Atoms and Small Molecules as Useful Chemical Reagents BY P. L. TIMMS School of Chemistry University of Bristol Bristol BS8 1TS Received 31st August 1973 The paper surveys some recent work on reactions of high temperature species with ordinary compounds on a liquid-nitrogen-cooled surface. Useful syntheses with metal vapours metal salt vapours and low-valency boron and silicon compounds are described. Some reactions which failed are also reported and the reasons for unpredictability in the low-temperature reactions are discussed. Over the last few years an increasing number of atomic and molecular species produced at high temperatures have been used successfully in chemical synthesis. 1-3 Potentially useful species include atoms of the metals and metalloids vapours of metal salts and sub-halides and sub-oxides of group 3 and 4 elements.These species may be reactive because they contain elements which are in unstable low oxidation states or which are co-ordinately unsaturated. The reactions are carried out in two main ways. On a preparative scale (> 100mg of the species) the species is formed under a high vacuum and condensed immediately on a surface cooled by liquid nitrogen at the same time as the vapour of another compound. The cold surface acts as a cryogenic pump for the vapours. As there are few collisions in the gas phase at the low pressure thermal decomposition of the added compound on the furnace or by hot vapour is avoided. The products are normally isolated on warming the co-condensate or if they are very unstable they may be reacted with other compounds added at a low temperature.On a very small scale high-temperature species and compounds are co-condensed in the presence of a huge excess of a noble gas at 4-20 K. The species are initially isolated in the inert gas matrix but slight warming permits diffusion and if reactions then occur they are followed spectroscopically. In this paper recent results obtained at Bristol are used to show the potentialities and limitations of the preparative scale method EXPERIMENTAL FORMATION OF HIGH TEMPERATURE SPECIES Methods of vaporizing metals salts and related compounds are well known because of the interest in making thin films of materials by vacuum evaporation. We prepare high temperature species by vaporization of condensed phases by gas-solid reactions and by decomposition of gases at high temperatures.(a) VAPORIZATION TECHNIQUES For species which are formed below 18Oo0C,we prefer to contain the evaporant in an inert crucible (usually alumina) heated resistivity by 18 s.w.g. Mo or W wire. The gauge of wire requires less than 50 A to reach the evaporation temperature so that heavy electrical leads are unnecessary. The crucibles can be efficiently thermally insulated so that radiative heat losses to the cold walls of the vacuum chamber are small. A crucible with a capacity of 1 ml will vaporize 50-100 mmol of most metals. 68 P. L. TIMMS 69 For the more refractory 2nd and 3rd row transition metals we have used electron bom- bardment vaporization but we find that secondary electrons often cause degradation of the compounds condensed with the metal vapours.(b) GAS-PHASE REACTIONS The preparation of the boron monohalides or the silicon dihalides by gas+solid reactions e.g.9 1300°,1 Torr Si(s)+ SiCI4(g) + 2SiCI2 is conveniently conducted in an inductively-heated graphite tube. The sub-halide emerges from the tube into a high vacuum and is immediately condensed at -196°C. Boron monochloride is prepared by flash thermolysis of diboron tetrachloride lOOO" 3-5 Torr B2CUQ) + BCI+BCIB. A short residence time for the vapour in a narrow quartz tube furnace has proved the best condition to avoid the thermodynamically favoured nucleation of solid boron. THE REACTION VESSEL Fig.1 shows a reaction vessel set up for studying co-condensation reactions of metal vapours. The glass vacuum chamber is pumped by a mercury diffusion pump with a speed at the chamber of about 15 1. sec-l. This pumping speed is more than adequate during Reectant inlets \ Water and eI ec tr i c i ty" Solution ATOMS AND SMALL MOLECULES co-depositionof a metal vapour and a compound which has a low vapour pressure at -196"C as the liquid-nitrogen-cooled walls of the vessel act as a cryogenic pump. If traces of per-manent gas are evolved during the deposition or if the compound added has an appreciable vapour pressure at -196"C a higher pumping speed is required to maintain the pressure below Torr ; above this pressure we find that thermal decomposition of the added compound usually becomes appreciable and this may liberate more permanent gas.We use stainless steel apparatuses with much higher pumping speeds when handling very volatile compounds. The apparatus of fig. 1 can be used to vaporize as much as 50 g of suitable metals Iike copper and to condense the vapours with compounds added through the central tube. This tube can be heated to vaporize slightly volatile compounds. The reaction products are recovered either by warming the vessel to room temperature with continuous vacuum pumping into an attached liquid-nitrogen-cooled trap or by adding solvents to the apparatus and sucking out the solution for work-up by standard techniques. The recovery of products is made simpler if a much smaller apparatus is used the vacuum vessel being a one-litre flask.After the condensation the flask and its contents can be handled like a piece of conventional apparatus. This form of metal evaporation system is the basis of an undergraduate teaching experiment at Bristol.* Furnaces which make high temperature species by gas+solid reactions or by flash thermolysis are mounted in vacuum systems similar to those used for metal evaporation. The identification of the products of the reactions has generally depended on a com- bination of infra-red n.m.r. and mass-spectrometric data with microanalysis. RESULTS SOME REACTIONS OF METAL VAPOURS The majority of our work has been with the following metals Cr Mn Fe Co Ni Cu Mo Pd Ag Sn Au (a) WITH BENZENE DERIVATIVES Chromium vapour reacts readily with benzene and its derivatives to give arene- chromium complexes.This route avoids the complication of solvents and the presence of Lewis acids inherent in conventional methods of making these complexes. Thus cumene and other alkylbenzenes give pure di-arene chromium complexes with chromium vapour with no isomerization of the arene~.~ We have prepared the first complexes in which hexafluorobenzene behaves as a six-electron donor; (C6F6)Cr(PF,) and (C$&r(C&) were made in low yield by condensing chromium vapour with the respective ligand mixtures. No (C6F6)2Cr could be isolated the co- condensate exploded. Skell et aL6 have reported the synthesis ofchromium complexes of chlorobenzene fluorobenzene and difluorobenzene; we find that the yield of (C6HSC1)@ is higher than witb any other arene 60-80 % even in the hands of under- graduates.We have also isolated small yields of (.n-pyridine)Cr(PF& and (n-hexamethyl- borazine)Cr(PF,) from chromium vapour and the mixed ligands ; reactions of pyridine with chromium by conventional routes yield only complexes with pyridine co-ordinated through the nitrogen atom. No "20-electron " diarene-iron complexes have yet been isolated from the reac- tions of arenes with iron vapour. Benzene and iron form an explosive product perhaps (C6H6)2Fe but mesitylene and iron give (C9Hl2)Fe(CgH,,) in moderate yield. Manganese vapour gives a low yield ofthe known compound (C6H6)Mn(C5H,) when condensed with benzene and cyclopentadiene.P. L. TIMMS (b) WITH OTHER UNSATURATED ORGANIC COMPOUNDS The reactions of metal atoms with cycloheptatriene are varied. Chromium gives good yields of (C7H8)2Cr ti or in the presence of trifluorophosphine (C,H,)cr(PF,),. Manganese cobalt and nickel vapours gave no simple product with cycloheptatriene but iron gave heptafulvalene (C7H6)2 and no organo-iron complex. Admixture of trifluorophosphine with cycloheptatriene still gave no products with manganese and nickel iron formed heptafulvalene and cobalt gave a good yield of (C7H7)Co(PF3) ; HCo(PF,) was also formed. This hydrogen abstraction in the presence of tri- fluorophosphine is also observed in the reaction of cobalt vapour propene and trifluorophosphine which gives (n-aUyl)Co(PF,),.The reaction of nickel vapour with tetra-ally1 tin causes a simple transfer of groups and bis(n-allyl) nickel is formed in good yield. (C) WITH PHOSPHINES The usefulness of trifluorophosphine as a ligand in metal vapour reactions is clear from the compounds described above. Although its reactions with most transition metals are simple iron causes defluorination and the products include Fe2(PF2)2(PF3)6 and Fe3(PF)2(PF3)9 in ad-dition to Fe(PF3)5. Trimethylphosphine complexes of iron cobalt and nickel are formed on con- densation but phosphine PH3 has proved more difficult. No Ni(PH3) was iso- lated from nickel vapour and phosphine. Much hydrogen was evolved on warming from -196°C; in the presence of trifluorophosphine Ni(PF3)3PH3 and Ni(PF3)2(PH3)2 were obtained.(d) WITH NO N2 AND CO Nitric oxide is a difficult ligand to use in co-condensation reactions because it has a vapour pressure of 0.1 Torr at -196°C and it is an oxidizing agent. On several occasions metal-nitric oxide co-condensates exploded on warming fi om -196°C. We have made the series of compounds Co(NO)(PF,), Fe(NO),(PF,), and Mn(NO),PF, hut we could not obtain Cr(N0)4,5 a compound which has been made photochemically. Nitrogen and carbon monoxide are effectively non-condensible at -196°C. Appreciable gettering of nitrogen was observed when nickel was evaporated under a pressure of 1 Torr of N2. However the condensate did not react with cyclo- octadiene suggesting that no discrete nickel-dinitrogen complex was present.Evaporation of palladium under a pressure of 1 Ton of carbon monoxide followed by addition of trifluorophosphine to the cold surface gave Pd(PF3) ; this indicates that a palladium carbonyl perhaps Pd(CO) as reported by Ogden,I was present at -196°C. (e) REACTION WITH HALOGEN COMPOUNDS We have studied reactions of copper silver and gold atoms with a variety of inorganic and organic halogen compounds. The atoms show selectivity and stereo- specificity. Thus copper atoms efficiently couple the boron atoms in boron trichloride or dichloroalkylboranes making this an excellent route to diboron compounds.* Phosphorus-phosphorus bonds can also be formed from phosphorus chlorine com- pounds and copper vapour but silicon chlorine compounds do not react.Con-densation of copper or silver vapour with R(-)sec-butyl chloride gave s,$(-)3,4- ATOMS AND SMALL MOLECULES dimethylhexane with preservation of 70 % of the optical activity and apparent inversion of configuration. Under the same co-condensation conditions sodium vapour gave an optically inactive 3,4-dimethylhexane. Gold vapour causes coupling of the alkyl groups from alkyl bromides but it does not react with alkyl chlorides. Tin vapour has proved to be rather unreactive towards alkyl halides despite the known stability of the alkyl tin halides. No reaction occurs at -196°C with alkyl chlorides and bromides although alkyl iodides give polymeric (RSnI),. Other research groups are actively studying metal atom dehalogenations and reports have recently appeared on lithi~m,~ zinc,l0 calcium,l and palladium l2 vapour reactions.The work of Lagow on Lithium stands apart for his reactions occur in the gas phase; thus when the flame formed by burning CC14 vapour in a stream of lithium atoms is quenched at -196"C the products behave chemically like Li,C the lithium atoms being readily displaced by other groups. METAL SALT VAPOURS AS REACTANTS Our results in this area are preliminary but it is clear that metal salt vapours are not ideal reactants at low temperatures. We condensed NiC1 and CoC1 vapours with a range of alkenes and dienes and observed polymerization of the organic compounds to high polymers reactions which we thought were catalyzed by the transition metal ions. However the same results were obtained with CaC1 vapour and the polymerization catalyst was found to be the proton liberated in the reaction co-condense HCl(g)+MC12(g) -+ H +MC1,.-196' The hydrogen chloride was formed in trace amounts on heating the " anhydrous " dihalides. Potassium cyanide vapour reacts on condensation with nickel or iron carbonyls to give cyano-carbonyls e.g. co-condense Fe(CO) +KCN(g) -+ K+[Fe(CO),CN]-. -196°C Stannous chloride vapour condensed with alkyl halides does not form alkyl tin halides. A slow reaction is known to occur between stannous halides and alkyl halides above 100°C. SOME REACTIONS OF BORON AND SILICON SPECIES The low-temperature reactions of the boron and silicon species have been studied in more detail than reactions of most of the metal atoms.The compounds provide unique routes to the higher boron and silicon fluorides silicon-boron halides and 1,4-dibora- and 1,4-disila-cyclohexanes and cyclohexadienes. We find the 1,4-diboracyclohexadienesto be powerful ligands to transition metals e.g. \/ \/ co-condense Ni(co)4 BF(g)+CH3C=CCH3 -+ FBoBF -+ FB 1 BF -196'C 0 /-\ /ll\ Ni /\ co co There are surprising differences in the low-temperature behaviour of related species e.g. SiF, SiC12 and SiO. On a cold surface SiFz appears to react mainly P. L. TIMMS in the form of a diradical SiF2SiF2 whereas SiC12 reacts as a monomer. Yields of organo-silicon products from reactions of SiClz are generally higher than those from reactions of SiF2.probably because the reaction mechanism is simpler for SiC12. Silicon monoxide 1s unique in being able to insert into C-H bonds as well as add across C-C multiple bonds. We have studied the chemistry of BC1 only since being able to make its precursor B2C14 in gram quantities from copper vapour and boron trichloride. In its reactions with organic compounds BCl closely resembles Sic],. We have found gaseous boron monoxide O=B-B-0 to react with boron trichloride or with sulphur tetrafluoride at -196°C to give high yields of B2C14 and B2F4 respectively. DISCUSSION Two main points emerge from the work at Bristol and from the work of other groups in this field. First the low-temperature co-condensation method is very effective and enables many compounds which were previously unknown or inaccess- ible to be made in useful quantities.There seems an assured future among synthetic chemists for this interface between " high-temperature chemistry " and " conventional chemistry". Secondly there is much to be understood about the interaction of high-temperature species and compounds on the cold surfzce. Much of the unpredictability of the low-temperature reactions stems from the rapidity with which the) must occur. The rate of recombination of metal atoms to give massive metal at -196°C is probably diffusion controlled. Reaction of the atoms with another compound is only competitive if the activation energy for the reaction is very low and the compound is present in excess. An activation energy of 1 kJ mol-1 can make a reaction quite slow at -196°C.Yields of products based on the metal do improve with an increasing ratio of compound metal. When atoms and molecules react sufficiently quickly the choice between reaction pathways can be governed by subtle kinetic factors rather than thermodynamic ones. Thus it is found that chromium and chlorobenzene give very high yields of (C6H5C1),Cr and not the thermodynanGcally more favoured dechlorination to diphenyl and chromous chloride ; nickel vapour causes efficient dechlorination. There is at present no reliable physical method for following the reactions that occur on co-condensation at -196°C. Studies with clean surfaces have little relevance to a process in which the rate of increase of film thickness may exceed 1000&.Matrix isola.tion spectroscopy can give information about the primary interactions between high-temperature species and compounds but its findings are in a sense artificial. For example Ozin j3 has reported the formation of Ni(N2)4 in an inert gas matrix below 30 K from nickel atoms and N2. We failed to make this compound by condensing nickel vapour in the presence of nitrogen gas on a surface at -196°C. The difference between these results may depend not as much on the temperature in the two experiments as on the isolated state of Ni(N2)4 molecules in the matrix. Decomposition of these molecules in the matrix can only give isolated metal atoms and N2 molecules; Ni(N2)4 will be more stable than the isolated components by approximately [4B(Ni-N,] kJ mol-l.On a surface at -196"C Ni(N2)4 can decompose to solid nickel and N2 ; it will be more stable than these components by approximately [4D(Ni-N,) -AH&,(Ni)] kJ mol-'. As the Ni-N2 bond energy is low Ni(N2)4 decomposes when not matrix isolated. The decomposition will be catalyzed by finely divided nickel on the surface; this catalytic effect probably ATOMS AND SMALL MOLECULES accounts for our failure to make the known compounds Ni(PH& and Cr(NO) which have some kinetic stability in the absence of catalysts. The research at Bristol has been generously supported by grants from the Science Research Council. P. L. Timms Adv. Inorg. Chem. Radiochem. 1972 14 121. J. J. Havel M. J. McGlinchey and P. S. SkeI1 Accts Chem.Res. 1973 6 97. P. L. Timms Accts Chem. Res. 1973 6 118. P. L. Timms J. Chem. Educ. 1972 49,782. R. Middleton J. R.Hull S. R. Simpson C. H. Tomlinson and P. L. Timms J.C.S. Dalton Trans. 1973 120. P. S. Skell D. L. Williams-Smith and M. J. McGlinchey J. Amer. Chem. Soc. 1973 95,3337. ’J. H. Darling and J. S. Ogden J.C.S. Dalton 1973 1079. * P. L. Timms J.C.S. DaZton Trans. 1972 830. R. J. Lagow J.C.S. Chem. Comm. 1972 1078. lo K. J. Klabunde M. Scott Key and J. Y. F. Low J. Amer. Chem. SOC.,1972,94,999. l1 K. J. Klabunde J. Y. F. Low and M. Scott Key J. Fluorine Chem. 1972 2,207. l2 K. J. Klabunde and J. Y. F. Low J. Organomet. Chem. 1973,51 C33. l3 H. Huber E. P. Kundy M. Moskovits and G. A. Ozin J. Amer. Chem. SOC.,1973,95,332.