A Chatt-Type Catalyst with One Coordination Site for Dinitrogen Reduction to Ammonia.

With [Mo(N2 )(P2 Me PP2 Ph )] the first Chatt-type complex with one coordination site catalytically converting N2 to ammonia is presented. Employing SmI2 as reductant and H2 O as proton source 26 equivalents of ammonia are generated. Analogous Mo0 -N2 complexes supported by a combination of bi- and tridentate phosphine ligands are catalytically inactive under the same conditions. These findings are interpreted by analyzing structural and spectroscopic features of the employed systems, leading to the conclusion that the catalytic activity of the title complex is due to the strong activation of N2 and the unique topology of the pentadentate tetrapodal (pentaPod) ligand P2 Me PP2 Ph . The analogous hydrazido(2-) complex [Mo(NNH2 )(P2 Me PP2 Ph )](BArF )2 is generated by protonation with HBArF in ether and characterized by NMR and vibrational spectroscopy. Importantly, it is shown to be catalytically active as well. Along with the fact that the structure of the title complex precludes dimerization this demonstrates that the corresponding catalytic cycle follows a mononuclear pathway. The implications of a PCET mechanism on this reactive scheme are considered.

The activation of molecular nitrogen has been of great interest over the last decades. This in particular refers to biological nitrogen fixation, which is mediated by the enzyme nitrogenase. Although the structure of this enzyme has been fully determined, the mechanism of the dinitrogen reduction and protonation is still the subject of current research. To mimic this process and elucidate its mechanism, various small‐molecule based model systems have been studied in detail. The earliest of these systems were established by Chatt and Hidai on the basis of molybdenum (bis)dinitrogen complexes with phosphine coligands. In 1985 Pickett et al. demonstrated an electrochemical synthesis of NH3 mediated by a tungsten complex. The first truly catalytic reduction of N2 to ammonia was achieved by Schrock et al. in 2003 using a triamidoamine molybdenum complex (Scheme 1) as catalyst, Cp*2Cr as reductant and LutH(BArF) as proton source. This system generated 7.6 equivalents of NH3, which clearly was a milestone in synthetic nitrogen fixation. A N2‐bridged dinuclear Mo system supported by pincer ligands, presented by Nishibayashi et al. in 2011, led to 23.2 (through modification up to several hundred)[ , ] equivalents of ammonia. While this group first employed LutH(OTf) as acid and Cp*2Cr as reductant, an even more powerful protocol was established in 2019, involving SmI2/H2O as reductant and proton source. This way, 4.350 equivalents of ammonia could be generated. In 2013, the first non‐molybdenum catalytic system for N2 reduction was presented by Peters et al., employing a BP3‐supported iron complex, KC8 as reductant and HBArF as acid.

Scheme 1

Molybdenum‐based model systems for synthetic nitrogen fixation.[ , ]

n class="Chemical">Molybdenum‐based model systems for synthetic n class="Chemical">nitrogen fixation.[ , ]

On the basis of the classic Chatt‐type bis(dinitrogen) Mo/W complexes containing diphosphine coligands (e.g., dppe or depe) the first mechanism for the transition‐metal mediated conversion of N2 to NH3 was formulated, the so‐called Chatt cycle (Scheme 2).

Scheme 2

Chatt cycle (solid arrows); dotted: dinuclear pathway with N−N cleavage.

This reactive scheme starts with the protonation of the parent N2 complex, leading to the hydrazido(2‐) complex. In the subsequent steps, one additional proton and two electrons are required to cleave the N−N bond and generate the first equivalent of ammonia.[ , ] This mechanism is very similar (but not identical) to the Schrock cycle, which is based on the Mo triamidoamine complex.[ , ] In this context it should be noted that both the Chatt‐ and the Schrock cycle involve N−N splitting at the level of NNH2‐ and NNH3‐complexes, whereas the dinuclear systems of Nishibayashi et al. (Scheme 1) mediate N−N cleavage of the parent N2‐complexes, leading to two nitrido intermediates, which subsequently are converted to NH3.[ , ] Recently, this scenario has also been evidenced in a classic Chatt system by Masuda and co‐workers, where the dinuclear MoI complex [{Mo(depe)2}2(μ‐N2)]2+ was found to split into two [Mo(N)(depe)2]+ cations by cleavage of the N−N bond (Scheme 3). Similar reactivities have been obtained with other dinuclear, dinitrogen‐bridged transition metal) complexes.[ , ]

Scheme 3

Formation of a dinuclear MoI complex from [Mo(N2)2(depe)2] via one‐electron oxidation, leading to a MoIV nitrido complex by dinitrogen cleavage; adapted from Masuda et al.

An important disadvantage of the original Chatt systems has been the fact that protonation of the dinitrogen complex involves exchange of one of the two N2 ligands by the conjugate base of the applied acid, causing a 50 % loss of bound substrate. Moreover, this anionic trans‐coligand had to be exchanged again at the end of the cycle leading to the bis(dinitrogen) complex, and MoI complexes formed as intermediates during that stage were found to be prone to disproportionation.[ , ] These mechanistic drawbacks have traditionally been invoked to rationalize that classic Chatt‐type systems, although in principle forming all relevant intermediates, are catalytically inactive towards the conversion of N2 to NH3. On the other hand, Nishibayashi et al. recently showed that Chatt complexes with mono‐ and bidentate ligands indeed catalyse the generation of ammonia from N2 if SmI2/H2O (or SmI2/alcohol) is used as reductant and proton source. Using cis,mer‐[Mo(NNH2)(OTf)2(PMePh2)3] as example for a NNH2 intermediate also led to catalytic amounts of NH3. From this observation it was inferred that the SmI2‐mediated reduction pathway of Chatt‐type complexes probably follows the Chatt cycle.

In view of the above‐mentioned problems of the classic Chatt complexes, we had in the past developed a series of molybdenum dinitrogen complexes in which the trans position is occupied by a donor atom of a multidentate ligand. These systems were intended to provide only one site for the coordination and reduction of N2 and avoid all other ligand exchange reactions occurring at the single Mo center. Initially, we had employed a combination of a tripodal (1) or a linear tridentate ligand (2) with a bidentate co‐ligand (Scheme 4) for this purpose. Compounds 2 and 1, however, suffered from isomerization and, respectively, instability of the tridentate phosphine ligand coordination upon protonation of the N2‐complex, which was ascribed to the fact that the trans‐donor is not fixed strongly enough to the center Mo atom. Later we succeeded combining the two described approaches into a unique pentadentate tetrapodal (pentaPod) phosphine ligand. Based on this concept, the molybdenum mono(dinitrogen) complex [Mo(N2)(P2 MePP2 Ph)] (3) was synthesized and characterized both experimentally and theoretically.

Scheme 4

[Mo(N2)(tdppme)(dmpm)] (1) [Mo(N2)(prPPHP)(dmpm)] (2), and [Mo(N2)(P2 MePP2 Ph)] (3).

We now discovered that reaction of 3 in THF with N2 gas at 1 atm, 180 equiv of SmI2 and 180 equiv of H2O gives 25.73±0.37 equiv of ammonia based on the molybdenum atom (43 % yield based on SmI2; Table 1). Replacing 14N2 by 15N2 in these experiments correspondingly leads to 15NH3 which was detected by 1H‐NMR as 15NH4Cl (cf. SI, Figure S1). To check if the pentadentate coordination of 3 is responsible for the catalytic activity, complexes 1 and 2 with tridentate or tripodal ligands were investigated under the same conditions. However, both only led to substoichiometric amounts (less than 2 equiv) of ammonia, which indicates decomposition of the complexes. In order to understand the different catalytic activities of 1, 2 and 3, the electronic and geometric structures of these systems are analysed in the following.

Table 1

Experimental and calculated spectroscopic and structural parameters of the employed molybdenum pentaphosphine complexes and ammonia formation in the presence of them.

Catalyst	NH3 productiona)	NN stretch [cm−1]	d(Mo‐Pax) [pm]	d(Mo‐N) [pm]	d(N‐N) [pm]	d(Mo‐Peq) av. [pm]	2 J(31PM,15Nα) [Hz]	3 J(31PM,15Nβ) [Hz]
[Mo(N2)(tdppme)(dmpm)] (1)	0.82±0.04	1979	244.54(16)	206.6(6)	106.9(8)	246.21	–	–
[Mo(N2)(prPPHP)(dmpm)] (2)	1.77±0.03	1974	240.15(6)	202.1(2)	111.6(3)	242.96	14.1	1.2
[Mo(N2)(P2 MePP2 Ph)] (3)	25.73±0.37	1929	238.68(12)	203.3(5)	109.9(5)	244.81	13.5	1.4
[Mo(NNH2)(P2 MePP2 Ph)]2+ (4)	26.14±0.32	1490[b]	261.17[b]	177.4[b]	131.7[b]	252.51[b]	23.2	7.6

[a] equivalents per Mo atom; N2 gas at 1 atm, 180 equiv of SmI2 and 180 equiv of H2O; [b] PBE0‐D3(BJ)/def2‐SVP.

Experimental and calculated spectroscopic and structural parameters of the employed n class="Chemical">molybdenum pentaphosphine complexes and n class="Chemical">ammonia formation in the presence of them.

n class="Chemical">NH3

d(Mo‐n class="Chemical">Pax)

2 J(n class="Chemical">31PM,n class="Chemical">15Nα)

3 J(n class="Chemical">31PM,n class="Chemical">15Nβ)

n class="Chemical">[Mo(N2)(n class="Chemical">tdppme)(dmpm)] (1)

–

–

n class="Chemical">[Mo(N2)(prPPHP)(n class="Chemical">dmpm)] (2)

n class="Chemical">[Mo(N2)(P2 MePP2 Ph)] (3)

[Mo(n class="Chemical">NNH2)(P2 MePP2 Ph)]2+ (4)

The key property of a molecular catalyst for synthetic nitrogen fixation is the activation of the N2 ligand, enabling its protonation and further reduction to ammonia. The most sensitive probe of this capability is the N−N stretching frequency. In this respect, complex 3 exhibits the highest activation (Table 1). As a matter of fact, its N−N stretching frequency is the lowest of all known Mo‐pentaphosphine complexes. In comparison, νNN of 1 and 2 are by 45 and 50 cm−1 higher, respectively. The activation of N2 is a function of the electron density on the Mo0 center, which in turn sensitively depends on the type of phosphine donors. Note that all three complexes have an equatorial Peq coordination of two PMe2 and two PPh2 groups, whereas the nature of the phosphine in trans‐position is different. In a first approximation, the activation of N2 in complexes 1–3 thus is a function of the axial phosphine donor Pax, and in view of the fact that electron donation increases within the sequence PPh3<PR2H

Besides these electronic factors, it is also of interest to analyze the Mo−N and Mo−Pax distances in 1–3. Importantly, 3 has the shortest Mo−Pax bond of all complexes, also being shorter than all Mo−Peq bonds (Table 1). The Mo−Pax distance of 2 is slightly longer, whereas that of 1 is much longer, getting similar to the Mo‐Peq values (≈2.4 Å). The short Mo−Pax distances of 3 and 2 indicate strong Mo−Pax bonds, which serves to transfer electron density to the Mo0 center. Remarkably, the Mo−N bonds are short in these complexes as well. This should lead to strong Pax‐Mo‐(N2) interactions which may be probed by 31P‐ and 15N‐NMR spectroscopy.

The 31P NMR spectrum of 3 shows an AA'MXX’ patternpan>, in agreement with its pentaphosphine environment (Figure 1 a). In order to obtain information regarding the coupling between the phosphine ligands and the N atoms of the coordinated N2, the isotopically labeled complex was synthesized. Additional couplings between the P donors and the Nα and Nβ atoms of the dinitrogen ligand are visible in the 31P‐NMR spectrum of N‐3 (Figure 1 b and c). The M signal, which belongs to Pax (Figure 1 d), exhibits much stronger couplings (2 J(31PM, 15Nα)=13.5 Hz, 3 J(31PM, 15Nβ)=1.4 Hz) than the phosphine groups Peq in cis‐position (2 J(31PA, 15Nα)=3.1 Hz, 3 J(31PA, 15Nβ) <1.0 Hz; 2 J(31PX, 15Nα)=3.0 Hz, 3 J(31PX, 15Nβ)=1.0 Hz; cf. SI, Figures S2, S3).

Figure 1

a) Experimental (in C6D6) 31P‐NMR spectra of 3 and b) a mixture of 3 and N‐3 (18 % 3). c) Simulated spectrum of N‐3. d) Overlay of experimental M signals of N‐3 and the mixture, showing the two‐bond (tertiary) 15N‐induced isotope effect (* 2Δδ=1.0 Hz, 6.2 ppb).

a) Experimental (in C6D6) n class="Chemical">31P‐NMR spectra of 3 and b) a mixture of 3 and N‐3 (18 % 3). c) Simulated spectrum of N‐3. d) Overlay of experimental M signals of N‐3 and the mixture, showing the two‐bond (tertiary) n class="Chemical">15N‐induced isotope effect (* 2Δδ=1.0 Hz, 6.2 ppb).

In the M part of the spectrum, an asymmetric positioning of the 15N (dd‐) signal with regard to the 14N signal deriving from residual 3 is noticed (Figure 1 d), which corresponds to a two‐bond (tertiary) 15N‐induced isotope effect on the chemical shift of the trans 31P nucleus (2 Δ 31P(15N)). We ascribe this phenomenon to the anharmonicity of the Mo‐(N2) potential, leading to a slight reduction of the Mo−Nα equilibrium bond distance if the mass of the N2 ligand is increased. This in turn increases the Mo−Pax bond length by virtue of the trans effect, causing an increased shielding of Pax. With an upfield shift of around 1 Hz (6.2 ppb) the two‐bond isotope shift across the metal center is in a range where usually one‐bond 14N→15N shifts (e.g., phosphoric acid amide: 9.6 ppb ) are observed. This indeed reflects a strong influence of the N2 coordination on the bonding of the P‐atom in trans position.

In order to elucidate a possible dependence of this effect on the electronic structure of the Mo‐(N2) complex, the 31P‐NMR spectra of complex 2, which also exhibits a short Mo−Pax bond (cf. Table 1), were re‐examined (cf. SI, Figures S4–S6). This analysis provided similar results ((2 Δ 31P(15N))=1.1 Hz (6.9 ppb). Moreover, the J(31PM,15N) coupling constants were determined to 14.1 (Nα) and 1.2 Hz (15Nβ), respectively, quite close to the values of 3 (Table 1). In case of complex 1 having the longest Mo−Pax bond of all three complexes, an analogous analysis was not possible due to its 31P‐NMR spectrum being of higher order (cf. SI, Figure S7). As the bonding situation drastically changes along the Chatt cycle, it also appeared of interest to explore a possible correlation between 2 Δ 31P(15N) and the electronic structure of the respective intermediates.

Protonation of 3 with [H(OEt2)2][BArF] in Et2O (“HBArF”) affords the NNH2 complex [Mo(NNH2)(P2 MePP2 Ph)][BArF]2 (4). This is, for example, evident from the vibrational spectra of solid 4 ( N‐4) showing N−H (15N−H) stretches at 3312 (3307) cm−1 (ν as(NH)) and 3200 (3198) cm−1 (ν s(NH)) as well as the disappearance of νNN at 1929 cm−1 (cf. SI, Figure S8 and Table S1; preliminary spectroscopic data of 4 were already given in ref. [23]). In analogy to 3, the 31P‐NMR spectrum of 4 exhibits an AA'MXX’ pattern, with chemical shifts and coupling constants modified with respect to the former (Figure 2 a; cf. SI, Figure S9–S13). This indicates that the pentaPod environment of 3 is retained upon protonation, a prerequisite for the catalytic activity of our system. Protonation was also performed with 3 containing a mixture of 14N2 and 15N2 (18 % 14N; see above). Again, the resulting 31P‐NMR spectrum (Figure 2 b) shows a superposition of the spectra mainly deriving from the 15N15NH2 complex (Figure 2 c) with small additional signals from the 14N14NH2 isotopomer. In contrast to the parent N2 complex 3, however, no 15N‐isotope effect on the 31P‐NMR shift is visible in the M‐part of 4/ N‐4 (Figure 2 d).

Figure 2

a) Experimental (in d10‐Et2O) 31P‐NMR spectra of 4 and b) a mixture of 4 and N‐4 (18 % 4). c) Simulated spectrum of N‐4. d) Overlay of experimental M signals of N‐4 and the mixture.

In order to interpret this result, we note that DFT predicts a hydrazido(2‐) configuration for 4 (cf. SI, Figure S14), with a triple bond between Mo and Nα. This is in contrast to classic Chatt‐type NNH2 complexes such as [MoF(NNH2)(diphos)2] where an isodiazene description was found to be more appropriate. The lack of 2 Δ 31P(15N) on δ(PM) suggests that the anharmonicity in the Mo≡N potential of 4 is much lower than in the Mo‐N2 bond of the parent dinitrogen complex 3. The triply bonded NNH2 ligand should exert a strong trans effect. This is supported by DFT calculations which indicate a significant elongation of the Mo−Pax distance in 4 with respect to 3, making it even longer than the Mo−Peq bonds (Table 1). Correspondingly, the protonation‐induced high‐field shift is much larger for the M signal than for the A and X signals (cf. SI, Figure S13 and Table S2).

The flexibility of the metal–E bond in trans‐position to the nitrogenic ligand has been considered by Peters et al. as an important criterion for the catalytic activity of their iron‐dinitrogen complexes supported by EP3 ligands (E=B, Si, C). In spite of the short Mo−Pax bond observed for the Mo‐dinitrogen complex 3 it appears that the pentaPod ligand framework is sufficiently flexible to allow elongation of the axial Mo−P bond in the NNH2‐complex 4.

Formation of the hydrazido(2‐) complex 4 is also evident from its 1H‐15N‐HMBC spectrum which clearly shows the ‐NNH2 moiety; i.e., a doublet in the 1H dimension with a 1 J(15Nβ, 1H) of 94.6 Hz and a corresponding triplet in the 15N spectrum (Figure 3; cf. SI for complete spectrum, Figure S15). In the 15N spectrum the couplings of Nβ to Nα (11.2 Hz) and the trans standing PM (7.6 Hz) are also observable (cf. SI, Figure S16, Tables S3 and S4).

Figure 3

Enlarged NβH2 part of the 1H‐15N‐HMBC spectrum of N‐4 in d10‐Et2O.

In analogy to 3, compound 4 was applied as a catalyst for the reduction of N2 at 1 atm with 180 equiv of SmI2 and 180 equiv of H2O in THF. As the stability of 4 in this solvent had been found to be limited, we generated 4 in situ in diethyl ether and subsequently added this to a solution of SmI2/H2O in THF. These experiments afforded 26.14±0.32 equiv of NH3 (Table 1), identical to the yield obtained with the N2 complex 3 within the error limit. This proves the role of the hydrazido(2‐) complex 4 as an intermediate in the catalytic conversion of N2 to NH3 mediated by 3 and suggests that the corresponding mechanism follows the Chatt cycle; e.g., avoids a direct N≡N cleavage (Scheme 2). Furthermore, the fact that dimerization of 3 is sterically hindered renders the existence of a dinuclear pathway (Scheme 3) improbable. A simulation of a corresponding MoI or Mo0 dimer leads to dissociation of one Mo−P bond (SI, Figure S17).

The usual formulation of the Chatt cycle starts with two protonations of the Mo0‐N2 complex, leading to the NNH2 complex (cf. Scheme 2); notably, 4 has been generated from 3 this way. On the other hand, the SmI2/water complex is known to react with protonatable/reducible substrates by proton‐coupled electron transfer (PCET). In this context, it has become customary to assess the N2‐reducing capacity of a catalytic nitrogenase model system by quoting the N−H bond dissociation free energy (BDFE) of the respective NNH (diazenido) complex (cf. Scheme 2). In order to exergonically transfer one electron and one proton to the N2 complex, the BDFE of the former has to exceed that of the employed PCET reagent or the effective BDFE of the employed acid/reductant combination, respectively.

To determine the N−H BDFE of the NNH‐intermediate for a given N2‐reduction catalyst, DFT calculations may be employed. Transfer of one electron and one proton to the Mo0‐dinitrogen complex leads to the neutral MoI‐diazenido(−) intermediate. An estimate of the corresponding energetics was obtained by DFT, simulating the reaction of [Mo(N2)(pentaPod)] with TEMPO‐H, a H‐atom transfer reagent having a well‐defined O−H BDFE of 65.2 kcal mol−1 in benzene, to give the [Mo(NNH)(pentaPod)] complex. Subtraction of the reaction TEMPO‐H→TEMPO+H leads to a N−H BDFE of 19.2 kcal mol−1 for the MoI‐diazenido(−) complex (Δr,theo G (solv, benzene), cf. SI, Table S6), which is somewhat lower than the O−H BDFE of SmI2/water (26 kcal mol−1). PCET from this reagent to the Mo0(N2) complex thus is slightly endergonic (ΔG 298=+6.8 kcal mol−1), but thermodynamically feasible.

In view of the fact that the diazenido(−) intermediates of the classic Chatt cycle correspond to MoII (and not MoI) species, we also theoretically investigated the formation of [MoII(NNH)(pentaPod)]+ by PCET from the corresponding cationic MoI(N2)‐complex. An analogous procedure as described above gives a N−H BDFE of 52.5 kcal mol−1 for the MoII‐diazenido(−) intermediate (Δr,theo G 298(solv, benzene), cf. SI, Table S6). This value well exceeds the BDFE of SmI2/water (see above), rendering PCET to the cationic [MoI(N2)(pentaPod)]+ complex highly exergonic (ΔG 298=−26.5 kcal mol−1). On the other hand, neutral [Mo0(N2)(pentaPod)] (3) was successfully employed as catalyst in our SmI2/water‐mediated N2‐to‐NH3 conversion experiments (see above). In the framework of a PCET mechanism it thus remains to be elucidated whether (and, if yes, how) our system switches from a pathway starting from a Mo0(N2) complex to an energetically more favourable reaction path that involves a mononuclear, cationic MoI(N2) intermediate.

In summary, three structurally related Mo‐N2 complexes with pentaphosphine environment have been investigated as catalysts for the conversion of N2 to NH3, using SmI2/H2O as protonating agent and reductant. Only the title complex [Mo(N2)(P2 MePP2 Ph)] (3) was found to be catalytically active. This is attributed to the fact that it exhibits the highest activation of N2 and the pentaPod coordination. The strong chelate effect of this ligand creates an inert and stable, yet flexible ligand environment allowing protonation and reduction of the Mo0‐N2 complex under retention of the pentaphosphine ligation. Protonation of the dinitrogen complex 3 leads to the hydrazido(2‐) complex 4 which was isolated and spectroscopically characterized. Importantly, 4 was also found to be catalytically active. Along with the fact that the Mo(N2)‐pentaPod complex precludes dimerization this demonstrates the existence of a mononuclear pathway along the Chatt cycle for the N2‐to‐NH3 conversion catalyzed by this system. The implications of a PCET mechanism on this pathway are considered.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.