摘要:

Quarterly Reviews No 3 Vol22 1968 Valence-shell Expansion in Sulphur Heterocycles By W. 0. Salmond The Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance Techniques By J. Burgess and M. C. R. Symons The Mass Spectra of Amino-acid and Peptide Derivatives By J. H. Jones Page 253 276 302 Mass Spectra of Organometallic Compounds By D. B. Chambers F. Glockling and J. R. C. Light 317 The Application of the Woodward-Hoflinann Orbital Symmetry Rules to Concerted Organic Reactions By G. B. Gill 338 Molecular Rearrangements Related to the Claisen Rearrangement By A. Jefferson and F. Scheinmann 391 The Chemical Society London Quarterly Reviews contains articles by recognised authorities on selected topics from general physical inorganic and organic chemistry. The Journal and Annual Reports interest primarily the research worker Quarterly Reviews is designed for a wider range of readers. It is intended that each review article shall be of interest to chemists generally and not only to workers in the particular field being reviewed. The submission of reviews for publication is welcomed but intending authors are advised to write in the first place to the Editor The Chemical Society Burlington House Piccadilly London W 1V OBN. Such pre- liminary communications should be accompanied by an outline of the ground to be covered (about two quarto pages) rather than by the completed manuscript. Price to non-fellows E4 10s. Od. per annum @ Copyright reserved by The Chemical Society 1968 Published by The Chemical Society Burlington House London. Printed in England by The Thanet Press Margate

ISSN:0009-2681    

DOI:10.1039/QR96822FP005

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

Valence-shell Expansion in Sulphur Heterocycles By W. G. Salmond UNIVERSITY CHEMICAL LABORATORY LENSFIELD ROAD CAMBRIDGE Valence-shell expansion in the sulphur atom though implicit in earlier studies of thiazole derivatives,l was first postulated explicitly in connection with thiophen2 Much of the subsequent evidence for this phenomenon has come however from work on open-chain compounds. A review3 of this area covers most of the literature to the end of 1959. The purpose of this Review is to survey the evidence accruing from studies of the heterocyclic systems particularly aromatic systems in which the d orbitals of the sulphur atom have been thought to be important in the bonding. It is not intended to present an exhaustive bibliography but to trace the developments which have proved most fruitful.Work on acyclic systems has been ignored except where it has been of particular relevance in a generalised treatment of the bonding of sulphur. Inorganic cyclic compounds such as thiazyl halides and the sulphur nitrides will not be discussed. 1 The Bonding Possibilities of Sulphur Although recent publicationsS have considered in some detail the bonding possibilities of sulphur it will still be advantageous to present a brief outline here. Sulphur belongs to the second short period and to the sub-group VIB of the Periodic Table. Its valence electrons are contained in the M shell and in its ground state the sulphur atom has the electronic configuration If following the lead of Craig and his sch001,~ we assume that the angular functions of s p and d orbitals are the same as in the free atom that is bond formation leaves these functions virtually unchanged then a preliminary discussion of the atomic orbitals associated with sulphur will lead naturally to a discussion of the bonding in molecules containing sulphur.The K and L shells of the sulphur atom are completely filled and the electrons contained therein are assumed to play no part in the bonding. The atomic 3s=3pz23py13p; . H. Erlenmeyer and H. M. Weber Helv. Chim. Ada 1938 21 863; H. Erlenmeyer H. M. Weber and P. Wiessmar ibid. p. 1017. V. Schomaker and L. Pauling J. Amer. Chem. SOC. 1939 61 1769. G. Cilento Chem. Rev. 1960 60 147. J. Nys and A. van Dormael Ind. chim. belge 1961 1109; A. B. Burg in ‘Organic Sulphur Compounds’ ed. N. Kharasch vol. I pp. 3040 Pergamon Press New York 1961 ; C. C.Price and S. Oae ‘Sulphur Bonding’ Ronald Press New York 1962. D. P. Craig A. Maccoll R. S. Nyholm L. E. Orgel and L. E. Sutton J Chern. Soc. 1954 332. 253 Valence-shell Expansion in Sulphur Heterocycles orbital of lowest energy in the valence shell is the 3s. This orbital which is spherically symmetrical and is completely filled is also normally assumed to play no part in bonding except of course where! sp or spd hybrid orbitals are involved. The orbitals of next higher energy are the three mutually perpendicular and orthogonal 3p orbitals 3pz 3py and 3p,. These orbitals of equal energy are those employed most commonly in organic sulphur compounds since they can be used with neighbouring ligands in the formation of a bonds (Figure la) or 7r bonds (Figure lb). In this Review we are concerned with sulphur bound to first-row elements (carbon nitrogen oxygen) so that the 7r bond formed will be of the 2w-3p7.r type (Figure lc).Owing to inner-shell repulsions this type of bond can be expected to be weaker than a bondof the2pn-2pn type which is of course the double bond normally met in organic chemistry. Of even higher energy are the five 3d orbitals. These orbitals are mutually orthogonal and in the isolated atom are degenerate. This degeneracy is lost in a ligand field however. In an octahedral field for instance those of highest energy are the 3 4 and the 3dXa+. The remaining three orbitals are of equal energy and are of the correct symmetry to be able to participate in 7r-bonding. They are the 3dz, 3d,, and 3 4 orbitals. Av bond arises if one of these orbitals is appropriately oriented with its positive and negative lobes equally inclined to the line of nuclear centres and requires am (or dn) orbital at the other centre (Figure Id).T h i s p - d r bonding is of particular relevance here. (a) 2p-2p u bond (c) 2p-3p Q bond Figure 1 (b) 2p-2p ?I bond (d) 2p-3d ?I bond In an important paper Craig and his co-workers5 have discussed theoretically the conditions necessary for the formation of chemical bonds involving d orbitals. They calculated overlap integrals for a series of different types of bond (for 254 Salmond example bonds involving different spd hybrid orbitals) and used these integrals as criteria of bond strength. Calculations based on the Slater ruless indicate that the d orbitals associated with the sulphur atom would be far too diffuse for bonding purposes and Craig points out that for dealing with the valence states of atoms use of Slater's rules may be inadequate.Certainly comparison with S.C.F. (self-consistent field) functions in some cases shows that the Slater d-orbital functions are too diffuse; but the same is found to hold for s andp functions when the same type of comparison is made. It was concluded therefore that results obtained by use of Slater's rules would still be expected to be at least qualitatively correct in unsophisticated comparative studies. Consequently it was assumed that when employed for bonding purposes d orbitals contract as a result of perturbation by the ligands. This idea was developed in later papers' and applied to compounds such as sulphur hexafluoride where o bonds are involved and whose structure is usually interpreted in terms of the configuration 3s3p33d2 for the sulphur atom.One of the ways in which this contraction can be brought about is by the development of a formal positive charge on the central atom by partial ionisation of the outer d electrons into the ligand orbitals. This positive charge causes the more polarisable 3d orbitals to contract and thus increases their ability to hybridise effectively with the 3s and 3p orbitals. This approach has been criticised by some workers* who computed S.C.F. functions for the sulphur atom in various spectroscopic states arising from the electronic configuration sp3d2 and found the radial maxima of the d functions to be within the bonding region and that therefore it was not necessary to invoke this type of orbital contraction.The view that d orbitals are of a favourable size for bonding has had support from others9 but more recent work still favours the requirement of orbital contraction.1° Indeed at the other end of the scale it has been claimedll that the contraction required (in aromatic sulphides for example) is too great to allow the use of d orbitals in bonding at all. In situations involving T bonding however the necessity for such perturbing effects is not so critical. In a p7r-d~ bond overlap is better when the d orbital is considerably more diffuse than the p orbital than when the two orbitals (p and d ) are of the same size. Put another way in a cr bond or in a T bond involving two p orbitals overlap is most highly developed between the nuclei and so will be greatest if the two orbitals are the same size; however d r a n d p orbitals overlap most efficiently when the lobes of the d orbital are almost above and below the nucleus of the other atom as indicated in Figure Id.This asymmetric overlap would be expected to give rise to a polar bond. ' D. P. Craig and E. A. Magnusson J. Chem. Soc. 1956 4895; D. P. Craig Chem. SOC. Special Publ. No. 12 1958 p. 343. * D. W. J. Cruikshank B. C. Webster and D. F. Mayers J. Chem. Phys. 1964 40 3733. 58 185. lo D. P. Craig and T. Thirunamachandran J. Chem. Phys. 1965,43 4183. l1 G. L. Bendazzoli and C. Zauli J. Chem. SOC. 1965 6827. J. C. Slater Phys. Rev. 1930 36 57. R. S. Mulliken J . Amer. Chem. SOC. 1950 72 4493; H. H. Jaffe J. Phys. Chem. 1954 255 Valence-shell Expansion in Sulphur Heterocycles For the C-S bond Craig found overlap values much larger than normal but this need not infer greater bond strength since by virtue of the low electro- negativities of carbon and sulphur overlap occurs in a region of weak nuclear field.It does however imply that in compounds such as thiophen suchpr-dv bonding may be considerable despite the low electronegativity of carbon and despite the non-development of a formal positive charge on the sulphur atom. 2 Thiophen A. Theoretical Studies.-For many years the electronic structure of thiophen was considered to be entirely analogous to that of furan. In this conception the 3s 3p, and 3p orbitals of sulphur are mixed to form three sp2 hybrid orbitals of which two serve to establish two 0 bonds with the ring carbon atoms and the third non-bonding contains the two electrons with antiparallel spin of the lone pair.There remain two electrons in the 2p orbital capable of entering into conjugation with the 2p electrons of the carbon atoms of the heterocycle. This gives rise to the situation illustrated on Figure 2a which is identical with that for furan except that 3p orbitals for sulphur are used in place of 2p orbitals in the oxygen case. In a pioneering extension of the Huckel molecular orbital (M.O.) theory12 to include heterocyclic systems this scheme was employed by Wheland and Pauling13 and later its use was implicit in an early valence-bond (V.B.) treatment of the chemical reactivity of thiophen.14 Schomaker and Pauling,2 comparing bond lengths resonance energy values and dipole moments in pyrrole furan and thiophen concluded that in thiophen the sulphur appeared to admit to some extent ten electrons into its valence shell in place of the usual octet in oxygen or nitrogen.The value of the C-S bond length in thiophen (obtained from electron diffraction) is 1.74 compared with the sum of the single-bond covalent radii of 1-81 A a shortening of 0.07 A. The corresponding decrease in furan is only 0.03 A. In a similar fashion the dipole moment of thiophen in benzene is 0.54 D and that of tetrahydrothiophen is 1-87 D a difference of 1.33 D. The corresponding difference for furan to tetrahydrofuran is 1.01 D. Further evidence for the stronger conjugation of the sulphur atom with the ring in thiophen than the oxygen atom with the ring in furan comes from thermochemical data. The resonance energies of the two compounds were found to be 31 and 23 kcal.mole-l respectively. These results were held as evidence that in a valence-bond description of the molecule the structures (1) are important contributors to the resonance hybrid of thiophen .. .. .. - - the added stabilisation resulting from this increased delocalisation accounting for the closer similarity of this compound to benzene. l2 E. Huckel 2. Physik 1931 70,204. lS G. W. Wheland and L. Pauling J. Amer. Chem. SUC. 1935 57 2086. l4 P. Daudel R. Daudel Buu-Hoi and M. Martin Bull. Suc. Chim. France 1948 1202. 256 Salmond This essentially qualitative valence-bond representation of thiophen was translated into a quantitative study by L~nguet-Higginsl~ in 1949 with the aid of the M.O. method. He demonstrated that the 3p orbital and two 3d orbitals (the 3d, and 3d,,) of sulphur could be mixed to give three hybrid pd2 orbitals two of which have the proper energy and symmetry characteristics required for entering into conjugation with the 2p orbitals of the neighbouring ring carbon atoms.Since moreover these two sulphur hybrid orbitals are defined to be mutually non-orthogonal the analogy with benzene becomes clear both are systems in which six electrons are associated with six atomic orbitals (instead of a system in which six electrons are associated with five orbitals as in the other scheme). The thirdpd2 orbital which is orthogonal to the other two is mainly 3d in character and is of too high an energy for it to be occupied in the ground state. The projection of the carbon 2p orbitals and the three sulphurpd2 hybrid orbitals on the molecular plane is shown in Figure 2b.Figure 2 Longuet-Higgins concluded that with the use of pd2 hybrid orbitals each atomic orbital contains unit charge as in benzene and in this way the pronounced aromatic character of thiophen compared with furan or pyrrole is readily explained. Since the overlap integrals for the C-S bond (which determine the degree of conjugation with the sulphur atom) are about 20% lower than the corresponding integrals for the C-C bond in benzene the ability of thiophen to show the characteristics of a diene albeit weakly is readily understood. Arising from this for example is a greater reactivity at the a rather than the ,8 carbon atoms. Finally he concluded that the dipole moment of thiophen is due predominantly to the unsymmetric character of the pd2 orbitals.This mode of reasoning with various and minor modifications has been applied to thiophen its substituted derivatives and its ring homologues.16 l5 H. C. Longuet-Higgins Trans. Faraday SOC. 1949 45 173. l6 M. G. Evans and J. de Heer Acta Cryst. 1949 2 363; J. de Heer J. Amer. Chem. SOC. 1954 76,4802; J. Metzger and F. Ruftler J. Chim. phys. 1954 51 52; L. Melander Arkiv Kemi 1955,8,361; 1957,11,397; Acta Chem. Scand. 1955,9,1400; K. Kikuchi Sci. Reports TGhoku Univ. First Ser. 1956 40 133; 1957 41 35; K. Maeda ibid. 1959 43 203; Bull. Chem. SOC. Japan 1960,33 304; J . Koutecky R. Zahradnik and J. Paldus J. Chim. phys. 1959,455; G. Milazzo and G. De Alti Gazzetta 1959,89,2479; R. Zahradnik C. Parkanyi V. Horak and J. Koutecky Coll.Czech. Chem. Comm. 1963,28 776. 257 Valence-shell Expansion in Sulphur Heterocycles Much controversy concerning whether the d orbitals of sulphur should be used in quantum-mechanical calculations remains however. Their use has proved useful in many cases but many workers feel that thiophen and related compounds can still be treated adequately without their inc1~sion.l~ Indeed subsequent to a V.B. treatment of thiophen it has been averred that calculations based on the Longuet-Higgins model are fallacious.ls Treatment of thiophen by the Free Electron method has also been claimedlg to give reasonable correlation with experiment. A notable recent paper20 set out to examine the possibility of inclusion of the d orbitals of sulphur in a quantum-mechanical treatment of thiophen and avoided the unsatisfactory approach of making an apparently arbitrary choice between their inclusion or exclusion.The S.C.F.M.O. method was extended to include more than one atomic orbital per atomic site and in this manner three models differing in the orbitals used on the sulphur atom were studied the first model used only the 3p orbital the second included the 3d, and 3dvz orbitals and the last included also the 4p orbital. It was concluded that the d orbitals do participate in then bonding to a small extent and although this participation is small it nevertheless affects the calculated properties (charge densities and electronic absorption spectrum) markedly. In spite of the many startling results obtained from theoretical calculations there exists still the doubt as to whether they have any real chemical meaning.Do theoretical calculations provide any real chemical information? Does it necessarily mean anything that in one case a p-model gives better correlation with experiment or in another case a d-model? Until questions such as these can be answered the results of theoretical calculations as reliable predictive aids will be treated with considerable reserve. B. Experimental Studies.-In order to elucidate the r6le of the d orbitals of the sulphur atom thiophen and its derivatives have been subjected to various experimental studies often with conflicting results. The technique most extensively used has been ultraviolet (u.v.) absorption spectrophotometry. The differencein the U.V. spectrum of the azine (2) and its carbocyclic analogues was held up as evidence for valence-shell expanson of the sulphur atom in the thiophen rings.21 Conversely the U.V.spectrum of the dibenzothiophen derivatives (3) seemed to indicate non-involvement of the d orbitakZ2 In a short review of the U.V. absorption spectra of aromatic sulphur compounds it was l7 G. Berthier and B. Pullman Compt. rend. 1950 231 774; S. Nagakura and T. Hosaya Bull. Chem. SOC. Japan 1952 25 179; M. M. Kreevoy J. Amer. Chem. SOC. 1958 80 5543; F. L. Pilar and J. R. Morris J. Chern. Phys. 1961,34,389; D. S . Sappenfield and M. M. Kreevoy Tetrahedron 1963 Suppl. 2 1963 157; A. J. H. Wachters and D. W. Davies Tetrahedron 1964,20,2841; N. Solony F. W. Birss and J. B. Greenshields Cunad. J. Chem. 1965 43 1569; D. T. Clark Tetrahedron Letters 1967 2889. A. Mangini and C.Zauli J. Chem. Soc. 1960,2210. lS T. N. Rekaseva Optika i Spektroskopija 1961,11,284. 2o M. J. Bielefield and D. D. Fitts J. Amer. Chem. SOC. 1966 88 4804. 21 H. H. Szmant and H. J. Planinsek J. Amer. Chem. SOC. 1950 72 4981. aa A. Mangini and R. Passerini Gazzetta 1954 84 635. 258 Salmond concluded that although involvement of the d orbitals in open-chain compounds is unlikely it is nevertheless probable in heterocyclic systems.23 Recent infrared and Raman studiesM of 2-halogenated thiophens favour the contribution of the structures (4) to the resonance hybrid of the molecule as postulated in earlier studies26 to explain the magnitudes of the dipole moments of these compounds. The high nuclear quadrupole resonance frequency of the chlorine atom in 2-chlorothiophen has been interpreted in terms of similar However recent e.s.r.work on the radical anions derived from dibenzothiophen and its oxygen and selenium analogues together with the U.V. absorption spectra and thepolarographic reduction potentials of thesecompounds are in firm disagreement with a model which incorporates valence-shell expansion of the hetero-at~m.~' Convincing ,evidence for through-conjugation via a sulphur atom seems to emerge from n.m.r. studies. The coupling constant between the 2- and the 5-hydrogen atoms (J2d in furan derivatives is 1.5 c./secas The corresponding Jza in thiophen compounds is 2.7 C . / S ~ C . ~ ~ suggesting more through-conjugation via the sulphur atom by means of structures such as (lc). Such structures (5) involving hyperconjugation of the methyl group with the d orbitals of sulphur have been invoked to explain the coupling of the methyl protons with the /3-olehic protons in methyl vinyl sul~hide.~O No coupling occurs with the a protons.Similar structures may also explain the coupling between the 2- and 6-protons in benzo[b]thiophen which is ca. 0-5 c./secgl The coupling J26 in benzo [blfuran has not been resolved.sg It is significant perhaps that in indolizine Jze is of the order of 0-5 c./secS3 Through-conjugation of the 2- and 6-protons in a manner analogous to benzo[b]thiophen is possible with indolizine but not with benzo[b]furan (cf. structures 6a by c). a8 A. I. Kiss Acta Phys. et Chem. Szeged 1960,45. ar M. Horak I. J. Hymans and E. R. Lippincott Spectrochim. Acta 1966 22 1355. as M. T. Rogers and T. W. Campbell J. Amer. Chem.SOC. 1955 77 4527. M. J. S. Dewar and E. A. C. Lucken J. Chem. SOC. 1959 426. R. Gerdil and E. A. C. Lucken J. Amer. Chem. SOC. 1965 87,213; 1966 88,733. a8 R. J. Abraham and H. J. Bernstein Canad. J. Chem. 1961 39 905. S. Gronowitz and R. A. Hoffman Arkiv Kemi 1958,13,279; C. A. Reilly Analyt. Chem. 1961 32 221; D. M. Grant R. C. Hirst and H. S. Gutowsky J. Chem. Phys. 1963 38 470. *O R. T. Hobgood G. S. Reddy and J. H. Goldstein J. Phys. Chem. 1963 67 110; M. C. Caserio R. E. Pratt and R. J. Holland J. Amer. Chem. SOC. 1966 88 5747. s1 K. Takahashi T. Kanda F. Shoji and Y. Matsuqci Bull. Chem. SOC. Japan 1965 38 508. *a P. J. Black and M. L. Heffernan Austral. J. Chem. 1965 18 353. ** P. J. Black M. L. Heffernaa L. M. Jackman Q. N. Porter and G. R. Underwood Austral. J. Chem. 1964,17 1128 259 Valence-shell Expansion in Sulphur Heterocycles A transient existence for the fully delocalised structure (7a) of which structures (7b c) are contributors to the resonance hybrid has been claimed.34 It is tempting certainly to write such a formulation for this compound but its high reactivity has also been attributed to a triplet ground states5 (calculations based on ap-model).But more important in such situations is the necessity for care in the interpretation of the experimental results. Sulphoxide (8) when dehydrated in presence of acetic anhydride and N-phenylmaleimide yields the adduct (9) where the dienophile has attacked the ring which was originally aromatic and which holds methyl substituents. Since methyl groups are known to increase the rate of some Diels-Alder reactions this behaviour was deemed evidence for the intervention of (7) as an intermediate.However such a dehydration and addition might easily follow the concerted Scheme 1 which does not require an intermediate involving d orbitals or a triplet ground state. Without a more thorough investigation of the reaction such a sequence cannot be dismissed. 34 M. P. Cava and N. M. Pollack J. Amer. Chem. SOC. 1967,89,3639. 35 D. T. Clark Tetrahedron Letters 1967 5257. 260 Salmond From the conflicting nature of the experimental results it appears that the use of d orbitals may not be universal; that there may be situations where for some reason their use is favoured and other situations where they are not. 3 Other Systems with Sulphur as the Sole Hetero-atom In a discussion of aromaticity in carbocyclic systems it is a logical step from benzene to tropylium so here the thiopyrylium ion follows thiophen.In a theoretical treatment of this compound it has been found useful to include the d orbitals of the sulphur atom.36 However calculations based on a p-model also agree well with the observed U.V. and n.m.r. results of various thiopyrylium ions.37 A similar analogy with tropone is found in y-thiopyrone which has been studied extensively in its anomalous relation to y-pyrone-anomalous that is in regard to the relation of acyclic ethers to sulphides. Evidence for the ability of sulphur to conjugate with a carbonium centre is manifest in the very high rate of hydrolysis of a-chloro-~ulphides~~ compared with the rate of hydrolysis of alkyl halides. This reaction proceeds via a carbonium-ion intermediate39 which is presumably stabilised by delocalisation of the positive charge the 3pn orbital of the sulphur atom overlapping with the vacant 2p7~ orbital of the adjacent carbon atom to give a 2pn-3pn bond.Were this interpretation correct a-chloro-ethers would be expected to hydrolyse** even more quickly since the 2p7r orbital of the oxygen atom would give rise to a 2 p 7 r - 2 ~ bond where overlap would be more substantial. This is the case and numerous examples of this greater electron-releasing power of oxygen than sulphur41 are to be found in the literature. The two compounds y-pyrone and y-thiapyrone however provide a notable exception to this behaviour. The dipole moment (3.82 D) of the y-pyrone (10) indicates a large contribution from the polar structure (lob) to the resonance The absence of a normal carbonyl frequency from the i.r.spectrum,43 the lack of normal carbonyl reactivity and the basicity of the compoundu s6 J. Koutecky Coll. Czech. Chem. Comm. 1959 24 1608. s7 T. E. Young and C. J. Ohnmacht J. Org. Chem. 1967,32,444 1558. 38 R. A. Peters and E. Walker Biochem J. 1923,17,260; H. Mohler and J. Hartnagel Helv. Chim. Acta 1941 24 564; 1942 25 859; H. Bohme and K. Sell Chem. Ber. 1948 81 123. s8 F. G. Bordwell G. D. Cooper and H. Morita J. Amer. Chem. SOC. 1957 79 376. 40 H. Bohme H. Fischer and R. Frank Annalen 1949 563 54. 41 C. K. Ingold and E. H. Ingold J. Chem. SOC. 1926 1310; E. L. Holmes and C. K. Ingold ibid. 1328. 42 E. C. E. Hunter and J. R. Partington J. Chem. SOC. 1933 87; F. Arndt G.T. 0. Martin and J. R. Partington ibid. 1935 602. 43 A. Ross Proc. Roy. SOC. 1926 A 113,213; L. Kahovec and K. W. F. Kohlrausch Chem. Ber. 1942 75 627. 44D. S. Tarbell and P. Hoffman J. Amer. Chem. SOC. 1954 76 2451. 261 Valence-shell Expansion in Sulphur Heterocycles are consistent with this polarisation. By analogy with the examples of acyclic compounds just given the y-thiapyrone should be less polarised in view of the suggested smaller electron-releasing ability of the sulphur atom. In fact the strong absorption in the double-bond region of the i.r. spectrum of the y-thiapyrone (11) is even lower44 and the dipole moment (4.40 D) is higher than the value for the corresponding y-pyrone. In this instance therefore the electron-releasing ability of sulphur is greater than that of oxygen and there appears to be a considerable contribution of the polarised structure (llb) to the resonance hybrid an assumption which is further supported by the remarkable resonance energy of y-thiapyrone estimated at 33 kcal.Although electron-releasing conjugation of this type does not prima facie involve the use of d orbitals of the sulphur atom this remarkable change in the order of events may well be explicable in terms of more d orbital interaction in a ring than in open-chain compounds perhaps because of more favourable geometry . Arguments of this nature have also been called upon to rationalise the n.m.r. spectra of pyrone and thiapyr~ne.~~ The midpoint of the spectra of these com- pounds is found to shift downfield on the replacement of oxygen by sulphur (contrary to the normal upfield shift in acyclic compounds) and this has been interpreted in terms of an increased ring current in the sulphur-bearing ring brought about by the increased ability of sulphur to conjugate via the d orbitals.In further support of this proposition the J26 coupling in thiapyrones (cu. 4.0 c./sec.) is in general greater than 4 6 in pyrones (ca. 1.2 c./s~c.).~' Further more distant analogues of tropone are provided by the isomeric benzothiepin dioxides (12)48 and(1 3).49 The compound (12) possesses no aromatic property and loses sulphur dioxide readily on heating; its isomer seems to be somewhat more stable but details of its properties are not yet available. However a highly substituted derivative of this system compound (14) seems to have no aromatic delocalisation in the seven-numbered ring as judged from its n.m.r.spectrum50 and the parent compound (15) appears to be completely non- aromatic.51 It is stable at ordinary temperatures but its n.m.r. spectrum is reminiscent of that of a conjugated olefin its U.V. spectrum is similar to that 46 L. Lorenz-Oppan and H. Sternitzke Z. Elektrochem. 1934 40 501. 4~ J. Jonas W. Derbyshire and H. S. Gutowsky J. Phys. Chem. 1965 69 1. 47 The couplings Jza and J, are wrongly assigned in ref. 46 but the arguments presented here are not changed; see R. E. Mayo and J. H. Goldstein Spectrochim. Acta 1967 23A 55. 48W. E. Truce and F. J. Lotspeich J. Amer. Chem. SOC. 1956 78 848. V. J. Traynelis and F. F. Love Chem. and Ind. 1958,439; J. Org. Chem. 1961,26,2728. 50 H. Hoffman and H. Westernacher Angew. Chem.Internat. Edn. 1966 958. 61 W. L. Mock J. Amer. Chem. SOC. 1967 89 1281. 262 Salmond of cycloheptatriene and it is readily hydrogenated. All of these properties together with its thermal instability argue against any contribution of the structure (1 5b). The analogue of cyclopentadienide the four-membered ring system (16) in which the d orbitals of the sulphur atom would be involved in any cyclical delocalisation has not yet been prepared and is predicted to be quite unstable.52 The thiete (17),53 the conjugate acid of this system has been prepared but preliminary experiments indicate its non-acidic character. It is perhaps not surprising therefore that a system incorporating both the thiepin dioxide and the thietide systems namely compound (18) was found to possess no aromatic stabilisation.54 Sulphur-containing analogues of cyclo-octatetraene are known and as expected are somewhat unstable. For example the dicarboxylic acid (19) loses sulphur very readily to yield naphthalene2,3-dicarboxylic acid.55 Strangely the diester (20) seems to be much more stable and was attributed some degree of resonance stabili~ation,~~ but this structure has been contesteds7 and in its place structure (21) was suggested. The n.m.r. spectrum of the benzothiepin derivative (22) is also consistent with lack of aromatic delocalisation in the heterocyclic ring.50 The double bond in derivatives of dibenzothiepin (23) is stated to be less ‘olefinic’ than that in dibenzocyclo-~ctatetraene.~~ Nevertheless e.s.r. studies of compound (23) itself suggest that more conjugation is still transmitted via the double bond than viu the sulphur bridge.59 In addition these compounds iose sulphur thermally to give phenanthrene derivatives.6o Similarly 1,2-dithiin (24)61 loses sulphur readily to yield thiophen derivatives m R. Zahradnik and C. Parkanyi Coll. Czech. Chem. Comm. 1965 30 3016. O9 D. C. Dittmer K. Takahashi and F. A. Davis Tetrahedron Letters 1967 4061. 54 R. Breslow and E. Mohacsi J. Amer. Chem. SOC. 1962 84 684. ss G. P. Scott J. Amer. Chem. SOC. 1953,75 6332. 56 K. Dimroth and G. Lenke Chem. Ber. 1956,89,2608. ti7 A. Schonberg and M. B. E. Fayez J. Org. Chem. 1958,23 104. m E. D. Bergmann and M. Rabinovitz J. Org. Chem. 1960 25 828. 6m M. M. Urberg and E. T. Kaiser J. Amer. Chem. SOC. 1967 89 5931. *l W. Schroth F. Billig and G. Reinhold Angew. Chem. Internat.Edn. 1967 6 698. J. D. Louden A. D. B. Sloan and L. A. Summers J. Chem. Soc. 1957 3814. 263 Valence-shell Expansion in Sulphur Heterocycles nor does it display any properties consistent with cyclical delocalisation. 1,4-Dithiin7 however is a stable compound and its benzo-derivative has been attributed some degree of aromatic character.s2 The parent compound although not exhibiting these aromatic properties nevertheless is not affected in many procedures which cause reaction in vinyl sulphides. The added stabilisation in this molecule has been put down to the involvement of d orbitals of the sulphur atom especially as the ring is known to be puckereda3 andp-orbital delocalisation therefore impossible (unless of course some o-overlap of p orbitals is admitted).64 A large resonance energy has been calculated for 174-dithiin in terms of ap-model though in a more sophisticated treatment still neglecting d-orbitals much of this resonance energy is lost.Lucken in his e.s.r. work on 1,4-dithiin radical cations and their derivativess5 has shown that a d-model gives a quantitatively better description than a p-model. He does point out however that the involvement of the d-orbitals of the sulphur atom may only be apparent owing to the probable non-planarity of the ring system allowing the interaction of the C-S antibonding orbital with the carbon 2pE orbitals,ss which would have a similar effect as the use of the d-orbitals. Systems with seven T-electrons formally associated with the ring have been invoked as intermediates in the dehydration of the sulphoxides (26) and (29) in the presence of N-phenylmaleimide and acetic anh~dride.~' Sulphoxide (26) gives rise to the adduct (27) but for reasons given earlier this need not imply the separate existence of the intermediate (28).Likewise although the formation ?- P i b-Ph %* See W. E. Parham in 'Organic Sulphur Compounds' ref. 4 p. 248. P. A. Howell R. M. Curtis and W. N. Lipscomb Acta Cryst. 1954 7 498. 64 See c.g. K. G. Untch J. Amer. Chem. SOC. 1963 85 345. rsE. A. C. Lucken J. Chem. SOC. 1962 4963; Theor. Chim. Ada 1963 1 397. J. F. A. Williams Trans. Faraday SOC. 1961 57 2089. 07 M. P. Cava N. M. Pollack and D. A. Repella J. Arner. Chem. SOC. 1967 89 3640; R. H. Schlessinger and I. S. Ponticello ibid. p. 3641 ; Tetrahedron Letters 1967 4057. 264 Salmond of the adduct (30) from sulphoxide (29) may argue against the structure (31a) for the intermediate it does not imply the validity of the structure (31b) as the intermediate.4 Thiabenzene Derivatives When phenyl-lithium and the thiapyrylium salt (32) are allowed to react the thiabenzene (33) is produced. This has been isolated as a deep purple amorphous solid which isomerises slowly to the isomeric thiapyran (34). Attempts to prepare S-alkylthiabenzene derivatives gave only the isomeric thiapyrans. When solutions of the thiabenzene (32) are treated with oxygen and then acidified the betaine (35) and thiophenol are produced indicating no doubt that the phenyl group from phenyl-lithium has become attached to sulphur and thus establishing the structure of the compound. The ring homologues of thiabenzene (36-39) are much more stable to heat to light and to oxygen.Nor do they revert to the isomeric thiapyrans. They are also less highly coloured than the parent compound (red rather than violet).s8 Ph Ph @Ph (32) The dipole moments of all five compounds in benzene lie in the range 1-50-1-88 D indicating a larger contribution from the covalent structure to the resonance hybrid than the ylide structure. Phe dipole moment of the ylide (40) is 6.2 D ] . ~ ~ Moreover n.m.r. data support an aromatic structure. Similarly the low basi~ities'~ compared with acyclic ylides and the resistance of compounds (37) and (39) to desulphurisation with Raney nickeI suggest marked stabilisation by through-conjugation in the six-membered ring. Commenting on the electronic structure of the S atom in these compounds Price has suggested that instead of a p3 arrangement for the a-framework around the suIphur atom the hybridisation changes to sp2 so that the thiabenzene ring and the S-phenyl ring become coplanar leaving the 3p orbital of the sulphur atom to conjugate with the 2p orbitals of the neighbouring 88 G.Suld and C. C. Price J. Amer. Chem. Suc. 1961,83 1770; 1962,84,2090; 2094; C.C. Price M. Flori T. Parasaran and M. Polk ibid. 1963 85 2278. 6B G. M. Phillips J. S. Hunter and L. E. Sutton J. Chem. SOC. 1945 146. 'O E. A. Blair Dim. Abs. 1965 26 1912. 265 Valence-shell Expansion in Sulphur Heterocycles carbon atoms. The lone pair would then occupy a 3d orbital. In the monocyclic compound such a planar array would be hindered by the adjacent phenyl groups the p3 geometry being maintained.The lone pair would then occupy the 3s orbital and any through-conjugation with the ring would have to take place via a p7r-d~ bond this conceivably being the source of the relative instability of this system. The U.V. spectra of thiabenzenes are similar to those of the related benzenoid systems so Price’s theorys8 concerning the electronic structure of the sulphur atom is supported by the additional absorption in the visible region causing excitation of one electron from a low-lying d orbital into a higher d orbital; especially is this so if the lone pair is contained in a d orbital. This hypothesis is supported by comparison of these thiabenzenes with the analogous pho~phabenzenes.~~ In the phosphabenzenes all the valence electrons associated with the phosphorus atoms are involved in the bonding so that any aromatic delocalisation in the ring must be transmitted via a pn-dn bond.It is perhaps significant therefore that the phosphabenzenes are much less stable and that thiabenzenes are highly coloured compounds whereas the phosphabenzenes are pale yellow. Moreover the phosphabenzene (41) is protonated readily in aqueous media to the conjugate acid phosphonium salt (42). The thiabenzenes are not protonated at all under these conditions. However thiabenzene S-oxides,72 for example compound (43) a remarkably stable substance exchange the ring protons readily in acidic media behaviour consistent with ‘ylide-character’. Four main views are held on the bonding in phosphom-containing compounds. Treating the electronic structure of phosphorus in phosphates Fukui and his considered one d r orbital (a d?s hybrid) and one electron for the phosphorus atom and Mason 74 in his work on phosphabenzene chose a model similar to that Craig uses for phosphonitrilic compounds for example compound ( 4 ) .7 5 Craig7s believes that the 3d, orbital of the adjacent nitrogen atoms 71 G. Markl Angew. Chem. Internat. Edn. 1963 2 153. 72 A. G. Hortmann J. Amer. Chem. SOC. 1965 87,4972; Y. Kishida and J. Ide Chcm. and Pharm. Bull. (Japan) 1967 15 360. 73 K. Fukui K. Morokuma and C. Nagata Bull. Chem. SOC. Japan 1960 33 1214. 74 S. F. Mason Nature 1965 205 495. 75 N. L. Paddock Quart. Rev. 1964,18 168; D. P. Craig and N. L. Paddock Nature 1958 181,1052. 76 D. P. Craig J. Chem. Soc. 1959 997. 266 Salmond (Figure 3a) to give a system in which there is complete cyclic delocalisation.In this type of delocalisation the 'aromatic sextet' loses its significance since the interactions of the d, orbital with the 2p orbitals on either side are of opposite sign. This leads to the result that the lower filled energy levels are degenerate (whereas in benzene of course the highest filled energy level is degenerate) and the delocalisation energy rises slowly with the number of T electrons (provided there is an even number) in the (AB) system. (A is a first row element which gives rise to a p orbital and B is a second row element providing a dv orbital). Dewar and his collaborator^,^^ however point out that the d,,, orbital ought to be included in the discussion. Assuming that the electronegativities of the d, and d, orbitals are equal and the orbitals thus being equivalent any linear combination of these orbitals would be suitable for this valence problem.Dewar suggested the sum and the difference. This gives rise to another two hybrid orbitals (designated d+ and d-) which are simply as if the previous two had been rotated in a clockwise direction through an angle of 45". This brings them into a much more favourable position for overlap with the adjacent 2p orbitals of the nitrogen atoms. In this scheme the 2p orbital of N(2) overlaps strongly with the d- orbital at P(1) and the d+ orbital at P(3) (Figure 3b). Now the d, and duz orbitals are orthogonal; so also therefore are d+ and d-. Delocalisation therefore takes place over three-centre (P-N-P) 'islands' and not over the complete cyclic structure as in benzenoid chemistry since conjugation is effectively interrupted at each phosphorus atom.This procedure has been critici~ed~~ because the dZz and d, orbitals are assumed to have the same electronegativities. Since the weakly bonded d orbitals '7 M. J. S. Dewar E. A. C. Lucken and M. A. Whitehead J. Chem. SOC. 1960,2423. 78 D. P. Craig M. L. Heffernan R. Mason and N. L. Paddock J. Chern. Soc. 1961 1376. 267 Valence-shell Expansion in Sulphur Heterocycles are easily polarisable the degree of polarisation clearly depends closely upon the environment. The d, orbital is tangential to the ring; the dv radial so the electronegativities are expected to differ. In a further paper Craig and Paddock79 admit to the involvement of d orbitals other than dz,. For example the lone pairs of the nitrogen atoms in an spy orbital are allowed to overlap with the vacant dxs-ya orbital of the phosphorus atom giving rise to n’-bondings0 around the periphery of the ring.Later the two approaches were compared and it was shown that the ‘island’ model was merely a special case of the cyclic model and that the delocalisation energy in the ‘island’ model is always less although the differences between the two may be small.81 Extending such arguments to 1 ,l-diphenylphosphazene (43 this compound may be stabilised by a fully delocalised n-electron system or by an ‘internal cyanine’ structure stabilisation. It is important to note that the phosphabenzene (45) has very similar U.V. absorption to the linear phosphonium salt (46) suggesting that the ‘island’ model is more correct in this case.82 In agreement a recent theoretical appraisal of the four models favours the Dewar or the Fukui models.83 By analogy those arguments which apply to phosphabenzenes may be extended to thiabenzenes and thence to other sulphur-containing heterocycles.5 Mixed Systems A. Thiazo1e.-While attempting to explain the base-catalysed exchange of the 4-methyl hydrogen atoms with deuterium in the thiazole (47) Erlenmeyer1 invoked structures such as (48). This appears to be the earliest reference to such structures involving quadrivalent sulphur. From U.V. studies of isomeric 2-styryl-4-aryl- and 2-styryl-5-aryl-thiaoles it has been concluded that valence-shell expansion may be imp~rtant.~~ To add to the confusion concerning the applicability of theoretical calculations recent studies of radical phenylation p-nitrophenylati~n,~~ and methylations6 of thiazole and its methyl derivatives accord well with recent calculations based on a d-models7 while disagreeing with a previous p-modeLs8 However the same experimental results are in agree- ment with a later theoretical treatment based on another p - m ~ d e l .~ ~ The anomalous downfield resonance of the 2-proton in the n.m.r. spectrum of thiazolegO may perhaps be due to feedback of the p electrons of the carbon- ?* D. P. Craig and N. L. Paddock J. Chem. SOC. 1962,4118. D. W. J. Cruickshank J. Chern. SOC. 1961 5486. D. P. Craig and K. E. Mitchell J. Chem. SOC. 1965 4682. sa G. Mbkl Angew. Chem. Internat. Edn. 1964 3 147. R. Vilceanu A. Balint and Z. Simon Nature 1968 217 61. 84 E. D. Sych and L. P. Umanskaya J. Gen. Chem.(U.S.S.R.) 1963 73. 86 J. Vitry-Raymond and J. Metzger Bull. SOC. chim. France 1963 1784. 86 H. J. Dou Bull. SOC. chim. France 1966 1678. E. Vincent and J. Metzger Bull SOC. chim. France 1962 2039. J. Metzger and A. Pullman Compt. rend. 1948 226 898; Bull. SOC. chim. France 1948 R. Zahradnik and J. Koutecky Coll. Czech. Chem. Comm. 1961 26 156. 1166; A. Pullman and J. Metzger ibid. p. 1021. *@ A. Taurins and W. G. Schneider Canad. J. Chem. 1960 38 1237. 268 Salmond nitrogen double bond into a vacant d orbital. This postulation may be supported by the 13C-H coupling constants observed for thiazolium salts (see below concerning thiamine) and from recent studies of the 13C-H coupling constants in vinyltin compoundsQ1 and n.m.r. studies in vinylsilicon compounds.92 In addition the JZ5 coupling constant in thiazole is 1.9 c./sec.and J2* is 0 c./sec. while the corresponding couplings in oxazole are both very small again suggesting the involvement of the d orbitals of the sulphur atom in transmitting conjugation. Considerable work has been done on the thiazolium ring system in connection with the commercially important cyanine dyes. The thiazole dyes (49-51) are - + I R Me Me W C H - C H = $ j q & t S-S Me (50) Et bt all deeper in colour than their oxygen analogues.93 This uniform behaviour led Knottg4 to suggest that valence-shell expansion in the heterocyclic ring was responsible a supposition which is supported by further U.V. studies of cyanine dyes (52) derived from 2-methyl-4-styryl- and 2-methyl-5-styryl-thiazoles where it was found that a styryl group in the 5-position of the thiazole ring is more feebly conjugated with the polymethine chromophore than one in the 4-p0sition.*~ P h - C H = C H g Y C H = C H - C H y S 1 CHZCH-Ph (52) N Et Et + I Cocarboxylase (53a) the pyrophosphate of thiamine (53b) is the coenzyme for a number of important biochemical processes.From deuterium exchange O1 D. J. Blears S. S. Danyhik and S. Cawley J. Organometallic Chem. 1966 6 284. oP R. T. Hobgood J. H. Goldstein and G. S. Reddy J. Chem. Phys. 1961,35 2038; R. T. Hobgood and J. H. Goldstein Spectrochim. Ada 1963 19 321; R. Summit J. J. Eisch J. T. Trainor and M. T. Rogers J. Phys. Chem. 1963,67 2362. 83 B. Beilenson N. I. Fisher and F. M. Hamer Proc. Roy. SOC. 1937 B 163 138; L. G. S. Brooker G. H. Keyes R. H. Sprague R. H. Van Dyke E. Van Lare G.Van Zandt F. L. White H. W. J. Cressman and S. G. Dent J. Amer. Chem. SOC. 1951,73 5332; E. B. Knott J. Chem. SOC. 1952 4099; R. A. Jeffreys ibid. p. 4823. *( E. B. Knott J. Chem. SOC. 1955,916. Bs E. D. Sych and E. D. Smaznaya J. Gen. Chem. (U.S.S.R.) 1963 68. 269 Valence-shell Expansion in Sulphur Heterocycles studies on simpler model thiazolium salts the action of thiamine has been explained in terms of ionisation of the 2-hydrogen of the thiazolium ring to give the intermediate thiazolium anion The exchanged deuterium was detected originally by mass spectral analysis and subsequently (unambiguously) by n.m.r. studies. Breslow stated that valence-shell expansion in the sulphur atom is an unlikely explanation for the stabilisation of the thiazolium anion owing to the unfavourable geometry required in the bent allenic structure at the 2-carbon atom (54b).It is true that such an ylide is not stabilised wholly by valence-shell expansion since imidazolium and oxazolium cations also exchange their 2-hydrogen atoms readily in the presence of base. The difference in the rates of exchange between oxazolium and thiazolium cations and the magnitudes of the 13C-H coupling constants of the 2-protons in the two systems however indicate that d-orbital interaction may well be occurring and the benzyne-like structure (54c) as a contributor of the resonance hybrid of the intermediate has been post~lated.~~ Calculations indicated that in the thiazolium ring a greater part of the charge (0.6) resides on the sulphur atom so that sufficient d-orbital contraction may occur to allow overlap with the adjacent sp2 orbital containing the lone pair.(The overlap of two sp2 orbitals has been suggested with respect to benzene;98 and the overlap of an sp2 orbital containing a lone pair with a vacant d orbital constitutes the important d- bonding in phosphonitrilic compounds). However these calculations were based on a p-model and it seems not altogether aesthetically satisfying to calculate the charge in the sulphur atom with such a model and then to proceed to say that the charge is sufficient to cause d-orbital contraction. Moreover the use of 13C-H coupling constants as a guide to the s-character of a bond has been criticised. 99 Olofson and his collaboratorslOO have substantiated considerably the case for d-a overlap by studying the rates of deprotonation of thiazole and related bases.They found an extraordinary rate enhancement when a proton is on a carbon atom adjacent to sulphur and that the position of the nitrogen (whether a or 18 to that proton) has little effect. Remarkably the 5-proton in isothiazole is ionised more readily than the 2-proton in thiazole. This behaviour is echoed in the thiadiazole series where H(5) in compound (55) is deprotonated more R. Breslow Chem. and Ind. 1957 893; J. Amer. Chem. SOC. 1957 79 1762; 1958 80 3719. 07 P. Haake and W. B. Miller J. Amer. Chem. SOC. 1963 85 4044. J. D. Roberts D. Semenov H. Simmons and L. Carlsmith J. Amer. Chem. SOC. 1956 78 601. G. J. Karabatsos and C. E. Orzech J. Amer. Chem. SOC. 1964 86 3574. 100 R. A. Olofson and J. M. Landesburg J. Amer. Chem. SOC. 1967,88,4263; R.A. Olofson J. M. Landesburg K. N. Houk and J. S. Michelman ibid. p. 4265. 270 Salmond quickly than H(2) in (56). All these results provide strong indications of d-a overlap with the sulphur atom. B. 1,2,5-Thiadiazole.-It has been claimedlOl that the stability of the well-known thiadiazole derivatives (57) and (58) implies a significant degree of quadrivalent character in the sulphur atom of these compounds. X-Ray crystallographic datalo2 seem to support this claim. The reactions of the parent compound (59) itself suggest that the structure (59b) is an important contributor to the resonance hybrid.lo3 However microwavelM and electron diffraction studieslos suggest that the structure (59a) in which the sulphur is bivalent is more important in the ground state. These two views are however not incompatible since the reactions could involve excited electronic and vibrational states as well as the ground state.A recent theoretical study has indicated that the quadricovalent character of the sulphur atom increases in the series (59) (57) and (60).lo6 It is noteworthy that the U.V. absorption spectra of (57) (Amax 310 nm.)lo7 and (61) (hm,,274 nm)lo8 are quite different. The analogous structures (57a b) and (61a b) can be drawn for both compounds but no structure analogous to (57c) can be drawn for the benzotriazole. On the other hand the U.V. spectra of the tricyclic systems (62)loQ and (63)110 which have seven n-electrons associated lol M. P. Cava and R. H. Schlessinger Tetrahedron Letters 1964 3815. lo2 V. Luzzatti Actu Cryst. 1951,4 193. lo3 L.M. Weinstock Diss. A h . 1959 19 3136; M. Carmack L. M. Weinstock and D. Shew Abs. 136th National Meeting of Amer. Chem. SOC. Atlantic City N.J. Sept. 1959 p. 3%. lo' V. Dobyns and L. Pierce J. Amer. Chem. SOC. 1963,85,3553. lo5 F. A. Momany and R. A. Bonham J. Amer. Chem. SOC. 1964 86 162. lo6 N. K. Ray and P. T. Narasimhan Theor. Chim. Acta 1966 5 401. lo' L. S. Efros and R. M. Levit J. Gen. Chern. (U.S.S.R.) 1955 25 183. Io8 D. Dalmonte A. Mangini R. Passerini and C. Zauli Gazzetta 1958 88 977. loo R. Dietz Chem. Comm. 1965 57; H. Behringer and K. Leiritz Chern. Ber. 1965 98 3196. M. J. Perkins J. Chem. SOC. 1964 3005. 271 Valen ce-shell Expansion in Sulphur Heterocycles with the heterocyclic ring are remarkably similar (h,, 658 and 655 nm. respectively) implying no contribution from the N=S=N bond system in the thiadiazine derivative.Also in the eight-r-heterocyclic system (64) sulphur is easily extruded to give diazaphenanthrene.lll The properties of these compounds thus strongly suggest delocalisation via the d electrons of the sulphur in six n-systems. This may be the reason for the observed increase in conjugation observed in the series (65-67) although interaction between the oxygen and sulphur atoms (676) as in furothiophthen derivatives (see later) is undoubtedly also important.l12 However e.s.r. studies of the radical anion derived from (57) based on two different d-models suggest that d-orbital interaction is not import ant .l13 C. Mesoionic Compounds.-AIthough a large number of mesoionic compounds have been prepared few physical measurements have been made with a view to evaluating their electronic structures.On the basis of U.V. solvent-shift studies114 the structure (68) has been claimed to be important in the resonance hybrid of the molecule and the ease of nucleophilic attack at the 5-position in the thiazole ring of compound (69) has been attributed to similar structures.l16 0 RS-#? s-s 0 Jx:.. Me 6 Thiothiophthens Treatment of diacetylacetone with phosporus pentasuIphide gives a compound to which Amdt116 assigned the structure (70). This structure seemed to be confirmed by chemical means when it was found that potassium hydrosulphide reacted with 2,6-dimethylthiapyrone to yield the oxygenated compound (71) which on treatment with phosphorus pentasulphide gave Arndt's compound.l17 ll1 G. R. Collins Diss. Abs.1966 4 0 3 ~ . 112 A. K. Kirby Tetrahedron 1966 3001; T. R. Lynch I. P. Mellor S. C. Nyburg and P. Yates Tetrahedron Letters 1967 373. 113 E. T. Strom and G. A. Russell J. Amer. Chem. SOC. 1965 87 3326; N. M. Atherton J. N. Ockwell and R. Dietz J. Chem. SOC. (A) 1967 771. 11* E. B. Knott J. Chem. SOC. 1955 937. 115 G. F. Duffin and J. D. Kendall J. Chem. SOC. 1956 361. 11* F. Arndt Rev. Fac. Sci Univ. Istanboul 1948 13 A 57; F. Arndt P. Nachtwey and J. Push Chem. Ber. 1925 58 1638. 117 G. Traverso and M. Sanesi Ann. Chim. (Italy) 1953,43 795; G. Traverso ibid. 1954,44 1018; Chem. Ber. 1958 91 1224. 272 Salmond Early n.m.r. work also seemed to c o b the structure (70) since two equivalent hydrogen atoms and two equivalent methyl groups were shown to be present.l18 However almost simultaneously both i.r.ll9 and X-ray diffraction studies120 indicated that the trisulphurated compound had the structure (72) and the oxygenated compound structure (73).The S-S bond length in (72) is 2.36 A greater than the S-S bond length of 2-00-2.10 A in disulphides121 but still much less than the sum of the van der Waals radii of 3-70 A between two sulphur atoms so bonding to some extent between the two outer with the inner sulphur atom is apparent. Moreover the C-C bond distances are all in the range 1.37-1.38 A the bond length of ‘aromatic’ double bonds. The compound was stated to exhibit ‘no-bond’ resonance which conferred aromatic properties on the rings. In the oxygenated compound (73) the S-S bond length of 2-12 A is much nearer normal suggesting solely the structure (73a) without resonance interaction.However the S-0 distance is 2.41 A halfway between the ‘normal’ S-0 single bond length and the sum of the van der Waals radii of the sulphur and oxygen atoms. This may be electrostatic in origin. The polarisation (736) can be envisaged followed by an electrostatic attraction between the negative oxygen and the positive ring perhaps with some covalent bonding by involvement of the d orbitals of sulphur. Where a full negative charge is available on the oxygen atom as for example in the amino-furothiophthen derivatives (see later) a sulphur oxygen bond does seem to be formed. The X-ray work on these compounds was c o n h e d by later n.m.r. studieslZ2 when it was demonstrated that the oxygenated compound (73) possessed non-equivalent methyl groups and hydrogen atoms.Moreover the proton resonances in the trisulphurated derivatives are consistently down-field from those in the furothiophthens implying a degree of aromaticity to the thiothio- phthens which is not present in the oxygen analogues. Once the structures of these compounds had been established the chemistry of the system developed rapidly and many thiothiophthens have been prepared by different routes and from isomeric starting materials indicating the delocalised nature of the system.123 Replacement of carbon by nitrogen as in 11* A. Bothner-By and G. Traverso Chem. Ber. 1957,90,453. ll0 G. Guillouzo Bull. SOC. chim. France 1958 1316. lZo S. Bezzi C. Carbuglio M. Mammi and G. Traverso Gazzetta 1958 88 1226; S. Bezzi M. Mammi and C. Carbuglio Nature 1958 182 247; M.Mammi R. Bardi G. Traverso and S. Bezzi ibid. 1961 192 1282. lal I. M. Dawson and J. M. Robertson J. Chem. SOC. 1948 1256; A. Hordvik Acta Chem. Scand. 1966 20 1885. 122 H. G. Hertz G. Traverso and W. Walter Annalen 1959 625 43; G. Pfister-Gillouzo and N. Lozac’h Bull. Sac. chim. France 1964 3254. la3 G. Pfister-Guillouzo and N. Lozac’h Bull. SUC. chim. France 1963 153; 1964 3252; E. Klinsberg J. Amer. Chem. SOL. 1963 85 3244; C. Portail and J. Vialle Bull. SOC. chim. France 1966 3187; H. Behringer H. Reimann and M. Ruff Angew. Chem. 1960 72,415; R. Pinel Y . Mollier and N. Lozac’h Bull. SUC. chim. France 1967 856; Y . Poirier and N. Lozac’h. ibid. p. 865; G. Duguay H. Quiniou and N. Lozac’h ibid. p. 2763; R. J. S. Beer D. Cartwright and D. Harris Tetrahedron Letters 1967 953; G.Claeson and J. Pederson ihid. p. 3283. 273 Valence-shell Expansion in Sulphur Heterocycles compounds (74) and (75) seems to have little effect on their properties.124 Theoretical calculations have been made by making use of p-models which invoke no-bond ~es0nance.l~~ However Maeda126 feels this type of resonance unlikely and considers the system to be truly bicyclic. He considers the central sulphur atom by virtue of hybridisation to be quadrivalent. The electronic configuration of the sulphur atom is then 3s23ps3d. One p orbital makes a o bond with the central carbon atom the second enters into the n-electron system of the molecule and the third mixes with a d orbital to yield two hybrid pd orbitals which overlap with twop orbitals of the adjacent sulphur atoms to complete the o-framework of the molecule.Such an arrangement he claims would explain both the S-S-C bond angles and the S-S bond lengths. P h f l N ) P h PhfNyN>l?h phwMe MemPh P h m phw P h m P h s-s s-s s s-s s -S (74) m 176) (n) s-s s-s s-s s-s s s-s s (4 (78) Gl (-1 Klin~bergl~~ has recently extended the thiothiophthen system to one containing four sulphur atoms. By allowing compounds (76) and (77) to react the salt (78) is obtained which can be looked upon formally as a cyanine-type structure resonating between the forms (78a) and (78b). X-Ray structural analysis128 indicates that the internal pair of sulphur atoms are less than 3.10 A apart (considerably less than the sum of the van der Waals radii) so that some bonding between these atoms seems to exist. Structures such as (78c) should therefore be included in the resonance hybrid.Selenium analogues of the thiothiophthen system have also been prepared.129 By comparing the different S-S bond distancesI3O in the unsymmetrical com- lB4 J. Derocque and J. Vialle Bull. SOC. chim. France 1967 3079; G. Lang and J. Vialle bid. p. 2865;. H. Behringer and D. Bender Chem. Ber. 1967 100 4027. 126 G. Giacometti and G. Rigatti J. Chem. Phys. 1959 30 1633; E. M. Shustorovich Zhur. obshchci Khim 1959 29,2459. lZ6 K. Maeda Bull. Chem. SOC. Japan 1960,33,1466; 196f,34,785 1 f66. 12' E. Klinsberg J. Heterocyclic Chem. 1966 3 243. 1*8 A. Hordvik Acta Chem. Scand. 1965 19 1253. l** G. Traverso Ann. Chim. (Italy) 1957 47 3 ; M. Sanesi and G. Traverso Chem. Ber. 1960,93,1566; J. H. van den Hende and E. Klinsberg J. Amer. Chem.SOC. 1966,88,5045. lSo A. Hordvik E. Sletten and J. Sletten Acta Chem. Scand. 1966 20 2001. 274 SaImond pound (79) with the equal S-S bond lengths in the symmetrical compound (72) Klinsberg has suggested more of a contribution of the structure (79a) to the resonance hybrid. By similar reasoning he points out that selenium is readily accommodated in this type of system and suggests that structure (80b) provides a greater contribution to the resonance hybrid of the molecule. The original X-ray work on compound (72) has been brought into question by workers who claim131 that thiothiophthens have a typical ‘short’ S-S bond length in the range 2-12-2.22 A and a typical ‘long’ distance (2.47-2.57 A). However comparison of the ‘long’ distance with van der Waals distance between two sulphur atoms still provides evidence for one-bond-no-bond resonance.It has been noted132 that the ‘thioamide’ (81) should be considered as a thiothiophthen and that the ‘amide’ (82) should be considered as having the open-chain structure since their U.V. spectra are related in the manner that appears completely characteristic for the thiophthens and furothiophthen~.~~~ Thus in neutral and acidic solutions the compounds (81) and (82) have U.V. spectra which are related in the same way as the spectra of say compounds (73) and (72). A remarkable feature of the compound (81) is that functionally the amino-group is acidic. The resistance of the anion to further attack by hydroxyl ion has been attributed to structures such as (83) where the d orbitals of sulphur are involved. Even more remarkable is the acidity of the amide (82) which is also stable to further attack by hydroxyl ion.The U.V. spectra of both compounds in alkali are so similar that the structure (84) for the oxygenated ion has been suggested as an important contributor to the hybrid. Another remarkable feature is the accommodation of oxygen in this system without a corresponding change in the spectra. As Klinsberg points out such an easy interchange of oxygen and sulphur in heterocyclic systems is virtually unprecedented and argues most cogently for the involvement of the d orbitals of the central sulphur atom in these compounds. Determination of the S-0 bond lengths must constitute one of the more immediate aims in the study of these anions. S. M. Johnson M. G. Newton I. C. Paul R. J. S. Beer and D. Cartwright Chem. Comm. 1967 11 70. 18* E. Klinsberg J. Org. Chem. 1966 31 3489. H. Behringer and R. Wiedenmann Tetrahedron Letters 1965 3705; H. Behringer M. Ruff and R. Wiedenmann Chem. Ber. 1964 97 1732; H. Behringer and D. Weber ibid. 2567. 275

ISSN:0009-2681    

DOI:10.1039/QR9682200253

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

The Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance Techniques By J. Burgess and M. C. R. Symons DEPARTMENT OF CHEMISTRY UNIVERSITY OF LEICESTER Recently the standard techniques for studying ions in solution have been supplemented by spectroscopic techniques in particular ultraviolet infrared and magnetic resonance methods and in this Review we wish to call attention to some of the ways in which magnetic resonance can shed light on the problems of ionic solvation and interactions between ions in solution. The variables from which information is sought include (i) chemical shifts or g-value variation (ii) the appearance of new resonance lines (iii) variations in spin-spin hyperfine coupling (iv) line widths and shapes. Often these pieces of information are merged. For example different species may be present in a solution which should in principle give rise to different resonances but the exchange of nuclei between these species may be so rapid that only a single merged line is detected.The former situation is described as ‘static’ and will be treated initially even though the bulk of the results may apply to the ‘dynamic’ case. Sometimes chemically interesting information is derived directly from n.m.r. and e.s.r. measurements. At other times a knowledge of the chemistry of the system is required before an understanding of the resonance phenomena can be achieved. Both aspects will be discussed in this Review but mathematical derivations will be omitted and the more physical aspects of relaxation phenomena largely ignored. Details of the instruments and experimental procedures are also omitted.Knowledge of the theory underlying the resonance phenomena will be assumed. Acid-base equilibria are also outside the scope of this Review. No attempt is made to cover the literature fully and our choice of illustrative examples is somewhat arbitrary. Fortunately a recent review by Hinton and Amisl on the applications of n.m.r. to ionic solvation covers recent literature very fully; kinetic aspects of the subject have also been comprehensively discussed.2 Of the techniques under consideration n.m.r. has been the more extensively applied to the study of ions in solution. We therefore devote the first half of the Review to the n.m.r. applications while the second half treats the more limited but nevertheless fruitful applications of e.s.r.J. F. Hinton and E. S. Amis Chem. Rev. 1967,67,367. a A. Loewenstein and T. M. Comer Ber. Bunsengesellschaft Phys. Chem. 1963 67 280; C. S. Johnson ‘Advances in Magnetic Resonance’ vol. 1 Academic Press New York 1965 p. 33. 276 Burgess and Symons 1 Nuclear Magnetic Resonance A. Shifts.-If a nucleus experiences a range of different bonding interactions then separate resonances will be observed over a range of frequencies; but if interchange between these states is rapid a single resonance which is the weighted mean of the individual resonances will be detected. Resonance frequencies are a function of medium effects on the induced diamagnetic or paramagnetic currents of the electrons associated with the nucleus. For example an effective gain of associated electrons will increase the shielding of a nucleus and cause a shift of the resonance to high field.(i) Solvent nuclei general. In an electrolyte solution solvent molecules will be in a variety of environments which may somewhat arbitrarily be divided into primary solvation shells of the ions secondary solvation regions and bulk solvent where solvent molecules are effectively out of range of ionic influences. If exchange of solvent molecules (or atoms) between these various environments were slow (7) > ca. lo4 sec.) five peaks (or multiplets) could be expected in the n.m.r. spectrum of a solvent nucleus. In practice the secondary solvation sphere merges into bulk solvent and solvent exchange between these environments and with primary solvation shells of anions is fast so that these four resonances are time-averaged to a single peak whose shift from the pure solvent resonance reflects the mean effect of the different environments of the primary anion and secondary solvation spheres.Rates of solvent exchange with cation primary solvation shells are also often fast in which case only one averaged resonance will be seen. However in certain instances especially at low temperatures a separate resonance due to solvent bonded to cations may be resolved; for example Mg2+ in methanol at -75°.3 From relative peak areas the cation solvation number can be determined. An alternative method for determining these solvation numbers is from molal shifts (cf. ref. 4). In most instances only a single averaged line is detected and the individual molar ionic shifts are deduced from the overall salt shifts by making some quite arbitrary assignment a popular one being that the shift due to NH,+ is zero.It does not seem to have been realised that a link however tenuous can be forged between these averaged results and the low-temperature resolved lines discussed above. Thus for magnesium perchlorate in methanol the cation shift can be directly estimated and thence the anion shift derived. In this case the shifts are comparable in magnitude but opposite in sign.5 From these results it should be possible by appropriate extrapolations to obtain values for many other ions. However differences between cation shifts or anion shifts can be meaningfully compared and this is done in Figure 1 for ions in watere and ammonia.' J. H. Swinehart and H. Taube J. Chem. Phys. 1962,37 1579.M. Alei and J. A. Jackson J. Chem. Phys. 1964,41 3402. R. N. Butler E. A. Phillpott and M. C. R. Symons Chem. Comm. 1968 371. ti (a) H. G. Hertz and W. Spalthoff 2. Elektrochem. 1959 63 1096; (b) M. S. Bergqvist and E. Forslind Acta Chem. Scand. 1962,16,2069; (c) Z. Luz and G. Yagil J. Phys. Chem. 1966 70 554. ' A. L. Allred and R. N. Wendricks J . Chem. SOC. (A) 1966 778. 277 Study of Ion-Solvent and Ion-lon Interactions by Magnetic Resonance KS- CS? Na+- R b+- -NH;*- Li+- -Br' -F' '70 'H 'H Water Ammonia Figure 1 Ionic molal shifts for IH and 1 7 0 resonances of water and 'H resonances of ammonia in electrolyte solutions. The ionic molal shift for the ammonium ion has been arbitrarily taken as zero in all cases Shoolery and Alder,* and most subsequent workers have discussed shifts for electrolytes in water in terms of two opposing contributions a low-field shift arising from polarisation of water molecules and a high-field shift from hydrogen-bond breaking.Assignation of high-field shifts to structure-breaking is fully consistent with the increasing high-field shift of the proton resonance of pure water as the temperature is raised and with the proton chemical shift of water v a p o ~ r . ~ (The shift for ice protons relative to those of water is unknown because the resonance line is too broad. A recent claim that the line is relatively narrow near 0" has been discredited;1° a small number of protons in an unknown J. N. Shoolery and B. J. Alder J Chem. Phys. 1955 23,805. * W. G. Schneider H. J. Bernstein and J. A. Pople J. Chem. Phys. 1958 28 601; E.R. Malinowski P. S. Knapp and B. Feuer J . Chem. Phys. 1966,45,4274. lo J. Clifford Chem. Comm. 1967 880. 278 Burgess and Symons state give rise to this line which has a frequency close to that from gaseous molecules.) The 170 resonance of water is also shifted to high-field by increasing temperature or by dilution with an inert (non-hydrogen-bonding) solvent.sc Shifts for electrolyte solutions imply that 1 1 electrolytes have a predominantly structurebreaking effect ; the opposite direction of shifts for other electrolytes implies dominance of polarisation. Proton ionic shifts5 (Figure 1) for the halide ions are increasingly to high-field going from fluoride to iodide. This is consistent with the decrease in polarisation as the ionic radius increases but may indicate some increase in structure-breaking also.For the alkaline earth ions (Mg2+ -+ Ba2+) a similar shift to high-field with increasing ionic radius is observed. The greater poIarisation effect of the dipositive alkaline earth cations than of the unipositive alkali metal cations is reflected in the lower-field of the former series. Figure 1 also shows ionic shifts in liquid ammonia solution. These are much smaller than those in aqueous solution and the order of effects of halide ions on solvent proton shifts is reversed. Differences in proton-shift behaviour between these two solvent systems are ascribed to the smaller degree of hydrogen bonding in liquid ammonia? The 1 7 0 n.m.r. spectra of diamagnetic salts in water consist of one line both for the rapidly and slowly exchanging cation cases.If cobalt@) perchlorate is added in most cases the resonance merely shifts but in the cases of the A3+ Ga3+ and Be2+ the 1 7 0 n.m.r. spectrum then consists of a large broad shifted peak and a small residual peak at the original position. [If dysprosium(m) is used instead of cobalt (n) there is still a large shift but little broadening of this peak?] For fast-exchange cations all the water in the system is exchanging rapidly between bulk solvent and the solvation shells of the cobalt(I1) and the diamagnetic cations giving an average paramagnetic shift for all the water. But for A13+ Ga3+ and Be2+ the slowly exchanging water in their primary hydration spheres does not come directly under the influence of the paramagnetic ion hence the small residual peak. Solvation numbers for AIJ+ and Be2+ have been estimated from peak areas in these experiments.lf Again it is not yet possible to determine absolute values for cation or anion shifts but as for proton shifts 170 molar shifts can be arbitrarily separated into self-consistent sets of ionic shiftsaC (Figure 1).It is interesting to compare trends for 1 7 0 and for lH shifts for certain series of ions. For instance the order of halide shifts is exactly reversed in the two cases presumably because of the differences in hydrogen-bonding interaction. Ion-water interactions appear to be more important than structure-breaking effects for 170 shifts. (ii) Solvent nuclei mixed solvents. For slow-exchange diamagnetic cations in binary solvent mixtures it should be possible to detect separate resonances arising from both solvents co-ordinated to the cations apart from the resonances for the bulk solvent.In solutions of aluminium chloride in for example aqueous dimethylformamidc dimethyl sulphoxide,12 and N-methyla~etamide,~~ signals l1 R. E. Connick and D. Fiat J. Chem. Phys. 1963,39 1349. la A. Fratiello and D. P. Miller Mol. Phys. 1966 11 37. la J. F. Hinton and E. S. Amis Chern. Comm. 1967 100. 279 Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance from the organic solvent bonded to the A13+ cation have been detected. Moreover in aluminium perchlorate solution in aqueous acetonitrile a series of peaks in the proton n.m.r. spectrum can convincingly be assigned to the series of solvated A13+ cations containing 1-6 molecules of primary solvating water.14 For soh tions of fast-exchange cations in some mixed solvents for instance aqueous pyridine15 and aqueous N-rnethylformamide,l2 variation of proton shift with solvent composition indicates solvation by both components.But in other cases such as aqueous alcohols12 a lack of shift variation over only a limited range of solvent composition does not rule out solvation by the organic component (Figure 2). Recent measurements in aqueous dioxan indicate some Non-aqueous component (mole fraction) Figure 2 Hypothetical variation of solvent chemical shgt due to solute with mole fraction composition of a binary aqueous solvent mixture. Such a plot could arise from preferential solvation by water; if observations are made only in the region A to B then no indication of solvation by the non-aqueous component is obtained whereas such solvation is indicated by observations over the whole range A 10 C solvation of the perchlorate anion by the dioxan.16 The relative magnitudes of shifts for different protons in the organic solvating molecule can yield information about bonding to the cation.In aqueous pyridine bigger displace- ments of the 18- than of the y-proton lines are consistent with the expected solvation at the nitrogen atom and consequent polarisation.ls In aqueous dimethylformamide the carbonyl proton resonance is displaced more than that for the methyl protons by addition of salts. This has been interpreted in terms of preferential interaction between the cations and the oxygen atom rather than the nitrogen atom of the dimethylformamide.12 (iii) Solvent nuclei solutions containing paramagnetic ions.Very much larger solvent shifts can often be detected when paramagnetic ions are involved. When there is hyperfine coupling between a nucleus and an unpaired electron l4 L. D. Supran and N. Sheppard Chem. Comm. 1967 832. l5 A. Fratiello and E. G. Christie Trans. Faraday SOC. 1965 61 306. l6 J. F. Hinton L. S. McDowell and E. S. Amis Chem. Comm. 1966 776. 280 Burgess and Symons the nuclear resonance is split into a doublet but because of the very large magnetic field associated with unpaired electrons these lines are so far removed from a normal resonance that they are not detected. If however reorientation of the electron in the magnetic field is rapid as is the case for example for Co(OH&,2+ a weighted averaged line will be detected shifted from the normal solvent resonance because of the Boltzmann distribution of the electron spin states.Such shifts are called contact or Knight shifts; as they are very large compared with normal chemical shifts they can be of great use in separating broad overlapping resonances (see e.g. p. 279 and refs. 4 and 11). The situation is otherwise very similar to that discussed above for diamagnetic ions. If solvent molecules have a long lifetime in contact with a rapidly relaxing paramagnetic ion then their nuclear resonance will be greatly shifted up- or down-field depending on the sign of the effective spin-density at this nucleus (contact shift) or upon the anisotropy of the g-tensor (pseudo-contact shift). From the shift it is often possible to calculate hyperfine coupling constants but care must be taken to make proper allowance for the pseudo-contact shift.However irrespective of the source of the shift the area under the resonance line can give a good estimate of the solvation number for example of Coz+ in methanol at low temperatures,17 and the effect of temperature on line-widths can give detailed kinetic information. Moreover detailed information about solvation in mixed solvents can be obtained; for example peaks in the low- temperature proton n.m.r. spectrum of Co2+ in methanol containing traces of water can be assigned1' to water and cis- and trans-methanol molecules in [CO(M~OH),(OH&]~+. If the electron spin relaxation is too slow (e.g. for Mn2+) no nuclear resonance will be detected from bound solvents and only the bulk solvent line will be detected. As the rate of exchange of co-ordinated solvent increases the usual line broadening and shift towards the weighted-mean frequency occurs until a single shifted line results (see Figure 4).This applies irrespective of the rate of electron spin relaxation. In contrast with the situation for diamagnetic ions the overall shift can now safely be assigned to the effect of the paramagnetic cation only since Knight shifts are so large. Specific examples are cited in ref. 1; further discussion is deferred until line-widths in these systems are considered. (iv) Solute nuclei. There are nuclei of spin 2 3 for all the alkali metals and all the halogens so it is possible to study the whole range of alkali halides in solution. Magnitudes of chemical shifts increase greatly from lithium1* through to caesium,1° and similarly increase with increasing size for the halogens (Table).However the sensitivity of chemical shift to solvent decreases rapidly as ionic size increases; lithiumlS and fluoridez0 resonances are very solvent- sensitive caesium20 practically unaffected. l7 Z. Luz and S. Meiboom J. Chem. Phys. 1964,40 1058. l8 G. E. Maciel J. K. Hancock L. F. Lafferty P. A. Mueller and W. K. Musker Inorg. Chem. 1966,5 554. lo C. Deverell and R. E. Richards Mol Phys. 1966 10 551. 2o A. Carrington F. Dravnicks and M. C. R. Symons Mol. Phys. 1960 3 174. 281 Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance Chemical shifrs in p.p.m. of alkali-metal nuclei in 3-molal aqueous solutions of alkali halides (from ref. 19) CSI - 57 CSI - 57 RbI - 21 CsBr - 36 KI -6 CSCl - 23 NaI -1 In aqueous solutions of alkali halides the chemical shift of the alkali metal increases with concentration.Up to about 4 molal concentrations the shift varies linearly with activity implying that interionic interactions are important in determining the magnitude of the shift. Indeed the lack of solvent-sensitivity of the caesium resonance compared with the large shifts produced by varying the nature and concentration of anionslg* 2o underlines the significance of these interionic interactions. The increasing magnitude of chemical shift with cation or anion size rules out a simple electrostatic explanation of these shifts. A more likely rationalisation of the observed shifts emerges from comparison of alkali halide solutions and the respective crystals illustrated for rubidium bromide in Figure 3.RbBr (mole fraction) Figure 3 Variation of *'Rb chemical shift with mole fraction in aqueous rubidium brvmide solution. This Figure relates this variation to the chemical shijit value in the crystal and variarion of chemical shift with pressure in the crystal ( ) experimental results; (- - - - - -) arbitrary interpolation between solution and crystal points 282 Burgess and Symons The magnitude and variation of the alkali and halogen shifts in the crystals under varying pressures have been satisfactorily interpretedz1 in terms of the Kondo-Yamashita theory,22 in which chemical shifts are controlled by overlap repulsive forces between the ions. Increasing proximity of ions leads to increased overlap and repulsion and thus to an increasingly low-field shift.In aqueous solution the resonances shift increasingly downfield as the concentration increases and interionic interactions increase. The plot of shift against mole fraction tends towards the shift for the crystal; moreover this graph tends in the same direction as that of the shift against pressure trend for the crystal (Figure 3). The Kondo-Yamashita theory thus predicts qualitatively the correct direction and order of magnitude of shifts for a given alkali halide and the observed largest shifts for largest ions. The anomalous position of lithium among the cations can be understood in terms of tenacious solvation inhibiting interionic interactions in the case of this very small cation. Concentration-dependence of 203Tl and 205Tl chemical shifts in aqueous solutions of thallium(1) and thallium(m) can be understood in terms of extensive ion-pairing or complex for ma ti or^.^^-^^ The continuous gradation from effects due to fortuitous encounters between ions through preferential ion-pairing to complex formation thus leads us beyond our brief into complex chemistry where the application of n.m.r.to a very wide range of solvent and solute nuclei has proved of great value in elucidating problems of composition structure and kinetics.l In mixed aqueous solutions of caesium fluoride 19F shifts correlate linearly with mole fraction composition for many solvent mixtures such as aqueous methanol and aqueous dimethylformamide. This behaviour implies smoothly changing statistically controlled variation of the fluoride solvation shell as solvent composition varies.20 As in the case of solvent proton shifts (see Figure 2) lack of variation of 19F shift with composition over a limited range as in aqueous methyl cyanide and aqueous isopropyl alcohol does not necessarily indicate lack of solvation of fluoride by the organic component.Solvent composition dependence of 7Li shifts1* has in several cases indicated solvation by both components of a binary solvent mixture. Lithium ions are particularly satisfactory for such studies since solvent effects on shifts are relatively large and correlate well with Kosower’s solvent polarity factor 2?6 Considerable information about interactions of tetra-alkylammonium anilinium and similar salts in solution has been obtained by studying the proton resonances of the cations. The differences in shift behaviour of the various protons for anilinium salts in non-polar solvents27 and for quaternary ammonium salts 21 R.Baron J. Chem. Phys. 1963 38 173; D. Ikenberry and T. P. Das J. Chem. Phys. 1966,45 1361 and refs. therein. 22 J. Kondo and J. Yamashita J. Phys. and Chem. Solids 1959 10,245. 23 H. S. Gutowsky and B. R. McGarvey Phys. Rev. 1953,91,81. 24 R. Freeman R. P. H. Gasser R. E. Richards and D. H. Wheeler Mol. Phys. 1959 2 75. 25 R. Freeman R. P. H. Gasser and R. E. Richards MoZ. Phys. 1959,2,301. 26 E. M. Kosower J. Amer. Chem. SOC. 1958 80 3253. 27 G. Fraenkel J. Chem. Phys. 1963,39 1614; G. Fraenkel and J. P. Kim J. Amer. Chem. SOC. 1966 88,4203. 2 283 Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance in nitrobenzene,* indicate significant ion-association with the anion in close proximity to the nitrogen atom of the cation.Also the observation that the 1-methylene and ortho-ring proton shifts of 1,4-diethylpyridinium salts are a function of the nature and concentration of anion in deuteriochloroform solution demonstrates significant ion association in this system.29 B. Line-widths and Shapes.-A large number of factors govern the width of an n.m.r. line which is usually treated in terms of two time constants T, the spin- lattice or longitudinal relaxation time and T, the spin-spin or transverse relaxation time. T is a measure of the interaction between spin energy and thermal motion of the molecule and its surroundings; it governs the degree of saturation. When this interaction is very efficient leading to rapid spin inversion each spin state has a very short lifetime and there is uncertainty-principle broadening of the resonance line.On the other hand T2 is associated with variations in the relative energies of the spin states. In both cases it is the time-dependence of some interaction acting on the spins that causes the relaxation. One important factor controlling line-widths will therefore be a fluctuation of position and orientation of solvent molecules relative to an ion or other solvent molecules. This coupling may be direct through magnetic fields or indirect via electric fields if the nucleus has a quadrupole moment. Other factors which can contribute are the anistropy of the chemical shift of a tumbling molecule spin-rotation coupling and the effect of paramagnetic materials. Anisotropy is particularly important in 19F resonances.Coupling to magnetic fields generated by rotating molecules is important when rotation is fairly free in a liquid. The greater the frequency of collisions the smaller the correlation time and the less effective is this coupling as a mechanism for relaxation. Thus we can say that motions having any frequency from the resonance frequency down to near zero will affect the line-width through T2 but only those close to resonance will involve T,. (i) Solvent nuclei. In principle considerable information relating directly tc the correlation times of the solvent molecules in the absence and presence of ions can be extracted from resonance studies of solvent nuclei. The situation has been reviewed in depth by Hertz.30 We turn to some more specifically chemical aspects by considering an equilibrium in which a given nucleus is transferred between environments in which it resonates at different frequencies.For slow rates of exchange between two sites two narrow lines are detected. As this rate is increased for example by an increase in temperature these individual lines broaden and eventually merge to a single broad line centred on the weighted-mean position. Further increase in rate gives rise to line-narrowing (Figure 4). This behaviour is well exemplified by the case of cobalt (a) perchlorate in methano1.l' The effect of temperature on shifts and line-widths is summarised in Figure 5. 28 R. L. Buckson and S. G. Smith J. Phys. Chem. 1964 68 1875. 29 R. J. Chuck and E. W. Randall Spectrochim. Acfu 1966 22 221. 30 H. G. Hertz in 'Progress in Nuclear Magnetic Resonance Spectroscopy' vol.3 ed. J. W. Emsley J. Feeney and L. H. Sutcliffe Pergamon Oxford 1967 p. 159. 284 Burgess and Symons I 2 3 4 Figure 4 Variation of ‘H n.m.r. spectra with temperature for a solution of a cation whose rate of solvent exchange varies from ‘slow’ (i.e. T+ > lo-* sec.) (1) through intermediate (2 and 3) to fast (4) as the temperature rises Figure 5 Variation of proton chemical sh$t for bulk solvent [u (s)] for solvent bound to cation [u (c)] and variation of line width ( A ) fur either with reciprocal temperature (1/T) for a solution where rates of solvent exchange at the cation vary from much slower to much faster than the n m r . frequency The points A to F correspond to identical temperatures in each diagram. The section A to B corresponds to exchange rates much lower than the n.m.r.frequency; from B to C both lines broaden; C to D the lines overlap and the peaks approach to coalesce at D; D to E involves exchange narrowing; and beyond E rates of exchange are much faster than the n.m.r. frequency. The steady upfield trend with rising temperature for cation-solvent shifts in the region A to B and for the averaged solvent shifts E to F arises from variation of the Boltzmann distribution between the two spin states. Kinetic data and Arrhenius parameters can be obtained from the section B to E Two pieces of chemical information can be extracted. These are the solvation number of the cation at low temperatures and from the region R - + E in Figure 5 rates and thence activation parameters for the solvent exchange.For cobalt(rr) in methanol the parallel behaviour of methyl and hydroxyl proton resonances indicates exchange of whole methanol molecules; in contrast the different behaviour of the two proton resonances for methanolic solutions of 285 Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance chromium(m) implies that only the hydroxyl proton exchanges rapidly the remainder of the methanol molecule remaining as expected M y bound to the cation. Indeed the n.m.r. study showed that proton exchange is faster for methanol co-ordinated to Cr3+ than for free methanol illustrating the electron- withdrawing properties of the cation. This chromium-methanol case illustrates the general desirability of studying solvent exchange processes by monitoring the ligating nucleus e.g.170 rather than lH for water or for alcohols. The relatively slow electronic relaxation for manganese(@ which precludes observation of bound solvent peaks in solutions of this cation (see p. 281) is with fast exchange of co-ordinated solvent molecules,31 the reason for the possible success of Swift and Sayre’s method for determining hydration numbers?2 Determination of solvation numbers by peak-area or molal-shift methods mentioned earlier is straightforward for slowexchange cations but these methods are not applicable to the majority of cations whose rates of solvent exchange are too fast for separate co-ordinated solvent n.m.r. signals to be detectable. Swift and Sayre treat linewidths of averaged proton signals from water containing a diamagnetic cation of unknown hydration number with the Mn2+ ion acting as a probe.The method depends on modifying the association time of Mn2+ and water and hence the proton resonance line-width by addition of the competing cation. In view of the sensitivity of line-widths in Mn2+ solutions to added ions and to ionic-strength variations it has not proved possible to determine hydration numbers of cations directly from solutions containing only Mn2+ and the cation of interest but an empirical relationship has been obtained for relating hydration numbers of cations to the known hydration numbers of A13+ (6.0) and Be2+ (4-0). Hydration numbers so determined are lower than expected for instance for Ca2f (4-3) Cd2+ (4.6); only for larger cations is the octahedral value of 6.0 approached [Ba2+ (5.7)]. The marked similarity of values for cations of similar size (e.g.Ca2+ Cd2+; Ba2+ Pb2f) and the increase with increasing ionic radius are considered to arise from a balance between six-co-ordination of cations and four-co-ordination in water itself. Smaller ions may be more readily accommodated geometrically into the water structure. Despite objections to the physical basis of the method33 the empirical relationship gives consistent re~ults.3~ It may be noted that recent ultrasonic35 and ‘Li n.m.r.36 studies support Swift and Sayre’s values. The surprising values are those of zero for both H+ and NH,+ cations; those values are likely to be accidental consequences of the definition of hydration number and the fact that protons are common to cation and solvent. This method of determining solvation numbers has been extended from aqueous solution to liquid ammonia in which the Ni2+ cation proves to be a suitable probe.37 31 T.J. Swift and R. E. Connick J. Chem. Phys. 1962 37 307. 32 T. J. Swift and W. G. Sayre J. Chem. Phys. 1966,44,3567. 33 S . Meiboom J . Chem. Phys. 1967 46,410. 31 T. J. Swift and W. G. Sayre J. Chem. Phys. 1967,46,411. 36 D. S. Allam and W. H. Lee J. Chem. Soc. (A) 1966 5 ; 1966,426. 86 J. W. Akitt and A. J. Downs Chem. Comm. 1966 222. T. J. Swift and H. H. Lo J. Amer. Chem. SOC. 1966 88 2994. 286 Burgess and Symons (ii) Solute nuclei. The discussion so far has covered n.m.r. spectra of solvent nuclei. The study of line-widths for solute nuclei has been less widely employed but can give pertinent information about solvation. Close approach of cation to anion as in ion-association or complex formation reduces the symmetry of the environments of the ions and thereby modifies relaxation times and line-widths.Much information on ion-solvent and ion-ion interactions in aqueous solution of alkali halides has been obtained from n.m.r. resonances of alkali metal and halogen nuclei.38 Ion-pairing and complex formation in systems as diverse as Na+ with keto- and hydroxy-acid~~~ In3+ and simple oxyanions,4O and Zn2+ Cd2+ or Hg2+ with bromide or iodide,4l have been monitored from line-width variations. From extensive studies of solutions of R4M& (M = By N As etc.) salts we shall mention for illustration the cases of Et4As+ and Et,MeAs+ salts in water and chloroform!2 The unsymmetrical Et,MeAs+ cation gives sharp proton resonances in water and in chloroform but symrnetrical Et4As+ gives sharp lines in chloroform broad in water.In chloroform ion-pairing is extensive which results in an asymmetric environment of the Et4As+ cation promoting relaxation and giving sharp lines. In water the symmetry of solvated Et4As+ is not disturbed by ion-pairing relaxation is slow and proton n.m.r. lines are broad. A more indirect approach to ion-association has been to study the effect of gegen-ions on rates of ferrocyanide-ferricyanide electron exchange determined by I4N line-widths and to correlate rate variations with changes in concentrations and reactivities of ion-~airs.4~ A link between n.m.r. and e.s.r. studies (p. 298) is provided by a 'Li n.m.r. study of the paramagnetic lithium salt of fluorenone dissolved in tetrahydro- f ~ r a n .4 ~ Line-broadening with increasing temperature and increasing anion concentration can be correlated with increasing rate of exchange of Li+ between the Knight-shifted ion-pair and the free (solvated) cation. C. Spin-Spin Coupling.-Spin-spin interaction between two magnetic nuclei occurs in many polyatomic ions for example in BF4- BH4- AsEt4+ and should in principle be affected by the environment. However variation in spin-spin coupling constants has been very little used as a probe for ion-ion or ion-solvent interactions. The tetrafluoroborate anion has been the most studied but even here it has proved difficult to explain the differences in behaviour observed for the sodium ammonium and silver salts in a variety of ~olvents.4~ It is noteworthy that spin-spin coupling constants correlate linearly with 19F chemical shifts.38 E.g. C. Deverell D. J. Frost and R. E. Richards Mol. Phys. 1965,9 565. 40 T. H. Cannon and R. E. Richards Trans. Faraday SOC. 1966,62 1378. r a A. G. Massey E. W. Randall and D. Shaw Spectrochim. Acta 1964 20 379. 43 M. Shporer G. Ron A. Loewenstein and G. Navon Znorg. Chem. 1965 4 361; A. Loewenstein and G. Ron Znorg. Chem. 1967 6 1604. 44 G. W. Canters H. van Willigen and E. de Boer Chem. Comm. 1967 566. pb R. Haque and L. W. Reeves J . Phys. Chem. 1966,70,2753. 0. Jardetzky and J. E. Wertz J. Amer. Chem. SOC. 1960 82 318. H. G. Hertz 2. Elektrochem. 1961,65,36. 287 Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance 2 EIectron Spin Resonance Perhaps the major chemical difference between the techniques of n.m.r.and e.s.r. as applied to the study of ionic solvation is that in the latter at least one component must be specifically paramagnetic. Apart from the transition-metal ions these are few and often very unstable. Nevertheless some of the most informative spectra have come from solutions of very unstable species such as the alkali-metal salts of m-dinitrobenzene. In general e.s.r. cannot be used to ‘look at’ solvent molecules although a solvent molecule bonded strongly to a paramagnetic ion can contribute a hyperfine splitting to the spectrum In practice such couplings are not generally resolved because of rapid exchange and n.m.r. studies of solvent nuclei then give more information. We start by discussing briefly the information that can be derived from studying g-shifts.These may be compared with the chemical shifts of n.m.r. but are in general less useful for solvation studies. It is the hyperfine coupling to paramagnetic nuclei discussed on p. 289 which is most informative in complete contrast with n.m.r. where spin-spin multiplets have been of relatively little direct use. One reason for this is that the isotropic hyperfine coupling constants derived from solution spectra are strongly dependent upon the spin-density on the nucleus concerned or on adjacent nuclei. Strong interaction can alter this spin-density and this is directly reflected in the hyperfine splitting. A result of outstanding significance in the field of ion-pairing was the detection of hyperfine multiplets derived from coupling to the nuclei of diamagnetic cations in solutions of paramagnetic anions.46 This is elaborated in section 2B which is based in part upon a recent more extensive review of this aspect of the problem.47 As with n.m.r.there is a lot of useful kinetic information hidden within the widths of lines and again although special saturation techniques can be used,48 most work has been carried out on slow-passage spectra. As with n.m.r. we distinguish two major sources of line broadening one involving only the tumbling motion of the ion and its neighbouring solvent molecules and the other involving chemical reactions. In many instances especially when there are large anisotropies of the magnetic parameters or when the spin-state of the ion is greater than 3 line- widths may be so great that no e.s.r. spectrum can be detected. It is just this situation that gives rise to a Knight-shifted n.m.r.line and hence the two techniques are complementary to each other in this sense. Indeed if e.s.r. lines are narrow it may be useful to broaden them deliberately in order to obtain n.m.r. information. Concentrated solutions where spin-exchange is rapid are often suitable for this purpose. 46 F. C. Adam and S. I. Weissman J. Amer. Chem. SOC. 1958 80 1518. 47 M. C. R. Symons 3. Phys. Chem. 1967 71 172. 48 D. C. McCain and R. J. Myers J. Phys. Chem. 1967 71 192. 288 Burgess and Symons A. Shifts in g-Values.-Since the g-values for organic radicals are close to 2.0023 solvent shifts have rarely been detected. One example for which small g-shifts were found on ion-pairing is that of the anion of benz~phenone.~~ The effect of the cation close to the oxygen is to reduce the spin-density thereon.Since the small positive deviation from the free-spin value is primarily caused by spin on oxygen this has the effect of reducing the g-value. Other examples of small g-shifts of this sort are given in Table 2 of ref. 47. A similar qualitative argument can be used to explain the shift from 2.00469 to 2.00541 on going from water to dimethyl sulphoxide observed by Zandstra.60 Alternatively one can say that hydrogen-bonding stabilises the non-bonding electrons relative to the 7r electrons thus increasing the n -+7r* energy gap. The presence of low-lying excited states causes the range of g-values for transition-metal complexes to be far more extensive and if one or more of the ligands is labile then varying the solvent will often result in ligand replacement and a consequent shift in the average g-value.In such a case either a set of lines from the individual complexes or a single averaged line will be detected. The widths of such lines will give information about the rates of ligand replacement in exactly the way obtained for chemically shifted n.m.r. lines. However the actual g-values are a rather complex property of the nature of the bonding to the transition-metal ion and are not of direct significance to the problem of solvation as discussed here. B. Changes in Hyperhe Coupling Constants.-@ Nuclei within the paramagnetic ion. These can arise from changes in the spin-density distribution induced by changes in solvation or from changes in the shape of the ion. For transition- metal complexes the complex itself is changed when one ligand displaces another and hence there can be quite drastic changes in the magnetic parameters .The first effect is generally quite small but can become dramatically large when the symmetry of the molecule is changed by the interaction. This occurs frequently on ion-pair formation when there is more than one site of high charge-density in the paramagnetic ion. Thus p-benzosemiquinones tend to hold a cation close to one oxygen at a time and this causes a drift in the charge and spin densities. The effect is very much more marked for rn-dinitrobenzene anions. Examples of these phenomena are now discussed in more detail. Serniquinones. Three different nuclei can be monitored IH 13C or 170 and in the case of the parent ion benzosemiquinone all have been investigated in a range of s ~ l v e n t s .~ l - ~ ~ The protons are remarkable for their relative insensitivity which is apparently carried right through to the diprotonated cation in sulphuric acid.54 This is neither true of the carbonyl-carbon isotropic 40 A. M. Reddoch J. Chem. Phys. 1965,43 3411. 50 P. J. Zandstra J. Chem. Phys. 1964 41 3655. 51 E. W. Stone and A. H. Maki J. Chem. Phys. 1962,36 1944. 52 E. W. Stone and A. H. Maki J. Amer. Chem. SOC. 1965 87,454. 53 W. M. Gulick and D. H. Geska J. Amer. Chem. SOC. 1966 88 4119. 64 J. R. Bolton and A. Carrington Proc. Chem. SOC. 1961 385. 289 Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance coupling nor of the 1 7 0 coupling which change in a complementary fashion. These changes55 swnmarised in Figure 6 are qualitatively in good accord with expectation.Small strongly interacting solvent molecuIes tend to pull negative charge on to the oxygen atoms thus lowering the spin-density on oxygen and increasing it on carbonyl carbon. Since the isotropic coupling to oxygen may be thought of as being made up from a positive contribution from spin on oxygen and a negative one from spin on carbon a shift in the spin-density from oxygen to carbon is additive in the overall changes in the hyperfine coupling constants. 0 Figure 6 1 7 0 Coupling constant of p-benzosemiquinone plotted against the corresponding ~arbonyl-~~C coupling constant in various solvent mixtures When a proton is attached to one of the oxygen atoms the pattern of hyperfine lines changes from a group of five to two groups of three lines as is to be expected from the change in symmetry.These monoprotonated neutral species have e.s.r. spectra which are quite sensitive to the nature of the ~ o l v e n t . ~ ~ - ~ ~ More pertinent to this Review is the fact that lithium salts in ‘poor’ solvents induce a similar asymmetry although differentiation between the two halves of the molecule is much less marked.59 This observation enables us to say 65 J. Oakes and M. C. R. Symons Trans. Faraday SOC. in the press. 66 T. A. Claxton T. E. Gough and M. C. R. Symons Trans. Faraday Soc. 1966 62 279. 67 T. E. Gough Trans. Faraday SOC. 1966 62,2321. 68 T. A. Claxton J. Oakes and M. C. R. Symons Trans. Faraday SOC. 1967 63,2125. T. A. Claxton J. Oakes and M. C. R. Symons Trans. Furaday Soc. 1968 64 596. 290 Burgess and Symons 9.5 tfo 90- 8.5 conclusively not only that ion-pairs are formed but that the cation is held closely to one oxygen at a time.The situation for larger cations is even more intriguing. At low temperatures similar asymmetric species are detected but as the temperature is raised lines broaden and merge to give spectra with alternating line-widths. This alternation is ascribed to a migration of the cation between the two equivalent sites and is discussed in detail on p. 298. A similar asymmetry induced by solvent molecules has not been detected in pure or mixed solvents despite the expectation that especially in mixed solvents there will be momentary asymmetry. Evidently it is normally too imprecise and the fluctuations are too rapid to be 'seen' in the e.s.r. specta.This is probably not the case for rn-dinitrobenzene anions as we shall see later. In order to gauge the influence of different solvents it is helpful to use some device for linking the diverse results from a wide range of techniques. This can be done by utilising the Z-value or ET scales which are based on shifts in ultraviolet spectra.60 Once again [cf. section IA (iv)] there is often a fairly good linear correlation despite the widely different factors involved. For example Rassar found a fair correlation between 2 and the 14N hyperfine coupling constant of (Ph),NO in a range of solvents.61 Results for the hyperfine coupling to 1 7 0 in benzosemiquinone anions55 correlated even better with 2 as is shown - - - Figure 7 Dependence of the 170 coupling constant of p-benzosemiquinone upon the Z value of the solvent 00 C.Reichardt Angew. Chem. Internar. Edn. 1965 4,29. R. Briere H. Lemaire and A. Rassat Bull. SOC. chim. France 1965 3273. 291 Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance P Q . -0 -3 The low sensitivity for a~ in both benzosemiquinone and durosemiquinone anions on diprotonation fits in with the very minor changes induced in these parameters on changing the solvent. In marked contrast diprotonation of 2,6- dimethylbenzosemiquinone results in a major change in both aH(methy1) and aH(ring) which suggests that the proton couplings for this anion might be strongly dependent upon the solvent. Further it has been found58 that mono- protonation occurs only at the unhindered oxygen and when ion-pairs are formed with alkali-metal cations these spend most of their time at theunhindered end of the anion.59 This suggests that interaction with solvent molecules might be somewhat stronger at the unhindered end although since both ends must interact such differentiation will be relatively small.Both such trends will operate in the sense of increasing aH(methy1) and decreasing aH(ring) on increasing the solvating power of the solvent and in fact quite large changes are detected.55 That more than one factor is involved is suggested by the plot Of [&(methyl) - aH(ring)] against (Figure 8 ) which is a very poor correlation in comparison with the trends in the 170 hyperfine coupling for benzosemiquinone. 0 0 0 0 0 0 0 8 0 Figure 8 Dependence of A A [ a ~ ( ~ e ) - a~(ring)] of 2 6-dimethyl-benzosemiquinone on solvent Z value Aromatic nitro-anions.The hyperfine coupling to 14N in nitrobenzene anions is extremely sensitive to the nature of the environment and again there is a rough correlation with the Z-value of the solvent. Solid-state studies strongly suggest that in contrast with the neutral molecule the nitro-group in the anion 292 Burgess and Symons is slightly pyramidal the more so the greater the spin-density on the group.62 There are indeed good theoretical reasons for this and any shift of spin density into the group will result in a greatly magnified increase in the 14N coupling because on bonding there is a direct admixture of 2s-character into the orbital of the unpaired electron which will add to the normal spin-polarisation term.63 It is interesting that in the limiting case of nitroalkane anions there is practically no solvent effect.This implies that the 14N coupling in the pyramidal RN0,- group is not very sensitive to small changes in spin-density within the NOz group itself. Both the rn- and p-dinitrobenzene anions have provided a wealth of information with respect both to solvation and to ion-pair formation. A key factor seems to be that their structures especially that of the rneta-anion are very readily distortable in a sense that may be described as ‘all or nothing’. Thus if a cation say becomes associated with one of the nitro-groups the 14N coupling involving this group becomes double the original value whilst that of the latter tends to zero. One possible explanation would be that the unpaired electron in the unperturbed ion is in a degenerate orbital and that the asymmetric perturbation lifts the degeneracy ; but whatever may be the reason for the phenomenon the meta-anion is certainly a very useful ‘probe’ for studying solvation.The situation is then comparable with that for the semiquinones ; strongly held cations such as Li+ remain at one end long enough for the asymmetric species to be detected (the ‘static’ case) but most other cations jump from one nitro-group to the other and this gives rise to a line-width alternation. However on cooling the ‘static’ situation can again be detected.64 In addition the normal gain or loss of spin-density shows up as trends in aN and plots of their shifts against the solvent 2-values give a fair correlation. One very important factor has emerged however which did not show up in the study of semiquinones.As we have stressed absence of hyperfine coupling to the cations cannot be correlated with absence of ion-pairs. Since at best this coupling is very small (Table 1 of ref. 47) we expect that only contact ion-pairs will show up in this way and that any form of solvent-separated ion-pair will not give a detectable coupling. In many cases especially when ‘good’ solvents are used the radical- anions under consideration are generated electrolytically by use of a large excess (ca. 0 . 1 ~ ) of a tetra-alkylammonium salt. It has been widely accepted that these large cations have little or no tendency to form ion-pairs and this has received support from the observation that aN for p-chloronitrobenzene anions was independent of the concentration of R4N+ although it changed markedly when other electrolytes were added.65 6a W.M. FOX J. M. Gross and M. C. R. Symons J. Chem. Soc. (A) 1966,448. 6s P. W. Atkins and M. C. R. Symons ‘Structure of Inorganic Radicals’ Elsevier Amsterdam 1967. 64 T. A. Claxton W. M. Fox and M. C. R. Symons Trans. Faraday Soc. 1967,63 2570. e6 T. Kitagawa T. Layoff and R. N. Adams Analyt. Chem. 1964,36,925. 293 Stua'y of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance However aN for p-dinitrobenzene anions in alcoholic solutions is very dependent upon the concentration of alkylammonium salts,65 66 although it is independent of the concentration of sodium salts except in the very dilute region (Figure 9). This means that R4N+ readily forms ion-pairs even in 'good' solvents such as methanol.The reason for the marked shift can be thought of as a loss of hydrogen-bonded solvent rather than as a perturbation by R4N+ and the lack of shift in such solvents as dimethylformamide arises because the two environments are now quite similar. I . I I I 0 0.01 0.02 0-03 0.04 Mole fraction of added salt Figure 9 Dependence of a N for p-dinitrobenzene on the concentration of added salt A NaOEt; B Bun,NBr Mixedsolvents. A considerable effort has been directed towards the interpreta- tion of trends in hyperfine coupling constants on changing from one solvent to another. A typical shift is given in Figure 10 for nitrobenzene anions,67 and the marked curvature indicating preferential solvation by water is typical of such plots. It can be compared with very similar trends observed for the maximum of the first absorption band of iodide ion in this solvent mixture.67 Sometimes a reasonable fit for such curves can be obtained from simplified models involving equilibria such as A- S1+ Sz + A- Sg + S1 66 J.Oakes and M. C. R. Symons Chem. Comm. 1968 294. 67 M. J. Blandamer and M. C. R. Symons Internat. Symp. Solvation Phenomena Calgary 1963. 294 Burgess and Symons to tT= I 2 14 I 0.5 Methyl cyanide (mole traction) Figure 10 Dependence of the nitrogen coupling constant aN for nitrobenzene anion on solvent composition for mixtures of methyl cyanide and water or more complex versions which take the presence of two solvation sites into account.52 68 However the agreement is often very poor and since alkyl- ammonium ions are usually present in high concentration the shifts observed must often be a rather complex mixture of solvation and ion-pairing effects.One result of interest to those concerned with the special behaviour of water is the rather complex shift behaviour of aH(ring) and aH(methyi) for 2,6-dimethylsemiquinone ions in water on the addition of t-butyl alcohol.69 (ii) To solvent nuclei. As we have stressed this is inevitably a minor section but we wish to mention two phenomena which may point the way to future work. One is that when hydrogen atoms are trapped in a matrix of xenon strong hyperfine coupling to six of the surrounding xenon nuclei was dete~ted.~' The other is that the e.s.r. spectra of frozen solutions of cobalt phthalocyanine in various amine solvents gives clear indication of 14N hyperfine coupling involving two co-ordinated solvent These well-defined results show that detailed structural information can be obtained in favourable circumstances.It seems likely that solid-state studies will help because the 68 J. Gendell J. H. Freed and G. K. Fraenkel J. Chem. Phys. 1962,37,2832. 6g M. J. Blandamer D. E. Clarke T. A. Claxton M. F. Fox N. J. Hidden J. Oakes M. C. R. Symons G. S. P. Verma and M. J. Wootten Chem. Comm. 1967,273. 70 J. M. Assour J. Amer. Chem. SOC. 1965,87,4701. 295 Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance rapid exchange of solvent molecules that normally results in an exchange- narrowed single line can be eliminated. The chance of actually resolving hyperfine coupling to neighbouring nuclei will be small but under favourable circumstances it seems that E.N.D.O.R.techniques will provide considerable information about solvent molecules in contact with paramagnetic ions which will complement n.m.r. results from the liquid phase. However great care has to be observed when trying to obtain frozen solutions because of the dangers of phase-separation or clustering which are both hard to detect and to avoid and which can seriously modify spectra. Further hyperhe parameters are frequently temperature-sensitive’l and so extrapolation has many pitfalls. C. Relaxation Studies.-Factors governing line-widths of e.s.r. spectra are quite comparable with those discussed above for n.m.r. We need a time- dependent process to cause a modulation; some of these are briefly outlined below particular attention being given to chemical processes.(i) Relaxation associated with tumbling. Generally both A and g are strongly anisotropic and hence even though the tumbling motion in solution is fast enough to cause complete averaging the resulting modulation can cause a relaxation and consequently broaden the line by a factor which will depend upon the magnitude of the original anisotropy. This is one of the most important sources of line-broadening for dilute solutions of S = 3 radicals. The situation is illustrated in Figure 11. From the viewpoint of solvation the important parameter that can be extracted at least in principle from the pattern of line-broadening that results is the tumbling correlation time of the radical or complex. The factors involved have been extensively studied.72 73 However care must be exercised in deriving such parameters because several other relaxation mechanisms may also contribute to the line-width.These include spin-rotational relaxation of the type discussed for n.m.r. above in which the rotating motion of the radical itself generates an indirect magnetic field due to the electrons whose rotation fails to keep up with the molecular framework. The effect will only be important when interactions with the solvent are weak and rotation can begin to approach the gas-phase situation. It is not easy to extract contributions made by this mechanism to the overall relaxation of a radical but it seem to be the major source of line-broadening for small neutral molecules such as The isoelectronic radicals SO2- and C10 make an interesting contrast in this respect since the former gives very narrow 33S hyperfine lines in aqueous solution under which conditions the individual components of the 35Cl and 75 71 W.M. Walsh J. Jeener and N. Bloembergen Phys. Rev. 1965 139 A 1338. 72 D. Kivelson J. Chem. Phys. 1964 41 1904. 73 C. P. Slichter ‘Principles of Magnetic Resonance’ Harper and Row New York 1963. ?‘P. W. Atkins Mol. Phys. 1967 12 133. 75 G. Nyberg Mol. Phys. 1967 12 69. 296 Burgess and Symons Q b C Figure 11 Diagrammatic representation of the connection between g- and A-anisotropy and the line-widths of liquid-phase e.s.r. spectra. (a) Typical liquid-phase spectrum for a radical with hyperfine coupling to a single nucleus with I = 1. (b) Shows how these spectra are linked. (cl Envelope spectrum for stationary radicals. 37Cl hyperfine lines from the latter are so broad that they are only just resolved.The reason for this remarkable contrast is almost certainly the fact that SO2- interacts very much more powerfully with the solvent because of its charge.76 Perhaps the most important aspect of this mechanism is that such broadening should increase with temperature in contrast with that due to the anisotropy of the g- and hyperfine-tensors. Yet another important source of broadening is commonly called the Orbach process which especially applies to radicals in near-degenerate states which are not coupled directly by spin-orbit ~oupling.~’ Another is the dynamic Jahn- Teller effect in which a Jahn-Teller distortion migrates so rapidly between equivalent distortions that an isotropic spectrum results.78 At very low temperatures in crystalline solids these distortions may be ‘static’ and as the temperature is raised lines are broadened before merging according to expectation.A typical example is the ion CU(H,O),~+ the distortion being an elongation along any one of the three ligand axes.78 To what extent this dynamic 76 P. W. Atkins A. Horsfield and M. C. R. Symons J. Chem. SOC. 1964 5220. 77 R. Orbach Proc. Phys. SOC. 1961 A 77 821. 78 M. C. M. O’Brien ‘Low Symposium on Paramagnetic Resonance’ Academic Press New York 1967 p. 323. 297 Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance distortion contributes to the line-widths of such complexes in fluid solution at room temperature is not really established. Tetragonal distortion of CU(H,O),~+ in aqueous solution has been studied by lH and 1 7 0 n.m.~.7~ Yet another factor of this general type that needs to be considered for radicals which deviate considerably from a spherical shape is restricted rotation.Again in crystal or glassy lattices such rotations can be detected readily a good example being NO2 trapped in rare-gas matrices in which rotation about the long axis of the molecule remains free so that two of the principal values of the I4N hyperfine tensors are effectively averaged leaving the third unchanged.** If rotation in one sense can occur far more effectively than in another this can clearly modify the extent of broadening of particular lines in an average spectrum. This has been discussed theoretically by Freed.81 We now turn to somewhat more tractable problems which have direct bearing upon chemical rections.The first involves the transfer of nuclei such as protons in acid-base reactions or alkali-metal cations in ion-pair equilibria. The second involves transfer of electrons between molecules and this is used as a link to a final brief consideration of spin exchange and triplet states. (ii) EquiZibria involving ion-pairs. To illustrate the way in which line-widths can be used to obtain kinetic information relating to ion-pairing we consider the equilibria (1)-(4). Equilibrium (1) will be characterised by the separate weighted spectra of A- and M+A- if the rates are slow but as the life-times of these species are reduced (on warming for example) each component line will broaden until ultimately a spectrum comparable with that of the solvated ion A- is obtained. This is because any given A- will combine with different M+ ions having arbitrary nuclear spin states.M+A- + M++A- (1) (2) A-Mf $ M+A- (3) (M+A-)1 + (M'A-)2 (4) M+ + A-M+ + M+A- + M+ Then (2) is simply the bimolecular equivalent of (1) and again loss of metal hyperfine coupling will be the main result as the lifetime of any one M+A- unit is reduced. This situation can readily be achieved by adding a diamagnetic salt MfB- and rates can be calculated from the excess of broadening.82 These rates do not directly relate to (2) however since competing ion-pair equilibria with B- will play an important r&e. In addition to giving kinetic information this reaction has important practical aspects in that it can be used to simplify a complex spectrum. However if it is unplanned one can be misled into concluding that ion-pairs are absent or that the metal hyperfine coupling is too small to detect.Reactions (3) and (4) are closely related and give rise when the rates are appropriate to specific line-broadening. In (3) an intramolecular jump of the 7Q W. B. Lewis M. Alei and L. 0. Morgan J. Chem. Phys. 1966,44,2409. 80 P. W. Atkins N. Keen and M. C. R. Symons J. Chem. SOC. 1962,2873. 81 J. H. Freed J. Chem. Phys. 1964,41,2077. 82 N. Hirota and S. I. Weissman J. Amer. Chem. SOC. 1964 86 2537. 298 Burgess and Symons cation between equivalent sites of the anion is envisaged as for example between the oxygens of p-benzosemiquinone or the nitro-groups of p-dinitro- benzene anions. As already described for slow exchange spectra characteristic of the asymmetric ion-pairs are obtained whilst for fast exchange spectra similar to those of the symmetrical anions result except that hyperfine coupling to the metal remains sharp since the same cation is involved in many migrations.For intermediate exchange rates some lines of the averaged spectrum will be broadened possibly beyond detection whilst others will have narrow and sharp components and the outermost lines remain sharp. The situation both for the initial broadening of the lines of the 'static' ion-pair and for the final narrowing of the merged lines for a radical having two normally equivalent nuclei with unit spin is fully described and illustrated in ref. 47. In the limit of extreme broadening the spectrum appears to be a 1 1 1 triplet having twice the normal hyperhe coupling. If no cation hyperfine lines are detected then the alternation may be caused either by migration of a cation possibly strongly solvated or by concerted redisposition of solvent molecules such that the negative charge is 'trapped' at one end of the anion or the other depending upon the orientation of the adjacent solvent molecules.We have been unable to detect such asymmetric solvation for p-duroquinone or p-benzosemiquinone anions even using mixed solvents with the strongly interacting component in low concentration but it probably does occur for solutions of m-dinitrobenzene 84 (W. E. Griffiths C. J. W. Gutch G. F. Longster J. Myatt and P. F. Todd have recently detected a large and a small 14N isotropic hyperfine coupling for aqueous solutions of m-dinitrobenzene anions showing that in water asymmetric solvation is long-lived.) Specific examples of cation migrations are reviewed in greater depth elsewhere?' Equation (4) also represents a cation migration but between different sites.One example is that of the 2,6-dimethyl-p-benzosemiquinone anion.59 Migration should now modulate all but part of the central hyperfine component. Also the outer lines of the quartet from a cation with spin 3/2 should broaden more than the inner pair. Because the populations of the two states now differ the unhindered site being greatly favoured by the cation interpretation of the different spectra is far more complicated. Important features are (i) that for slow exchange two species should be detected but the one in low concentration may be masked by the other; (ii) the lines for the unfavoured species will broaden first; (iii) when only one species is detected slow exchange broadening can be distinguished from fast exchange in that for the former all exchanging components are equally broadened.These considerations apply equally well to equilibria of type (4) in which contact and solvent-separated ion-pairs are involved. The interesting results J. H. Freed and G. K. Fraenkel J. Chem. Phys. 1964,41,699. 8' C. J. W. Gutch and W. A. Waters Chem. Comm. 1966,39. 299 Study of Ion-Solvent and Ion-Ion Interactions by Magnetic Resonance of Hirota for alkali-metal anthracene salts may be illustrative of such reactions.86 (iii) Electron-transfer processes. We now consider yet another equilibrium (5) involving M+A- ion-pairs. As the rate of such a process increases so the individual hyperfine components of A- broaden and merge to a single line which ultimately can become very narrow.If the reaction involves M+A- rather than just A- and if hyperfine multiplets from M+ are detected then the exchange-averaged feature will also be such a multiplet. This was observed by Ward and Weissman86 for such systems as sodium naphthalenide and naphthalene. *A + M+A- + *A-M+ + A These reactions are of significance to solvation studies in that the major barrier to electron-transfer is usually the fact that solvent molecules are in quite different states in the vicinity of the reactants. To some extent it may be possible for suitably polarised solvent molecules to be ‘transferred‘ rather as the cation is but otherwise reactions must await fortuitous arrangements when the solvent favours equally the electron on either site.Hence reaction energies and entropies have a direct bearing on solvent organisation. This is in marked contrast with the situation for spin exchange. A fair model would now be to consider the approach of two A- radicals with or without their gegen-ions. When there is effective overlap between the orbitals of the two anions the electrons will exchange but now there is no solvation barrier and hence rates are very much greater being in general diffusion controlled. Generally the broadening of hyperfine lines and ultimate appearance of an exchange-narrowed single line which follows exactly the pattern for single electron transfer is simply a nuisance and is avoided by spectroscopists by using very dilute solutions. If however gegen-ions do participate the ‘transition state’ may be greatly stabilised and the resulting ion-cluster may have a life of sufficient length for direct detection.The sort of problems that then arise have been briefly reviewed el~ewhere.~’ Suffice it to say that generally two different units seem to occur commonly. In one the radicals are about 6-7 A apart and the fluid spectra are very broad singlets. The frozen solutions however give well resolved fine-structure features in the g = 2 region from which the effective spin-separation can be calculated. In the weakly interacting species the spins are usually too distant for resolution of the fine-structure features in the solid state but the fluid spectra are quite well resolved. The most significant result is then that two alkali-metal cations are involved showing that neutral ion-tetramers are present.The interaction in these ‘triplet-state’ species is almost entirely of a classical dipolar nature and the line-broadening in solution follows from this. For transition-metal complexes having S > 3 however the coupling between electrons occurs largely by a spin-orbit coupling mechanism which is far more (5) eaN. Hirota J. Phys. Chem. 1967 71 127. 86 R. L. Ward and S. I. Weissman J. Amer. Chem. Soc. 1957 79 2086. 300 Burgess and Symons effective. The result is that in the solid-state resonance features are spread over such a huge range of field that many are beyond the range of detection. Such complexes often cannot be detected at all in fluid solution. Factors controlling line-widths have been considered by Carrington and Luckhurst.*' A. Carrington and G. R. Luckhurst Mol. Phys. 1964 8 125. 301

ISSN:0009-2681    

DOI:10.1039/QR9682200276

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

The Mass Spectra of Amino-acid and Peptide Derivatives By J. H. Jones DYSON PERRINS LABORATORY OXFORD UNIVERSITY 1 Introduction The various types of absorption spectroscopy which are such powerful tools in most areas of organic chemistry find limited application in peptide chemistry; these methods reveal the main structural features but give no information about the amino-acid sequences of proteins although they are used in a semi-empirical manner to procure data about secondary structure. In the elucidation of the primary structure of a protein it is necessary to determine the amino-acid sequences of a large number of small peptides which are obtained by partial degradation. Although chemical methods for the determination of oligopeptide amino-acid sequences have attained a very high level of sophistication (e.g.the recent introduction of repeated Edman degradation of resin-bound peptides2) they are time-consuming and tedious. There has therefore been considerable interest during the last few years in the behaviour of amino-acid and peptide derivatives on electron impact since mass spectrometry provides a rapid means requiring very small amounts of material (0.2 pg. is sufficienp) of discovering the amino-acid sequences of oligopeptides. For discussion of the principles of instrumentation and of the fragmentation of ionised organic molecules the reader is directed to a previous Quarterly Review and standard texts.s In the compilation of this Review papers published up to the end of 1967 have been considered. 2 Amino-acids and their Derivatives A. Free Amino-acids.-The zwitterionic nature of free amino-acids renders them very involatile and thermal reactions can occur at the high temperatures required for vaporisation.However this difficulty has been partly overcome by E. Y. Spencer ch. 15 in ‘Techniques of Organic Chemistry’ vol. XI part 11 ed. K. W. R. A. Laursen J . Amer. Chem. SOC. 1966 88 5344. M. Barber P. Powers M. J. Wallington and W. A. Wolstenholme Nature 1966 212 R. I. Reed Quart. Rev. 1966 20 527. (a) K. Biemann ‘Mass Spectrometry’ McGraw-Hill New York 1962; (b) H. C. Hill ‘Introduction to Mass Spectrometry’ Heyden and Son Ltd. London 1966; (c) F. W. McLafferty ‘Interpretation of Mass Spectra’ W. A. Benjamin Inc. New York 1966; (a) H. Budzikiewicz C. Djerassi and D. H. Williams ‘Mass Spectrometry of Organic Compounds’ Holden-Day Inc.San Francisco 1967; (e) J. H. Beynon ‘Mass Spectrometry and its Applications to Organic Chemistry’ Elsevier Amsterdam 1960. Bentley Interscience New York 1963. 784. 302 Jones improved techniques for the direct introduction of samples into the ion source.6 The most important fragmentation is the formation of ‘amine fragments’ (a) by loss of the carboxyl group,6 after primary ionisation at the nitrogen atom6 (Scheme 1 R‘ = H). Side-chain cleavages are discussed in a separate section. -(b2R‘) + .+ n NH -CHR- C0,R’ - NH,= CHR Scheme 1 Mass spectrometry can be used for quantitative analysis of amino-acid mixtures:8 the method depends on comparison of the relative intensities of peaks peculiar to each component. However if any particularly involatile amino-acids (e.g. tryptophan) are present the contribution of these constituents to the spectrum is too small for accurate relative intensity measurement.Qualitative analysis of amino-acid mixtures can also be performed by using mass spectrometry to identify characteristic decomposition products after separation by gas-liquid chromatography (g.1.c.) of the products of controlled pyroly~is.~ Time-of-flight mass spectrometerslO have been favoured in work with free amino-acidsll l2 and other biologically important compounds such as nucleosides,13 and the fact that these instruments can be made in a compact form suitable for rocket transport gives a possible method for detecting life-characteristic molecules in extraterrestrial environments. Comparison of the mass spectra of ordinary leucine and leucine recovered after equilibration with pyrid~xal-alum-D,O~~ revealed a deuteriation pattern which was consistent with the generally accepted15 Schiff-base mechanism for pyridoxal-catalysed transamination.Another example illustrating one of the pitfalls of mass spectrometry is the case of the yellow pigment (I) isolated from South American butterflies. The ion of highest mass-to-charge ratio (rn/e) in the spectrum was erroneously identified as the molecular ion and the structure (11) was assigned.16 It was later found that (I) was converted into (11) by cyclisation with loss of ammonia in the heated inlet system.17 a G. Junk and H. Svec J. Amer. Chem. SOC. 1963 85 839. K. Heyns and H. F. Griitzmacher Annalen 1963 667 194. G. A. Junk and H. J. Svec Analyr. Chim. Acta 1963 28 164. C. Merritt and D.H. Robertson J. Gas Chromatog. 1966 5 96. lo D. B. Harrington ‘Advances in Mass Spectrometry’ ed. J. D. Waldron Pergamon Press Oxford 1959 p. 249. l1 K. Biemann and J. A. McCloskey J. Amer. Chem. SOC. 1962,84,3192. l2 N. Martin NASA report No. CR-68768 (Chern. Abs. 1967 66 1587). la K. Biemann and J. A. McCloskey J. Amer. Chem. SOC. 1962 84 2005. l4 G. A. Junk and H. J. Svec J. Org. Chem. 1964 29 944. l5 A. E. Braunsteh ‘The Enzymes’ vol 2 ed. P. D. Boyer H. Lardy and K. Mryback Academic Press New York 1960 p. 137. l6 K. S. Brown J. Amer. Chem. SOC. 1965 87,4202. 1’ K. S. Brown and D. Becker Tetrahedron Letters 1967 1721. 303 The Mass Spectra of Amino-acid and Peptide Derivatives B. Amino-acid Alkyl Esters.-Because of their greater volatility a-amino- esters18 l9 are more suitable for mass spectrometric investigation than the free acids.The fragmentation mechanisms of these compounds have been discussed at length in a well-known treatise20 and it suffices here to say that the most intense peaks are usually the amine fragments (a). Mixtures of amino-acid esters can be analysed quantitatively21 by use of the principle outlined above for mixtures of the free acids. Such procedures do not compare favourably with orthodox methods based on ion-exchange chromatography and colorimetric estimation,22 but may find use when only very limited amounts of material are available. Mass spectrometry has been used to distinguish between possible structures for the y-hydroxy-a-amino-acids obtained by degradation of toxins from Amanila phalloides. The acids were converted into the corresponding lactone hydrochlorides which underwent sufficient thermal dissociation at the source temperature for the spectra of the free bases to be observed.23 Another example is the determination of the structure (111) of lysopine (an amino-acid isolated in very small amount from crown gall tissue) from the mass spectrum of its ethyl ester.% CH,.CH.N H.CH.CO,H CO2H ' C H21PNH2 C. N-Protected Amino-acids and their Derivatives.-The mass spectra of N-acetylamino-acids' and their alkyl esters25 all show an intense acetyl ion (m/e 43) and the chief common feature is loss of the carboxyl (or alkoxycarbonyl) group to give an acyliminium ion (b) which then ejects keten with formation of an amine fragment (a) (Scheme 2). In addition side-chain cleavages characteristic of the amino-acid concerned are observed (see p.309). Scheme 2 K. Biemann J. Siebl and F. Gapp J. Amer. Chem. SOC. 1961 83 3795. Is C.-0. Andersson R. Ryhage S. Stalberg-Stenhagen and E. Stenhagen Arkiv Kemi 1962 19 405. 2o Ref. 5 (a) p. 260. a1 K. Biemann and W. Vetter Biochem. Biophys. Res. Comm. 1960 2 93. 22 D. H. Spackman W. H. Stein and S . Moore Analyt. Chem. 1958,30 1190. 24 K. Biemann G. G. J. Deffner and F. C . Steward Nature 1961 191 380. P. Pfaender and T. Wieland Annulen 1966 700 126. (2-0. Andersson R. Ryhage and E. Stenhagen Arkiv Kemi 1962 19 417. 304 Jones N-Trifluoroacetylamino-acid methyl esters are suitable derivatives for the separation of amino-acid mixtures by g.1.c.F6 and the constituents of the mixture can be identified by passing the eluant from the g.1.c.column directly into the ion source of a fast-scanning mass spectrometer. This technique ('Combination g.1.c.-mass spe~trometry'~~) has been used for analysis of the mixture produced by trifluoroacetylation and esterification of the amino-acids obtained by acid hydrolysis after treatment of butyl-lithium with atomic nitrogen.28 The mass spectra of the methyl esters of a few N-formylamino-acids2B and one N-isovalerylamino-acidm have been recorded. N-Benzyloxycarbonyl derivatives31 give spectra consisting essentially of the superimposed spectra of benzyl alcohol and the corresponding isocyanate (formed by pyrolysis) if the source temperature exceeds ca. 2OO0 but if the sample is sufficiently volatile for a spectrum to be obtained at a lower temperature the protecting group can either break down to give a tropylium or it can lose a benzyloxy-radical.Cleavage of the C-CO bond gives ions (c) which then expel carbon dioxide with concomitant migration of the benzyl group (Scheme 3). Rearrangements involving the ejection of neutral Scheme 3 fragments from non-terminal positions are of some importance because of their mechanistic interest and relevance to the 'element mapping' technique :% a detailed review of such processes has appeared.34 Phthaloylamino-acids35 have a tendency to sublime and good spectra are obtained at moderate source temperatures. A number of pathways can be discerned but the most interesting is the loss of novel neutral fragments from the intense ions (d) formed by loss of the carboxyl group from phthaloylamino- acids with an alkyl side chain [e.g.the mechanism which has been suggested for the case of ion (d) from phthaloylvaline (Scheme 4)]. Trimethylsilylation is a general procedure for the preparation of volatile derivatives from compounds with polar functionalities such as hydroxyl 2* F. Weygand B. Koeb A. Prox M. A. Tilak and J. Tomida 2. physiol. Chern. 1960 322 38. 27 F. A. J. M. Leemans and J. A. McCloskey J. Amer. Oil Chemists' SOC. 1967 44 11. 28 J. Winkler quoted in ref. 61. 20 K. Heyns and H. F. Griitzrnacher 2. Naturforsch. 1961,16b 293. 30 K. Tanaka and K. J. Isselbacher J Biol. Chem. 1967 242 2766. 31 R. T. Aplin J. H. Jones and B. Liberek J. Chem. SOC. (C) 1968 1011. 32 S . Meyerson P. N. Rylander E. L. Eliel and J. D. McCollum J . Amer. Chem. Soc. 1959 81,2606. 33 K. Biemann P. Bommer and D. M. Desidero Tetrahedron Letters 1964 1725.s4 P. Brown and C. Djerassi Angew. Chem. Znternat. Edn. 1967 6 477. 36 R. T. Aplin and J. H. Jones Chem. Comm. 1967 261. 305 The Mass Spectra of Amino-acid and Peptide Derivatives Scheme 4 amino- and carboxyl groups (e.g. g.1.c. work with carbohydrate^,^^ amino- acids:’ etc.) but only incomplete details of the mass spectra of a few trimethylsilylamino-acid trimethylsilyl esters are Mass spectrometry has been recommended as a means of analysing mixtures of 2,4-dinitrophenylamino-a~ids,3~ which are produced in Sanger’s N-terminal amino-acid deterrninati~n.~~ A second application of mass spectrometry in conjunction with ‘wet’ methods of sequence analysis is in the identification of phenylthiohydantoins,4l which are produced by Edman degradati~n.~~~ 4g 3 Peptides which give only Amino-acids on Hydrolysis A.Free Peptides.-Free peptides are very involatile but the mass spectra of a few free oligopeptides (e.g. gly~ylleucyltyrosine~~) have been reported. Dipeptide~~~ 46 undergo thermal cyclisation to dioxopiperazines and the sequential individuality of the amino-acids is thereby lost. Because of their very low vapour pressures and susceptibility to thermal decomposition peptides are normally subjected to chemical modification before mass spectrometric examination. B. Chemical Modamtion of Peptides for Mass Spectrometry.-Reduction of peptides or acetylpeptides with lithium aluminium hydride (Scheme 5 ) gives LiAlH H-( N H . CH R.CO),-O H + H-( N H.CH R.CH2) n-0 H Scheme 5 polyamino-alcohols which are fairly volatile separable by g.l.c.and give simple mass spectra.47 The principal ions are of type (e) formed by cleavage 01 to the s6 C. C. Sweeley R. Bentley M. Makita and W. W. Wells J . Amer. Chem. SOC. 1963 85,2497. s* R. M. Teeter paper presented at the ASTM Committee E-14 Mass Spectrometry Conference New Orleans 1962. 39 T. J. Prenders H. Copier W. Heerma G. Dijkstra and J. F. h e n s Rec. Trav. chim. 1965 85 216. 40 F. Sanger Biochem. J. 1945 39,507; 1949,45 563. 41 N. S. Wulfson V. M. Stepanov V. A. Puchkov and A. M. Zyakoon Zzvest. Akad. Nauk S.S.S.R. Ser. khim. 1963 1524. 42 P. Edman Acta Chem. Scand. 1950,4,283. 44 Ref. 5 (a) p. 285. 45G. A. Junk and H. J. Svec. J. Amer. Chem. SOC. 1964 86,2278 47 Ref. 5 (a) p. 284. E. D. Smith and H. Sheppard Nature 1965 208 878. P. Edman Acta Chem. Scand.1956,10 761. K. Heyns and H. F. Griitzmacher Annalen 1963,669,189. 306 Jones nitrogen atoms. As a series of ions (e) corresponding to stepwise loss of amino- acid units from the carboxyl end are observed the sequence of the original peptide can be deduced. The alternative a cleavage generates a positive charge on a primary carbon atom and the resulting ions (f) are of lower abundance than the ions (e). When the polyamino-alcohol contains side-chain hydroxyl groups further chemical modification (thionyl chloride treatment followed by lithium aluminium deuteride reduction) becomes necessary and the complexity of the chemical treatment is probably responsible for the fact that this method has not gained popularity. + + [-NH-CHR - - NH=CHR] e [cH,- NH -. - CH,= E;H - 1 f Techniques for the direct insertion of samples into the ion source make it possible to examine peptide derivatives with intact amide bonds and all recent work has been on N-protected peptides.N-Protection eliminates the zwitterionic character and it is in any case necessary to mark the N-terminal amino-acid so that ions with the same sequence as the original peptide up to the point of cleavage can be identified. In addition esterification of the carboxyl terminal (methyl,"8 or t-buty150 esters) gives increased volatility though it has been stated51 that this is advantageous in marginal cases only. Andersson's preliminary studies on trifluoroacetyl peptide esters52 were extended by Stenhagen:* who was the first to point out the implications of the mode of fragmentation of the peptide bond for sequence determination and the cleavage patterns were further clarified by extensive investigations on acetylpeptides,46 trifluoroacetylpeptide and peptide esters with fatty acid (C > acyl groups.54 Techniques for the acylation and esterification of peptides on a very small scale have been described.55* 66 Other N-protected 48 E.Stenhagen 2. analyt. Chem. 1961,181,462. 49 V. G. Manusadzhyan A. M. Zyakoon A. V. Chuvilin and Ya. M. Varshavskii Izvest. Akad. Nauk Arm. S.S.R. Ser khim. 1964,17 143. 50 M. M. Shemyakin Yu. A. Ovchinnikov A. A. Kiryushkin E. I. Vinogradova A. I. Miroshnikov Yu. B. Alakhov V. M. Lipkin Yu. B. Shetsev N. S. Wulfson B. V. Rosimov V. N. Bocharev and V. M. Burikov Nature 1966 211 361. 51 K. Biemann C. Cone B. R. Webster and G. P. Arsenault J. Amer. Chem.SOC. 1966 88 5598. 52 (2-0. Anderson Acta Chem. Scand. 1958,12 1353. 5s F. Weygand A. Prox H. H. Fessel and K. K. Sun 2. Naturforsch. 1965 20b 1169. 54 E. Bricas J. van Heijenoort M. Barber W. A. Wolstenholme B. C. Das and E. Lederer Biochemistry 1965 4 2254. 55M. Senn R. Venkataraghavan and F. W. McLafferty J. Amer. Chem. SOC. 1966 88 5593. 56 A. A. Kiryushkin Yu. A. Ovchinnikov M. M. Shemyakin V. N. Bocharev B. V. Rosinov and N. S. Wulfson Tetrahedron Letters 1966 33. 307 The Mass Spectra of Amino-acid and Peptide Derivatives peptides which have been examined include benzyIoxycarbonyl,3l phthaloyIYs1 etho~ycarbonyl,~~ he~anoyl,~’ and 2,4-dinitro~henyl~~ derivatives. When functional side chains are present there are difficulties owing to decreased volatility and additional routes for thermal degradation.Arginine- containing peptides were for these reasons outside the realm of mass spectrometric investigation until recently when it was shown that conversion of arginine residues into 6-(2-pyriminidyl)ornithine residue@ (Scheme 6) gave derivatives - NH-CH CO C HN” ‘NH~ Scheme 6* * In this and other schemes and formulae a wavy line chain. NAN R ~ R indicates the remainder of the peptide with sufficient volatility and simple mass spectra.60 Side-chain amino-groups (lysine ornithine) and carboxyl groups (aspartic and glutamic acids) are modified simultaneously with the terminal groups during the acylation- esterification procedure. 50 Alcoholic side-chains (serine threonine) can be left free or converted into their 0-acetyl derivativesYg1 but with some tyrosine- containing peptides methylation of the phenolic group is essential in order to obtain satisfactory spectra.62 Cysteine peptides have only been examined with the thiol function protected by means of a methoxycarbonylmethy150 or benzyle3 group.Although peptides containing unmodified asparagine glutamine histidine or tryptophan residues have been investigated sucessfully the presence of more than one of these amino-acids prevents the observation of spectra with peptides greater than tetra pep tide^.^^ Even in the absence of polar side-chains there is a limit (at present eight62 or nines0 64 amino-acids) to the length of peptide chains which can be investigated. It is general experience that acetylpeptides containing N-substituted amino-acids are more volatile than ordinary acylpeptides.62 This is because the principal reason for the low volatility of acylpeptides is inter-chain hydrogen bonding and it was therefore 57 J.P. Flikwert W. Heerma H. Copier G. Dijkstra and J. F. Arens Rec. Truv. chim. 1967 86 293. 58T. J. Penders W. Heerma H. Copier G. Dijkstra and J. F. Arens Rec. Trav. chim. 1966 85 879. 59 T. P. King Biochemistry 1966 5 3454. 6o A. A. Kiryushkin M. Yu. Feigina E. I. Vinogradova Yu. A. Ovchinnikov M. M. Shemyakin Yu. B. Alakhov N. A. Aldanova B. V. Rosinov and V. M. Lipkin Experientiu 1967 23 428. 62 J. van Heijenoort E. Bricas B. C. Das E. Lederer and W. A. Wolstenholme Tetrahedron 1967 23 3403. 83 E. Bayer C. Jung and W. Konig 2. Naturforsch. 1967 22b 924. 64 M. Barber P. Jolles E. Vilkas and E. Lederer Biochem. Biophys. Res. Comm. 1965 18 469.K. Heyns and H. F. Grutzmacher Fortschr. Chem. Forsch. 1966 6 536. 308 Jones suggesteds5 that methods for the replacement of the peptide hydrogen atoms would ameliorate the volatility problems. Treatment of acylpeptides with methyl iodide in dimethylformamide in the presence of silver oxide66a results in quantitative methylation of the peptide nitrogen atoms and side-chain functionalities yielding permethyl derivatives which are more volatile than the original acylpeptides and give excellent mass spectra. The fact that the reaction is heterogeneous is a disadvantage but providing the sample requirement can be drastically reduced from the present level (ca. 2 mg.s6b) this procedure will probably prove to be of great value. C. Features of the Spectra Due to the Side Chains.-All amino-acid and peptide derivatives show fragmentations which are characteristic of the side chains.Loss of the side chain via a McLafferty rearrangement6’ occurs with leucine isoleucine and valine (e.g. Scheme 7) and the observation of peaks Scheme 7 corresponding to loss of olefin fragments is diagnostic of these amino-a~ids.~~ Alkyl side chains can also be partly or wholly lost as neutral radicals and leucine differs sufficiently from isoleucine in this respect to permit distinction. 53 Functional groups in the position are eliminated (probably by a thermal reaction) e.g. serine derivatives usually show [M- H,O ] peaks. Sometimes thermal degradation is useful thus in the spectra of a-(7-methylglutamyl) peptides all fragments (m) retaining the glutamic acid residues are accompanied by (m- 18) ions corresponding to pyrolysis of the peptide (Scheme 8) but such Scheme 8 ions are not observed in the spectra of the isomeric y-(a-methylglutamyl) peptides.68 When the side chain is of the type aryl-CH,- (phenylalanine tyrosine histidine tryptophan) side-chain cleavage occurs with preferential 6 5 E.Lederer and B. C. Das ‘Peptides’ ed. H. C. Beyerman A. van de Linde and W. Massen van den Brink North Holland Publishing Co. Amsterdam 1967 p. 131. c6 (a) B. C. Das S. D. Gero and E. Lederer Biochem. Biophys. Res. Comm. 1967,29 211; (b) D. S. Millington personal communication. 67 Ref. 5b p. 66. 68 A. A. Kiryushkin A. I. Miroshnikov Yu. A. Ovchinnikov B. V. Rosinov and M. M. Shemyakin Biochem. Biophys. Res. Comm. 1966 24 943. 309 17te Mass Spectra of Amino-acid and Peptide Derivatives retention of charge by the side-chain moiety.69 Simple fission of the C,-C bond gives ions of type [aryl-CH,]+ which in the case of the [C,H7]+ ion32 at least have the aromatic tropylium ion structure.Alternatively transfer of one of the benzylic hydrogen atoms gives rise to characteristic conjugated ions as shown in Scheme 9. Cleavages involving the side chain are always prominent NH H -cptcti-coM -(-c:,) -(ion-) ti I - [PhCH=CH C o d - [PhCH=CH]’ 2 r,(? H Ph Scheme 9 features in the mass spectra of peptides containing aromatic amino-acid~,~~ and usually dominate the spectra of tryptophan derivative^.^^ If an acylpeptide methyl ester has any basic heterocyclic side-chains the ions (m) containing the heterocyclic ring often have satellite (rn + 14) ions.60s 65n 6o These peaks probably arise by a thermal transfer of a methyl group from the C-terminal ester function to the basic nitrogen atom.Similar complications have been observed in the mass spectrum of the indole alkaloid v~acamine.~~ Other types of side-chain also have their characteristic modes of cleavage (e.g. ring contraction of proline derivatives50) but these are not yet as fully documented as those mentioned above. The generalisation has been made72 that ‘the specific behaviour of the individual amino-acids ‘is more strongly expressed the closer the particular amino-acid is to the N-terminus’ but the basis for this statement is not clear as no systematic comparison of sequential isomers has been made. Indeed in the case of some simple acetyldipeptides containing leucine,46 the reverse appears to be true as expulsion of isobutene from the molecular ions of these compounds is more marked when leucine is in the C-terminal position.D. Fragmentation of the Peptide Bond.-Ionised peptide chains rupture at the amide bonds in two main modes:46 507 537 54 (i) Cleavage of the CO-N bond (or at the C-terminus the CO-0 bond) gives acylium ions (8) which then lose the next amino-acid residue either by concerted loss of carbon monoxide and a neutral imine fragment or via the acyliminium ion (h) (Scheme 10). When R1 = H (i.e. glycine residues) the ions (h) are of lower abundance than in other cases as the electron-donating effect of alkyl groups is a stabilising influence. The acylium ions (g’) formed by (n = 1) and y (n = 2) peptides do not lose carbon monoxide as this would give an ion (h’) in which the positive charge could not be delocalised as in (h) the absence of ions (h’) has therefore K.Heyns and H. F. Griitzmacher Annalen 1966 698,24. 70 P. Pfaender Annalen 1967 707 209. ‘lD. W. Thomas and K. Biemann J Amer. Chem. SOC. 1965 87 5447. 72 M. M. Shemyakin Yu. A. Ovchinnikov and A. A. Kiryushkin ‘Peptides’ ed. H. C. Beyerman A. van de Linde and W. Massen van den Brink North Holland Publishing Co. Amsterdam 1967 p. 155. 3 10 Jones C B A I [- CO~NH I Scheme 10 been recommended as a criterion for distinguishing p-alanyl p-aspartyl and 7-glutamyl peptides from their a-linked isomers.62 The ions (h) also decompose by hydrogen transfer giving an amine fragment and a neutral substituted keten (cf. Scheme 2). (ii) Cleavage of the C-CO bond is generally less favoured than the alternative fission between carbon and nitrogen but occurs with retention of charge by either moiety (Scheme 11).E. Determination of Amino-acid Sequence.-It follows from the above discussion that the mass spectrum of an acylpeptide ester will contain a series of ions (go) (g3 (gz) etc. and a series (I+,) (hl) (h& etc. corresponding to loss of 0 1 2 etc. amino-acid residues. Clearly if these ions (the 'sequence ions') can be identified the amino-acid sequence of the peptide can be deduced. The 31 1 The Mass Spectra of Amino-acid and Peptide Derivatives sequence ions frequently comprise only a small proportion of the total ion current but their recognition can be facilitated by judicious choice of the protecting groups. Thus acylation of the N-terminus with an equimolecular mixture of acetic and trideuterioacetic acids62 causes the ions with an intact N-terminus ( i e .the sequence ions) to appear as pairs of peaks of equal intensity separated by 3 mass units. An equimolecular mixture of hepta- and octa-decanoic acidsM can be used in a similar manner with the additional advantage that the sequence ions appear in the high mass range clear of the majority of the other fragments. Interpretation of the spectra can also be simplified by comparing the fragmentation of the methyl and trideuteriomethyl esters or the acylpeptide which identifies ions retaining the C-terminu~.~~ Because the mass spectra of peptide derivatives are so complex data-processing techniques are of great importance. Weygand’s ‘Differen~schema’~~ consists of a systematic examination of the difference in mass between the peaks in the spectrum.Starting from th: molecular ion a search is made for difi‘erences which correspond to loss of amino-acid residues and the relationships between the sequence ions are checked by use of metastable73 ions i.e. broad low-intensity peaks due to ions which decompose during acceleration. The mass (m*) of such ions is ca. m22/m1 where m is the mass of the parent ion and rn that of the fragment ion so that observation of a metastable ion can be used to relate the peaks at m and V Z ~ . ~ ~ Recent technical advances74 mave made it possible with the aid of computers to determine complete high-resolution mass spectra (i.e. spectra in which the m/e of every peak is determined to a degree of accuracy which permits unambiguous assignment of elemental composition).Because the primary results come from a computer and because these spectra comprise such an enormous amount of precise information it is logical to extend the application of the computer to analysis of the spectra. In the computer program devised by McLafferty el ~ f . ~ ~ the N-terminal amino-acid is identified by checking combinations of the exact mass of the N-protecting group (43.01539 for CH,.CO- etc.) and the exact masses of each of the possible amino-acid residues against the observed exact masses. The computer then identifies the next amino-acid by the same procedure and continues the process until addition of the exact mass of the C-terminal protecting group locates the molecular ion. The program can also accommodate cases where no molecular ion is observed by searching for typical fragment ions such as [M-H20],J- [M-CO,].+ etc.The logic of the method of Biemann et aL61 is similar but in this case the N-terminal amino-acid is identified by checking combinations of the exact mass of the protecting group plus -NHCH- with the exact masses of each of the possible side chains against the observed exact masses. The subsequent sequence is determined analogously. The program of Barber et ~ l . ~ on the other hand elaborates the sequence starting from the 73 Ref. 5e p. 251. 74F. W. McLafferty Science 1966 151 641; also relevant literature cited in refs. 3 51 and 55. 312 Jones C-terminus the need for a molecular ion is a limitation but this program is more versatile than the two mentioned above as it applies to cyclopeptides and cyclodepsipeptides as well as linear peptides.The programs of Biemannsl and M~Lafferty~~ aim to eliminate all chemical examination by determining not only the sequence but also the identity of the amino-acids which seems unnecessarily restrictive as amino-acid analyses are now easily obtained from very small samples Barber’s3 program can use any available chemical information. The use of high-resolution spectra makes it unnecessary to use special pro- tecting groups for the identification of the sequence ions even a simple N-acetyl group gives the sequence ions unique masses which are determined with a degree of accuracy which eliminates any ambiguity. Further the subjective element of interpretation is removed because the computer considers all the peaks in the spectrum irrespective of their intensity.F. Cyc1opeptides.-Apart from cyclodipeptides (2,5-dio~opiperazines),~~ 46 only a small number of cy~lopeptides~~ 75 has been examined. These compounds exhibit comparatively abundant molecular ions the cross-linked cyclopenta- peptide malformin B,76 (IV) is an outstanding example in this respect. 1 V)* a I I e-C)s-Val-C)s-Leu r- * In this and other formulae abbreviated designations for amino-acids and their mode of use in representing peptide derivatives is as recommended in I.U.P.A.C. Information Bulletin No. 25 1966. Cyclopeptide rings open on electron impact giving acyclic ions which lose amino-acid residues in a stepwise manner in the same way as linear peptides. The amino-acid sequence can be deduced from the mass spectrum but the occurrence of ring opening at more than one site complicates ths interpretation.The case of the cyclononapeptide (V)77 isolated from linseed is instructive. Partial methanolysis of the peptids followed by trifluoroacetylation gave a mixture of trifluoroacetyl-di -tri- and -tetra-peptide methyl esters which were separated by g.1.c. The amino-acid sequences of these degradation products were determined by mass spectrometry by use of the ‘Differenzschema’ (see p. 312) and consideration of the ways in which the sequences overlapped indicated structure (V) which was consistent with the mass spectrum of the intact cyclopeptide. I (V) Leu-lie-Ile-Leu-Val-Pro-Pro- Phe-Phe r 75 B. J. Millard Tefrahedron Leffers 1965 3041. 76 S. Takeuchi M. Senn R. W. Curtis and F.W. McLafferty Phytochemisfry 1967 6 287. 77 A. Prox and F. Weygand ‘Peptides’ ed. H. C. Beyerman A. van de Linde and W. Massen van den Brink North Holland Publishing Co. Amsterdam 1967 p. 158. 313 The Mass Spectra of Amino-acid and Peptide Derivatives 4 Peptides which are not Constructed Exclusively from Amino-acids. The mass spectra of peptides of this type ('conjugated' peptides such as pep t idolipids glyco-pep t ides pep tide-alkaloids e tc. and depsipep t ides) have been of particular interest for two main reasons (i) certain special aspects of the spectra (especially of peptidolipids) often make them relatively simple to interpret so that this area has been a convenient testing ground for ideas about mass spectrometric sequential analysis; and (ii) mass spectrometry is well suited for solving structures with unusual features whereas conventional methods of sequence determination are only well developed for normal peptides containing the amino-acids commonly found in proteins.The first amino-acid sequence to be determined by mass spectrometry was that of fortuitines* (VI) an acylnonapeptide ester isolated from Mycobacterium Me Me Ac Ac C H 3. (CH 2) 12 .CO. Val-Leu-Val-Val-Leu-Th I " r T h r-Aia-Pro.OMe (VI) I t~ = 18,20 furtuitum. The mass spectrum contained a prominent and complete set of sequence ions which were easily recognised because (i) they all occurred at m/e > 400 clear of most of the other ions; and (ii) all ions with an intact N-terminus appeared as pairs of equally intense peaks separated by 28 mass units owing to the fact that the lipid group was an equimolecular mixture of homologues.A further example is afforded by the peptidolipid isolated from Mycubacterium johnei for which a tetrapeptidolipid structure had been considered78 on the basis of hydrolytic degradation. Mass spectr~rnetry,~~ however indicated the pentapeptide structure (VII) the amino-acid analysis was incorrect because the hindered isoleucylisoleucine peptide bond is only hydrolysed under extreme conditions. C H 3. (C H 2) %. CO. P h e-I I e -I I e-P h e-A1 a.0 M e (VII) n = 14,16,ia 20 Peptidolipin NA (VIII),80 Val6-peptidolipin NA,81 and cc-aminobutyryll- peptidolipin NAS2 (a series of closely related peptidolipid lactones from Nocardia asteroides) all underwent ring opening on electron impact by loss of carbon I 1 CH,.(CH,),,.CH.CH,.CO.Thr-Vai-Ala-Pro-olle-Ala-Thr I (VIII) 78 G. Laneele and J. Asselineau Biochim. Biophys. Acta 1962 59 731. 79 G. Laneele J. Asselineau W. A. Wolstenholme and E. Lederer Bull. SOC. chim. France 1965,2133. 8o M. Barber W. A. Wolstenholme M. Guinand G. Michel B. C. Das and E. Lederer Tetrahedron Letters 1965 1331. 81 M. Guinand M. J. Vacheron G. Michel B. C. Das and E. Lederer Tetrahedron 1966 Suppl. No. 7 271. 82 M. Guinand G. Michel B. C. Das and E. Lederer Vietnamica Chim. Acta 1966 37. 314 Jones dioxide from the ester function giving open-chain ions the subsequent fragmentation of which revealed the amino-acid sequences. The antibiotic Staphlomycin S (IX),83 which is particularly interesting because of the predominance of unusual amino-acids behaved in a similar manner the molecular ion lost carbon dioxide and a hydrogen atom to give the acylpeptide acylium ion (i).The stepwise splitting of (i) confirmed the amino-acid sequence Et I Me 0 p h a c 0 . n I - NH.cH.co-Pro~Phe - N CO-NH.~-CO- (IX) OH 0 which had been suggested from other evidence. A group of peptidolipid lactones isolated from the genus Isaria has also been inve~tigated,~~ but the spectra of these compounds were more complex than in the examples given above. Thus the confirmation of the structure of isariin (X)85 required an examination of C H 3( C H 2) 8C H . CO. G I y-Val - Leu-Al a-Val - I 0 the spectra of partial degradation products in addition to that of the intact natural product. of the mass spectra of cyclodepsipeptides containing more than one ester link but as the fragmentation pathways have been summarised in a previous Quarterly Review,a8 no further discussion will be given here.Examination of mycoside Cbag has shown that mass spectrometry is also useful in structural work with glycopeptides and investigations of some synthetic N-acyla~noacyl-2-deoxy-2-aceta~do-3,4,6-tr~-~-acetyl-~-~-glucosylam~nes in- dicate that the presence of an amino-acid-hexosamine link in a glycopeptide can be inferred from the mass spectrum of a suitable derivati~e.~~ A combination of mass spectrometry and chemical degradation has proved There have been a number of detailed studiesa6s 83 A. A. Kiryushkin V. M. Burikov and B. V. Rosinov Tetrahedron Letters 1967 2675. 85 W. A. Wolstenholme and L. C. Vining Tetrahedron Letters 1966 2785. 86 C. H. Hassal and J. 0. Thomas Tetrahedron Letters 1966 4485.N. S. Wulfson V. A. Puehkov V. N. Bocharev B. V. Rozinov A. M. Zyakoon M. M. Shemyakin Yu. A. Ovchinnikov V. T. Ivanov A. A. Kiryushkin E. I. Vinogradova M. Yu. Feigina and N. A. Aldanova Tetrahedron Letters 1964 95 1 . L. H. Briggs B. J. Fergus and J. S. Shannon Tetrahedron Suppl. No. 8 (l) 1966 269. D. W. Russell Quart. Rev. 1966 20 559. E. Vilkas A. Rojas B. C. Das W. A. Wolstenholme and E. Lederer Tetrahedron 1966 22 2809. O0 L. Mester A. Schimpl and M. Senn Tetrahedron Letters 1967 1697. 315 3 The Mass Spectra of Amino-acid and Peptide Derivatives very useful in the assignment of the related cyclic peptide alkaloids scutianin (XI),91 ~andamine,~~ ~izyphin,~~ and ceanthonine B.94 Me I Me.LPhe -Pro-N H.CH.CO-Phe-NH.CH =CH-C6H4-O- I I L H 1 5 Conclusion Mass spectrometry is now firmly established as a powerful tool for the solution of structural problems involving conjugated peptides.With peptides of the type obtained by partial hydrolysis of proteins however improvements in procedures for the preparation and separation of suitable derivatives are required before the full potential of the method can be harnessed. Many workers are applying themselves assiduously to the remaining problems and we can expect an increasing swing towards mass spectrometric sequential analysis in the near future. Since this review was completed an important new method (treatment with mythyl iodide in dimethylsulphoxide containing sodium methylsulphinylmethide) for the permethylation of peptides for mass spectrometric purposes has been described.95 I thank Balliol College Oxford for a Junior Research Fellowship and Drs. R. T. Aplin I. Eland and G. T. Young for discussions. g1 R. Tschesche R. Welters and H.-W. Fehlhaber Chem. Ber. 1967 100 323. 92 M. Pais X. Monseur X. Lusinchi and R. Goutarel Bull. SOC. chim. France 1964 817. 93 E. Zbiral E. L. Menard and J. M. Muller Helv. Chim. Acfu 1965 48 404. g4 E. W. Warnhoff J. C. N. Ma and P. Reynolds-Warnhoff J . Amer. Chem. Soc. 1965 87 4198. 96 E. Vilkas and E. Lederer Tetrahedron Letters 1968 3089. 316

ISSN:0009-2681    

DOI:10.1039/QR9682200302

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

Mass Spectra of Organometallic Compounds By D. B. Chambers F. Glockling and J. R. C . Light CHEMISTRY DEPARTMENT THE UNIVERSITY DURHAM The basic theory1 and organic aspects2 of mass spectrometry have been summarised recently and sufficient is known of the behaviour of organometallic compounds under electron impact to show the major differences from organic molecules. Problems connected with the identification of ions containing polyisotopic elements are first considered and this is followed by a discussion of appearance potential measurements. General aspects of the fragmentation behaviour of organometallic compounds are then summarised and this is followed by a treatment of first main group and then transition-metal compounds. The range of compounds covered includes the metalloids (Si and As) and transition-metal carbonyls but excludes boron and phosphorus.Aston3 used the volatile alkyls and carbonyls of metals to determine their isotopic constitution and for compounds containing polyisotopic metals the relative isotope abundances produce characteristic patterns of ions which are readily re~ognised,~ and can assist in identifying species in low-resolution spectra. With increasing numbers of polyisotopic atoms the calculation of abundance patterns and mass combinations is more readily performed by computer method^.^ In comparing the abundances of ions containing polyisotopic elements with other types the contributions from each isotope combination must be summed. For example a spectrum showing three peaks of relative height 2 1 1 due to lZ7I+ 81Br+ and 79Br+ corresponds to a 1 1 ratio of I+ to Br+ (81Br = 49% 7DBr - 51 %).Similarly in obtaining relative ion abundances allowance must be made for the 13C content of each ion. Although much information can be obtained by studying mass spectra at low resolution (1 lOOO) there are some problems for which high resolution (1 10,OOO) facilities are essential since these allow distinctions to be drawn based on packing-fraction mass differences. Fragmentation processes of ions containing polyisotopic elements sometimes produce overlapping patterns and identification of each component then requires high-resolution mass measurements on selected ions. This situation arises even in simple molecules such as tetramethylgermane where the ions GeCH,+ GeCH,+ - GeCH,+ R. I. Reed Quart. Rev. 1966 20 527 and refs.therein. H. Budzikiewicz C. Djerassi and D. H. Williams ‘Mass Spectrometry of Organic Compounds’ Holden-Day Inc. San Francisco 1967. F. W. Aston ‘Mass Spectra and Isotopes’ 2nd edn. Edward Arnold and Co. London 1942. * D. B. Chambers and F. Glockling J. Chem. SOC. (A) 1968 735; J. R. C. Light and F. Glockling J. Chem. SOC. (A) 1968 717; D. B. Chambers F. Glockling and M. Weston J. Chem. SOC. (A) 1967 1759 or refs. therein. A. Carrick and F. Glockling J. Chem. SOC. (A) 1967 40. 317 Mass Spectra of Organometallic Compounds GeCH2+. GeCHf and GeC+* form a complex pattern at low resolution. Identification can also be a problem with compounds of monoisotopic metals. For example the spectrum of diethylberyllium shows a fragment ion at mass 37 which in the absence of high-resolution facilities could reasonably be assigned to C2H4Be+ or C,H+.Diffuse ‘metastable’ peaks of low abundance often appear in low-resolution mass spectra usually at non-integral masses. These result from the decomposition of ions in the field-free region between the magnetic and electrostatic analysers (in a double-focusing instrument) and for a transition (l) if there is no internal m1+ + m2+ + (ml - ni2) (1) energy release the apparent mass1 of the metastable peak m* is given by (2). Metastable peaks are not always symmetrical and for ions containing polyisotopic elements the isotope abundance pattern appears in the metastable peak. If m1 is of high mass or if the neutral fragment (ml - mJ eliminated is large individual peaks coalesce. If the neutral fragment contains a polyisotopic element the pattern is distorted in various ways.4 Variations in the ionizing energy between 30 and 70 ev does not generally alter ion abundances significantly.Below 30 ev fragment ions disappear from the spectrum as the threshold energy for a given process is reached. Appearance Potentials Heats of formation and bond dissociation energies can be measured or compared from appearance-potential measurements using the cycle shown in Scheme 1 ,l D 0 X Y - x* + Y. XY+.X+ + Y. Scheme I in which I @ . ) is the ionisation potential of X-; A (X+)xy is the appearance potential of X+ from the molecule X Y ; and D (X-Y) and D (X-Y)+ are the dissociation energies of the molecule and molecular ion. This gives the following relationships (3) and (4) in which El and E2 represent excess of kinetic energy J.H. Beynon and A. E. Fontaine Z. Nururforsch. 1967 22 334 or refs. therein. 318 Chambers Glockling and Light produced in the dissociation process together with any excitation energy of the products X+ and Y.. A similar cycle can be written for the formation of Y+ and X*. There is a considerable body of evidence1 that if I (X.) <I (Y*) then the ion X+ and radical Yo are formed without excess of energy; El and E may then be taken as zero but in general bond energies derived from appearance potentials are likely to be upper limits. The appearance potentials of metal ions M+ in the spectra of many organometallic compounds have also been used to determine heats of formation and mean metal-carbon bond energies. The validity of using appearance-potential measurements for the derivation of thermodynamic quantities depends on the correct assignment of the process involved in the threshold measurement and in general more reliance can be placed on values derived from appearance potentials of ions formed by less extensive dissociative processes.Ionisation potentials of organometallic compounds obtained from electron- impact measurements decrease as the size of the attached organic group increases. For methyl derivatives of the main-group metals ionisation requires less energy as the atomic number of the metal atom increases.' For peduoroalkylarsines* variations of the observed ionisation potentials have been correlated with the inductive and electron-withdrawing effects of groups attached to arsenic. Ionisation potentials of metal carbonyls are approximately the same as those of the free metal which suggests that the electron is removed from an essentially non-bonding 0rbita1.~ This is further substantiated by mean metal-carbon bond dissociation energies which are slightly lower in the molecular ion than in the molecule (Table 1).Ionisation potentials of biscyclopentadienyl metal complexes also lead to the conclusion that an electron is removed from a non-bonding orbital localised Table 1 Mean metal-carbon bond dissociation energies in neutral molecules and molecular ions (kcal. mole-l) (from ref. 9) E(M-CO)+ E(M-CO) Thernio a chemical Ni(CO)& 37 43 35.2 FdCO) 32 33 27-7 Cr(C0)6 31 31 27.1 w(c0)6 48 50 42-1 Mo(C0)6 39 42 35.9 VCO) 31 34 UF. A. Cotton A. K. Fischer and G. Wilkinson,J. Amer. Chem. SOC. 1959 81 800. R. E.Winters and R. W. Kiser J. Organometallic Chem. 1967 10 7 or refs. therein. D. R. Bidinosti and N. S. McIntyre Canad. J. Chem. 1967,45 641 or refs. therein. * W. R. Cullen and D. C. Frost Canad. J. Chem. 1962,40,390. 319 Mass Spectra of Orgammetallic Compounds on the metal and this is in agreement with the calculated molecular orbital energy levels in ferrocene.1° Comparatively few absolute bond dissociation energies have been determined by electron-impact methods since ionisation potentials of organometallic radicals are mostly unknown. The ionisation potential of the Mn(CO),. radical (9.26 ev) has been determined directly and combined with the appearance potential of Mn(CO),+ leads to a value of 19-22 kcal. moleq1 for D [(CO),Mn-Mn(CO),]. Although this is the energy to form two Mn(CO),.radicals the radicals may not have the same geometrical configuration.ll Ther- mal decomposition of iron pentacarbonyl at a high source temperature (250- 300") produces Fe(CO),* radicals and their ionisation potential (8.48 ev) can be measured. At low source temperatures (60-70") the appearance potential of the ion Fe(CO),+ may be found giving D [(CO),Fe-CO] as 6-1-6.9 kcal. mole-1.12 Mean metal-carbon bond dissociation energies for metal carbonyls obtained from the appearance potential of the metal ion assuming its formation to be as in eqn. ( 5 ) are shown in Table 1. The differences from thermochemically determined bond energies may be ascribed to the fragmentation process occurring with release of excess of energy. Average metal-ring bond energies in biscyclo- pentadienyl complexes have been derived in the same wayl0,l3 [E (M-C,H,) = 90 f 6 (V Cr); 72 f 6 (Mn); 75 f 6 (Fe); 112 f 6 (Ru); 79 f 6 (Co); 71 f 6 (Ni) kcal.mole-l]. The values for ferrocene and nickelocene are in good agreement with the thermochemically determined value of 77 kcal. mole-l. Heats of formation of molecules can be calculated from appearance-potential measurements if the heat of formation of the ligand or ligands is known. Applied to biscyclopentadienyls the values agree well with calorimetric measurements.1° There are some examples where appearance potentials provide evidence for a particular decomposition process occurring at the threshold energy. For example in cyclopentadienylmetal carbonyls C,H,M(CO) (M = Coy Mn V) the appearance potential of the metal ion suggests that the overall process leading to its formation is (6).14 M(CO) + e + M+ + nCO + 2e (5) (C,H,)M(CO) + e M+ + C3H3 + C,H2 + nCO + 2e (6) Appearance potentials of trimethylsilyl ions Me,Si+ derived from various trimethylsilyl compounds have been determined but it is only recently that dissociation energies of Me$-X bonds have been e~a1uated.l~ The ionisation potential of trimethylsilyl radicals (7.1 ev) has been determined from the appearance potential of the trimethylsilyl ion derived from hexamethyldisilane 10 J.Miiller and L. D'Or J. Organometallic Chem. 1967 10 313 or refs. therein. l1 L. I. B. Haines D. Hopgood and A. J. Poe J. Chem. SOC. (A) 1968 421 or refs. therein. l2 S. Pignataro and F. P. Lossing J. Organometallic Chem. 1968 11 571 1967 10 531 or refs. therein. l3 H.S. Hull A. F. Reid and A. G. Tunbull Inorg. Chem. 1967 6 805 or refs. therein. l4 R. E. Winters and R. W. Kiser J. Organometallic Chem. 1965 4 190. l5 S. J. Band I. M. T. Davidson I. L. Stephenson and C. A. Lambert Chem. Comm. 1967 723 and refs. therein. 320 Chambers Glockling and Light and the kinetically determined dissociation energy of the Si-Si bond in Me6Si2. An alternative method of determining ionisation potentials and heats of formation of organometallic radicals is applicable to metal-metal compounds such as Me,Sn.SnMe3.l6 The heat of formation of the ion SnMe,+ is first found from the appearance potential of SnMe,+ derived from a trimethylstannyl compound whose standard heat of formation is known. The heat of formation of the trimethylstannyl radical can then be obtained from the appearance potential of SnMe,+ derived from hexamethyldistannane and the heat of formation of hexamethyldistannane (7).The ionisation potential of SnMe * [AHf(SnMe,+)- AHf(SnMe,.)] has been estimated as 6.54 ev16 and that of the triphenyltin radical as 5-51 f 0.5 e ~ ? ~ A(SnMe3+)sn2Me8 = AHf(SnMe,+) + AHf(SnMe,*) - AHf(Sn,Me,) In cases where the ionisation potential of organometallic radicals is unknown it is still possible to compare bond dissociation energies from appearance potential differences and it has been shown for example that the Ge-Sn bond strength in Ph,Sn-GeMe is 12 f 3.2 kcal. mole-l stronger than the phenyltin bond in tetra~henyltin.~ Bond energies in molecular ions are obtained directly from eqn. (3) by measuring the ionisation potential of the molecular ion I ( X Y ) in addition to A(X+)xy.Illustrative examples are given in Table 2 and for mono- nuclear metal carbonyls wide variations have been reported; these may be due to thermal decomposition at the source temperature used.12 (7) Table 2 Approximate bond-dissociation energies in molecular ions (kcal. mole-l)* V Cr Mn Fe Co Ni Mo W D(C5H,M+-C5H5)10*12 123 151 87 153 180 125 D( C6H6M+-c6H6)12 81 D(C5H5MS-C&6)l2 122 D [(co).M+-co 39 11-28 5-34 16 22-34 32 D(MeM+-Me) 45(Be)17 53(Zn)7 30(Hg)7 D(Me,M+-Me)7 23(Al) 58(Sb) D(E t ,Sn+-E t) 14 D(PhEt,Sn+-Ph) 10 *For references see text. General Discussions of Spectra The importance of mass spectroscopic studies on organometallic compounds apart from the derivation of thermodynamic quantities lies in establishing molecular formulae and providing information on molecular structure.Most of the following discussion is concerned with the fragmentation behaviour of different types of compound. It is only when a number of compounds of a given type have been examined that one can proceed with any confidence to A. L. Yergey and F. W. Lampe J. Amer. Chem. SOC. 1965,87,4204 or refs. therein. 321 Mass Spectra of Organometallic Compounds assign a unique structure when various isomers are possible and this is the main justification for the empirical approach to mass spectrometry. For example the digermane Et,(C,H,,)Ge shows a metastable transition for the elimination of hexene which confirms its constitution. However in the absence of a background knowledge of how alkyldigermanes fragment elimination of hexene could be attributed to the isomeric structure Et,(Bu),Ge,.17 Many of the differences between electron-impact behaviour of organometallic and organic compounds arise from the low M-C and M-H bond strengths compared with those of C-C and C-H.The greater electronegativity of carbon and hydrogen than of metals means that when a positive ion decomposes the charge is likely to remain with the metal-containing fragment and for most organometallic com- pounds a high proportion of the ion-current is cariied by metal-containing species. There are nevertheless examples of transitions in which a metal-containing fragmelit is eliminated either as a neutral molecule or a radical. Compounds in which hydrocarbon ions feature extensively (e.g. mercury- and beryllium- alkyls) are those which decompose thermally at the source temperature before ionisation.Hence the experimental conditions under which spectra are recorded can profoundly affect ion abundances. There are instances where the ionisation potential of a hydrocarbon radical is lower than that of the metal and this may account for the high abundance of the C7H,+ (tropylium) ion in the mass spectrum of tetrabenzylgermane (equation 8). C,H,Ge+ -> C,H,+ + Ge (8) The question of whether a bonding or non-bonding electron is removed in the primary ionisation process is not clearly reflected in ion abundances. Organo-derivatives of the Group IVB elements generally show low-abundance molecular ions compatible with removal of an electron from an M-C bond. However molecular ions are of high abundance in beryllium alkyls where it must be a bonding electron which is removed.Mercury zinc and aluminium alkyls also produce abundant molecular ions. With main-group organometallic compounds molecular ions usually decompose by radical elimination thereby producing even-electron ions R4M + e -+ R4M+. -F R,M+ + R. (9) e.g. reaction (9). In subsequent fragmentation processes the dominant feature (as in organic compounds) is the tendency to form further even-electron ions. Thus R,M+ ions (M = Si,Ge,Sn,Pb) lose alkene or R molecules to a greater extent than a further radical which would give the odd-electron ion R,M+. (reaction 10). l7 D. B. Chambers G. E. Coates F. Glockling J. R. C. Light and P. D. Roberts unpublished observations. 322 Chambers Glockling and Light Although there is a paucity of bond-energy data it appears that with unsymmetrical organometallic compounds of Group IV the weakest bond in the molecular ion is the same as in the molecule and this is the bond most susceptible to cleavage.The abundance of metal ions M+ depends on the strengths of M-C M-H and M-M bonds as well as on the ionisation potential of the metal and among the Group IV metals these ions are only abundant for tin and lead where bond energies and ionisation potentials are low. The variety and abundance of metal hydride ions formed from almost all types of organometallic compound is quite striking especially with metals such as mercury and lead. Transition-metal complexes such as carbonyls and cyclopentadienyls often show molecular ions of high abundance and the fragmentation behaviour is not dominated by odd- and even-electron considerations.This difference may be accounted for if primary ionisation involves the removal of an electron from an essentially non-bonding orbital. Relative metal-carbon bond strengths profoundly influence fragmentation as in 7.r-cyclopentadienylmetal carbonyls which fragment by successive loss of carbon monoxide before cleavage or fragmentation of the cyclopentadienyl group ; [D(n-C5H5-M) > D(M-CO) 1. Similarly in metal-metal bonded carbonyls the way the M-M bond strength increases with atomic weight in series such as Mn < MnRe < Re is reflected in the proportion of the ion-current carried by ions having the M-M bond intact. The extent to which structures can be assigned to ions is quite limited. The high energy of the electron beam (normally 70 ev) means that unimolecular rearrangement processes readily occur and these can invalidate structural conclusions.Metal-metal bonded compounds are particularly prone to undergo rearrangement processes especially in compounds where the metal-metal bond is strong relative to others in the molecule. Despite these limitations mass spectrometry has provided substantiating evidence for novel structures in addition to being the most definitive method of establishing molecular compositions. Main-group Metals Lithium-The mass spectra of lithium alkyls are of interest since their vapours are associated. None of the compounds so far reported18 produces a molecular ion but the identity of parent ions can be deduced from appearance-potential measurements of fragment ions. In this way it has been demonstrated that ethyl-lithium vapour consists of tetramer and hexamer units both of which decompose (1 1) upon ionisation.Similar observations show that lithiomethyl- trimethylsilane LiCH,SiMe, vapour is tetrameric. Et,Li + e __+ (Et,-,Li,)+ + Et - + 2e (1 1) In the mass spectra of lithium alkyls the most abundant ions are Li,R+; Is G. E. Hartwell and T. L. Brown Inorg. Chem. 1966 5 1257 and refs. therein. 323 Mass Spectra of Organometallic Compounds ion abundances and appearance potentials suggest that the decomposition processes (12) are most favourable when x is even. Beryllium Magnesium Zinc and Mercury.-This group of metal alkyls shows many trends which reflect changes in ionisation potential and M-C and M-H bond energies. With the exception of the di-t-butyl all beryllium alkyls are associated through Be-C-Be bridging-bonds usually forming cyclic dimers in the vapour state.The low energy of these bridge-bonds (probably less than 10 kcal. per C-Be) means that only under the most carefully controlled spectrometer conditions are dimer or trimer ions observed. The abundance of all beryllium-containing ions but especially dimeric and trimeric species falls with increasing source temperature owing to dissociation of bringing bonds and thermal decomposition before ionisation.17 Mass spectrometry has provided the only direct evidence that beryllium dialkyls (R = Et Prn Pri) are associated in the vapour phase.17 Diethyl- beryllium even shows the trimeric ion Et,Be,+ whilst Et,Be,+ is extremely abundant at low source temperature (< 50"). The fragmentation of beryllium alkyls reflects the greater stability of even-electron ions as in the sequences (13) and (14).The higher alkyls also show transitions in which C-C and C-H -Et* -GHi - GH' Et,Be,+. -> Et,Be,+ -> Et,Be,H+ -+ EtBe,H,+ R,Be+- + RBe+ + R. (1 3) (14) bonds are cleaved as in (15) and (16). Alkane elimination from odd-electron But,Bef* -+ ButBeCMe,+ + Me. ButBe+ $ C,H - Hz - H* C,H,Be+ ___+ C,H,Be+ __+ C,H,Be+ monomer ions is a favoured process (17) in marked contrast to the behaviour (C,H,,+,),Be+* + CnH,,Be+- + CnH2n+2 (17) of Group IVB metal alkyls. With di-isopropylberyllium the C,H,Be+' ion is the most abundant in the spectrum but appearance-potential measurements suggest that its formation is accompanied by considerable rearrangement possibly to an allylberyllium hydride as in (18).Alkene elimination from the monomer 324 Chambers Glockling and Light is always observed although the hydride ion produced is usually of low abund- ance. Where the ionisation potential of a hydrocarbon radical is less than that of beryllium elimination (19) of the metal may be observed. C3H3+ + C3H,+ + Be (19) [IC,H;) = 8.25; I(Be) = 9.32evI Biscyclopentadienylmagnesium'1° is mentioned under transition metals. Several zinc7J7 and mercury7 alkyls have been examined. They form molecular ions of high abundance and with mercury alkyls (Me Et Bun) hydrocarbon ions make a large contribution to the total ion current though part of this may arise from thermal decomposition before ionisation. Low-abundance hydride ions (MH+ and RMHf) are observed but the major processes are simple bond cleavages (20).R,M+. + RM+ __+ M+- (20) Aluminium.-Aluminium alkyls are also associated in the vapour phase and dimeric ions have been detected for Me,A17,17 at low source temperature but not for Et3A1. By contrast compounds with strongly bridging groups19 such as (Et,AlOEt) and (R,AlNCPh,) produce parent dimer ions in high abundance even at a source temperature of 200". but metastable peaks are observed for more complex rearrangement processes17 such as (21) and (22). Dimethylaluminium hydride vapour contains dimeric Trimethylaluminium fragments mainly by successive loss of methyl radicals H2A1+ + C2H4 Al+ + C,H Me,Al+ and trimeric species at SO" but its mass spectrum shows only a minute amount of trimer ion Me,Al,H,+ whereas dimer ions presumably hydrogen-bridged are of high abundance.17 These fragment partly by elimination (23,24) of methylaluminium hydride molecules.Me,Al,H,+ + Me,Al+ + MeAlH Me,Al,H+ __+ Me,Al+ + Me,AlH (23) (24) Silicon Germanium Tin and Lead.-Organo-derivatives of ~ i l i ~ ~ n g e r ~ n a n i u m ~ ~ ~ ~ . ~ ~ tin4,7,16s20 and lead4*7y20 have been the subject of lS K. Wade and B. K. Wyatt J. Chem. SOC. (A) 1967 1339. 2o J. J. Ridder and G . Dijkstra Rec. Trav. chim. 1967 86 737. 21 C. A. Hird Analyt. Chem. 1961 33 1786. 22 G. Fritz H. J. Buhl and D. Kummer 2. anorg. Chem. 1964 327 165 and refs. therein. 23 F. Aulinger Colloq. Spectros. Intern. Sth Lucerne Switzerland 1960 1959 267 and refs. therein. 24 J. Silbiger C. Lifshitz J. Fuchs and Mandelbaum J. Amer. Chem. SOC. 1967 89,4308 and refs. therein.325 Mass Spectra of Organometallic Compounds numerous studies. Their mass spectra have many features in common; the differences observed are qualitatively related to the decrease in ionisation potential of the metal with increasing atomic number and to a progressive decrease in the metal-carbon and -hydrogen bond strengths. For example metal hydride ions are least abundant with lead and tetraethyl-lead eliminates butane from the Et,Pb+ ion whereas a similar process is rarely observed with the lighter Group IVB metal alkyls. Molecular ions are usually of low abundance and radical elimination by M-X bond cleavage (25) is a dominant decomposition path for molecular and other odd-electron ions. R,M+* -> R3M+ 3- R * (25) In unsymmetrical molecules such as R,MR’ the relative probability of the various bond-cleavage processes depends on the bond strengths in the molecule ion and on the relative stabilities of the radicals and ions produced.Even so it appears that the radical most readily eliminated is that which is most weakly bonded to the metal in the neutral molecule. For example in Me,SnEt and Ph,SnEt the major ion is produced by ethyl loss from the parent whilst in Me2SiC12 the ion MeSiCl,+ is more abundant than Me,SiCI+. The relative bond dissociation energies are D(Me,Sn-Me) > D(Me,Sn-Et) D(Me,Sn-Ph) > D(Me3Sn-Et),16 D(Ph,Sn-Ph) > D(Ph,Sn-Et),4 and D(Me,Si-CI) > D(Me,Si-Me).15 Similarly in metal-metal bonded compounds of the type R,M.M’R’ (M M’ = Si Ge Sn) extensive cleavage of the metal-metal bond occurs only when it is the weakest in the neutral molecule.For R,M-M’Ph compounds the Ph,M‘+ion is always the more abundant irrespective of the metal. This may be ascribed to the combined effect of various factors the ionisation potential of Ph,M’ radicals may be lower than those of R,M*; charge delocalisation over the phenyl groups will stabilise the Ph,M’+ ion and there are fewer low-energy decomposition routes available €or Ph,M’ + than for R,M+. In methylpolygermanes Me2n+2Gen (n = 2 4 5 or 6) loss of a methyl radical is followed by successive elimination of Me,Ge units. Whereas cyclic organosilanes (Me,SiCH& produce only one major fragment ion due to loss of methyl from the parent linear silanes of the type Me3SiCH2(SiMe2CHz)p2- SiMe also fragment by cleavage of methylene-silicon bonds. Radical elimination from even-electron ions is rarely observed and only a few examples are supported by metastable peaks such as (26) and (27).Ph,GeMe+ - i Ph,Ge+. + Me- (26) Fragmentation by cleavage of C-C and C-H bonds is most apparent with organosilanes although tolylgermanes produce ions owing to methyl loss. In R,M compounds doubly- and triply-charged ions are either absent or of low abundance but with methylsiloxanes silazanes and related compounds having more than one metal atom they are quite intense indicating the removal of a 326 Chambers GIockling and Light methyl group together with one electron per silicon atom. In germoxanes (R,Ge),O doubly charged ions are more apparent than in digermanes R,Ge,. Alkene elimination is an important fragmentation process for all a-bonded metal alkyls frequently giving strong metastable ions.It is a general reaction for even-electron ions containing the grouping R,CHCH,M and is probably a p-elimination process (28). -C& -C&h -CzK Et,M+ -+ Et2MHS .-* EtMH,+ + MH,f H' Alkene elimination from odd-electron parent ions has not been observed; on mechanism (28) it requires a pentaco-ordinate intermediate which may be energetically unfavourable compared with radical elimination. More complex alkene-elimination processes (29) are shown by benzyl and methyl compounds. (RCH,),Ge+ -+ RCH,GeH,+ + RCH=CHR (R - H Ph) (29) Molecule elimination by cleavage of two M-X bonds is most favourable for the heavier metals and comparatively few organosilanes show metastable peaks for alkane or hydrogen elimination. This again is a reflection of M-C and M-H bond energy changes in the group.The reactions (30-35) show the range of neutral molecules formed in this way. For reactions such as Et3Pb+ -+ EtPb+ + C4HIo Ph3Mf ,-* PhM+ + Ph2 Ph2MH+ -* PhM+ 4- C6H6 Ph,MCl+ .-+ PhM+ + PhCl (PhCH,),Me,GeSi+ ~-* PhCH,Ge+ + Me3SiCH2Ph (PhCH&,GeD+ -* PhCH2Ge+ + PhCH,D RMH,+,-* RM+ + H2 (32) it is almost certainly the M-H bond which is cleaved since this decom- posit ion (3 6) is met as t able-confinned for t ribenz yldeu t er i ogermane . This type of fragmentation is sometimes observed from odd-electron ions when the pro- duct (also odd-electron) ions are often of high abundance as in reactions (37-40). Cleavage of one M-X bond in even-electron metal aryl ions (41) occurs to a slight extent. A similar decomposition involving methane loss is observed with methylsilazanes and cyclic silicon-methylene compounds.The only odd-electron ions showing a parallel process (42) are the triphenyltin halides. Ph,M+- __+ Ph2M+* + Ph Et,GeH+* 0_3 Et,Ge+- + QH MeSiH+. + Si+. + CH Ph,MX+* + Ph,+. + PhMX 327 Mass Spectra of Organometallic Compounds Many low-abundance ions are formed by processes which do not involve cleavage of bonds to the metal. For example the higher alkyls eliminate smaller alkene fragments (43). Lead allcyls in particular show abundant ions formed by rearrangement of alkyl groups and in tetra-n-propyl-lead the MePb+ ion is the most abundant in the spectrum. In some cases it is not possible to determine whether metal-carbon bonds are cleaved. For example dimethylsilacyelohexane produces an abundant ion owing to loss of ethylene (44).Similarly phenylmetal ions degrade partly by successive elimination (49 (46) of acetylene Acetylene loss is a high-energy process which disappears at 20 ev and diminishes in the order Si > Ge > Sn > Pb. The extent of hydrogen-molecule elimination (47) (48) follows the same order and may be ascribed to the progressive decrease in M-C bond strength. Ph,M+ + Ci,Hi&+ + H2 PhMf ___+ C,H,M+ + H2 (47) (48) Methylene elimination from the trimethyltin ion is a metastable-supported process and tetraethyl derivatives of germanium tin and lead show low-abundance ions which are best ascribed to elimination of methylene (49). RzMEtf- + R2MCH3' + CH (R = Et Ph) (49) Arsenic Antimony and Bismu th.-The trip h e n ~ l - ~ and t r imet hy 1-der iva t iveP have been examined and molecular ion abundances diminish in the order As > Sb > Bi whereas the metal-ion abundances M+ follow the reverse order as expected.Loss of H (or 2H *) from R2M+ ions is observed for the arsenic and antimony compounds but not with bismuth. Transition Metals The mass spectra of organotransition metal compounds are classified according to the types of ligand present since these produce the main characteristics of the spectra the influence of the metal being more subtle. 25 D. E. Bublitz and A. W. Baker J. Organometallic Chem. 1967 9 383. 2363. R. G. Kostyanovsky and V. V. Yakshin Zzvest. Akad. Nauk. S.SS.R. Ser. khim. 1967 328 Chambers Glockling and Light Carbony1s.-The mononuclear carbonyls of iron nickel chromium molybdenum tungsten and ~ a n a d i u m ~ ~ ~ ~ ~ ~ show prominent molecular ions decomposing by successive loss (50) of carbon monoxide.Appearance-potential measurements indicate that electron-impact ionisation of carbonyls occurs by removal of an essentially non-bonding electron and stepwise loss of carbonyl groups is often confirmed by the presence of metastable peaks.27 Metal ions M+ are commonly the most abundant in the spectra but doubly-charged molecular- and fragment- ions are also formed and these decompose by stepwise loss (51) of carbon m o n ~ x i d e . ~ ~ ~ M(CO),2+ + M(C0),-12+ + CO (51) The heavier metal carbonyls show carbide ions M(CO),C+ and these decompose by loss (52) of ~ a r b o n y l . ~ J ~ ~ ~ ~ ~ ~ M(CO),C+ 4 M(CO),-,C+ + CO (52) Differences in the fragmentation of bi- and poly-nuclear carbonyls provides evidence about the extent of metal-metal bonding.For purely metal-metal bonded carbonyls [such as Mn,(CO),, Re2(CO) and ReMn(CO) 0]12,28 a high proportion of the ion current is carried by ions containing the metal-metal bond and this increases with the atomic weight of the metal (Table 3). This trend and the observation that doubly-charged dimetallic ions are more abundant for the heavier metals reveals the greater ability of the heavier metals to sustain metal-metal bonding in higher oxidation states. Of this series only manganese has a metastable peak for cleavage (53) of the M-M bond. Mn,(CO),,+ ___+ Mn(CO),+ + Mn(CO) (53) Compounds with bridging carbonyl groups and only weak metal-metal interaction show extensive cleavage to mononuclear ions. For example in Fe,(CO) the ions Fe,(CO),-,+ are all of low abundance and the spectrum is dominated by the ions Fe(CO),-,+.In dicobalt octacarbonyl although there are bridging carbonyl groups the metal-metal interaction is stronger and dimetallic species are more abundant .12,28 Differences in bonding are implied by the mass spectra of the carbonyls M3(CO),2 (M = Fe Ru 0s)l2 (Table 3) and only the iron complex has bridging carbonyl groups. In CO,(CO), a high proportion of the ion-current is carried by CO species.12 Osmium tetroxide and Os,(CO), form a complex Os,O,(CO), which has been identified solely from its mass spectrum. The molecular ion forms the base peak and ions due to successive loss of all 12 carbonyl groups are of high abundance but the corresponding doubly-charged ions are absent.However doubly-charged ions OS,(CO),~+ (n = 0 - 1 1 ) are present but not the singly- 27 R. E. Winters and J. H. Collins J. Phys. Chem. 1966 70 2057 or refs. therein. 28 R. E. Winters and R. W. Kiser J. Phys. Chem. 1965 69 1618 or refs. therein. 329 Mass Spectra of Organometallic Compounds Table 3 Relative ion-abundmces of M, M, and M species (from ref. 12) M3 MZ M Mn,(CQ),o - 59 41 ReMn(CQ)l - 71 29 96 4 Fe3(C0)12 36 4 60 Re2(CO)10 - Ru3(C0)12 93 5 2 OS,(CO),~ 100 0 0 charged species. These unusual observations can be explained by elimination (54) of the negative Os0,- ion and subsequent carbonyl e l i m i n a t i ~ n . ~ ~ ~ ~ Hydrocarbon Complexes.-Bis-r-cyclopentadienyl compounds have been ex- tensively studied (Fe Co Ni V Cr Ru Re 0s).lo 30 The molecular ion is always of high abundance and metal-ring bond cleavage processes also produce ions of high abundance (Scheme 2).Fragmentation of the cyclopentadienyl Scheme 2 groups occurs mainly by elimination of acetylene but minor ions are produced by loss (54a) (55) of Ha and CH,. and other hydrocarbons. The isoelectronic cyclopentadienyls of Fe Ru and Os and the isoelectronic hydrides (C,H,),TcH (C,H,),ReH and (C,H,),WH,1° produce similar abundances for ion groups containing the same number of carbon atoms. Some biscyclo- pentadienyls show ions at higher mass than the molecular ion which have been ascribed to ion-molecule reactions (56). Examination of the mass spectrum of a mixture of ferrocene and nickelocene reveals ions such as (C,H,),FeNif whilst ferrocene also shows ions resulting from fragmentation of (C5H5),Fe2+.B. F. G. Johnson J. Lewis I. G. Williams and J. Wilson Chem. Comm. 1966 391. ao F. W. McLafferty Analyt. Chem. 1956 28 306.. 330 Chambers Glockling and Light (56) The ionic cyclopentadienyls (C,H,),Mn and (C,H,),Mg produce similar fragment ions to those of then-bonded complexes although in very different relative abundance indicative of the structural differenceslO [(C,H,),Mn + 19; C,H,Mn+ 29; Mn+ 25%; (C,H&Mg+ 21; C5H,Mg+ 38; Mg+ 31x1. With increasing ring substitution in ferrocene the abundance of metal-containing ions dim in is he^.^^ The weaker metal-ring bonding in benzene complexes is probably reflected in the isoelectronic series (C,H,),Fe C,H,MnC,H, and (C,&),Cr for which the most abundant metal-containing species are (C,H,),Fe+ C,H,Mn+ and Cr+.In bisbenzenechromium the abundance of ions produced by fragmentation of benzene is high and the doubly-charged molecular ion [a chromium(I1) species ] has also been reported.lo,12 Mass spectrometry has proved important in the characterisation of more complex cyclopentadienyls such as (I) for which the molecular ion forms the base peak and metastable transitions are observed for successive loss of two methyl groups.32 Similarly the zirconium oxide complex33 [(C,H,),ZrCl ],O gives many ions containing the Zr,O unit. Me i e Fe Some n-ally1 complexes are too unstable to show molecular ions under electron-impact and field-ionisation mass spectrometry has been used to minimise fragmentation processes. Electron-impact fragmentation of the bis-n-ally1 complexes of nickel causes ethylene elimination and loss of ally1 radicals from the parent whereas in the platinum analogue elimination of propene predominate^.^^ 31 D.T. Roberts W. F. Little and M. M. Bursey J. Amer. Chem. SOC. 1967 89 4917 or refs. therein. 32 W. E. Watts J. Organometallic Chem. 1967 10 191 or refs. therein. 33 A. F. Reid J. S. Shannon J. H. Swann and P. C. Wades Austral. J. Chem. 1965 18 173. 34 J. K. Becconsall B. E. Job and S. O'Brien J. Chem. SOC. (A) 1967,423 331 Mass Spectra of Organometallic Compounds Carbonyl Hydrides.-In the mass spectra of metal carbonyl hydrides hydrogen is strongly retained by parent and fragment ions. Manganese carbonyl hydride shows two series of ions HMn(CO),-,+ and Mn(CO),-,+ in which elimination of a hydrogen radical competes with loss of carbon monoxide.12 Polynuclear carbonyl hydrides lose carbon monoxide more readily than hydrogen as in H,Mn,(CO), where the base peak is due to H,Mn,(CO),+.In similar rhenium and ruthenium hydrides [H,Re,(CO), and H,Ru,(CO), 3 the molecular ions are of high Polynuclear carbonyl hydrides pose an acute analytical problem and the number of hydrogen atoms attached to a metal cluster cannot usually be predicted. For most compounds of this type molecular compositions are convincingly established from their mass spectra. In the complex borohydride H,B2Mn,(CO),, six of the hydrogen atoms form bridging B-H-Mn bonds whilst the other occupies a bridging position across an Mn-Mn bond. Its mass spectrum shows that all of the hydrogen atoms are retained until four carbonyl groups have been lost and in general it appears that bridging hydrogen atoms are even more strongly retained than those occupying terminal positions.36 The structures of several unusual compounds derive support from their mass spectra.For example the iron and ruthenium complexes HMCo,(CO), (M = Fe Ru) have been assigned structure (11) in which the hydrogen atom is enclosed in the metal cage. Their mass spectra show the molecular ion followed by successive loss of six carbonyl groups. Slight hydrogen loss is then observed while the remaining six carbonyl groups are cleaved. Only when the tetrahedron of metal atoms CO,M is broken does hydrogen loss become an important process. Although hydrogen is initially bonded to iron or ruthenium low abundance cobalt hydride ions (HCo,+ and HCo,+) are observed.,’ Hydrocarbon-Metal Carbony1s.-The mass spectra of complexes having different ligands bonded to a transition metal reflect differences in metal-ligand 35 B.F. G. Johnson R. D. Johnston J. Lewis and B. H. Robinson J. Organometallic Chem. 1967 10 105 or refs. therein. 36 J. M. Smith K. Mehner and H. D. Kaesz J. Amer. Chem. SOC. 1967 89 1759 or refs. therein. 37 M. J. Mays and R. N. F. Simpson Chem. Comm. 1967 1024 or refs. therein. 332 Chambers Glockling and Light bond strengths. In cyclopentadienylmetal carbonyls the carbonyl groups are lost before cyclopentadienyl although in some cases the ion MCO+ can be detected.14 Cyclobutadieneiron tricarbonyl is similar in that carbonyl groups are lost before fragmentation or cleavage of cyclob~tadiene.~~ In benzenechromium tricarbonyl the ions Cr(CO),-,+ are of low abundance whilst in cyclopenta- dienylnickel nitrosyl the ion NiNO+ is only 1.5 % as abundant as the base peak C,H5Ni+.12 In C,H,Mo(CO),NO both carbonyl groups are lost before the nitrosyl In molybdenum and tungsten complexes such as Et,GeM(CO),C,H a high proportion of the ion cuIrent is carried by species containing the GeM group and molecular ions are of high abundance.The first fragment ions are those formed by loss of carbonyl or an ethyl radical and further degradation (57) takes place by ethylene and carbonyl elimination. Et,GeM(CO),C,H,f -+ Et,GeM(CO),C,H,+ and Et,GeM(CO),C,H,+ (57) A most unusual feature in the spectrum of the trimethylgermyltungsten complex is that the parent ion decomposes by loss of a methyl iadical or carbon dioxide40 (Scheme 3). The influence of bridging carbonyl groups on the proportion of Scheme 3 Me,GeW(CO),C,H,+ + Me* Me,GeW(C,O)C,H,+ + CO (or CO + 0) Me,GeW(CO),C,H,+ 7 dimetallic ions has been referred to and further illustrations are provided by the dimeric compounds [C,H,Mo(CO),] and [C,H,Fe(CO) 1,.In the former the most abundant ion is (C,H,),Mo,+ whereas in the carbonyl-bridged iron complex dimer ions are of low abundance and the base peak is C5H,Fe+.l2 These complexes form ion-molecule associates as with ferrocene.1° In cyclo- hexadieneiron tricarbonyl loss of molecular hydrogen is a significant process probably yielding wbenzenecarbonyl ions.39 Carbonyl Halides.-Manganese- and rhenium-carbonyl halides (CO),MX and Fe(CO),I show comparable loss of carbonyl and halide from the molecular ions although MXf ions are of considerable abundance.In halogen-bridged dimers such as (CO),Mn,X the Mn,X unit persists unfragmented until loss of carbonyl groups is complete. As with the hydrides there seems to be an appreciable difference between ease of elimination of terminal and bridging liga11ds.1~~~~ Carbonyl-phosphorus Complexes.-Differences in the donor-acceptor properties of phosphines relative to carbon monoxide influence fragmentation behaviour s8 R. G. Amiet P. C. Reeves and R. Pettit Chem. Comm. 1967 1208. 39 M. A. Haas and J. M. Wilson J . Chem. SOC. (B) 1968 104 or refs. therein. 40 A. Carrick and F. Glockling J. Chem. SOC. (A) 1968 913. 333 Mass Spectra of Organometallic Compounds considerably. In carbonyltriphenylphosphine complexes of molybdenum and tungsten Ph,PM(CO), the triphenylphosphine group is eliminated only after complete loss of carbon monoxide and there is little fragmenta- tion of the phosphine.By contrast the chelating phosphine complexes (Ph,PCH,CH,.PPh,)M(CO) lose ethylene forming the ion (Ph,P),M+. In biiiuclear complexes with tertiary phosphines [R,PMn(CO) I, the abundance of dimetallic species is low compared with unsubstituted binuclear carbonyls and arylphosphines enhance this effect more than alkylphosphines. Binuclear phosphite complexes [(RO),PMn(CO) 1 lose carbonyl groups more readily than phosphite and metastable peaks are apparent for the loss of four and two carbonyl groups1 (Scheme 4). Scheme 4 L,Mn,(CO),,+ + 2CO L,Mn,+ + 4CO L,Mn,(CO),+ ’ Several phosphino-bridged complexes of Cr Fe Mn Mo and W have been examined (111 and IV) and loss of carbonyl groups leaves the M,P,Me nucleus intact.For compounds of the type (IV) mononuclear metal ions present in low abundance are probably formed by the process (57).41 Me2 [(CO),MPMe,],+--+(CO),M(PMe,),+ + M(CO) (57) (IV) Carbonyl-sulphur Complexes.-Sulphur is a very strong ligand to many transition metals and the trinuclear complex Fe,(CO),S loses all the carbonyl groups leav- ing the ion Fe,S,+ as the base peak of the spectrum.42 Similarly thio-bridged metal carbonyls such as Fe2(C0)6S2R2 Mn,(CO)8S,R, and Re,(CO)8S,Ph resemble phosphorus and halide analogues in retaining the M,S,R2 structure until loss of carbonyl groups is completed. Alkene loss is then observed if R is ethyl or butyl. Trimeric and tetrameric compounds show similar feature^.,^^^^ 41 B. F.G. Johnson J. Lewis J. M. Wilson and D. T. Thompson J. Chem. SOC. (A) 1967 1445 or refs. therein. 42 S. R. Smith R. A. Krause and G. 0. Dudek J. Inorg. Nuclear Chem. 1967 29 1533. 43 M. Ahmad G. R. Knox F. J. Preston and R. I. Reed Chem. Cornm. 1967 138. 334 Chambers Glockling and Light Other Substituents.-Elimination of nitric oxide from complexes such as Co(CO),NO and Fe(CO),(NO) competes with carbonyl loss at all stages. In the phosphorus trichloride complex Co(CO),(NO)PCl the molecular ion loses a chloride radical as well as the three neutral ligands.12 In transition-metal- Group IVB metal complexes like (Me,Sn),Ru(CO), a methyl radical is lost before a carbonyl groupg4 whereas in Ph,SnMn(CO) the ion of highest mass Ph,SnMn+ shows successive loss of phenyl radical^.^ Mass spectrometry has provided support for a number of unusual structures.The compound (PhN),Fe,(CO) formed in the reaction between Fe3(CO), and phenyl isocyanate shows the molecular ion and fragments involving loss of only six carbonyl groups followed by the elimination of phenyl isocyanate. This behaviour is completely compatible with structure (V).45 A furthei novel example is the ruthenium carbonyl carbide RU,(CO)~,C in which the ‘carbide’ atom lies close to the centre of an irregular octahedron of ruthenium atoms with both bridging and terminal carbonyl groups.45a The mass spectrum is understandable in terms of this structure since the molecular ion is observed followed by stepwise elimination of all 17 carbonyl groups giving finally the very abundant ion Ru,C+. No singly-charged ions of lower mass were pIesent but abundant doubly-charged ions Ru,C(CO),-,,2+ were observed.46 a-Bonded Complexes-Comparison of the mass spectrum of the a-benzyliron complex C,H,(CO),FeCH,Ph with that of the pentafluorophenyl analogue suggests that on ionisation only the former is transformed into then-bonded tropylium group., Methylmanganese pentacarbonyl loses a methyl radical and carbonyl groups competitively and several manganese hydride ions are produced in low abundance.The trifluoromethyl analogue behaves differently in that the molecular ion is far more abundant and the highest-mass fragment ion is formed by loss of F*. Rearrangement ions containing Mn-F bonds ale of appreciable abundance. Methylsulphonylmanganese pentacarbonyl MeSO,Mn(CO) does not show ions due to successive elimination of carbonyl groups most of the 44 J.D. Cotton S. A. R. Knox and F. G. A. Stone Chem. Comm. 1967,965. 45 J. A. J. Jarvis B. E. Job B. T. Kilbourn R. H. B. Mais P. G. Owston and P. F. Todd Chem. Comm. 1967 1149. 45a R. Mason and W. R. Robinson Chem. Comm. 1968 468. 46 B. F. G. Johnson R. D. Johnston and J. Lewis Chem. Comm. 1967 1057. 47 M . T. Bruce. J. Orgunometallic Chem. 1967 10,495 or refs. therein. 335 Mass Spectra of Organometallic Compounds ion current being carried by the species SO,Mn(CO),+ Mn(CO),+ Mn(CO),f and Mn+.,* Rearrangement Ions.-The fragmentation processes so far discussed involve simple bond cleavage with in some cases migration of a hydrogen atom. However as with organic electron-impact induced migration (58) of function groups is quite common.[ ,,’’‘c’’‘*s ] + ,c A - C + + B b - 4 + - A’____-__- B (5 8) Rearrangement ions are often of high abundance and this makes the use of mass spectrometry ambiguous for distinguishing isomers of for example A,M.MB and it also poses problems for ‘element mapping’ by computer.49 Compounds of the type A,M-M’B (M M’ = Si Ge Sn) show all possible rearrangement ions A,MB+ AMB,” MB,+ B,M’A+ BM’A,+ and M’A,+. Rearrangement may occur in the molecular ion or in fragment ions containing the metal-metal bond. It is also possible that rearrangement ions are formed by a synchronous process in which transfer or interchange of groups is accompanied by fission of the metal-metal bond. In the above series of compounds (M # M’) rearrangement ions are most abundant for silicon and least for tin.Migration reactions are observed in other types of metal-metal bonded compounds such as the iron-tin complex n-C,H,(CO),Fe.SnPh (Scheme 5). Scheme 5 PhFef + C,H,Sn. C,H,Sn+ + PhFe. C,H,Fe. SnPh+ Similarly the PhMnf ion is observed in the spectrum of Ph,Sn.Mn(CO),. Transfer of an initially n-bonded cyclopentadienyl group from a transition- to a non-transition-metal is a common process and for [n-C,H,(CO),Fe ],SnCl the cyclopentadienyltin ion is the most abundant in the spectrum.12 Trialkylgermyl-molybdenum and -tungsten complexes n-C,H,(CO),M.GeR behave similarly forming the ion C5H,Ge+.40 Transfer of carbonyl groups must occur in various rhenium carbonyls [Re,(CO),, (CQ),Re-Mn(CO), H3Re,(C0)4] since Re(CO),+ is formed in low a b u n d a n ~ e . l ~ ~ ~ ~ Phosphorus- sulphur- and oxygen-metal complexes show ions which result from rearrangement processes e.g.(59) (60). C,H,CrS,CrC,H,+ -+ (C5H,),Cr+ + CrS,12 (59) MeSO,Mn(CO),+ -+ MeMn(CO) + SO,,* (40) M. J. Mays and R. N. F. Simpson J. Chem. SOC. (A) 1967 1936 or refs. therein. 4i3 P. Brown and C. Djerassi Angew. Chem. Internat. Edn. 1966 6 477. 336 Chambers Glockling and Light Migration of groups from carbon to a metal has also been observed. For example Si-0 and Si-N compounds such as Me,Si.O.CH,Ph and Me3Si.NHCH2Ph produce phenylsilyl ions2 whilst substituted arene complexes of the type ~T-C,H;CORM(CO) (M = Mn Cr) form RMf and C,H,MRf ions.12 Rearrangement processes can also result in the formation of metal-metal bonds as in examples (61) (62) although thermal decomposition followed by ionisation may also O C C U ~ .~ ~ ~ ~ C,H5(CO)Fe(CO)2Fe(CO)C,H5+ -+ (C,H,),Fe,+ + (C,H,),Fe+ (61) (62) Hg [Mn(CO) 12+ - Mn,(CO)+,- Group-migration reactions are common in fluorocarbon-metal compounds presumably because of the high strength of metal-fluorine bonds. Tetrakis- (pentafluorophenyl)germane4 produces (C,F,),GeF+ in high abundance contrasting with the low abundance of Ph2GeH+ derived from tetraphenyl- germane. Migration of fluorine from carbon to metal produces high abundance ions in the mass spectra of fluorocarbon derivatives of ~ilicon,~ arsenic,5o iron,4s3g,47*51 mangane~e,~,,~~ and rhenium,12 and elimination of the neutral metal fluoride is often observed as in (63) and (64). (C,F,),Ge+ -+ C& + GeF. (F,C),AsCF2+ __+ C,F,+ + AsF (63) (64) 50 R. C. Dobbie and R. G. Cavell Inorg. Chem. 1967 6 1450. 51 R. B. King J. Amer. Chem. SOC. 1967 89 6368. 337

ISSN:0009-2681    

DOI:10.1039/QR9682200317

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

The Application of the Woodward-Hoffmann Orbital Symmetry Rules to Concerted Organic Reactions By G. B. Gill DEPARTMENT OF CHEMISTRY UNIVERSITY OF NOTTINGHAM NOTTINGHAM NG7 2RD Introduction Among the many and varied reactions of olefins certain processes may be properly classified as concerted reactions. The importance of these reactions in organic synthesis is beyond doubt for included within the definition ‘concerted’ are the well-known Diels-Alder and Cope reactions and the Claisen rearrange- ment. There are three main categories of concerted reaction (i) intramolecular electrocyclic reactions (ii) intermolecular cycloaddition reactions and (iii) sigmatropic rearrangements.lP2 The reactions are subject to thermal and/or photochemical control and usually proceed in a highly stereospecific manner.Despite the synthetic utility of these reactions until recently their mechanisms were not at all well understood. In many cases the rates of the reactions are little affected by the nature or on the presence or absence of solvent or catalyst and intermediate products are rare. Thus extensive mechanistic studies have fur- nished comparatively little detailed information. One general feature of the reactions is the characteristic negative entropy of activation. This is indicative of a high degree of atomic order in the transition state of such reactions. Recently Woodward and Hoffmann,lp2 Fukui? and Longuet-Higgins and Abrahamson4 have interpreted these concerted reactions in terms of the sym- metry properties of the reactant and product energy levels. The definitive rules developed by Woodward and Hoffmann are of particular interest to the organic chemist for they not only explain the large majority of these reactions known at present but permit predictions to be made on reactions hitherto not s t ~ d i e d .~ ~ ~ This Review is largely devoted to the qualitative aspects of orbital symmetry relationships and in particular to examining the application of the Woodward- Hoffmann rules to concerted organic reactions. The rules are based on the results of extended Huckel calculations,6 and mark an important achievement of R. B. Woodward and R. Hoffmann J. Amet. Chem. SOC. 1965 87 (a) 395; (6) 2046; (c) R. Hoffmann and R. B. Woodward Accounts of Chemical Research h e r . Chem. SOC. (a) K. Fukui Tetrahedron Letters 1965 2009; Bull. Chem. Soc. Japan 1966 39 498; (b) H.C. Longuet-Higgins and E. W. Abrahamson J. Amer. Chem. Soc. 1965,87,2045. P. Millie Bull. SOC. chim. France 1966,4031 ; E. M. Kosower “An Introduction to Physical Organic Chemistry” Wiley New York 1968; M. Orchin and H. H. Jaffe “The Importance of Antibonding Orbitals” Houghton Mifflin Co. Boston U.S.A. 1967; J. J. Volmer and K. L. Servis J. Chem. Educ. 1968 45 214. 4385; (4 2511; (e) 4389; see also L. Salem ibid. 1968 90 543 553. Publ. 1968 vol. 1 p. 17. K. Fukui and H. Fujimoto Tetrahedron Letters 1966 251 ; (c) K. Fukui ibid. 1965 2427. R. Hoffmann J. Chem. Phys. 1963,39 1397; 1964,40,2480. 338 Gill molecular orbital theory. The underlying theoretical principles of course require a considerable familiarity with the usual computational methods of molecular orbital theory.The elegant approach developed by Longuet-Higgins and Abrahamson however only requires a knowledge of the symmetry properties of the reactant and product energy levels and their relative energies. By the application of Group Theory the reactant and product energy levels can be correlated graphically and the conditions (i.e. d or hu) necessary to perform a particular reaction can be deduced from an inspection of the correlation diagram. Thermally allowed reactions for example are characterised by a complete correlation of ground-state energy levels in the reactants with the corresponding ground-state energy levels in the products. Strictly speaking the correlation diagrams are only applicable to systems possessing some symmetry since the construction of a diagram for an unsym- metrical system (i.e.one without a plane or two-fold axis of symmetry) can lead to a totally erroneous conclusion. The detailed examination of levels throughout a reaction avoids pitfalls of this type,2 and in the following discussion the Woodward-Hoffmann rules are applied to several reactions where important symmetry properties are absent. The reactions of course must be concerted and in photochemical processes particularly it can be difficult to be completely sure that this prerequisite is met. A theoretical treatment applicable to unsymmetrical systems has been described.' The Article is divided into the following sections (1) Intramolecular electro- cyclic reactions namely 4n n-electron systems and 4n + 2 n-electron systems; (2) Intramolecular fragmentation reactions; (3) Iptermolecular cycloaddition reactions namely thermally controlled cycloadditions photochemically con- tro!led cycloadditions and metal-catalysed cyclo-oligomerisations; (4) Sigma- tropic rearrangements namely those of order [l j ] and [i j ] ; ( 5 ) miscellaneous reactions; and ( 6 ) conclusions.1 Intramolecular Electrocyclic Reactions It is necessary first briefly to consider the possible stereochemical features of intramolecular electrocyclic reactions. The terminology used throughout is that suggested by Woodward and Hoffmann.lS2 Consider the cyclisation of a linear fully conjugated olefin containing k n-electrons and substituted at the terminal carbon atoms by the groups A-D as shown in Figure 1. The acyclic olefin is drawn such that the lobes of the n-orbitals are parallel to the plane of the paper whereas the carbon chain (assumed planar) is perpendicular to this plane; In order that a new a-bond may form between the termini the terminal n-lobes must interact.This can only arise when concerted rotation about the C(l)-C(2) and C(k - 1)-C(k) bonds brings these terminal lobes sufficiently close for orbital overlap to occur. The rotation may occur in two physically distinct senses (a) conrotatory or (b) dis- rotatory giving rise to the products (I) or (11). In certain cases one can dis- H. E. Zimmerman J. Amer. Chem. SOC. 1966 88,1564 1566. 339 Woodward-Hofman Orbital Symmetry Rules to Concerted Organic Reactions p& P I c A It I (b) Figure 1 P 7’ tinguish two possible conrotatory or disrotatory modes. Thus in case (b) if the cyclic olefin (11) has A=D = CH and B=C = H ring opening could in principle yield a trans,truns-1 ,k-dimethylpolyene or its cis,cis-isomer.laY2 In general these electrocyclic reactions proceed in a highly stereospecific manner under thermal or photochemical control and the product of the thermal reaction is different from that of the photochemical process.Thus under one form of control path (a) is followed whereas in the alternative energy pathway the mode (b) applies. In circumstances where there are two possible products as in the above case (11; A=D = CH,; B=C = H) one product usually pre- dominates for reasons of simple steric demand (i.e. the trans,tvans-1 &-dimethyl- polyene should be the major or only product). The stereochemical pathway is predominantly determined by the symmetry properties of the highest occupied molecular orbital (containing two electrons in thermal ground-state reactions or one electron in the photochemical first excited state p r o c e ~ s e s ) .~ ~ ~ ~ ~ * This bald statement is based upon the results of extensive calculations,1,2,6 and upon the examination of symmetry state correla- tion diagram^.^,^^^^^ An account of these methods cannot be given here but the basic results can be represented as in Figure 2 which indicates the essential features of orbital symmetry at the termini of 4n and 4n + 2 .zr-electron systems. The development of a bonding interaction between the termini can only arise if a concerted rotation about the C(l)-C(2) and the C(k - 1)-C(k) bonds causes an overlap of the terminal lobes such that there is no sign inversion in the region of overlap.The rotations must therefore occur in the physical senses indicated 340 Gill h Thermal Y J - x Photochemical 4n systems 4n + 2 systems Figure 2 Orbital symmetries and symmetry elements for the ring closure of 4n and 4n -l- 2 .rr-electron systems. A symmetry axis (CzY) is designated for conrotatory ring closure a symmetry plane (uyz) for disrotatory ring closure. in Figure 2. The conrotatory mode is characterised by a two-fold axis of sym- metry (C2J and the disrotatory mode is characterised by a symmetry plane (cry ,). The Woodward-Hoffmann rules for intramolecular electrocyclic reactions may be summarised as in Table 1 according as to whether i k is an odd (4n + 2 system) or an even (4n system) integer.1a*2~4p5 Odd-electron systems (i.e. radicals) behave similarly to even-electron systems possessing one further electron.341 Woodward-Hofnian Orbital Symmetry Rules to Concerted Organic Reactions Table 1 k Thermal Photochemical 4n Conro ta tory Disro tatory 4n + 2 Disro tatory Conrotatory n = 0 1 2 3 . . . . . Charged systems behave similarly to electrically neutral systems containing the same number of n-electrons. The rules are analogous to the spectroscopic selection rules in that they are based on symmetry and are neither a necessary nor sufficient condition for the relevant phenomena to be observed. However when a particular mode is predicted but cannot operate because of the molecular geometry the alternative mode may operate under much more highly energetic conditions and a non-concerted mechanism may then apply.A. 4n 7r-Electron Systems.-(i) Thermal control (conrotatory mode allowed). In this category the most common examples are for the butadiene-cyclobutene interconversion. Many of these examples will not be considered here because the substitution pattern (or lack of it) of the substrate does not allow a distinc- tion to be drawn as to whether a conrotatory process or a disrotatory process has occurred since only one product is possible. Many more examples of the thermal ring fission of cyclobutenes are known than the reverse ring closures which are comparatively rare processes. When the butadiene moiety forms part of a medium-sized alicyclic ring (2 C8) ready thermal cyclisation occurs if the 1,3-diene system has a cis,trans geometry. The products (111) and (IV) suffer 80' > 175O (2) disrotatory (or non-concerted) ring cleavage to yield the cis,cis-monocyclic dienes but only at considerably higher temperatures.8 When a cis,cis-l,3-diene unit is part of an alicyclic ring (3 C,) conrotatory cyclisation to the trans-fused bicyclic systems should be possible [e.g.equation (3)] but examples of this type have not been reported. * K. M. Shumate P. N. Newman and G. J. Fonken J. Amer. Chem. SOC. 1965,87,3996. 342 GiII As mentioned above the thermal ring cleavage of cycIobutenes is a well- documented process. In the case of simple cyclob~tenes~-~~ the essentials can be summarised as in equations (4) and (5). The predicted products are formed in each case. In equation ( 5 ) two products are possible in each case the trans,trans- compounds and the cis,cis-compounds.The former are produced almost ex- clusively owing to the lower steric compression in the transition state when the su bstituents rotate outwards. (4) R = H A = H R = H A = c1 R = H A=Me R = M e A = M e R = H R =Ph A = Ph A = CO,Me B = Mes B = Me" B = C0,Me18 B = Cl'O B = MeloBlg B = Ph" R =H A = CI R = H A=Me R = M e A = M e R = H A = COaMe R =Ph A =Ph B a l 5 - 1 7 B = Me" B = Me"J91S B = C0,Me18 B = Ph" * E. Gil-Av and J. Shabtai J. Org. Chem. 1964 29 257; H. M . Frey Trans. Farahy SOC. 1964 60 83. lo R. Criegee D. Seebach R. E. Winter B. Borretzen and H. A. Brune Chem. Bet. 1965 98 2339. l1 R. E. K. Winter Tetrahedron Letters 1965 1207. le R. Criegee and K. Noll Annalen 1959 627 1. l3 E. Vogel Angew. Chem. 1954 66 640; Annalen 1958 615 14. l4 H. H. Freedman G. A. Doorakian and V.R. Sandel J. Amer. Chem. SOC. 1965,87,3019. l6 M. Avram I. Dinulescu M. Elian M. Fircagiu E. Marica G. Mateescu and C. D. Nenitzescu Chem. Ber. 1964 97 372. l7 G. F. Emerson L. Watts and R. Pettit J. Amer. Chem. SOC. 1965 87 131. R. Criegee W. Horauf and W. D. Schellenberg Chem. Be?. 1953,86 126. 343 Woodward-Hofman Orbital Symmetry Rules to Concerted Organic Reactions Bicyclo[l,l,O]butane (V; a=b=c=d = H) is thermolysed to butadiene.Is In considering the energetics of the process Wiberg and Lavanishl*” conclude that ring opening occurs by way of a concerted conrotatory path. However the stereospecificity of the process remains to be demonstrated for example by utilising (V; a=c = ,H; b=d = lH)lsa since with (V; a=c = C0,Me; b=d = H; H = Ph or a=d = C0,Me; b=c = H; H = Ph) the results are somewhat confused (the predicted product is not obtained in the former case but is in the latter process).18 In certain cases the terms conrotatory and disrotatory are meaningless for example in the ring cleavages of cyclobuten-3-ones and 3-methylenecyclo- butenes.Nevertheless the processes can be highly stereospecific. This has been demonstrated for the ketone system (VI). Under thermal control the trans- products (VII) are obtained whereas the cis-isomers of (VII) are formed in the corresponding photochemical rea~ti0ns.l~ (7) (a) R = R’ = C1 (b) R = Me R’ = C1 (c) R = C1 R’ = Me A large amount of work has been carried out on the thermal cleavage of cyclo- butenes in which the cyclobutcne moiety forms part of a bicyclic or tricyclic system. Such reactions are expected to be difficult since a cis,trans-l,3-diene l8 (a) K.B. Wiberg and J. M. Lavanish J. Amer. Chem. Soc. 1966 88 5272; see also K. B. Wiberg and G. Szeimies Tetrahedron Letters 1968 1235; (6) I. A. D’yakonov V. V. Razin and M. 1. Komendantov Tetrahedron Letters 1966 1127 1135. l9 J. E. Baldwin and M. C . McDaniel J. Amer. Chem. Sac. 1967 89 1537; see also E. Gil-Av and J. Herling Tetrahedron Letters 1967 1. 344 Gill contained within a ring would be formed in the conrotatory process. However (VIII) and (IX) ring-openZ0 without special difficulty since a trans-double bond can be accommodated by the ten-membered rings. 200° ____) gas phase 140° A gas pheee However when the second ring of the bicyclic system possesses rather fewer atoms conrotatory ring fission is clearly impossible.Ring opening must therefore occur by way of the symmetry-forbidden disrotatory mode or by a heterolytic or homolytic pathway which are predicted to require highly energetic condi- tions.laY2 This conclusion has been substantiated in a number of cases. Thus for compounds of type (X),10,11921-24 in which the two rings are necessarily fused cis ring opening only occurs at high temperatures. A similar situation pertains to the related heterocyclic and carbocyclic systems (XI-XV) which require high temperatures respectively 350",25 500" (unchanged),2s 500°,27 500" (un- changed),28 and 360°,29 for ring opening to occur. It appears that these difficult ring fissions can be facilitated under mild condi- tions by the presence of transition-metal The mixing of the substrate 2o P.Radlick and W. Fenical Tetrahedron Letters 1967 4901 ; see also J. J. Bloomfield ibid. 1968 587. 21 (a) M. R. Willcott and E. Goerland Tetrahedron Letters 1966 6341 ; (b) W. G. Dauben and R. L. Cargill Tetrahedron 1961 12 186. 22 J. Rigaudy and P. Courtot Tetrahedron Letters 1961 95; P. Courtot Ann. Chim. (France) 1963 8 197. 23 R. Criegee and H. Furrer Chem. Ber. 1964 97 2949. 24 R. Askani Chem. Ber. 1965 98 2322. p6 P. R. Story and S. R. Fahrenholtz J. Amer. Chem. SOC. 1965 87 1623. R. F. Childs and A. W. Johnson J. Chem. SOC. (C) 1967 874. E. Vogel R. Erb G. Lenz and A. A. Bothner-By Annaien 1965 682 1. G. J. Fonken Chem. and Ind. 1961 1575. M. Jones and S. D. Reich J. Amer. Chem. SOC. 1967 89 3935. F. R. Mango and J. H. Schachtschneider J . Amer. Chem. SOC. 1967 89,2484.31 (a) H. Hogeveen and H. C. Volger J . Amer. Chem. SOC. 1967 89 2485; (b) H. Hogeveen and H. C. Volger Chem. Comm. 1967 1133. 32 W. Merk and R. Pettit J. Amer. Chem. SOC. 1967 89 4787 4788. Woodward-Hofrnan Orbital Symmetry Rules to Concerted Organic Reactions X ( X ) (10) n = 3 R = H X = H11,21 n = 4 R = H X- H10#'1 R = H X =Ph2* R = Me X = MeaP R =Me X =Has R = Me X = C0,Mea4 R = Me X = C0,R2* and the metal electronic energy states converts the reaction into a symmetry- allowed process. Thus a number of tricyclo[4,2,0,02~5]octa-3,7-dienes are readily isomerised at low temperatures to cyclo-octatetraenes in the presence of silver@ and copper(1) ions.32 Reaction (1 1) is typical. Ag+. roam temp. Nevertheless there are a number of such isomerisations which occur in the absence of metal catalysts and at anomalously low temperatures (i.e.low activation energies). Thus bicyclo[2,1 ,O]pent-Zene is converted into cyclo- pentadiene at room temperature in carbon tetrachloride solution (ti = ca. 2 J X . ) ~ ~ and Dewar benzene is isomerised to benzene under similar conditions (room temperature t ) = ca. 2 days in ~yridine).~~ Both of these bicyclic com- 83 J. I. Brauman L. E. Ellis and E. E. van Tamelen J. Amer. Chem. SOC. 1966,88 846. s4 E. E. van Tamelen and S. P. Pappas J . Amer. Chem. Suc. 1963,85,3297; E. E. vanTamelen Angew. Chem. Znternat. Edn. 1965 4 738. 346 Gill pounds are highly strained and in the case of Dewar benzene the formation of the low-energy benzene molecule may cause a further lowering of the activation energy for the valence isomerisation.Indeed it is perhaps surprising that Dewar benzene is so rekztively stable in view of the large favourable enthalyy change (AH = -60 kcal. mole-l) for the isomerisation. In many of the cases in which ring fission of the cyclobutene ring of a bi- or tri-cyclic system occurs at a rela- tively low temperature in contravention of the rules a lowering of the activation energy for the isomerisation may be traced to the release of ring strain in the reactant or to a substantial increase in conjugation in the product. Typical examples of this behaviour are illustrated by the compounds 0(VI-XXI).10~35-38 Activation energy data (where available) and the characteristic reaction tem- perature (T10-4) at which the rate of cleavage is lo4 sec.-l are included.1° The data are clearly far from complete but the introduction of a further ring (XX) txvr) tXVl1) 195" 205" dHiS(kcal./mole) 42 - Ref.10 10 (XVIII 1 31 85" 35 ]i-prR ]i-T--fJ-c6H~l~ C%R 0 ( X l X I (XX 1 ( X X I ) ( X X i I ) dHl(kcal./mole) 27.7 - - T10-4 98" ca. 100" complete reaction at 210" Ref. 36 37 38 or of a double bond capable of conjugating with the developing 1,3-diene system (XVIII XIX) has the predictable effect of 1oweringdHt and as compared with (XVI). The activation parameters for (XXII) a more satisfactory reference compound have apparently not been determined. The easy thermal isomerisa- tion of the azabicycloheptadiene (XXIII) to N-ethoxycarbonyla~epine~~ (a non- planar molecule) is more difficult to explain. Further experimentation in this area seems highly desirable. 35 D. Seebach Chem.Ber. 1964 97 2953. 36 R. Crjegee and F. Zanker Angew. Chem. Internat. Edn. 1964 3 695. 37 G. Fonken and U. Mehrotra Chem. andInd. 1964,1025; W. G. Dauben and R. M. Coates J. Amer. Chem. SOC. 1964 86 2490. 38 J. B. Bremner and R. N. Warrener Chem. Comm. 1967,926. 39 L. A. Paquette and J. H. Barrett J. Amer. Chem. Soc. 1966 88 1718; see also L. A. Paquette and R. W. Begland ibid. p. 4685. 347 4 Woodward- Hofman Orbital Symmetry Rules to Concerted Organic Reactions 100 min. I (12) In the category of 4n welectron systems reactions other than cyclobutene- butadiene interconversions are relatively uncommon. The conservation of orbital symmetry demands that the concerted cyclisation of the pentadienyl cation to the cyclopentenyl cation should occur by the conrotatory pathway. Such cyclisa- tions have been observed:* and the stereoselectivity has now been established in one case.The treatment of 1,1 ’-dicyclohexenyl ketone with phosphoric acid. yields a mixture of two ketones with the gross structures (XXIV) and (XXV). The stereochemistry of (XXIV) which is diagnostic of the stereospecificity of the cyclisation of the intermediate carbonium ion (XXVI) has been shown to be as indicated by the structure (XXIVa). Hence the conrotatory cyclisation of (XXVI) is clearly demonstrated.4l 0 0 m i” ( x x v r 1 + t x x v ) J (13) The rules further predicted l a p 2 that the thermal isomerisation of the cyclo- propyl anion to the ally1 anion should follow a concerted conrotatory course. This has now been verified by Huisgen and his co-workers who studied the thermolysis of both cis- and trans-dimethyl 1 -(4-methoxyphenyl)aziridine-2,3- dicarboxylate in the presence of efficient dipolarophiles (e.g.dimethyl acetylene- dicarboxylate).42 The intermediate azomethine ylids (XXVII) and (XXVIII) are thus efficiently intercepted and the stereospecificity of the initial ring-opening can be deduced from the stereochemistry of the essentially pure adducts (XXIX) 40 G. A. Olah C. U. Pittman and T. Y. Sorensen J. Am&. Chem. Soc. 1966,88,2331; T. S. Sorensen ibid. 1965,87 5075; Canad. J. Chem. 1965,43,2744; N. C . Deno C. U. Pittman and J. 0. Turner J. Amer. Chem. SOC. 1965 87 2153. 41 R. B. Woodward in “Aromaticity” Chem. SOC. Special Publ. No. 21 1967 p. 217. 43 R. Huisgen W. Scheer and H. Huber J. Amer. Chem. SOC. 1967 89 1753. 348 Gill and (XXX) since dipolar cycloaddition is known to proceed stereospecifically cis.Af p Ar ./L$ = tooo * H ) ~ c o s H e *"Q 'C0,tle - . . M OaC C W C M e0,C H H ' PI eO,C COAMe (xx VI I. 1 \ f (XXIX) MeO,C.C~C.CO,Me Q n e 0 ) H Thermal cyclisations in systems in which n > 1 are almost unknown. This is not surprising since in the case n = 2 cyclisation would produce an eight- membered ring and for n = 3 a twelve-membered ring. Such processes would be attended by a highly adverse entropy factor as well as a substantially increased activation energy. One possible example of an n = 2 system is provided by the hydrogenation of traqtrans- and of cis,trans-deca-2,8-diene-4,6-diyne over a Lindlar catalyst. In addition to a mixture of over-hydrogenation products the trans,trans compound yields the bicyclic compound (XXXI) whereas (XXXII) is produced from cis-tran~-deca-2,8-diene-4,6-diyne.~ The formation of (XXXI) for example can be rationalised as in the sequence (15).The conversion (XXXIII) Gd"' 'n. ( X X X I ) r i c ( X X X l l ) Me Me (xxxr) Is (a) E. N. Marvell and J. Seubert J. Amer. Chern. Soc. 1967 89 3377; (b) Dahmen and H. Huber ibid. p. 7130 and refs. therein. d Me R. (1 5 ) Huisgen A. 349 Woodward-Hofman Orbital Symmetry Rules to Concerted Organic Reactions -+ (XXXI) is also a thermally allowed process (4n + 2 system disrotation). These conclusions are in agreement with the results of the thermal isomerisation of pure 2,4,6,8-de~atetraenes.~~~ (ii) Photochemical control (disrotatory mode allowed). As in the previous section the majority of examples are of the butadiene-cyclobutene interconver~ion.~~ Indeed the photochemical cyclisations of butadienes is a valuable synthetic route to cyclobutenes.The simple acyclic systems have been extensively but the stereochemistry of the products have not always been defined. The majority of the definitive examples are to be found in systems in which the ljutadiene chromophore forms part of a ring. Several of these are summarised in reactions (16) and (17).21,33,39,45-53 The formation of (XXXVc) is surprisingly = 1,33 2 46 3 Zlb,46 4 45a,47,48 and 549,513 9 7 1 (==J / CARL H t x x x v ) R" (1 7) (a) R1 Ra = H; X = CH,,21b945a CHz.CH2:' 0:' NCOaEt3* (b) R1 Ra = Me; X = 039 (c) R1 = OMe; Ra = H; X = CH,53 specific and this seems to be a general rule for 1-substituted cycloheptatrienes when the substituent is an electron-releasing group (e.g.R2 = H; R1 = Me 44 R. Steinmetz Fortschr. Chem. Forsch. 1967 7 445. 45 (a) R. Srinivasan J. Amer. Chem. SOC. 1962 84 4141 ; 1963 85 4045; (6) K. J. Crowley Tetrahedron 1965 21 1001. 46 H. Prinzbach and E. Druckrey Tetrahedron Letters 1965 2959. 47 R. S. H. Liu J. Amer. Chem. SOC. 1967 89 112; see also D. I. Schuster B. R. Sckolnick and F.-T. H. Lee ibid. 1968 90 1300. 48 W. G. Dauben and R. L. Cargill J. Org. Chem. 1962,27 1910. 49 K. M. Schumate and G. J. Fonken J. Amer. Chem. SOC. 1966 88,1073. 50 K. M. Schumate P. N. Neuman and G. J. Fonken J. Amer. Chem. SOC. 1965,87,3996. 51 W. R. Roth and B. Peltzer Angew. Chem. Internat. Edn. 1964 3 440. 62 J. M. Holovka and P. D. Gardner J. Amer. Chem. SOC. 1967,89,6390. 53 G. W. Borden 0. L.Chapman R. Swindell and T. Tezuka J. Amer. Chem. SOC. 1967 89 2979. 350 Gill OMe SMe and NMe,).54 Similarly the irradiation of h~motropylidene~~ and 2,3-h0motropone~~ yields only one of the two possible isomers arising from the photochemical disrotatory process in each case. This result has been ascribed to secondary steric forces which raises the activation energy of the alternative disrotatory pathways.56 In certain cases it appears that the observed photochemical reaction may in fact comprise two discrete steps. Thus the acetophenone-sensitized irradiation of (XXXIV; n = 4)47 yields the cis,trans-isomer of (XXXIV; n = 4) which cyclises by the conrotatory mode at 80" to give a bicyclic product identical with that produced in the direct unsensitized reaction of (XXXIV). In connexion with this observation Liu4' has suggested that the sensitized photochemical cyclisation of 1,l '-bicy~lohexenyl~~ also occurs by way of a two-step reaction (18).The product (XXXVI) is also formed in the direct (unsensitized) reaction (XXXVI) (1 8) by the disrotatory pathway.57 A trans-fused cyclobutene ring may be formed from monocyclic systems possessing a cis,trans-l,3-diene unit if the ring is sufficiently large. Thus irradiation of cis,cis-l,3-cyclononadiene W V I I ) yields both a cis- and a trans-fused cyclobutene (19). The products are formed by disrotatory cyclisations of (XXXVII) and its &,cis-isomer (XXXVIII) which itself is produced in an initial and rapid photo-eq~ilibrium.~~ Likewise Faren- h o r ~ t ~ ~ has suggested that Mobius benzene (Le. cis,cis,trans-cyclohexa-1,3,5- triene) is possibly the intermediate in the photochemical isomerisation of benzene to Dewar benzene.Alternative explanation^^,^^ of this process however appear more attractive. An outstanding example of the use of these reactions in preparative organic 54 A. P. Ter Borg E. Razenberg and H. Kloosterziel Chem. Comm. 1967 1210. 55 W. R. Roth and B. Peltzer Annalen 1965 685 56. 56 L. A. Paquette and 0. Cox J . Amer. Chem. Soc. 1967 89 5633. 57 W. G. Dauben R. L. Cargill R. M. Coates and J. Saltiel J . Amer. Chem. SOC. 1966 88 2742. 68 E. Farenhorst Tetrahedron Letters 1966 6465. 59 D. Bryce-Smith and H. C. Longuet-Higgins Chem. Comm. 1966 593. 35 1 Woodward-Hoflman Orbital Symmetry Rules to Concerted Organic Reactions chemistry is furnished by the elegant synthesis (20) of Dewar benzene.34 The irradiation of substituted cyclohexadiene anhydrides yields mainly aromatic hydrocarbons.The latter reaction can be suppressed by using the Corresponding The similar reaction of a-pyrone (21)60 promises to be a useful and u o 0 (20) simple route to cyclobutadiene (isolated as the iron tricarbonyl complex).61 8 + co Cyclopropyl anions have also been shown to follow the predicted disrotatory course. Thus the irradiation of the cis- and trans-aziridines discussed earlier (eqn. 14) in the presence of dimethyl acetylenedicarboxylate yields the expected products ; cis-aziridine - (XXX) trans-aziridine - 0(XIX).42 The rules are also obeyed for systems in which n = 2 although few such reactions are known. Upon irradiation W I X ) ring-expands to give [16]-ann~lene,~~ an example of double disrotation and (XL) under similar conditions is converted into 9-ethoxycarbonylcyclonona-l,3,5,7-tetraene which is further transformed under the influence of heat and light.63 (XXXIX) ( X L I B.(4n + 2)~-Electron Systems.-(i) Thermal control (disrotatory mode allowed). The simplest system here is provided by the case n = 0 corresponding to the cyclopropyl-ally1 cation interconversion. Molecular orbital calculations1a~2 indicate a very high stereospecificity for the ring-opening if the ionisation of a suitable leaving group (e.g. tosylate or halide) and disrotation are completely O0 E. J. Corey and J. Streith J. Amer. Chem. SOC. 1964 86 950. 62 G. Schroder and J. F. M. Oth Tetrahedron Letters 1966 4083. 03 G. J. Fonken and W. Moran Chem. and Znd.1963 1841. M. Rosenblum and C. Gatsonis J. Amer. Chem. SOC. 1967 89 5074. 352 concerted. The experimental investigations of De Puy and his co-workers have led them independently to the same conclusion.gq The cyclopropane substituent groups trans to the leaving group (X) rotate outwards (or cis-substituents rotate inwards) with concerted loss of X-. In this way the departure of X- is anchimeric- ally assisted by rear-side orbital overlap.64 These preferred modes of disrotation are represented in equation (22) and it should be noted that only acyclic ally1 Allylic products cations are under consideration. The situation is therefore precisely defined (XLI) should react more quickly than (XLII) in identical experimental condi- tions since the steric compression of the incoming groups of the cis-compound raises the energy of cisoid transition state.This conclusion has been amply verified by various solvolysis studies.@,65 When ionisation of X and ring-opening are simultaneous the cyclopropyl cation is not a true intermediate since it does not have a finite lifetime. In a discrete cyclopropyl cation the positively charged carbon atom and the atoms bonded to it presumably adopt a planar configuration. On symmetry grounds alone both of the possible disrotatory modes are identical in these circumstances and relatively non-specific ring-opening may be found. In this connexion Kirmse and Schiitte have shown that both cis- and trans-2-phenylcyclopropanediazonium ions decompose in methanol to give trans-cinnamyl methyl ether.66 The result is consistent with the intermediacy of the 2-phenylcyclopropyl cation in the reaction.In the case of the concerted solvolytic cleavage of cyclopropyl tosylates and halides the above ideas can be extended to include bicyclo[n,l ,O]alkyl tosylates and halides. The predictions for the bicyclic systems are essentially the reverse of those for the monocyclic corn pound^.^^ Thus halides and tosylates with the endo configuration (XLIII) should react rapidly by the concerted mechanism since the cis-ally1 cation can be favourably accommodated in a ring structure. C. H. DePuy L. G. Schnack J. W. Hausser and W. Wiedemann J. Amer. Chem. SOC. 1965 87,4006; S. J. Cristol R. M. Sequeira and C. H. De Puy ibid. p. 4007; C. H. DePuy L. G. Schnack and J. W. Hausser ibid. 1966,88 3343; J. W. Hausser and N. J. Pinkowski ibid.1967 89 6981. P. von R. Schleyer G. W. Van Dine U. Schollkopf and J. Paust J. Amer. Chem. SOC. 1966 88 2868 and refs. therein. W. Kirmse and H. Schutte J. Amer. Chem. SOC. 1967 89 1284. 353 Woodward-Hoflman Orbital Symmetry Rules to Concerted Organic Reactions The trans ring opening required for the exo-compound (XLIV) is clearly im- possible for small rings and a more conventional type of transition state might be expected for the exo-isomers. In agreement with this generalisation it has x \ (XLlIl) been found that many endo-bicyclo[n,l ,O]alkyl tosylates and halides react more quickly than the corresponding exo-compounds at least for n = 2 or 3.64,65,s7,ss When more bridging atoms are present (YE 2 4) in the exo-compounds (XLIV) the situation becomes more complex. As n increases through 4 to 5 and 6 in (XLIV; X = OTs) the acetolysis rates show a sharp increase.In the endo- compounds there is a corresponding sharp decrease in rate. An intermediate state has been suggested for these cases which electronically is somewhere inter- mediate between an allyl and a cyclopropyl cation.68 With increasing n the allyl character becomes more important and it is probable that the trans-acetates are the first formed products. The addition of acetic acid to the double bond could then result in trans -+ cis isomerisation to yield the observed cis- cyclenyl-3-acetates. The nature of this intermediate state has been the subject of a further communi~ation.~~ The addition of halogenocarbenes to double bonds is frequently employed as a first step in syntheses involving ring expansion.The implication of the above results is obvious. If a monohalogenocarbene is employed both endo- and exo- isomers are obtained and if the bicyclic compound possesses few bridging atoms (n small) then only the endo-compound reacts to furnish the required p r o d u ~ t . ~ ~ ~ When the number of bridging atoms is large then the exo-compound is the more reactive. In such circumstances it is preferable to employ a dihalogenocarbene so that an endo- and an exo-halogen is guaranteed. This method has formed the basis of a novel meta-cyclophane synthesis (24).70 The trans-allylic system is readily accommodated by the medium-sized ring.71 The majority of the remaining thermally controlled electro-cyclic reactions of (4n + 2) m-electron systems correspond to the situation where n is unity that is the hexatriene-cyclohexadiene interconversion.The pentadienyl radical is also a 67 M. S. Baird and C. B. Reese Tetrahedron Letters 1967 1379; L. Ghosez P. Laroche and G. Slinckx ibid. p. 2767; L. Ghosez G. Slinckx M. Glineur P. Hoct and P. Laroche ibid. p. 2773; C . W. Jefford and R. Medary ibid. 1966,2069 2792; C . W. Jefford E. H. Yen and R. Medary ibid. p. 6317. 6* U. Schollkopf K. Fellenberger M. Patsch P. von R. Schleyer T. Su and G. W. Van Dine Tetrahedron Letters 1967 3639. 69 von W. Kutzelnigg Tetrahedron Letters 1967 4965. 70 W. E. Parham and J. K. Rinehart J. Amer. Chem. SOC. 1967 89 5668. 71 W. E. Parham and R. J. Sperley J Org. Chem. 1967,32,924 926. 354 Gill n = 8 or 10. (412 + 2) system and is known to cyclise to give the cyclopentenyl radical.72 The stereochemistry of the reaction has not been investigated.The pentadienyl anion however cyclises by the expected disrotatory mode (25).73 I The thermal isomerisation of simple acyclic hexatrienes follows the expected disrotatory pathway even though the steric interactions [(26) (27)] between the incoming groups must be quite large.74,75 This must be particularly true in the rlc a (26) and other products (27) 72 K. W. Egger and S. W. Benson J. Amer. Chem. SOC. 1966,88,241. 73 P. R. Stapp and R. F. Kleinschmidt J. Org. Chem. 1965,30 3006. 74 E. N. Marvell G. Caple and B. Schlatz Tetrahedron Letters 1965 385. 75 E. Vogel W. Grimme and E. Dinne Tetrahedron Letters 1965 391. 355 Woodward-Hofmn Orbital Symmetry Rules to Concerted Organic Reactions well-known thermal conversion of precalciferol into isopyrocalciferol (XLV) and pyrocalciferol (XLVI).76 It was here that Oosterhoff made the first sugges- tion of the possible importance of orbital symmetry relationships in such p r o c e s s e ~ .~ ~ ~ ~ ~ The hexatriene-cyclohexadiene interconversion is particularly easy when the 1,3,5-triene unit forms part of a ring system. Several of these processes are summarised in reactions (29) and (30). The cyclisations of the hexatrienes (XLVII; n = 1; X = -0- -N< and -CR,-) have been extensively s t ~ d i e d . ~ ~ ~ ~ - ~ ~ The benzene oxide-oxepine equilibrium has been clearly estab- lished by nuclear magnetic resonance studies and the equilibrium constant varies enormously with solvent polarity and to some extent with temperat~re.~~ A - (29) H H H n = 3,7s or 4.,O However azepines and simple cycloheptatrienes exist almost entirely in the monocyclic form (XLVII; X = :N-R or CH,; n = 1).In the latter case the norcaradiene stability is favoured by the incorporation of one or both of the double bonds into a condensed aromatic system,s3 by bridging the C(l) and 76 E. Havinga and J. L. M. A. Schlatmann Tetrahedron 1961 16 146. T7 E. Vogel W. A. Boll and H. Gunther Tetrahedron Letters 1965 609; H. Gunther ibid. p. 4085; E. Vogel and H. Gunther Angew. Chem. Internat. Edn. 1967 6 385. 78 A. P. Ter Borg E. Razenberg and H. Kloosterziel Rec. Trav. chim. 1966 85,774. 79 W. L. Mock J. Amer. Chem. SOC. 1967 89 1281. Benzene is the product of the reaction but it is probable that the bicyclic sulphone is intermediate in the change. 80 R. Huisgen and F.Mietzsch Angew. Chem. Internat. Edn. 1964,3,83; E. Vogel H. Kiefer and W. R. Roth ibid. p. 442. 81 D. S. Glass J. W. H. Watthey and S . Winstein Tetrahedron Letters 1965 377. 8s A. C. Cope A. C. Haven F. L. Rampand and E. R. Trumball J. Amer. Chem. Soc. 1952 74 4867; see also ref. 43. 8s R. Huisgen and G. Juppe Chem. Ber. 1961,94,2332; E. Muller H. Kessler and H. Suhr Tetrahedron Letters 1965 423; H. Nozaki M. Yamabe and R. Noyori Tetrahedron 1965 21 1657; E. J. Corey H. J. Burke and W. A. Remers J. Amer. Chem. Soc. 1955,77,4941; E. Vogel D. Wendisch and W. R. Roth Angew. Chern. Internat. Edn. 1964,3,443. 356 Gill C(6) positions with a three-atom bridge8* (but not four- or five-atom bridges77) or by the presence of electron-withdrawing substituents at C(7).85,8s Norcaradienes which do not possess one of the above features are rare but 2,5,7-triphenylnorcaradiene is reported to be moderately stable.87 However when the bridging atom spans the 9- and 10-positions of the naphthalene nucleus (XLVIII) the bicyclic lor-electron systems (XLIX) are the predominant isomers (31).77,88,89 The parent l0n-electron system [lo]-annulene is not stable (xLvr11) ( X L I X ) (31) X = O,ss CH2,80 N.CO.CH,"b at normal temperatures and cyclises readily to cis-9,10-dihydronaphthalene the expected isomer of a thermal reaction.The Woodward-Hoffmann rules are therefore highly successful in the inter- pretation of the reactions of 4n + 2 n-electron systems run under thermal control. Occasionally the (4n + 2) system involved in the change may only appear as a transient intermediate in the reaction but its presence seems probable from the nature of the final product [e.g.reactions (32)29 and (33)90]. (33) 84 E. Vogel W. Wiedemann H. Kiefer and W. F. Harrison Tetrahedron Letters 1963 ; 673 ; P. Radlick and W. Rosen J. Amer. Chem. SOC. 1966 88 3461; R. Dams T. Threlfall M. Pesaro and A. Eschenmoser Helv. Chim. Acta 1963 46,2893. E. Ciganek J. Amer. Chem. SOC. 1965 87 652 1149; M. A. Battiste and T. J. Barton Tetrahedron Letters 1967 1227. 86 E. Ciganek J. Amer. Chem. SOC. 1967 89 1454 1458. T. Mukai H. Kubota and T. Toda Tetrahedron Letters 1967 3581. (a) A. Shani and F. Sondheimer J. Amer. Chem. SOC. 1967 89 6310; (b) E. Vogel N. E. Vogel and H. D. Roth Angew. Chem. Internat. Edn. 1964 3,228. S . Mazamune C. G. Chin K. Hojo and R. T. Seidner J .Amer. Chem. Soc. 1967 89 Biskup W. Pretzer and W. A. Boll Angew. Chem. Internat. Edn. 1964 3 642. 4805; see also K. Grohmann and F. Sondheimer ibid. p. 71 19. 357 Woodward-Hofman Orbital Symmetry Rules to Concerted Organic Reactions (ii) Photochemical control (conrotatory mode allowed). Although the photo- chemical transformations of 4n + 2 n-electron systems are well known virtually all of the reactions are of the hexatriene-cyclohexadiene type (i.e. n = 1) so that the diversity of the processes is not very great. The irradiation of trans,cis trans-2,4,6-oct at riene yields trans-5,6-dime t hyl- 1,3 -cyclo hexadiene. 91 On this basis it seems likely that the photochemical cyclisation of cis-stilbene and of related systems gives as the initial product a trans-4a,4b-disubstituted phen- anthrene (L).92-94 The stereochemistry about the newly formed c3-bond has not often been proven and in the simple cases (R = H) the lability of the plioto- product in the presence of traces of oxygen makes its isolation experimentally exceedingly difficult.With appropriate substitution these difficulties are partially avoided.95 P The photochemical ring fission of cyclohexadienes is well known and is frequently reversible. Occasionally the bridging of the butadiene unit to form a cyclobutene may occur in competition with the ring-opening to a hexatriene. The cyclobutene formation is not a photochemically reversible process since the absorbing chromophore is destroyed by the r e a c t i ~ n . ~ ~ ~ ~ Thus pyrocalciferol (XLVI) and isopyrocalciferol (XLV) upon irradiation yield the cyclobutenoid compounds (LI) and (LII) respe~tively.~~ However ergosterol (LIII) which differs from the above only in the stereochemistry of the groups about the C(9>-C(lO) bond is converted into pre-ergocalciferol (LIV) which in turn is further transformed under the influence of light into tachysterol (LV) and lumisterol (LVI) as outlined in reaction (34).98 The cis-ring fusion present in the pyrocalciferol and isopyrocalciferol molecules is much more amenable to cyclo- butene formation than is the case for ergosterol or lumisterol which are trans- fused.9l G. J. Fonken Tetrahedron Letters 1962 549. 82 K. A. Muszkat and E. Fischer J. Chem. SOC. (B) 1967 662 and refs. therein. 98 A. Padwa and R. Hartman J. Amer. Chem. SOC. 1966,88,3759. 94 C. E. Loader and C. J. Timmons J.Chem. SOC. (C) 1967 1343 1457 1677 and refs. therein. 95 V. Boekelheide and J. B. Phillips J. Amer. Chem. SOC. 1967 89 1695; J. B. Phillips R. J. Molynew E. Sturm and V. Boekelheide ibid. p. 1704; V. Boekelheide and T. Miyasaka ibid. p. 1709; H. Blashke and V. Boekelheide ibid. p. 2748. 96 p. de Mayo and S. T. Reid Quart. Rev. 1961 15 393. 97 W. G. Dauben and G. J. Fonken J. Amer. Chem. SOC. 1959 81 4060. 98 E. Havinga R. J. de Kock and M. P. Rappoldt Tetrahedron 1960 11 276. 358 Gill (LVII) q z .... 0. (35) In some cases the conrotatory ring fission to give the hexatriene may be followed by a purely thermal (disrotatory) cyclisation process. A scheme based on these lines accounts for the photo-isomerisation of isodehydrocholesterol to coprosta-6,8-dienols6 and for the conversion (LVII) -+ (LVIII).gg The photo- chemical ring cleavages may in suitable cases be further complicated by therm- ally controlled sigmatropic rearrangement.96,98~100 Such processes are considered below.The conrotatory ring opening of 5,6-disubstituted cyclohexa-l,3-dienes under photochemical control should give a cis,cis,trans-hexatriene if the 5,6-substituents are cis to one another whereas an all-cis-hexatriene or a trans,cis,trans-hexatriene should result with trans-5,6-substituents. In this connexion Vogel and his collaborator^^^ have found that (LIX) is isomerised to cis,cis,trans-cyclonona- 1,3,5-triene which cyclises at room temperature by the disrotatory pathway to the trans-isomer of the starting material. Likewise (LX) yields (LXI),75 and trans-9,lO-dihydronaphthalene upon irradiation at very low temperatures gives detectable amount of [lo]-annulene.Between - 190" and room temperature the [lo]-annulene cyclises (38) to cis-9,10-dihydronaphthalene.lo1 One interesting 99 E. J. Corey and A. G. Hortmann J. Amer. Chem. SOC. 1965,87,5736. loo R. L. Autrey D. H. R. Barton A. K. Ganguly and W. H. Reusch J. Chem. SOC. 1961 3313. Io1 E. E. van Tamelen and T. L. Burkoth J. Amer. Chem. SOC. 1967 89 151. 359 Woodward-Hofman Orbital Symmetry Rules H ( L I X ) to Concerted Organic H Reactions - 8 (LX) (37) H U k n feature of reaction (37) is the simultaneous formation of cis-bicyclo[6,1 ,O]nona- 2,4,6-t13ene’~ by a thermally controlled conrotatory cyclisation (4n v-electron system). 2 Intramolecuhr Fragmentation Reactions The concerted fragmentation reactions of olefinic systems are essentially the reverse of intermolecular addition reactions.Since the fragmentations have several features in common with these intermolecular additions which are dis- cussed below and the intramolecular reactions outlined above they may be conveniently considered here. Basically some of the fragmentation reactions fit into the general pattern of reaction (39) where Y represents a group which furnishes one or more of its constituent atoms for the actual ring skeleton of (LXII). To deduce the orbital symmetry requirements it is simpler to consider the retro-reaction that is the 360 Gill addition of :Y to the polyene (LXIII). The total number of electrons involved in the change (LXII) + :Y -4 (LXIII) is 4 + 2(k - 1) + 2 = 2k + 4 and comprises the four n-electrons of the terminal double bonds of (LXIII) the n-electrons of the internal (conjugated) double bonds of (LXIII) and the two electrons from Y which are here represented as a lone pair.If k is an even number then the reaction is that of a 4n system which under thermal control should follow a conrotatory pathway. When k is odd a disrotatory cyclisation of the (4n + 2) system is to be expected in a thermally controlled reaction. In other words if half the total number of electrons involved in the change is an even number the reaction is conrotatory whereas if half the total number of electrons is odd the disrotatory mode is to be expected in thermal processes. When k is zero the molecule (LXII) is a three-membered ring and the theory predicts a non-concerted fragmentation with possible loss of stereospe~ificity.~~~~~ Many of these thermal fragmentations are however highly stereospecific (LXII; k = 0 ; Y = :N = N >SO2 2 = 0 etc.) but this is not necessarily in- consistent with a non-concerted reaction p a t h ~ a y .~ ~ ~ ~ The definitive cases are for the case k = 1 and a disrotatory cleavage is predicted. Thus the retro-Diels-Alder reaction provides ample evidence for this conclusion. One outstanding example of this process is provided by the ingenious synthesis (40) of benzocyclopropene.103a Here the group (Y) of (LXII) accom- modates the Dair of electrons in a newly-formed n-bond. When Y represents + - (40) but a single ring-bonded atom the two electrons must be accommodated as a non-bonding lone pair. Several examples of this type have been recently dis- closed.Cyclopentadienone dimer upon heating to 120” suffers loss of carbon monoxide from the 10-position. The large positive entropy of activation (AS$ = 10 e.u.) indicates a concerted reaction which because of the geometry of the molecule must follow a disrotatory pathway.lM The thermally induced loss lo2 R. Hoffmann and R. B. Woodward 150th National Meeting American Chemical Society Atlantic City September 1965 Abstr. p. 8 s ; 15 1 st National Meeting American Chemical Society Pittsburgh March 1966 Abstr. ~ 2 8 ~ 1 0 9 . lo3 J. P. Freeman and W. H. Graham J. Amer. Chem. SOC. 1967 89 1761 and refs. therein; see also L. A. Paquette and L. S. Wittenbrook ibid. p. 4483. lo3 (a) E. Vogel W. Grimme and S. Korte Tetrahedron Letters 1965 3625. lo4 J. E. Baldwin Canad.J. Chem. 1966 44 2051. 361 Woodward-Hoflman Orbital Symmetry Ru Ies to Concerted Organic Reactions of nitrogen and sulphur dioxide from (LXIV) and (LXV) occurs in similar stereospecific reactions (41) (42).lo5-lo7 By the principle of microscopic rever- sibility the retro-reactions should also be concerted and stereospecific. This is only possible chemically for the case of Y = SO, and has been entirely con- firmed for the reverse of reaction (41).lo6 The addition of sulphur dioxide to cis,trans-2,4-hexadiene requires more forcing conditions which results in some isomerisation.lo6 The trans-cyclic compounds (LXV) require higher temperatures for complete reaction on account of the greater steric demand of the incoming methyl group at the transition state. Fragmentation of (LXIV) and (LXV) under photochemical control should proceed by way of the conrotatory mode.It is difficult however to get sufficient energy into these molecules because of the nature of the absorbing chromo- phoric groups (i.e. the energy required for the 7~ - 7 ~ * transition is very large). The problem can be overcome by the use of a sensitizer (e.g. benzene). Thus (LXIV; Y = SO,) upon sensitized irradiation yields the isomeric 2,4-hexadienes (LXVI) (LXVII) and (LXVIII) in the ratio 15:75:10 whereas with (LXV; Y = SOz) the ratio is 60:25:15.108 The evidence indicates that the fragmentation occurs mainly from an electronically excited triplet state before crossing to the ( L X V I 1 ( L X V l I ) tLXVi11) ground state. From (LXIV) and (LXV) (both Y = SO,) the predicted products are respectively (LXVII) and (LXVI) which are the major components of the D.M. Lemal and S. D. McGregor J. Amer. Chern. SOC. 1966 88 1335. lo8 S. D. McGregor and D. M. Lemal J. Amer. Chem. SOC. 1966 88,2858. lo7 W. L. Mock J. Amer. Chem. SOC. 1966 88 2857. lo8 J. Saltiel and L. Metts J. Amer. Chem. SOC. 1967 89 2233. 362 Gill mixtures of isomeric 2,4-hexadienes in each case. The other dienes are accounted for by cis-trans isomerisation initiated by electronically excited (triplet) SO molecules. The results are therefore in accord with the expected conrotatory ring fission but there is some uncertainty concerning the predictive power of the rules for electronically excited states of this nature. In a similar connexion the electronic structure of trimethylenemethane produced by the thermal or photochemical decomposition of 4-methylene-1-pyrazoline has been discussed.The production of the ground-state triplet of trimethylenemethane by the photochemical process is in accord with the orbital symmetry arguments.10g The decomposition of the compounds (LXII k = 2) have yet to be studied in detail. In the case of Y = SO the indications are that reaction (39) is concerted in both forward and reverse directions. Thus cis-hexatriene and sulphur dioxide react at room temperature to give 2,7-dihydrothiepin 1,l-dioxide. The pyrolysis of the latter yields only the cis-hexatriene and sulphur dioxide.79 Other concerted cycloadditions have been examined theoretically.2J02 An example is given by the generalised equation (43). It can be readily shown by R.CH2- (CH=CH)k-CH,.R -t CH,= CH- (CH =CH)k-l-CH=CH + RR (43) an adaptation of the group-theoretical approach of Longuet-Higgins and Abraham~on,~ that concerted cis-elimination in reaction (43) which is charac- terised by a plane of symmetry (ovz) in the transition state (LXIX) is only allowable when k is an odd integer.When k is even or zero concerted elimina- tion is predicted to occur by way of a trans-mechanism and the transition state (LXX) possesses a two-fold axis (C2J of symmetry. When k is even but small the concerted trans-elimination may be exceedingly difficult since the close approach of the two groups R (to facilitate bonding) may cause a severe de- coupling of the n-electron system. A similar situation pertains to ring systems in which the groups R are situated trans to one another. The rules are precisely reversed for the photochemical reactions and may also require further modifica- tion if inversion of geometry of the group R is a real possibility.These conclusions are nicely demon~trated~l by the pyrolyses of 1,4-cycIo- lo9 W. T . Borden Tetrahedron Letters 1967 259. 363 Woodward- Ho ffman Orbital Symmetry Rules to Concerted Organic Reactions hexadiene and 1,3-cyclohexadiene. The first-named compound (k = 1 system) undergoes a smooth unimolecular conversion into benzene and hydrogen in a process of low activation energy. However the 1,3-diene (k = 2 system) is converted into benzene and hydrogen only at high temperatures in a reaction in which radical intermediates are implicated.l1° The production of pyridines in the reaction of cyanogen and of related compounds with 1,3-dienes111 may also proceed by way of the concerted 1,4-cis-elimination (44) of hydrogen from a dihydropyridine intermediate.The pyrolytic dehydrogenation of cyclopentene to cyclopentadiene also occurs predominantly by way of the symmetry-allowed (44) 1 +elimination. 112 The hydrogen transfer from di-imide to the termini of polyenes has also been considered within the context of orbital symmetry relationships.2,102 The reactions are formally similar to the reverse of (43) except that the two electrons of the imide N=N must also be taken into account. Therefore the predictions are that 1,2-addition should be cis (a mode of addition which had been known for some time113) and 1,4-addition should be trans (not yet demonstrated). Further concerted eliminations or additions might well yield to similar theore- tical arguments.Fukui has outlined the relevance of his frontier orbital method to the consideration of stereospecific heterolytic and homolytic additions to multiple bonds.3c 3 Intermolecular Cycloaddition Reactions The most common reactions in this category are dimerisations although a few trimerisations and tetramerisations are known. The selection rules are most readily deduced (by calculation or by group-theoretical arguments) for cyclo- addition reactions for which there is a plane of symmetry (azv) at the transition state as in (LXXI) (LXXII) and (LXXIII).lb In the reactions new a-bonds are formed equivalent in number to the n-bonds consumed. The number of olefinic v-electrons which participate in the reactions are given by the quantities p q and r.Hence the total number of n-electrons involved in the changes represented by (LXXI)-(LXXTII) are ( p + q) ( p + 2 4 and (p + q + 2r) respectively. In their simplest context the symmetry rules require that for reactions run under R. J. Ellis and H. M. Frey J. Chem. SOC. (A) 1966 553; S. W. Benson and R. Shaw Trans. Faraday SOC. 1967 63 985; S. W. Benson and R. Shaw J. Amer. Chem. SOC. 1967 89 5351. ll1 G. J. Janz “1,4-Cycloaddition Reactions” ed. J. Hamer Academic Press London 1967 p. 97. 112 J. E. Baldwin Tetrahedron Letters 1966 2953. 113 C. E. Miller J. Chem. Educ. 1965 42 254. Gill thermal control B(p + q) &(p + 2q) or &(p + q + 2r) should be odd (ix. 4n -1- 2 systems) whereas these quantities should equate to an even number (i.e. 4n systems) for allowed photochemical processes.Actually these con- clusions are only correct for completely cis-additions on all components. In some cases the trans-addition of one or more of the olefinic components in the cyclo- addition process may be possible and the orbital symmetry requirements are accordingly modified. In triple or more complicated additions these topological Table 2 (P +4) Thermal Photochemical 4n cis-trans cis-cis trans-cis trans-trans 4n + 2 cis-cis cis- trans trans- trans trans- cis distinctions can become very complex (e.g. there are three distinct cis-truns- trans combinations) particularly when the symmetry plane is absent from the cycloaddition.2 Completely cis-additions are assumed in the following discussion since cis-trans or trans-trans additions are very uncommon.2 By convention the values (p + q) (p + 2q) or ( p + q + 2r) are used to classify the reactions so that (4 + 2) (6 + 4) and (8 + 2) processes are allowed thermally and (2 + 2) (4 + 4) and (6 + 2) processes should only occur if one olefin is in the lowest electronically excited state.Likewise (2 + 2 + 2) and (2 + 2 + 2 + 2) are representative of a thermally allowed trimerisation and a photochemically allowed tetramerisation respectively. It is convenient to con- sider the thermal reactions first. A. Thermally Controlled Cycloaddition Reactions.-(4 + 2) Reactions. This group comprises the Diels-Alder reaction and related processes which have been the subject of several r e v i e ~ s . l l ~ - ~ ~ ~ It is sufEcient therefore to consider recent examples. 11* J. Hamer ed. “1 ,4-Cycloaddition Reactions” Academic Press London 1967.115 A. Wassermann “Diels-Alder Reactions” Elsevier Amsterdam 1965. llE R. Huisgen R. Grashey and J. Sauer “Chemistry of the Alkenes” ed. S. Patai inter- science New York 1964 p. 739. 117 J. G. Martin and R. K. Hill Chem. Rev. 1961 61 537. 11* S. B. Needleman and M. C. Chang Kuo Chem. Rev. 1962 62,405. 119 M. A. Ogliaruso M. G. Romanelli and E. I. Becker Chem. Rev. 1965 65,261. 365 Woodward-Hoflman Orbital Symmetry Rules to Concerted Organic Reactions In addition to the above selection rules in Table 2 Woodward and Hoffrnann have deduced that (4 + 2) reactions should be characterised by endo-transition states.lc This result which is in accord with experience arises from a considera- tion of the secondary orbital interactions a viewpoint which has been recently criticised.12* Nevertheless the facts remain and there are very few exceptions to the Alder endc-addition rule for reactions in which such a distinction may be drawn.Onc case of exo-addition arises in the reaction (45) of furan with maleic anhydride.121 Further work has shown that the endo-isomer is also formed but in solution and at moderate temperatures it is converted into the more stable exo-isomer by dissociation and recombination.122 With maleic imide the stereo- (45) chemistry of the addition can be controlled more readily by the appropriate choice of reaction temperature.123 Cyclo butadienes also dimerise in a partially non-stereospecific manner since both syn- and anti-dimers are formed (reaction 46).12* When the diene is gener- ated by a non-ambiguous route for example as from the cyclobutadiene-iron tricarbonyl complex the amount of the syn-dimer far outweighs the amount of the anti-cornpo~nd.~~~ Cyclobutadiene is highly reactive both as a diene and as a 2 sYn anti (46) dienophile and it is perhaps not surprising that the secondary interactions (if indeed they are important) are of only minor significance in the overall energy change.The iron tricarbonyl complex of cyclobutadiene is a particularly con- venient source for the free diene and a study of its reactivity towards various dienophiles has led Pettit and his co-workers to the discovery of several interest- ing and important preparative methods to compounds containing the cyclo- butane ring. Typical examples include the synthesis of Dewar benzenes,126 tricyclo[4,2,0,02~5]octa-3,7-dienes,32 and cubane (reaction 47).127 120 W.C. Herndon and L. H. Hall Tetrahedron Letters 1967 3095. lZ1 R. B. Woodward and H. Baer J. Amer. Chem. SOC. 1948,70 1161. lZ2 F. A. L. b e t Tetrahedron Letters 1962 1219. 123 H. Kwart and I. Burchuk J. Amer. Chem. SOC. 1952 74 3094. 124 R. Criegee Angew. Chem. 1962 74 703 and refs. therein; P. S. Skell and R. J. Peterson J. Amer. Chern. Soc. 1964 86 2530 and refs. therein. lZ5 L. Watts J. D. Fitzpatrick and R. Pettit J. Amer. Chem. Soc. 1966 S8 623. lZ6 L. Watts J. D. Fitzpatrick and R. Pettit J . Amer. Chem. SOC. 1965 87 3253. lP7 J. B. Barborak L. Watts and R. Pettit J. Amer. Chem. SOC. 1966 88 1328. 366 Gill 0 (47) In recent years development of the Diels-Alder reaction as a general synthetic method has frequently led to the use of less conventional dienes (e.g.cyclo- butadiene) and dienophiles. The nitrogen analogue of N-phenylmaleimide (LXXIV) has been reported to be a potent dienophile,lZ8 and the reactivity of cyclopropeneslZg and cyclopropan~nes~~~-~~~ has attracted some attention. The cyclopropanones most probably react as the enolate dipole and upon reaction with dienes cyclic 7-membered ring ketones can be prepared. The reaction of 1,2-dibromo-l,2-diphenylpropan-2-one with sodium iodide in acetonitrile in the presence of furan yields a mixture of the cis- and trans-adducts (LXXV) and (LXXVI).131 A possible reaction pathway can be formulated as in (48) but the experimental results are not inconsistent with a non-concerted mechanism. In any event if the reaction is concerted the observed stereospecificity cannot be Ph 7% 6s Br ( L X X V ) H (48) accounted for by the symmetry-allowed secondary interactions of energy levels since such an interaction is not possible in these lz8 R.C. Cookson S. S. H. Gilani and I. D. R. Stevens J. Chem. SOC. (C) 1967,1705. M. A. Battiste and T. J. Barton Tetrahedron Letters 1967 1227. 130 W. B. Hammond and N. J. Turro J . Amer. Chem. SOC. 1966 88 2880. 131 R. C. Cookson M. J. Nye and G. Subrahmanyan J. Chem. SOC. (C) 1967,473. For example see H. Tanida T. Tsuji and T. Irie J. Amer. Chem. SOC. 1967 89 1953; 3 67 Woodward-Hofman Orbital Symmetry Rules to Concerted Organic Reactions The product of the thermal reaction of tetracyanoethylene and N-ethoxy- carbonylazepine was originally assigned the structure (LXXVII) a (6 + 2) adduct.ls2 On the basis of spectroscopic and X-ray studies the structure has (LXXVII) ( L X X V I I I ) been amended to (LXXVIII) the thermally allowed (4 + 2) a d d ~ c t .l ~ ~ A similar conclusion has been recorded for the adducts derived from N-methoxy- ~arbonylazepines,1~~ and also from 3-bromo-N-methoxycarbonylazepine.155 The irradiation of tropone in ether or acetonitrile solution yields among other products a (4 + 2) dimer.136s137 Since this is a disallowed process it seems probable that this product arises from a reaction involving the addition of a trans or Mobius-type tropone molecule to a second molecule of tropone in the ground state.ls7 Alternatively the process may be non-concerted.ls8 The considerations of orbital symmetry are not necessarily restricted to the reactions of neutral stable molecules but encompass the concerted reactions of radicals and of anions and cations.Thus reactions (49) and (50) are allowed thermal processes of the (4 + 2) type whereas (51) is a (2 + 2) reaction and should not occur under thermal control. The thermolysis of perezone (LXXIX) K. Hafner Angew. Chem. Internat. Edn. 1964 3 165. lS5 J. H. van den Hende and A. S . Kende Chem. Comm. 1965 384. J. E. Baldwin and R. A. Smith J. Amer. Chem. SOC. 1965 87 4819. lS5 R. A. Smith J. E. Baldwin and I. C. Paul J. Chem. SOC. (B) 1967 112. 136 A. S. Kende J. Amer. Chem. SOC. 1966 88 5026. lS7 T. Tezuka Y. Akasaki and T. Mukai Tetrahedron Letters 1967 1397 5003. lS8 A. S . Kende and J. E. Lancaster J. Amer. Chem. SOC. 1967 89 5283; G. R. Ziegler and G. S. Hammond ibid.1968,90 513. 368 Gill to the pipetzols (LXXX)139 is most probably a reaction of type (49):l although the allowed (2 + 2 + 2) process [(53); see later] cannot be entirely discounted as an alternative pathway. The hydroxycyclopropanone product (LXXXI) on / ( L X X I X ) (LXXX) might readily be converted into (LXXX) in the presence of fortuitous acidic or basic impurities. The ene synthesis140 is related to the Diels-Alder reaction in that it is essen- tially a (4 3. 2) cycloaddition which requires the movement (54) of the double bond in the ene or diene. The predictable endu-additi~n~l has been observed for the reactions of cis-but-2-ene cyclopentene and trans-but-Zene. The cis-olefins yield mainly the threo-adduct whereas for the trans-olefin the major adduct has the erythru-c~nfiguration.~~~ (6 + 4) Reactions.Since the publication of the Woodward-Hoffmann rules a few examples of (6 + 4) thermal cycloadditions have been reported. It is perhaps significant in view of the recent criticism120 of the suggestion that the product stereochemistry is dependent upon secondary orbital interactions in the transition state that the predicted em-cycloaddition occurs in all cases. Clearly further theoretical approaches are required before it becomes known if these criticisms are justified. In (6 + 4) reactions the operation of highly adverse entropy effects are likely to be of crucial importance to the possible occurrence of such reactions. It is not surprising therefore that in all of the known (6 + 4) cycloadditions lS0 E. R. Wagner R. D. Moss R. M. Brooker J.P. Heeschen W. J. Potts and M. L. Dilling Tetrahedron Letters 1965 4233. 140 W. R. Roth Chimia 1966 20 229. 141 J. A. Berson R. G. Wall and H. D. Perlmutter J. Amer. Chem. Suc. 1966,88 187. 369 Woodward-Hoflman Orbital Symmetry Rules to Concerted Organic Reactions the reacting systems are embedded within rings. In this way the termini of the olefins are held close enough together for sufficient time for the reaction to proceed at a useful rate. The reaction of cyclopentadiene with tropone gives the exo-adduct (LXXXJI),142 and similarly 2,5-dimethyl-3,4-diphenylcyclopentadienone reacts with tropone and cycloheptatriene to furnish (LXXXIII) and (LXXXIV) respecti~ely.~~ (LXXXV) i;, -. (LXXXII) R = R' = H; X = CH (LXXXIII) R = Me; R' = Ph; X = :C = 0 A product arising from the cis (6 + 6) dimerisation of N-ethoxycarbonyl- azepine is formed at 200".This anomalous compound arises from the thermal rearrangement of the initially formed (4 + 2) dimer (LXXXV) which can be isolated at lower (1 30") temperatures.lg3 Other N-substituted azepines yield (6 + 6) dimers upon thermolysis and the intermediacy of the (6 + 4) adduct has also been demonstrated for the cyano-compound (LXXXV; -CO,.Et = CN).14 The cis (6 + 4) compound obtained by the irradiation of tropone in acetonitrile is formed in a non-concerted rea~ti0n.l~~ (8 + 2) Reactions. In the only reactions of this type which have been observed four of the n-bonds have been conformationally frozen within ring structures. 5,6-Dimethylenecyclohexa-l,3-diene which can be generated by the pyrolysis of the thiophen dioxide (LXXXVI) reacts with N-phenylmaleimide to give the imide (LXXXVIL).145 The thermolysis of the cyclobutene (LXXXVIII) in the ( L X X X V J ) (LXX*VlI) 142 R.C. Cookson B. V. Drake J. Hudec and A. Morrison Chem. Comm. 1966 15; S . Ito Y . Fujise T. Okuda and Y . Inone Bull. Chem. Soc. Japan 1966,39,1351. 143 L. A. Paquette and J. H. Barrett J. Amer. Chem. SOC. 1966 88 2590. 144 A. L. Johnson and H. E. Simmons J. Amer. Chem. Soc. 1966,88,2591; 1967,89,3191 145 M. P. Cava and A. A Deana J. Amer. Chem. SOC. 1959 81,4266. 370 Gill presence of maleic anhydride to give (LXXXIX)14s is essentially a similar reac- tion. Of interest is the stereochemistry of the product (LXXXIX). If cis-addition in the h a 1 step is assumed then ring-opening of the cyclobutene must occur by the expected conrotatory pathway.The reaction between 7-methylenecyclo- hepta-l,3,5-triene and dimethyl acetylenedicarboxylate furnishes the azulene Ph Ph ' (LXXXIX) (XCI) when the reaction mixture is worked up in the presence of 0~ygen.l~' That the (8 + 2) cycloadduct (XC) is formed initially is therefore highly probable.41 The treatment of 1,2,3,4,5,6-hexaphenylpentalene with excess of dimethyl acetylenedicarboxylate at 1 60" gives a diethoxycarbonylhexaphenylazulene formulated as (XCII a or b).148 The reaction most probably involves an initial (8 + 2) addition,41 followed by the fission of the cyclobutene ring. If these considerations are correct then the product should be (XCIIa). Thermal cycloadditions of order higher than (8 + 2). The only reaction of this type which has come to the Reviewer's attention is that which occurs between bicyclohepta-2,4,6-triene- 1 -ylidene and tetracyanoethylene.The product (XCIII) has clearly been derived from a trans (14 + 2) cycloaddition rea~ti0n.l~~ The orbital symmetry requirements are precisely reversed for trans-cycloadditions 146 R. Huisgen and H. Seidl Tetrahedron Letters 1964 3381. 147 W. von E. Doering and D. W. Wiley Tetrahedron 1960 11 183. 148 E. Lc Goff J. Amer. Chem. SOC. 1962 84 3975. 149 W. von E. Doering et al. cited in ref. 41. 371 Woodward-Hoflman Orbital Symmetry Rules to Concerted Organic Reactions Ph Ph (a) R = Ph R’ = COsMe (b) R = CO,Me R’ = Ph so that (59) is a symmetry-allowed process. However the treatment of bicyclo- penta-2,4-dien-l-ylidene (XCIV) with tetracyanoethylene does not yield a (10 + 2) adduct but instead a product resulting from a double Diels-Alder (4 + 2) process.150 The required trans (10 + 2) cycloaddition is probably not possible for the less flexible fulvalene system but the cis (10 + 2) photochemical addition seem feasible.7,8-Dimethylenecyclo-octa- 1 3 5- triene reacts readily with various dienophiles by addition across the exocyclic double bonds. The cis-fused adducts (ie. when maleic anhydride or p-benzoquinone are the dienophiles)151 are formed by way of a (4 + 2) and not the disallowed (10 + 2) addition. The failure of attempted reactions of 1,2-diphenyl-3,4-methylenecyclobutene with dienophileP2 may be now ascribed to the operation of an adverse orbital symmetry requirement for the (6 + 2) thermal processes. 160 W. von E. Doering U.S. Dept.Corn. Office Tech. Serv. P.B. Rept. 34 No. 3 8 pp. 1960; Chem. Abs. 1962,56 5883e. lS1 J. A. Elk M. V. Sargent and F. Sondheimer Chem. Comm. 1966 508. 152 A. T. Blomquist and Y. C. Meinwald J. Amer. Chem. Soc. 1957 79 5316. 372 Gill ( 2 + 2 + 2) Reactions. The probability of a termolecular collision of three ethylenic molecules in the precise orientation and with sufficient energy to undergo a cyclic trimerisation is negligible. It is therefore necessary that at least two of the olefinic systems be held within a fairly rigid cyclic molecular frame- work and in the correct relative orientation. Most of the examples are provided by the reactions of dienophiles with norbornadiene.l= The observed endo- addition (60)154 is in agreement with the orbital symmetry arguments. 1,3,5,7- Tetramethylenecyclo-octane a less rigid system reacts with tetracyanoethylene by a similar (2 + 2 + 2) mechanism (61).155 B.Photochemically Controlled Cycloaddition Reactions.-The photochemical cycloadditions leading to carbocyclic systems have been recently re- ~ i e ~ e d ~ J ~ ~ ~ ~ ~ so few examples will be considered here. (2 + 2) Reactions. There is a dearth of examples for the photodimerisation of simple acyclic ethylenic compounds as in reaction (62).15' Rather more common are the (2 + 2) reactions of acyclic non-conjugated olefins in which both of the lS3 A. T. Blomquist and Y. C. Meinwald J. Amer. Chem. SOC. 1959 81,667; H. Heaney and J. M. Jablonski Tetrahedron Letters 1967 2733; F. W. Grant R. W. Gleason and C. H. Bushweller J. Org. Chem. 1965 30 290. 154 R. C. Cookson J.Dance and J. Hudec J. Chem. SOC. 1964 5416. lS5 J. K. Williams and R. E. Benson J Amer. Chem. SOC. 1962 84 1257. lS8 W. L. Dilling Chem. Rev. 1966 66 373. lS7 D. R. Arnold and V. Y . Abraitys Chem. Comm. 1967 1053. 373 Woodward-Hofman Orbital Symmetry Rules to Concerted Organic Reactions n-systems belong to the same molecule,156 and the photodimerisation of con- jugated and small ‘cyclic ~ l e f i n ~ . ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Energy transfer frequently a thorny problem is often achieved by the use of a sensitizer and dimerisation may then occur by way of the triplet ~ t a t e . ~ J ~ ~ J ~ ~ The (2 + 2) photo-cycloaddition reaction is particularly useful in the syn- thesis of cage C O ~ ~ O U ~ ~ S . ~ ~ ~ ~ ~ ~ J ~ ~ J ~ ~ Thus norbornadiene its derivatives and related compounds undergo ready cycloaddition to quadricyclenes upon direct or sensitized i r r a d i a t i ~ n .~ J ~ ~ J ~ ~ J ~ ~ Reaction (63) is a recent example of this type of process.161 In a similar manner facilitated the synthesis of cubanes,12’ (63) (2 + 2) photochemical reactions have homocubanes,ls2 and ‘basketanes’,les besides a whole host of other types of cage C O ~ ~ O U ~ ~ S . ~ Benzene a molecule which does not yield to the Diels-Alder reaction can be persuaded to react photochemically with dimethylacetylenedicarboxylate and related c o m p o u n d ~ ~ J ~ ~ ~ ~ ~ ~ for example reaction (64). The intermediate (2 + 2) adduct (XCV) is implicated in these reactions. On the other hand the reactive 111 I q n L w diene cyclopentadiene will react in the (2 + 2) manner with dichloromaleic anhydride.ls5 l f 8 0.L. Chapman Adv. Photochem. 1963 1 323. R. Srinivasan Adv. Photochem. 1964 4 113. 160 S. J. Cristol and R. L. Snell J. Amer. Chem. Soc. 1958 80 1950; G. S. Hammond N. J. Turro and A. Fisher ibid. 1961 83 4674; P. G. Grassman D. H. Aue and D. S. Patton ibid. 1964 86 4211; D. M. Lemal E. P. Gosschlink and S. D. McGregor ibid. 1966 88 582; W. G. Dauben and R. L. Cargill Tetrahedron 1961,15,197; H. Prinzbach W. Eberbach and G. von Veh Angew. Chem. Internat. Edn. 1965 4 436. 161 E. Payo L. CortBs J. Mantech C . Rivas and G. de Pinto Tetrahedron Letters 1967 241 5. 162 J. C . Barborak and R. Pettit J. Amer. Chem. SOC. 1967 89 3080. 163 S. Masamune H. Cuts and M. G. Hogben Tetrahedron Letters 1966,1017; W. G. Dauben and D. L. Whalen ibid. p. 3743. 164 E.Grovenstein and D. V. Rao Tetrahedron Letters 1961 148; D. Bryce-Smith and J. E. Lodge Proc. Chem. SOC. 1961 333; J. Chem. SOC. 1963,695. 165 H.-D. Scharf Tetrahedron Letters 1967 423 1 . 374 Gill The concerted ( 2 + 2) reactions should according to the Woodward-Hoff- mann rules occur only under photochemical activation (cis-addition assumed) a condition which also applies to the retro-reactions. Of course not all reactions which yield cyclobutanes and cyclobutenes are concerted and many proceed under purely thermal control by way of biradical or dipolar inter- m e d i a t e ~ . ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ However some thermal (2 + 2) reactions have been reported which appear to be concerted processes. In each case a d-electron metal atom is present and the processes are symmetry-allowed on account of the interaction of olefin and metal electronic energy levels.30 Thus quadricyclene is rapidly converted into norbornadiene in the presence of various complexes of platinum palladium and Similarly p-dichlorohexamethylbicyclo[2 2 Olhexa- 2,5-dienedirhodium catalyses the conversion of hexamethylprismane into hexa- rnethylbicyclo[2,2,0]hexa-2,5-diene.31 The reaction of iron or ruthenium tricarbonyl complexes with hexafluoroacetone (XCVI; Y = 0) and with 1,l- dicyano-2,2-bis(trifluoromethyl)ethylene [XCVI; Y = C(CN),] gives the corre- sponding (2 + 2) adducts (XCVII).168 I I M I c d j The conversion of norbornene into the exo,trans,exo-dimer (XCVIII) only occurs if both a cuprous salt (e.g.Cu,Br,) and irradiation are employed. Apparently a termolecular collision is required and this prerequisite is accounted for in terms of an electronically excited tetrahedral copper complex containing three molecules of norbornene and one halogen atorn.le9 If the reaction is con- certed then an interesting point can be raised is the (2 + 2) stereospecific reaction subject to thermal control with the involvement of the metal atom or is the process essentially photochemical ? The answer presumably depends upon the electronic configuration of the norbornene-cuprous bromide complex formed concomitantly with (XCVIII).166 J. D. Roberts and C. M. Sharts Org. Reactions 1962 12 1. 16' P. D. Bartlett et al. J. Amer. Chem. SOC. 1964 86 616 622 628. 16* M. Green and D. C . Wood Chem. Comm. 1967 1062. 169 D. J. Trecker R. S . Foote J. P. Henry and K. E. McKeon J.Amer. Chem. SOC. 1966 88 3021. 375 Woodward-Hofmn Orbital Symmetry Rules to Concerted Organic Reactions (4 + 4) Reactions. The photodimerisation of 2-pyrid0nes,l~~ a-pyrones,171 2-rnetho~ynaphthalene,~~~ and anthra~enesl~~ by formal (4 + 4) processes are well known.158 The dimers are obtained by cis-addition across the positions indicated in (XC1X)-(CII). The reaction of benzene with butadiene to give (CIII) is a more recent case of this type of rea~ti0n.l~~ The seemingly (4 + 4) dimerisation of 1,3-butadienes in the presence of nickel salts is now known to involve two discrete steps a (2 + 2) metal-catalysed reaction followed by a Cope rearrangement (67).175 (6 + 2) Reactions. This is a very rare type of process. The irradiation of solu- tions of tropone yields a (6 + 2) dimer among other but the cycloaddition may not be ~0ncerted.l~~ Ziegler and Hammond have described a (6 + 2) cycloaddition in the photorearrangement of 1,4-epoxy-l,4-dihydro- naphthalene to benzlf10xepin.l~~~ (6 + 6) Reactions.The irradiation of a dilute sulphuric acid solution of tropone yields the dimer (CIV).13' C. Metal-catalysed Cyclo-o1igomerisations.-The Repp6 cyclo-octatetraene synthesis which is thought to involve the concerted tetramerisation of acetylene 170 L. A. Paquette and G. Slomp J. Amer. Chem. SOC. 1963 85 765. 171 P. de Mayo and R. W. Yip Proc. Ciiem. Soc. 1964 84 and refs. therein. 172T. S. Bradshaw and G. S. Hammond J. Amer. Cliem. Soc. 1963 85 3953; see also E. Vogel W. Grimme W. Meckel and H. J. Riebel Angew. Ciien?. Internat. Edn. 1966 5 590; W. von E. Doering and J.W. Rosenthal J . Amer. Chem. Soc. 1966 88 2078. 173 D. E. Applequist and R. Searle J. Amer. Chem. Soc. 1964 85 1389. 174 K. Kraft and G. Koltzenburg Tetrahedron Letters 1967 4357. 176 E. Vogel Annalen 1958 615 1 ; see also G . Satori V. Turba A. Valvassori and M. Riva Tetrahedron Letters 1966 21 1. 376 Gill has been considered in the context of a symmetry-allowed thermal (2 + 2 + 2 + 2) reaction in which the metal electronic energy levels are intimately im- p l i ~ a t e d . ~ ~ The trimerisation of 1,3-butadiene to cyclododeca-l,5,9-triene under similar circ~mstances~~~ is possibly a (4 + 4 + 4) process. The dimerisation of norbornadiene to (CV) in the presence of iron pentacarbonyl and light177 appears to be an obvious (2 + 2 + 2 + 2) reaction but the precise nature of the involvement of the metal atom is not known.4 Sigmatropic Rearrangements A sigmatropic change is one which involves the migration of a o-bond flanked by one or more n-electron systems to a new position within the molecule in an uncatalysed intraniolecular process. The reaction order of these rearrangements is dependent upon the relationship of the a-bond to the termini of the n-system@). Reaction (68) is classified as a sigmatropic shift of order [l illd since the group R' migrates from C(l) to CU). On the other hand the reaction represented by (68) (69) is one of order [i j]le since the o-bond which is cleaved is that joining C(i) and C(j). In the latter instance the carbon framework is numbered starting from the terminus of each n-system. Thus (CVI) is representative of the transition ( C Y J I ) (69) ICY!) 176 G.Wilke Angew. Chem. Znternat. Edn. 1963 2 105. R. C. Cookson J. Hudec and R. 0. Williams ibid. p. 373. D. M. Lemal and K. S . Shim Tetrahedron Letters 1961 368; C. W. Bird D. L. Colinese 377 Woodward-Hofman Orbital Symmetry Rules to Concerted Organic Reactions state of a [ l 51 shift whereas (CVII) illustrates that for a [3 31 rearrangement. A. Sigmatropic Rearrangements of Order [l j].-The orbital symmetry argu- ments are based upon the assumption that these systems [e.g. (CVI)] can be treated as a pair of interacting radicals namely R’ and the framework radical. By the above definition the radical(s) derived from the carbon framework must be of the odd-alternant type and will therefore possess one non-bonding electronic energy level.In thermal reactions this level is occupied by the odd electron and Figure 3 represents the general symmetry properties of the orbital. 0 0 0 0 0 0 0 0 0 Figure 3 1 2 3 4 5 6 7 . . . . On account of the non-bonding property the wave function has nodes at the even-numbered carbon nuclei. Two different transition-state geometries have been defined for [l j ] sigma- tropic rearrangements.ld In the suprafacial route the migrating atom or group (R’) is associated at all times with the same face of the rr-system; in the antara- facial process the migrating group (R’) is passed from the top face of one carbon terminus to the bottom face of the other. Assuming that the migrating atom or group (R’) employs a a-type orbital Figure 3 reveals that [l 31 [l 71 [l 111 . . . . shifts [i.e. 8(j - 3) is zero or even] must be antarafacial whereas [l 51 [l 91 .. . . shifts [i.e. B(j - 3) is odd] must occur by the suprafacial route.ld Clearly all [l 31 antarafacial shifts will be impossible as will be [l j ] antara- facial shifts in rings of moderate size on account of the serious uncoupling of the polyenyl radical orbitals. In reactions involving the first excited state concerted shifts are governed by rules which are precisely the reverse of the above. This arises from the different symmetry properties of the lowest excited-state levels of the polyenyl radicals. Thermal [1 31 shifs (antarafacinl). As might be expected there is no known example for the migration of a-type radicals (e.g. H). Berson and Nelson have found that the thermolysis of [exo-2H7]bicyclo[3,2,0]hept-2-en-endo-6-yl acetate (CVIII) yields exo-cis-6-deuterionorborn-2-en-5-yl acetate (CIX) a reaction which involves the concerted SzlgrafaciaZ migration (70) of a n-type radi~a1.l~~ In rr-type migrations in which the selection rules are precisely reversed owing to rear-side orbital overlap inversion at the migrating centre is expected.ld This 178 J.A. Berson and G. L. Nelson J. Amer. Chem. SOC. 1967 89 5503 and refs. therein. 378 Gill . I . AcO l* expectation is realised in reaction (70). The size of the groups attached to the migrating atom are probably of critical importance because of their steric inter- action with the allyl framework. Photochemical [l 31 shifts (suprafacial). Of the photochemical [l 31 shifts that have been reported a few appear to involve the concerted suprafacial migra- tions of a-type centres.The most clear-cut example is provided by the sequence (71). The cis-trans equilibration of the migrating allyl group seems to be an integral part of the rearrangement since neither the starting material nor (CX) were isomerised under the reaction conditions.179 The photochemical conversions of verbenone into chrysanthenone,’80 bull- valene into (CXII),181 and (CXIII) into (CXIV)lS2 might likewise occur by concerted suprafacial shifts. The cobalt-catalysed conversion of allylbenzene into trans-pr~penylbenzenel~~ has been suggested30 as an example of a supra- facial [l 31 shift which occurs thermally because of the electronic involvement of the metal atom. The formation of 2-phenylcycloheptatriene from the irradia- tion of the 7-phenyl isomerls4 could involve a [l 31 hydrogen shift.The altern- ative possibility of two successive [l 71 shifts seems more probable although 1 -phenylcycloheptatriene is not isolated from the reaction mixture since the maximum linear conjugation of the framework radical seems to be generally preferred. 17a R. F. C. Brown R. C. Cookson and J. Hudec Chem. Comm. 1967,823. la0 J. J. Hurst and G. H. Whitham J. Chem. SOC. 1960,2864. 181 M. Jones J. Amer. Chem. SOC. 1967,89,4236; see also R. N. Warrener and J. B. Bremner Rev. Pure Appl. Chem. 1966 16 117. 182 W. G. Dauben and W. T. Wipke Pure Appl. Chem. 1964 9 539. la3 L. Roos and M. Orchin J. Amer. Chem. Soc. 1965 87 5502. la4 A. P. Ter Borg and H. Kloosterziel Rec. Trav. chim. 1965 84,241. 379 5 Woodward-Hofman Orbital Symmetry Rules to Concerted Organic Reactions Thermal [l 51 shifts (suprafacial).There are many examples of thermal [l 51 shifts of hydrogen particularly for ring systems and the concerted suprafacial pathway is beyond doubt for the migration of this atom. Considerably less is known about the migratory aptitude of other atoms and groups. Much of the earlier work on dienyl and homodienyl hydrogen shifts both in acyclic (74) and cyclic (75) systems has been the subject of an excellent and concise review.185 The thermolysis of 3,4-epoxycyclo-octene and of 3,4-epoxy- cycloheptene yields respectively cis,cis-3-oxa- 1,4cyclononadiene and cis,cis- 3-oxa-l,4-cyclo-octadiene. These reactions are analogous to the homodienyl H (75) n =O-5 [I 51 hydrogen shifts.lS6 The n-bonds of carbonyl groups also appear to be capable of participating in homodienyl hydrogen shifts for example (76).lS7 lS6 D.S. Glass R. S. Boikess and S. Winstein Tetrahedron Letters 1966 999; see also W. R. Roth and J. Konig Annalen 1966 699 24. lg6 J. K. Crandall and R. J. Watkins Tetrahedron Letters 1967 1717. lS7 R. M. Roberts and R. G. Landolt J. Amer. Chem. SOC. 1965 87,2281. 380 Gill The migration of a hydrogen atom specifically from the cis-methyl group has been demonstrated by a deuterium-labelling experiment.lsa +y* n H '1 (76) Me.CO.CH,.CH,.C(Me) = CH2 The migration of hydrogen atoms in the thermal reactions of 7-substituted cycloheptatrienes occur by way of [l 51 shifts although superficially the [l 71 and [l 31 pathways also seem possible. The [l 51 suprafacial migrations which occur as predicted by the Woodward-Hoffmann theory are summarised in reaction (77).189-195 H Few examples of the [l s] migrations of groups other than hydrogen have been reported.The definitive cases are again for medium-size ring systems. Upon heating to 300" (gas phase) 3,7,7-trimethylcycloheptatriene is equilibrated with the 1,7,7- and 2,7,7-isomers. Although no experimental basis is available for deciding whether the interconversions proceed through biradical or con- certed [l 51 shifts the latter can be rationalised by the scheme (78),lg6 which is also in accord with the results of deuterium-labelling experiments. However 7,7-dicyanonorcaradiene is isomerised above 100" first to 4,7-dicyanocyclo- heptatriene and thence to 1,4- and to 1,5-dicyanocycloheptatriene. A series of concerted [ l 51 shifts (79) appears to afford the most reasonable interpretation of these observations.86 The apparently different behaviour of a methyl sub- 188 R.M. Roberts R. N. Greene R. N. Landolt and E. W. Heyer J . Amer. Chem. SOC. 1965 87 2282. leu A. P. Ter Borg H. Kloosterziel and N. Van Mews Proc. Chem. SOC. 1962 359; Rec. Trav. chim. 1963 82 717. lgo K. W. Egger J. Amer. Chem. SOC. 1967 89 3688. lgl A. P. Ter Borg E. Razenberg and H. Kloosterziel Rec. Trav. chim. 1965,84,1230. lg2E. Weth and A. S. Dreiding Proc. Chem. SOC. 1964 59. lg3 T. Nozoe and K. Takahashi Bull. Chem. SOC. Japan 1965,38,665. lg4 A. P. Ter Borg and H. Kloosterziel Rec. Trav. chim. 1965 84,245; see also ref. 81. lgS R. W. Murray and M. L. Kaplan J. Amer. Chem. SOC. 1966 88 3527. lg6 J. A. Berson and M. R. Willcott J. Amer.Chem. SOC. 1966 88 2494. 381 Woodward-Hogman Orbital Symmetry Rules to Concerted Organic Reactions NC H n (79) stituent and a cyano-substituent [e.g. (78) versus (79)] may be due to the smaller steric requirement of the linear nitrile group but this cannot be the complete answer since (CXV) and (CXVI) are isomerised to the norcaradiene (CXVII). A scheme (80) analogous to (78) appears to be operative.86 Likewise the methyl- 7,7-dicyanonorcaradienes are thermally isomerised by this circulatory mechanisin.lS7 Further heating of (CXVII) produces (CXVIII) the [l 51 CN-shift product. 86 (80) (cxv 1 0 ) lg7 J. A. Berson P. W. Grubb R. A. Clark D. R. Hartter and M. R. Willcott J. Amer. Chem. SOC. 1967 89 4076. 382 Gill Atoms other than carbon should be able to migrate with reasonable facility.ls5 The ready thermal isomerisations of the phosphine (CXIX)lg8 and the amino- pentadiene (CXx)lg9 are examples of the possible extension of these processes.(CYI x) Ph I UPh He \,, 90' _L* Photochemical [I 51 shifts (antarafacial). In accord with the predicted antara- facial migration there appears to be no recorded example of [l 51 shifts in ring system where the antarafacial process is clearly impossible. The alternative suprafacial migration of groups which can utilise a low-lying n-orbital also appears to be unknown. In the irradiation of some acyclic 1,3-dienes products of [1 51 shifts are occasionally obtained. Other photochemical processes (e.g. ring closure to cyclobutenes) however usually afford the major reaction path- way. Thus 2-methyl-trans-l,3-pentadiene upon irradiation yields (besides the cis-isomer of the starting material) 4-methyl-l,3-pentadiene among other products.Likewise 1 -cyclohexyl-l,3-butadiene (cis- and trans-isomers) is converted into (CXXI; cis and trans).45b a (CXXI ) Thermal [l 71 shifts (anfarafacial). There appears to be no known example of a thermally initiated [l 71 shift in cyclic systems even for atoms which could migrate by the suprafacial pathway by using a -type orbital The [l 71 migra- tion of hydrogen atoms in the thermal reactions of acyclic hepta-1,3,5-trienes has been known for some time but the systems studied are very similar to one lS* T. J. Katz C. R. Nicholson and C. A. Redly J. Amer. Chem. SOC. 1966 88 3832. lo9 H.-W. Bersch and D. Schon Tetrahedron Letters 1966 1141. 383 Woodward-Hoflman Orbital Symmetry Rules to Concerted Organic Reactions another.Perhaps the best known of these reactions are the thermal equilibra- tions (83) of vitamin D and precal~iferol,~~ and of calciferol-precalciferol analogues.200 The compound (CXXII) which is obtained by irradiating methyl- a P ACO (OX% i r ) (84) dehydroursolate acetate is rearranged by heat (84) in a manner analogous to reaction (83).loo The [l 71 antarafacial migration of hydrogen atoms reasonably accounts for these observations. In these molecules the heptatrienyl unit must adopt a spiral conformation in the transition state (CXXIII) a geometrical disposition of groups which is admirably suited to the antarafacial hydrogen shift. Photochemical [l 71 shifts (suprafacial). The photochemical excitation of cyclo- heptatrienes furnishes compounds which result from the [l 71 migration of a substituent group located at C(7).Most usually this group is a hydrogen atom. These [l 71 hydrogen ~ h i f t ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ * ~ ~ which are summarised in (85) un- doubtedly proceed by the suprafacial pathway because of the presence of the ring system. The distribution of the products depends on the nature of the J. L. M. A. Schlatmann J. Pot and E. Havinga Rec. Trav. chim. 1964 83 1173. 201 W. von E. Doering and P. P. Gaspar J. Amer. Chem. SOC. 1963 85 3043. 202 R. Roth Angew. Chem. Znternat. Edn. 1963 2 688. 384 Gill H (85) x = D,184,301,202 Me,304 et,aO2 OMe,68 c H 184 C8H4-4-(7-C,H,),'06 8 6 group X and not all of the isomeric cycloheptatrienes are formed in each of the above cases.For example with the 7-phenyl compound as starting material only 2-phenylcycloheptatriene is isolated. This compound could arise by two consecutive [l 71 H-shifts or by a [l 31 H-shift. In these migrations the frame- work radical usually assumes the maximum degree of linear conjugation1& so that the [l 31 shift seems less probable. The photochemically initiated [l 71 shifts of hydrogen atoms in l-substituted cycloheptatrienes have also been investigated. The proportions in which the isomeric products are formed is again dependent upon the nature of the 1- substituent in the starting materiaLM These reactions are essentially the final two stages of (85). The migration of methyl groups may be observed in suitable cases for example (86)203 and (87),204a but data on other types of [l 71 migration (87) are lacking.The successes of the rules (which are based on an acyclic model) for cyclic systems is almost fortuitous. During the concerted migration of a group (R) formerly attached to C(n) of a C,H,,+,R monocycle the cyclic framework radical (C,H2,+ ,) is completely cyclically conjugated and its electronic energy states are distinctly different from those of the analogous acyclic system. How- ever the symmetry restrictions based on the more realistic cyclic model are in good agreement with those formulated for the less satisfactory acyclic model except that in the cycloheptatriene system [l 31 [l 51 and [l 71 shifts are per- mitted under photochemical Experiment does not allow a choice a03 L. B. Jones and V. K. Jones J. Amer. Chem. SOC. 1967,89 1880 [the conclusions of foot- note (9) of this paper are incorrect].204 (a) 0. L. Chapman and S. L. Smith J. Org. Chem. 1962 27 2291 ; (b) A. G. Anastassiou Chem. Comm. 1968 15. 385 Woodward-Hofman Orbital Symmetry Rules to Concerted Organic Reactions between the two predictions at present since only the photochemical [l 71 shifts have been observed. In a molecule such as 3-t-butylcycloheptatriene however a single photochemical [l 51 hydrogen shift would alleviate a considerable amount of steric strain and a study of such a system might be worthwhile in the above connexion. Other [l,jJ shifts. Migrations of order higher than [l 71 are likely to be difficult even in ring systems because of the need for near planarity (i.e. substantial through-conjugation) of the framework radical. Recently however the thermal sigmatropic migrations of 1-substituents of 1-substituted (i.e.CH,) and 1,19- disubstituted (i.e. C0,Et) nickel tetradehydrocorrins have been described. The migrations may be formally regarded as [l 171 (for CH,) and [ 1 51 or [ 1 171 (for C0,Et) shifts respectively. Preliminary Huckel calculations indicate205 that the highest occupied orbital of the metal-free macrocycle has the correct sym- metry for the suprafacial migration from (Cl) to C(2). B. Sigmatropic Rearrangements of Order [i j].-Uncatalysed intramolecular sigmatropic changes of order [i j ] have also been considered by Woodward and Hoffmann in a further comrnuni.cation.le Their theoretical treatment assumes that these systems can be likened to the intimate interaction of a pair of polyenyl radicals possessing respectively i and j conjugated atoms.A topological distinc- tion on both framework radicals is thus possible and the main theoretical con- clusions can be summarised as in Table 3 (s = suprafacial a = antarafacial interactions).2 Table 3 li i l Thermal Photochemical [3,31 s-s S-LZ [3,51 S-a s-s ~ 5 1 s-s S-Q The thermal [3 31 migrations are more widely known as the Cope and Ciaisen rearrangements. Both reactions are of great theoretical and preparative import- ance and accordingly the amount of experimental work which has been,2°6,207 and continues to be,208 carried out on [3 31 shifts is enormous. Of particular 206 R. Grigg A. W. Johnson K. Richardson and K. W. Shelton Chem. Comm. 1967 1192. S. J. Rhoads in “Molecular Rearrangements” ed. P. de Mayo Interscience New York 1963 Part I p.655. 207 E. Vogel Angew. Chem. Internat. Edn. 1963 2 1; W. von E. Doering and W. R. Roth ibid. p. 1 1 5. 208 For example see B. Miller J. Amer. Chem. SOC. 1965,87,5515; S. F. Reed J . Org. Chem. 1965 30 1663; E. Vogel R. Erb G. Lenz and A. A. Bothner-By Annalen 1965 682 1 ; H. A. Staab and F. Vogtle Chem. Ber. 1965,98,2681,2691; H. A. Staab and C. Wunsche ibid. p. 3479; A. Viola E. J. Iorio K. K. Chen G. M. Glover U. Nayak and P. Kocienski J. Amer. Chem. SOC. 1967,89,3462; W. von E. Doering and W. R. Roth Tetrahedron 1963 19 715; G. Schroder Angew. Chem. Internat. Edn. 1963 2 481; A. F. Thomas Chem. Comm. 1967 947; L. A. Paquette T. J. Barton and E. B. Whipple J. Amer. Chem. SOC. 1967,89 5481 ; G. M. Blackburn W. D. OlIis J. D. Plackett C. Smith and I. 0. Sutherland Chem.Comm. 1968 186. 386 Gill interest within the framework of the Woodward-Hoffmann theory are those experimental observations which allow a direct test of the theoretical predictions. One such prediction is that in thermally controlled [3 31 sigmatropic rearrange- ments there should be a distinct preference for a chair-like transition stateale Some thirteen years previously Doering and Roth had recorded such a prefer- ence for the particular transition-state geometry represented by (CXXIV) from the results of their classical work on the thermolysis of mesu- and (j-)-3,4- dimethylhexa-1 ,5-dienes.207~209 The high stereospecificity of [3 31 migrations has been further confirmed by work on the thermal reactions of optically active N-allylamines [amino-Claisen rearrangement (89)I2l0 and of optically active hepta-l,Sdienes [Cope rearrangement In each case two products are formed which possess opposite configurations not only at the double bonds but also at the centres of asymmetry.The optical purity of each product is very Ph I Ph I Ph Ph (89) Ph Ht I I high. These observations are only consistent with four-centre chair-like transi- tion states [e.g. (CXXV) and (CXXVI) for reaction (90)l2I1 for the migration and not with the alternative six-centre boat conformations. 209 W. von E. Doering and W. R. Roth Tetrahedron 1962 18,67. alo R. K. Hill and N. W. Gilman. Tetrahedron Letters 1967 1421. R. K. Hill and N. W. Gihan Chem. Comm. 1967,619. 387 Woodward-Hofman Orbital Symmetry Rules to Concerted Organic Reactions Alternative theoretical treatments of the observed preference for the four- centre transition state conformation in concerted thermal [3 31 migrations have been d i s c ~ s s e d .~ ~ ~ ~ ~ The photochemical [3 31 migrations should proceed by way of a suprafacial- antarafacial interaction of the two allyl systems in the transition state a type of interaction which is not favourable. There is no clear-cut example of a con- certed photochemical [3 31 migration.213 However [3 51 shifts should occur with reasonable facility under photochemical control and it seems likely that a migration of this type has been observed. Upon irradiation allyl 2,6-dimethyl- phenyl ether (isotopically labelled mainly at the position indicated *) is partially converted into 2,6-dimethyl-4-(prop-3-enyl)phenol(91). The radioactive tracer is approximately equally distributed between C(l) and C(3) of the allyl system in (CXXVII).214 This result can be explained in terms of two competing migrations; a two-stage thermal migration (i.e.para-Claisen rearrangement) and a one-step concerted [3 51 shift passing through an eight-membered transition state under photochemical control. If the two mechanisms are of about equal importance a 50% scrambling of the label would be expected. A thermal [3 5] rearrange- OH ment in a related system21S seems less likely to involve a one-step concerted reaction passing through an eight-membered transition state.ldp2O6 Very recently a [5 51 concerted migration (92) has been described,216 although any preference for a chair-like four-centre transition statele has yet to be detected. 5 Miscellaneous Reactions The orbital symmetry arguments have been extended to rationalise and predict virtually every concerted organic reaction.ls2 Examples include sigmatropic changes within ionic s p e c i e ~ l ~ ~ ~ ~ ' valence tautomerisms of cyclic conjugated 212 M.Simonetta and G. Favini Tetrahedron Letters 1966 4835; M. Simonetta G. Favini C. Mariani and P. Gramoccioni J. Amer. Chem. SOC. 1968 90 1280. 213 See for example H. N. Subba Rao N. P. Damordaran and S. Dev Tetrahedron Letters 1967,227. 214 K. Schmid and H. Schmid Helv. Chim. Acta 1953 36 687. 21K P. Fahrni and H. Schmid Helv. Chim. Acta 1959 42 1102. 216 Gy. Frater and H. Schmid Helv. Chim. Acta 1968 51 190. 217 N. P. Phelan H. H. Jaff6 and M. Orchin J. Chem. Educ. 1967,44,626. Gill R (92) (CXXVIII) (a) R = R’ = H ; (b) R = H R’ - Me; (c) R = Me R’ = H.ole fin^,^,^^ double terminal additions of acetylene to polyenes,2,102 and general- ised Cope rearrangements.2,102 Because of the lack of experimental verification of the predictions in many of these cases it would be premature to consider these systems here. 6 Conclusions The agreement between experimental observation and the predictions of Wood- ward and Hoffmann’s elegant theory are extremely satisfactory. The reper- cussions of orbital symmetry arguments have already brought marked changes in the philosophy of the molecular orbital rnethod.2,21s The rules (which one must remember are permissive for a particular concerted reaction but not obligatory) will undoubtedly be of great assistance in certain areas of preparative organic chemistry.I thank Professor A. W. Johnson for interest and encouragement. The Review has developed from notes prepared for a post-graduate lecture course at the University of Nottingham. Many discussions which have been shared with Drs. R. E. Grigg B. W. Bycroft T. J. King and M. R. Willis are warmly acknowledged. m8 R. Hoffmann and R. A. Olofson J. Amer. Chem. SOC. 1966 88,943; R. Hoffmann A. Imamura and G. D. Zeiss ibid. 1967 89 5215. 389

ISSN:0009-2681    

DOI:10.1039/QR9682200338

出版商:RSC    

年代:1968

数据来源: RSC

摘要:

Molecular Rearrangements Related to the Claisen Rearrangement By A. Jefferson* DEPARTMENT OF CHEMISTRY UNIVERSITY OF SALFORD ENGLAND F. Scheinmannt OROANISCH-CHEMISCHES INSTITUT DER UNIVERSITAT ZURICH SWITZERLAND The Claisen rearrangement involves the thermal transformat ion of an allyl vinyl ether into a homoallylic carbonyl compound (Scheme 1) by a concerted intramolecular pathway. The all-carbon analogue is known as the Cope rearrangement. The Claisen rearrangement is better known for the rearrangement Scheme I of allyl phenyl ethers [(3) -+ (4) -+ (5) and (3) + (6) - (7)] where the vinyl moiety is now an integral part of an aromatic ring and this reaction has been extensively investigated.l The more unusual aspects however have been largely beyond the scope of previous reviews and it is now proposed to deal with some extensions of the Claisen rearrangement.Some emphasis will be given to synthetic aspects. This Review is not comprehensive but selects some examples which deviate from the Claisen rearrangement and where possible correlates them in mechanistic terms. Fortunately the related rearrangements can be readily understood by first a brief consideration of the essential steps of the mechanism of the Claisen rearrangement and by use of the selection rules for sigmatropic changes. 1 Mechanism of the Claisen Rearrangement The thermal rearrangement of allyl phenyl ether (3) proceeds by an intra- molecular pathway to give an ortho-dienone (4) which rapidly enolises to the * Present address Department of Chemistry Western Australia Institute of Technology Perth Western Australia.t On leave from the University of Salford England. 391 6 Molecular Rearrangements Related to the Claisen Rearrangement 2-allylphenol (5). The net result is that the y-carbon atom of the allyl phenyl ether now becomes directly attached to the benzene ring in the formation of 2-allylphenol (this process is often referred to as ‘inversion’ of the allyl group and in this sense the term gives no information about the stereochemical properties of the rearrangement) and this has been demonstrated by labelling experiments and the use of substituents in the allyl group (Scheme 2). When the ortho positions are substituted (e.g. R3 = alkyl) and enolisation is impossible a Cope type rearrangement gives apara-dienone (6) and hence a 4-allylphenol(7). The para Claisen rearrangement occurs with two reversals of ends of attachment of the allyl group and labelling experiments show that the carbon atom that was attached to the ether oxygen atom is now directly attached to the benzene ring.l Kinetic results show that both ortho and para rearrangements are first order reactions and negative entropies of activation support the involvement of highly ordered cyclic transition states.Further details about the transition state are vague because the Claisen rearrangement is largely insensitive to the probes for ionic and free-radical character. Since the rearrangement cannot be classified in classical terms as either a homolytic or a heterocyclic reaction it has been referred to as a reaction which proceeds by a ‘no mechanism’ pathway.lb CHR’ OH ( 5 ) ‘CH q! (7) ZHP ( 6 ) CHR’ Scheme 2 The stereochemical relationship between reactant and product in the Claisen rearrangement can be derived by considering a transition state where the six atoms concerned in bond and electron redistribution can adopt either a quasi-chair (a) or a quasi-boat (y) conformation.This leads to the conclusion that the rearrangement of trans- and cis-ay-dimethylallyl phenyl ether (10 and 11) gives (a) D. S. Tarbell ‘The Claisen Rearrangernentin Organic Reactions’ ed. R. Adams J. Wiley New York 1944 vol. 11 p. 1. (b) S. J. Rhoads ‘Rearrangements Proceeding through “No Mechanism Pathways” ’ in ‘Molecular Rearrangements’ ed. P. de Mayo Interscience New York 1963 vol. I p. 655. (c) H. J. Shine ‘Aromatic Rearrangements’ Elsevier Amsterdam 1967 p. 89. 392 Jeflerson and Scheinmann largely the trans-2-(a~-dimethylallyl)phenol (12) and that asymmetry can be induced.Thus Goering and Kimoto2 showed that R( +)-trans-acy-dimethylallyl phenyl ether gave a mixture of cis- (1 3) and mainly trans-2-(ay-dimethylallyl)- phenol (12). Hydrogenation of this mixture gave S(-)-2-(2-pentyl)phenol. In an independent study3 it was shown that both cis- (1 1) and trans-ay-dimethylallyl phenyl ether (10) gave largely the truns-2-(ay-dimethylallyl)phenol (12) on thermal rearrangement. These results do not distinguish between the quasi-boat and quasi-chair transition states since both conformations lead to the same result. This can be illustrated by use of modified Newman projection formulae (14-17). The three carbon atoms of the ally1 group and their substituents are represented by the thickened lines to show that they are in a plane above but parallel to the benzene ring.Two possible quasi-chair transition states (14) and (15) are shown for the trans-ether (10) but conformation (14) will require the lower activation energy because the non-bonding interaction of the methyl groups is at a minimum. For the quasi-boat conformations (16 and 17) similar considerations favour (16). Both conformations (14 and 16) lead to the trans-2-(ay-dimethylallyl)phenol (1 2). From the above results it is not possible to exclude the quasi-boat conformation in the transition state but more definitive experiments which favour a quasi-chair conformationga will be the subject of another review.4b The advantages of the stereospecificity of the Claisen rearrangement in synthetic work are immediately obvious not only can the rearrangement be used to create a new stereochemical centre by asymmetric induction but in H.L. Goering and W. I. Kimoto J. Amer. Chem. SOC. 1965 87 1748. E. N. Marvell J. L. Stevenson and J. Ong J . Amer. Chem. SOC. 1965 87 1267. (a) H. Scmid Chemical Society Symposium on Stereochemistry Sheffield December 1966. (b) Chem. in Britain to be published. 393 Molecular Rearrangements Related to the Claisen Rearrangement addition it provides a valuable method for correlating configurations especially in alicyclic ~hernistry.~ 2 Sigmatropic Rearrangements The Cope and Claisen rearrangements have recently been classified by Woodward and Hoffmann6 as sigmatropic changes of the order [i,j] where i and j each corresponds to 3.The order [iJ] refers to the extent of migration of a sigma bond along one or more 7r-electron systems during a concerted electron and bond reorganisation process where the actual distance traversed is i - 1 and j - 1 atoms. Thus in the Claisen and Cope rearrangements the bond which appears to connect two ‘allyl’ systems at atoms designated 3,3 appears to move down two n-electron systems to positions 1,l (Scheme 1). It can be shown by use of the phase relationships of the highest occupied molecular orbital that for rearrangements of the order [ i j ] in which both i and j are greater than unity thermal changes are symmetry-allowed when i + j = 4n + 2 while photochemically excited transformations are permitted when i + j = 4n. This rule requires I odification for migrations to a charged site.For a pictorial interpretation of the rule based on Hiickel molecular orbital theory consider the welectron system over the three carbon atoms of an allyl group.’ These can be described by three molecular orbitals #2 and #3 and it must be remembered that no more than two paired electrons can occupy a given orbital. The three energy levels described as bonding non-bonding and anti-bonding are diagrammatically represented in Figure 1. A Anti-bonding orbital g - i z 1 N Non-bonding orbital T l B Bonding orbital Figure 1 R. K. Hill and A. G. Edwards Tetrahedron Letters 1964 3239. R. B. Woodward and R. Hoffmann J. Amer. Chern. Soc. 1965,87,2511,4389 and personal communication from Professor R. Hoffmann. For an introduction to molecular orbital theory see J. D. Roberts ‘Notes on Molecular Orbital Calculations’ W.A. Benjamin New York 1962; A. Streitweiser jun. ‘Moleculaf Orbital Theory for Organic Chemists’ John Wiley and Sons New York 1961. 394 Jeferson and Scheinmann The positive and negative signs refer to the phase of the wave function and bonding can only OCCUT between atomic orbitals of the same phase. In the allyl radical which has three electrons indicated by arrows $2 represents the highest occupied molecular orbital. If it is assumed that the Cope rearrangement proceeds by formation and combination of allyl quasi-radicals in the transition state the diagram (18) of the highest occupied molecular orbitals shows that the [3,3] change is permitted whereas the [1,3] sigmatropic shift is not. [Experimental dataaa show that the conclusions are identical for the rearrangement of allyl vinyl ethers.] In this concept of the transition state atoms 1,l and 3,3 are associated in order that relative phases of the allyl orbitals are maintained but the manner in which the two molecular orbitals perturb one another has not been considered.The Figure is consistent with the (18) stereochemical requirements for intramolecular allylic rearrangements which demands that bond breaking and bond formation both occur on the same face of the allyl group. This is classified as a suprafacial migration in the terminology of Woodward and Hoffmann.6 For the Claisen rearrangement of allyl phenyl ether and for the ally1 migration in the all-carbon analogue 4-phenylb~t-l-ene~~~ the highest occupied molecular orbital for the seven n-electrons of the phenoxyl and benzyl radicals will be given by $4 in each case (19 and 20).9 (Ally1 migration to the ortho position in allyl phenyl ether or in 4-phenylbut-l-ene can be regarded as either a [3,3] or a [3,7] sigmatropic change depending on which way the sigma bond is considered to migrate along with the aromatic ring.Where there is a choice in agreement with the views of Professor Roald Hoffmann the lowest possible numbers have been assigned to a sigmatropic change. This means that all Claisen rearrangements can be regarded as [3,3] changes regardless of whether the reaction occurs in aliphatic or polynuclear aromatic compounds. For the theoretical interpretation all the 7~ 9 - electrons in the orbitals must v4 benzyl y4 phenoxyl be considered (a) Y . Pocker Proc. Chem. SOC. 1961 141; (b) W.von E. Doering and R. A. Bragole (a) G. WagniBre personal communication; (b) E. Heilbronner and P. A. Straub ‘Hiickel Tetrahedron 1966 22 385. Molecular Orbitals’ Springer-Verlag Berlin 1966. Molecular Rearrangements Related to the Claisen Rearrangement which necessitates the use of aromatic orbitals in aromatic rearrangements.) It can be seen that the highest occupied molecular orbitals for the benzyl and phenoxyl radicals are similar and the only differences that emerge are due to the position of the nodal planes (shown by broken lines) with respect to the substituent atom. Thus the phase of the wave function changes sign between the substituent atom and the ortho positions and changes again at the para position. The consequences of these features in the Claisen rearrangement are now clear.The transition state will be represented by z,b2 from the allyl radical and z,b4 from the phenoxyl radical (21). By the very nature of the wave functions allyl migration to the ortho position is allowed thermally and this may be followed by a further rearrangement of the allyl group to the para position. However since the phase of the wave function is the same at both ortho positions an ortho-ortho migration with ‘inversion’ of an allyl group cannot occur thermally by a concerted mechanism. An alternative step-wise mechanism will be discussed in this Review to account for products arising from apparent ortho-ortho migrations. n (21) A general rule for sigmatropic changes which includes migrations to charged sites and takes account of the suprafacial route can equally be given in terms of the total number of electrons involved in the transition state.Thus if there are (4n + 2) r-electrons in the transition state a sigmatropic change is thermally allowed by the suprafacial route. For rearrangements which occur under photochemical stimulus in the first excited state these rules may be reversed. It is a consequence of the Woodward-Hoffmann rules that a Claisen rearrangement which proceeds with ‘inversion’ by photochemical excitation cannot be a concerted process. 3 Rearrangement of Ally1 Aryi Ethers Rearrangement to the para Position in the Absence of one ortho Substituent.- The thermal rearrangement of allyl aryl ethers to thep-allylphenol (Scheme 2) is possible by two [3,3] sigmatropic transformations and in principle can occur whenever the usually rapid enolisation of the ortho-dienone intermediate is hindered or when the enolisable hydrogen is replaced by a substituent.396 Jeflerson and Scheinmann The 120 examples collected for an early Reviewla show that an allyl group migrates almost exclusively to an ortho position if one is free. The few exceptions quoted 1 -( 3,3-dimet hyl allyl)-2-methoxyphenyl ether allyl 2-hydroxyphenyl ether and allyl 2,3-methylenedioxypheny1 ether all gave para rearrangement products. In a more recent review l b Rhoads pointed out that in the case of allyl aryi ethers which are substituted at only one ortho position a mixture of products can arise and should be expected from competitive ortho and para rearrangement. Marvell and his co-workersl0 confirmed this suggest ion by showing that in the rearrangement of allyl 2-alkylphenyl ethers (22) the predominating product in each case 6-allyl-2-alkylphenol(23) was accompanied by a small amount (ca.10%) of the 4-allyl-2-alkylpheno1 (25). The size of the alkyl group which varied from methyl to t-butyl had a surprisingly small influence on the ratio of the products formed. The ratio of ortho to para product is dependent on the relative rates of migration to the two ortho positions and the partitioning of the dienone (24) between return to the ether (22) and migration to the para position. Migration of the allyl group to the para position of ethers possessing a vacant ortho position has been used as a synthetic route to natural phenols with yy-dimethylallyl side chains. Thus the Claisen rearrangement of the yy-dimethylallyloxyxanthone (26)11 gave the expected om-dimethylallylphenol OH ' 0 0 q / OH 4 ( 24 QR lo E.N. Marvell B. Richardson R. Anderson J. L. Stephenson and T. Crandall J. Org. Chem. 1965,30 1032. l1 (a) E. D. Burling A. Jefferson and F. Scheinmann Tetrahedron 1965 21 2653; (6) A. Dyer A. Jefferson and F. Scheinmann J. Org. Chern. 1968,33 1259. 397 Molecular Rearrangements Related to the Claisen Rearrangement (27) but the main product (28) was due to an intramolecular para migration and gave the dimethyl ether of ugaxanthone (from the heartwood of Symphonia globulifera L.).12 The circumstantial evidence here would suggest that the para rearrangement product was not formed from an initial rearrangement to the blocked ortho position because with simple l-allyloxyxanthonesll~ lS no para migration could be detected.Thus a more plausible explanation is that rearrangement to the free ortho position would yield a dienone (29) where steric factors allow a further [3,3] sigmatropic shift to become a competitive process with enolisation to the 2-(aa-dimethylallyl)xanthone (27). Rearrangement to the para Position in Ally1 Phenyl Ethers with no ortho Substituents.-Borgulya Hansen Barner and Schmid14 observed that both ortho and para rearrangement can occur when both ortho positions are unsubstituted (30+31 + 32). The observations were made with the rearrangement of y-methyl and y-phenylallyl phenyl ethers (30) having substituents in the 3,5-positions of the benzene ring. The ortho :para ratio depends on the substituent size in the ally1 moiety and the nature of the solvent used.On heating 3,5-dimethylphenyly-methylallyl ether (30a) in NN-diethylaniline for 16 hours 79% of the ortho isomer (31a) and 21 % of thep-methylallylphenol (32a) were obtained whereas with 3,5-dimethylphenyl y-phenylallyl ether (30d) the yields of ortho (31d) and para (32d) rearrangement products were 11.5% and 88.5% respectively. The effect of solvent was demonstrated by also rearranging 3,5-dimethylphenyly-methylallyl ether in decalin benzonitrile and dimethylformamide under comparable conditions with the result that the ortho :para ratio of the products changed from approximately 1 1 (decalin) to 60 1 (in dimethylformamide). In addition in contrast to most Claisen rearrangements an isotope effect was observed in the enolisation of the o-dienone (33 or 34).Thus replacement of the aromatic protons by deuterium in 3,5-dimethylphenyl y-methylallyl ether (30a) gave more than double the amount of para rearrangement. Clearly therefore steric factors may retard the usually rapid enolisation process. More recent studies have shown that electronic factors can also influence the enolisation of the ortho-dienone (33 or 34). Thus introduction of halogens into the 3,5-positions where they can exert a strong -1 effect favours enolisation and gives a high ortho para ratio. As expected from these considerations the order of the effect is C1> Br > I > CH,.16 In a previous reviewlC it was stated that the same yield of ortho- and para- allylphenols was obtained in the rearrangement of 3,5-dimethylphenyl y-methylallyl ether (30a) and 3,5-diethylphenyl y-methylallyl ether (30b).An inspection of the data14 from both rearrangements carried out by heating in l1 H. D. Locksley I. Moore and F. Scheinmann J. Chem. Soc. (0 1966,2265. l8 P. Scheinmann and H. Suschitzky Tetrahedron 1959,7,31. l* J. Borgulya H.-J. Hansen R. Barner and H. Schmid Helv. Chim. Ada 1963,46,2444. versity of Zurich 1967. (a) Gy. FrAter and H. Schmid unpublished work; (b) Gy. Fritter Ph.D. Thesis Uni- 398 Jeferson and Scheinmann I a ; R= Me X= Me b ; R= Me,X= Et c ; R= Me,X=MeO d ; R = P h X = M e X o* I G CHR ( 3 2 1 diethylaniline at 186.5 O for 16 hr. shows that there are appreciable differences. Thus by replacing the methyl groups at the 3,5-positions by ethyl groups the extent of para migration increases (from 21-29%). The mechanism for the para rearrangements must also account for the fact that ay-dimethylallyl 3,5-dimethylphenyl ether rearranges to give only the ortho Claisen product.An examination of a possible transition state (36) for the para rearrangement provides an explanation. Thus if rearrangement to give the ortho-dienone (33) occurs in the likely stereochemical pathway the side chain will be attached in a pseudo-axial conformation to the cyclohexadienone (33). Interaction of the methyl group on the side chain with the neighbouring groups on the ring will hinder formation of the pseudo-equatorial conformer (34) and therefore hinder enolisation to the o-allylphenol. H x *; C Hl Free rotation of the side chain in the pseudo-axial conformation (33) gives a conformer (35) which has the correct orientation for the Cope-type rearrangement to the para-dienone.If it is assumed that the Cope rearrangement goes through the transition state with a quasi-chair conformation the diagram (36) implies that replacement of either hydrogen atom on the unsubstituted methylene of the ally1 moiety will cause a 1,3-diaxial type interaction with the groups on the ring. The experimental results suggest that such an interaction is sufficient to prevent the para rearrangement. Similar 1 Sinteractions are also possible in a quasi-boat transition state. 399 Molecular Rearrangements Related to the Claisen Rearrangement That steric hindrance is the driving force for the para rearrangement is further illustrated by the allylphenol rearrangement. Thus for example 2-a-methylallyl- 3,5-dimethoxyphenol (31c) gave on heating in diethylaniline for 48 hr.the thermodynamically more stable 4y-methylallyl-3,5-dimethoxyphenol (32c) in 80 % yield. Such a reaction presumably occurs by formation of an ortho-dienone (e.g.,35) followed by the Cope-type rearrangement. The failure of the 2-(aa-dimethylallyl)xanthone (27) to rearrange further is thus attributed to hydrogen bonding which prevents formation of an ortho-dienone.lla The previous examples may give the impression that in fact one meta substituent at least is required for the para rearrangement. However recently Scheinmann Barner and Schmid16a have shown that yy-dimethylallyl phenyl ether (37); rearranges in diethylaniline to 4-(~-dimethylallyl)phenol(39) in high yield; the other product (41) isolated arises from an abnormal Claisen rearrangement.Claisen reported,16 but without experimental details that the yy-dimethylallyl ether of phenol gave the expected ortho product (40) when heated in sodium carbonate. Attempts to repeat this work failed to give 2-(aa-dimethylallyl)phenol (40).16a Clearly steric interactions in the ortho-dienone (38) and the 2-(aa-dimethylallyl)phenol (40) [which is the intermediate for the abnormal product (41)] allow further sigmatropic rearrangements to give thermodynami- cally more stable products. Product control can again be exercised by solvent reaction in dimethylformamide favours formation of 2-(a#3-dimethylallyl) phenol (41) (89 %) by influencing the enolisation stage (38+40). 3 ' (39) The Out-of-ring Claisen Rearrangement.-That the Cope rearrangement may follow a Claisen rearrangement if such possibilities exist can account for a more striking variation of the Claisen rearrangement.In 1926 Claisen and Tietzel' found that a migrating ally1 group of a phenyl ether became attached l6 (a) F. Scheinmann R. Barner and H. Schmid unpublished work; (b) L. Claisen F. Kremers F. Roth E. Tietze J. prakt. Chem. 1922 105 65. l7 L. Qaisen and E. Tietze Annalen 1926,449 89. 400 Jeferson and Scheinmann to the central carbon atom of an o-propenyl side chain (e.g. 42 -+ 45). Lauerl* and Schmidlg and their co-workers both showed using different methods that rearrangement of the allyl group occurred without net ‘inversion’. Thus the American group rearranged the crotyl ether (42; R = CHd while the Swiss workers labelled the allyl ether (42; R = H) with 14C at the y-carbon atom and in addition showed that the reaction was intramolecular by carrying out a crossing experiment.By analogy with the mechanism for the para Claisen rearrangement it was suggested that the out-of-ring migration is a two-cycle process to form first the dienone (43) which rearranges to another dienone (44). Thus two ‘inversions’ occur during the course of the reaction. This mechanism + CH,CH=CHR 1 -= leeH-CH,CH=CHR 14 1 (43) (441 Me is therefore another example of two consecutive [3,3 ] sigmatropic rearrangements. However the Woodward-Hoffmann selection rules and the experimental data do not exclude the alternative mechanism (42 += 44) whereby formation of the dienone (44) occurs in one stage because this migration represents a [1,5] sigmatropic shift of an allyl group. Such a mechanism is not supported by experience since an allyl group does not migrate in this manner.To confirm the mechanism involving two [3,3] sigma- tropic shifts it would be necessary to trap the dienone (43) possibly as the maleic anhydride adduct. The ability of the allyl group to migrate in stages in the para rearrange- ment and probably in the out-of-ring migrations led Nickon and Aaronoff to investigate multistage rearrangements in which the allyl group can traverse W. M. Lauer and D. W. Wujciak J. Amer. Chem. Soc. 1956 78 5601. l9 K. Schmid P. Fahmi and H. Schmid Helv. Chim. Actu 1956 39 708. 401 Molecular Rearrangements Related to the CIaisen Rearrangement even greater distances?O Thus pyrolysis of the allyl ether of 2,6-dimethyl- 4-propenylphenol (46) in ethanol at 200" followed by catalytic hydrogenation gave 2,6-dimethyI-4-(2-methylpentyl)phenol (47) by three step-wise migrations of an allyl group (Scheme 3).To test whether the olefinic unit acting as the 0 CH = CHCH 0 allyl acceptor in out-of ring rearrangements could be part of a second phenyl ring the allyl ethers of several 2- and 4hydroxybiphenols were prepared.2O How- ever on pyrolysis these allyl ethers disproportionated to the corresponding parent phenols showing that the second phenyl ring was acting as a blocking group and not as an allyl acceptor. Other interesting examples of out-of-ring Claisen rearrangements have been reported by Makisumu210 who studied the migrations in nitrogen heterocylic compounds. By heating the allyl ether of 4-hydroxy-2,3-dimethylquinoline (48; R = H) at 200" migration to the meta methyl group (49) occurred in high yield (92%).The two other products that were isolated (50 and 51) resulted from cyclisation (49+ 50) and from para migration to nitrogen. If the side chain in the 2-position was extended it was shown that migration always occurred to the activated a-carbon atom of the side chain. Subsequent work21b showed that rearrangement to the meta side chain and para rearrangement were competitive processes which occurred from the same dienone-type inter- mediate (52) (Scheme 4). Other Claisen rearrangements in nitrogen heterocyclic systems have recently been reviewed by Thyagarajan.22 A. Nickon and B. R. Aaronoff J. Org. Chem. 1964 29 3014. *1 (a) Y . Makisumi Tetrahedron Letters 1964 699; (b) 1964 1635. la B. S. Thyagarajan 'Claisen Rearrangements in N-Heterocycles' in 'Advances in Heterocyclic Chemistry' ed.A. Katritzky Academic Press New York 1967 vol. 8 p. 143. 402 Jeflerson and Scheinniann OCH,CH= CH Q&H3 CH. CH,C€I- C H I (49) I R 148 i //d J tI The ortho-ortho Rearrangement.-It has been previously stated that an ortho-ortho rearrangement is not allowed by the Woodward-Hoffmann selection rules.s Despite this a number of ortho-ortho rearrangements have been postulated in order to account for experimental phenomena and reaction products.lb These rearrangements should now be interpreted as reactions which occw by a stepwise mechanism quite different from the concerted ortho to para migration discussed previously. In studying the reversibility of the Claisen rearrangement Schmid et aZ.23 observed that on heating a [yJ4C] allyl ether of 2,4,6-trimethylphenol (534 the radioactivity became distributed between the a and y-carbon atoms of the allyl group (53b) (Scheme 5).2b This phenomenon could be best accounted for by the occurrence of an ortho-ortho rearrangement with reversal of the ends of attachment (54a g54b).Scheme 5 as P. Fahrni and H. Schmid Helv. Chim. Actu 1959 42 1102. 1 Me ( 5 6 b ) 403 Molecular Rearrangements Related to the Claisen Rearrangement In the rearrangement of the but-2-enyl ethers of 16- and y-tocopherol (55a and 556) re~pectively~~ products arising from the expected ortho rearrangement (55c and 55e) and also from rearrangement without net inversion (55d and 55f) were isolated in good yieId. Although the rearrangements were accompanied by cleavage of the ethers (55a and 5%) to give the original tocopherols (55g and 55h) all attempts to detect ionic free-radical or intermolecular character failed.Thus Green McHale and MarcinkiewiczN concluded that the but-2-enyltocopherols are formed by an intramolecular ortho-ortho migration of an o-a-methylallyldienone intermediate. N.m.r. analysis was not available for this study and the structures of the reaction products were assigned on the basis of infrared analysis and colour reactions. a ; b ; R' = Me Ra = H R3 = CH2.CH:CH.CH3 R' = H Ra = Me R3 = CHa.CH:CH.CHs R1 = CH(CH3).CH:CH2 R2 = Me R = H R2 ' 0 Me f ; R1 = CH2.CH:CH.CH3 R2 = Me R = H Me ( 5 5 ) g ; R 1 = M e R 2 = R 3 = H R1 = Rs = H R2 = Me R 3 0 ~ c 1 6 H 3 3 c d ; ; R' R1 = = Me Me R2 = = CH(CHJ.CH:CHa CH2CH:CH.CH3 R3 R3 = = H H e ; h ; Another example of a possible ortho-ortho rearrangement was reported by Dinan and Tie~kelmann~~ who showed that the Claisen rearrangement of 2-but-2-enyloxypyridine in absence of solvent gives in addition to the expected 1,2-dihydro-1- and 3-(1-methylallyl)pyridonesY a product with no 'inversion' l-but-2'-enyl-l,2-dihydro-2-pyridone.An ortho-ortho migration of lY2-dihydro- 3-(l-methylallyl)pyridone may account for the latter reaction product. It has recently been suggested4q that the ortho-ortho rearrangement occurs in a stepwise manner by formation of an internal Diels-Alder adduct (Ma 56b) from the ortho-dienone (54a 54b). Cleavage of the four-membered ring would then complete the ortho rearrangement with inversion (54a + 54b). As yet an internal Diels-Alder adduct has not been isolated from the rearrangements of ally1 aryl ethers.However the rearrangement of 2,6-dimethylphenyl propargyl ether does give an internal Diels-Alder adduct (see p. 411)26 and this result will undoubtedly stimulate further work to isolate or trap the intermediates in the ortho-ortho rearrangement. The Abnormal Claisen Rearrangement.-In 1936 Lauer and Filbert reported that the rearrangement of y-ethylallyl phenyl ether (57) gave 2-(ay-dimethylallyl)- phenol (59).27 This type of reaction became known as the abnormal Claisen rearrangement. The expected product 2-(a-ethylallyl)phenol (58) was in fact also present28 and further work showed conclusively that it could be converted 24 J. Green S. Marcinkiewicz and D. McHale J. Chem. SOC. (0 1966 1422; Proc. Chem. SOC.1964 228. F. J. Dinan and H. Tieckelmann J. Org. Chem. 1964 29 892. 26 H. Schmid and J. Zsindely Helv. Chim. Acta. 1968 in the press. 27 W. M. Lauer and W. F. Filbert J . Amer. Chem. SOC. 1936 58 1388. 28 C. D. Hurd and M. A. Pollack J. Org. Chem. 1938 3 550. 404 Jeferson and Scheinmann into the abnormal product. Thus Marvell Anderson and Ong29 demonstrated that for the abnormal reaction to occur a hydroxyl group and an ally1 group must be located ortho to each other since the methyl ether (58; OMe instead of OH) and 4-(~-ethylallyl)-2,6-dimethylphenol both failed to undergo an abnormal rearrangement. The mechanism (57)-(60) was suggested in which formation of the abnormal product occurs by the reversible intramolecular hydrogen transfer of the spirocyclopropylcyclohexadienone intermediate (60).This mechanism was verified by the labelling experiments which revealed further subtle features. Thus work with 2-(oc-methylallyl)-4-methylphenol labelled with 14C at the a-methyl group showed that the radioactivity became distributed between the a-methyl group and the y-carbon atom of the side-chain.30 By use of a deuterium label it was demonstrated that deuterium could be exchanged between the terminal methylene and phenolic hydroxyl groups and that no incorporation of the label occurred at the a- and ,&carbon atoms of the sidechain." It was also shown that deuterium had been incorporated into the a-methyl substituent but at a much slower rate than in the terminal methylene group. The abnormal Claisen rearrangement of o-allylphenols is a general reaction which in the case of unsubstituted side-chains could only be demonstrated by the exclusive deuterium exchange between the phenolic hydroxyl group and the olefinic methylene group.The mechanistic pathway for the abnormal Claisen rearrangement is summarised for the rearrangement of 4-methyl-2-(a-methylallyl)phenol in Scheme 6. Formation of (trans-61) should be favoured over formation of (cis-61) since in the former case steric interaction of the cis-methyl substituents during formation of the cyclopropane ring will be avoided. It is for this reason that deuterium exchange is faster at the terminal methylene group than at the a-methyl group. In agreement with this explanation it was shown that the incorporation of deuterium at both sites 29 E. N. Marvell D. R. Anderson and J. Ong.J. Org. Chem. 1962,27 1109. so A. Habich R. Barner R. M. Roberts and H. Schmid Helv. Chim. Ada 1962 45 1943; similar conclusions were reached by W. M. Lauer and T. A. Johnson J. Org. Chem. 1963 28 2913. 81 A. Habich R. Barner W. von Philipsborn and H. Schmid Helv. Chim. Ada 1965,48 1297. 405 Molecular Rearrangements Related to the Claisen Rearrangement Me (cis - 61) M i O/D 0 A e Me Scheme 6 I C’H ‘CH h! e (trans - 61) H ”i’ (65) could be equalised by creating conditions in which both methyl groups in the cyclopropane ring are likely to be cis to one another as in the intermediate (cis-61). This occurred in the rearrangement of 3,5-dimethyl-2-(a-methylallyl)- phenol where the presence of the 3-methyl substituent would raise the free energy of activation for the formation of the trans-dimethylcyclopropane intermediate 406 Jeferson and Scheinmann (62).The geometrical isomerism of a reaction product can also be altered by the abnormal Claisen rearrangement as demonstrated by the interconversion of cis- and trans-2-crotyl-3,5-dimethylphenol (30a) and likewise cis- and trans-Z(ay-dime t hylallyl)-3,5-dimet h ylp hen01 at 200 O ?2 In the examples that have been discussed the abnormal Claisen rearrangement occurs at a slower rate than the rearrangement of the allyl aryl ether and there has been no difficulty in isolating the normal product. This is not always the case. By heating the 3,3-dimethylallyl ether of oestrone (63) in diethylaniline none of the expected product (64) could be isolated.= Instead 2-(ap-dimethyl- al1yl)oestrone (66) was isolated as the only rearrangement product and its formation was accounted for by the rapid transformation of 2-(aa-dimethylallyl)- oestrone (64) into the spiro-dienone (65) which could then revert into the more stable phenol (66).Thus when the rearrangement of the ether was carried out in dimethylaniline containing butyric anhydride the normal CIaisen product (64) was trapped as its butyric ester [64; CH,(CHJ2C02 instead of OH]. The phenol (64) was isolated by hydrolysis and rapidly isomerised in hot diethylaniline to the abnormal product (66). The abnormal Claisen rearrangement can now be used to account for hitherto unexpected rearrangement products from 3,3-dimethylallyl aryl ethers. Thus rearrangement of 2-(yy-dimethylally- loxy)-l,4-naphthaquinone (67) gave 2-(a~-dimethylallyl)-3-hydroxy-l,4-naphtha- quinone (68).= The thermal rearrangement of ethyl 4-(~~-dimethylallyloxy)- benzoate gave isoprene and ethyl 4-hydroxybenzoate but in addition 5-ethoxycarbonyl-2,2,3-trimethyl coumaran (69) was formed,3b presumably by cyclisation of an abnormal rearrangement product.36 From a synthetic standpoint products arising from an abnormal Claisen rearrangement must be expected whenever an allyl ether contains a 7-alkyl substituent.The limiting cases appear only when the intrinsic geometry necessary for the intramolecular hydrogen transfer is missing. Thus in cases where the hydroxyl group is strongly hydrogen bonded to a carbonyl group (e.g. 27) the o-allylphenol does not undergo further rearrangement even after lengthy heating. The fact that the abnormal Claisen rearrangement occurs by a thermal concerted intramolecular suprafacial transfer of hydrogen enables it to be classified as a [1,5] sigmatropic rearrangement for migration of hydrogen from spiro-dienone to phenol (e.g.83 Gy Frater and H. Schmid Helv. Chim. Acta 1966 49 1957. 33 A. Jefferson and F. Scheinmann Chem. Comm. 1966 239. 84 R. G. Cooke Austral. J. Sci. Res. 1950 3 481. 8s W. M. Lauer and 0. Moe J. Amer. Chem. SOC. 1943 65 289. *6 For examples of other abnormal Claisen rearrangements see W. M. Laucr G. A. Doldouras R. E. Hileman and R. Liepins J. Org. Chem. 1961,26,4785. 407 Molecular Rearrangements Related to the Claisen Rearrangement 60 -+ 59 and 60 -+ 58). By the principle of microscopic reversibility the reverse reaction (e.g. 59 -+ 60 and 58 +- 60) can also be considered as [1,5] shifts?'@ In classifying the abnormal Claisen rearrangement as a [1,5] sigmatropic change the lowest possible order has been assigned (see p.394). The reactions (58) -+ (60) and (59) -+ (60) can be regarded as [1,5] changes when one neglects the a-carbon atom of the side-chain in the assumption that it is a poor insulator of conjugation. The ally1 sidechain and the benzene ring are homoc~njugated.~~~ The reaction has general application and occurs in aliphatic compounds. Thus 1 -acetyl-2-alkylcyclopropanones (70) undergo easy ring-opening on heating to produce homoallylic methyl ketones (71) only when the acyl and alkyl groups are cis to one another.37 The all-carbon analogue the transformation of cis-1 -methyl-2-vinylcyclopropane into cis-hexa-lY4-diene occurs readily at 1600.38 (70) (71) 3 Formation of a Stable Dienone and the Retro-Claisen Rearrangement Although the Claisen rearrangement in principle is a reversible reaction its preparative value is in formation of allylphenols or homallylic carbonyl compounds.The rearrangement of 1 -allyl-2-allyloxynaphthalene (72) was exceptional in that a stable dienone (73) was isolated in 55% yieldFQ This product partly reverted to the starting material (72) when heated at 194'. With 2-allyloxy-l-(a-methylallyl)naphthalene in boiling dimethylaniline a dienone was not isolated but its presence was inferred by formation of 1 -allyl-2-(~-methylalloxy)naphthalene. Only rearrangement to C( 1) is possible since no rearrangement occurs to the 3-position in 2-allyloxynaphthalenes. Similar inequality of two ortho positions is observed in rearrangement of allyloxy-quinolines and -isoquinoline~.~~ 37 (a) R.M. Roberts R. G. Landolt R. G. Greene and E. W Heyer J. Amer. Chem. SOC. 1967 89 1404 and refs. therein; (b) see S. Winstein Non-classical Ions and Homo- aromaticity' in 'Aromaticity' Chem. SOC. Special Publ. 1967 No. 21 p. 1. 38 R. J. Ellis and H. M. Frey Proc. Chem. SOC. 1964 221; W. Grime Chem. Ber. 1965 98 756. 89 J. Green and D. McHale Chem. and Znd. 1964 1801. 40 H. Win and H. Tieckelmann J. Org. Chem. 1967 32 59 and refs. therein. 408 Jeferson and Scheinmann An interesting case of a retro-Claisen rearrangement occurring in theoretical yield was reported by Ansell and Leslie.4l The reaction of 2,3-dimethylbutadiene with o-chloranil gave the Diels-Alder adduct (74) which in boiling benzene rearranged to the more stable aromatic compound (75).In the aliphatic series a reverse Claisen rearrangement probably arises in the equilibrium (76) + (77).4a Here the ring structures create particularly favourable entropy factors for valence tautomerism but molecular models show that both isomers are under ring strain. It was established that seven parts of the bicyclo [3,1 ,O]hex-2-ene-6-endoformaldehyde (76) coexist in equilibrium with three parts of 2-oxabicyclo [3,2,1 ]octa-3,6-diene (77).a2 4 Rearrangement of Pentadienyl Phenyl Ethers Consider the highest occupied molecular orbital for the pentadienyl radical the phases at C(l) and C(5) are the same but there is a change of phase at C(3) (Figure 2). To predict the outcome of the rearrangement of pentadienyl phenyl Figure 2 ether by the thermal sigmatropic process it is necessary to consider #s of the pentadienyl moiety with #4 for the phenoxyl radical (see p.394). The same phase between oxygen and the para position would allow the pentadienyl group to undergo a [5,5] migration with ‘inversion’. Further since there is a change of phase between the oxygen and the ortho position of #* for the phenoxyl group it is only C(3) of the pentadienyl group which can become attached to 41 M. F. Ansell and V. J. Leslie Chem. Comm. 1967 949. M. Rey and A. Dreiding Helv. Chim. Acta 1965 48 1985. 409 Molecular Rearrangements Related to the Claisen Rearrangement the ortho position by a suprafacial process. The experimental work of Friter and S ~ h m i d ~ ~ is in complete accord with the orbital symmetry requirements.Thus rearrangement of pentadienyl phenyl ether (78a) gives the p-pentadienyl- phenol (80a) and also the o-allylphenol (79a). Introduction of a methyl group at either the 01- or o-position of the pentadienyl ether (78b and 78c) gave largely the same ortho product (846 = 84c) but different principal para products (80b and 80c). [In rearrangement of (78b) both the trans-o-allylphenol (79b) and its cis isomer were formed. As expected for the ortho rearrangement of a-alkylallyl phenyl ethers the trans product predominated (see p. 392). ] Thus consistent with a direct para migration with 'inversion' an a-methyl group in the ether (78b) resulted in p-pentadienylphenol with the methyl group at the w-position (80b) similarly the w-methylpentadienyl phenyl ether (78c) gave largely the a-methylpentadienylphenol (80c).These results preclude the formation of the p-pentadienylphenols (80) from an ortho-dienone intermediate and in addition an allylphenol rearrangement could be excluded since the o-allylphenols (79) could not be converted into their para isomers (80). Thus in the rearrangement of pentadienyl phenyl ethers (78) [3,3] and [5,5] sigmatropic changes are competitive reactions but the reaction of the higher order is significantly faster 6s 43 R' R' + O k R 2 I OH I (79) I r" 5. Rearrangement of Propargyl Vinyl and Aryl Propargyl Ethers Propargyl vinyl ethers (e.g. 82) undergo a Claisen-type rearrangement to give allenic carbonyl compounds (e.g. 84).u The reaction conditions are more severe than for the ally1 vinyl ether rearrangement but this is understandable if it is assumed that the planar transition state (e.g.83) for the formation of the allene carbonyl compound is not so favourable. In accordance with this concept of the transition state asymmetry can be induced into the allene by the starred asymmetric atom in the ether (82). Thus starting from (S)-but-3-yn-2-01 reaction with 2-methylpropanol formed the acetal (81) which on passage over silica at 210" gave (R)-2,2-dimethylhexa-3,4-dienol (84)?6 An elegant synthesis4s of pseudoionone (89) has been achieved which involves a 4s Gy. Frater and H. Schmid Helv. Chim. Acfa 1968 51 190. 44 D. R. Taylor Chem. Rev 1967 67 317 and refs. therein. 46 E. R. H. Jones J. D. Loder and M. C. Whiting Proc. Chem. SOC. 1960,180. 46 G. Saucy and R. Marbet Neb. Chim. Acta 1967,50 1158. 410 Jeferson and Scheinmann (CH-C- CHMe0I2CHCHMe2 rearrangement of a propargyl vinyl ether (87).In this method the propargyl vinyl ether (87) is prepared in situ by acid-catalysed trans etherification of the vinyl ether (86) with the alcohol (85) and in these circumstances migration occws under mild conditions. In the rearrangement of aryl propargyl ethers (90) in diethylaniline an o-allenylphenol (91) has as yet not been isolated and the corresponding chromene (92) is the only pr0duct.4~ Iwai and Ide47 concluded that direct cyclisation to form the pyran ring occurred without prior rearrangement to the a-allenylphenol (91). Their data can also be accommodated by formation and rapid cyclisation of an o-allenylphenol (91) since contrary to a previous report48 o-allenylphenol (91; X = R = H) on heating is converted into chromen (92; X = R = H).49 The rearrangement of 1,4-diaryloxybut-2-yne (e.g.93) gives the benzofuro- benzopyran (94) and the chromen (95) is believed to be an intermediate.s0 r71. Iwai and J. Ide Chem. Pharm. Bull. Japan 1962,10,926; 1963,11 1042. 48 R. Gatrtner J. Amer. Chem. SOC. 1951,73 4400. 4s J. Zsindely and H. Schmid unpublished work. soB. S. Thyagarajan K. K. Balasubrammian and R. Bhima Rao Tetrahedron 1967 23 1893. 41 1 Molecular Rearrangements Related to the Claisen Rearrangement CH,OPh I Ill I C H2 OPh (93 1 The studies on the rearrangement of 2,6-dimethylphenyl propargyl ether (96) provide strong circumstantial evidence for the formation of an allene-dienone intermediate (97). The tricyclic ketone (98) was isolated and its formation was best rationalised by postulating that the allene (97) from a Claisen rearrangement undergoes an internal Diels-Alder-like reaction (Scheme 7).49 M e a M e OCH,-C ZCH Me@ 0 Me@ 0 4/ (97) \ (98) ' (96) Scheme 7 6 The Rearrangement of Benzyl Vinyl Ethers The success of the Claisen rearrangement of allyl phenyl ethers where the vinyl moiety is part of an aromatic ring stimulated studies in the rearrangement of benzyl vinyl ethers where the allylic double bond is incorporated in a benzene ring.Benzyl vinyl ether (99u) however failed to give o-tolylacetaldehyde (1OOa) and similar failures were recorded with naphthalene and anthracene analogues.51 Instead benzyl vinyl ether gave 3-phenylpropanal. Modifications in the aromatic portion or alternatively in the vinyl moiety however led to the expected 63 Thus 3,5-dimethoxyphenyl isopropenyl ether (996) gave 80 % of 2,4-dimethoxy-6-methylphenylacetone (100b) when heated at 240" but data on the mechanism have not been reported.62 3,5-Dimethoxybenzylacetone was also isolated and this product probably arises from a free-radical scission-recombination mechanism.It thus appears that the mesomeric donation of electrons by the methoxyl groups to the benzene ring promotes the rearrangement of the benzyl vinyl ether (996). By the use of the acetal of NN-dimethylacetamide (101) it is possible to prepare 'in situ' an aminovinyl benzyl (or allyl) ether (e.g. 103) which undergoes a Claisen rearrangement at 140-1 go0.= The Claisen intermediate is probably formed by alcohol exchange of the vinyl ether (102). By use of benzyl alcohol and the acetal of NN-dimethylacetamide (101) and heating at 1 80" NN-dimethyl-o- tolylacetamide (104) was isolated in 50% yield.In this case resonance stabilisation of the carbonyl function by the dimethylamino-group helps to favour product formation.53 61 A. W. Burgstahler L. K. Gibbons and I. C. Nordin J. Chem. SOC. 1963 4986. 6a W. J. Le Noble P. J. Crean and B. Gabrielson J. Amer. Chem. SOC. 1964 86 1649. A. E. Wick D. Felix K. Steen and A. Eschenmoser Helv. Chim. Ada 1964,47 2425. 412 Jeferson and Scheinmann 7 AUyl Migration in the Wittig Rearrangement The Wittig rearrangement involves the base-catalysed rearrangement of certain benzyl and diphenylmethyl ethers to alcohols." The allyl groups migrate under mild conditions by a cyclic intramolecular pathway with inversion of the allyl group as shown by the rearrangement of allyl quinaldyl ethers (105 3 106)55 and allyl 9-fluorenyl ethers (107 -+ as their anions.C HR' // -CH (108) The reactions can be considered as [2,3] sigmatropic rearrangements which involve migration of a negative charge. Since there are six electrons in the transition state (or e.g. eighteen electrons if all thev-electrons in the aromatic orbitals are included for rearrangement of allyl Pfluorenyl ether) the processes are thermally allowed (see p. 395). From this example it is clear that the Woodward-Hoffmann rule i + j = 4n + 2 for thermal [i j ] sigmatropic changes is obeyed only if there is no migration of charge. For migration of D. J. Cram 'Fundamentals of Carbanion Chemistry' Academic Press New York 1965 Y.Makisumi and S. Notzumoto Tetrahedron Letters 1966 6393. p. 230. ria V. Schollkopf and K. Fellenberger Annulen 1966 698 80. 413 Molecular Rearrangements Related to the Cldsen Rearrangement charge as in the Wagner-Meerwein and the above Wittig reactions the rule requires modification to i + j = 2k + 3 (where k = 0 1 2 3 . . .). The Hiickel molecular orbital coefficients for dibenzofulvene have been calculated and it can be shown that the anionic rearrangement of the all-carbon analogue of allyl 9-ffuorenyl ether is a symmetry-allowed process. Thus the highest occupied molecular orbital for the anion of dibenzofulvene is t,b8 (109)579 9 b and the transition state for migration of an allyl radical requires change of phases between C(9) of the ring and the substituent atom.f 4- 8 Nitrogen Sulphur and Phosphorus Analogues of the Claisen Rearrangement Before 1961 nitrogen sulphur and phosphorus analogues of the Claisen rearrangement were unknown. An early attempt to rearrange N-allylaniline at 275 O gave aniline and propene. 58 Marcinkiewicz Green and M a m a l i ~ ~ ~ estimated that the activation energy for rearrangement of N-allylaniline is about 6 kcal./mole higher than for the rearrangement of allyl phenyl ether. Thus at the higher temperatures required for allyl migration in N-allylaniline preferential allyl cleavage occurred.59 However it was successfully predicted that N-allyl-l-naphthylamine would rearrange to 2-allyl-1-naphthylamine because Claisen rearrangements in the naphthalene series require a lower energy of activati~n.~~ Since the orbital symmetry requirements do not forbid the rearrangement of N-allylanilineGo by a concerted mechanism the problem of allyl migration essentially is how to reduce the activation energy required to reach the transition state.It is known that allyl viny161 and allyl aryl etherss2 rearrange more readily when proton acids or Lewis acids are present probably because the resulting positive charge on the oxygen atom is further delocalised in the transition state. It has been reported that N-allylaniline gives 2-all~laniline~~ and that N-allyl-2,6-xylidine gives 4-allyl-2,6-~ylidine~~ in the presence of zinc chloride Unfortunately there were no data to show whether the rearrangements occurred by a concerted mechanism or by an intermolecular pathway. It has recently been observed that 1 -p-bromophenyl-2-isopropenylaziridine (110) is smoothly converted into the benzazepine (111) in high yield.s5 In this 67 C.A. Coulson and A. Streitweiser ‘Dictionary of n-Electron Calculations’ W. H. Freeman and Co. San Francisco 1965 p. 140. 68 F. L. Canahan and C. D. Hurd J. Amer. Chem. Soc. 1930,52,4586. 59 S. Marcinkiewicz J. Green and P. Mamalis Tetrahedun 1961 14 208. 6o Ref. 57 p. 263. 61 G. Saucy and R. Marbet Helv. Chim. Acta 1967,50,2091,2095. 6a H. Schmid Gazzetta 1962 92 968 and unpublished work. C. D. Hurd and W. D. Jenkins J. Org. Chem. 1957 22 1418. 64 M. Elliot and N. F. Janes J. Chem. Soc. 1967 1780. 65 P. Scheiner J. Org. Chem. 1967 32,2628. 414 Jeferson and Scheinmunn case it was suggested that an amino-Claisen rearrangement was made feasible by the relief of aziridine ring strain (12-14 kcal./moIe) in the transition state.66 The rearrangement of N-allyleneamines provides a method for alkylating aldehydes at the a-position without the use of an alkylating agent or a strong base.66 Thus enamines prepared from N-allylamines (e.g.1 12) and a-disubstituted aldehydes rearrange quantitatively on heating at 200-250" to give an imine (e-g. 113) which on hydrolysis yields an aldehyde (e.g. 114). The rearrangement proceeds with inversion and in a stereospecific manner.66 A variation of this reaction was reported by Ficini and Barbara.67 An aminal(116) prepared by the reaction of an ynamine (115) with an allylamine rearranged at 280" to give the amidine (1 17). Ally1 phenyl sulphidc (118) on heating in either NN-diethylaniline or quinoline gives 2-methylthiocoumaran (122) and thiochroman (121).680 No 2-allylthiophenol was obtained but it may be one of the intermediates in the formation of the heterocyclic compounds (121 and 122).68b Thus while 2-allylthiophenol gave both the thiochroman (121) and 2-methylthiocoumaran (122) on heating under the same conditions as used for ally1 phenyl sulphide the proportions of the cyclic products were quite different.It was therefore R. K. Hill and N. W. Gilman Tetrahedron Letters 1967 1421. (a) H. Kwart and E. R. Evans J. Org. Chem. 1966 31 413; (b) H. Kwart and H. M. L~'J. Ficini and C. Barbara Tetrahedron Letters 1966 6425. Cohen ibid. 1967 32 3135. 41 5 Molecular Rearrangements Related to the Claisen Rearrangement suggested by Kwart and his co-workers68 that allyl phenyl sulphide can also cyclise to (119) and that octet expansion occurs to give a thiiran intermediate (120).Opening of the three-membered ring can occur in two ways to give both the 2-methylthiocoumaran and the thiochroman (Scheme 8). The later work68 suggests that in the thio-Claisen rearrangement the amine solvent plays an important role in the reaction. The rearrangement of allyl and propargyl phosphites resembles the Claisen rearrangement in that migration to the phosphonate occurs with ‘inver~ion’.~~ There may be some duality of mechanistic pathways in some cases since rearrangement of diethyl crotyl phosphite (123) gave largely the cc-methyl- allylphosphonate (124a) but in addition some of the crotylphosphonate (1246) was formed.70 9 Developments in Aliphatic Claisen Rearrangements Although aromatic Claisen rearrangements have been studied in most detail there have been some notable synthetic applications in aliphatic chemistry.The introduction of an angular group with the correct stereochemistry in polycyclic terpene and steroid syntheses has never been easy,” but with a Claisen rearrangement of the appropriate ally1 vinyl ether (e.g. 125 126) an angular group can be introduced in a stereospecific manner.72 The preparation of an allyl vinyl ether involves an ether-exchange reaction of a vinyl ether with an allyl alcohol,’3 and recently the procedures have been 68 A. N. Pudovik and I. M. Aladzheva Doklady Akad. Nauk S.S.S.R. 1963 151 1110. 70 A. L. Lemper and H. Tieckelmann Tetrahedron Letters 1964 3053. 71 J. W. Cornforth ‘The Total Synthesis of Steroids’ in ‘Progress in Organic Chemistry ed.J. W. Cook Butterworths London 1955 vol. 3 p. 1. 78 W. H. Watanabe and L. E. Conlon J. Amer. Chem. SOC. 1957,79,2828. M. Torigoe and J. Fisham Tetrahedron Letters 1963 1251. 416 Jefferson and Scheinmann OAc simplified. Thus by the method of Eschenmoser (see p. 411),= or by the acid-catalysed reaction of a vinyl carbinol (127) with a vinyl ether (128),61 it is unnecessary to isolate the ally1 ether (129) in the formation of unsaturated carbonyl compound (130) in high yield (Scheme 9). FOH + R’ R3 (130) R‘ R’ (129) Scheme 9 y8-Unsaturated amides (134) can be prepared by the reaction of an ynamine (131) with an allylic alcohol (132) in the presence of boron trifluoride ether complex below 30°.s7 The reaction is complete within 1 hr. and high yields are obtained from primary allylic alcohols.The reaction can be understood by postulating a Claisen rearrangement of a complexed ynamine-ally1 alcohol adduct (1 33). &\C=CH (134) Me An elegant application of two [3,3] sigmatropic changes has been used by Thomas in the synthesis of or-sinensal (137) which was isolated from the Chinese orange Citrus sinensis L.74 The basis of the synthesis was a Claisen 74A. F. Thomas Chern. Cornm. 1967 946 and unpublished work. 417 Molecular Rearrangements Related to the Claisen Rearrangement rearrangement of an ether (135) which gives an aldehyde (136). Free rotation as indicated would allow the 1,5-diene system to assume the correct stereochemistry for a Cope reaction to give a sinensal. If chair-like transition states are assumed for the Claisen and Cope rearrangements stereochemical the reaction product can be derived.information about Thus from the Claisen rearrangement the resulting aldehyde can be rotated about the single bond to give two possible chair-like conformations (138) and (139) which would lead to two different sinensals (140 and 141) from Cope rearrangements. However the conformation with the large R group equatorial (139) will be preferred and this will lead to the trans double bond at C(6) (140). If it is assumed that under the conditions of the reaction (50 hr. at 98") that the aldehyde (140) would isomerise to the trans isomer then on the basis of the synthesis a-sinensal was postulated as 2,6-dimethyI-lO-methylene- dodeca-2-trans,6-trans,ll-trien-l-al (1 37). The consecutive Claisen and Cope '(141) RAH rearrangements are analogous to the para Claisen rearrangement in the aromatic series and the scope of the reaction with aliphatic compounds is under investigation.For aldehyde intermediates with an a hydrogen atom the rate of double-bond migration to the a/3 position can be competitive and even much faster than the Cope rearrangement.'* 41 8 Jefferson and Scheinmann 10 Concluding Remarks This survey was prompted by our observations during synthetic work that some Claisen rearrangements gave unexpected products and that it was necessary to examine the structural features which lead to related intramolecular rearrangements. The choice of solvent has often been underestimated from our experience a tertiary aromatic amine such as NN-diethyl- or NN-dimethyl- aniline has proved to be a most versatile solvent for aromatic Claisen rearrangements.Phenolic solvents or even phenolic reaction products in neutral solvents or in absence of solvent may catalyse the elimination of a substituted allyl group from its phenyl ether or cause the migration of the double bond in 2-allylphenols and cyclisation to c ~ u m a r a n ~ ~ or chroman derivatives.la Dimethylallyl groups for example are particularly sensitive and care must be taken in purification of the ether.12 The acid-catalysed cyclisation of 2-allylphenols to coumaran or chroman derivatives follows the rules of Markownikoff addition and these ring-closures can be used to check the structure of the side-chains. It is remarkable that although the Claisen rearrangement was discovered in 1912 new observations and developments in this area of chemistry continue.The rate of progress during the last few years has been particularly enhanced by the use of gas and thin-layer chromatography to detect unusual products. Structural analysis by physical methods especially n.m.r. enable the fate of an allyl group to be established rapidly without the ambiguity of the more classical methods. The Woodward-Hoffmann rules now provide a new theoretical basis for correlating earlier experimental results and for stimulating further research. Appendix added in Proof New work by Herriott and M i s l o ~ ~ ~ has shown that allyl phosphinates can rearrange to allyl phosphine oxides by an intramolecular concerted pathway with complete specificity. The duality in reaction pathways reported previously70 (see page 415) can be avoided by selecting milder experimental conditions which favour only the sigmatropic process.Ally1 sulphenates (142) rearrange to allyl sulphoxides (143) by a concerted cyclic pathway and this is well supported by the activation parameters and labelling experiment^.^^ The rearrangements of allyl phosphinates and sulphenates are formally analogous to the allyl migration in the Wittig reaction (see Section 7) in that they are 12 31 sigmatropic processes. In this treatment the unshared p-electrons on phosphorus and respectively sulphur fulfill the role of the carbonion electrons for six rr-electrons to participate in the transition state. '13 A. W. Herriot arid K. Mislow Tetrahedron Letters 1968 3013. 77 P. Bickart F. W. Carson J. Jacobus E. G. Miller and K.Mislow J. Amer. Chem. SOC. 1968 90 in the press. 75A. T. Shulgin and A. W. Baker J. Org. Chem. 1963 28 2468. 419 Molecular Rearrangements Related to the Claisen Rearrangement e.g. ,07 0- :HZ +\ ,CH CH3C,H4-S FHZ f=t CH,CIH4-S’ ‘-&H CH (142) (143) Attention has been focussed on the rearrangement of allylic sulphur y l i d ~ ~ ~ (e.g. 144+ 145) because of their possible role in C-C bond formation in t 3 - s\ R R= Me,H,Me C= CH- the biosynthesis of squalene. These reactions also show a formal resemblance to ally1 migration in the Wittig reaction where the oxygen atom is now replaced by the sulphur moiety. Rearrangement involves a [2 31 sigmatropic bond migration with six m-electrons in the transition state. A special case of Diels-Alder adducts (146) undergoing a reverse Claisen re- arrangement has been reported by Hughes and Williams.79 The adducts (146) from reaction of fulvenes and cis-hex-3-en-2 5-dione form dihydropyran derivatives (148) by Scheme 10.Scheme 10 ’* R. Bates and D. Feld Tetrahedron Letters 1968 417; B. M. Trost and R. LaRochelle Tetrahedron Letters 1968 3327; G. M. Blackburn W. D. Ollis J. D. Plackett C. Smith and I. 0. Sutherland Chem. Comm. 1968 186; J. E. Baldwin R. E. Hackler and D. P. Kelly Chem. Comm. 1968 537. 79 M. T. Hughes and R. 0. Williams Chem. Comm. 1968 587. 420 Jeflermn and Scheinmann Steric compression of the cis-endo diacetyl substituents in the adduct (146) and extension of conjugation in the transition state (147) due to a developing buta- diene system are responsible for rearrangement. More recent work on the isomerisation of bicyclo [3 1 01 hex-2-ene-6-endo formaldehyde (76) shows that above 110" in addition to the reverse Claisen rearrangement (page 408) a series of [l 51 hydrogen shifts occur.8o This Review was initiated in 1965 and finalised while one of us (F.S.) was on leave at the University of Zurich. We thank Professor Hans Schmid and Dr. Gy. FrAter Dr. H.-J. Hansen and Mr. J. Zsindely for many stimulating discussions and for permission to include unpublished work. We also thank Professors Roald Hoffmann G. Wagnih-e K. Mislow A. Dreiding and Dr. A. F. Thomas for personal communications. We are grateful to the University of Salford for the award of a Fellowship (to A. J.) and for leave of absence (to F. S.). We thank the Royal Society for an award (to F. S.) in their Euro- pean Fellowship Programme. F. Bickelhaupt W. L. de Graaf and G. W. Kiump Chem. Comm. 1968 53; A. Dreiding and M. Rey unpublished work. 421

ISSN:0009-2681    

DOI:10.1039/QR9682200391

出版商:RSC    

年代:1968

数据来源: RSC