Ferrocene—how it all began

Journal of Organometallic Chemistry 637–639 (2001) 3–6www.elsevier.com/locate/jorganchemFerrocene—how it all beganPeter L. PausonDepartment of Pure and Applied Chemistry, Uniersity of Strathclyde, Thomas Graham Building, Cathedral Street, Glasgow G1 1XL, UKtropolones truly fit into this category has remainedcontroversial, they were certainly topical and of muchinterest in this context at the time. My obvious interestsfor my first independent researches were therefore directedtowards further work on either tropolones orother non-benzenoid systems.(Peter L. Pauson)With the rest of my family, I had come to the UK in1939 as a refugee from Nazi persecution of the Jews.After completing my secondary education, I enteredGlasgow University in 1942 to study Chemistry. Myinclination towards the organic side was greatly enhancedand my approach to research was moulded bythe inspiring teachings of T.S. Stevens. 1 The offer of aresearch studentship led me to Sheffield University in1946 to do my PhD under the direction of ProfessorR.D. Haworth. 2 My task was to study the chemistry of‘purpurogallin’—the brilliant red compound obtainedby oxidation of pyrogallol. We succeeded in establishingits structure as the benzotropolone (1) and showedthat further oxidation followed by decarboxylation ledto -methyltropolone (2), the first synthetic and fullycharacterised simple tropolone.The concept of ‘non-benzenoid aromatics’ was stillrather new and, although the question as to whether1 Discoverer of the Stevens Rearrangement, the McFadyen–Stevens aldehyde synthesis, and the Bamford–Stevens reaction.2Best remembered for the Haworth phenanthrene synthesis.On completion of my degree, I had moved toDuquesne University 3 in 1949. There, I read with interest,but also scepticism, R.D. Brown’s paper (Nature,1950), which suggested that fulvalene (3) might showaromatic properties. The challenge to either prove or3 This Catholic university in Pittsburgh was better known for itsbasketball team than for its academic achievements. Its chemistrystudents were mostly pre-meds, but it offered BS and MS degrees inchemistry although not yet PhD. I had applied in response to anadvertisement in Chemical and Engineering News and, after their firstselected candidate withdrew at short notice, I was offered the post ofAssistant Professor without interview.0022-328X/01/$ - see front matter © 2001 Published by Elsevier Science B.V.PII: S0022-328X(01)01126-3

4P.L. Pauson / Journal of Organometallic Chemistry 637–639 (2001) 3–6disprove his suggestion was heightened by my belief thatthe compound might be accessible in just two steps: thecoupling of two molecules of cyclopentadienylmagnesiumbromide to give the dihydro compound 4, followedby dehydrogenation. This, therefore, was the project Ioffered to Tom Kealy when he expressed an interest, in1951, in working with me for his MS degree.The reductive coupling of Grignard reagents RMgXby transition metal halides to give hydrocarbons RR isa general reaction that is known to be particularly easywhen R is an allylic group. Although other halides, e.g.CoCl 2 , were regarded as better, we chose FeCl 3 for tworeasons: (i) It is a common and ether-soluble reagent inits anhydrous form, whereas all the other candidateswould have had to be made from hydrates; and (ii) iffulvalene were truly aromatic, the oxidising power ofFe(III) might suffice to effect the dehydrogenation stepand thus give us fulvalene in a one-pot process.(T.J. Kealy)Tom and I did the crucial experiment together one dayin July 1951: making EtMgBr in ether, at the same timecracking the dimer of and redistilling cyclopentadiene,which was then added to the Grignard solution todisplace ethane and generate cyclopentadienylmagnesiumbromide, finally adding the FeCl 3 . Standardworkup of this mixture by pouring onto ice and ammoniumchloride produced a yellow ether layer—not theyellow aqueous layer expected if the iron(III) had remainedunreduced. Moreover, a few drops of the yellowsolution that had escaped onto the outside of theseparating funnel evaporated, leaving yellow crystals.Could we have made a yellow hydrocarbon, perhapseven the desired fulvalene? It seemed too good to be true.We soon had a batch of purified crystals and, notwishing to imply that the results should add up to 100%,these were sent for C,H microanalysis as containing C,H, and O. When the results came back a few days later,the large deficit from 100% clearly required an elementheavier than oxygen and it needed little arithmetic tofind that these results accurately fitted C 10 H 10 Fe. Wenow had to analyse for iron, both qualitatively andquantitatively. The compound dissolved in strong sulphuricacid but was largely recovered unchanged ondilution. We had to boil with strong nitric acid before wecould get the standard qualitative tests for iron to work,and even this was insufficient to give accurate quantitativeresults. These were only obtained after fuming todryness with HClO 4 .What was this remarkable substance we had made?We were well aware of the common belief that bondsbetween transition metals and hydrocarbon groupswould always be unstable—based on the failure of allrecorded attempts to form such links. Yet, here we hada compound, formed from two C 5 H 5 units and an ironatom, which was not only isolable but also stable to hightemperatures and unaffected by water and by bothstrong acids and alkalis. True, we knew of the exceptionsto the alleged non-existence of organo-transition metalcompounds: Zeise (1827) had made K[PtCl 3 (C 2 H 4 )] andsome methylplatinum compounds had been described—but platinum was seen as ‘atypical’; Hein (1919) haddescribed ‘polyphenylchromium salts’ and Reihlen(1930) had reacted iron pentacarbonyl with butadiene togive C 4 H 6 Fe(CO) 3 . These substances tended to be ignoredsimply because their structures were unknown.Whatever our product might be, it was clearly somethingvery novel and unexpected, and we thereforereported our findings in a note to Nature—despatchedwithin a week of getting that C,H analysis. We wrote thestructure as 5.I attended the large IUPAC Congress in New York acouple of months later and there handed a sample of ourcrystals to J.M. Robertson, the distinguished X-raycrystallographer who was Professor of Physical Chemistryat Glasgow University.Soon several people independently guessed the correctstructure. The first of these was W. von E. Doering,whom I visited at Columbia University immediatelyafter the IUPAC Congress and told about our finding.Unfortunately, I failed to understand his suggestion,which included the possibility that magnetic susceptibilitymight provide evidence; I was too reluctant to exposemy ignorance and my lack of understanding and wascontent to await the definitive answer expected fromX-ray work. In the meantime, through circumstancesbeyond his control, 4 no results were forthcoming fromRobertson before several others undertook X-ray structuralwork after our note appeared in Nature. Ironically,one of these was Jack Dunitz, next to whom I was sittingat the IUPAC meeting while listening to Robertson’stalk and waiting to give him the sample.4 Robertson was on his way to Cornell University as that year’s‘Baker Lecturer’ and preoccupied with writing the book which wasexpected of holders of this visiting appointment.

P.L. Pauson / Journal of Organometallic Chemistry 637–639 (2001) 3–6 5In those days, the Chemical Society published lists ofpapers accepted for publication in its journal. About amonth before our Nature note was due, I read of apaper on ‘Dicyclopentadienyliron’ by Miller, Tebbothand Tremaine of the British Oxygen Co. I wrote to Dr.Miller and told him that we appeared to have the samecompound, and so it proved when he sent me a proofcopy of the J. Chem. Soc. paper. It also showed thatthey had used a radically different route and that theirpaper had been submitted before we ever tried thereaction, but it was to appear in the February 1952issue of J. Chem. Soc., whereas Nature’s faster methodof publication gave us the apparent precedence with a1951 publication date. In an accompanying note Millermentioned that they had had the compound for about3 years. Even this, however, was quite possibly not thefirst preparation. The following tale was told about ayear later by E.O. Brimm of Linde Air Products.Brimm, who had done some work on metal carbonylsand other organometallics, was one of several industrialchemists who, on reading our report, wanted to makesome of the compound. Therefore he enquired of acolleague at Union Carbide (Linde’s parent company)whether they had any cyclopentadiene. The reply thatthey no longer did so was accompanied by the statementthat, some years previously, they had terminatedwork on the cracking of cyclopentadiene because ayellow sludge clogged the iron pipes they used. Theyhad not attempted to isolate or analyse this but hadkept a bottle of it; it was ferrocene. The British Oxygengroup’s synthesis was, in fact, not dissimilar as it involvedpassing cyclopentadiene over a heated, iron-containingammonia synthesis catalyst.As is well-known, the correct structure of dicyclopentadienylironwas put forward in two independent publications:one by E.O. Fischer and W. Pfab, the other byG. Wilkinson, M. Rosenblum, M.C. Whiting, and R.B.Woodward. Both groups had repeated our preparationand made physical measurements that strongly supportedthe ‘Doppelkegel’ or ‘sandwich’ structure butfell short of full proof. Fischer relied chiefly on preliminaryX-ray data indicating that the molecules arecentrosymmetric, while the Harvard group cited thesingle C–H frequency in the IR and the diamagnetism.Only weeks went by before two independent completeX-ray structure determinations were published. 5At Woodward’s suggestion that the compound mightbe aromatic, Whiting and Rosenblum went on to showthat it readily undergoes Friedel–Crafts acylation and,on this basis, suggested the name ‘ferrocene’—with the‘ene’ ending implying aromaticity. We had indeed prepareda novel, non-benzenoid aromatic, but a veryunexpected addition to that class!5By Dunitz and Orgel and by Eiland and Pepinsky.Both Fischer and Wilkinson very soon started addingbis-cyclopentadienyls of other transition metals; TomKealy had tried nickel and cobalt but was thwarted bythe insolubility of their anhydrous halides, which preventedtheir reaction with the ethereal solution of cyclopentadienylmagnesiumbromide. Fischer overcamethis problem by turning to the hexammine complexes ofthese metals and Wilkinson by using the acetylacetonates.In the meantime I was spending the academic year1951–1952 at the University of Chicago as a postdoctoralfellow working on peroxide chemistry with M.S.Kharasch. My next move was to Harvard where I wasappointed to a ‘Du Pont Fellowship’ to work independently.Although my application for this had beenbased on a scheme to synthesise the alkaloid colchicine(a tropolone), Professor Bartlett, as Chairman of thedepartment, made clear that I was free to do whateverI liked. It cannot have been unexpected that by then Iwas anxious to play a part in the work on metalloceneswhich was in full swing in both Woodward’s andWilkinson’s laboratories.En route to Harvard, I visited the Du Pont laboratoriesat Wilmington. There, Dr Victor Weinmayr wasdoing work on the organic chemistry of ferrocene. Ontalking about other metals, he told me that nickelworked and that I should feel free to use this informationprovided that I did not disclose its source.Combining this hint with what I knew already, I thereforeemployed nickel acetyl acetonate in my first experimentsat Harvard and, by the time Wilkinson andCotton both returned from their summer vacations, Iwas able to show them a sample of the beautifulemerald-green crystals of nickelocene. They showedthat it is paramagnetic and we extended our jointstudies to the benzo derivatives of ferrocene and of thecobaltocenium cation, which I prepared from indene bysimilar techniques, leaving them to do all the physicalmeasurements.During this and the following 2 years, Fischer andWilkinson made bis-cyclopentadienyl–metal and cyclopentadienylmetalcarbonylcompounds of most of thetransition metals. It seemed almost inevitable that theywere in constant competition as to who would be firstwith the next, fairly obvious target. 6At the end of my year at Harvard, I returned to theUK to take up a lectureship in organic chemistry atSheffield University. There, I soon had the benefit ofPhD students to do all the hard work for me. Weconcentrated on the organic aspects. The first student(G.D. Broadhead), after showing that ferrocene undergoesacetylation very much faster than anisole, foundthat it would undergo formylation under Vilsmeyer6 This competition effectively ceased in 1955, when Fischer’s success(with W. Hafner) in making bis-benzenechromium led him toconcentrate heavily on arene complexes.

6P.L. Pauson / Journal of Organometallic Chemistry 637–639 (2001) 3–6conditions (as also shown independently by Rosenblum)and then tried azo-coupling, thereby finding theunexpected arylation. The second student (B.F. Hallam),after some work on cyclopentadiene–iron carbonylcomplexes, took on the task of verifying mybelief that Reihlen’s above-mentioned butadienetricarbonylironis also a -complex. By repeating its preparationand also making the cyclohexadiene complex, heopened up the whole field of olefin, diene, and trienecomplexes—but that is the beginning of a larger storybeyond the scope of the present topic.It was undoubtedly the availability of good methodsof structure determination coupled with the greatertheoretical understanding which made possible the explosivegrowth of interest in organo-transition metalchemistry following directly from the discovery of ferrocene,whereas the earlier findings had led nowhere.Personally, I feel extremely lucky to have been involvedfrom the start and I owe a huge debt of gratitude to asuccession of co-workers who made it possible for meto participate with great enjoyment in these developmentsthroughout these 50 years.