Dinitrogen Cleavage and Hydrogenation by a Trinuclear Titanium Polyhydride Complex

See allHide authors and affiliations

Science  28 Jun 2013:
Vol. 340, Issue 6140, pp. 1549-1552
DOI: 10.1126/science.1238663

Titanium Cleaver

A century after its discovery, the Haber Bosch process is still used to produce ammonia from nitrogen for fertilizer. Nonetheless, the process requires high temperature and pressure, and chemists continue to look for synthetic analogs to microbial nitrogenase enzymes, which have managed to slice through the N2 triple bond under ambient conditions for millennia. Most efforts in this vein have relied on a boost from the reducing power of alkali metals. Shima et al. (p. 1549; see the Perspective by Fryzuk) instead explored the reactivity of a titanium hydride cluster, which cleanly slices through N2 at room temperature and incorporates the separated N atoms into its framework. Though ammonia was not produced, the system offers hope in the search for mild nitrogen reduction catalysts.


Both the Haber-Bosch and biological ammonia syntheses are thought to rely on the cooperation of multiple metals in breaking the strong N≡N triple bond and forming an N–H bond. This has spurred investigations of the reactivity of molecular multimetallic hydrides with dinitrogen. We report here the reaction of a trinuclear titanium polyhydride complex with dinitrogen, which induces dinitrogen cleavage and partial hydrogenation at ambient temperature and pressure. By 1H and 15N nuclear magnetic resonance, x-ray crystallographic, and computational studies of some key reaction steps and products, we have determined that the dinitrogen (N2) reduction proceeds sequentially through scission of a N2 molecule bonded to three Ti atoms in a μ-η122-end-on-side-on fashion to give a μ2-N/μ3-N dinitrido species, followed by intramolecular hydrogen migration from Ti to the μ2-N nitrido unit.

Dinitrogen (N2) is the most abundant component (78%) of Earth’s atmosphere and is largely chemically inert under ordinary conditions. Certain microbial organisms can reduce N2 to ammonia (NH3) by using nitrogenase enzymes at ambient temperature and pressure (17). In this transformation, six electrons and six protons are required to produce two equivalents of NH3 per N2. Industrially, NH3 is produced in ~108 tons/year quantities from N2 and H2 by the Haber-Bosch process, in which H2 serves as the source of both electron and proton (810). This process requires relatively harsh conditions (350° to 550°C and 150 to 350 atm) to activate N2 on the solid catalyst surface, making it energy intensive. Indeed, the Haber-Bosch ammonia synthesis consumes more than 1% of the world’s annual energy supply. Both the biological and Haber-Bosch processes are thought to take place through the cooperation of multiple metal sites.

To further explore the mechanism of N2 reduction at the molecular level and thereby develop milder chemical processes for ammonia synthesis, extensive studies on the activation of N2 with organometallic complexes have been carried out over the past decades. By use of low-valent transition metal species or a combination of transition metal complexes with strong reducing reagents, such as KC8, Na/Hg, or Mg, the activation of N2 has been achieved under mild conditions (1122). The catalytic transformation of N2 to NH3 has also been accomplished by using a Mo-N2 complex (23, 24). However, these reaction systems generally require a stoichiometric excess of strong reducing agents and extra proton sources to afford NH3. An alternative approach is the direct reduction of N2 by transition metal hydrides, which avoids the use of extra reducing agents and proton sources and may provide an entry to homogeneous catalyst systems for the synthesis of NH3 from a mixture of N2 and H2. Previously, various metal hydride complexes have been reported for the activation of N2 (25). However, most of these N2-activating hydrides were mononuclear transition metal complexes and did not lead to N−N bond cleavage. A binuclear niobium tetrahydride complex has been reported to enable N−N bond cleavage with loss of two H2 molecules, but the hydrogenation of the resulting nitrido species did not take place (26). A metal hydride complex that can induce both N≡N bond cleavage and N–H bond formation remains unknown, and the use of a polynuclear rather than binuclear metal hydride complex for the activation of N2 has not been reported to date. In view of the fact that both Haber-Bosch and biological ammonia syntheses likely rely on the cooperation of multiple metal sites in the activation and hydrogenation of N2, the investigation of the reactivity of multimetallic hydride complexes with N2 is of great interest and importance. We report here a trinuclear titanium polyhydride complex that reacts with N2 through N≡N bond cleavage and N–H bond formation under mild conditions without additional reducing agents or proton sources. The Ti-bound N2 activation products and some key reaction steps have been elucidated by 1H and 15N nuclear magnetic resonance (NMR), x-ray crystallographic, and computational studies.

We have previously reported that the hydrogenolysis of the C5Me4SiMe3-ligated half-sandwich rare-earth dialkyl complexes such as [(C5Me4SiMe3)Ln(CH2SiMe3)2(THF)] (Ln indicates Sc, Y, Gd, Tb, Dy, Ho, Er, Tm, Yb, or Lu; Me, methyl group; and THF, tetrahydrofuran) with H2 could easily afford the corresponding tetranuclear rare-earth octahydride complexes of a general formula {[(C5Me4SiMe3)Ln]4(μ-H)8} (2729). The analogous tetranuclear zirconium and hafnium octahydride complexes {[(C5Me4SiMe3)M]4(μ-H)8} (M = Zr, Hf) could also be obtained similarly by the hydrogenolysis of the alkyl precursors [(C5Me4SiMe3)M(CH2SiMe3)3] (30). In an attempt to synthesize a titanium analog, we carried out the hydrogenolysis of the titanium alkyl complex [(C5Me4SiMe3)Ti(CH2SiMe3)3] (1) under similar conditions (4:1 H2:N2 mixture at 5 atm) in an autoclave. A mixed di-imido/tetrahydrido tetranuclear titanium complex {[(C5Me4SiMe3)Ti]43-NH)22-H)4} (2) was obtained in 90% yield as dark purple crystals (Fig. 1A), whereas the expected octahydride complex {[(C5Me4SiMe3)Ti]4(μ-H)8} was not observed. In this reaction, one N2 molecule was formally reduced to two [NH]2– imido units by H2. The hydrogenolysis of 1 in the presence of 15N2 afforded the isotopically enriched 15N-imido complex {[(C5Me4SiMe3)Ti]43-15NH)22-H)4} (2-15N) [15N NMR, with a chemical shift in parts per million (δ) of 52.7 (using MeNO2 as a standard reference) and a coupling constant of NH (JNH) = 66.5 Hz]. Similarly, the reaction of 1 with D2 and N2 gave the corresponding deuterated analog {[(C5Me4SiMe3)Ti]43-ND)22-D)4} (2-d6). These results suggest that the NH imido units in 2 are formed by the hydrogenation of molecular nitrogen with H2. Protonolysis of 2 and 2-15N with anhydrous hydrochloric acid almost quantitatively afforded almost NH4Cl and 15NH4Cl, respectively, together with the formation of [(C5Me4SiMe3)TiCl3].

Fig. 1 N2 activation by Ti complexes 1 and 3.

(A) Hydrogenolysis of 1 in the presence of N2 affords an imido complex 2 (Cp′ = C5Me4SiMe3), the x-ray core structure of which is shown in the square frame (Cp′ ligands omitted for clarity; the N1–Ti bond lengths, with standard deviations in the parentheses, are given in angstroms). Hydrogenolysis under an N2-free atmosphere gives mainly a trinuclear heptahydride complex 3 together with a tetranuclear octahydride 4. (B) Reaction of 3 with N2 at room temperature (rt) gives the nitrido/imido/dihydrido complex 5, which on reaction with 15N2 at 180°C affords the tri-imido/nitrido complex 6-15N2. The x-ray core structures of 3, 5, and 6 (Cp′ ligands omitted) are shown in the square frames. The oxidation states of the Ti metals in 3, 5, and 6 are assigned formally.

To further clarify the mechanism for the formation of 2, we then carried out the hydrogenolysis of 1 with H2 (4 atm) under N2-free conditions, which afforded a nitrogen-free trinuclear titanium heptahydride complex {[(C5Me4SiMe3)Ti]33-H)(μ2-H)6} (3) in 69% yield as dark brown crystals (Fig. 1A) together with a small amount of the tetranuclear titanium octahydride complex {[(C5Me4SiMe3)Ti]4(μ-H)8} (4) (10%, dark purple crystals) (31). An x-ray diffraction study revealed one μ3-H and six μ2-H ligands in 3, with each Ti atom also coordinated by a C5Me4SiMe3 ligand (Fig. 1B). Formally, one of the three Ti atoms in 3 should be in the 4+ oxidation state, and two Ti atoms should be in the 3+ oxidation state in view of the total negative charge (10) of all the ligands. However, there is no significant difference in the three Ti–Ti separation distances, which exhibit an average value (2.6479 Å) comparable with those found in the Ti(III) hydride complexes reported to have metal–metal bonding interactions (32). A density functional theory (DFT) study on a model compound {[(C5H4SiH3)Ti]33-H)(μ2-H)6} (3m) implied that the electron cloud of the highest occupied molecular orbital (HOMO) is delocalized among the three Ti metals, with Wiberg bond indexes (WBI) for the Ti−Ti bonds of 1.1058 (Ti1−Ti2), 1.1213 (Ti1−Ti3), and 1.1233 (Ti2−Ti3), respectively, consistent with Ti–Ti bonding interactions in 3.

In agreement with the presence of Ti(III)–Ti(III) bonding interactions observed by the x-ray and DFT analyses, 3 exhibited diamagnetic behavior and showed well-resolved signals in the 1H NMR spectrum in toluene-d8. The seven hydrides appeared equivalently as a singlet at δH 2.66 over a temperature range from 22° to –80°C, suggesting rapid μ2-H/μ3-H site exchange.

The tetranuclear Ti(III) octahydride complex 4 is an analog of the zirconium and hafnium complexes {[(C5Me4SiMe3)M]4(μ-H)8} reported previously (30), neither of which showed activity toward N2 at room or even high temperatures (~120°C). In contrast, the trinuclear mixed valance Ti(III)/Ti(IV) heptahydride complex 3 showed high reactivity with N2, affording the imido/nitrido complex {[(C5Me4SiMe3)Ti]32-NH)(μ3-N)(μ2-H)2} (5) in 91% yield as dark blue crystals, upon exposure to a N2 atmosphere (1 atm) at room temperature for 12 hours (Fig. 1B). Single crystals of 5 suitable for x-ray diffraction study were obtained by recrystallization in THF. In the solid-state structure, the nitrido atom (N1) is bonded to three Ti atoms, with one N–Ti bond [N1-Ti1: 2.071(5) Å] significantly longer than the other two almost equivalent N–Ti bonds [N1–Ti2: 1.853(5), N1–Ti3: 1.883(5) Å] (Fig. 1B). The NH imido ligand (N2) bridges one of the three Ti–Ti sides of the Ti3 triangle [Ti2–N2: 1.937(6), Ti3–N2: 1.926(6) Å], and the other two Ti–Ti sides are each bridged by a μ2-H ligand [Ti1–H1: 1.91(4), Ti3–H1: 1.90(4) Å; Ti1–H2: 1.92(4), Ti2–H2: 1.91(4) Å].

The reaction of 3 with 15N2 under the similar conditions afforded the corresponding 15N-enriched complex {[(C5Me4SiMe3)Ti]32-15NH)(μ3-15N)(μ2-H)2} (5-15N), confirming that the imido and nitrido units in 5 are formed by the reduction of N2. The 15N NMR spectrum of 5-15N in THF-d8 at –50°C showed two broad signals at δ 402.9 and δ 46.9, which are assignable to the nitrido (μ3-N) and imido (μ-NH) units, respectively (33). The 1H NMR spectrum of 5-15N at –50°C showed a doublet at δ 17.62 with JNH = 63.6 Hz for the imido group.

To gain more information on the formation of 5, we monitored the reaction of 3 with 15N2 by 1H and 15N NMR spectroscopy in THF-d8 at low temperatures (Fig. 2). The reaction of 3 with 15N2 took place even at –30°C, leading to formation of a dinitrogen complex {[(C5Me4SiMe3)Ti]3-(μ-η122-15N2)(μ-H)3} (7-15N) with release of two equivalents of H2 (observed at δH 4.5) [the x-ray structure of a trimetallic Ti N2 complex was reported recently; see (34)]. In this process, N≡N was formally reduced to [N2––N2–] by four electrons generated by reductive elimination of two molecules of H2 from 3. The 15N NMR spectrum of 7-15N at –30°C showed two doublets at δN 73.1 and 262.9 with JNN = 21.5 Hz. This JNN value and the large difference in chemical shift between the two 15N NMR signals of 7-15N are comparable to those observed in the binuclear tantalum end-on-side-on dinitrogen complex {[NPN]TaH}2N2 {where [NPN] = PhP(CH2SiMe2NPh)2, δN –20.4 and 163.6, and JNN = 21.5 Hz} (35). The three hydride ligands in 7-15N gave two singlets with a 1:2 integration ratio at δH 9.73 (1 H) and –13.80 (2 H) in the 1H NMR spectrum at –30°C. When the temperature was raised from –30° to –10°C, nitrogen–nitrogen bond cleavage took place to give the dinitrido (N3–) complex {[(C5Me4SiMe3)Ti]33-N)(μ2-N)(μ2-H)3} (8-15N) almost quantitatively in 2 hours with disappearance of 7-15N. This transformation was accompanied by the oxidation of the two Ti(III) sites to two Ti(IV) units. The 15N NMR spectrum of 8-15N at –50°C showed two singlet peaks at δN 593.4 and 444.8, which could be assigned to the two bridging nitrido ligands. No cross-peak with any protons was observed by 1H-15N two-dimensional NMR spectroscopy. The three hydride units in 8-15N showed one triplet and one doublet at δH 4.65 (1 H) and 2.67 (2 H) with JHH = 28.0 Hz, respectively, in the 1H NMR spectrum at –70°C. When the temperature was raised to 20°C, one of the two nitrido units in 8-15N was hydrogenated (or protonated) by an H ligand, yielding the mixed imido/nitrido/dihydrido complex 5-15N. In this reaction, a hydride (H) is oxidized to a proton (H+), whereas the two Ti(IV) ions bridged by the hydride are formally reduced to Ti(III) (Fig. 2A), demonstrating that a hydride ligand can serve as a formal proton source through metal reduction. The related N–H bond formation in the reactions of zirconium-N2 and hafnium silylimido species with H2 has been reported previously (15, 17, 36).

Fig. 2 NMR monitoring of reaction kinetics of 3 with N2.

(A) Observed intermediates in the reaction with 15N2: oxidation states of the Ti metals are assigned formally. (B) Conversion versus time curves at the indicated temperatures. The solid lines are interpolations of the experimental data.

To have a better understanding of the mechanistic details, we performed DFT computations on a model compound of 3, namely [(C5H4SiH3)3Ti3H7] (3m). The Kohn-Sham orbital analysis revealed that the lowest unoccupied molecular orbital (LUMO) of 3m concentrates on the Ti3 atom, facilitating access of N2 in an end-on manner (Fig. 3A) (37). After the coordination of N2 to Ti3, the rearrangement of some hydride ligands takes place, leading to release of one molecule of H2 and formation of the pentahydride/dinitrogen complex [(C5H4SiH3)3Ti3H5(μ-η12-N2)] (3m′-N), in which the dinitrogen is bonded to two Ti atoms (Ti2 and Ti3) in a side-on-end-on fashion (Fig. 3B). The whole process is exergonic by 3.89 kcal/mol. Subsequently, release of another molecule of H2 from 3m′-N takes place to give the trihydride/dinitrogen complex [(C5H4SiH3)3Ti3H3(μ-η122-N2)] (7m), which is equivalent to the dinitrogen complex 7-15N observed experimentally. The N–N bond cleavage in 7m then occurs via a transition state TS78 to give the dinitrido complex [(C5H4SiH3)3Ti3H33-N)(μ2-N)] (8m). This process is accompanied by migration of a Ti–H bond from Ti3 to Ti1. The subsequent migration of a μ2-H ligand, which bridges Ti1 and Ti2, to the μ2-N nitrido atom in 8m affords the imido/nitrido product [(C5H4SiH3)3Ti3H23-N)(μ2-NH)] (5m). To see whether N–H bond formation could precede N–N bond cleavage, we also computed the energetics of migration of an H atom to an N atom in 7m (dashed line, Fig. 3B). However, this reaction path requires overcoming an energy barrier as high as 47.55 kcal/mol and is therefore kinetically less favorable. These computational results are in good agreement with the experimental observation of the dinitrido/trihydrido intermediate species 8-15N as described above (Fig. 2).

Fig. 3 Computational analysis of the reaction of 3 with N2.

(A) Observed intermediates in the reaction with 15N2: oxidation states of the Ti metals are assigned formally. (B) Conversion versus time curves at the indicated temperatures. The solid lines are interpolations of the experimental data.

The sequential N–N bond cleavage and N–H bond formation observed in the present reaction of the titanium hydride cluster 3 with N2 stands in contrast with the reaction mechanisms previously observed in other homogeneous or surface-supported organometallic N2-activating systems or the FeMo nitrogenase enzymes, in which N–H bond formation generally took place before N–N bond cleavage (5, 38). In the heterogeneous Haber-Bosch process, N2 reduction is also thought to take place on the catalyst surface first through N≡N bond cleavage then followed by hydrogenation of the resulting nitrido species (8, 38), although details are not clear because of the difficulty in identifying the true active sites and reaction intermediates.

The dihydrido/imido/nitrido complex 5 is stable at room temperature. However, when it was heated at 180°C under N2 (1 atm) overnight, further incorporation and reduction of N2 took place to give the tri-imido/nitrido complex {[(C5Me4SiMe3)Ti]33-N)(μ2-NH)3} (6) in 85% yield. When 5 was heated with 15N2 under the same conditions, the 15N-enriched analog {[(C5Me4SiMe3)Ti]33-15N)(μ2-15NH)(μ2-NH)2} (6-15N2) was obtained (Fig. 1B). The 15N NMR spectrum of 6-15N2 showed a singlet at δN 424.6 and a doublet at δN 101.3 with JNH = 64.0 Hz, which could be assigned to a μ3-N nitrido and a μ2-NH imido unit, respectively. These results suggest that the newly incorporated N2 molecule is split into a μ3-N nitrido unit and a μ2-NH imido unit, whereas the nitrido unit originally existing in 5 is hydrogenated to a μ2-NH imido group. No apparent reaction between 5 or 6 and H2 (up to 8 atm) was observed at room or higher temperatures (up to 150°C). However, when the hydrogenolysis of the trialkyl complex 1 with H2 was carried out in the presence of 1 equiv of 5, the tetranuclear di-imido complex 2 was formed quantitatively (Fig. 1B), possibly through hydrogenation of the nitrido group of 5 with a mononuclear Ti hydride species such as (C5Me4SiMe3)TiH3 formed in situ by hydrogenolysis of 1 (39). These results could account for the formation of 2 in the hydrogenolysis of 1 with H2 in the presence of N2 (Fig. 1A), in view of the facile formation of 3 in the hydrogenolysis of 1 in the absence of N2 and the high reactivity of 3 with N2 to give 5. Although the origin of the unusually high reactivity of the trinuclear mixed valence Ti(IV)/Ti(III) heptahydride complex 3 is subject to further studies, our findings demonstrate that hydride ligands in a metal hydride cluster can serve as the source of both electron and proton and that multimetallic transition metal hydride complexes can serve as a platform for nitrogen fixation.

Supplementary Materials

Materials and Methods

Figs. S1 to S26

Tables S1 to S3

Scheme S1

References (4050)

References and Notes

  1. The hydrogenolysis of 1 under a higher H2 pressure (80 atm) at 60°C led to formation of a larger amount of the octahydride 4 (~20%) and a smaller amount of the heptahydride 3 (~50%).
  2. A similar DFT study showed that the reaction of the tetranuclear Ti octahydride complex 4 with N2 is less favored, because its LUMO is distributed on the four Ti atoms with competitive orbital contribution and is unsuitable in orbital shape for an overlap with the HOMO of N2 (see fig. S26 and table S3 in the supplementary materials).
  3. The hydrogenation of the nitrido group of 5 with a mixed alkyl/hydride species formed in situ by partial hydrogenolysis of 1 followed by further hydrogenolysis of the alkyl species could not be ruled out in this case.
  4. Acknowledgments: This work was supported by a Grant-in-Aid for Young Scientists (B) (no. 21750068) and a Grant-in-Aid for Scientific Research (S) (no. 21225004) from the Japan Society for the Promotion of Science, an Incentive Research Grant from RIKEN, and grants from the National Natural Science Foundation of China (nos. 21028001 and 21174023). We gratefully appreciate access to the RIKEN Integrated Cluster of Clusters and the Network and Information Center of Dalian University of Technology for computational resources. Metrical parameters for the structures of compounds 2 to 6 are available free of charge from the Cambridge Crystallographic Data Centre under reference nos. CCDC-937384 to 937388. Correspondence on DFT calculations should be sent to Y.L. ( Z.H., T.S., and S.H. conceived and designed the experiments. S.H. carried out most of the experiments. T.S. carried out the x-ray analyses and part of the experiments. G.L., X.K., and Y.L. performed the DFT calculations. Z.H., T.S., and S.H. analyzed the data and co-wrote the manuscript. Z.H. directed the project.
View Abstract

Navigate This Article