₂ tautomer in an equilibrium mixture of [CpRuH₂(dmpe)]⁺ and [CpRu(H₂)(dmpe)]⁺. The H₂ ligand has greater kinetic acidity because deprotonation of an H₂ complex involves no change in coordination number or oxidation state. Thus, H₂ gas can be turned into a strong acid: free H₂ is an extremely weak acid [pK_a ∼ 35 in THF (86)], but binding it to an electrophilic cationic metal increases the acidity spectacularly, up to 40 orders of magnitude. The pK_a can become as low as −6, i.e., η²–H₂ can become more acidic than sulfuric acid as shown by Morris (25, 26, 82) and later Jia (36). Electron-deficient cationic H₂ complexes with electron withdrawing ligands such as CO and short HH bonds (<0.9 Å), i.e., [Re(H₂)(CO)₄(PR₃)]⁺ (87) are among the most acidic. Positive charge increases acidity: W(CO)₃(PCy₃)₂(H₂) is deprotonated only by strong bases (88), but on oxidation to [W(CO)₃(PCy₃)₂(H₂)]⁺ becomes acidic enough to protonate weakly basic ethers (89). Such ability is relevant to processes such as ionic hydrogenation and the function of metalloenzymes such as hydrogenases (H₂ases).

Complexes with H₂ ligands are highly dynamic, and cis interactions, which are hydrogen-bonding-like interactions between η²–H₂ and a cis hydride observable in the solid state (3, 20, 42, 90), facilitate solution exchange processes (Eq. 7). The intermediate is a “trihydrogen” complex (91, 92). Although not isolated, evidence exists for its intermediacy in facile H-atom exchange in ReH₂(H₂)(CO)(PR₃)₃ (93), which can be exceedingly fast even at −140°C in hydrido(H₂) complexes (94–99). The barrier for hydrogen exchange in IrClH₂(H₂)(PⁱPr₃)₂ is only 1.5 kcal/mol even in the solid state (95, 96).

Can direct transfer of hydrogens from an H₂ ligand occur in catalytic hydrogenation? Although difficult to prove conclusively, there is evidence in ionic hydrogenation where an organometallic hydride, e.g., CpMoH(CO)₃, plus a strong acid, e.g., HO₃SCF₃, reduce ketones (100, 101). An acidic H₂ complex is involved in proton transfer (Scheme 7). An impressive example of catalysis employing heterolysis of H₂ is the asymmetric hydrogenation of ketones to alcohols catalyzed by the ruthenium system of Nobel Laureate Ryoji Noyori (102, 103). Other σ-bonds can be cleaved heterolytically, particularly on electrophilic metals (3, 40, 42). For coordinated SiH bonds, the bond becomes polarized Si(δ⁺)–H(δ⁻), i.e., the Si becomes positively charged (Scheme 8). Very reactive silylium ions are eliminated; they scavenge nucleophiles such as water or abstract fluoride from anions such as B(C₆F₅)4− (104). Similarly, a coordinated BH bond in a BH₃·PMe₃ ligand in [Mn(CO)₄(PR₃)(BH₃·PMe₃)]⁺ cleaves to give H⁻ (forming MnH(CO)₄(PR₃)) and “[BH₂·PMe₃]⁺” (105).

Scheme 7.

Scheme 8.

Can CH bonds in alkanes bind to electrophilic M to form a σ-alkane complex that can be split heterolytically? Proton transfer to a cis ligand (or anion) could take place followed by functionalization of the resultant methyl complex. Increased acidity of CH bonds in transient alkane complexes analogous to that for coordinated HH bonds may be important in alkane activation such as conversion of methane to methanol, a holy grail in chemistry well addressed in this special feature and the prolific work of Bercaw, Periana, and Bergman. In 1965, Chatt discovered OA of an arene CH bond to a metal complex and in 1976 predicted that “in 25 years methane will be the most popular ligand in coordination chemistry,” as noted by Shilov (106). As can be seen, this prediction has become true. As in H₂ activation, alkane σ-complexes should be intermediates, astonishingly even in reaction media as harsh as sulfuric acid at 200°C in Pt^II-catalyzed methane to methanol conversions (107, 108), despite the weak binding energy of CH₄ to metals [≈10 kcal/mol (109)].

Molecular binding and heterolysis of H₂ on metal surfaces and small metal clusters is rarely observed because formation of hydrides is favored. H₂ binding to a stepped Ni(510) surface containing unsaturated sites was seen by electron energy-loss spectroscopy (110) and is the first step in hydriding other surfaces (111, 112). H₂ also ligates at low T in small clusters such as Cu₃(H₂) (113), Pd(H₂) (114), and similar species (115). Oxides adsorb and activate H₂, including Cr₂O₃, MgO, and ZnO even at 25°C; some of these could involve molecular binding. (η²–H₂)CrO₂ has been prepared by cocondensation of CrO₂ molecules with H₂ in Ar at 11 K and photoisomerized to HCrO(OH), ostensibly through H₂ heterolysis (116). RuO₂ (111) has also been found to bind H₂ at 85 K (ν_HH = 2960 cm⁻¹; calcd d_HH = 0.89 Å) (117) suggesting that, as for H₂ on Ni surfaces, the binding of H₂ is similar to that in organometallics. Zeolites can bind H₂ (118, 119), notably to the extraframework iron in Fe–ZSM5 at 110 K. Research at the interface between heterogeneous and homogeneous catalysis (41) includes employing H₂ interactions as probes for the catalytic sites in both regimes (120–122). An elegant example is the demonstration that Ir(CO)Cl(PPh₃)₂ catalyzes hydrogenation of unsaturated compounds both in solution and solid state through an H₂ complex (123).

Activation of H₂ on Biological and Nonmetal Systems

H₂ases are redox enzymes in microorganisms that catalyze H₂ ⇌ 2H⁺ + 2e⁻ to either use H₂ as an energy source or dispose of excess electrons as H₂ (124–127). Biologically unprecedented CO and CN ligands are present in the dinuclear active site of iron-only H₂ases (128) that are remarkably organometallic-like and have been extensively modeled for biomimetic H₂ production (126, 127, 129–132) (see Sketch 2).

Sketch 2.

This site presumably transiently binds and heterolytically splits H₂, most likely at a site trans to bridging CO, where a proton transfers to a thiolate ligand as in Eq. 6 or other Lewis-basic site (127). Such heterolysis has recently been shown to occur on a mononuclear Fe complex with a pendant nitrogen base (132). Nature apparently designed these enzymes billions of years ago to use the CO ligand, whose strong trans influence favors reversible H₂ binding and heterolysis (3, 40). An H₂ complex of a H-ase model, [Ru₂(μ-H)(μ-S₂C₃H₆)₂(H₂)(CO)₃(PCy₃)₂]⁺, is known, albeit with Ru instead of Fe (133).

H₂ can also be activated at nonmetals, e.g., the bridging sulfides in Cp₂Mo₂S₄ that react with H₂ to form SH ligands perhaps via a 4-center S₂H₂ transition state (134). Metal-free hydrogenation of ketones on strong bases such as t-BuOK occurs under harsh conditions, apparently through base-assisted heterolysis of H₂ (135, 136). Thus, H₂ is a very weak acceptor (Lewis acid) through electron donation to its σ* orbital and can interact with the O in alkoxide or metal oxides and undergo heterolysis (3). Significantly, the first example of reversible splitting of H₂ on a nonmetal center has been found (137). The phosphine borane in Scheme 9 has a strong Lewis acidic center (boron) linked to a Lewis basic site (phosphorus). It is likely that H₂ heterolysis takes place at boron where proton transfer from an H₂-like complex to the basic phosphorus site occurs to form a phosphenium–borate.

Scheme 9.

H₂ Storage and Production: A Glance to the Future

H₂ is a fuel of the future, but vexing challenges exist. Materials for H₂ storage are difficult to design because, although H₂ can readily be extruded from a variety of compounds, it can be difficult to add back. The materials also must be light and contain >6% by weight H₂, reducing prospects for known facile reversible systems such as metal–H₂ or hydride complexes. Amine borane, H₃NBH₃, is a popular candidate and also combines both Lewis acidic (B) and basic (N) centers. Here, however, these centers are directly bonded, whereas the acidic and basic sites are separated by linkers in the phosphine-borane in Scheme 9. The metal-free aspect is relevant because precious metals such as platinum are often used in catalysis and can be environmentally unfriendly as well as costly or in short supply. Materials such as metal-organic frameworks (MOFS) (138–140) are now being examined for H₂ storage and have huge surface area capable of binding large numbers of H₂ molecules. Neutron scattering studies by Eckert are critical in determining whether H₂ binds to unsaturated metal centers as in organometallics and/or is physisorbed in the framework. Calculations indicate complexes with multiple H₂, i.e., Cr(H₂)₆ may be stable (141), and species such as [M(H₂)_n]⁺ have a fleeting gas phase existence (142), but isolation in condensed phases will be problematic.

Production of H₂ fuel from water by means of solar energy is of high interest (143). Catalysis may involve H₂ complexes at least as intermediates, and H₂ complexes have been implicated in solar energy conversion schemes based on photoreduction of water (144). Industrially important water gas shift and related H₂-producing reactions undoubtedly proceed through transient H₂ complexes (145). Biomimetic H₂ production, particularly solar driven (photocatalysis), is also a challenge and may take a cue from models of the active site of H₂ase coupled with models of nature's photosystems (129–131, 143). Here the formation of HH bonds from protons and electrons, the microscopic reverse of H₂ heterolysis, will be crucial in leading to formation of H₂ and is very rapid at the Fe sites in H₂-ases. Coupling model catalysts with photochemical water splitting will require fine-tuning of electrochemical potentials for tandem catalysis schemes.