Catalytic Reduction of Dinitrogen to Ammonia at a Single Molybdenum Center

See allHide authors and affiliations

Science  04 Jul 2003:
Vol. 301, Issue 5629, pp. 76-78
DOI: 10.1126/science.1085326


Dinitrogen (N2) was reduced to ammonia at room temperature and 1 atmosphere with molybdenum catalysts that contain tetradentate [HIPTN3N]3– triamidoamine ligands {such as [HIPTN3N]Mo(N2), where [HIPTN3N]3– is [{3,5-(2,4,6-i-Pr3C6H2)2C6H3NCH2CH2}3N]3–} in heptane. Slow addition of the proton source [{2,6-lutidinium}{BAr′4}, where Ar′ is 3,5-(CF3)2C6H3]and reductant (decamethyl chromocene) was critical for achieving high efficiency (∼66% in four turnovers). Numerous x-ray studies, along with isolation and characterization of six proposed intermediates in the catalytic reaction under noncatalytic conditions, suggest that N2 was reduced at a sterically protected, single molybdenum center that cycled from Mo(III) through Mo(VI) states.

The reduction of dinitrogen (N2) to ammonia (NH3) by various nitrogenase enzymes is one of the most fascinating transition metal–catalyzed reactions in biology (112). Six electrons and six protons produce two equivalents of NH3 per N2 in discrete steps at 1 atm of ambient pressure and mild temperatures, with the aid of one or more transition metal centers (Fe, Mo, or V) within those nitrogenases. Although nitrogenases have been studied for decades (primarily the Fe/Mo nitrogenase), it is still not known today how they accomplish this feat.

With the discovery of the first N2 complex of a transition metal in 1965 (13) came the hope that many N2 complexes could be prepared and that an abiological catalytic reduction of N2 at ambient pressure and temperature with protons and electrons at a well-defined transition-metal site would be forth-coming (1422). Hundreds of N2 complexes are now known, but only a few reports of the catalytic reduction of N2 to NH3 have appeared (18, 2327). No reduction of N2 has been accomplished with a relatively mild reducing agent, and no system has revealed many details of the N2 reduction steps. The most mechanistically elaborated system that contains N2 and reduced-N2 ligands has been a series of W(0) and Mo(0) phosphine complexes (1416, 21). Although examples of almost all of the proposed intermediates for reduction of N2 at a single metal center have been isolated, no catalytic reaction to give NH3 [in the presence of protons and electrons (28)] that uses these relatively well-defined systems has been established since they were discovered more than 30 years ago.

We have been studying the chemistry of N2 complexes for two decades, especially those of Mo and W (29), focusing on chemistry that we believe is relevant to N2 reduction and that involves these metals in relatively high oxidation states [M(III) to M(VI)]. Recently, we prepared and began to explore the chemistry of Mo complexes that contain a triamidoamine ([(ArNCH2CH2)3N]3– is [ArN3N]3–, where Ar is aryl) ligand (30, 31). In order to prevent formation of what we believe to be relatively stable and unreactive bimetallic [ArN3N]Mo-N=N-Mo[ArN3N] complexes, maximize steric hindrance in a monomeric species, and provide increased solubility of the complexes, we synthesized species that contain a [HIPTN3N]3– ligand, where HIPT (hexa-iso-propyl-terphenyl) is 3,5-(2,4,6-i-Pr3C6H2)2C6H3 (Scheme 1) (32, 33). Starting with MoCl (where Mo is [HIPTN3N]Mo), we showed that we could prepare many intermediates in a hypothetical reduction of N2, all of which contain the same [HIPTN3N]3– ligand. These intermediates include paramagnetic Mo(N2) (1); diamagnetic Mo-N=N-H (2); diamagnetic {Mo=N-NH2}{BAr′4}, where Ar′ is 3,5-(CF3)2C6H3 (3); diamagnetic Mo≡N (4); diamagnetic {Mo=NH}{BAr′4} (5); and paramagnetic {Mo(NH3)}{BAr′4} (6). Extensive 15N labeling studies, nuclear magnetic resonance (NMR) studies, and x-ray studies [of 1, 4, and 6 (33) and 2, 3, and 5 (34)] all reveal a trigonal pocket in which N2 and its reduced products are protected to a marked degree by three 2,4,6-i-Pr3C6H2 rings clustered around the pocket (Scheme 1). Compounds 1 through 6 are among the 13 or more that might take part in a catalytic reduction of end-on bound N2 by the stepwise, alternating addition of six protons and six electrons (Fig. 1). The intermediates shown in Fig. 1, in which the oxidation state of the metal varies between Mo(III) and Mo(VI) (29), are analogous to those proposed originally by Chatt (14) for lower oxidation-state Mo and W phosphine complexes.

Scheme 1.
Fig. 1.

Proposed intermediates in the reduction of dinitrogen at a [HIPTN3N]Mo (Mo) center through the stepwise addition of protons and electrons.

We also showed (32) that it is possible to prepare several compounds in high yield from others through the use of {2,6-lutidinium}{BAr′4} as the proton source and cobaltocene as the electron source in C6D6. For example, the addition of 1.0 equivalent of {LutH}BAr′4 and 2.0 equivalents of Co(η5-C5H5)2 (CoCp2) in benzene to Mo(N2) yields Mo-N=NH essentially quantitatively. [For CoCp2, the reversible half-wave redox potential (E0′) = –1.33 V versus [FeCp2]+/0 in CH2Cl2 (35).] This conversion is made possible by what is believed to be an initial protonation of an amido nitrogen [not the N2 (34)] followed by an electron transfer, a type of proton-coupled electron transfer reaction (36). In the presence of 7.0 equivalents of {LutH}{BAr′4} and 8.2 equivalents of CoCp2, Mo(N2) is converted in C6D6 into a mixture of compounds in which {Mo(NH3)}{BAr′4}isthe major species (∼60%). Although we found that {Mo(NH3)}{BAr′4} is not reduced to Mo(NH3) by CoCp2 in C6D6, {Mo(NH3)}{BAr′4} reacts with 3 equivalents of CoCp2 in C6D6 to give an equilibrium mixture of {Mo(NH3)}{BAr′4} (90%) and Mo(N2) (10%) after 18 hours in a sealed NMR tube. Therefore, it seemed plausible that an actual catalytic conversion of N2 to NH3 at a single metal center might finally be realized with a stronger reducing agent than CoCp2, under the appropriate conditions.

Decamethylchromocene [Cr(η5-C5Me5)2, or CrCp*2] was chosen as the reducing agent on the basis of its demonstrated ability (in contrast to CoCp2) to reduce {Mo(NH3)}+ completely to Mo(NH3) in C6D6; CrCp*2 is a stronger reducing agent than CoCp2 by 0.13 V in CH3CN (37). However, initial experiments suggested that formation of Mo(N2) from Mo(NH3) under N2 was relatively slow (minutes to hours). As expected, we found that CrCp*2 was oxidized rapidly by {LutH}{BAr′4} in C6D6, in which {LutH}{BAr′4} is soluble. Thus, it became apparent that, in order to achieve catalytic conversion of N2 to NH3 with any efficiency (in terms of electrons consumed), it would likely be necessary (i) to slow down the reaction between the proton source and the reducing agent relative to reactions that involve Mo species, and (ii) to allow sufficient time to convert Mo(NH3) to Mo(N2). Therefore, we felt that it would be most desirable to add {LutH}{BAr′4} and CrCp*2 to the catalyst in solution at a slow, controllable rate, in order to maintain Mo intermediates in excess of both acid and reductant. The choice of heptane as the solvent ensured that the concentration of sparingly soluble {LutH}{BAr′4} in solution would be low. (In contrast, all cationic Mo derivatives in this family that we have isolated, such as 3, 5, or 6, as BAr′4 salts, are soluble in alkane solvents.) The suspension was then stirred vigorously as a solution of CrCp*2 was added with a syringe pump, over a period of 6 hours. In order to be certain that all ammonia (in the gas phase, in solution, or as an ammonium salt) could be collected and measured accurately and with precision, we designed and constructed a self-contained glass reactor in which the reducing agent could be added by means of a magnetically driven syringe over a period of several hours. Great care was taken to purify solvents and reagents. We demonstrated that a reaction set up in a drybox in fact could be run outside the drybox with no significant change in result (38).

The results of several runs on the scale of 36 equivalents of CrCp*2 and 48 equivalents of {LutH}{BAr′4} are listed in Table 1. Formation of 7.56 to 8.06 equivalents of NH3 in the presence of four different Mo derivatives (1, 2, 4, or 6; 16 runs total) suggests that NH3 was formed catalytically with respect to Mo from N2. In order to eliminate the remote possibility that the triamidoamine ligand and/or 2,6-lutidine or 2,6-lutidinium might serve as a nitrogen source for the NH3, the reduction was carried out under 15N2. The result was formation of 8.18 equivalents of 15NH4Cl with an 15N isotopic enrichment that was indistinguishable from that of commercially available 15NH4Cl (>98% 15N), according to its 1H NMR spectrum in DMSO-d6 (where DMSO is dimethyl sulfoxide) (Fig. 2). Addition of the reducing agent over 25 s (followed by stirring for 7 hours) resulted in a poorer yield of NH3 (2.83 equivalents), which suggests that one or more side reactions, such as protonation of CrCp*2, takes precedence over N2 reduction under these circumstances. Preliminary studies of the reduction of {Mo(NH3)}{BAr′4} under N2 to give Mo(N2) suggest that the slowest step in Fig. 1 is conversion of Mo(NH3) to Mo(N2), although the rate of addition of the reductant is believed to limit the rate of reduction in the experiments in which CrCp*2 is added over a period of 7 hours (Table 1). It is not yet known whether the reaction is limited to about four turnovers under the conditions we describe here or what species are present at the end of a reaction.

Fig. 2.

1H NMR spectra (DMSO-d6) of (A) the mixture of NH4Cl and 2,6-LutHCl obtained from the reaction of Mo(N=NH) with 48 equivalents of {2,6-LutH}{BAr′4} and 36 equivalents of CrCp*2 under an atmosphere of 14N2, (B) the mixture obtained from the analogous reaction of Mo(15N=15NH) under an atmosphere of 15N2, and (C) authentic 15NH4Cl (>98% 15N) in the presence of 2,6-LutHCl. ppm, parts per million.

Table 1.

The results of catalytic reduction of N2. Unless otherwise indicated, all runs were done at 23° to 25°C and 1 atm of N2, by dropwise addition with constant stirring of 10.0 mL of a solution of CrCp*2 in heptane (36 equivalents relative to Mo) at a rate of 1.7 ml per hour to a mixture of the Mo compound, 48 equivalents of {LutH}{BAr′4} and 0.6 ml of heptane, followed by stirring for 1 hour. Ammonia was isolated as a mixture of solid NH4Cl and 2,6-LutHCl and analyzed by the indophenol method (3840). The theoretical yield is based on the amount of NH3 possible with the reducing equivalents available. Numbers in parentheses in columns 2 and 4 are the standard deviations, σ. Equiv., equivalents; expt, experimental.

Mo compound Equiv. NH3 (expt/theory) No. of runs Yield NH3, %
[HIPTN3N]Mo(N2) (1) 7.56 (11)/12 6View inline 63(1)
[HIPTN3N]Mo(N=NH) (2) 7.73 (15)/12.33 4View inline 63(1)
[HIPTN3N]MoΞN (4) 7.97 (23)/12 3View inline 66(2)
{[HIPTN3N]Mo(NH3)}{BAr′4} (6) 8.06 (21)/12.67 3View inline 64(2)
[HIPTN3N]Mo(15N=15NH) under 15N2 8.18/12.33 1View inline 66
[HIPTN3N]Mo(N2)View inline 2.83/12 1 24
  • View inline* Two runs outside the drybox (7.54 and 7.62 equivalents), two runs inside (7.55 and 7.62 equivalents), and two runs inside in the dark (7.69 and 7.36 equivalents); average = 7.56 equivalents, σ = 0.11.

  • View inline 7.75, 7.93, 7.61, and 7.62 equivalents; average = 7.73, σ = 0.15.

  • View inline 8.22, 7.95, and 7.75 equivalents; average = 7.97, σ = 0.23.

  • View inline§ 8.08, 8.26, and 7.84 equivalents; average = 8.06, σ = 0.21.

  • View inline >98% 15NH4Cl by 1H NMR, as described in the text and Fig. 2.

  • View inline CrCp*2 was added over a period of 25 s, followed by stirring for 7 hours.

  • To the best of our knowledge, the efficiencies (the yield of NH3 relative to that expected by theory on the basis of reducing equivalents) of the most successful of these experiments (63 to 66%) are second only to that of Fe/Mo nitrogenase (75%). (Nitrogenases consume two or more reducing equivalents in side reactions that make H2, although turnovers are essentially unlimited in nitrogenases in general.) Furthermore, the catalytic activity (with respect to Mo) is attained with the weakest reductant of all abiological systems that have been reported. [For CrCp*2, E°′ can be estimated to be about –0.90 V versus the normal hydrogen electrode (NHE) (35); the reducing power of biological reducing agents is limited to about –0.46 V versus NHE (2). We consider it highly likely that N2 is being reduced at a sterically protected, single Mo center and that the relevant oxidation states are Mo(III) to Mo(VI) (Fig. 1). In spite of x-ray studies that have focused attention on the seven-Fe cluster in Fe/Mo nitrogenase as the site of N2 reduction (12), we believe that reduction of N2 in Fe/Mo nitrogenase at the single Mo center, favored before the structure of Fe/Mo nitrogenase was elucidated through x-ray studies, again must be considered a strong possibility.

    Supporting Online Material

    Materials and Methods

    Fig. S1

    References and Notes

    References and Notes

    View Abstract

    Stay Connected to Science

    Navigate This Article