Nitrogen Fixation (Incorporating the Liversidge Lecture, delivered in London, 18th November, 1971)* By J. Chatt, F.R.S. and G. J. Leigh A.R.C. UNIT OF NITROGEN FIXATION, UNIVERSITY OF SUSSEX, BRIGHTON, BNl 9QJ Nitrogen fixation is topical and the subject of many detailed reviews. Here the intention is to give a concise account of the present state of the subject in the light of our experience and that of our colleagues with reference to reviews and very recent papers only. We shall be concerned with developments over the past decade, which has seen a great surge of interest in the fixation of molecular nitrogen (i.e. dinitrogen) under mild conditions such as might simulate those found in biological nitrogen-fixing systems. Active extracts were obtained from bacteria only eleven years ago and the past six years have seen the development of the whole of the co-ordination chemistry of dinitrogen. This chemistry has added remarkably little to our knowledge of how the natural system works, but there is sufficient circumstantial evidence to suggest that it is relevant.There are two main approaches to the study of nitrogen fixation. One is to study the natural nitrogen-fixing systems found in bacteria and some other very simple forms of life. The other is to attempt to make chemical models of the natural system to see whether any fixation of nitrogen can be achieved under mild reaction conditions such as one would expect to find in bacteria. 1 Biological Nitrogen Fixation This, the more direct approach, has yielded very important biochemical informa- tion and much of interest to the chemist.The biochemical literature is vast (references 1-8 are recent and representative?) and here we can only summarize that of interest to chemists. Only very primitive bacteria and some blue-green algaee fix nitr~gen.l-~ * The lecture was also given in St. Andrews, 1 lth November, Aberystwyth, 25th November, 1971, Aberdeen, 8th March, and Glasgow, 9th March, 1972. t Phrases in square brackets in references are descriptions of the contents of review articles. ‘The Chemistry and Biochemistry of Nitrogen Fixation’, ed. J. R. Postgate, Plenum Press, London, 1971. R. W. F. Hardy and R. C. Burns, Ann. Rev. Biochem., 1968, 37, 331. [Enzymology and chemistry of biological nitrogen fixation.] R.W. F. Hardy and E. Knight, in ‘Progress in Phytochemistry’, ed. L. Rheinhold and Y.Liwshitz, Wiley, London, 1968, p. 407. [Biochemistry of nitrogen fixation.] J. R. Postgate, Symposium SOC.Gen. Microbiol,, 1971, 21, 287; ref. 1, p. 161. [Physiological chemistry of nitrogen fixation.] R. W. F. Hardy, R. C. Burns, and G. W. Parshall, Adv. Chem. Ser., 1971, No. 100, 219. [Biochemistry of nitrogen fixation.] 121 Nitrogen Fixatioti The bacteria have yielded most biochemical information. They are of two main types, symbiotic and free-living. The symbiotic, e.g. rhizobium, are the more important.10 They have normal bacterial lives in the soil but they do not fix nitrogen there. However, certain plants, especially the legumes, grown in soil deficient in fixed nitrogen, exude some substance which attracts a specific species of rhizobium to the root hairs. The organisms enter along the axis of the hair infecting certain cells and causing nodules to grow on the roots.The bacteria multiply briefly within a membrane in the infected cells, then change shape, increase in volume up to 40times, and lose the power of reproduction. In this altered state they are called 'bacteroids' and they can fix nitrogen, providing the plant with ammonia in exchange for carbohydrate. In contrast, the free-living bacteria live and fix nitrogen in the soil but are only important ecologically in barren soils.11 They are the more easily cultured in the laboratory and most of our information concerning the chemistry and biochemistry of the natural process of nitrogen fixation has been obtained from them.In 1930, Bortells showed that molybdenum is essential for nitrogen fixation by bacteria and is closely associated with the metabolism of dinitrogen. The bacteria grow well in the absence of added molybdenum if they are provided with a source of ammoniacal nitrogen but are unable to metabolize dinitrogen in the absence of molybdenum in excess of impurity levels. Indeed, in the presence of ammoniacal nitrogen the bacteria do not produce nitrogenase, the nitrogen-fixing enzyme. However, if the source of ammonia is removed, the bacteria can develop nitrogenase and metabolize dinitrogen within 1-10 h.The amount of molybdenum required by the bacteria is often very small, for example the bacteria Beoerinckia indica require only 0*004--0.034 p.p.m. for half optimal growth, and in culture 0.1 p.p.m. is ample. Vanadium, but no other metal, can be used by some cultures of bacteria in place of molybdenum. Iron, like molybdenum, is an essential trace element and magnesium ions are also necessary for nitrogen fixation. The low solubility of dinitrogen in water does not limit the rate of fixation in bacteria, which is independent of pressure down to 0.3 atm. It seems that the enzyme has an enormous affinity for molecular nitrogen. The temperature is also somewhat critical. Bacteria work best between 15 and 25 "C. They are very slow at 0 "C and most cut offsharply at about 35-40 "C.Studies using I6N2show that the first recognisable product from the natural system is ammonia. There is no evidence for the intermediate formation of 'R. W. F. Hardy and R. C. Burns, in 'Inorganic Biochemistry', ed. G. Eichorn, Elsevier, Amsterdam, 1972. [Bioinorgank chemistry of nitrogen fixation.] 'R. H. Burris, Proc. Roy. SOC.,1969, B172, 339; ref. 1, p. 105. [Fixation by free-living mic- robes: enzymology.] a R. M. Cox and P. Fay, Proc. Roy. SOC.,1969, B172, 357. [Fixation by blue-green algae.] a W. D. P. Stewart, 'Nitrogen Fixation in Plants', Athlone Press, London, 1966. lo F. Bergerson,Proc. Roy. SOC.,1969, B172,401; H.J. Evans and S. A. Russell, ref. I, p. 191. [Fixation in leguminous systems.] l1 W.D. P. Stewart, Proc. Roy. SOC.,1969, B172, 367. Chatt and Leigh hydroxylamine, hyponitrous acid, di-imine, hydrazhe, or any of the other substances which, on the basis of thermodynamic studies, might be intermediates. The dinitrogen in its intermediate stages of reduction is held by the enzyme, only ammonia being freed. Indeed, hydrazine is not reduced by the enzyme but poisons it. It is often said that the bacteria are very efficient nitrogen fixers because they operate under mild conditions of temperature and pressure, with carbohydrate or similar reducing agent. However, in normal laboratory culture, they consume enormous amounts of energy in the form of carbohydrate; by this criterion they are highly inefficient. The most efficient fix only about 15 mg of nitrogen per lg of glucose consumed, but laboratory systems devised to simulate natural condi- tions more precisely suggest that bacteria may actually be up to three times more efficient than this.It is claimed tha# the treatment of soil with waste molasses stimulates nitrogen fixation. The enzyme nitrogenase, when extracted and tested in vitro, requires consider- able amounts of adenosine triphosphate (ATP) to function. Approximately 15 moles ATP are hydrolysed to reduce 1 mole of dinitrogen to 2 of ammonia, a free energy dissipation in the region of 105 kcal mole-’ of dinitrogen. The seemingly high energy requirement of the reduction is puzzling. If dinitrogen and dihydrogen were to react to equilibrium at atmospheric pressure and temperature, heat would be evolved and the equilibrium concentration of ammonia would approach 100%.Thus only an efficient catalyst is necessary and no additional energy in any system which can generate dihydrogen. All nitrogen-fixing bacteria contain hydrogenase and can generate dihydrogen (although not all hydrogenase-containing bacteria fix nitrogen). The bacteria do not utilize dihydrogen as such, and indeed, dihydrogen is a competitive inhibitor of nitro- gen fixation. Presumably dihydrogen can compete with dinitrogen for the active site in the enzyme. The nitrogenase system is not as specific as most enzyme systems. It will reduce a number of dinitrogen-like molecules, such as acetylene, which are competitive inhibitors of nitrogen fixation.Carbon monoxide is not reduced, but is a strongly competitive inhibitor. Table 1 lists such ‘dinitrogen analogues’ and their reactions with the bacteria or enzyme extracts. Xn some of these the triple bond is reductively split as in dinitrogen, but acetylene is reduced to ethylene free from ethane or methane, and dideuterioacetylene yields cis-dideuterioethylene. All those substances in Table 1 which have a well-developed co-ordination chemistry also have a high affnity for class b transition-metal ionsla although other such substances, e.g. C2H4 or P(CH20H)s, do not inter- fere with the nitrogenase system, perhaps for steric reasons. Only small rod- like molecules interfere ;methylacetylene is reduced more slowly than acetylene and ethylacetylene slower still.Dimethylacetylene and diphenylacetylene are not reduced at all. S. Ahrland, J. Chatt, and N. R. Davies, Quarr. Rev., 1958, 12, 265. [Class a and class b metal ions.] Nitrogen Fixation Table 1 Substances which interact with the nitrogenase system Substance Interaction or reduction product Relative ratesa NZ 2NH3 1.0 CZH2 C2H4 4-0 N3-NH3, Nz 3.0 N2O Nz,HzO 3.0 MeNC CH4, MeNHz, traces of CzHz, CzHa, CzH6, etC. 0.8 MeCN C2H6 0.004 EtCN C3H8, NH3 0.003 CHZyCHCN C3H6, C3H8, NH3 0-2 CN-CH4, NH3, MeNHz 0.6 HZ Competitive inhibition of nitrogen fixation co Strong inhibition of nitrogen fixation NO Inhibition NZH* In hi bi tion Obtained from the rate of formation of reduction products from an enzyme saturated with substrate5 The reduction of acetylene to ethylene provides a very quick quantitative test for nitrogen fixation because the ratio of the two gases is readily deter- mined in minute concentration by gas-liquid chromatography.It is often useful to measure fixation in soil under natural conditions, or to screen bacteria for nitrogen-fixing activity, but positive acetylene tests on bacteria should always be confirmed by mass spectrometric tests of lsNz metabolism. No anomalous result has yet been reported from the acetylene test. Before 1960 all attempts to obtain active nitrogen-fixing cell-free extracts from bacteria failed, leading to the view that nitrogenase was very unstable. Then Carnahan, Mortenson, Mower, and Castle, working with Clostridium pasteurianum, discovered that it was essential to exclude dioxygen from the bacterial extracts.Their discovery sparked the present phase of interest in nitrogen fixation. They also found that nitrogenase alone did not cause the reduction of dinitrogen. The system also needed an electron carrier, such as ferredoxin, and a reducing agent (sodium pyruvate) which appeared to be somewhat specific because, as later discovered, it activated an ATP-generating system. The whole had to be buffered to about neutral pH. Later Bulen, Burns, and LeComte showed that sodium dithionite is an excellent artificial reducing agent which can replace both the electron carrier and the more natural reducing agent.They produced a very simple method for the assay of nitrogenase. Sodium dithionite and an ATP-generating system (adenosine diphosphate, creatine phosphate, and creatine kinase) were added to the buffered bacterial extract under argon. This caused the evolution of dihydrogen at a rate which was measured. In a parallel experiment under dinitrogen the rate of dihydrogen evolution was lower. The decrease in rate of dihydrogen evolution was exactly Chatt and Leigh equivalent to the rate of nitrogen fixation. An assay by this method can be done in 20 min, and its discovery facilitated the concentration and purification of nitrogenase fractions, although it has now been superceded by the acetylene test. The nitrogenase system produces electrons at a constant rate, regardless of the substrate, so that the total reduced product (e.g.dihydrogen plus ammonia or other reduction products, depending on the substrate) has the same chemical equivalence. Even under carbon monoxide, when the reduction of substrates is inhibited, dihydrogen is evolved as if no substrate were present. The relative rates of reduction in Table 1 indicate the efficiency of reduction. Thus acetylene, needing two electrons for reduction, should be reduced three times as fast as dinitrogen, which requires six. In fact it is reduced four times as fast, and proportionately less dihydrogen is produced. Thus it is reduced more efficiently. Nitrogenase from several sources has now been examined and has been obtained from three genera of bacteria in a state of what is believed to be high purity.It can be separated into two proteins, one of molecular weight 2-3 x lo5 and the other around 7 x 104; these may be broken down into sub-units. The larger protein contains molybdenum and iron, and the smaller only iron, as essential metallic constituents. There is still some doubt about the molecular weights and the exact metal content of the two proteins and of the character of the sub-units. Neither protein shows nitrogenase action on its own but a mix- ture is immediately active. There is obviously no winding together of protein chains or other slow process involved in their concerted action. The large pro- tein from one species of bacteria can be activated by the small protein from another if the bacteria are sufficiently closely related, but the activity falls as the difference between the two types of bacteria increases, This would indicate that the two proteins are essentially the same but have peripheral differences which prevent their fitting together when they come from widely different bac- teria. The nomenclature of the proteins is somewhat confusing as almost every worker in the field appears to have developed his own.The larger protein has been variously called enzyme 1, protein 1, fraction 1, component 1, molybdo-ferredoxin, azofermo, Fe-Mo protein, and Mo-Fe protein; the smaller has been referred to as enzyme 2, protein 2, fraction 2, component 2, azo ferredoxin, azofer, and Fe protein.We shall call them the molybdenum-iron (or Mo-Fe) and the iron (or Fe) proteins. The Mo-Fe protein is brown in aqueous solution; it appears to contain 1 or, less probably, 2 atoms of molybdenum and around 15 atoms of iron per molybdenum atom. One mole of dihydrogen sulphide per iron atom is liberated by acid. A solution of the Fe protein is yellow and contains 2 atoms of iron and 2 ions of ‘labile sulphide’. The proteins are unusual only in containing a somewhat higher than normal proportion of sulphur, leading to the suggestion that perhaps the iron and molybdenum are in a sulphur environment. There is no direct evidence as to how the metals are held, or even whether they are directly involved with dinitrogen during the fixation process. Mossbauer spectra of 67Fe-labelled proteins under various conditions and with various Nitrogen Fixation substrates have not given any definitive information, but they suggest that the iron atoms do not react directly with the dinitrogen and probably occur in pairs.It appears that, when all conditions for nitrogenase activity are satisfied (ATP, reductant etc.), the Fe protein catalyses a reduction of some of the iron in the Mo-Fe protein. E.s.r. spectra have shown signals typical of non-haem iron but no reproducible signals have been obtained from molybdenum. The electronic, o.r.d., etc. spectra have not yet given any useful information. The oxidation state of molybdenum is unknown and iron is probably present as iron(1r) and iron@) in the working enzyme.Both proteins are oxidized by air. If not over-exposed to air, the Mo-Fe protein can be reactivated by sodium dithionite, but the Fe protein is destroyed irreversibly. The aerobes, azotobacter, have a high respiratory rate and it has been suggested that this is necessary to consume dioxygen, so as to protect their nitrogenase. There is a limit to this mechanism of protection and as soon as the dioxygen tension in the environment becomes too high they switch off their nitrogen-fixing activity as though they had pulled down shutters to keep out the dioxygen (and dinitrogen). When the dioxygen tension drops the azotobacters start to fix nitrogen again at the critical threshold tension of dioxygen. The Mo-Fe protein has been obtained crystalline from Azotobacter vinelandii, but from no other source.The crystals are colourless and have curved faces, thus they are not suitable for X-ray structure determination. They mush very readily to a brown, highly active solution of the large protein. The two nitrogenase proteins have sometimes been attributed distinct func- tions, one as a dinitrogen-trapping agent and the other as the reducer, but apart from indications from Mossbauer spectroscopic methods, there is no definite evidence concerning their functions. The absence of reactions with bulky molecules related to those in Table 1 suggests that access to the active site is probably restricted. The site appears to have extreme class 6 metal character. The production of C2hydrocarbons from the reduction of methyl isocyanide has been held to support the hypothesis of a bimetal site, the two adjacent metal atoms being necessary to bring together the two terminal carbon atoms of the isocyanide, but it is equally consistent with a free-radical mechanism and only one site of attack.Indeed, most transition- metal ions in the presence of reducing agent produce some C2hydrocarbons from isocyanides. Studies of the relative rates of reduction of various substrates and of dinitrogen by nitrogenase fractions, and of the effects of inhibitors on the reductions and on hydrogen evolution, have given much valuable information about the biochemistry of nitrogen fixation. This is summarized well by Burris;' such studies give no hint as to the chemical mechanism of the reduction of dinitrogen on nitrogenase. 2 Chemical Nitrogen Fixation Although such evidence as described above points to the activation of dinitrogen by metal ions in the natural system, it is conceivable that highly reactive or strained organic species might use their energy to activate dinitrogen.Thus the Chatt and Leigh phenylsulphenium ion (PhS+) reacts with Nain trace amounts to form a material which may contain [PhSNJ+. Diazomethane is photolysed reversibly, carbene must thefore react with dinitrogen. Finally a salt, formulated NaJC {CHOP- (S)CS](I, is said to react with dinitrogen ‘greedily’ to give a product in which the nitrogen atoms are still linked. We have failed to confirm this last.None of these routes is likely to provide a ready source of fixed nitrogen. Studies of biological fixation suggest that it occurs at a metal site which is sensitive to dioxygen. The site may be in an anaerobic and/or anhydrous region of the enzyme, but there is no evidence concerning this. However, the rate of dihydrogen evolution from the enzyme increases in the absence of a substrate which indicates that protons can reach the reducing site. Most attempts to produce chemical model systems have been based on the assumption that moly- bdenum and iron are key elements. This has led, over the past quarter of a century, to the study of the interaction of dinitrogen with mixtures of transition- metal compounds and the development of three main types of chemical approach.These are now all being studied in various laboratories, and in chronological order of initiation they are as follows: (a) Reduction with aqueous reducing agents, Dinitrogen is bubbled through, or compressed on to, mixtures of reducing agent and transition-metal catalysts dissolved or suspended in water, and after reaction, if any, the solution is tested for ammonia or other reduced products such as hydrazine. (b) Nitriding. The aprotic reduction of dinitrogen is brought about by a transition-metal compound mixed with a very strong reducing agent such as an alkali metal. The product behaves as a nitride. (c) Study of dinitrogen complexes. This is a fundamental study of the inter- action of dinitrogen with transition-metal salts, which appears to be relevant to the functioning of the enzyme.Obviously these three approaches are related. Thus dinitrogen complexes are probably precursors to nitrogen fixation by reduced transition-metal systems, and may be intermediates in the reduction of dinitrogen to ammonia in the biological system. However, the chemical work at present falls neatly into the three sections noted above. Because it provides fundamental background information, the chemistry of dinitrogen complexes will be taken first. A. Dinitrogen Complexes.-Dinitrogen complexes are the only type of com- pound formed by elementary nitrogen under mild reaction conditions, i.e. in water or alcohols at ordinary temperature and pressure. There are two classes: terminal-dinitrogen complexes in which the dinitrogen molecule occupies one co-ordination place in a transition-metal complex, and bridging-dinitrogen complexes where dinitrogen bridges two metal atoms, one or both of which may be transition metals.Their general chemistry and their relation to bio- logical and abiological nitrogen fixation have been reviewed in detail.lS-O1 In J. Chatt, Proc. Roy. SOC.,1969, B172,327; Pure Appl. Chem., 1970,24,425.[Chemistry of nitrogen hation.] J. E. Fergusson and J. L. Love, Rev. Pure Appl. Chem., 1970,20,33.[Dinitrogencomplexes.] Nitrogen Fixation this outline we shall consider mainly those aspects relevant to nitrogen fixation. Terminal-dinitrogen Complexes. The electronic structure of dinitrogen is similar to that of carbon monoxide.The filled orbital of highest energy in both molecules has o-symmetry and low energy, but owing to the greater electronegativity of nitrogen relative to carbon, dinitrogen is a poorer electron donor. Its ionization potential of 15.6 eV is almost equal to that of argon and considerably higher than that of carbon monoxide (14.0 eV). Nevertheless, it forms a range of mononuclear complexes which are in many ways similar to those formed by carbon monoxide, and a similar bonding scheme has been postulated for them (Figure 1). The dinitrogen complexes are, however, far less numerous. Figure1 Bonding scheme for dinitrogen complexes. Electrons are donated from the ditritrogen 3ogorbital into a hybrid acceptor orbital on the metal ion, and a d (or dp hybrid) orbital of the metal donates electrons back into the anti-bonding dinitrogen lx,-orbital The first dinitrogen complex, [RU(NH,),(N,)]~+, was isolated by Allen and Senoff in 1965, during the attempted synthesis of [RU(NH,)~]~+ by the reaction of hydrazine with ruthenium trichloride trihydrate in water.Its solid salts were characterized by a very strong band in their i.r. spectra at about 2100 cm-I, the precise frequency being dependent upon the counter-anion. About 25 mononuclear complexes of dinitrogen have now been characterized. All have the dinitrogen bound in end-on configuration (i.e. they are monohapto-complexes) with the M-N-N system essentially linear in all cases investigated by X-ray analysis (see Table 2).In the complexes, the N-N distance is increased by less than 0.015 A from that in N2itself (1.098 A). The elements italicized in Table 3 are those which are at present known to form crystalline mononuclear dinitrogen complexes. They lie about a line drawn from nickel to tungsten in the Periodic Table, the most stable Complexes being formed by elements towards the tungsten end. All but a very few of the dinitrogen complexes obey the effective atomic number rule. A representative selection is shown in Table 4. It is evident that l6 Yu. G. Borodko and A. E. Shilov, Russ. Chem. Rev., 1969, 38, 355. [Dinitrogen complexes and nitrogen fixation.] R. Murray and D. C. Smith, Co-ord. Chem. Rev., 1968, 3, 429. [Activation of molecular nitrogen.]A.D. Allen and F. Bottomley, Accounts Chem. Res., 1968, 1, 360. [General review of dinitrogen complexes.] lEG.J. Leigh, Preparative Inorg. Reactions, 1971.7, 165. [Synthesis of dinitrogen complexes.] J. Chatt and R. L. Richards, ref. 1, p. 57. [Dinitrogen complexes and nitrogen fixation.] zo G. J. Leigh, ref. 1, p. 19. [Abiological nitrogen fixation.] *l G. Henrici-Olivk and S. Olive, Angew. Chem. Internat. Edn., 1969, 8, 650. [Non-enzymatic nitrogen fixation.] Chatt and Leigh Table 2 X-Ray strirctural data on some dinitrogen complexes A N-N (A) 1.16 M-N (A) 1 a80 M-N-N (") 175 1 * 101(1 2) 1.784(1 3) 178(2) N 1.03 1 * 12(08) 2-1O(04) 2-1qol) --180 1.1 24( 15) 1.928(6) 178*3(5) 1-106(11) 1.894(9) 179-3(9) 1 *055(30) 1 -966(21) 177( 1) Two crystallographically independent molecules in the unit cell; b en = ethylenediamine;distances and angles refer to the complexed dinitrogen dinitrogen differs from carbon monoxide mainly in its tendency to form mono- or occasionally bis-dinitrogen complexes, and only with metals in moderately low oxidation states (o-II), whereas carbon monoxide also tends to form poly- carbonyl complexes with metals in exceptionally low oxidation states (-r1-0).In its behaviour as a ligand, dinitrogen is thus closer to organic isonitriles than to carbon monoxide. The dinitrogen stretching frequencies, v(N2),in the i.r. spectra of dinitrogen complexes parallel those of v(C0) in the analogous carbonyl complexes, con- sistent with similar bonding systems for both carbon monoxide and dinitrogen in their complexes.Theoretical considerations, as well as i.r. and Mossbauer spectroscopy, suggest that dinitrogen is both a weaker a-donor and a weaker r-acceptor than carbon monoxide. In its terminal complexes, the dinitrogen molecule is strongly polarized, as indicated by the strong v(N2)bands in their i.r. spectra. The charge difference between the two nitrogen atoms in [ReCI(N,)-(Ph2PCH2CH2PPh,),] has been estimated from its X-ray-induced electron emission spectrum to be about 0.4 electrons. Dinitrogen might also form side-on, or dihapto-complexes, but none have been characterized. The main evidence concerning their stability comes from a study of the isomerization of [RU(NH,),(~~N~~N)]~+ to [RU(NH~)~(~~N~~N)]~+, which has an activation energy of about 21 kcal mol-l in aqueous solution.The aquation of this ion with loss of dinitrogen has an activation energy of 28 kcal mol-l. The isomerization occurs by tumbling of the dinitrogen molecule, passing through the dihupto-position. The energetics of the process indicate that the acti- vation energy of displacement of dinitrogen from the dihapto-position by water is of the order of 7 kcal mol-l, i.e. the dihapto-dinitrogen is almost free. It thus seems unlikely that dihapro-complexes could be stable relative to monohapto- complexes at any temperature, although they might exist at very low tem-perat ures. Terminal-dinitrogen complexes are usually prepared by the reduction of a transition-metal complex under dinitrogen, or from a two or three nitrogen atom chain in a ligand attached to the meta1.18~*g Dinitrogen is very specific as 129 Nitrogen Fixation to the type of transition-metal site which it will occupy and this appears always to possess strong class b character.The stabilities of dinitrogen complexes are very dependent upon the co-ligands with dinitrogen, and it is critical to have the right co-ligands in the reduced complex. The only guide to their choice is experience. Thus, the ruthenium(I1) and osmium(I1) dinitrogen complexes are particularly stable with five ammonia molecules as co-ligands {i.e. [(Ru,Os)-(NH3)S(N2)]2+}and not with many others, except ethylenediamine. A11 metals along the W-Ni line (Table 3) give isolable dinitrogen complexes with tertiary Table 3 The occurrence of dinitrogen complexes Ti V Cr Mn Fe co Ni Zr Nb Mo Tc Ru Rh Pd Hf Ta W Re 0s Ir Pt organic phosphines, usually also with hydride or halide ions as co-ligands.The lighter transition elements (Fe, Co, and Ni) tend to use hydride and the heavier (Re, Os, and Ir) use halide. Molybdenum(o), tungsten(o), and nickel(o) give dinitrogen complexes with phosphines only as co-ligands (Table 4). Only rarely have tertiary arsines been satisfactory co-ligands, and never yet have organic sulphides. The critical nature of the co-ligands is well illustrated in the series of complexes [Ir(N2)C1(PR3)2] (PR8 = tertiary phosphine). The complex is readily produced by reaction (1) when R = Ph, but analogous products containing similar phosphines such as methyldiphenylphosphine and tri-p-tolylphosphine are too unstable to be isolated in pure condition.Only two reasonably extensive series of stable dinitrogen complexes have been discovered. These are of the types [ReCI(N,)(PR,),] and [OsCI,(N,)(PR,),]. The critical dependence of stability on co-ligands with dinitrogen explains why the discovery of dinitrogen complexes was so long delayed in spite of many attempts to obtain them during the 1940's and ~O'S,and also why there are so few relative to the numbers of the analogous carbonyl complexes. For example, the carbonyl analogues of the above iridium(1) complex can be prepared from almost any tertiary phosphine, and are relatively stable substances.Typical reactions (2)-(7) of formation of dinitrogen complexes are as follows: (a) By direct reduction: THF [MoCI,(THF),] + Ph2PCH,CH2PPh, + Na-Hg + N2+ 15 "C trans-[Mo(N,),(Ph,PCH,CH,PPh,)J + NaCl (2) Et,O [Co(acac),] + AlEt, + PPh, + N2+[CoHCN,)(PPh,),] (3)0 "C Chatt and Leigh Table 4 Some typical terminal-dinitrogen complexes Metal electron Compound trans-[Mo(N2)2-(Ph ,PCH ,CH 2PPh 2) 2] [Mo(.rr-PhMe)(N,)(PPh,),l cis-IW(Na)d?'Me2Ph)J trans-[ReCl(N 2)(PMe2Ph)41 [ReCI(N,)(Ph,PCH,CH,PPh,),] + Colour Orange-yellow Orange Yellow Yellow Green or figuration con-ds ds de ds ds v(N2)a (cm-l) 2020, 1970(Nujol) 2005(toluene) 1931, 1998(benzene) 1925(chloroform) 2060(chloroform) ds 2105,2020, 1935 (chloroform) Yellow Orange Pale da ds d6 2057(Nujol) 2090(Nujol) 21 30(water) yellow White ds 2147(Nujol) Pale yellow White d6 ds 2220, 2190 (not stated) 2082(benzene) White Red Black Red- ds d@ dl0 d8 2057(benzene) 2093(THF) 1875(THF) 2096(Nujol) orange Yellow d8 21 O5(chloroform) Light d8 2152 (not stated) yellow Orange-yellow do 2076(hexane) d1 not i .r.-act ive ds 1910 (solid state, Raman) da 2100 (solid state, Raman) 2040 (not stated) not i .r.-active 2028( benzene) 0 Solvent or dispersion medium in brackets; b v(N3 and v(C0) are extensively coupled; cen = ethylenediamine; d Cy = cyclohexyl; i.r.-active 0,)probably due to dissociation in solution to [Ni(PCy,),(N,) J Nitrogen Fixation [FeHC1(Et2PCH2CH2PEt2),]+ Na [SPh,] + N, --+ [FeH(N,)(Et,PCH,CH,PEt,),] [BPh,] + NaCl (5) (c) By degradation of a nitrogen chain.Allen and Senoff's preparation and the formation of the iridium complex (see above) are of this type. A further example is : No\ MeOH[.(Ph3P)2ClzRe=N-N=C-P~)] + PR, reflux ?rans-[ReCI(N,)(PR3),] + PhC0,Me + PPh, + HCI (6) The azide ion has also been used as a source of co-ordinated N, and provides the best preparation of Allen and Senoff's compound. [Ru(NH~)~(H~O)]~++ N,-+[Ru(NH3),(N2)l2++ 4-N2 + H2O (7) Many terminal-dinitrogen complexes which are formed as impure products by such reactions as the above and recognised by their strong i.r.bands are too unstable to be isolated. Most reagents react with dinitrogen complexes with displacement of dinitrogen and formal oxidation of the metal, but certain ligand molecules such as carbon monoxide displace dinitrogen without the oxidation. The reaction (8) of dihydrogen with [CoH(N,)(PPh,),], and similarly with a number of related complexes, may be relevant to the natural pracess. [CoH(N&.PPh3)3] H2*[CoH3(PPh3)31 N2 (8) This reversible reaction occurs in ethanol and shows that dinitrogen and di- hydrogen can compete for the same metal site. Also carbon monoxide, which is a strong inhibitor of nitrogen fixation in the natural system, displaces dini- trogen irreversibly from this cobalt complex. Thus, if nitrogenase contained a metal site of the type in the cobalt complex, the competitive inhibition of nitrogen fixation by dihydrogen and the strong inhibition by carbon monoxide would be allowed for.When dideuterium is used in place of dihydrogen in reaction (8) it replaces all 18 o-hydrogen atoms from the phosphine ligands as well as the hydrogen on the metal, indicating that reversible formation of 0-C-Co bonds with loss of dihydrogen occurs. The less-stable dinitrogen complexes readily lose dinitrogen to leave highly reactive metal-containing species, e.g. [CoH(PPh,),], often excellent olefin polymerization catalysts. Dinitrogen complexes are especially sensitive to Chatt arid Leigh oxidation with loss of dinitrogen, but a few of the very stable complexes, e.g., trans-[ReCl(N,)(PMe,Ph),], are able to retain their dinitrogen and yield a pure product, e.g., trans-[ReCI(N,)(PMe,Ph),]+, when the metal is oxidized to the next highest oxidation state.The N-N stretching frequency is then raised by about 80-100 cm-l, owing to weaker back-bonding from the more positive metal of the oxidized species. All attempts to reduce well-defined terminal-dinitrogen complexes to ammonia have failed. Very strong reducing agents capable of producing nitride from elementary nitrogen in aprotic solvents are capable of producing nitride from the complexes, but this may occur through preliminary dissociation of the complex. A very few dinitrogen Complexes are sufficiently stable for ligand exchange reactions without loss of dinitrogen.One of the most remarkable, reaction (9), occurs in boiling toluene. trans-[ReCl(N,)(PMe,Ph),] + 2PhzPCH2CH2PPh--+ trans-[ReCl(N,)(Ph,PCH,CH,PPh,),] + 4PMe2Ph (9) Although the less stable dinitrogen complexes react with potential ligand molecules with irreversible displacement of dinitrogen, some react reversibly, e.g. reaction (10). This reaction is important because it shows that if dinitrogen were to be reduced on some metal sites, the ammonia produced could be displaced by dinitrogen so that the cycle of reduction would continue. The terminal nitrogen atom of co-ordinated dinitrogen in some of the more stable complexes is an electron donor, and reacts with various electron acceptor systems to form bridging-dinitrogen complexes (see below).Stable analogues of the dinitrogen complexes are usually obtained from carbon monoxide, organic cyanides, or organic isocyanides. In all such cases the stretching frequency of the triple bond is considerably lower than in the usual run of complexes of these ligands, indicating a very strong back-donation of d-electrons from the metal atom into the 7r-anti-bonding orbitals of the triply bonded system. This appears to be the main characteristic of a dinitrogen- bonding site. It has a high d-electron density on the metal, and this is probably responsible for the weak bonding of normally strong ligands such as water and ammonia, which have no mechanism for accommodating the d-electronic charge. Since the donor orbital on dinitrogen has very low energy, the a-acceptor on the metal will also have to be on a low energy for good mixing.Thus, for the binding of dinitrogen the metal must provide a very low energy a-orbital to match the o-orbital of the dinitrogen, and it must also have filled d-orbitals of sufficiently high energy to combine with the anti-bonding 7r-orbitals of the dinitrogen. These are conditions difficult to satisfy simultaneously. Their ful-fillment depends critically on the metal, the co-ligands with dinitrogen, and their positions relative to dinitrogen. Minor changes in any of these render the site much less capable of holding dinitrogen. Nitrogen Fixation Although the terminal nitrogen atom of some dinitrogen complexes serves as an electron donor to many acceptor systems, it does not appear to do so to the proton. If it did, the reduction of the dinitrogen ligand would be com- paratively easy. However, since the electron density on the dinitrogen ligand is enhanced largely by electron drift from the d-orbitals of the metal, it is evident that the metal must also have a high electron density.Thus protic acids appear to protonate the metal, leading to its oxidation and the release of the dinitrogen. It therefore appears unlikely that dinitrogen in terminal-dinitrogen complexes can be reduced as a consequence of attack by protons or by hydride ions. There is no evidence that dihydrogen is effective either. It may be that the electronic condition of bridging-dinitrogen is more favourable for reduction.Bridging-dinitrogen Complexes. Dinitrogen is also like carbon monoxide in being capable of bridging between metal atoms. Carbon monoxide normally bridges between two class b transition-metal atoms through the carbon atom, as in the usual form of (Co,(CO),]. However, in a very few cases it bridges through carbon and oxygen in structures of the type M-CEO-M1, where M is a class b transition metal and M1a strong class a acceptor, e.g. Me3AI. In contrast, only the M-N=N-M1 type of bridging has been found for dinitrogen and it occurs whether M and M1 are both strong class b transition-metal acceptors, or only M, with M1 a strong class a acceptor, e.g. Me,AI, CrCl,(THF),, or PF,. The first bridged complex was the ion [(NH3)5Ru(N2)Ru(NH3)5]4+,prepared in aqueous solution by a remarkable series of reactions: [RU@TH,),CI]~++ Zn + 2H+ -[Ru(NH3),(H,0)J2++ Zn2++ H, (11) [Ru(NH3)s(H20)la+-t-N2 -+[R~(NH~)S(N~)~~+ (12)+ H2O + +[Ru“H 3) 6(w2 + [Ru(NH3) 5(H20)l2 + -NNH 3) ,RU(N,)RU(NH 3) 6J4 + H20 (13) In this bridged ion the Ru(N,)Ru system is essentially linear, but the N-N separation is insignificantly different from that in [Ru(NH,),(N,)]~+, and only marginally greater than in dinitrogen (see Table 2).In solid [Ru(NH,),(N,)J- [BFJ,, v(N,) (Raman spectrum) is only 31 cm-l higher than in the correspond- ing salt of the dinuclear ion. Thus dinitrogen seems to have been altered very little on forming a bridge. A kinetic study of reaction (13) supports this by show-ing that the dinitrogen complex has about the same affinity for the aqua-complex as has dinitrogen itself. Other similar bridged species where dinitrogen connects closed-shell atoms are WH3) ,Ru(N 2)OS(NH3) 514 +, [(C,H,)(PPh 3) 2Mo(N2)Mo(PPh3) o(GH and [(P~,P),CO(N,)CO(PP~,)~].~~They have had little study.There is also a nickel complex with tricyclohexylphosphine which dissociates in solution : [(CY3P),Ni(N,)Ni(PCY,),I+ “i(N,)(PCY,)*l + “i(PCY3)J 22 M.L. H. Green and W. E. Silverthorn, Chem. Comm., 1971, 557. 2J M. Aresta, C. F. Nobile, M. ROSS~,and A. Sacco, Chem. Comm., 1971, 781. Chatt and Leigh In addition to the above bridged complexes, there is a series of dinitrogen- bridged complexes derived from [ReCl(N,)(PMe,Ph),], which has the lowest v(N,) (1925 cm-l) so far observed in any neutral complex.This may be an indication of a greater back-donation of d-electrons in this complex than in any other, with consequent raising of the electron density on the dinitrogen ligand. It appears to have donor properties very similar to those of acetonitrile, and interacts with a great number of acceptor molecules of all types, sometimes with evolution of dinitrogen (e.g. with BCI,), but often to form strongly coloured products which are essentially adducts. In these the N2molecule is believed to bridge between the rhenium atom of the original complex and the acceptor atom of the added reagent. Owing to the asymmetry of the product, v(N,) is i.r.-active and occurs at lower frequencies than in the parent rhenium complex. Table 5 Table 5 Adducts of trans-[ReCl(N,)(PMe,Ph),] with various Lewis acids Acid derivatives with Ratio configurationdo-d2 Colour v(N,) (cm-l) Re:acceptor atom Blue 1805 1 :1 Red not 1 :2 Brown assignable 1695 1 :I 1795 1 :1 Blue 1680 1 :2 [MoCI,(PMe,Ph) 2] Purple not 1 :2 assignable PFS Pale pink 1640 1 :1 Acid derivatives with configurationsd3-d10 or without available d-orbitals [CrCl3(THF)J Purple 1875 1 :1 CoCl,(THF),.s Blue 1855 a [Pt,Cb(PEt3)21 AlMe, Yellow Pale 1890 1894 a 1 :1 yellow a Dissociates upon work-up,pure adduct not obtained lists some of the acceptor molecules which are known to react, together with v(N,) of the product and the number of acceptor atoms in the rhenium dini-trogen-bridged product.Combination may occur with displacement of a ligand, e.g. with [CICla(THF),](reaction 14),or by direct addition, as with PFs (reaction 15). [ReCI(N,)(PMe,Ph),] + [CrCl,(THF),] -+[(PMe2Ph)4ClRe(Na)CrCI,(THF)2] [ReCl(N,)(PMe,Ph)J + PF6 -[(PMe2Ph),ClRe(N,)PFSJ + THF (14) (1 5) 135 Nitrogen Fixation These adducts and the symmetrical, bridged compounds first discussed appear to be of very similar type. The rhenium-containing adducts fall into two distinct classes, those where v(N2) is lower than in the mononuclear rhenium complex by <100 cm-l (formed by d3-d10 acceptors), and those where the lowering is considerably greater, up to 350 cm-l (formed by do, dl, and d2 acceptors).In some cases, e.g., 1 :2 adducts with titanium or molybdenum derivatives, it is not possible to assign a frequency to v(N2), presumably because it has been lowered into the region of the other ligand frequencies. The original rhenium complex can be recovered from all of these adducts by hydrolysis of the added molecule or otherwise, and so the N2molecule has remained intact. The reason for the formation of two classes of adduct is apparent from the simplified n-molecular-orbital scheme for the Re-N-N-M1 system given in Figure 2. The N2molecule provides the four electrons necessary to fill the le orbitals and the rhenium atom the six electrons necessary to fill the lb2, and the 2e orbitals which are anti-bonding on N2.Thus when the acceptor atom M1is in 3e (non-bonding) 9b2 I I I wI (non-bonding) -. I '"2 I 2eI I I 1 le Figure 2 Suggested molecular-orbital scheme for the x-system of an adduct of trans-[ReCI(N,)-(PMe,Ph),] with a transition-metal derivative Chatt and Leigh a door d1state, it can, by withdrawing electrons from the bonding system of the N, and feeding none back into the .rr-orbital system, cause a great lowering of v(N,). However, when it is a d3A10system, the additional electrons it contains must enter the 3e orbitals which are bonding on dinitrogen so compensating for any electron withdrawal from the N-N a-system or le r-system, and caus- ing very little lowering of v(N,) in the adduct. Thus, when both metals bridged by dinitrogen have closed electronic shells as in the compIex [(NH,),-Ru(N,)Ru(NH3)J4+, v(N,) is almost unchanged from that of the mononuclear analogue [Ru(NH,),(N,)]~ +.One of the most interesting of the [ReCI(N,)(PMe,Ph),] adducts is that with MoOC1,L (L = Et,O, PMe,Ph, etc.) attached to the terminal nitrogen atom (see Table 4). The low v(N,) may be a reflection of a weak N-N bond. It seems likely that, were it possible to attach two moieties of MoOC1,L type, one to each end of the dinitrogen molecule in a reducing medium, they might pull it apart. This is probably the basis of formation of nitride complexes from dinitrogen by transition-metal species which do not have closed electron shells. It suggests that transition metals with open electron shells should form dinitrogen complexes which might be unstable and reactive.Evidence is now appearing that dinitrogen can indeed form complexes with d1electronic systems. Thus there is electrochemical evidence that titanium(@ chloride in dimethyl sulphoxide or propylene carbonate forms complexes with dinitr~gen,,~and more recently a d1 complex, probably [(n-C6H5),PhTi(N2)- TiPh(r-C5H5),], has been It is formed by reaction of [TiPh(.rr-C,H,),] in toluene solution with dinitrogen. Reaction occurs at about -80 "C, and the solid complex is stable below 0 "C. None of the above dinitrogen complexes have been reduced to ammonia or to hydrazine. Indeed, since the 4e orbitals are on a higher energy than the corre- sponding anti-bonding orbitals in free dinitrogen (see Figure 2), the dinitrogen in the complexes is more difficult to reduce to nitride than is elementary nitrogen. Nevertheless, it is now becoming apparent that some very unstable bridging- dinitrogen complexes formed in aprotic media can be reduced to hydrazine and ammonia.These will be considered at the end of the next section. 3 Nitriding Lithium, magnesium, the alkaline-earth metals, and most transition metals of Groups IV-VII form nitrides from dinitrogen exothermically, generally at a dull red heat; indeed, titanium is reported to burn with incandescence when so heated. At room temperature, nitriding occurs rapidly on chemically clean metal surfaces but it does not occur in the bulk metal or is very slow, except on lithium and the alkaline-earth metals.These build up a visible layer, or even nitride completely through the bulk of the metal (e.g. Ca). If fresh lithium metal is continually exposed to dinitrogen, for example, by very slow addition of ethyl bromide to lithium in boiling dry pentane under dinitrogen, yields of UD to 40% T. C. Franklin and R. C. Byrd, Inorg. Chem., 1970,9,986. J. H. Teuben and H. J. de Liefde-Meijer, Rec. Truv. chim., 1971,90,360. I37 Nitrogen Fixation Li,N based on the metal can be obtained. Hydrolysis of the lithium nitride gives ammonia with traces of hydrazine. Workers with Zeigler catalysts (e.g. TiCl, + AIEt, in petrol) for polymeriza- tion of olefins became aware of systems which nitride at room temperature during the 1950’s.Dinitrogen can partially inhibit catalysis of olefin polymeriza- tion by some Ziegler-Natta systems. When the catalysts are formed under dinitrogen at atmospheric pressure and room temperature, the precipitated catalyst contains a small percentage of nitrogen which appears as ammonia when the catalyst is hydrolysed. This was discovered independently and deve- loped mainly by the work of Vol’pin in Russia26 and later, again independently, by Van Tamelen in the United States.a7 They showed that the reaction of a simple transition-metal compound, for example titanium tetrachloride or di-isopropoxydichloride,with a strong reduct- ant such as a reactive organometallic compound or an alkali metal, in ethereal soIution, yields products which react with dinitrogen.Details of the reaction are still uncertain, but generally when an organometallic reductant is used, there is an induction period after mixing the reagents before the absorption of dinitrogen begins, and during this period saturated and unsaturated hydrocarbons are evolved. Presumably organo-transition-metal compounds are formed and these decompose to form hydrides or lower-oxidation-state derivatives of the transi- tion metal. It is one or more of these species which react with dinitrogen to produce complexes, which on solvolysis with protic solvents give ammonia, sometimes with hydrazine. Generally, the nitrogen-containing products behave as solutions or suspensions of highly reactive nitride complexes.A highly active and much studied system which is homogenous is formed by the reaction of bis(.rr-cyclopentadieny1)dichlorotitaniumwith ethylmagnesium bromide. It shows an induction period, and the rate of reaction with dinitrogen is dependent on pressure. Russian workers have tended to use 100-150 atm, although others claim that pN, for half the maximum reaction rate is about 0.5 atm. Dinitrogen uptake is a maximum for rather over 4 mol of organometallic reagent or reductant per titanium atom in the above system and is partly in- hibited by strongly solvating solvents. Generally, carbon monoxide, acetylene, olefins, and dihydrogen inhibit fixation as they do in the natural system, indicat- ing that the fixation probably occurs on a class b metal site.However, some of the systems derived from titanium appear to show enhanced fixation in the presence of dihydrogen, which has been ascribed to the hydrogenation of dini-trogen. Dioxygen in small amounts is said not to affect the uptake of dinitrogen by some systems. The systems also reduce ‘dinitrogen analogues’ similarly to the reduction by nitrogenase.2s Any very strong reducing agent appears to be able to activate the titanium M. E. Vol’pin and V. B. Shut, Organometallic Reactions, 1970, 1, 55. [General review of nitriding reactions. 1 E. E. Van Tamelen, Accounts Chem. Res., 1970,3,361. [Systems for chemical modification of dinitrogen.J E. E. Van Tamelen, H. Rudler, and C. Bjorkiand, J. Amer. Chem. Soc., 1971, 93, 3526; G.J. Leigh, unpublished observations. Chatt and Leigh and other early transition-metal systems; Grignard reagents in ether, triethyl- aluminium, metallic sodium in diglyme, lithium-naphthalene in tetrahydrofuran, or reduction at a nichrome cathode have been used. The fixed nitrogen in the system behaves as nitride, and can be so reactive that even tetrahydrofuran solvolyses it slowly to ammonia. Undoubtedly dinitrogen complexes, possibly bridged-dinitrogen species, M-NEN-M (M = metal), are formed as inter- mediate~.~~-~~In titanium systems, titanium(r1) is believed to be the active species, and e.s.r. studies certainly suggest that the principal species in titanium systems is diamagneti~.~~ However, titanium complexes are so labile that any oxidation state which may be needed for nitrogen fixation could be produced by disproportionation of oxidation states, e.g.2TiII +Tio + TiIv. Despite much work and speculation, the mechanism of the reactions involved in the conversion of dinitrogen into nitride are still uncertain, but it is not now thought that hydridic species are important in the system. Most of the systems are not catalytic in the true sense, because solvolysis with destruction of the active species is needed to liberate ammonia. However, by controlled solvolysis followed by removal of the ammonia, a further cycle of reduction, dinitrogen absorption, and solvolysis can be made. Van Tamelen and co-workers have shown that the titanium retains activity through about nine cycles in the 2-propoxydichlorotitanium-diglyme-sodium system using propan- 2-01 for solvolysis.Vol'pin and co-workers, using a non-protic Lewis acid, aluminium tribromide, were able to demonstrate the truly catalytic effect of titanium by treating dini- trogen with a mixture of titanium tetrachloride, metallic aluminium, and alu- minium tribromide at 50 "C,either in the absence or in the presence of a solvent, e.g. benzene. As much as 200 mol of ammonia per mol of TiCl, was obtained after hydrolysis. It is clear that a system for the catalytic nitriding of aluminium had been evolved, since aluminium is not nitrided in the presence of its bromide alone. Circumstantial evidence suggests that titanium@) halides produced by the reaction with aluminium form an active adduct TiXz,nAlBr3, and that this takes up dinitrogen to form eventually a nitride which, with aluminium tri- bromide, generates titanium tribromide and forms aluminium nitride. The titanium(m) is then reduced again to titanium(@ and recycled.The above nitriding systems are also capable of producing nitrogen-containing organic compounds. Thus the bis(wcyclopentadieny1)dichlorotitanium system with an excess of phenyl-lithium in ether solution under dinitrogen at room temperature gives, after hydrolysis, 0.15 mol of aniline per atom of titanium, as well as ammonia (0.65 mol) and trace amounts of u-aminobiphenyl; both formation of a benzyne intermediate and insertion of N2into a titanium--carbon ns M.0. Broitman, N. T. Denisov, N. I. Shuvalova, and A. E. Shilov, Kinetika i Kataliz, 1971, 12, 504, and references therein. *O E. E. Van Tamelen, D. Seeley, S. Schneller, H. Rudler, and W. Cretney, J. Amer. Chem. SOC.,1970, 92, 5251, and references therein. s1 J. E. Bercaw and H. Brintzinger, J. Amer. Chem. SOC.,1971, 93, 2045; R. H. Marvich and H. Brintzinger, J. Amer. Chem. SOC.,1971,93,2046. sn E. Bayer and V. Schurig, Chem. Ber., 1969, 102, 3378. Nitrogen Fixation bond have been suggested as a mechanism, but dinitrogen has not yet been inserted into any stable transition-metalrarbon bonds. An alternative synthesis of organonitrogen compounds also supports the idea that the reduced titanium systems contain nitride. Thus when bis(7rcyclo- pentadieny1)titanium dichloride is reduced with magnesium metal in tetra- hydrofuran under dinitrogen, and the resultant mixture is allowed to react with ketones, secondary amines are formed upon subsequent hydrolysis :33 N3-f R'RCO ,If,O, RR'CHNH2 Apparently only titanium systems react in this way.Di-n-butyl ketone gives 5-nonyl- and di-(5-nonyl)-amines in about 25 % yield based on the total amount of nitrogen originally fixed. Benzaldehyde yields benzylamine and dibenzylamine. This reaction may be compared with that of lithium nitride with acid anhydrides to give nitriles or with benzoyl chloride to give triben~amide.~~ In these cases, nitride ions are almost certainly involved. Probably some combination of catalytic nitridation with this kind of nitride ion reaction could lead to the large- scale synthesis of organonitrogen compounds direct from molecular nitrogen.By careful control of the working-up, Van Tamelen and his co-workersYo have been able to catch the reduced complexed dinitrogen before the N-N bond has been completely severed. Under these conditions, hydrazine is also a solvo- lysis product. Lithium and barium nitrides are also known to produce some hydrazine upon solvolysis. Recently, from the Shilov have come some related developments concerning some very unstable, presumably bridging- dinitrogen, complexes of both early and late transition metals. The complexes may be related to [(~T-C~H~)~P~T~(N~)T~P~(~T-C~H~)~]discussed in the previous section.Apparently, however, that complex cannot be solvolysed to ammonia or hydrazine.26 The Shilov school has prepared a bright-blue solution by the reaction of [TiC12(7r-C6H6)2] with an excess of isopropylmagnesium chloride in ether at -60°C under dinitrogen. As with the phenyl complex26 this solution does not take up dinitrogen at room temperature but, on cooling, dinitrogen is absorbed and on warming is evolved again. However, if the cold solution con- taining dinitrogen is solvolysed with hydrogen chloride, hydrazine is obtained. Similarly [TiC12Et2] on treatment with ethylmagnesium bromide in ether at -80 "C produces a solution which contains a dinitrogen-trapping species. At -60 "C dinitrogen is released, but treatment with hydrogen chloride at -60"C before the dinitrogen is released produces hydrazine.Interestingly, the later transition metals also appear to undergo this type of reaction. When isopropylmagnesium chloride is added to ferric chloride in the presence of triphenylphosphine in ether, it also produces a solution which does not absorb dinitrogen at room temperature, but at -40°C and atmospheric pressure it does so to produce a red solution. This red solution, on solvolysis at ss E. E. Van Tamelen and H. Rudler, J. Amer. Chem. Soc., 1970, 92, 5253. s4 P.E. Koenig, J. M. Morris, E. J. Blanchard, and P. S. Mason, J. Org. Chem., 1961, 26, 4777. Chatt and Leigh -40 "C with hydrogen chloride, gives up to 10%of its complexed dinitrogen as hydrazine and the remainder as dinitrogen.On the other hand, if the red solution is allowed to warm to room temperature it becomes brown, the complex de- composes and dinitrogen is evolved, +N2per iron atom. Hydrazine is obtained only in the presence of triphenylphosphine. In its absence there is only very slight formation of a dinitrogen complex, but if the reaction is runat 90 "C and at over 20 atm pressure of dinitrogen in absence of triphenylphosphine, nitriding occurs, and yields of ammonia after hydrolysis reach 30% based on FeCl,, assuming one nitrogen atom per iron atom. The addition of triphenylphosphine reduces the yield of ammonia from the above reaction. The reaction of moly- bdenum(v) chloride with isopropylmagnesium chloride in ether at low tem- peratures similarly yields, on hydrolysis, ammonia and hydrazine.The above work has only just appeared and has not had appraisal in other laboratories. It is interesting in showing that easily dissociable, but nonetheless reactive, dinitrogen complexes can be formed at very low temperatures, although they are not formed at room temperature, and that they can be derived from transition metals whether at the beginning or near the end of the transition-metal series. It appears also that the formation of a dinitrogen complex precedes nitridation, but no well-characterized dinitrogen complex has been degraded to nitride, and the exact requirements for such a reaction of complexed dinitrogen are not yet clear.2o 4 Reduction by Aqueous Reducing Agents This has often been attempted, and many strong reducing agents in the presence of derivatives of transition metals such as molybdenum have been reported to produce minute traces of ammonia from dinitrogen. Typical aqueous systems are dihydrogen with platinum black or in the presence of reduced molybdenum compounds, fresh metallic iron rusting in dinitrogen with traces of dioxygen, and acidic solutions of sodium molybdate, sodium borohydride, and thio- organics, with or without ferrous salts.The apparent production of traces of ammonia from dinitrogen must always be checked by reduction of 15N2,and this relatively recent method of testing has been applied to very few systems. Spurious results are likely because the Nessler test for ammonia is not sufficiently specific and, where large volumes of dinitrogen are used, the system may scavenge from the gas phase traces of ammonia, or oxides of nitrogen which are subse- quently reduced.The apparent production of ammonia has even been traced to the rubber tubes used to convey the dinitrogen. Two types of aqueous system are now known to reduce dinitrogen, the one to hydrazine and the other to ammonia. The essential catalyst to produce hydrazine is a reduced molybdenum or vanadium salt in the presence of a substantial proportion of magnesium ions.35 Typically, dinitrogen is reduced by an aqueous or aqueous-alcoholic solution of sodium molybdate or molybdenum(v) tri- 36 A. Shilov, N. Denisov, 0. Efimov, N. Shuvalov, N. Shuvalova, and A. Shilova, Nurure, 1971,231,460.Nitrogen Fixation chloride oxide, mixed with titanium(iI1) chloride, at pH greater than or equal to 10.5. The mole ratio of magnesium to titanium for optimum reduction is 1:2. Hydrazine is produced at room temperature and atmospheric pressure of dinitrogen, but at elevated temperatures and pressures (50-100 "C,50-150 atm) yields of hydrazine as high as 100 mol per molybdenum atom have been obtained, and at the higher temperatures some systems produce ammonia. The reductant titanium(Ir1) can be replaced by vanadium(1x) or chromiwn(n). The system is poisoned by carbon monoxide, and indirect evidence suggests that carbon monoxide forms a 1:1 adduct with the molybdenum, which is in the oxidation state III. The mixture is heterogenous.It has been suggested that the function of the magnesium is to keep the Ti"' ions in the hydroxide gel apart so that their oxidation with dihydrogen evolution is retarded by the formation of Ti-0-Mg-O-Ti species. The following scheme has been proposed: This system parallels the natural system in many respects. However, its efficiency, as measured by the yield of reduced dinitrogen per molybdenum atom, is at best about 1% that of the natural system. Vanadium(n) can take the place of both the molybdenum and the titanium in the above system and at alkaline pH rapidly reduces the dinitrogen to hydrazine. At 10 atm pressure, essentially quantitative reduction based on vanadium according to the following equation is claimed: 4V2++ N2 + 4H2O --+ 4V3++ NaHI + 40H-The rate of reduction is faster than in the molybdenum system.Both systems also reduce acetylene to ethylene (and further to ethane). The molybdenum and vanadium systems apparently contain active metal compounds in the d3configuration. They are thus more deficient in electrons than those systems forming non-reducible stable dinitrogen complexes, in which at least one metal atom has a closed electron shell. They also differ from systems which nitride, in that the latter contain the metal atoms in exceptionally low oxidation states and even as finely divided metal. The second type of system known to reduce dinitrogen is the result of an attempt to produce chemical models on the basis of the scanty knowledge that nitrogenase contains iron, molybdenum, sulphide, and thiol groups.Schrauer Chatt and Leigh and his co-workers3* have reported that organic thiols, sodium molybdate, and ferrous sulphate in the presence of a reducing agent, e.g. Na2S20, or NaBH,, give good reduction of those dinitrogen analogues which are reduced by the natural system. There is a close similarity to the natural system except that reduction is not strongly inhibited by carbon monoxide. By optimizing condi- tions for acetylene reduction, they have produced a system which gives trace amounts of ammonia from dinitrogen at 2000 p.s.i. pressure, e.g., 3-5 pmol of NH, from about 5 mmol Na,MoO, + 2.5 mmol thioglycerol + 0-1 mmol FeSO,,SH,O and 0.25 g NaBH, in 50 ml water.The optimum ratio of moly- bdenum to iron is 2 :1, and no ammonia is obtained in the absence of moly- bdenum. Copper, nickel, or palladium could not be used in place of iron. Working with 15N2, Hill and Richards3' have been able to demonstrate the production of 2-3 pmol of 16NH3 under Schrauzer's conditions using cysteine, molybdate, and ferrous sulphate, but at 1 atm of 15N2. No 15NH3 was obtained when no ferrous sulphate was added. Further, using 2-aminoethanethiol and one hundredth the quantity of catalyst Na,MoO, (60 pmol), FeSO, (1 pmol), NaBH, (0.39 mmol) in water (1.5 ml) they have produced about 90 pmol of 16NH3 from 15N2 at atmospheric pressure. Shilov's, Schrauzer's, and the results from our laboratory are all very new. They demonstrate unequivocally that dinitrogen can be reduced catalytically in an aqueous environment to give substantial yields of hydrazine or ammonia and, with further development, may provide important and useful methods of nitrogen fixation.POSTSCRIPT. One of us has learnt more about the aprotic reduction of dini- trogen in the interesting system developed by the Shilov school, in discussion with Professor A. E. Shilov. The bright-blue solution mentioned above is obtained in a more concentrated state from ~iC1(.rr-C5H5),] than from the dich- loride, because the former is more soluble in ether. A blue solid complex [{Ti(Pri)(n-C,H5),),@I2)]has been isolated from it. Like its phenyl analogue, hydrogen chloride solvolyses it to give dinitrogen and not hydrazine in ether at -100 "C.However, if it is first treated with an ethereal solution of isopropyl- magnesium chloride, and then with hydrogen chloride at -100"C, hydrazine is obtained from it quantitatively (see Chem. Comm,, 1971, 1590). The active complex [Fe2(Pri),(PPh3),(N,)H(Et,0)n] has been isolated as a red solid from the ferric chloride system (see above). This decomposes slowly at 0 "C, and at -40 to -100"Cit is solvolysed by hydrogen chloride in ether to give a 10%yield of hydrazine, the remaining dinitrogen being evolved. Similar solvolysis at 0°C gives only dinitrogen as does solvolysis by cold methanol. Pretreatment with Grignard reagent is not necessary for the production of hydrazine from this complex (see Chem. Comm.,1971, 1185).By treating ferric chloride in very cold ether with phenyl-lithium (10-20 mol) G. N. Schrauzer, G. Schlesinger, and P. H. Doemeny, J. Amer. Chem. Soc., 1971,93,1803,and references therein. 37 R. E. E. Hill and R. L. Richards, Nature, 1971, 233, 114. Nitrogen Fixation under dinitrogen, a red complex is formed in solution, which by cold solvolysis with hydrogen chloride yields about 56 76 of hydrazine based on iron, assuming two iron atoms per molecule of bound dinitrogen. Ammonia and dinitrogen are produced together with the hydrazine. This work will be published in Kinetika i Katalk Fleischer and Krishnaniurthy (personal communication) report that sodium rneso-tetra(p-sulphonatophenyl)porphinatocobalt(III) in aqueous solution yields 06-0.8 mol NH3 by passage of air through its aqueous solution and with one addition of sodium hydroborate every 24 h; we have confirmed this.They claim that most of the ammonia is derived from the dinitrogen of the air, and some from the porphin. Hill and Richards (personal communication) report that their 2-aminoethane- thiol system depends for its action on some unknown factor, and that pure reagents give only marginal fixation, if any. When the fixation of 15N2 occurs on an appreciable scale, one third to one half as much 14NH3 is formed as 15NH3. It appears that some catalytically initiated condensation with elimination of ammonia from 2-aminoethanethiol is necessary to produce the ligand which forms the nitrogen-fixing catalyst complex, but the conditions for its formation are obscure. Sellmann (Angew. Chern. Internat. Edn., 1971, 10,919) has reported a dinitro- gen manganese complex, [(CO), (.rr-C,H,)(N,)Mn].