Article | Open

Remarkable catalytic activity of dinitrogen-bridged dimolybdenum complexes bearing NHC-based PCP-pincer ligands toward nitrogen fixation

  • Nature Communications 8, Article number: 14874 (2017)
  • doi:10.1038/ncomms14874
  • Download Citation
Received:
Accepted:
Published online:

Abstract

Intensive efforts for the transformation of dinitrogen using transition metal–dinitrogen complexes as catalysts under mild reaction conditions have been made. However, limited systems have succeeded in the catalytic formation of ammonia. Here we show that newly designed and prepared dinitrogen-bridged dimolybdenum complexes bearing N-heterocyclic carbene- and phosphine-based PCP-pincer ligands [{Mo(N2)2(PCP)}2(μ-N2)] (1) work as so far the most effective catalysts towards the formation of ammonia from dinitrogen under ambient reaction conditions, where up to 230 equiv. of ammonia are produced based on the catalyst. DFT calculations on 1 reveal that the PCP-pincer ligand serves as not only a strong σ-donor but also a π-acceptor. These electronic properties are responsible for a solid connection between the molybdenum centre and the pincer ligand, leading to the enhanced catalytic activity for nitrogen fixation.

Introduction

Nitrogen is an essential element for all living things on earth. Since most of the nitrogen atoms on earth exist as the form of inert dinitrogen gas, the fixing of molecular dinitrogen is necessary to be utilized. The industrial dinitrogen fixation system, called the Haber–Bosch process, plays an important role in producing ammonia from dinitrogen gas today1. The operation of the process, however, requires high temperature and high pressure, resulting in large consumption of fossil fuels1. On the other hand, nitrogenases transform dinitrogen gas into ammonia under ambient reaction conditions, where the active sites of nitrogenases include iron, molybdenum and vanadium as essential transition metals2,3,4,5. Studies on the active sites of nitrogenases are considered to be important to elaborate an efficient artificial system for the catalytic ammonia formation from dinitrogen gas6,7,8.

Despite intensive efforts for the transformation of dinitrogen gas using transition metal–dinitrogen complexes as catalysts under mild reaction conditions9,10,11,12,13,14,15,16, only a few systems have succeeded in the catalytic formation of ammonia from dinitrogen gas17,18,19,20,21,22,23,24,25,26,27,28. In 2003, Yandulov and Schrock29,30,31,32 reported the first successful example of the catalytic conversion of dinitrogen gas into ammonia using a molybdenum–dinitrogen complex as a catalyst and in 2013 Peters and co-workers33,34,35,36,37,38,39 reported the iron-catalysed transformation using an iron–dinitrogen complex as a catalyst. We also found that several dinitrogen-bridged dimolybdenum complexes such as [{Mo(N2)2(PNP)}2(μ-N2)] (2; PNP=2,6-bis(di-tert-butylphosphinomethyl)pyridine)40,41,42 and molybdenum–nitride complexes bearing PNP-type pincer ligands43 or mer-tridentate triphosphine44 worked as more effective catalysts towards ammonia formation under ambient reaction conditions, where up to 63 equiv. of ammonia were produced based on the molybdenum atom of the catalyst.

During our continuous study, we have realized three promising clues to develop more effective catalysts. The first clue is the introduction of an electron-donating group to the pincer ligands to increase the backdonating ability of the molybdenum atom to the dinitrogen ligand. In fact, dinitrogen-bridged dimolybdenum complexes bearing the electron-donating group-substituted PNP-pincer ligands worked as more effective catalysts in our previous reaction system41. The second clue is the inhibition of the dissociation of the pincer ligand from the molybdenum atom to increase the stability of the molybdenum complex. We generally observed the dissociation of the PNP-pincer ligand after the ammonia formation in the catalytic reaction40. The third clue is the preservation of the dinitrogen-bridged dimolybdenum core to promote the catalytic ammonia formation from the coordinated dinitrogen43. Density functional theory (DFT) calculations demonstrated that the dinitrogen-bridged dimolybdenum structure plays a vital role in the protonation of a dinitrogen ligand, where one molybdenum moiety of the dinuclear molybdenum–dinitrogen complex works as a mobile ligand to the other molybdenum moiety as an active site43.

Taking account of these clues, we have now planned to design an N-heterocyclic carbene- (NHC-)45,46 and phosphine-based PCP-type pincer ligand (a PCP-type pincer ligand composed of NHC and two phosphines) as a tridentate ligand in place of the so far employed PNP-type pincer ligand for preparing a new molybdenum–dinitrogen complex. It is known that NHC works as a stronger electron-donating ligand than pyridine and binds to a transition metal centre more strongly than pyridine47,48. In this article, we demonstrate that dinitrogen-bridged dimolybdenum complexes bearing PCP-pincer ligands [{Mo(N2)2(PCP)}2(μ-N2)] (1) worked as effective catalysts towards ammonia formation under ambient reaction conditions, where up to 230 equiv. of ammonia were produced based on the catalyst (115 equiv. of ammonia based on the molybdenum atom of the catalyst). This is so far the most effective catalytic reduction of dinitrogen gas into ammonia under ambient reaction conditions using transition metal–dinitrogen complexes as catalysts.

Results

Preparation and characterization of 1

On the basis of our proposal, we designed two types of dinitrogen-bridged dimolybdenum complexes bearing PCP-type pincer ligands with two tert-butyl groups on each phosphorus atom (Fig. 1a). One is the complex bearing methylene linkers between the NHC skeleton and the phosphorus atom, where a similar PCP-pincer ligand bearing two phenyl groups on each phosphorus atom has quite recently been reported by Rieger and co-workers49. The other is the complex bearing ethylene linkers, where similar PCP-pincer ligands bearing two phenyl groups on each phosphorus atom have already been reported by some research groups50,51,52,53,54,55.

Figure 1: Design and synthesis of dinitrogen-bridged dimolybdenum complexes.
Figure 1

(a) Newly designed PCP-type pincer ligands (PCP, phosphorus–carbene–phosphorus). (b) Synthesis of dinitrogen-bridged dimolybdenum complexes 1a1c. (c) ORTEP drawings of 1a (left) and 1b (right). Thermal ellipsoids are shown at the 50% probability level. Hydrogen atoms and solvated molecules are omitted for clarity.

According to our previous procedure40,41,42, we newly prepared three dinitrogen-bridged dimolybdenum complexes bearing the PCP-type pincer ligands. Treatment of [MoCl3(PCP)] (3a, PCP=1,3-bis((di-tert-butylphosphino)methyl)benzimidazol-2-ylidene (Bim-PCP[1]); 3b, PCP=1,3-bis(2-(di-tert-butylphosphino)ethyl)imidazol-2-ylidene (Im-PCP[2]); 3c, PCP=5,6-dimethyl-1,3-bis((di-tert-butylphosphino)methyl)benzimidazol-2-ylidene (Me-Bim-PCP[1])) with 6 equiv. of Na–Hg in tetrahydrofuran (THF) at room temperature for 17 h under an atmospheric pressure of molecular dinitrogen gave the corresponding dinitrogen-bridged dimolybdenum complexes [{Mo(N2)2(PCP)}2(μ-N2)] (1a; PCP=Bim-PCP[1], 1b; PCP=Im-PCP[2], 1c; PCP=Me-Bim-PCP[1]) in 37%, 53% and 46% yields, respectively (Fig. 1b). Detailed synthetic procedures for ligand precursors are included in Supplementary Methods and synthetic procedures for metal precursors 3a3c are included in Methods section. These dinitrogen-bridged dimolybdenum complexes were characterized by 1H and 31P{1H} NMR. Detailed molecular structures of these complexes 1a1c were determined by X-ray crystallography (Fig 1c for 1a and 1b; Supplementary Fig. 1 for 1c), which were similar to those of the dinitrogen-bridged dimolybdenum complexes bearing PNP-pincer ligands40,41,42. However, the bond lengths, the bond angles and dihedral angles were significantly different between 1a and 1b according to the nature of the linkers in the PCP-pincer ligands (Supplementary Tables 3 and 4). The bond lengths defined by Mo(1)-C(1) of 1a and 1b were 2.064(2) Å and 2.153(4) Å, respectively, and the bond angles defined by P(1)-Mo(1)-P(2) of 1a and 1b were 153.95(3)° and 163.18(8)°, respectively. On the other hand, the dihedral angles defined by N(1)-C(1)-Mo(1)-N(5) of 1a and 1b were 82.35(5)° and 43.74(7)°, respectively. The shortened bond length of Mo(1)-C(1) of 1a suggests the stronger π-backdonation from the molybdenum centre to the NHC unit due to the almost-perpendicular orientation of the NHC unit. On the other hand, the longer CH2 linker of Im-PCP[2] forces the twisted coordination of the NHC and is likely to weaken the π-backdonation in 1b. Further information on this topic is discussed based on DFT calculations (vide infra).

Infrared spectra of 1a1c in the solid state showed a strong absorption peak assignable to terminal dinitrogen ligands at 1,978, 1,911 and 1,969 cm−1 respectively. The single peak of each complex corresponds to the dinitrogen-bridged dimolybdenum structure, as determined by X-ray crystallography. Compared with the infrared spectrum of 1b, those of 1a and 1c showed the peak at much higher frequency due to the strong π-backdonation from the molybdenum centre to the NHC. The infrared spectra of 1a and 1c in THF solution showed one strong absorption peak assignable to terminal dinitrogen ligands at 1,979 and 1,973 cm−1, respectively. Comparison of the infrared spectra of 1a and 1c in the solid state with that in the solution state indicates that the dinitrogen-bridged dinuclear structures of 1a and 1c are preserved even in solution. Furthermore, the 15N{1H} NMR spectrum of 15N2-labelled 1a in C6D6 under 15N2 showed two singlet and one doublet signals; δ 7.2 (s, Mo–NN–Mo), −13.0 (d, 1JN–N=5.4 Hz, Mo–N≡N), −32.0 (br s, Mo–N≡N), which are consistent with the dinuclear structure (Fig. 2a)40,56. In contrast, the infrared spectrum of 1b in THF solution showed two peaks assignable to terminal dinitrogen ligands, suggesting that the structure of 1b in THF is no longer the same with that in the solid state. To obtain more information on real species of 1b in the THF solution, the 15N{1H} NMR spectrum of 1b was measured in THF-d8 solution under an atmospheric pressure of 15N2 gas. The spectrum showed two doublet and two double triplet signals; δ 1.2 (d, 1JN–N=6.0 Hz, Mo–N≡N(equatorial)), −20.8 (d, 1JN–N=6.1 Hz, Mo–N≡N(axial)), −27.8 (dt, 1JN–N=6.0 Hz and 2JN–P=1.6 Hz, Mo–N≡N(equatorial)), −30.5 (dt, 1JN–N=6.1 Hz and 2JN–P=2.3 Hz, Mo–N≡N(axial); Fig. 2b), demonstrating the formation of the corresponding mononuclear dinitrogen complex [Mo(N2)3(Im-PCP[2])] (1b′) in THF. In fact, these spectroscopic features of 1b′ in THF are consistent with those of similar mononuclear dinitrogen complexes such as mer-[Mo(N2)3L3] structures57,58. The instability of 1b in solution may be derived from the steric repulsion between the two molybdenum moieties bearing Im-PCP[2].

Figure 2: 15N{1H} NMR spectra of 1a and 1b.
Figure 2

(a) 15N{1H} NMR spectrum of 15N2-1a in C6D6 under 15N2. (b) 15N{1H} NMR spectrum of 15N2-1b′ in THF-d8 under 15N2.

As shown in Fig. 1, we prepared two types of the PCP-pincer ligands, where Bim-PCP[1] has a benzimidazol-2-ylidene skeleton and Im-PCP[2] has an imidazol-2-ylidene skeleton. Unfortunately, we were unable to synthesize the corresponding two PCP-pincer ligands based on the same NHC skeleton bearing linkers of the different lengths between the NHC skeleton and each phosphorus atom (Bim-PCP[2] and Im-PCP[1] in Fig. 1a). However, we consider that the presence or absence of benzene ring has little influence on either the thermodynamic stability of the Mo–N≡N–Mo structures or the electron-donating ability of the pincer ligands for the following reasons. To assess the influence of the benzene moiety in Bim-PCP[1] on the thermodynamic stability of the dinitrogen-bridged dimolybdenum structure, we calculated a model complex [{Mo(N2)2(Im-PCP[1])}2(μ-N2)], where the benzimidazol-2-ylidene skeleton in Bim-PCP[1] is replaced by the imidazol-2-ylidene skeleton. The optimized distance of the Mo–N2(bridging) bond and its bond dissociation energy (BDE) are 2.105 Å and 18.4 kcal mol−1, respectively, both of which are almost identical to those calculated for 1a (2.108 Å and 18.8 kcal mol−1). The definition of BDE is described in the Methods section. On the other hand, Tuczek and co-workers52,53 have previously prepared two PCP ligands based on benzimidazol-2-ylidene and imidazol-2-ylidene, where these ligands have similar σ-donating ability. Gusev59,60 has previously estimated the donor ability of various NHC ligands based on the computational evaluation of νCO (A1) of [Ni(CO)3(NHC)]. The author showed that 1,3-dimethylbenzimidazol-2-ylidene and 1,3-dimethylimidazol-2-ylidene have almost the same values of νCO (2,057 and 2,054 cm−1, respectively), suggesting that the electron-donating ability of the NHC ligands was scarcely influenced by the difference between benzimidazol-2-ylidene and imidazol-2-ylidene. We therefore expect that the introduction of a benzene ring to the NHC skeleton little influences either the thermodynamic stability of the Mo–N≡N–Mo structure or the electron-donating ability of the pincer ligand.

We then performed DFT calculations to elucidate how the length of linkers connecting the NHC skeleton with the PtBu2 groups in Bim-PCP[1] and Im-PCP[2] influences the thermodynamic stability of the dinitrogen-bridged dimolybdenum structures in 1a and 1b. Figure 3 shows optimized structures of dimolybdenum complexes 1a and 1b. The N≡N stretching frequencies of terminal dinitrogen ligands calculated for 1a (2,012 cm−1) and 1b (1,969 cm−1) reproduced the experimental trend. The Mo–N and N–N distances of a terminal dinitrogen ligand are calculated to be 2.032 and 1.137 Å for 1a and 2.016 and 1.142 Å for 1b, respectively, indicating that the terminal dinitrogen ligand in 1b is more activated than that in 1a. As a result, the BDE of a Mo–N2(terminal) of 1a (11.9 kcal mol−1) is considerably lower than that of 1b (16.7 kcal mol−1). On the other hand, the Mo–N distance of the bridging dinitrogen ligand of 1a (2.108 Å) is shorter than that of 1b (2.133 Å). The BDE of the Mo–N2(bridging) bond of 1a is 18.8 kcal mol−1, which is more than twice as high as that of 1b, 9.0 kcal mol−1. The very low BDE of the Mo–N2(bridging) bond of 1b can be associated with the experimental observation that the dinuclear complex 1b is labile to be separated into two mononuclear complexes in solution.

Figure 3: Optimized structures of dinuclear complexes.
Figure 3

Bond distances are shown in Å. The values of νNN present the N≡N stretching frequencies of terminal dinitrogen ligands. Hydrogen atoms are omitted for clarity.

Differences in thermodynamic stability of the Mo–N≡N–Mo structure between 1a and 1b can be rationalized by optimized structures of the corresponding mononuclear dinitrogen complexes. Figure 4 presents space-filling models of [Mo(N2)3(PCP)] (1a′; PCP=Bim-PCP[1], 1b′; PCP=Im-PCP[2]). Comparison of the Mo–N distance for the equatorial dinitrogen ligand of 1a′ (2.084 Å) with that of 1b′ (2.041 Å) suggests that 1b′ bearing Im-PCP[2] strongly binds dinitrogen at the equatorial position. Contrary to the BDEs of the Mo–N2(bridging) bond calculated for dinuclear complexes 1a and 1b, the BDE of the Mo–N2(equatorial) bond of 1b′ (21.5 kcal mol−1) is almost the same with that of 1a′ (21.2 kcal mol−1). The dramatic decrease in the Mo–N2(equatorial) BDE of 1b′ in the formation of the dinuclear structure can be ascribed to steric hindrance caused by the tert-butyl groups on each phosphorus atom in Im-PCP[2]. The optimized structure of 1a′ bearing Bim-PCP[1] with the methylene linkers has the P–Mo–P bond angle of 151.2°, while that of 1b′ bearing Im-PCP[2] with the ethylene linkers has the bond angle of 164.3°. The extension of the CH2 linkers in 1b′ forces the tert-butyl groups on each phosphorus atom in Im-PCP[2] to project towards the space surrounding the equatorial dinitrogen ligand (Fig. 4). As a result, the formation of a thermodynamically stable dimolybdenum complex bearing Im-PCP[2] is encumbered by steric repulsions between tert-butyl groups in two mononuclear molybdenum units facing each other.

Figure 4: Space-filling models of mononuclear Mo–N2 complexes 1a′ and 1b′.
Figure 4

The dashed lines represent the projection of tert-butyl groups on phosphine atoms.

Catalytic nitrogen fixation using 1 as catalysts

The catalytic reduction of molecular dinitrogen into ammonia using 1 as catalysts was carried out according to the following procedure of the previous method40,41,42. To a mixture of 1 and 2,6-lutidinium trifluoromethanesulfonate (96 equiv. to 1; [LutH]OTf) as a proton source in toluene was added a solution of metallocene (72 equiv. to 1) as a reductant in toluene via a syringe pump at room temperature over a period of 1 h, followed by stirring at room temperature for another 19 h under an atmospheric pressure of dinitrogen. After the reaction, the amounts of ammonia and molecular dihydrogen were determined by indophenol method61 and gas chromatography (GC), respectively. The yields of ammonia and molecular dihydrogen were calculated based on the metallocene. Typical results are shown in Table 1. In all cases, no formation of other products such as hydrazine was observed at all.

Table 1: Catalytic formation of ammonia from dinitrogen gas employing 1a or 1b as a catalyst.

First, we carried out the catalytic reaction in the presence of 1a as a catalyst using either cobaltocene (CoCp2; Cp=η5–C5H5), decamethylchromocene (CrCp*2; Cp*=η5–C5Me5), and decamethylcobaltocene (CoCp*2) as reductants, to produce 5.7, 17.6 and 11.8 equiv. of ammonia based on the catalyst, respectively (Table 1, runs 1–3). In the absence of a reductant, only 0.2 equiv. of ammonia were produced based on 1a (Table 1, run 4). We have previously obtained the result that 12.2 equiv. of ammonia were produced based on the catalyst from the reaction with CrCp*2 as a reductant in the presence of [{Mo(N2)2(PNP)}2(μ-N2)] 2 as a catalyst (Table 1, run 9)40. This means that 1a promoted the catalytic nitrogen fixation more effectively than 2. Separately, we confirmed the direct conversion of molecular dinitrogen into ammonia by using 15N2 gas in place of normal 14N2 gas (see Supplementary Methods for detailed procedure).

In stark contrast to the catalytic activity of 1a, 1b did not work so effectively towards the formation of ammonia under the same reaction conditions. When CoCp2, CrCp*2 and CoCp*2 were used as reductants, only 1.4, 3.2 and 2.9 equiv. of ammonia were produced based on the catalyst, respectively (Table 1, runs 5–7). In the absence of a reductant, 1.5 equiv. of ammonia were produced based on 1b (Table 1, run 8).

Next, we investigated the influence of a proton source in the catalytic nitrogen fixation using 1a as a catalyst. Typical results are shown in Table 2, where larger amounts of both reductant and proton source were employed in order to sharpen the difference among the results (see Supplementary Methods for the detailed procedure). The catalytic reaction using larger amounts of reductant CrCp*2 (360 equiv. to 1a) and proton source [LutH]OTf (480 equiv. to 1a) afforded 79 equiv. of ammonia based on the catalyst (Table 2, run 1). When 2-picolinium trifluoromethanesulfonate ([PicH]OTf; Pic=2-picoline) was used in place of [LutH]OTf, only a small amount of ammonia was produced based on the catalyst (Table 2, run 2). On the other hand, 2,4,6-collidinium trifluoromethanesulfonate ([ColH]OTf; Col=2,4,6-collidine) worked rather effectively, where 61 equiv. of ammonia were produced (Table 2, run 3). When a non-coordinating anion BArF4 (ArF=3,5-(CF3)2C6H3) was used in place of OTf in [LutH]OTf, only a small amount of ammonia was produced (Table 2, run 4). These results indicate that the use of [LutH]OTf as a proton source is an essential factor to achieve the high performance of 1a as a catalyst.

Table 2: Catalytic formation of ammonia from dinitrogen gas using 1a as a catalyst.

On the basis of the results shown in Tables 1 and 2, we carried out the catalytic reaction using much larger amounts of CrCp*2 and [LutH]OTf as a reductant and a proton source to the catalyst, respectively (see Supplementary Methods for the detailed procedure). The reaction using 1,440 equiv. of CrCp*2 as a reductant and 1,920 equiv. of [LutH]OTf as a proton source in the presence of 1a as a catalyst under ambient reaction conditions gave 2.0 × 102 equiv. of ammonia based on the catalyst (Fig. 5). The catalytic activity of 1a towards the formation of ammonia is an order of magnitude greater than that of [{Mo(N2)2(PNP)}2(μ-N2)] 2 (up to 23 equiv. of ammonia based on 2). Ammonia was obtained in 42% yield together with molecular dihydrogen (14% yield). Furthermore, a higher catalytic activity has been achieved when 1c was used as a catalyst, where up to 2.3 × 102 equiv. of ammonia based on the catalyst were produced under similar reaction conditions (Fig. 5). We have not yet obtained the exact reason why 1c worked as the more effective catalyst than 1a, but we consider that the introduction of two methyl groups to the benzimidazol-2-ylidene skeleton in the PCP-pincer ligand may increase the backdonating ability of molybdenum centres to the coordinated dinitrogen ligand and activated the terminal dinitrogen ligands more strongly than 1a. The infrared spectrum of 1c in the solid state showed a strong absorption peak assignable to terminal dinitrogen ligands at 1,969 cm−1, which is lower than that of 1a (vide supra). Previously, we reported that the introduction of electron-donating groups such as methyl and methoxy groups to the pyridine ring of the PNP-pincer ligand in 2 markedly enhanced the catalytic activity under the same reaction conditions41.

Figure 5: Catalytic formation of ammonia using larger amounts of a reductant and a proton source in the presence of 1a or 1c as a catalyst.
Figure 5

A solution of CrCp*2 in 5 ml of toluene was added to a mixture of the catalyst and [LutH]OTf in 1 ml of toluene at room temperature over a period of 1 h (for 1a) or 5 h (for 1c), followed by stirring at room temperature for another 19 h (for 1a) or 15 h (for 1c) under 1 atm of dinitrogen gas. The amounts of ammonia and hydrogen (equiv.) are based on the catalyst. Yields are based on CrCp*2.

The time profile of the catalytic reactions using 1a and 1c as catalysts was monitored (see Supplementary Methods for the detailed procedure). Typical results are shown in Fig. 6 together with the time profile using 2 as a catalyst41,42. The turnover frequency (TOF) for ammonia formation, which was determined as mols of ammonia (based on the catalyst) produced in initial 1 h, was 42 h−1 for 1a and 53 h−1 for 1c. The TOFs for ammonia using 1a and 1c are ca. 2.5 and 3.1 times larger than that using 2 (17 h−1), respectively. This result indicates that the dinitrogen-bridged dimolybdenum complexes bearing PCP[1]-type pincer ligands such as 1a and 1c have the effective performance not only on the catalytic activity but also on the rate for ammonia formation.

Figure 6: Time profiles of the formation of ammonia from dinitrogen gas.
Figure 6

A solution of CrCp*2 (0.72 mmol) in toluene was added to a mixture of 1a or 1c (0.0033, mmol) and [LutH]OTf (0.96 mmol) at room temperature over a period of 1 h under 1 atm of dinitrogen gas, followed by stirring for the indicated time (0.33, 0.67, 1, 2 and 20 h). A solution of CoCp2 (2.16 mmol) in toluene was added to a mixture of 2 (0.010 mmol) and [LutH]OTf (2.88 mmol) at room temperature over a period of 1 h under 1 atm of dinitrogen gas, followed by stirring for the indicated time (0.33, 0.67, 1, 2 and 20 h). The amount of ammonia (equiv.) is based on the catalyst.

For comparison of the stability of the dinitrogen-bridged dimolybdenum complexes bearing PCP-pincer ligands with that of PNP-pincer ligands, we carried out the following catalytic reactions using 1a and 2 as catalysts. After the formation of 47 equiv. of ammonia based on the catalyst from molecular dinitrogen following the same procedure of the time profile experiment using 1a, the same amounts of [LutH]OTf and CrCp*2 were further added at room temperature, and the mixture was stirred for another 2 h to afford a total of 69 equiv. of ammonia based on the catalyst (Fig. 7; see Supplementary Methods for the detailed procedure). In this reaction system, 22 equiv. of ammonia were produced from a further reaction of molecular dinitrogen with excess amounts of a proton source and a reductant. This experimental result indicates that the active species derived from 1a remain even after the catalytic reaction. In fact, no free PCP-pincer ligand was observed from the reaction mixture after the catalytic reaction using 1a as a catalyst, suggesting that the active species derived from 1a were still active. In sharp contrast, no additional ammonia was produced from similar treatment using 2 as a catalyst (Fig. 7), where free PNP-pincer ligand was observed from the reaction mixture after the catalytic reaction using 2 as a catalyst40. These results indicate that the stability of 1a is much improved compared with that of 2.

Figure 7: Reactions of further addition of proton source and reductant.
Figure 7

‘NH3 first’ and ‘NH3 second’ were collected in separated runs. Each ‘NH3 first’ is the same value as the time profile experiment. Each ‘NH3 second’ is collected by the following procedure. A solution of a reductant (CrCp*2 for 1a and CoCp2 for 2; 216 equiv) in toluene (4 ml) was added to a mixture of 1a or 2 (0.0033, mmol for 1a and 0.010 mmol for 2) and [LutH]OTf (288 equiv) at room temperature over a period of 1 h under 1 atm of dinitrogen gas, followed by stirring for 20 h. Then, [LutH]OTf (288 equiv) was added in one portion and another solution of the same reductant (216 equiv) in toluene (4 ml) was added over a period of 1 h, followed by stirring at room temperature for another 1 h under 1 atm of dinitrogen gas. The difference between the amount of ammonia obtained in this experiment and the ‘NH3 first’ is the ‘NH3 second’. The amount of ammonia (equiv.) is based on the catalyst.

Comparison of PCP and PNP ligands

In this section, we compare the electronic properties and reactivity of 1a and 2. We have previously reported that the catalytic activity of 2 was improved by the introduction of electron-donating groups to the 4-position of the pyridine ring in PNP41. In this report, DFT calculations demonstrated that the introduction of electron-donating groups to PNP enhances the backdonating ability of molybdenum centres and thereby leads to activation of dinitrogen ligands. As described in the Introduction, the NHC-based PCP ligand was expected to work as a strong electron donor to activate dinitrogen ligands coordinated to the molybdenum centre. For understanding the geometric and electronic structures of 1a and 2 upon the coordination of the pincer ligands, mononuclear molybdenum complexes 1a′ and 2′ were investigated in detail.

Figure 8a compares the electron-donating ability of Bim-PCP[1] and PNP in terms of differences in atomic charge (Δq) between the dinitrogen complexes (1a′ and 2′) and the free ligands (Bim-PCP[1] and PNP). The charges of the Mo atom and three N2 ligands obtained with the natural population analysis (NPA)62 were set to zero for the free ligands, and hence the sum of the charges is identical to the Δq value of the Mo(N2)3 moiety. The gross NPA charges of the Mo(N2)3 moiety in 1a′ and 2′ can be regarded as the amount of electron donated from the pincer ligands during complexation. To evaluate the electron-donating ability of Bim-PCP[1] and PNP, the Δq values of the PtBu2 groups and the carbene or pyridine moiety containing the methylene linkers are separately given in Fig. 8a. The Δq values of the PtBu2 groups are identical in both Bim-PCP[1] and PNP (+0.29), indicating that the electron-donating ability of Bim-PCP[1] and PNP is controlled by the carbene and pyridine moieties. Since the Δq values of the carbene and pyridine moieties are +0.23 and +0.12, respectively, the pincer ligands donate 0.81e (Bim-PNP[1]) and 0.70e (PNP) to the Mo(N2)3 moiety during complexation. As we expected, the NHC-based pincer ligand exhibits a stronger electron-donating ability than the pyridine-based one from a viewpoint of atomic charge.

Figure 8: Electronic properties of mononuclear molybdenum–dinitrogen complexes.
Figure 8

(a) Changes in the NPA atomic charge (Δq) in the coordination of the pincer ligands to the Mo(N2)3 moiety. The values of Δq are obtained as differences between mononuclear Mo–N2 complexes (1a′, 2′ and 1b′) and free ligands (Bim-PCP[1] for 1a′, PNP for 2′ and Im-PCP[2] for 1b′). (b) Spatial distribution of frontier orbitals of 1a′ and 2′ that contribute to σ donation from the pincer ligand to Mo. (c) Spatial distribution of frontier orbitals of 1a′ and 2′ that contribute to π back donation from Mo to both equatorial and axial dinitrogen ligands. The molecular structures are rotated by 90° along the Mo–N2(equatorial) bond from those in Fig. 8b. (d) A schematic drawing of the bonding interactions between the Mo atom and the carbene C atom in 1a′.

Optimized structures and BDEs between the molybdenum centre and dinitrogen ligands also reflect the strength of the electron-donating ability of Bim-PCP[1] and PNP. Table 3 summarizes geometric parameters around the molybdenum centre in mononuclear molybdenum–dinitrogen complexes 1a′ and 2′, together with the BDEs of the Mo–N2(axial) and Mo–N2(equatorial) bonds. The Mo-C(carbene) bond distance (2.099 Å) in 1a′ is significantly shorter than the Mo–N(pyridine) distance (2.240 Å) in 2′. The Mayer bond order (b.o.)63 of the Mo-C(carbene) bond is calculated to be 0.91, which is much larger than that of the Mo–N(pyridine) bond (0.39). The presence of a strong bonding interaction between the molybdenum centre and Bim-PCP[1] is consistent with the experimental fact that 1a bearing Bim-PCP[1] works as a long-lived catalyst for the catalytic nitrogen fixation with a high turnover number compared with 2 bearing PNP (vide supra). On the other hand, the strong trans influence of the carbene ligand weakens the Mo–N2(equatorial) bond in 1a′. The Mo–N2(equatorial) bond distance of 2.084 Å (b.o.=0.50) in 1a′ is much longer than that of 2.018 Å (b.o.=0.62) in 2′, and the BDEs of the Mo–N2(equatorial) bond are 21.2 kcal mol−1 for 1a′ and 30.1 kcal mol−1 for 2′. Interestingly, the coordination of Bim-PCP[1] to molybdenum would influence all dinitrogen ligands at both trans- and cis-positions so as to weaken all the Mo–N2 bonds. The Mo–N2(axial) bond distances (b.o.) are calculated to be 2.034 Å (0.53) for 1a′ and 2.024 Å (0.54) for 2′. The BDE of the Mo–N2(axial) bond of 1a′ (12.5 kcal mol−1) is also lower than that of 2′ (14.0 kcal mol−1). A similar trend was observed for the Mo–N2(bridging) and Mo–N2(terminal) bonds in dimolybdenum–dinitrogen complexes 1a and 2. The bond dissociation energies are 18.8 and 11.9 kcal mol−1 for 1a, both of which are smaller than those obtained for 2 (24.9 and 14.4 kcal mol−1).

Table 3: Selected bond distances in Å and Mo–N2 BDEs in kcal mol−1 of mononuclear molybdenum–dinitrogen complexes.

The origin of the weaker Mo–N2 bonds in 1a′ and 1a is understood by looking at frontier orbitals responsible for the bonding between the molybdenum centre and the carbene C atom of Bim-PCP[1]52,53,64,65. As depicted in Fig. 8b, the HOMO-6 (1a′) and HOMO-5 (2′) contribute to a σ-bond between the Mo atom and the carbene C atom (or the pyridine N atom). The large size of the lobe between the Mo and C atoms indicates that Bim-PCP[1] works as a strong σ-donor compared to PNP. The HOMO-1 in Fig. 8c mainly contributes to π-backdonation from an out-of-plane d orbital of Mo to a π* orbital of dinitrogen ligands. The backdonation from metal to dinitrogen is essential for the activation of dinitrogen upon the formation of metal–dinitrogen complexes. Occupation of the HOMO-1 strengthens all of the Mo–N2 bonds in 1a′ and 2′ because of their symmetrical structures. By comparing the HOMO-1 of 1a′ and 2′, one can find a bonding interaction between the Mo atom and the carbene C atom through π-backdonation from the d orbital of Mo to the vacant p orbital of C perpendicular to the carbene ring in 1a′. This backdonation decreases the amount of electron transferred to both the equatorial and axial dinitrogen ligands, leading to the lower BDEs of the Mo–N2 bonds in 1a′ (1a). As presented in Fig. 8a, the Δq value of the dinitrogen ligands in 1a′ (−0.26) is smaller than that in 2′ (−0.31) in spite of the electron-donating ability of Bim-PCP[1] superior to PNP. The backdonation from the Mo atom to the carbene C atom also contributes to the strong binding of Bim-PCP[1] to Mo.

On the other hand, the Mo-C bond distance (b.o.) in 1b′ bearing Im-PCP[2] (2.178 Å (0.79)) indicates that the Mo-C bond in 1b′ is weaker than that in 1a′, as summarized in Table 3. Owing to the longer CH2 linkers, the coordination of the carbene moiety to the molybdenum centre in 1b′ is highly twisted compared to 1a′; the dihedral angles of N(1)-C(1)-Mo(1)-N(3) are 69.6° for 1a′ and 43.8° for 1b′ (Supplementary Fig. 21). The twisted coordination of the carbene moiety in 1b′ reduces the overlap between the out-of-plane d orbital of the Mo atom and the vacant p orbital of the carbene C atom. As a result, Im-PCP[2] works only as a very strong σ-donor (0.90e donation to Mo; Fig. 8a). The gross NPA charge on the dinitrogen ligands (−0.36) as well as the large BDE of the Mo–N2(axial) bond of 1b′ (14.3 kcal mol−1) implies that the coordination of Im-PCP[2] to the molybdenum centre effectively activates the coordinated dinitrogen ligands. However, we theoretically confirmed that the mononuclear molybdenum–dinitrogen complexes such as 1b′ and [cis-Mo(N2)2(Im-PCP[2])] cannot be protonated by LutH+, similar to 1a′ and [Mo(N2)3(PNP)]43. All attempts to optimize a product complex comprise the protonated 1a′ (1b′), and Lut resulted in formation of a reactant complex comprising 1a′ (1b′) and LutH+, even though the optimization started from a structure with the N2…H+ distance of 5 Å. Thus, the lack of the catalytic activity of 1b for nitrogen fixation can be attributed to the thermodynamic instability of the dinitrogen-bridged dimolybdenum structure, as mentioned in the former section.

On the basis of the catalytic reaction pathway previously proposed for nitrogen fixation using 2 (ref. 43), we have theoretically investigated possible reaction pathways catalysed by 1a. In the present article, we particularly focus on the first protonation process shown in Fig. 9, since the protonation of a terminal dinitrogen ligand in 2 by [LutH]OTf is energetically the most unfavourable process in the catalytic cycle43. In the calculated reaction pathway, a terminal dinitrogen ligand in 1a is first protonated by LutH+ (1aA-PCP), and then the dinitrogen ligand trans to the generated NNH group is eliminated (A-PCP→B-PCP). The protonation of 1a yielding A-PCP is endothermic by 8.1 kcal mol−1 with an activation energy of 8.3 kcal mol−1. This energy profile indicates that proton detachment from A-PCP can easily occur like the PNP system. On the other hand, the following N2 elimination yielding B-PCP is exothermic by 5.2 kcal mol−1 with a low activation energy of 4.0 kcal mol−1. The coordination of OTf to B-PCP is highly exothermic by 20.7 kcal mol−1, and thus the whole reaction pathway leading to C-PCP is energetically downhill. Comparison of the energy profiles of the PCP and PNP systems suggests that the reactivity of the dinitrogen complexes 1a and 2 with [LutH]OTf would not be a major factor for rationalizing the high catalytic activity of 1a.

Figure 9: A possible reaction pathway and energy profiles of the first protonation process on a terminal dinitrogen ligand in 1a (highlighted in yellow) and 2.
Figure 9

Energy changes (activation energies in parentheses) are presented in kcal mol−1.

Discussion

On the basis of our previous findings on the catalytic nitrogen fixation, we have newly designed and prepared novel dinitrogen-bridged dimolybdenum complexes bearing NHC and phosphine-based PCP-pincer ligands, Bim-PCP[1] and Im-PCP[2]. The dimolybdenum–dinitrogen complexes bearing Bim-PCP[1] as PCP-pincer ligands have been found to work as so far the most effective catalysts towards the ammonia formation from molecular dinitrogen under ambient reaction conditions, where up to 230 equiv. of ammonia were produced based on the catalyst (115 equiv. of ammonia based on the molybdenum atom of the catalyst). The superior activity of dimolybdenum–dinitrogen complexes bearing Bim-PCP[1] included the high TOF for ammonia formation and the catalyst stability. DFT calculations on 1 reveal that Bim-PCP[1] as a PCP-pincer ligand serves as not only a strong σ-donor but also a π-acceptor. These electronic properties are responsible for a solid connection between the molybdenum centre and the pincer ligand, leading to the enhanced catalytic activity for nitrogen fixation.

Methods

General information

Detailed experimental procedures, characterization of compounds and the computational details can be found in the Supplementary Figs 1–21, Supplementary Tables 1–10 and Supplementary Methods. Cartesian coordinates are available in Supplementary Data 1.

Synthesis of [MoCl3(PCP)] (3a3c)

A typical procedure for the preparation of 3a is described below. To a mixed solid of 1,3-bis((di-tert-butylphosphino)methyl)-1H-benzo[d]imidazol-3-ium hexafluorophosphate (7a, 1.16 g, 2.00 mmol) and KN(SiMe3)2 (559 mg, 2.80 mmol) was added toluene (40 ml), and the resulting suspension was stirred for 20 min at room temperature. [MoCl3(thf)3] (733 mg, 1.75 mmol) and toluene (15 ml) were added to the suspension and stirred at 80 °C for 19 h. The solvent was removed under vacuum, and the residue was washed with hexane (5 ml × 2), toluene (10 ml) and hexane (5 ml × 2). The solid was dried under vacuum. The solid was extracted with CH2Cl2 (10 ml × 1, 5 ml × 7), recrystallized from CH2Cl2–hexane and dried under vacuum to afford 3a·0.5CH2Cl2 (712 mg, 1.05 mmol, 60%). Anal. Calcd. for C25.5H45Cl4MoN2P2 (3a·0.5CH2Cl2): C, 45.08; H, 6.68; N, 4.12. Found: C, 45.41; H, 6.61; N, 4.44. Crystals suitable for preliminary X-ray analysis were prepared by recrystallizing from CH2Cl2–hexane to give 3a·CH2Cl2 as orange-brown crystals.

3b: Recrystallization from CH2Cl2–hexane gave 3b·0.5CH2Cl2 as orange crystals. 33% yield. Anal. Calcd. for C23.5H47Cl4MoN2P2 (3b·0.5CH2Cl2): C, 42.94; H, 7.21; N, 4.26. Found: C, 43.04; H, 7.44; N, 4.19. Crystals suitable for X-ray analysis were prepared by recrystallizing from 1,2-dichloroethane–hexane to afford 3b·1/3C6H14. The structure is included in Supplementary Fig. 2 and selected bond lengths and angles in 3b are included in Supplementary Table 6.

3c: Recrystallization from CH2Cl2–hexane afforded 3c·0.5CH2Cl2 as orange crystals. 48% yield. Anal. Calcd. for C27.5H49Cl4MoN2P2 (3c·0.5CH2Cl2): C, 46.69; H, 6.98; N, 3.96. Found: C, 46.58; H, 6.79; N, 4.07. The structure is included in Supplementary Fig. 3, and selected bond lengths and angles in 3c are included in Supplementary Table 7.

Synthesis of [{Mo(N2)2(PCP)}2(μ-N2)] (1a1c)

A typical procedure for the preparation of 1a is described below. To a suspension of Na–Hg (0.5 wt% Na, 13.8 g, 3.00 mmol) in THF was added [MoCl3(Bim-PCP[1])]·0.5CH2Cl2 (341 mg, 0.501 mmol), and the resulting suspension was stirred under atmospheric pressure of N2 at room temperature for 17 h. The supernatant suspension was filtered through Celite, and the solvent was removed under vacuum. The resulting solid was extracted with benzene (5 ml) and filtered through Celite. The filter cake was washed with benzene (2 ml × 9), and the solvent of the combined solution was removed under vacuum. The resulting solid was washed with pentane (2 ml × 20) to give 1a·1.3C4H8O·0.4C5H12 as a dark purple solid, where 1.3 equiv. of THF and 0.4 equiv. of hexane were determined by 1H NMR (141 mg, 0.102 mmol, 37%). Analytically pure sample was prepared by recrystallization from THF at −18 °C to afford 1a.

1a: 1H NMR (C6D6): δ 7.01–6.98 (m, 4H, ArH), 6.75–6.71 (m, 4H, ArH), 3.84 (s, 8H, NCH2P), 1.38 (pseudo t, 3JP–H=5.7 Hz, 72H, PtBu2). 31P{1H} NMR (C6D6): 105.6 (s, PtBu2). Infrared (KBr, cm−1): 1,978 (s, νNN). Infrared (THF, cm−1): 1,979 (s, νNN). Anal. Calcd. for C50H88Mo2N14P4: C, 50.00; H, 7.38; N, 16.33. Found: C, 49.66; H, 6.94; N, 14.18. The lower content of nitrogen is considered to be due to the labile property of the coordinated dinitrogen ligand in 1a under the analytical conditions.

1b: Recrystallization from benzene–hexane afforded 1b·2/3C6H14 as dark-brown crystals. 53% yield. Anal. Calcd. for C50H101.33Mo2N14P4 (1b·2/3C6H14): C, 49.44; H, 8.41; N, 16.14. Found: C, 49.82; H, 8.40; N, 15.35. 1H and 31P{1H} NMR were measured in THF-d8 as a mixture of 1b and 1b′. 1H NMR (THF-d8): 1b, δ 6.76 (s, 2H, NCHCHN), 1.44 (pseudo t, 3JP–H=11.2 Hz, PtBu2), 0.95 (pseudo t, 3JP–H=11.2 Hz, PtBu2): 1b′, δ 6.85 (s, 2H, NCHCHN), 4.29–4.19 (br m, 4H, NCH2), 1.98–1.91 (br m, 4H, CH2P), 1.18 (br s, 36H, PtBu2). 31P{1H} NMR (THF-d8): 1b, δ 71.1 (s, tBu2P): 1b′, δ 69.2 (s, tBu2P). Infrared (KBr, cm−1): 1,911 (s, νNN for 1b). Infrared (THF under N2, cm−1): 2,041 (m, νNN for 1b′), 1,945 (s, νNN for 1b′).

1c: Reprecipitation from THF–hexane afforded 1c·0.5C6H14 as a brown solid. 46% yield. Crystals suitable for X-ray analysis were prepared by recrystallization from THF at −18 °C to afford 1c. The structure is included in Supplementary Fig. 1, and selected bond lengths and angles are included in Supplementary Table 5. 1H NMR (C6D6): δ 6.68 (s, 4H, ArH), 3.89 (s, 8H, NCH2P), 2.26 (s, 12H, ArCH3), 1.41 (pseudo t, 3JP–H=5.5 Hz, 72H, PtBu2). 31P{1H} NMR (C6D6): δ 105.8 (s, PtBu2). Infrared (KBr, cm−1): 1,969 (s, νNN). Infrared (THF, cm−1): 1,973 (s, νNN). Anal. Calcd. for C57H103Mo2N14P4 (1c·0.5C6H14): C, 52.65; H, 7.98; N, 15.08. Found: C, 52.80; H, 7.70; N, 13.37.

Catalytic reduction of dinitrogen to ammonia under N2

In a 50 ml Schlenk flask were placed 1a·1.3C4H8O·0.4C5H12 (12.9 mg, 0.00970, mmol) and 2,6-lutidinium trifluoromethanesulfonate [LutH]OTf (247 mg, 0.960 mmol). Toluene (1.0 ml) was added under N2 (1 atm), and then a solution of CrCp*2 (232 mg, 0.719 mmol) in toluene (4.0 ml) was added to the stirred suspension in the Schlenk flask with a syringe pump at a rate of 4.0 ml h−1. After the addition of CrCp*2, the mixture was further stirred at room temperature for 19 h. The reaction mixture was evaporated under reduced pressure, and the distillate was trapped in dilute H2SO4 solution (0.5 M, 10.00 ml). Aqueous solution of potassium hydroxide (30 wt%; 5 ml) was added to the residue to fully liberate ammonia, and the mixture was distilled into another dilute H2SO4 solution (0.5 M, 10.00 ml). The amount of NH3 present in each of the H2SO4 solutions was determined by the indophenol method.61

Computational method

DFT calculations were performed to search all intermediates and transition structures on potential energy surfaces using the Gaussian 09 programme66. Similar to the previous study43, we adopted the B3LYP* functional, which is a reparametrized version of the B3LYP hybrid functional developed by Reiher et al.67. For optimization, the Stuttgart–Dresden pseudopotentials and 6–31G(d) basis sets were chosen for the Mo atom and the other atoms, respectively. To determine the energy profile of the first protonation process, we performed single-point energy calculations at the optimized geometries using the 6–311+G(d,p) basis sets in place of the 6–31G(d) basis sets. Zero-point energy corrections were applied for energy changes (ΔE) and activation energies (Ea) calculated for each reaction step. Solvation effects (toluene) were taken into account by using the polarizable continuum model in the single-point energy calculations68. More details are described in Supplementary Methods. Throughout the paper, the BDE of an Mo–N2 (terminal, axial or equatorial) bond is defined as the energy change for dissociation of the corresponding dative N2 ligand, for example, [{Mo(N2)2(PNP)}2(μ-N2)]→[{Mo(N2)(PNP)}-NN-{Mo(N2)2(PNP)}]+N2. The BDE of the Mo–N2 (bridging) bond is defined as the energy change for separation of a dimolybdenum complex into two mononuclear Mo–N2 complexes, such as [{Mo(N2)2(PNP)}2(μ-N2)]→[Mo(N2)3(PNP)]+cis-[Mo(N2)2(PNP)].

Data availability

The X-ray crystallographic coordinates for structures reported in this article have been deposited at the Cambridge Crystallographic Data Centre (CCDC), under deposition number CCDC 1482254 (1a), 1482255 (1b), 1482256 (1c), 1482258 (3b) and 1482259 (3c). These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. All other data are available from the authors upon reasonable request.

Additional information

How to cite this article: Eizawa, A. et al. Remarkable catalytic activity of dinitrogen-bridged dimolybdenum complexes bearing NHC-based PCP-pincer ligands toward nitrogen fixation. Nat. Commun. 8, 14874 doi: 10.1038/ncomms14874 (2017).

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  1. 1.

    Ammonia Synthesis Catalysts: Innovation and Practice World Scientific (2013).

  2. 2.

    Current status of structure function relationships of vanadium nitrogenase. Coord. Chem. Rev. 237, 23–30 (2003).

  3. 3.

    et al. Evidence for interstitial carbon in nitrogenase FeMo cofactor. Science 334, 940 (2011).

  4. 4.

    et al. X-ray emission spectroscopy evidences a central carbon in the nitrogenase iron-molybdenum cofactor. Science 334, 974–977 (2011).

  5. 5.

    , , & Radical SAM-dependent carbon insertion into the nitrogenase M-cluster. Science 337, 1672–1675 (2012).

  6. 6.

    , , , & Mechanism of nitrogen fixation by nitrogenase: the next stage. Chem. Rev. 114, 4041–4062 (2014).

  7. 7.

    & Recent advances in the chemistry of dinitrogen complexes. Chem. Rev. 95, 1115–1133 (1995).

  8. 8.

    & Dinitrogen coordination chemistry: on the biomimetic borderlands. Chem. Rev. 104, 385–402 (2004).

  9. 9.

    , , & Reactivity of dinitrogen bound to mid- and late-transition-metal centers. Eur. J. Inorg. Chem. 2015, 567–598 (2015).

  10. 10.

    Recent progress in transition-metal-catalyzed reduction of molecular dinitrogen under ambient reaction conditions. Inorg. Chem. 54, 9234–9247 (2015).

  11. 11.

    , & Interplay between theory and experiment for ammonia synthesis catalyzed by transition metal complexes. Acc. Chem. Res. 49, 987–995 (2016).

  12. 12.

    , , , & Binding of dinitrogen to an iron-sulfur-carbon site. Nature 526, 96–99 (2015).

  13. 13.

    & A terminal N2 complex of high-spin iron(I) in a weak, trigonal ligand field. J. Am. Chem. Soc. 137, 8940–8943 (2015).

  14. 14.

    et al. Dinitrogen activation upon reduction of a triiron(II) complex. Angew. Chem. Int. Ed. 54, 1499–1503 (2015).

  15. 15.

    , , , & Metal-mediated production of isocyanates, R3EN=C=O from dinitrogen, carbon dioxide, and R3ECl. Angew. Chem. Int. Ed. 54, 10220–10224 (2015).

  16. 16.

    , , , & Conversion of dinitrogen into acetonitrile under ambient conditions. Angew. Chem. Int. Ed. 55, 4786–4789 (2016).

  17. 17.

    & Nitrogen fixation in solution. Coord. Chem. Rev. 144, 69–145 (1995).

  18. 18.

    Catalytic reduction of molecular nitrogen in solutions. Russ. Chem. Bull. 52, 2555–2562 (2003).

  19. 19.

    Reductive silylation of molecular nitrogen via fixation to tris(trialkylsilyl)amine. J. Am. Chem. Soc. 94, 9266–9267 (1972).

  20. 20.

    , , & Preparation and properties of molybdenum and tungsten dinitrogen complexes. 25. Catalytic conversion of molecular nitrogen into silylamines using molybdenum and tungsten dinitrogen complexes. J. Am. Chem. Soc. 111, 1939–1940 (1989).

  21. 21.

    Activation of nitrogen for organic synthesis. J. Organomet. Chem. 689, 4210–4227 (2004).

  22. 22.

    et al. Molybdenum-catalyzed transformation of molecular dinitrogen into silylamine: experimental and DFT study on the remarkable role of ferrocenyldiphosphine ligands. J. Am. Chem. Soc. 133, 3498–3506 (2011).

  23. 23.

    , & Catalytic dinitrogen reduction at the molybdenum center promoted by a bulky tetradentate phosphine ligand. Angew. Chem. Int. Ed. 53, 14206–14210 (2014).

  24. 24.

    et al. Catalytic silylation of dinitrogen with a dicobalt complex. J. Am. Chem. Soc. 137, 4638–4641 (2015).

  25. 25.

    et al. Cobalt-catalyzed transformation of molecular dinitrogen into silylamine under ambient reaction conditions. Chem. Eur. J. 21, 8905–8909 (2015).

  26. 26.

    , & N2 reduction into silylamine at tridentate phosphine/Mo center: catalysis and mechanistic study. ACS Catal 5, 6902–6906 (2015).

  27. 27.

    et al. Catalytic transformation of dinitrogen into ammonia and hydrazine by iron-dinitrogen complexes bearing pincer ligand. Nat. Commun. 7, 12181 (2016).

  28. 28.

    et al. Direct transformation of molecular dinitrogen into ammonia catalyzed by cobalt dinitrogen complexes bearing anionic PNP pincer ligands. Angew. Chem. Int. Ed. 55, 14291–14295 (2016).

  29. 29.

    & Catalytic reduction of dinitrogen to ammonia at a single molybdenum center. Science 301, 76–78 (2003).

  30. 30.

    et al. Molybdenum triamidoamine complexes that contain hexa-tert-butylterphenyl, hexamethylterphenyl, or p-bromohexaisopropylterphenyl substituents. An examination of some catalyst variations for the catalytic reduction of dinitrogen. J. Am. Chem. Soc. 126, 6150–6163 (2004).

  31. 31.

    Catalytic reduction of dinitrogen to ammonia at a single molybdenum center. Acc. Chem. Res. 38, 955–962 (2005).

  32. 32.

    Catalytic reduction of dinitrogen to ammonia by molybdenum: theory versus experiment. Angew. Chem. Int. Ed. 47, 5512–5522 (2008).

  33. 33.

    , & Catalytic conversion of nitrogen to ammonia by an iron model complex. Nature 501, 84–87 (2013).

  34. 34.

    & Catalytic reduction of N2 to NH3 by an Fe-N2 complex featuring a C-atom anchor. J. Am. Chem. Soc. 136, 1105–1115 (2014).

  35. 35.

    & Low-temperature N2 binding to two-coordinate L2Fe0 enables reductive trapping of L2FeN2 and NH3 generation. Angew. Chem. Int. Ed. 54, 532–535 (2015).

  36. 36.

    & Fe-N2/CO complexes that model a possible role for the interstitial C atom of FeMo-cofactor (FeMoco). Proc. Natl Acad. Sci. USA 110, 15898–15903 (2013).

  37. 37.

    , & A 106-fold enhancement in N2-binding affinity of an Fe2(μ-H)2 core upon reduction to a mixed-valence FeIIFeI state. J. Am. Chem. Soc. 136, 13853–13862 (2014).

  38. 38.

    & An Fe-N2 complex that generates hydrazine and ammonia via Fe=NNH2: demonstrating a hybrid distal-to-alternating pathway for N2 reduction. J. Am. Chem. Soc. 138, 4243–4248 (2016).

  39. 39.

    , & A synthetic single-site Fe nitrogenase: high turnover, freeze-quench 57Fe Mössbauer data, and a hydride resting state. J. Am. Chem. Soc. 138, 5341–5350 (2016).

  40. 40.

    , & A molybdenum complex bearing PNP-type pincer ligands leads to the catalytic reduction of dinitrogen into ammonia. Nat. Chem. 3, 120–125 (2011).

  41. 41.

    et al. Catalytic formation of ammonia from molecular dinitrogen by use of dinitrogen-bridged dimolybdenum-dinitrogen complexes bearing PNP-pincer ligands: remarkable effect of substituent at PNP-pincer ligand. J. Am. Chem. Soc. 136, 9719–9731 (2014).

  42. 42.

    et al. Nitrogen fixation catalyzed by ferrocene-substituted dinitrogen-bridged dimolybdenum-dinitrogen complexes: unique behavior of ferrocene moiety as redox active site. Chem. Sci. 6, 3940–3951 (2015).

  43. 43.

    et al. Unique behaviour of dinitrogen-bridged dimolybdenum complexes bearing pincer ligand towards catalytic formation of ammonia. Nat. Commun. 5, 3737 (2014).

  44. 44.

    et al. Catalytic reduction of dinitrogen to ammonia by use of molybdenum-nitride complexes bearing a tridentate triphosphine as catalysts. J. Am. Chem. Soc. 137, 5666–5669 (2015).

  45. 45.

    , , & An overview of N-heterocyclic carbenes. Nature 510, 485–496 (2014).

  46. 46.

    Nolan S. P. (ed.) N-Heterocyclic Carbenes Wiley-VCH (2014).

  47. 47.

    & The development of L2X2Ru=CHR olefin metathesis catalysts: an organometallic success story. Acc. Chem. Res. 34, 18–29 (2001).

  48. 48.

    & N-heterocyclic carbenes as supporting ligands in transition metal complexes of N2. Dalton Trans. 45, 874–880 (2016).

  49. 49.

    , , & Toward new organometallic architectures: synthesis of carbene-centered rhodium and palladium bisphosphine complexes. Stability and reactivity of [PCBImPRh(L)][PF6] pincers. Inorg. Chem. 54, 9517–9528 (2015).

  50. 50.

    , , & A new tridentate pincer phosphine/N-heterocyclic carbene ligand: palladium complexes, their structures, and catalytic activities. Inorg. Chem. 43, 6822–6829 (2004).

  51. 51.

    & Chemistry of the PCNHCP ligand: silver and ruthenium complexes, facial/meridional coordination, and catalytic transfer hydrogenation. Organometallics 24, 1692–1702 (2005).

  52. 52.

    , , , & Small-molecule activation with molybdenum(0) complexes supported by mixed inidazol-2-ylidene/phosphanyl hybrid ligands—electronic and structural consequences of substituting a phosphane by a carbene group. Eur. J. Inorg. Chem. 2013, 3943–3955 (2013).

  53. 53.

    , , , & Molybdenum(0)-carbonyl complexes supported by mixed benzimidazol-2-ylidene/phosphine ligands: influence of benzannulation on the donor properties of the NHC groups. J. Organomet. Chem. 770, 61–68 (2014).

  54. 54.

    , , , & Di- and trinuclear gold complexes of diphenylphosphinoethyl-functionalised imidazolium salts and their N-heterocyclic carbenes: synthesis and photophysical properties. Chem. Eur. J. 21, 601–614 (2015).

  55. 55.

    , , , & Molybdenum complexes supported by mixed NHC/phosphine ligands: activation of N2 and reaction with P(OMe)3 to the first meta-phosphite complex. Chem. Eur. J. 21, 1130–1137 (2015).

  56. 56.

    & Dinitrogen complexes of tungsten with bulky phosphine co-ligands. Preparation and crystal structures of [{W(N2)2(PEt2Ph)3}2(μ-N2)], trans-[W(N2)2(PEt2Ph)4]·C4H8O, and [W(η6-C6H5PPrn2)(N2)(PPrn2Ph)2] and their reactions to give hydrazine or ammonia. J. Chem. Soc. Dalton Trans. 245–252 (1986).

  57. 57.

    , & Preparation of mer-[Mo(N2)3(PPrn2Ph)3] from trans-[Mo(N2)2(PPrn2Ph)4]; its X-ray crystal structure, spectroscopic properties, and protonation to give ammonia. J. Chem. Soc. Dalton Trans. 1591–1594 (1986).

  58. 58.

    , & Synthesis and protonation of N-heterocyclic-carbene-supported dinitrogen complexes of molybdenum(0). Organometallics 34, 3414–3420 (2015).

  59. 59.

    Donor properties of a series of two-electron ligands. Organometallics 28, 763–770 (2009).

  60. 60.

    Electronic and steric parameters of 76 N-heterocyclic carbenes in Ni(CO)3(NHC). Organometallics 28, 6458–6461 (2009).

  61. 61.

    Phenol-hypochlorite reaction for determination of ammonia. Anal. Chem. 39, 971–974 (1967).

  62. 62.

    et al. NBO 5.9 Theoretical Chemistry Institute, University of Wisconsin (2009).

  63. 63.

    Charge, bond order and valence in the AB initio SCF theory. Chem. Phys. Lett. 97, 270–274 (1983).

  64. 64.

    & How important is backbonding in metal complexes containing N-heterocyclic carbenes? Structural and NBO analysis. Eur. J. Inorg. Chem. 2011, 5025–5035 (2011).

  65. 65.

    & Quantifying and understanding the electronic properties of N-heterocyclic carbenes. Chem. Soc. Rev. 42, 6723–6753 (2013).

  66. 66.

    et al. Gaussian 09, Revision C.01 Gaussian, Inc. (2011).

  67. 67.

    , & Reparametrization of hybrid functionals based on energy differences of states of different multiplicity. Theor. Chem. Acc. 107, 48–55 (2001).

  68. 68.

    , & Quantum mechanical continuum solvation models. Chem. Rev. 105, 2999–3094 (2005).

  69. 69.

    , , & Synthesis, reactivities, molecular structures, and physical properties of paramagnetic bis(tetraphenylcyclopentadienyl) complexes of vanadium, chromium, cobalt, and nickel. Organometallics 6, 1703–1712 (1987).

  70. 70.

    & A kinetic study of the effect of substituents on the rate of formation of alkylpyridinium halides in nitromethane solution. J. Chem. Soc. 1960, 1885–1895 (1960).

Download references

Acknowledgements

The present project is supported by CREST, JST. We thank KAKENHI (Nos JP26288044, JP26105708, JP15K13687, JP15H05798 to Y.N., No. JP24109014 to K.Y. and No. JP26888008 to H.T.) from JSPS and MEXT. A.E. and S.K. are recipients of the JSPS Predoctoral Fellowships for Young Scientists. We also thank the Research Hub for Advanced Nano Characterization at The University of Tokyo for X-ray analysis.

Author information

Affiliations

  1. Department of Systems Innovation, School of Engineering, The University of Tokyo, Bunkyo-ku, Tokyo 113-8656, Japan

    • Aya Eizawa
    • , Kazuya Arashiba
    • , Shogo Kuriyama
    • , Kazunari Nakajima
    •  & Yoshiaki Nishibayashi
  2. Institute for Materials Chemistry and Engineering, Kyushu University, Nishi-ku, Fukuoka 819-0395, Japan

    • Hiromasa Tanaka
    • , Yuki Matsuo
    •  & Kazunari Yoshizawa
  3. Elements Strategy Initiative for Catalysts and Batteries, Kyoto University, Nishikyo-ku, Kyoto 615-8520, Japan

    • Kazunari Yoshizawa

Authors

  1. Search for Aya Eizawa in:

  2. Search for Kazuya Arashiba in:

  3. Search for Hiromasa Tanaka in:

  4. Search for Shogo Kuriyama in:

  5. Search for Yuki Matsuo in:

  6. Search for Kazunari Nakajima in:

  7. Search for Kazunari Yoshizawa in:

  8. Search for Yoshiaki Nishibayashi in:

Contributions

K.Y. and Y.N. directed and conceived this project. A.E., K.A., S.K. and K.N. conducted the experimental work. H.T. and Y.M. conducted the computational work. All authors discussed the results and wrote the manuscript.

Competing interests

The authors declare no competing financial interests.

Corresponding authors

Correspondence to Kazunari Yoshizawa or Yoshiaki Nishibayashi.

Supplementary information

PDF files

  1. 1.

    Supplementary Information

    Supplementary figures, supplementary tables, supplementary methods and supplementary references.

  2. 2.

    Peer review file

Word documents

  1. 1.

    Supplementary Data 1

    Information of Cartesian Coordinates.

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Creative Commons BYThis work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/